The origin of structural variety of alkyne complexes of d8 metals. An example of structural isomerism

Pdyhedron Vol. 9, No. H/16, pp. 1.367-1881, Printed in Great Britain

1990 0

0277-5387/W S3.00+.00 1990 Pergamon Press pk

VARIETY OF ALKYNE THE ORIGIN OF STRUCTURAL COMPLEXES OF d8 METALS. AN EXAMPLE OF STRUCTURAL ISOMERISMT GREGORY MARINELLI,

WILLIAM

and KENNETH

E. STREIB, JOHN G. CAULTONI

C. HUF’FMAN

Department of Chemistry and Molecular Structure Center, Indiana University, Bloomington, IN 47405, U.S.A. and MICHEL

R. GAGNE and JOSEF

TAKATS

Department of Chemistry, University of Alberta, Edmonton, Alberta, Canada T6G 2G2 and MIC!I-@LE DARTIGUENAVE Laboratoire de Chimie de Coordination,

CNRS, 205 Route de Narbonne, 3 1077 Toulouse, France and

CATHERINE

Laboratoire

CHARDON,

SARAH

de Chimie Thiorique,

A. JACKSON

and ODILE EISENSTEINZ

Centre de Paris-Sud, 91405 Orsay, France

Abstract-The X-ray structures of three metal alkyne complexes are presented: Ir(PMe, Ph),(MeCzMe)+, M(C0).,(F3CC2CF3) (M = Ru, OS). The iridium(I) complex adopts an unusual structure (“CS”) which differs notably from that expected for four-coordinate d8 ML4 square-planar complexes. The structure resembles that of a square pyramid, with essentially C, symmetry, with a short apical M-P bond (2.236 vs 2.31 w for the basal M-P bonds), short M-C (2.01 A) and long C-C bonds (1.306 A), which indicate a strong metal-alkyne interaction. The ruthenium(O) and osmium(O) complexes adopt the expected pseudo-octahedral structure. The unusual CS structures, previously reported for cobalt(I) and iron(O) complexes, contrast with the square-planar structure of platinum(I1) alkyne complexes. Fluxionality is present in the iridium complex, as well as in those of cobalt, and is attributed to a rapid intramolecular site exchange among inequivalent phosphorus ligands. The iridium complex reacts with additional ligands like alkynes and HZ. Extended Hiickel calculations are presented to rationalize the structural and reactivity aspects of the whole family of d* ML,(alkyne) species (M = Pt”, Ir’, Co’, Fe’). It is shown that the fourelectron donation of an alkyne ligand is responsible for the unusual CS structure of certain of these ML,(alkyne) species. The alkyne moieties induce a four-electron destabilization in the square-planar structure which is large for diffuse d orbitals such as those of iridium(I), cobalt(I), iron(O), but not of platinum(I1). The fluxionality of the complexes is shown to be associated with an easy rotation of the alkyne about the metal-alkyne midpoint,

t This paper is dedicated to Richard Fenske for the advice and stimulus he provided to his student, K.G.C. $ Authors to whom correspondenceshould be addressed. 1867

G. MARINELLI et al.

1868

accompanied by a breathing motion of the M-L bonds of the ML, metal fragment. It is shown that unsymmetrical alkynes stabilize a conformation of the complex in which the ligand is turned by 90” with respect to the one presented here. The reactivity towards additional ligands is shown to be associated with the presence of a low-lying LUMO pointing away from the apical bond. Effects of substituents at the alkyne and of ligands at the metal are discussed. It is shown that it is the residual four-electron destabilization in M(CO),(alkyne) complexes which is responsible for the lability of CO ligands and thus for the ligand exchange process.

A number of ligands have the ability to exist in structurally isomeric forms (e.g. acyl, NO, C02, SOZ, alkyl, 1). Such isomerism is associated with a change in the number of ligand electrons shared with the metal valence orbitals. Most often the isomerit bonding forms are identifiable by clearly defined structural changes, but there are a few cases where the structural changes are quite subtle and thus difficult to perceive even when accurate structural parameters are known. One such example is the alkyne ligand. Here, in the absence of clearly diagnostic structural changes with change in electron donor number, molecular stoichiometry is used to evaluate the donor number of the alkyne moieties. Thus, it is conventional to perform a valence electron count on the fragment ML,, to which the alkyne is attached. If it is a 16-electron fragment, alkyne--+M donation from only one of the two orthogonal alkyne rt orbitals is assumed (“two-electron donor alkyne”). If it is a 14-electron fragment, donation from both alkyne x orbitals may take place. The reason for the qualification on this latter conclusion is that, for certain early and late metals, a 16-valence electron count is found in

M-N-O

__

M-N\ .?’

z

Ml1

0 M-C’

'0

C4

many isolable compounds. For example, in the case to be considered here, iridium(I) and platinum(I1) are frequently “content” with a 16-valence electron count, and thus the above method does not lead to a clear deduction about the alkyne-metal bonding in such ML,(alkyne) complexes. An interesting question that immediately arises concerns the consequences of having a ligand able to donate more electrons to the metal than this centre may accept. For instance, the well-known tendency of two-electron donor terminal alkynes to rearrange to vinylidene ligands [eq. (l)] may be rationalized by noting that such rearrangement diminishes the repulsion between the unused alkyne x-system and the filled metal d orbitals.’ Internal alkynes in two-electron donor situations do not have this outlet, however. R

M-ii -

M=C=CHR

(1)

C H

The present work offers structural results on some relevant compounds, an analysis of the bonding in several of the problematic situations identified above, and a rationalization of the observed chemical reactivity for each. *

0

0 EXPERIMENTAL

0

M-s;

/p

z=

0

H

H M-C!

,‘,#H

M-Fi 0

z=L

H

Mc\c. L"'H H

1 Scheme 1.

General procedures

All manipulations were performed under purified N2 or Ar using standard Schlenk techniques or a Vacuum Atmospheres glove box. Solvents were dried and deoxygenated by distillation over Na/Kbenzophenone ketyl under N2. Methylene chloride was distilled over P205 under N2, then subjected to a minimum of three freeze-pump-thaw degassing cycles to remove residual 02. Proton and phosphorus NMR spectra were

1869

An example of structural isomerism obtained on a Nicolet NT-360 at 360 and 146 MHz, respectively. Carbon NMR spectra were recorded on either a Nicolet NT-360 (91 MHz) or Bruker AM-500 (125 MHz). Proton chemical shifts are reported in units of 6 (ppm downfield from tetramethylsilane) and were measured relative to residual solvent protons. Carbon shifts are reported in units of 6 (ppm downfield from tetramethylsilane) and were measured relative to solvent absorption. Phosphorus shifts are reported in units of 6 (ppm downfield from external 85% H,PO,). Preparation of [Ir(PMe,Ph),(Me02CC2C0,Me)J BF4 (1) Reaction of [Ir(PMesPh)&Me2)]BF4 with MeO,CC,CO,Me. [Ir(PMezPh)&Me2)]BF4 (51 mg, 0.084 mmol) was placed in a Schlenk flask under Nr. THe solid was then dissolved in CHzClz (10 cm3), after which MeO,CC,CO,Me (0.042 cm3, 0.34 mmol) was added by syringe. The solution was stirred for 10 h, resulting in no apparent colour change. Solvent and excess MeO,CC,CO,Me was removed in vucuo, quantitatively giving [Ir (PMe,Ph),(MeO,CC,CO,Me),]BF, as a red oil. (2) By reaction of [IrH4(PMe2Ph)3]BF4 with MeO$C&O,Me. To a CHrCl, solution con-

taming [IrH4(PMe2Ph)3]BF4 (0.085 mmol) was added Me02CC2C0,Me (0.05 cm3, 0.41 mmol, 5 equiv) by syringe. After 30 min of stirring, the originally pale yellow solution had changed to dark orange. The solvent was removed in vucuo, and THF (4 cm3) was added to precipitate a red solid. The solution was concentrated to half its original volume to promote further precipitation, and the product was filtered. ‘H NMR (CD&l,): 7.4-7.1 (m, 15H), 3.91 (s, 6H), 3.78 (s, 6H), 1.87 (d, J= 11 Hz, 6H), 1.39 (vt, J= 6, 12H); 3’P NMR (CD2C12) -41.9 (d, J= 16), -57.3 (t, J = 16) ; 13C {‘H} NMR (CDJlr, 125 MHz): 160.1 (s), 160.0 (s), 128-131 (phenyl), 81.9 (d, J = 9 Hz), 79.4 [d (J = 24) of t (J= 36)], 54.24 (s), 53.36 (s), 17.98 (d, J= 36), 11.71 (vt, J = 37). X-ray

structure determination

PMe2Pb1BF4 Bond distances (A) Pl l-Cl2 Pll-Cl3 Pl l-Cl4 P2o-C21 P2O-C22 P2O-C23 F34-B33 F35-B33 F36B33 F37-B33

2.3090(H) 2.3119(H) 2.2355(U) 2.016(5) 2.014(6) 1.487(8) 1.306(8) 1.485(8) 1.826(6) 1.826(6) 1.828(6)

1.824(7) 1.817(7) 1.826(6) 1.809(6) 1.809(6) 1.813(6) 1.377(8) 1.390(8) 1.391(10) 1.393(10)

Bond angles (“) P2-Irl-Pll P2-Ir l-P20 P2-Ir l-C30 P2-Irl-C31 PI I-Irl-P20 Pl I-Irl-C30 Pll-IrlX31 P20-Irl-C30 P20-Irl-C3 C30-Irl-C3 Ir l-C3O-C29

1 1

106.14(5) 90.62(5) 137.35(16) 104.37(16) 94.07(5) 99.37(17) 135.07(16) 121.34(15) 117.80(16) 37.82(22) 145.1(4)

[Ir(MeC,Me)

A suitable crystal was cleaved from a larger piece grown by layering Et,0 onto a l,Zdimethoxyethane solution of the compound. The crystal was mounted using silicone grease and transferred to a goniostat where it was cooled to - 139°C for characterization and data collection

Table 1. Selected bond distances (A) and angles (“) for [Ir(Me&Me)

Irl-P2 Irl-Pll Irl-P20 Irl-C30 Irl-C31 C29-C30 c3O-C3 1 C3 l-C32 P2-C3 P2-C4 P2-C5

of

0+WV31BF4

Irl-C3O-C31 c29-C3O-C3 Irl-C31-C30 Irl-C31-C32 C3O-C31-C32 F34-B33-F35 F34-B33-F36 F34-B33-F37 F35-B33-F36 F35-B33-F37 F36B33-F37

1

71.0(3) 143.7(6) 71.2(3) 149.8(4) 138.9(6) 110.4(5) 109.2(6) 108.8(7) 109.3(7) 109.5(6) 109.716)

1870

G. MARINELLI

(6” < 28 < 45”).3 A systematic search of a limited hemisphere of reciprocal space revealed intensities with Laue symmetry and systematic absences consistent with space group P2,/n. After correction for absorption, data processing gave a residual of 0.027 for 529 unique intensities which had been measured more than once. Four standards measured every 300 data showed no systematic trends. The structure was solved using a combination of direct methods (MULTAN78) and Fourier techniques. After the non-hydrogen atoms had been located and refined, a difference Fourier map revealed most of the hydrogen atoms. All hydrogen atoms are placed in calculated positions prior to the final cycles of least-squares in which the nonhydrogen atoms were refined with anisotropic thermal parameters and the hydrogen atoms were refined with isotropic thermal parameters. The final difference map had a residual near iridium of 2 e/A3. All other residual peaks were well under 1 e/A3. The results? of the X-ray study are shown in Table 1 and Figs 1 and 2. Additional details are available as supplementary material. Refined C-H distances range from 0.84(7) to 1.18(8) A.

X-ray difiaction (M = Ru, OS)

study of M(C0)4(F,CC2CF3)

A suitable crystal of the osmium compound was located and transferred to the goniostat using inert atmosphere handling techniques and cooled to - 155°C for characterization and data collection.3 A systemetic search of a limited hemisphere of reciprocal space located a set of diffraction maxima with symmetry and systematic absences corresponding to the unique monoclinic cell P2Jn. Subsequent solution and refinement of the structure confirmed this choice. Data were reduced to a unique set of intensities and associated sigmas in the usual manner. There was not significant decomposition at - 155°C based on the intensities of four standards monitored every 2 h. The structure was solved by a combination of direct methods (MULTAN78) and Fourier techniques. Data were corrected for absorption before the final anisotropic refinement. A final difference Fourier indicated several other peaks of intensity

t Supplementary material available, anisotropic thermal parameters, parameters of the crystal and data collection, and all distances and angles have been deposited with the Cambridge Crystallographic Data Centre, University Chemical Laboratory, Lensfield Road, Cambridge CB2 IEW, U.K.

et al.

Fig. 1. ORTEP view of the cation Ir(MeC,Me)(PMe, Ph)3+. omitting hydrogens and showing selected atom labelling. This view emphasizes the approach to squarepyramidal geometry, with iridium lying above the approximate planar array of Pl 1, P2, C30 and C3 1.

1.8-2.2 e/A3 in the vicinity of the osmium atom and no other features. The large “ripples” near the osmium are probably due to an inaccurate absorption correction. The results? of the structural study are shown in Table 2 and Fig. 3. The ruthenium compound is isomorphous and the data collection proceeded as for M = OS. No absorption correction was necessary. A final different Fourier was featureless, the largest peak being 0.7 e/A3. The atom numbering is identical to that of the osmium analogue. RESULTS

Structure of Ir(MeC,Me)(PMe,Ph),BF, Ir(MeC,Me)(PMe,Ph),BF, exists in the solid state as non-interacting Ir(MeC,Me)(PMe,Ph),+ and BF4- ions. The closest approach of fluorine to iridium is 5.54 A. The coordination geometry about iridium in the cation (Fig. 1) is clearly not planar or even distorted planar. Because of the large disparity in P-Ir-P angles (90.6, 94.0 and 106.1”) and Ir-P distances (2.309, 2.312 and 2.236 A), it is equally inappropriate to characterize the coordination geometry as distorted tetrahedral. The one (idealized, non-crystallographic) symmetry element of the cation is a mirror plane containing P20, Ir and the C30-C31 midpoint. The phosphorus in this mirror plane has a dramatically shorter Ir-P distance than those involving the pair of “symmetry related” phosphines. To this “mirror symmetry” description can be added the observation that P2,

An example of structural isomerism

1871

Fig. 2. Stereo stick figure and space-filling drawings of Ir(MeC,Me)(PMe,Ph),+ from the direction opposite P20. The hatched atom is iridium. All C-H distances have been drawn at 1.08 A.

Pl 1 and alkyne carbons C30 and C31 are nearly

MeO,CC,CO,Me as a &and on the Ir(PMe2Ph)3+ planar (to f 0.04 A), with iridium 0.49 8, above this fragment plane (Fig. 1). Thus, because of the unique position Reaction of Ir(H2)(H)2(PMezPh)J+ with dimeof P20, the coordination geometry might be called thy1 acetylenediacarboxylate, under conditions quasi-trigonal planar (alkyne midpoint considered used for the synthesis of Ir(MeC2Me)(PMe2Ph)3’, as one site) to which has been added a tightly bound gives a bis-alkyne adduct Ir(MeOzCCzCOzMe)z P20 on the pseudo-three-fold axis. Alternatively (alkyne carbons considered as individual ligands), it (PMezPh),+ as the BF4 salt. This molecule shows ‘H, 13C and “P NMR spectra consistent with the might be described as square pyramidal. Consistent following trigonal bipyramidal structure. The with both descriptions is the fact that the coormolecule is stereochemically rigid both with resdination sphere is fairly open (but shielded by two phenyl groups) tram to P20 (Fig. 2). There are no + Ir..* o&o-H distances short enough to suggest agostic donation; the shortest distance is 3.24 A. pE The Ir-P2 and Ir-Pll distances are indisE C02Me tinguishable from the Rh-P distances in E Rh(PMe,Ph),+,4 so it is the Ir-P20 distance which is unusually short. While the alkyne C3O-C3 1 disPE tance (1.306(8) A) is lengthened from that in the free alkyne (1.213( 1) A),5 such lengthening is not pect to phosphorus site exchanges and alkyne rotadramatically greater than in other coordinated tion. We have reported the analogous molecule alkynes. 6,7 Likewise, the H&-C%C angles here Ir(C2H4)2(PMezPh)3+,9 and have commented (138.9 and 145.7”) are quite unexceptional for coorthere on its rigidity. This analogy of Me02CC2 dinated alkynes. The Ir-C distances, however COzMe to ethylene shows clearly that Me02CC2 (2.01 A), are remarkably shorter than the single CO*Me can serve as a formal two-electron donor. bond Ir-C distances in fac-IrMe,(PMezPh)3 Finally, this same complex can be synthesized by (2.159 A).” Both the unusual coordination displacement of Me&Me from Ir(MeC,Me) geometry and the very short 1r-C and Ir-P20 (PMe2Ph)3+, thus demonstrating that the butyne distances require explanation. complex undergoes attack by nucleophiles.

p-+

I,++’

/ v-

=

1872

Table 2. Bond distances (A) and angles (“) for M(C0)4(F$X2CF,) (M = Ru, OS) Bond distances (A) M = OS Osl-C2 1.976(17) Osl-C4 1.974(15) Osl-C6 1.964(13) Osl-C8 1.964( 14) Osl-Cl4 2.149(12) Osl-Cl5 2.135(12) FlO-Cl3 1.346(17) Fll-Cl3 1.335(19) F12--C13 1.317(17) F17-Cl6 1.322(17) FlS-Cl6 1.317(17) F19-Cl6 1.312(19) 03-C2 1.113(20) 05-x4 1.125(19) 07-C6 1.130(17) 09-X8 1.117(17) C13-Cl4 1.475(18) C14-Cl5 1.276(19) C15-Cl6 1.460(19)

M = Ru 1.961(5) Rul-C2 1.967(5) Rul-C4 Rul-C6 1.967(5) 1.971(5) Rul-C8 2.128(4) Rul-Cl4 Rul-ClS 2.121(4) 1.323(5) FlO-Cl3 1.328(6) Fll-Cl3 F12-Cl3 1.333(5) 1.319(5) F17-Cl6 1.329(5) F18-Cl6 1.312(6) Fl9-Cl6 1.131(6) 03-X2 1.130(6) 05-C4 1.129(6) 07--X6 1.126(6) 09-C8 1.473(6) C13-Cl4 1.276(6) C14-Cl5 C15-Cl6 1.464(6)

Bond angles (“) C2-Rul-C4 C2-Rul-C6 C2-Rul-C8 C2-Rul-Cl4 C2-Rul-Cl5 C4-Rul-C6 C4-Rul-C8 CX-Rul-Cl4 C4-Rul-Cl5 C6-Rul-CS Cd-Rul-Cl4 C6-Rul-Cl5 C8-Rul-Cl4 C8-Rul-C15 Cl”Rul-Cl5 Rul-C2-03 Rul-C4-05 Rul-CX-07 Rul-C8-09 FlO-C13-Fll FlO--C13-F12 FlO-C13-Cl4 Fl l-C13-F12 Fll-C13-Cl4 F12-C13-Cl4 Rul-C14-Cl3 Rul-C14-Cl5 C13-Cl4-Cl5 Rul-C15-Cl4 Rul-C15--C16 C14-Cl5-Cl6 F17-Cl6-F18 F17-C16-F19 F17-Cl6-Cl5 F18-ClbF19 Fl&Cl6-C15 F19-Cl6-Cl5

93.66(22) 91.10(19) 98.68(18) 148.17(17) 113.22(17) 172.74(19) 93.18(19) 84.56(19) 85.39(17) 91.52(20) 88.51(17) 87.69(18) 113.15(17) 148.10(17) 34.96(16) 177.6(4) 175.9(4) 178.6(4) 177.6(4) 106.0(4) 106.0(4) 111.9(4) 107.5(4) 112.2(4) 112.8(4) 145.4(3) 72.21(26) 142.3(4) 72.83(25) 145.3(3) 141.8(4) 104.8(4) 105.6(5) 112.2(4) 107.2(4) 113.9(4) 112.5(4)

c2-0s1-C4 C2-OS l-C6 C2-Osl-C8 C2-Osl-Cl4 C2-Osl-Cl5 C4-Osl-C6 C4-Osl-C8 C4-Osl-C14 C4-Osl-Cl5 C6-Osl-C8 c6-Osl-Cl4 cx-Osl-Cl5 C8-Osl-C14 C8-Osl-Cl5 c14-Osl-Cl5 Osl-C2-03 0s1-C4-05 0s1-C6-07 Osl-C8-09 FlO-Cl3-Fll FlO-C13-F12 FlO-C13-Cl4 Fl l-C13-F12 Fll-C13-Cl4 F12-C13-Cl4 Osl-c14-Cl3 Osl-C14-Cl5 c13-C14-Cl5 0s1-C15-Cl4 Osl-C15-Cl6 C14-C15-Cl6 F17-C16-F18 F17-C16-F19 F17-Cl6-Cl5 F18-Cl&F19 F18-C16-Cl5 F19-ClbC15

92.5(8) 92.2(8) 95.2(7) 150.6(7) 116.0(7) 172.6(6) 93.0(6) 84.7(6) 84.8(6) 92.1(6) 88.4(5) 88.0(5) 114.2(5) 148.8(5) 34.7(5) 173.6(19) 175.8(14) 179.3(13) 175.8(14) 106.4(13) 105.8(12) 111.0(12) 108.2(13) 112.0(13) 113.1(12) 146.2( 11) 72.1(8) 141.7(13) 73.3(8) 146.2(10) 140.6(13) 105.3(12) 105.3(15) 112.5(12) 107.0(14) 114.1(12) 112.0(12)

An example of structural isomerism F 19

3 Fl2

Fig. 3. ORTEP drawing of Ru(CO)~(F,CC$F& showing atom labelling. The isomorphous osmium analogue is labelled identically.

Structure of M(C0)4(F3CC2CF3)

(M = Ru, OS)

The ruthenium and osmium complexes, M(C0)4(F3CC&F3), form isomorphous crystals from pentane at -78°C. The structure (Fig. 3) of the ruthenium compound has the greater precision, due to the large influence of absorption for osmium, but the molecular dimensions of the two are indistinguishable within a 3a criterion. The axial and equatorial M-C distances are equal, as are the M-C(alkyne) distances and the alkyne c-=C distances. The molecules have a faithful approach to CZVsymmetry. A distortion from the idealized angles of a trigonal bipyramid is the reduction of the equatorial/equatorial intercarbonyl angle (C2-M-CS) to 95.2(7)’ (OS) and 98.68( 18)” (Ru) and indicates that these molecules are perhaps better regarded as six-coordinate octahedral species

+

with the M(F3CC2CF3) fragments approaching the metallacyclopropene bonding extreme. This view is reinforced by the solution “rigidity” of these compounds at ambient temperature and above as evidenced by their i3C NMR spectra. lo The corresponding angle in the analogous 0s(C0)&Ie3SiC2 SiMe,) complex is larger (108.8(3)“) in accord with the weaker Os-Me,SiC,SiMe, interaction. This observation is further corroborated by the larger C-C-R angle of the coordinated alkyne in the latter (av. 156(2) vs 142(l)“) and longer M-C(alkyne) distances (av. 2.26(l) vs 2.13(2) A). Also noteworthy is the bending of the axial carbonyls towards the alkyne: C4M-C6 = 172.74(19)” for ruthenium and 172.6(6) for osmium, as previously observed in Os(CO), (Me3SiC2SiMe3).” Two points bear mention in comparison to the bond lengths in Ir(MeC2Me)(PMe2Ph),+. In the iridium complex, the C&C distance is longer by 0.030 8, (five ESDs) and the M-C(alkyne) distances are shorter by 0.12 A (20 ESDs). The alkyne is thus more tightly bound to the Ir(PMe,Ph),+ fragment that it is to M(C0)4 (M = OS, Ru). Analysis of the orbital interactions

For conventional ligands all 4d8 and 5d* ML4 complexes are square-planar. The introduction of an alkyne as a ligand induces a major change in the structure. More interestingly, the change of structure depends also on the nature of the metal. The known iridium(I) (2, 412), cobalt(I) (3)13 and iron(O) (5) ’ 4 alkyne complexes uniformly adopt a structure with approximate Cs symmetry (“CS”). They are distorted away from Td symmetry and resemble a

P \

R’(O)C\

p\

1873

,..!,r-,_,/ R

gi3J-/&R

P’

R

PPl,:,-/~R

<

P=P&Ph

P = PMe,

P = P(P-tdyl),

R-Me

R=Ph

R=CO&le

R P = P(OMe), R=Ph

R’ = CH,CH,

3

2

Ph

4 +

5

‘BU

Ph

‘Bu

Ph P = Pt&=&H5

6

R=aryl

1874

G. MARINELLI

square pyramid. The metal-to-apical ligand bond length is especially short. In all these complexes, the metal-alkyne distance is very short, and the C-C bond is notably stretched which indicates a strong metal-alkyne interaction. In contrast, the platinum(I1) *5(e.g. 6) alkyne complexes are all squareplanar (“SP”) with the axis of the alkyne perpendicular to the PtL3 plane. Why should isoelectronic compounds have different structures and different reactivity? These are the points we will address here. Our analysis is based on extended Hiickel calculations. Atomic and structural parameters are given in the Appendix. The results of the calculations of the SP and CS structures for model complexes MP,(acetylene) (M = Pt”, I?, CoI, Fe’, P = PHJ are given in Table 3. Only platinum prefers the square-planar structure ; all other metals favour the CS geometry, in good agreement with the experimental result. As a background for understanding the reason for the trends in Table 3, we must consider the energy difference between a SP and a CS structure with purely cr ligands. We start the analysis with a study of the simplest model of an ML4 8’ complex in which all ligands are H ~. In the CS structure the M-H bonds (1.7 A) were aligned with the M-P and M-acetylene midpoint vectors of the title iridium structure. Calculations on IrHd3- and PtH4*- show the square-planar structure to be strongly favoured by about 67 kcal mall ’ over the CS for both metals. This large preference is easily accounted for by the consideration of the d splitting, since the energy of the occupied d block faithfully reflects the total energies. In a square-planar structure, three nonbonding orbitals are below a weakly antibonding zz (z axis perpendicular to the molecular plane). Above this set is the highly destabilized x2-y2 which is pointing toward the four ligands. In a CS

Table 3. Relative total energies (kcal mol-‘) of M(PH,),(acetylene) for SP and CS structures. Bond lengths were taken from the stable isomer and were kept equal in the other isomer. For the CS structure the experimental angles were chosen. The angles at the metal for the CS structure for platinum(I1) were chosen identical to that of iridium(I) Metal

SP

cs

Pt” Ir’ Co’ Fe0

0.0 1.9 11.3 19.6

0.3 0.0 0.0 0.0

et al.

structure the d splitting is close to that of a T,, complex : ’ ’ two non-bonding orbitals are below a set of three destabilized ones which are not degenerate, due to the absence of a C3 axis. Assuming a closed shell electronic configuration, the squareplanar structure has six electrons in non-bonding orbitals and two in a weakly antibonding one. In contrast, housing eight electrons in the CS structure results in a significant destabilization for four of them. Thus four a ligands prefer a square-planar arrangement for any metal. The occurrence of another structure is then undoubtedly due to rceffects. Square-planar and CS alkyne complexes To understand the factors that favour each structure we will compare the complexes of the two atoms of the third period, platinum(I1) and iridium(1) in which variations of bond lengths are expected to be minimal. The interaction diagrams for the acetylene ligand in the square-planar (lefthand side) and CS (right-hand side) geometries are presented in Fig. 4. Since the diagrams for platinum and iridium look similar and differ by only small changes in energy, we have shown the interaction diagrams for iridium only. Comparison between the two metals relies on numerical evaluations and will be discussed in the text. Note that from now on the z axis is along the metal-to-acetylene midpoint direction. We will start the discussion with the squareplanar geometry. This structure has been considered numerous times. I7 However, in order to discuss the bonding in both SP and CS structures, it is necessary to describe the bonding in SP that is germane to our analysis. In a T shape ML3,17 three non-bonding (xz, yz, xy) orbitals are just below a weakly antibonding orbital which is essentially oriented perpendicular to the fragment plane ( y2 in our coordinated system). Higher above is an empty a hybrid which is mostly extended toward the empty site. Bending the alkyne lifts the degeneracy of the rc orbitals (al, r+ rr*,,,n*J. Due to the high symmetry (C,,) of the complex, the interaction diagram is very simple to build. The znllorbital interacts very weakly with y2 and strongly with the empty a hybrid. This last interaction is responsible for the metal-acetylene a bond. The empty acetylene rc*,,interacts with the occupied yz orbital. This interaction is responsible for the metal-ligand backbonding and is stabilizing. The actylene rcl has only the symmetry of the occupied xz orbital (7). This interaction between two occupied orbitals induces a four-electron destabilizing interaction. No bond results from the interaction and it actually contributes to a decrease of

An example of structural isomerism

1875

eV

Fig. 4. Interaction diagram for a square-planar (SP) structure (left-hand side) and for the experimental IPP, “t.-..~+..-^c,...T-DIU0 u\+

the total bonding between the two fragments. The higher empty orbital of acetylene, A*~, has the proper symmetry to stabilize the occupied metal xy orbital. However the interaction is very small since it is based on a 6 type overlap. It will not be considered further.

w7 The analysis for the CS structure begins with the orbitals of the CS MP3 fragment which resembles that of a ML3 piece of an octahedron with C,, symmetry. The symmetry labels that will be used are those of the C, symmetry point group. Three low-lying (la’, 2a’, ICI”), essentially non-bonding orbitals close in energy, are below a set of three hybrids which are made of a well-known mixture of d, s and p metal orbitals. l8 These three orbitals separate into a set of two (26, 3a’), which would have been degenerate in C3vsymmetry, and a highlying 4~‘. In the present geometry of the CS MP3 fragment, 2~” is below 3~’ and can be considered as the HOMO of the fragment. As will be evident later, the relative order of these two orbitals will play a central role in the rotational orientation of the

acetylene ligand. These three orbitals, which are M-P antibonding, are strongly hybridized away from these ligands and are thus perfectly suited for a strong interaction with the incoming acetylene ligand. In C, symmetry, the four acetylene orbitals are la’ (nnl), 2~’ (xl,), la” (x*,,) and 2~” (a*J. The low symmetry of the complex might have resulted in a complicated interaction diagram. However, the analysis remains relatively simple due to consideration of local symmetry. The nil (2~‘) orbital overlaps essentially with the occupied metal la’ and the empty 4~’ orbitals while it encounters the nodal plane of the filled 2~’ and empty 3~‘. The interaction between nI,and the metal la’ orbital is destabilizing but it is fully compensated by the strong stabilization caused by 4~‘. This last interaction is responsible for the metalacetylene Q bond. The empty z*,, orbital (la”) overlaps with the occupied metal la” and 2~” orbitals giving rise to the metal-acetylene backbonding (the later interaction is represented in 8). Both of these interactions are stabilizing but 2~” plays a preferential role since it is closer in energy to A*,,. We will see later the consequences of this privileged role. The occupied 7~~(la’) acetylene orbital is in fact well suited to overlap with the occupied metal 2~’ and the empty metal 3~’ orbitals, since all of these orbitals have a spatially close nodal plane. In

1876

G. MARINELLI

et al.

indeed indicate that the interaction energy involving xl1equally favours the square-planar geometry for both metals. In consequence, x,_ and R*,,are the only orbitals of the acetylene that need to be considered. Let us then first compare the interactions that give rise to rr systems in both geometries. Since we are mostly interested in the unusual iridium structure, numerical values in the following section will be given for iridium(I). The x1 orbital is only participating in four-electron destabilization in SP while in CS it is able to find an empty orbital (3a’, 9), into which to donate electrons (0.29 e- are transferred), thus relieving a part of the four-electron destabilization. This effect favours CS. Metal-to-acetylene backbonding occurs from a non-bonding d orbital in the SP geometry, whereas in the CS structure it utilizes a high-lying hybrid orbital (1.1. eV higher in energy than the non-bonding d) pointing toward the acetylene ligand (8). Therefore for both energy and overlap (0.20 in SP, 0.26 in CS) reasons, the backbonding is stronger in 9 8 CS. Larger electron-transfer (0.53 e- in SP, 0.76 The metal is thus bonded to the acetylene via in CS) into R*,, in the CS structure confirms this three bonds, one 0 and two n. This accounts for the analysis. short 1r-C distance (2.01 A) when compared to Both of these effects oppose the e interactions, other 1r-C bond lengths (e.g. 2.159 A in fac- which favours SP. The outcome of these comIrMe3(PMe,Ph),).8 The acetylene tends to give elec- petitive effects can only be established by numerical trons to the metal via both its xl1and a, orbitals. calculations. In agreement with experiments, the As nicely discussed in a recent review I9 this is usu- calculations show that the iridium complex is more ally described as a four-electron donation, although stable for a CS structure, while the platinum comno such complete electron-transfer is ever at work. plex prefers a SP geometry (Table 3). Why is that In the square-planar geometry, the acetylene cannot so? behave like a four-electron donor and is at most a The main difference between the two metals relevant to our study is that the d orbitals of iridium(I) two-electron donor via xl1since there is no empty metal orbital available to receive electrons from are at a higher energy and more diffuse than those x,_. In the CS structure, the four-electron donating of platinum(I1). As a consequence, the backbonding capability can be fully used since the metal has from 2a” into n*,, is larger for iridium than for the proper number of empty orbitals to receive the platinum due to a smaller energy gap and a better electrons. overlap.? The situation is difficult for the case of Now that we have built both interaction x1. The four-electron destabilization is exclusively diagrams, we need to discuss why some metals pre- controlled by overlap” and thus influences the fer the square-planar structure and others the CS diffuse iridium orbitals more than the contracted one. Recall that the G effect strongly favours the platinum orbitals. Consequently there is larger built square-planar geometry and that the preference for in destabilization in the SP structure for iridium CS can only be due to II bonding. For this reason, than for platinum. In the CS structure, the electronthe interaction of the acetylene xl1orbital with the transfer from xn, into 3a’ is similar for both metals metal fragment cannot make the difference between because the larger energy gap between the two the two structures since it participates only in the orbitals in the iridium case is compensated by the formation of the metal CJbonds. Calculations do larger overlap (0.14 for Pt, 0.16 for Ir). It appears then that, in order to overcome the large preference for the square-planar structure t Note that the interaction with la” follows the same resulting from the (r interactions, both of the above trend but we will consider only 2a” since, being higher in factors must be present for the CS structure to be energy and hybridized towards acetylene, it is responsible preferred. It requires first of all a large four-electron destabilization in the square-planar geometry and for the main interaction.

contrast, the metal la’ and 4a’ orbitals are essentially pointing toward the nodal lane of x1 and therefore do not interact with it. The interaction between x1 and metal 2a’ is destabilizing, while that with the empty 3a’ is stabilizing. This last interaction, represented in 9, produces electron-transfer from rcl into the empty metal orbital and forms a second type of rr bond between the acetylene and the metal. Therefore in the CS geometry, the fourelectron destabilization between A, and a metal d lone pair, is in part compensated by a two-electron stabilization. Finally, the last orbital to consider is 7c*,. In CS geometry, as in the square-planar structure, this orbital plays essentially no role since it is only involved in a 6 type overlap with the occupied 2a” orbital.

J@z-

1877

An example of structural isomerism

a large ability for back donation to stabilize CS. It is clear that the absence of a four-electron destabilization for olq?n complexes leads to a squareplanar geometry independent of the metal and the extent of back donation into the rr* orbital. Numerical calculations do indeed confirm this point. In summary, it is the four-electron donation of an alkyne ligand which is responsible for the “unusual ” CS structure of ML,(alkyne) complexes

and any effect (e.g. increasing metal oxidation state or lowering the principal quantum number) which acts to quantitatively diminish the overlap between n1 and the occupied metal d orbital will return the structure to the classical square-planar one. The decrease in principal quantum number from iridium to cobalt might alter the ground-state structure, but the numerical calculations show that the CoP,(acetylene)+ (3) remains CS. This is because the shorter metal-carbon distances have maintained sufficient overlap with n,. It follows that the isoelectronic iron(O) (5) complex is also CS. Alkyne rotation : JEuxionality

Ground-state

structure

and

markably Co[PhC,(n-pentyl)](PMe,),+ (11)21 has the alkyne eclipsing one Co-P bond (“vertical”). However, interconversion between the horizontal and vertical isomers is facile since these cobalt complexes were reported to be highly fluxional. Flwtionality is also present in the iridium complex described in this work. The 31P, ‘H and 13CNMR spectra of Ir(MeC,Me)(PMe,Ph),+ at 25°C all indicate that there is rapid intramolecular site exchange among inequivalent phosphorus ligands. The data exclude dissociation of phosphine or butyne as the mechanism for such fluxionality. The barrier for this process is small, since the 3‘P{ ‘H} NMR spectrum remains a singlet even at - 95°C. Pl

+

” Ph

HORIZONTAL

Ph

P3’ G-4, CO-PI = 2127A

CUP1 = 2166A

cc-~2(3) = 22091A

Cc-P2(3)-2.131,2.16.3A

Pl -Co-P2 = 96”

PI-CO-P2

= 103” a”.

P2cc-P3 = 104”

P2ceP3

= 96”

10

The structure observed for cobalt necessitates consideration of another structural parameter : rotation of the alkyne about its midpoint. Both Ir(MeC,Me)(PMe,Ph),+ and Co(PhC,Ph)(PMe,),+ (10)13 have the alkyne oriented perpendicular to the mirror plane of ML3 (“horizontal”). Re-

+

\ p2 ,,w... co-

11

First let us point out that the P-M-P angles are significantly different in the horizontal and vertical structure. Comparing the cobalt complexes, the Pl-Co-P2(3) angle is about 96” (90’ for Ir) and the P2-M-P3 angle is 104” (106” for Ir) in the

VERTICAL

Fig. 5. Interaction diagram limited to the most relevant MOs for the horizontal (left-hand side) and the vertical (right-hand side) complex. In this diagram the structure of ML, corresponds to that observed in the horizontal isomer.

G. MARINELLI

1878

horizontal structure and, respectively, 103 and 95” in the vertical structure. As we will see there is, in fact, a pair of P-M-P angles best adapted to each conformer. As a first step, we rotate the acetylene about its midpoint from the horizontal to the vertical orientation keeping the MP3 angles at the values of the horizontal structure. Figure 5 shows that, during this rotation, nI and 7c*,,have interchanged their partners on the metal. In the vertical conformation 3a’ and 2a” interact with n*,, and x1, respectively. The rotation is energetically unfavourable since the stabilization of 2a” or 3a’ by n*,, is equivalent and thus (3a”+&~*,,) is always higher in energy than (2a”+ ETC*,,). The increased stabilization of nI in the vertical structure is too small to compensate the above effect. The vertical conformation would thus be more stable if 3a’ were below 2a” in the ML3 fragment. The angles between the phosphines determine the relative energies of these two orbitals. Since the 2a” and 3a’ are composed to a large extent of d orbit&, the maximum of destabilization by the ligands occurs when the angle between the ligands is close to 90”. Thus a 90” angle between P2 and P3 destabilizes mostly 2a”, while a 90” angle between Pl and P2(3) destabilizes mostly 3a’ (we maintain a mirror plane through the Ir-Pl bond). In case the angles between the phosphines depart from this ideal 90” value, the higher destabilization is associated with the P-Ir-P angle that is closer to 90” as summarized in Fig. 6. Therefore the angular relaxation of the MP3 fragment is a determining factor for the stability of the vertical conformation. It appears from the calculations that the vertical conformation becomes more stable when the geometry of MP3 is relaxed to that observed in the experimental structure. The rotation between horizontal and vertical conformers also alters M-P bond lengths. In the horizontal conformation, the M-PI bond is significantly shorter than M-P2(3) while this is not

1

1

3a % '3V

-

28" 2w3

b

-

1 1 28"

2

+

,*.. d@ 3

3a

a

+k

* k '39

b

Fig. 6. Relative energy of the (2u”, 3~‘) pair for (a) > 90°] and PI-Ir-P2(3) = 90”, P2-k-P3 [PI-Ir-P2(3) > go”, P(2)-Ir-P3 = 90’1.

(b)

et al.

so in the vertical conformer as shown in 10and 11. The unequal electron occupation of 2a” and 3a’ in the acetylene complex in the two conformations is responsible for this fact. The 2a” orbital is equally antibonding with P2 and P3 while 3a’ is much more antibonding with Pl than with P2(3) as represented in Fig. 6. In the horizontal conformation the electrons which are involved in the acetylene-Ir n bonds go into (nl +&3a’) which is mostly centred on the acetylene part and in (2a”+e~*,,) which is mostly centred onto 2a”. Going to the vertical conformation, the electron occupation is reversed since the two MOs are now made up of (nl + e2a”), mostly on the acetylene and of (3a’+en*,,), mostly on the metal 3a’ orbital. Consequently turning the acetylene by 90” weakens mostly M-P1 and makes the three M-P bond lengths more equivalent. Although a better estimation of the structure of the MP3 fragment accompanying the rotation of the acetylene is beyond the validity of the EHT method, the trend in the bond angle and bond length relaxation is well reproduced. This conclusion provides an unusually detailed view of the occurrence and the energetic importance of ML3 “breathing” during acetylene rotation. Note however that the difference in energy between 2a” and 3a’ is small since the MP3 geometry is not far from CJu symmetry, where the two orbitals would be rigorously degenerate thus allowing free rotation of the acetylene. The consequence is a small activation energy for alkyne rotation in all situations, in full agreeement with observed rapid fluxionality of all molecules discussed here. Since there is a strong correlation between the angles within the MP3 substructure and the orientation of acetylene, what is the factor which will eventually determine the preferred structure? Since symmetrical alkynes prefer to be horizontal while the asymmetrical one prefers the vertical position, the influence of alkyne asymmetry was next investigated. The phenyl substituent destroys the symmetry of the n orbitals of alkyne and increases the coefficient of the carbon which is far from the phenyl ring in the frontier orbitals.22 Recall at this point that the most important interaction between MP, and acetylene is that of the back donation into x*,,. The interaction is thus optimal when the large lobe of 3a’ (i.e. opposite the M-PI bond) overlaps with the large lobe of n*,, as shown in 12. This is achieved in the vertical conformation when the phenyl ring is syn to Pl. Changes within the ML3 fragment also influence the preferred structure. Consider the horizontal structure of the IrP2(acyl)(alkyne) complex (4) where the acyl ligand replaces Pl. Since the acyl

1879

An example of structural isomerism

12 group is a stronger c donor than a phosphine, the 3a’ orbital is always higher than 2a” (the 2a”-3a’ energy gap in the fragment Ir(acyl)P,+ is considerably larger than in IrP3+). The lowering of 2a” with respect to 3a’ is even increased by the angles between the ligands (105” between the two phosphines and 92 and 95” between the acyl and the two phosphines, i.e. a structure similar to the IrP3+ horizontal complex). The horizontal conformation with the acyl group occupying the site of Pl is thus observed. Although steric bulk might also influence these two angles (in particular in opening the angles between the phosphines and contracting those between the acyl and the phosphine), in this case also there is a nice correlation between the intimate structure of the metal fragment and the orientation of the acetylene.

antibonding combination of 3a’ and xn, and is mostly centred on the 3a’ orbital. This interaction is of moderate strength so that the LUMO remains at low energy and is also strongly developed tram to Pl. Since the molecule approximates a squarepyramidal geometry the region tram to Pl is also the most sterically accessible. But why is the structure a square pyramid, that is why are P2 and P3 nearly in the Ir-acetylene plane? The dominant interaction between the metal fragment and the acetylene ligand is that between the high-lying occupied 2a” and x*,,, the one between the deeper la” and R*,,being much smaller due to the larger energy gap. In order to maximize the dominant interaction, 2a” should point towards R*,, as illustrated m 14. This is achieved by putting P2 and P3 in the Iracetylene plane. Attendant diminution of the interaction between 2a” and x*,, is of less importance. The LUMO in the square-pyramid type complex is at a particularly low energy since the overlap between 3a’ and x1 is smaller than in a more Td type structure.

14

Reactivity The Ir(PMe,Ph),(MeC,Me)+ complex reacts easily with another alkyne or with HZ. The related cobalt complex 3 adds MeCN to give Co(PMe,), (MeCN)(Ph&Ph) + . l3 All of these Lewis bases will seek out the LUMO of the metal complex, yet facile nucleophilic attack on an 18-electron complex might be viewed as unexpected. Therefore we now analyse the energy and spatial orientation of the LUMO of IrP,(acetylene)+ in the experimental (horizontal) conformation. The LUMO, 13 of the complex is made of the

13

7 It might be asked why this complex is not planar since MeOzCCzCOzMe is more adequately considered as a two-electron donor. However, the effect of x, is not offset totally and it is clear that it is not possible to determine in a reliable manner the limit between the SP and CS structures. The presence of iridium by itself substituted by strong 0 donating ligand may have made the d block sufficiently diffuse to create a significant four-electron destabilization.

Tuning thefour-electron destabilization by an alkyne substituent and by ligand effects The above discussion of da ML3(alkyne)q species calls into question the related family of compounds M(CO)Aalkyne) (M = Fe, Ru, OS), several of whose structures are reported here. Are these 20valence electron species? In particular, what are the factors which control the thermodynamics of eq. (2)? There is a clearly defined preference for L = ML,(alkyne) 1

ML,(alkyne) + L

(2)

phosphine to favour the right-hand side of eq. (2), while L = CO favours the left-hand side. What electronic factors are at work? In addition, how is it possible to diminish the four-electron repulsion expected in compounds such as 0s(C0)4(alkyne). The introduction of acceptor ligands such as COzMe on the alkyne in the plane of the bent alkyne delocalizes the rrl orbital. Its overlap with metal orbitals in both the SP and CS structures is thereby significantly diminished. Thus the four-electron destabilization is considerably decreased (it is mostly controlled by the overlap) and it could behave as a two-electron donor.? This explains the otherwise curious production of Ir(Me0&C2C02 Me)z(PMezPh)3+, a bis-alkyne complex, under

G. MARINELLI

1880

conditions which produce monoalkyne complexes for phenyl and alkyl substituted alkynes. We can now see that the undetected intermediate Ir(Me02CC~COzMe)(PMezPh)s+, of CS structure, has a very low-lying LUMO since 3a’ is not significantly destabilized by x1. It thus easily adds an additional alkyne ligand. Interestingly the neutral mono dicarbomethoxy complex of IrP2(acyl) complex whose structure has been discussed above is stable. ‘* Its lack of reactivity relative to the pr&ent cationic IrP, fragment may be due to the strong CJdonating effect of the acyl that takes the place of Pl. The corresponding 3a’ and consequently the LUMO of the monoalkyne complex is higher, which diminishes its affinity for an additional ligand. Alternatively one can diminish the four-electron destabilization by delocalizing the “offending” metal orbital. This can be achieved by introducing CO ligands on the metal. The xz lone pair is delocalized into the four x*co orbitals and thus its overlap with the alkyne n1 orbital is only weakly destablizing. However, the molecule shows reactivity characteristics which can be attributed to the residual four-electron destabilization. While the M(CO),(alkyne) species (M = Ru, OS ; alkyne = HC 2 H 923Me,SiC SiMe I’ F3CC2CF3”) have been characterized by *the fell range of modern techniques, it is important to recall that, with the exception of the F3CC2CF3 derivatives, the compounds are thermally unstable and indeed more so than the related M(C0)4L complexes (L being the classical two-electron donor CO or H2C5CH21 1*23). They thus show evidence of the four-electron repulsion predicted by simple electron counting rules. The observation that Ru(CO),(Me3SiC2SiMe3) readily deposits Ru3(CO), 2 and that the decomposition is slowed down by added free Me3SiC2SiMe3 ’ ’ reveals that alkyne dissociation is one way in which the instability is manifested. In addition, however, the unit M(CO)3(alkyne) represented an intermediate with special stability due to the availability of the four-electron donor alkyne unit. This can account for the rapid incorporation of 13C0 into M(CO),(alkyne) (alkyne = HC2H, F3CC2CF3)10~23and, at least in part for the unusual facile condensation reactions of these compounds with other l&electron species.L0,23 Equilibrium (2) can thus be easily displaced towards one or the other side by a fine tuning of the four-electron destabilization. CONCLUSION It is well accepted that isomeric forms of certain ligands can furnish different numbers of electrons

et al.

to the metal. From this work it is now clear that

the mirror-image situation can be observed on the metal: isomeric fragments can accept a different number of electrons. Thus a T shape ML3 d8 fragment is a two-electron acceptor while a pyramidal ML3 d8 fragment is a four-electron acceptor (this is why butadiene and other four-electron conjugated dienes coordinate to a pyramidal ds ML3 fragment). An encounter between a ligand and a metal fragment may thus induce structural changes in both partners to optimize the bonding/accepting capability of each. A search for other examples of adaptation of ML, fragment geometry to the active partner ligand should be fruitful. Acknowledgemen&--This work was supported by the U.S. NSF, the French CNRS, and the European Economic Community. The Laboratoire de Chimie ThCorique is associated with the CNRS (URA 506) and is a member of ICMO and IPCM. The Laboratoire de Chimie de Coordination is associated with Universid Paul Sabatier. We thank Scott Horn for skilled technical assistance.

REFERENCES 1. K. R. Birdwhistell, T. L. Tonker and J. L. Templeton, J. Am. Chem. Sot. 1987,109, 1401. 2. Preliminary results have been reported : G. Marinelli, I. El-Idrissi Rachidi, W. E. Streib, 0. Eisenstein and K. G. Caulton, J. Am. Chem. Sot. 1989,111,2347. 3. J. C. Huffman, L. N. Lewis and K. G. Caulton, Znorg. Chem. 1980,19,2755. 4. E. G. Lundquist, W. E. Streib and K. G. Caulton, Znorg. Chim. Actu 1989, 159, 23. 5. M. Tanimoto, K. Kichitsu and Y. Morino, Bull. Chem. Sot. Jpn 1969,42,2519. 6. S. D. Ittel and J. A. Ibers, Adv. Organomet. Chem. 1976, 14, 33. 7. S. Otsuka and A. Nakamura, Adv. Organomet. Chem. 1976,14,245. 8. E. G. Lundquist, K. Folting, J. C. Huffman and K. G. Caulton, Polyhedron 1988,7, 2171. 9. E. G. Lundquist, K. Folting, W. E. Streib, J. C. Huffman, I. Eisenstein and K. G. Caulton, J. Am. Chem. Sot., in press. 10. M. R. Gagne and J. Takats, Organometallics 1988, 7, 561. 11. R. G. Ball, M. R. Burke and J. Takats, Organometallics 1987, 6, 1918. 12. B. J. Rappoli, M. R. Churchill, T. S. Janik, W. M. Rees and J. D. Atwood, J. Am. Chem. Sot. 1987, 109, 5145. We cannot agree with the conclusions of these authors that the alkyne is a two-electron donor. 13. B. Capelle, M. Dartiguenave, Y. Dartiguenave and A. L. Beauchamp, J. Am. Chem. Sot. 1983, 105, 4662. 14. T. V. Harris, J. W. Rathke and E. L. Muetterties, J. Am. Chem. Sot. 1978,100,6966. 15. (a) R. Uson, J. Fornies, M. Tomas, B. Menjon and

An example of structural isomerism

16.

17.

18.

19.

1881

20. L. Salem, J. Am. Chem. Sot. 1968,90,543. 21. A. Bouayad, M. Dartiguenave, M.-J. Menu, Y. Dartinguenave, F. Belanger-Garitpy and A. L. Beauchamp, Organometallics 1989, 8, 629. 22. 0. Eisenstein, J.-M. Lefour, N. T. Anh and R. F. Hudson, Tetrahedron 1977,33, 523. 23. M. J. Bum, G.-Y. Kiel, F. Seils, J. Takats and J. Washington, J. Am. Chem. Sot. 1989,111,6850. 24. J. H. Ammeter, H.-B. Btirgi, J. C. Thibeault and R. Hoffmann, J. Am. Chem. Sot. 1978,100,3686.

A. J. Welch, J. Organomet. Chem. 1986,304, C24; (b) B. W. Davies and N. C. Payne, Can. J. Chem. 1973,51,3477; (c) B. W. Davies and N. C. Payne, J. Organomet. Chem. 1975,102,245; (d) G. R. Davies, W. H. Hewertson, R. H. B. Maise, P. G. Owston and C. G. Patel, J. Chem. Sot. 1970, 1873 ; (e) A. L. Beauchamp, F. R. Rochon and T. Theophanides, Can. J. Chem. 1973,51, 126. T. A. Albright, J. K. Burdett and M.-H. Whangbo, Orbital Interactions in Chemistry, p. 305. John Wiley, New York (1985). T. A. Albright, R. Hoffmann, J. C. Thibeault and D. L. Thorn, J. Am. Chem. Sot. 1979, 101, 3801; 0. Eisenstein and R. Hoffmann, J. Am. Chem. Sot. 1981,103,4308. M. Elian, M. M. Chen, D. M. P. Mingos and R. Hoffmann, Znorg. Chem. 1976, 15, 1148 ; T. A. Albright, P. Hofmann and R. Hoffmann, J. Am. Chem. Sot. 1977,99,7546. J. L. Templeton, Ado. Organomet. Chem. 1989, 29, 1.

APPENDIX The extended Htickel calculations were carried out by using the weighted Hii formula. 24The atomic parameters are given in Table 4. The geometries of the calculated complexes were adapted from the experimental structures.

Table 4. Atomic parameters. A double expansion (coefficients C) is used for the d orbitals Atom

Orbital

Hii (eV)

[,

r;,

C,

C,

Ir

6s 6~ 5d

- 8.600 - 4.900 - 12.17

2.500 2.200 5.796

2.557

0.63506

0.5556

4s 4P 3d

-9.210 - 5.290 - 13/18

2.000 2.000 5.550

2.1

0.56786

0.60586

6s 6~ 5d

- 9.077 - 5.475 - 12.59

2.554 2.554 6.013

2.696

0.63338

0.55128

4s 4P 3d

-9.100 -5.320 - 12.60

1.900 1.900 5.350

2.000

0.55052

0.6260

3s 3P

- 18.60 - 14.00

1.600 1.600

co

Pt

Fe

P