PAPER www.rsc.org/dalton | Dalton Transactions

Stable Lewis acid chelate of a bis(imido) tungsten compound and implicationsfor a-olefin dimerisation catalysis: a DFT study†

Sven Tobisch*

Received 30th October 2007, Accepted 28th January 2008First published as an Advance Article on the web 28th February 2008DOI: 10.1039/b716672a

Dimerisation of a-olefins by a catalyst system comprising of a bis(imido) tungsten compound and aLewis acid component has been investigated by density functional theory employing a catalyst modelthat closely mimics the real system together with ethylene chosen as a prototypical substrate. This studydisclosed that the Lewis acid preferably forms four-membered chelates by bridging acrosstungsten–imido linkages. Complexation at one imido group is predicted as the favourable, stable modeof Lewis acid association onto a bis(imido) tungsten(VI) compound. The computational analysisrevealed that the cocatalyst affects the energetics of the several steps to a different extent. Formation ofthe five-membered tungstanacycle and the further growth is moderately influenced, this is in contrast toits profound influence on metallacycle degradation. AlClMe2 chelation of both tungsten–imido groupsis seen as being most effective in smoothing the energetics of the dimer-generating route. The predictedenergy profile of the most accessible pathway, however, is at odds with observed catalytic abilities. Thisstudy indicates that the removal of one imido ligand by the Lewis acid is a viable process, therebysuggesting that mono(imido) tungsten compounds are effective in dimerisation catalysis of the reportedcatalyst system.

Introduction

The dimerisation of a-olefins by homogeneous catalysts representsan attractive, direct route towards specific higher olefins.1 Severalcatalyst systems have been reported for dimerising a-olefins, mostof which yield branched dimers.2 Certain iron,3,4 cobalt3–5 andnickel6 catalysts have been described exhibiting abilities for lineardimerisation. The traditional migratory insertion/b-eliminationmechanism is generally believed to be operative for late transitionmetal systems, with linear dimer formation dictated by opposite(1,2) and (2,1) regiochemistry of first and second insertion.2 Onthe other hand, metallacycle intermediates have been identifiedas participating in chromium-mediated ethylene oligomerisationand polymerisation.7 Significant research efforts in both academiaand industry has led to the development of several chromiumsystems capable of the selective trimerisation of ethylene to 1-hexene8 or alternatively selectivity towards the tetramer, 1-octene.9

The application of group 6 metals in olefin dimerisation, however,is rather limited.2,10

Tungsten imido compounds are a notable exception. Whenexposed to Lewis acids of AlClnRm type, these compounds arecapable of dimerising a-olefins.11 As an example, treatment of WCl6

with two equivalents of aniline and the required amount of basegives initially rise to [W(NPh)2Cl2],12 which then leads presumablyin situ to the [W(NPh)2R2] bis(imido) tungsten compound 1 inthe presence of the cocatalyst. This catalyst system is reportedto be highly selective to dimerisation and moderately active.11 Incontrast to the more usual huge excess of activators for various

Sasol Technology (UK) Ltd, Purdie Building, North Haugh, St Andrews,Fife, UK KY16 9ST† Electronic supplementary information (ESI) available: Scheme S1 andcomputational details. See DOI: 10.1039/b716672a

other oligomerisation catalyst systems,2 here a specific amount ofLewis acid (∼15 molar equivalents) is required. Thus, the cocat-alyst is likely to have a distinct function in dimerisation catalysisthat extends its presumed function in precatalyst activation.

In order to unravel the role of the Lewis acid in the productivedimerisation cycle we decided to undertake a computational studyaimed at, firstly, probing the aptitude of the Lewis acid to establishspecific interactions with bis(imido) tungsten intermediates andsecondly, assessing what influence the associated cocatalyst exertson individual steps of the plausible mechanism shown in Scheme 1.The productive cycle starts after transformation of 1 into the[W(NPh)2(olefin)2] active catalyst species 2 with oxidative additionof the two olefin moieties. Concomitant ring closure affordsthe tungstana(VI)cyclopentane intermediate 3. Degradation ofthe metallacycle gives rise to the dimer product. The stepwisemechanism is most accessible for the conformationally rigidfive-membered metallacycle,13 comprising consecutive b-H ab-straction and reductive CH elimination with an intervening W-H–alkenyl intermediate 4. Repeated olefin insertion into suc-cessively enlarged metallacycle fragments (to first give rise tothe tungstana(VI)cycloheptane intermediate 5) and subsequentdegradation would lead to larger oligomers. Furthermore, severalconceivable paths for 1 → 2 transformation have been probedcomputationally.

Computational model and method

Model

Herein is reported the computational survey of the productivecycle for a-olefin dimerisation by bis(imido) tungsten compounds,together with several possible steps to be involved during 1 →2 conversion. The catalyst model was chosen in such a fashion

2120 | Dalton Trans., 2008, 2120–2127 This journal is © The Royal Society of Chemistry 2008

Publ

ishe

d on

28

Febr

uary

200

8. D

ownl

oade

d by

Lom

onos

ov M

osco

w S

tate

Uni

vers

ity o

n 22

/05/

2013

21:

52:1

8.

View Article Online / Journal Homepage / Table of Contents for this issue

Scheme 1 Catalytic cycle for a-olefin dimerisation by bis(imido) tungstencompounds, exemplified for ethylene.

to closely mimic the real catalyst system.11 This study employed[W(NC6H5)2(C2H4)2] and AlClMe2 components, together withethylene as a prototypical olefin substrate. We have performedan extensive DFT simulation of configurationally and conforma-tionally different reaction pathways involving up to two AlClMe2

moieties, the most favourable of which are reported herein. Forall the studied intermediates, Lewis acid association/dissociationand also the conversion between different association modes isindicated as being facile, to not be linked with a substantialenthalpy barrier. This has been examined in a linear-transitapproach, as the exact localisation of the corresponding TSstructures is complicated by the flatness of the PES in these regions.

Key species of the dimerisation reaction course were denotedconsistently with Arabic numbers as shown in Scheme 1, whilstScheme 4 contains the notation of species involved in 1 → 2conversion. We restrict the presentation of located key stationarypoints to unbound bis(imido) tungsten species and type II Lewisacid adducts (see below), as representative examples.

Method

All DFT calculations were performed with the program packageTURBOMOLE14 using the TPSS density functional15 in conjunc-tion with flexible basis sets of triple-f quality. For W we used theStuttgart–Dresden quasirelativistic effective core potential (SDD)with the associate (8s7p6d)/[6s5p3d] valence basis set.16 All otherelements were represented by Ahlrichs′ valence triple-f TZVPbasis set17 with polarisation functions on all atoms. The goodto excellent performance of the TPSS functional for a wide rangeof applications has been demonstrated previously.18

All stationary points were located by utilising analyti-cal/numerical gradients/Hessians according to standard algo-

rithms without imposing any symmetry constraints and wereidentified exactly by the curvature of the potential-energy surfaceat these points corresponding to the eigenvalues of the Hessian.The reaction and activation free energies (DG, DG‡ at 298 Kand 1 atm) were evaluated according to standard textbookprocedures19 using computed harmonic frequencies. The influenceof the solvent (chlorobenzene)11 was taken into explicit consider-ation by making use of a continuum model. All the key stationarypoints were fully located with inclusion of solvation by theCOSMO method.20 In order to account for real condensed-phasereaction conditions, the entropy costs for Lewis acid and olefinsubstrate association/dissociation processes was approximated asbeing two-thirds of its gas-phase value. This approximation wassuccessfully applied in previous investigations on group 4 metal-assisted selective olefin oligomerisation,13b,21 and is considered toreasonably estimate the true entropy contribution in condensedphase. Full details of the computational methodology employedare given in the ESI†.

Results and discussion

The first part of this study assesses what influence the Lewis acidexerts on the thermodynamics and kinetics of relevant steps of thedimerisation course. Several possible paths for 1 → 2 precatalystactivation will be explored afterwards.

Aptitude for Lewis acid association

As a first step towards a deeper understanding of the catalystsystem we start with probing the aptitude of the Lewis acid forcomplexation onto the crucial tungstanacycle intermediate 3. Upto two AlClMe2 moieties have been considered for saturation of theimmediate proximity of the tungsten centre. Several conceivableassociation modes have been tested. The located most stable forms(Fig. 1) are characterised by a four-membered chelating interactionof the Lewis acid with the tungsten–imido bond. This type ofLewis acid chelate is rarely found in the literature.22,23 Possible,alternative modes of AlClMe2 association onto 3 through one ortwo dative W–Me interactions are found to be less stable by at least7.5 kcal mol−1. Mountford and co-workers have recently reportedthe X-ray structure of an AlMe3 adduct of a titanium imido methylcation featuring two bridged methyl groups, together with DFTcalculations indicating that bridging the titanium–imido bond iselectronically favourable.23

Fig. 1 Most stable form of various modes of Lewis acid associationonto 3. Bond distances (pm), Wiberg bond orders (italic type) and relativestability (DG in kcal mol−1) are given.

The structural, electronic and energetic features summarised inFig. 1 provide a closer insight into Lewis acid adduct formation.The dianionic imido ligand preferably acts as a net six-electron

This journal is © The Royal Society of Chemistry 2008 Dalton Trans., 2008, 2120–2127 | 2121

Publ

ishe

d on

28

Febr

uary

200

8. D

ownl

oade

d by

Lom

onos

ov M

osco

w S

tate

Uni

vers

ity o

n 22

/05/

2013

21:

52:1

8.

View Article Online

Table 1 Calculated intrinsic free-energy profile (DG‡/DG in kcal mol−1) for relevant steps

Lewis acidassociation modea

2 → 3b

(DG‡/DG)b-H abstr.c

(DG‡)4c

(DG)RE-CHcd

(DG‡/DG)3 → 5c

(DG‡/DG)

0 7.5/−1.6 37.4 11.5 72.4/— 23.2/−9.6I 6.3/1.1 34.3 9.3 57.1/— 22.6/−9.5II 10.9/−1.9 24.4 15.7 36.5/−7.4 33.4/−1.4

a See Fig. 1. b Energies relative to 2. c Energies relative to 3. d The heat of RE-CH calculated for unbound species and type I adducts is unrealistic owingto the highly unsaturated nature of the [W(NPh)2(1-butene)] product intermediate and is thus omitted.

donor to the metal centre, formally adopting a r2p4 configuration,with the metal=N–R bond angle is near-linear.24 The nitrogen lonepairs of the two tetrahedrally disposed imido groups in 3 share thesame vacant metal orbital, such that each imido ligand can onlypartly donate its lone pair to the electron-deficient tungsten centre.This is manifested in two distinctly bent imido moieties ( W=N–R = 161 ◦) having equally elongated W=NR bonds (3-0, Fig. 1).As some proportion of the lone pair’s density remains localised oneach nitrogen centre, the arylimido–tungsten linkage can easily beattacked by the Lewis acid. Species 3-I illustrates the susceptibilitytowards AlClMe2 association. Bridging across the tungsten–imidobond diminishes its bond strength as indicated by its substantiallengthening and the smaller bond order. This is compensated forby the strengthened W=NR bond of the unbound imido moiety,which now can act as a net six-electron donor ( W=N–R =179 ◦), together with the newly formed N→Al and Cl→W dativebonds. As a net result, AlClMe2 forms a stable chelate with oneof the imido groups, which is driven by a thermodynamic force of6.0 kcal mol−1. Complexation at both imido groups (3-II), however,is found to be less favourable, since the new dative bonds are notstrong enough to balance the W=NR bond weakening. The DFTcalculations predict a stable chelate for the Lewis acid to associateat one imido group through N→Al and Cl→W dative bonds(Fig. 2). This favourable mode of Lewis acid adduct formationwith a bis(imido) tungsten compound can thus be anticipated asbeing isolable by X-ray diffraction.

Fig. 2 Structural parameters (A, ◦) of the optimised adduct 3-I.

Influence of the Lewis acid on individual elementary steps

Having analysed the aptitude of Lewis acid complexation froma thermodynamic point of view, we now investigate how thisaffects the kinetic profile of individual steps shown in Scheme 1.25

The intrinsic energy profile for formation of 3 and its stepwisedegradation or alternative growth is collected in Table 1 forunbound bis(imido) tungsten species (Table 1, entry 1) andAlClMe2 adducts of type I, II (cf. Fig. 1). The key stationarypoints associated with these steps are displayed in Fig. 3–5 forunbound (top) and type II adduct (bottom) species, respectively.

Unbound bis(imido) tungsten species. It is perhaps informativeto first inspect the situation where unbound bis(imido) tungstenspecies participate in the dimerisation course, hence without beinginfluenced by the Lewis acid.

The oxidative addition under C–C bond formation of the twocoplanar olefin moieties commencing from 2-0 passes through atransition-state (TS) structure that emerges at a distance of 2.00 Afor the new C–C bond. This step is seen to be remarkably facile,having a barrier of 7.5 kcal mol−1, and slightly exergonic, henceproceeding in a reversible fashion.

The first b-H abstraction along the stepwise path for degrada-tion of 3-0 does proceed through a TS structure that exhibits analready almost fully established hydrido–tungsten bond togetherwith an emerging olefinic subunit (Fig. 4, top). The decay ofTSbH-ab leads to 4-0, where the alkenyl moiety is g1-coordinatedto the metal by its alkenyl terminus. Subsequent reductive dimerrelease evolves through a TS structure (Fig. 4, top) that occurs ata distance of 1.40 A for CH bond formation. The dimer productmay coordinate back to the metal centre by its double bond atlater process stages, being displaced afterwards by incoming olefinsubstrate to recreate 2, or as an alternative leaving directly themetal’s coordination sphere to furnish 2 after substrate uptake. Incontrast to 2 → 3 olefin coupling, distinctly less favourable kineticsis predicted for degradation of 3-0. The reductive elimination hasthe higher barrier and hence dictating the overall degradationkinetics. Although the presence of two strong donor imido groupscan be predicted to make the RE step kinetically expensive, thecalculated barrier of 72.4 kcal mol−1 is unexpectedly huge (Table 1).

Of the two consecutive steps involved in the enlargement of themetallacycle fragment, the first olefin uptake onto 3-0 has a verysmall, if present at all, enthalpy barrier.26 Hence, the second olefininsertion into the metallacycle fragment determines the overallenergetics of the 3 + C2H4 → 5 process. The associated TS structurecomprises a quasi-planar four-membered cis arrangement of themetallacycle’s WVI–C bond and the coplanar oriented ethylene,with the newly formed C–C r-bond has a distance of 2.27 A.Ethylene insertion into the metallacycle of 3-0 is feasible (DG‡ =23.2 kcal mol−1, Table 1) and leads to the next larger metallacyclefragment 5-0 in an exergonic process that is driven by a thermo-dynamic force of 9.6 kcal mol−1.

2122 | Dalton Trans., 2008, 2120–2127 This journal is © The Royal Society of Chemistry 2008

Publ

ishe

d on

28

Febr

uary

200

8. D

ownl

oade

d by

Lom

onos

ov M

osco

w S

tate

Uni

vers

ity o

n 22

/05/

2013

21:

52:1

8.

View Article Online

Fig. 3 Selected degrees of freedom (A) of key stationary points for 2 → 3 oxidative ethylene coupling to afford tungstana(VI)cyclopentane intermediates3-0 (top) and 3-II (bottom).

Fig. 4 Selected degrees of freedom (A) of key stationary points for stepwise degradation of tungstanacycle intermediates 3-0 (top) and 3-II (bottom).

The following conclusions can be drawn regarding rival pathsthat commences from 3-0. The insuperable RE-CH barrier of72.4 kcal mol−1 does not coincide with the observed activityfor dimerisation.11 On the other hand, ethylene insertion hasmoderate kinetics, hence clearly indicating the strong preferenceof 3-0 to grow rather than to degrade. As a consequence thedimer-generating route is inaccessible and remains entirely closed

throughout. All these aspects clearly militate against unboundbis(imido) tungsten species as being responsible for the observeddimerisation catalysis.

AlClMe2-complexed bis(imido) tungsten species. We are nowgoing to investigate how the Lewis acid affects individual steps, by

This journal is © The Royal Society of Chemistry 2008 Dalton Trans., 2008, 2120–2127 | 2123

Publ

ishe

d on

28

Febr

uary

200

8. D

ownl

oade

d by

Lom

onos

ov M

osco

w S

tate

Uni

vers

ity o

n 22

/05/

2013

21:

52:1

8.

View Article Online

Fig. 5 Selected degrees of freedom (A) of key stationary points for 3 + C2H4 → 5 growth of tungstanacycle intermediates 3-0 (top) and 3-II (bottom).

limiting the discussion of structural aspects to type II Lewis acidadducts, as representative examples.

Following the minimum-energy pathway for oxidative couplingto start from 2-II some re-orientation of the Lewis acid chelatesoccurs. One of the chelates is partially broken whilst passingthrough the TS, but becomes re-established in 3-II (Fig. 3, bottom).It should be stressed here again that all processes related toreorganisation of AlClMe2 association are seemingly facile (seeComputational model and method section). As far as energeticsare concerned, the oxidative coupling appears to be the step thatis least influenced by the Lewis acid. Irrespective of the AlClMe2

association mode, metallacycle formation is remarkably facile andreversible, thus unlikely to affect the catalytic behaviour.

This contrasts with metallacycle degradation, where the cocata-lyst exerts a profound influence. The presence of two strong donorimido ligands in 3-0 renders the b-H abstraction and the reductiveelimination, in particular, rather demanding kinetically. The facileb-H abstraction necessitates an empty tungsten orbital as far aselectronic grounds are concerned. Such an orbital, however, isunavailable in 3-0, thereby giving rise to a significant barrier of37.4 kcal mol−1 (Table 1, entry 1). Lewis acid complexation at oneof the imido groups diminishes its donor ability, which is howeverlargely counteracted by the unbound imido moiety (see above) andas such causes a rather small drop of the barrier (Table 1, entry 2).AlClMe2 bridging across both tungsten–imido linkages is seen tobe most effective to facilitate the b-H abstraction. This lowers theactivation energy by 13.0 kcal mol−1 when compared to unbound3-0 (Table 1, entry 3). Interestingly, none of the chelates needsto be cleaved in order to accommodate the transferred H atom.As revealed from Fig. 4, the most accessible pathway for stepwise

degradation of 3-II sees some facile re-orientation of the bridgingAlClMe2 moieties, but with the two chelates remaining intactthroughout the process. Reduction of the imido’s donor capacityupon AlClMe2 complexation, which enhances the electrophilicityof the tungsten centre, affects the kinetics of the RE step in an evenmore pronounced fashion. Starting from an insuperable barrier inthe absence of the Lewis acid (Table 1, entry 1), a regular, steadydecrease of the activation energy is predicted upon increasing thenumber of associated AlClMe2 moieties. The stabilising role ofadditionally coordinated olefin molecules instead of AlClMe2 hasbeen explicitly probed, but found to be less effective than the Lewisacid.

Ethylene insertion into the metallacycle fragment of 3-II goesalong with the rupture of the Cl→W bond in one of the chelates,whilst traversing through the TS, but is rebuilt in the seven-membered tungstanacycle intermediate 5. AlClMe2 associationat both imido groups causes an increase in activation energyby 10.2 kcal mol−1 and a reduction in reaction heat (relative to3-0 + C2H4 → 5-0). This behaviour does not solely originatefrom electronic factors, but must also be attributed to some stericinteractions associated with olefin incorporation at the congestedtungsten centre (Fig. 5, bottom). In contrast, the energy profilefor 3-0/3-I + C2H4 → 5-0/5-I are almost degenerate; again owingto the regulating influence of the uncomplexed imido group in 3-I(see above).

Energy profile for the dimer-generating route

Having assessed thus far what influence the Lewis acid does exertonto individual steps of Scheme 1, we are now going to make a link

2124 | Dalton Trans., 2008, 2120–2127 This journal is © The Royal Society of Chemistry 2008

Publ

ishe

d on

28

Febr

uary

200

8. D

ownl

oade

d by

Lom

onos

ov M

osco

w S

tate

Uni

vers

ity o

n 22

/05/

2013

21:

52:1

8.

View Article Online

to the observed dimerisation abilities by analysing the absolutekinetics for the reaction route that leads to the dimer product.This is shown in Scheme 2 for unbound bis(imido) tungstenspecies and type I and II adducts. Note that 3-0, 3-I, 3-II are inrapid equilibrium, due to facile AlClMe2 association/dissociation(see Computational model and method section). As revealedfrom Scheme 2 the most accessible pathway for dimer formationstarts from 3-II. All the other pathways are precluded owingto computed kinetics that is significantly less favourable. Thereductive elimination is the kinetically more expensive of the twoconsecutive steps having a total barrier of 48.1 kcal mol−1 (relativeto 3-I). This is in obvious conflict with the observed moderateactivity.11 Moreover, the 3 + C2H4 → 5 growth of the metallacyclehas a barrier of 22.6 kcal mol−1 along the favourable pathway forethylene insertion into 3-I (Table 1). In light of the predicted strongpreference of the five-membered metallacycle to grow furtherthan to undergo degradation (DG‡ = 25.5 kcal mol−1) it mustbe concluded that the dimer-generating route cannot be accessed.

Scheme 2 Gibbs free-energy profile for the dimer-generating route. Solidlines indicate the most accessible pathway.

The apparent contradiction between observed catalytic abilitiesand the computationally assessed energy profile raises questionswhether bis(imido) tungsten species are actually responsible fordimerisation. A major factor that causes the unfavourable kineticsof metallacycle degradation is the presence of two strong imidodonor groups. Although the imido donor ability is significantlyreduced upon Lewis acid association, there is still a substantialoverall free energy of activation connected with the most accessiblepathway. As shown before, the intermediate character of thetungsten–imido bond in bis(imido)–WVI compounds (cf. species3-0) makes them susceptible to Lewis acid complexation to formstable adducts, such as 3-I. On the other hand, the Lewis acidmay also act to remove an imido group. Electrophilic attack ofbis(imido)–MoVI compounds by an alkyl halide has been reportedto easily abstract one of the imido groups, whilst the secondimido group remains unaffected.27 The viability of the consecutiveimido group removal from [WVI(=NPh)2Cl2] (0a) by the Lewisacid has been probed exemplarily for the gross reactions shown

in Scheme 3. Displacement of the first imido ligand is onlyslightly uphill by 3.1 kcal mol−1 (DG for reaction 1), whilst,not unexpected, a strong thermodynamic penalty of 31.7 kcalmol−1 (DG for reaction 2) is against the removal of the secondimido group from the mono(imido) tungsten intermediate 0b. Thepossible assistance of the ligand exchange by excess Lewis acidmoieties has been tested by evaluating the aptitude of AlClMe2

association onto intermediates 0a, 0b, 0c. Adducts of 0b and 0cbearing one or two chelates across WVI–imido, WVI–Me or WVI–Me linkages, respectively, are found to be less stable than theseparated fragments, thereby indicating that the removal of thesecond imido ligand is not facilitated by excess Lewis acid. Incontrast, AlClMe2 association across one of the tungsten–imidobonds in 0a is favourable by 2.4 kcal mol−1, but a slightly morestable adduct (DG = −3.0 kcal mol−1) has two chelates acrosstungsten–imido and tungsten–chlorine bonds. The displacementof the imido group from 0a is thus likely to commence fromthis adduct and furnish unbound 0b in a facile process that isuphill by 6.0 kcal mol−1. However, abstraction of the remainingimido ligand from 0b is seen to be prevented on thermodynamicgrounds, which is in agreement with observations on bis(imido)–MoVI compounds.27

Scheme 3 Imido ligand abstraction from [WVI(NPh)2Cl2] by the Lewisacid.

The indicated viable conversion of bis(imido)–WVI intomono(imido)–WVI compounds by the Lewis acid opens an alterna-tive reaction channel proceeding through mono(imido) tungstenspecies. This channel has a smooth reaction profile, which isconsistent with observed catalytic abilities.28 As a conclusion,this leads us to suggest that not the anticipated bis(imido)tungsten species but mono(imido) tungsten species are effectingdimerisation catalysis in the reported catalyst system.

Probing conceivable paths for 1 → 2 conversion. In orderto enhance the insight into the catalytic system further, severalconceivable paths for 1 → 2 transformation have been assessed.Scheme 4 reports the Gibbs free-energy profile for type II Lewisacid adducts, whereas the full account of data is included in theESI† (Scheme S1).

One avenue starts from 1′ with reductive ethane elimination tofurnish the bis(imido)–WIV species A, which would lead directly to2 after uptake of two olefin moieties. The reductive CC-elimination(1′ → A + C2H6) is indicated as being slow, having a significantbarrier of 45.5 kcal mol−1. The alternative ethylene insertion intoone of the WVI–Me bonds (1′ + C2H4 → B) is more facile andaffords the [WVI(NPh)2Me/Pr] intermediate B in an exergonicprocess. A multitude of steps are imaginable to commence fromB, of which the following have been examined: (a) B → A +C4H10 reductive butane elimination, (b) B + C2H4 → D ethylene

This journal is © The Royal Society of Chemistry 2008 Dalton Trans., 2008, 2120–2127 | 2125

Publ

ishe

d on

28

Febr

uary

200

8. D

ownl

oade

d by

Lom

onos

ov M

osco

w S

tate

Uni

vers

ity o

n 22

/05/

2013

21:

52:1

8.

View Article Online

Scheme 4 Condensed Gibbs free-energy profile (kcal mol−1) of alternativereaction routes for transformation of [WVI(NPh)2Me2] compound 1 intothe [WVI(NPh)2(C2H4)2] active catalyst species 2, to be assisted by twoassociated AlClMe2 moieties (association mode II). The activation andreaction free energies for individual steps are given relative to the respectiveprecursor. The favourable route is indicated by bold reaction arrows.

insertion into the WVI–propyl linkage, (c) formation of the WIV–carbene species E by methane a-elimination (B → E + CH4), whichmay afterwards be involved in coupling or metathesis steps withthe olefin substrate, and (d) reductive b-CH elimination (B →C + CH4) with liberation of methane. As revealed from Scheme 4,the last step is most accessible among the competing steps onboth kinetic and thermodynamic grounds. The final propylenedisplacement in C by two ethylene molecules is slightly downhill.

The examination of possible paths for 1 → 2 conversion suggeststhe following sequence of steps (Scheme 3): 1′ + C2H4 → B; B →C + CH4; C + 2C2H4 → 2 + C3H6. The entire process (1′ +3C2H4 → 2 + CH4 + C3H6) has a largest overall barrier of 34.0 kcalmol−1 and is driven by a thermodynamic force of 27.9 kcal mol−1

following the realistic scenario where two AlClMe2 moieties assistthe conversion.

Concluding remarks

Herein is reported the DFT examination of relevant elementarysteps of a-olefin dimerisation by a bis(imido) tungsten–Lewis acidcatalyst system by employing a catalyst model that closely mimicsthe real system and simulating authentic reaction conditions. Theextensive DFT simulation of configurationally and conformation-ally different association modes unveiled that AlClMe2 preferablyforms four-membered chelates with the imido groups throughN→Cl and Cl→W dative bonds. Complexation of one imidogroup is predicted as the favourable, stable mode of Lewis acidassociation onto a bis(imido) tungsten(VI) compound (cf. Fig. 2),which can be anticipated as being isolable by X-ray diffraction.

Several conceivable paths for 1 → 2 conversion have beenassessed and a practicable sequence of steps has been suggested.

The cocatalyst is found to affect the energetics of individual stepsof the dimerisation process to different extents. The oxidative cou-pling of two olefin molecules is remarkably facile and reversible,and appears to be least influenced by the Lewis acid. A morepronounced, but still moderate influence is found for the growthof the five-membered metallacycle intermediate. This contrastswith its degradation, which is greatly affected by the Lewis acidcomponent. AlClMe2 association across both tungsten–imidolinkages appears as being most effective in smoothing the energyprofile for the dimer formation.

The assessed energy profile for the most accessible pathwayof the dimer-generating route together with the indicated strongpreference of the five-membered metallacycle to grow than todegrade, however, are at odds with experiment. On the otherhand, the present study revealed a viable path for abstractionof one imido group by the Lewis acid, thereby making an alter-native, kinetically feasible reaction channel proceeding throughmono(imido) tungsten compounds accessible. This leads us tosuggest that mono(imido) tungsten(VI) species but not anticipatedbis(imido) tungsten(VI) species are effecting dimerisation catalysisin the reported catalyst system. Further investigations to clarifythe significance of bis(imido) tungsten(VI) species in olefin dimeri-sation are underway.

Acknowledgements

The author wishes to thank Prof. Malkin (Slovak Akademyof Sciences, Bratislava, Slovakia) and Prof. Kaupp (Universityof Wurzburg, Germany) for making the MAG-ReSpect moduleavailable to us, and also Prof. Helgaker (University of Oslo,Norway) for permission to use some of the DALTON routines.

Notes and references

1 (a) W. Keim, Angew. Chem., Int. Ed. Engl., 1990, 29, 235; (b) J.Skupinska, Chem. Rev., 1991, 91, 613; (c) Y. Chauvin, and H. Olivier,in Applied Homogeneous Catalysis with Organometallic Compounds, ed.B. Cornils, and W. A. Herrmann, VCH, New York, 1996, pp. 258–268;(d) G. W. Parshall, and S. D. Ittel, in Homogeneous Catalysis. TheApplications and Chemistry of Catalysis by Soluble Transition MetalComplexes, Wiley, New York, 2nd edn, 1992, pp. 72–85.

2 (a) For recent reviews see:S. D. Ittel, L. K. Johnson and M. Brookhart,Chem. Rev., 2000, 100, 1169; (b) S. Mecking, Angew. Chem., Int. Ed.,2001, 40, 534; (c) V. C. Gibson and S. K. Spitzmesser, Chem. Rev., 2003,103, 283.

3 (a) B. L. Small and A. J. Marcucci, Organometallics, 2001, 20, 5738;(b) B. L. Small, Organometallics, 2003, 22, 3178; (c) B. L. Small and R.Schmidt, Chem.–Eur. J., 2004, 10, 1014.

4 K. P. Tellmann, V. C. Gibson, A. J. P. White and D. J. Williams,Organometallics, 2005, 24, 280.

5 R. D. Broene, M. Brookhart, W. M. Lamanna and A. F. Volpe, J. Am.Chem. Soc., 2005, 127, 17194.

6 (a) B. Ellis, W. Keim and P. Wasserscheid, Chem. Commun., 1999, 337;(b) P. Wasserscheid and M. Eichmann, Catal. Today, 2001, 66, 309.

7 (a) T. Agapie, S. J. Schofer, J. A. Labinger and J. E. Bercaw, J. Am.Chem. Soc., 2004, 126, 1304; (b) M. J. Overett, K. Blann, A. Bollmann,J. T. Dixon, D. Haasbroek, E. Killian, H. Maumela, D. S. McGuinnessand D. H. Morgan, J. Am. Chem. Soc., 2005, 127, 10723; (c) A. K.Tomov, J. J. Chirinos, D. J. Jones, R. J. Long and V. C. Gibson, J. Am.Chem. Soc., 2005, 127, 10166.

8 (a) A. Carter, S. A. Cohen, N. A. Cooley, A. Murphy, J. Scutt andD. F. Wass, Chem. Commun., 2002, 858; (b) D. S. McGuiness, P.Wasserscheid, W. Keim, C. Hu, U. P. Englert, J. T. Dixon and C. Grove,

2126 | Dalton Trans., 2008, 2120–2127 This journal is © The Royal Society of Chemistry 2008

Publ

ishe

d on

28

Febr

uary

200

8. D

ownl

oade

d by

Lom

onos

ov M

osco

w S

tate

Uni

vers

ity o

n 22

/05/

2013

21:

52:1

8.

View Article Online

Chem. Commun., 2003, 334; (c) D. S. McGuinness, P. Wasserscheid, W.Keim, D. H. Morgan, J. T. Dixon, A. Bollmann, H. Maumela, F. M.Hess and U. Englert, J. Am. Chem. Soc., 2003, 125, 5272.

9 A. Bollmann, K. Blann, J. T. Dixon, F. M. Hess, E. Killian, H.Maumela, D. S. McGuiness, D. H. Morgan, A. Neveling, S. Otto,M. Overett, A. M. Z. Slawin, P. Wasserscheid and S. Kuhlmann, J. Am.Chem. Soc., 2004, 126, 14712.

10 (a) H. M. Menapace, N. A. Maly, J. L. Wang and L. G. Wideman,J. Org. Chem., 1975, 40, 2983; (b) H. Olivier and P. Laurent-Gerot,J. Mol. Catal. A: Chem., 1999, 148, 43.

11 M. J. Hanton, and R. P. Tooze, WO Pat., 2005/089940 (SasolTechnology (UK) Ltd), 2005; the {1 + Lewis acid} catalyst system ishighly selective to dimerisation (with %(higher oligomers + polymers)<1) and moderately active (N t ∼110 (mol olefin) (mol W)−1 h−1, 60 ◦C).

12 (a) R. R. Schrock, R. T. DePue, J. Feldman, K. B. Yap, D. C. Yang,W. M. Davis, L. Park, M. DiMare, M. Schofield, J. Anhaus, E.Walborsky, E. Evitt, C. Kruger and P. Betz, Organometallics, 1990,9, 2262; (b) H. H. Fox, K. B. Yap, J. Robbins, S. Cai and R. R. Schrock,Inorg. Chem., 1992, 31, 2287.

13 (a) A. N. J. Blok, P. H. M. Budzelaar and A. W. Gal, Organometallics,2003, 22, 2564; (b) S. Tobisch and T. Ziegler, Organometallics, 2003, 22,5392.

14 (a) R. Ahlrichs, M. Bar, H.-P. Baron, R. Bauernschmitt, S. Bocker, N.Crawford, P. Deglmann, M. Ehrig, K. Eichkorn, S. Elliott, F. Furche,F. Haase, M. Haser, H. Horn, C. Hattig, C. Huber, U. Huniar, M.Kattannek, A. Kohn, C. Kolmel, M. Kollwitz, K. May, P. Nava, C.Ochsenfeld, H. Ohm, H. Patzelt, D. Rappoport, O. Rubner, A. Schafer,U. Schneider, M. Sierka, O. Treutler, B. Unterreiner, M. von Arnim, F.Weigend, P. Weis, H. Weiss, TURBOMOLE V5.8, Quantum ChemistryGroup, University of Karlsruhe, Germany, 1988; (b) R. Ahlrichs, M.

Bar, M. Haser, H. Horn and C. Kolmel, Chem. Phys. Lett., 1989, 162,165; (c) O. Treutler and R. Ahlrichs, J. Chem. Phys., 1995, 102, 346.

15 J. Tao, J. P. Perdew, V. N. Staroverov and G. E. Scuseria, Phys. Rev.Lett., 2003, 91, 146401.

16 D. Andrae, U. Haußermann, M. Dolg, H. Stoll and H. Preuß, Theor.Chim. Acta, 1990, 77, 123.

17 A. Schafer, C. Huber and R. Ahlrichs, J. Chem. Phys., 1994, 100,5829.

18 (a) V. N. Staroverov, G. E. Scuseria, J. Tao and J. P. Perdew, J. Chem.Phys., 2003, 119, 12129; (b) Y. Zao and D. G. Truhlar, J. Chem. TheoryComput., 2005, 1, 415; (c) F. Furche and J. P. Perdew, J. Chem. Phys.,2006, 124, 044103.

19 D. A. McQuarrie, Statistical Thermodynamics, Harper & Row, NewYork, 2003.

20 A. Klamt and G. Schuurmann, J. Chem. Soc., Perkin Trans. 2, 1993,799.

21 S. Tobisch and T. Ziegler, J. Am. Chem. Soc., 2004, 126, 9059.22 T. J. Crevier and J. M. Mayer, Angew. Chem., Int. Ed., 1998, 37, 1891.23 P. D. Bolton, E. Clot, A. R. Cowley and P. Mountford, J. Am. Chem.

Soc., 2006, 128, 15005.24 T.R. Cundari, Chem. Rev., 2000, 100, 807.25 Tungstanacycle imido intermediates have at least two potential sites for

reaction with an olefin, viz. the W–C and W=NR bonds. The W–Cbond appears to be more reactive towards the olefin, which will bedetailed in a forthcoming paper.

26 Examination by a linear-transit approach gave no indication that thisprocess is associated with a significant enthalpy barrier.

27 (a) E. A. Maatta and R. A. D. Wentworth, Inorg. Chem., 1979, 18,2409; (b) D. A. Wigley, Prog. Inorg. Chem., 1994, 42, 239.

28 S. Tobisch, Organometallics, 2007, 26, 6529.

This journal is © The Royal Society of Chemistry 2008 Dalton Trans., 2008, 2120–2127 | 2127

Publ

ishe

d on

28

Febr

uary

200

8. D

ownl

oade

d by

Lom

onos

ov M

osco

w S

tate

Uni

vers

ity o

n 22

/05/

2013

21:

52:1

8.

View Article Online