DaltonTransactions

Dynamic Article Links

Cite this: Dalton Trans., 2012, 41, 9182

www.rsc.org/dalton PAPER

Mechanistic elucidation of the yttrium(III)-catalysed intramolecularaminoalkene hydroamination: DFT favours a stepwise σ-insertivemechanism†

Sven Tobisch*

Received 15th April 2012, Accepted 1st June 2012DOI: 10.1039/c2dt30814b

A comprehensive computational mechanistic study regarding intramolecular hydroamination (HA) ofaminoalkenes mediated by a recently reported class of highly active cyclopentadienyl-bis(oxazolinyl)-borate {Cpo}YIII alkyl compounds is presented. Two distinct mechanistic pathways of catalytic HAmediated by rare earth and alkaline earth compounds have emerged over the years, describingamidoalkene → cycloamine conversion proceeding through a stepwise σ-insertive mechanism or aconcerted non-insertive N–C/C–H bond forming pathway. Notably, both mechanisms account equally forreported distinct process features. Non-competitive kinetic demands revealed for the concerted aminoproton transfer associated with N–C ring closure, which commences from a {CpoM}Y(NHR)·(NH2R)substrate adduct and evolves through a six-centre TS structure, militates against a proton-triggered non-insertive pathway to promote HA for the rare earth catalyst at hand. A stepwise σ-insertive pathway,featuring rapid and reversible olefin insertion into the Y–N amido σ-bond, linked to a less facile andirreversible intramolecular Y–C azacycle tether aminolysis, is found to prevail energetically. The assessedeffective barrier for turnover-limiting aminolysis matches the empirically determined Eyring parameterwell and the computationally estimated primary KIE is close to the observed values. A recentcomputational study revealed a similar scenario for an analogous tris(oxazolinyl)borate {ToM}Mg system.Valuable insights into the catalytic structure–reactivity relationships have been unveiled by a comparisonof {CpoM}Y(NHR)- and {ToM}Mg(NHR)-catalysed hydroaminations.

Introduction

Catalytic hydroamination (HA), the direct addition of an N–Hbond across an unsaturated carbon–carbon linkage, offers facileand atom-economical access to an industrially relevant organo-nitrogen commodity and fine chemicals.1 The intramolecularhydroamination/cyclisation constitutes a particularly powerfuland concise route to functionalised nitrogen heterocycles.Significant research efforts over the past two decades have estab-lished rare earth complexes as highly active catalysts for thisprocess; their versatility has been broadly demonstrated.2 Furthercatalyst development, however, critically relies on a preciseunderstanding of the operative mechanistic pathway.

With regard to HA mediated by rare earth and alkaline earthcompounds, two distinct mechanistic pathways have emergedover the years as a result of comprehensive mechanistic explora-tion of organolanthanide-catalysed intramolecular hydroamina-tion pioneered by Marks’ group.1k Substantiated by a wealth of

both experimental1k and computational3 evidence, it is com-monly thought that close interaction with the electropositive rareearth activates the unsaturated C–C linkage towards nucleophilicamido attack. This σ-insertive mechanism2b (Scheme 1, cycle A)involves N–C ring closure through the insertion of a CvClinkage into the metal–N amido σ-bond (3′ → 4) followed bymetal–C azacycle bond aminolysis (4·S → 5). Whilst distinctivereactivity trends observed for aminoalkene and aminoalkyne sub-strates1k,2b are indicative of a turnover-limiting cyclisation, com-putational mechanistic studies are in support of Ln–C aminolysisas being turnover limiting for aminoallene3c,3d and amino-diene3e,3g substrates.

As a particularly intriguing aspect, rare earth-catalysed hydro-amination often proceeds with large primary kinetic isotopeeffects (KIEs) of N-proteo versus N-deutero aminoalkene sub-strates2b,2u and sometimes with distinct isotope effects on diaster-eoselectivity.2b On the basis of these observations, a proton-assisted insertion mechanism was originally suggested2b and themechanistic detail was refined later as proceeding though a mul-ticentre transition state (TS) structure that describes concurrentN–C bond formation and metal–C bond aminolysis (Scheme 1,cycle B).4,5 The non-insertive pathway affords the cycloamineproducts in a single step commencing from a rare earth–amidosubstrate adduct (3·S). In sharp contrast to the σ-insertive

†Electronic supplementary information (ESI) available: Additionaldetails. See DOI: 10.1039/c2dt30814b

School of Chemistry, University of St Andrews, Purdie Building, NorthHaugh, St Andrews, United Kingdom KY16 9ST.E-mail: st40@st-andrews.ac.uk; Fax: (+44) 1797-383-652

9182 | Dalton Trans., 2012, 41, 9182–9191 This journal is © The Royal Society of Chemistry 2012

Publ

ishe

d on

19

June

201

2. D

ownl

oade

d by

Uni

vers

ity o

f C

alif

orni

a -

Sant

a C

ruz

on 2

2/10

/201

4 12

:56:

46.

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

mechanism, close proximity of the CvC linkage and the rareearth is no longer required along the non-insertive pathway,since the delivery of an amino proton activates primarily theCvC linkage to undergo concurrent N–C ring closure.

Computational studies revealed that the non-insertive pathwayis energetically less favourable than the σ-insertive mechanismin Cp*2Ln(NHR)-catalysed cyclisation of aminoallenes3d andaminodienes,3g but indicated that it may become operational forcatalysts with sterically demanding ligand architectures. A con-certed N–C/C–H bond forming mechanism was suggested forsome cyclopentadienylphosphazene rare earth,6 tris(oxazolinyl)-borate magnesium,7a diketiminate group 27b and phenoxyaminemagnesium7c catalysts. Noteworthy, a recent computationalstudy8 described the concerted non-insertive pathway as a viablemechanistic alternative for alkaline earth-catalysed HA of amino-alkenes, but concluded that a σ-insertive mechanism prevails fora tris(oxazolinyl)borate magnesium7a system. On the basis ofobserved reactivity and activity trends cyclopentadienyl-salicyl-oxazoline zirconium(IV)9a and cyclopentadienyl-bis(oxazolinyl)-borate zirconium(IV)9b catalyst systems were proposed to cycliseaminoalkenes through a non-insertive six-centre mechanism

(similar to cycle B in Scheme 1), but this was discarded for theformer in favour of a cycloaddition mechanism. Complementaryexperimental9c and computational10 mechanistic elucidation pro-vided compelling evidence for the operation of a proton-trig-gered non-insertive cyclisation mechanism for aminoalkene HAby a ZrIV(NHR)2 catalyst with a sterically encumbering tetheredbis(ureate) ligand.

Recently, Sadow and co-workers reported a cyclopentadienyl-bis(oxazolinyl)borate {CpoM}YIII alkyl compound 2 (CpoM =bis(4,4-dimethyl-2-oxazolinyl)(cyclopentadienyl)phenylborate)as a highly active precatalyst for intramolecular hydroaminationof both primary and secondary 2,2-disubstituted aminoalkenes(Scheme 2).11 Detailed mechanistic and kinetic studies of a{Cpot-Bu}Y alkyl congener 2* revealed an empirical formalsecond-order rate law, describing a first order dependence on theconcentration of both catalyst and substrate together with somesubstrate inhibition. A general mechanistic scenario shown ineqn (1),11 which comprises rapid catalyst–substrate association–dissociation equilibria (K1, Ka), prior to the turnover-limitingevent commencing at a cat·sub adduct (k2), is in accord withthe empirical rate law, but other possibilities exist. The

Scheme 1 Plausible mechanistic pathways for rare earth-catalysed intramolecular aminoalkene hydroamination, exemplified for {CpoM}Y(NHR) 3′as catalytically relevant species and aminoalkene substrate 1 ({CpoM} = {PhB(C5H4)(Ox

Me2)2}).

This journal is © The Royal Society of Chemistry 2012 Dalton Trans., 2012, 41, 9182–9191 | 9183

Publ

ishe

d on

19

June

201

2. D

ownl

oade

d by

Uni

vers

ity o

f C

alif

orni

a -

Sant

a C

ruz

on 2

2/10

/201

4 12

:56:

46.

View Article Online

turnover-limiting step involves N–H bond cleavage, as evi-denced by an observed significant primary KIE. Remarkably,α-NH2 by α-ND2 substrate substitution also effects a distinctdecrease in enantioselectivity.11 Taking all these findingstogether, Sadow and co-workers suggested the operation of thenon-insertive pathway (Scheme 1, cycle B).11 Although itappears tempting to conclude from the isotopic perturbation ofenantioselectivity that the stereochemistry-determining N–Cbond formation involves a N–H(D) bond, it does not eliminatethe stepwise insertive pathway (Scheme 1, cycle A) a priori.Given that the N–H bond cleavage does not occur at the insertivecyclisation step, as it does in Y–C bond protonolysis, the isotopeeffect on enantioselectivity strongly suggests that insertive cycli-sation is reversible and faster than the subsequent aminolysiswhen proceeding through cycle A. It is noted that a similar scen-ario has recently been put forth by Hultzsch for a related inter-molecular hydroaminoalkylation.12 This process also displays amarked perturbation of enantioselectivity upon substrate deutera-tion, whilst the stereochemistry-determining step does notinvolve the element–D bond. Rapid and reversible insertivecyclisation linked to a less facile and irreversible Y–C bondaminolysis along pathway A in Scheme 1 accounts for all theobserved features, as does pathway B.

ð1Þ

In light of the remaining uncertainty as to which of the path-ways in Scheme 1 prevails, we employed density functionaltheory (DFT) calculations as an established and predictivemethod. Its particular value for unravelling mechanistic intrica-cies in rare earth-catalysed hydroamination has been demon-strated by several groups.3,13 Having the focus on rivalmechanistic pathways, this study reports a comprehensive scru-tiny of intramolecular HA of prototype 2,2-dimethyl-4-penten-1-amine aminoalkene 1 (S) by [{CpoM}YCH2SiMe3] 2 withoutimposing structural simplification on any of the key speciesinvolved. Aspects of the mechanism of stereoinduction by chrial2* will be subjected to forthcoming studies. The employed DFTmethodology (meta-GGATPSS in conjunction with basis sets oftriple-ζ quality, see the computational methodology section),which has previously been thoroughly benchmarked and suc-cessfully applied on related 4f,3g 5f,14 and alkaline earth8

element-mediated HA, simulated authentic reaction conditionsadequately and mechanistic discussions were based on Gibbsfree-energy profiles.

The computational examination presented herein favours astepwise σ-insertive mechanism that entails fast and reversibleolefin insertion into the Y–N amido σ-bond, linked to irrevers-ible intramolecular Y–C azacycle tether aminolysis as the mostaccessible pathway. The assessed effective activation barrier forturnover-limiting aminolysis and the associated primary KIEmatches observed values well and moreover the derived rate lawis consonant with the empirical rate law obtained experimentally.On the other hand, a computed less favourable kinetic profilemilitates against the operation of the alternative pathway, whichdescribes N–C ring closure triggered by concurrent aminoproton delivery at the CvC linkage, proceeding through a six-centre TS structure for the catalyst at hand.

Results and discussion

This study starts with an examination of the transformation ofyttrium alkyl starting material 2 into the catalytically relevantY(NHR) compound. Key features of the Y–N amido stepwiseσ-bond insertive and concerted non-insertive pathways ofScheme 1 are discussed next and the two mechanisms are com-pared and contrasted thereafter. The accessibility of the yttriumcentre is likely to profoundly control the energy balance betweenthe rival pathways. The aptitude for substrate (S) and cycloamine(P) binding has therefore been probed explicitly for each inter-mediate involved.

Catalyst initiation

Effective HA entails the initial smooth conversion of yttriumalkyl starting material 2 into the catalytically competent[{CpoM}Y(NHR)] compound. Fig. 1 collates the assessedenergy profile for protonolytic Y–C bond cleavage by the sub-strate. The initial facile substrate association15 is slightly down-hill in energy and leads to adduct 2·S with an intact (κ2-Ox2)Yligation of the two oxazoline rings.16 An isomer featuring oneamine N–H stabilised dissociated oxazoline ring is found to beat higher energy (Fig. S1 in the ESI†). Commencing from 2·S,Y–C bond aminolysis evolves through a metathesis-type TS[2·S–3] that decays thereafter into (κ2-Ox2){Cpo

M}Y(NHR) 3through facile liberation of SiMe4. Fig. 1 reveals a moderatebarrier of 15.9 kcal mol−1 (relative to 2·S) for the protonolyticexpulsion of the alkyl ligand at 2, thereby characterising itstransformation into the catalytically relevant Y(NHR) complexas a facile step, which is furthermore driven by a largethermodynamic force and, hence, is irreversible. The possiblesupportive involvement of a second substrate molecule has beenprobed but was found to be ineffective (Fig. S1 in the ESI†).

Scheme 2 Intramolecular HA of a primary 2,2-disubstituted aminoalk-ene mediated by [{CpoM}YCH2SiMe3] 2 and chiral [{Cpot-Bu}-YCH2SiMe3] 2* ({CpoM} = {PhB(C5H4)(Ox

Me2)2}; {Cpot-Bu} = {PhB-

(C5H4)(S-Oxt-Bu)2).

Fig. 1 Y–C bond aminolysis at 2 by substrate 1.17a

9184 | Dalton Trans., 2012, 41, 9182–9191 This journal is © The Royal Society of Chemistry 2012

Publ

ishe

d on

19

June

201

2. D

ownl

oade

d by

Uni

vers

ity o

f C

alif

orni

a -

Sant

a C

ruz

on 2

2/10

/201

4 12

:56:

46.

View Article Online

The {CpoM}Y(NHR) compound

Amine association onto the catalytically competent [{CpoM}Y-(NHR)] compound is examined first (Fig. 2 and S2 in the ESI†).The favourable substrate-free species 3′ features an amidoalkeneunit that forms a weak chelate by orienting its double bond prox-imal to yttrium, thereby adopting a geometry that is bestdescribed as a four-legged piano-stool. A three-legged piano-stool isomer 3 with a monohapto N-ligated amidoalkene moietyis found higher in energy. As the stability of 3′ with a chelateY(NHR) unit indicates, a moderately sterically encumbering{CpoM}Y ligand backbone renders the amine association afford-able energetically. Species with one or two adducted substrateand pyrrolidine molecules have been located. Although a d0

yttrium centre favours the coordination of an amine over anolefin (in terms of enthalpy), the substrate adducts 3·S and 3′·Swith a monohapto ligated and a chelated amidoalkene unit,respectively, are placed above 3′ in free energy, due to theentropy penalty associated with amine complexation. The associ-ation of a single amine molecule inverts the relative stability ofthe substrate-free species 3′ and 3. Now, 3·S is found somewhatfavoured over 3′·S, with both featuring an intact (κ2-Ox2)Y lig-ation. On the other hand, the association of another amine mole-cule involves dissociation of one oxazoline ring, to be stabilisedby an amine H-bonded contact. Not unexpectedly, the energygap between species with a (κ1-NHR)Y (3·(S)2) and (κ2-NHR)Y(3′·(S)2) ligated amidoalkene increases substantially upon associ-ation of a second amine molecule (Fig. S2 in the ESI†). Thecomplexation of cycloamine and substrate follows a similarpattern, with the primary amine binding more strongly. Asrevealed from Fig. 2, DFT predicts that the catalytically compe-tent [{CpoM}Y(NHR)] compound is predominantly present asspecies 3′ on entry to the catalytic manifold, together with lesspopulated mono-amine (3·S, 3′·S) and bis-amine (3·(S)2)adducted species, all of which are expected to participate inrapid amine association/dissociation equilibria.15

The σ-insertive mechanism

We start with an examination of the insertive pathway(Scheme 1, cycle A) that involves migratory olefin insertion intothe Y–N amido σ-bond to proceed at 3′ or its substrate adductsand subsequent intramolecular protonolytic cleavage of the Y–Cazacycle tether bond by an yttrium bound substrate molecule. Itis noted that a feasible insertive cyclisation necessitates an effec-tive activation of the olefin unit towards a nucleophilic amido

approach by its close interaction with the electropositive rareearth.

Migratory olefin 1,2-insertion into the Y–N amido σ-bond.Various trajectories for insertive cyclisation commencing from 3′and its mono- and bis-amine adducted species have been exam-ined. The prevalent {CpoM}Y(NHR) species 3′, which has noadditional amine molecule coordinated (Fig. 2), is the precursorfor the most accessible pathway. It evolves through a four-centreTS structure (TS[3′–4]) describing an yttrium-mediated olefininsertion into the Y–N amido σ-bond, which occurs at a distanceof 1.96 Å for the emerging C–N bond (Fig. 3). Commencingfrom 3′ with an already suitably attached double bound, no sub-stantial structural reorganisation takes places whilst traversingthe (κ2-Ox2)Y TS[3′–4]. Hence, this process has a modest acti-vation energy of 8.9 kcal mol−1. Following the reaction pathfurther, TS[3′–4] decays into (κ2-Ox2)Y yttrium alkyl intermedi-ate 4, which has the azacycle bound to yttrium through itsmethylene tether and the N donor centre. Because the newlyestablished azacycle yttrium coordination cannot completelycompensate for the cleavage of the Y–N(amido) bond, it places4 3.4 kcal mol−1 above 3′ in free energy (Fig. 4).

The species participating in a pathway with one coordinatedsubstrate molecule all feature a (κ2-Ox2)Y ligation. As is indi-cated by a similar stability of 3·S and 3′·S (Fig. 2), a relativelyopen {CpoM}Y ligation makes the yttrium centre with onebound amine molecule accessible for the olefin unit at stillaffordable energy costs. Indeed, pathways without (3′ → 4, ΔG‡/ΔG = 8.9/3.4 kcal mol−1) and with one adducted substrate mol-ecule (3·S → 4·S, ΔG‡/ΔG = 9.5/4.4 kcal mol−1, relative to 3·S)exhibit intrinsic energy profiles of comparable magnitude(Fig. 4). Thus, it is not the activation barrier for C–N ringclosure, but the unfavourable thermodynamics of 3′ + S → 3·Ssubstrate coordination that renders insertive cyclisation at 3·Sless probable. Likewise, the participation of another spectatorsubstrate molecule via 3·(S)2 → 4·(S)2, which promotes amine-assisted dissociation of one oxazoline ring, drives the process tobe substantially less affordable.

Migratory olefin insertion into a highly polar Y–N amidoσ-bond at 3′ has a remarkably modest barrier of 8.9 kcal mol−1 toovercome and is somewhat endergonic. It characterises insertivering closure for a {CpoM}Y(NHR) catalyst species with aneasily accessible yttrium centre as a kinetically highly facile,reversible step that favours 3′. We note that previous compu-tational studies revealed a kinetically easy and reversible inser-tion step that proceeds without the involvement of spectatoramine molecules for analogous tris(oxazolinyl)borate mag-nesium ({ToM}Mg)8 and lanthanocene3a systems. In light of theless accessible compact magnesium centre the assessed barrier of16.2 kcal mol−1 becomes understandable, whilst for a moder-ately bulky Cp2La(NHR) catalyst a free-energy of activation of12.9 kcal mol−1 was reported.

Y–C azacycle tether aminolysis. The necessary complexationof a substrate molecule cis to the Y–C linkage at 4 results insome relaxation within the (κ2-Ox2)Y backbone, whilst adductspecies with one amine-stabilised dissociated oxazoline ring arefound at higher energy (Fig. S6 in the ESI†). Several trajectoriesfor intramolecular H transfer have been probed (Scheme 3). The

Fig. 2 Various forms of the catalytically relevant {CpoM}Y(NHR)compound ({CpoM} = {PhB(C5H4)(Ox

Me2)2}).15,17b

This journal is © The Royal Society of Chemistry 2012 Dalton Trans., 2012, 41, 9182–9191 | 9185

Publ

ishe

d on

19

June

201

2. D

ownl

oade

d by

Uni

vers

ity o

f C

alif

orni

a -

Sant

a C

ruz

on 2

2/10

/201

4 12

:56:

46.

View Article Online

minimum-energy pathway commences from 4′′·S and describesprotonolytic cleavage of an already weakened axial Y–C tetherbond that is trans to the Cp ring, whilst aminolysis of an equator-ial Y–C bond at the more stable isomer 4·S (that is generatedthrough 3·S → 4·S cyclisation) is disfavoured kinetically. Inneither case is the enthalpy for amine complexation large enoughto compensate for its entropy penalty,15 which places both 4·Sand 4′′·S above 4 in free energy (Fig. 4). Aminolysis evolvesthereafter through a metathesis-type (κ2-Ox2)Y TS structure (TS-[4′′·S–5], Fig. 5) that describes the cleavage of an already suit-ably polarised N–H bond and concurrent C–H bond formation.All attempts to localise an isomeric TS structure with one amine-stabilised dissociated oxazoline ring were unsuccessful, but indi-cated that such a TS structure is to be expected at higher energy.The most accessible 4 + S → 4′′·S → 5 pathway for Y–C bond

aminolysis has a barrier of 14.3 kcal mol−1 (relative to {4 + S})and affords (κ2-Ox2) [{Cpo

M}Y(NHR)(NHRR)] compound 5 ina process that is downhill by 8.4 kcal mol−1 (relative to {4 + S},Fig. 6). The facile displacement of pyrrolidine P by substrate Sat 5 (ΔG = −7.1 kcal mol−1), renders aminolysis even morefavourable thermodynamically.

Similar to what is found for the insertive cyclisation step, ami-nolysis also appears to be appreciably more facile for the presentcatalyst with a strongly trans directing Cp ring when comparedto an analogous {ToM}Mg(NHR) system featuring a more effec-tively protected compact alkaline earth. It results in a DFTassessed barrier of 18.4 kcal mol−1 (relative to a magnesiumalkyl intermediate similar to 4) and involves a substrate-assisteddissociation of one oxazoline ring.8 In contrast, a computationalstudy using a Cp2La(NHR) catalyst model reported a free-energy

Fig. 3 Selected structural parameters (angstroms) of located TS structures for σ-insertive cyclisation through pathways without and with adductedsubstrate molecules.18

Fig. 4 Migratory olefin insertion into the Y–N amido σ-bond at 3′ and its substrate adducted species 3·S, 3·(S)2.17b

9186 | Dalton Trans., 2012, 41, 9182–9191 This journal is © The Royal Society of Chemistry 2012

Publ

ishe

d on

19

June

201

2. D

ownl

oade

d by

Uni

vers

ity o

f C

alif

orni

a -

Sant

a C

ruz

on 2

2/10

/201

4 12

:56:

46.

View Article Online

of activation of 10.9 kcal mol−1 (relative to a thermodynamicallyfavourable substrate adduct like 4·S) for La–C bond protonolysis,which proceeds with an initial exergonic substrate association.3a

In the light of the apparent structural simplification, it makes itdifficult to reconcile the assessed energetics with the experimen-tally studied Cp*2La(NHR) catalyst.

Overall, the assessed smooth energy profile for a catalyticallyrelevant {CpoM}Y(NHR) compound shown in Fig. 6 is indica-tive of a kinetically feasible and irreversible intramolecular Y–Cbond protonolysis step along a 4 + S → 4′′·S → 5 pathway. Par-ticipation of another coordinated substrate molecule, however, isunlikely to facilitate the process, because it does not stabilise thereactant, product or TS structures. It is worth noting that theinaccessibility of a 4·(S)2 → 5·S pathway primarily originatesfrom the unfavourable thermodynamics for complexation ofanother substrate molecule, since the intrinsic energy profile ofthe two pathways (4′′·S → 5, ΔG‡/ΔG = 10.0/−12.7 kcal mol−1

and 4·(S)2 → 5·S, ΔG‡/ΔG = 10.4/–15.2 kcal mol−1) are remark-ably similar (Fig. 6).

The proton-triggered non-insertive cyclisation mechanism

The rival mechanistic pathway B in Scheme 1 that describesN–C ring closure triggered by concurrent amino proton deliveryat the CvC linkage, thereby achieving amidoalkene → cyclo-amine transformation in a single step is examined next. DFTrevealed an affordable energy profile for the two trajectoriesshown in Scheme 4 involving N–C bond formation through anaxial olefin approach at a Y–C amido σ-bond, triggered by aproton delivered from an equatorially bound substrate, or alterna-tively describing concomitant ring closure and proton transferoccurring at equatorial sites. The single-step amidoalkene →cycloamine conversion commences from 3·S and evolvesthrough a six-centre TS structure that constitutes N–C5 bond for-mation with concurrent proton transfer from an yttrium coordi-nated and, hence, acidified substrate molecule onto the adjacentolefin–C6 centre. As exemplified in Fig. 7 for the more accessi-ble trajectory of equatorial olefin approach, the located TS-[3·S–5] features a (κ2-Ox2)Y ligation with a N–C5 distance of1.93 Å and vanishing N–H (1.24 Å) and emerging C–H(1.53 Å) bonds, indicative of a nearly half-complete protontransfer. The TS decays thereafter into (κ2-Ox2) [{CpoM}-Y(NHR)(NHRR)] compound 5.

It is perhaps instructive to compare the structural aspects ofthe located TS structures for the crucial steps of rival pathwaysA and B in Scheme 1, viz. Y–C azacycle tether aminolysis andproton-triggered non-insertive cyclisation. The TS[4′′·S–5] hasthe azacycle tether still bound at the yttrium (Fig. 5) and thusdescribes an yttrium-assisted Y–C bond protonolysis. Hence, itsviability (and also that of insertive cyclisation, Fig. 3) criticallyrelies on the accessibility of the rare earth. In contrast, proton-triggered cyclisation via TS[3·S–5] takes place outside of the

Fig. 5 Selected structural parameters (angstroms) of the located TSstructures for protonolytic Y–C bond cleavage through pathways withone or two adducted substrate molecules involved.18

Fig. 6 Y–C azacycle tether aminolysis at the yttrium alkyl intermediate through pathways with one or two adducted substrate moleculesparticipating.17b

Scheme 4 Alternative trajectories for the proton-triggered non-inser-tive N–C ring closure.

Scheme 3 Alternative trajectories for Y–C bond aminolysis at 4·S.

This journal is © The Royal Society of Chemistry 2012 Dalton Trans., 2012, 41, 9182–9191 | 9187

Publ

ishe

d on

19

June

201

2. D

ownl

oade

d by

Uni

vers

ity o

f C

alif

orni

a -

Sant

a C

ruz

on 2

2/10

/201

4 12

:56:

46.

View Article Online

immediate vicinity of the yttrium centre, as indicated by therather long Y–C6 distance of 3.80 Å. Moreover, aminolysis andnon-insertive cyclisation steps favour an alignment of reactingmoieties that is different. Aminolysis preferably involves anaxial Y–C bond via TS[4′′·S–5] and benefits from a trans dis-posed Cp ligand (Scheme 3, Fig. 6). However, its influence natu-rally vanishes for non-insertive cyclisation and as a result atrajectory for concurrent N–C bond formation and amino protondelivery at equatorial sites via TS[3·S–5] prevails energetically(Scheme 4, Fig. 8). It has a barrier of 27.3 kcal mol−1 toovercome and affords 5 in a process that is downhill by5.0 kcal mol−1 (Fig. 8). Thus, proton-triggered non-insertivecyclisation is irreversible as is Y–C bond aminolysis and bothsteps furnish compound 5. Similar to previous discussion for theaminolysis step, the association of another spectator substrate

molecule, which provokes dissociation of one oxazoline ring,does not stabilise the reactant, product or TS structures andadditional amine molecules are therefore unlikely to participatein this step.

A recent DFT study8 reported a kinetically viable (ΔG‡ =30.2 kcal mol−1, relative to a Mg(NHR) species similar to 3) andirreversible pathway for concerted non-insertive cyclisation for arelated [{ToM}Mg(NHR)] compound. Whilst comparing the twocatalysts it is worth noting that the barriers for insertive cyclisa-tion and aminolysis steps are markedly lower by 7.3 kcal and7.5 kcal mol−1 for the yttrium system, whereas the gap narrowsdown to 2.9 kcal mol−1, again in favour of the yttrium system,for non-insertive cyclisation. A more accessible rare earth at thesterically relatively open catalyst at hand is, among others, amajor factor affecting this behaviour.

Comparison of mechanistic pathways

This section compares and contrasts the mechanistic alternativesshown in Scheme 1. Fig. 9 collates the assessed free-energyprofiles for the σ-insertive (right) and non-insertive (left)mechanistic pathways for intramolecular HA of aminoalkene 1by a catalytically competent [{CpoM}Y(NHR)] compound. Non-insertive cyclisation through a six-centre TS[3·S–5] structure forN–C ring closure triggered by concurrent amino proton deliveryis irreversible and has a total barrier of 27.3 kcal mol−1. Thispathway is consistent with the empirical rate law and theobserved large primary KIE. Nonetheless, the distinct deviationof the DFT assessed barrier19 from a value of 18.8 kcal mol−1

reported11 for cyclisation of a 2,2-cyclohexyl substituted4-aminoalkene S′ by {Cpot-Bu}Y alkyl congener 2*, togetherwith a predicted pronounced kinetic penalty versus a classicalσ-insertive pathway (ΔΔG‡ = 9.6 kcal mol−1 between turnover-limiting events, Fig. 9) strongly militates against the operation ofthe non-insertive mechanism.

On the other hand, the energetically prevalent σ-insertivepathway has a total barrier of 17.7 kcal mol−1 (relative to{CpoM}Y(NHR) 3′, which corresponds to the catalyst restingstate) to overcome for turnover-limiting Y–C azacycle tetheraminolysis that matches the empirically determined Eyring par-ameter11 well. DFT estimates a classical KIE of 3.5 to be associ-ated with TS[4′′·S–5].20 Taking tunnelling into account,20b it

Fig. 7 Selected structural parameters (angstroms) of the optimisedstructures of key stationary points for non-insertive N–C ring closurewith concurrent amino proton transfer at the double bond at 3·S and3·(S)2.18

Fig. 8 Non-insertive N–C ring closure with concurrent amino proton delivery at the olefin unit at 3·S and 3·(S)2.17b

9188 | Dalton Trans., 2012, 41, 9182–9191 This journal is © The Royal Society of Chemistry 2012

Publ

ishe

d on

19

June

201

2. D

ownl

oade

d by

Uni

vers

ity o

f C

alif

orni

a -

Sant

a C

ruz

on 2

2/10

/201

4 12

:56:

46.

View Article Online

results in a primary KIE of 3.7, which is close20d to an observedvalue of 2.6 ± 0.4 for cyclisation of S′ by 2*. The herein pre-sented computational analysis provides sound evidence that ami-noalkene HA by a catalytically relevant [{CpoM}Y(NHR)]compound proceeds through the σ-insertive pathway and entailsthe following steps (Scheme 5): (1) highly facile and reversible3′ ⇆ 4 migratory olefin insertion into the Y–N amido σ-bond ata Y(NHR) catalyst species 3′ (K3) that favours the reactant; 3′corresponds to the catalyst resting state and is in rapid equili-brium with its substrate-adducted species 3·S and 3·(S)2. (2) Ir-reversible turnover-limiting 4 + S → 5 Y–C bond protonolysis tobe followed by (3) facile and exergonic 5 → 3′ + P liberation ofthe cycloamine product to regenerate the active catalyst species.

A simplified enzyme-like rate law with an additional term thataccounts for substrate inhibition (see Scheme 5 and the ESI†)describes this scenario. It is consistent with the experimentallyderived empirical second-order rate law.11

It is worth noting that the scenario outlined in Scheme 5 bearsgreat similarity to the one proposed for an analogous tris(oxazo-linyl)borate magnesium system,8 including a facile and revers-ible σ-insertive ring closure that favours the M(NHR) catalystspecies with irreversible and turnover-limiting aminolysis occur-ring thereafter. These {CpoM}Y(NHR)- and {ToM}Mg(NHR)-catalysed hydroaminations display other similar features, com-prising large primary KIEs of N-proteo versus N-deutero amino-alkene substrates and empirical formal second-order rate laws,

Fig. 9 Overall reaction profile for intramolecular hydroamination of aminoalkene 1 by {CpoM}Y(NHR) catalyst species 3′ proceeding via alternativemechanistic pathways. Pyrrolidine product liberation through 5 → 3′ + P is included ({CpoM} = {PhB(C5H4)(Ox

Me2)2}).

Scheme 5 Proposed mechanistic scenario for the {CpoM}Y(NHR)-mediated intramolecular aminoalkene hydroamination ({CpoM} = {PhB(C5H4)-(OxMe2)2}).

This journal is © The Royal Society of Chemistry 2012 Dalton Trans., 2012, 41, 9182–9191 | 9189

Publ

ishe

d on

19

June

201

2. D

ownl

oade

d by

Uni

vers

ity o

f C

alif

orni

a -

Sant

a C

ruz

on 2

2/10

/201

4 12

:56:

46.

View Article Online

describing a first order dependence on the concentration of bothcatalyst and substrate together with some substrate inhibition. Itcontrasts with a bis(ureate) Zr(NHR)2 system that is also exhibit-ing significant primary KIEs and an empirical second-order ratelaw, for which complementary experimental9c and compu-tational10 studies provided compelling evidence for the operationof a proton-triggered non-insertive cyclisation mechanism. DFTpredicts a kinetic gap between the relevant proton-triggered non-insertive cyclisation and aminolysis steps that diminishesfrom 9.6 kcal mol−1 for {CpoM}Y(NHR)-catalysed HA to5.0 kcal mol−1 for {ToM}Mg(NHR)-catalysed8 HA of amino-alkene 1. It can be viewed as a manifestation of apparent differ-ence in spatial demands imposed on the involved TS structuresand in requirements to access the central element.

With regard to possible strategies for the development of moreeffective rare earth HA catalysts, several aspects arise from theanalysis presented thus far. Firstly, a non-insertive pathway prob-ably prevails for ligand architectures that are designed to slowdown the σ-insertive cyclisation step through protection of therare earth against olefin approach and concomitantly ensuringthat a metal-bound amine α-H is sufficiently acidic. However,such tailor-made catalysts are likely to display a diminished per-formance. Secondly, systems featuring a proper balance betweenM–amido bond polarity, rare earth capacity to polarise the olefinand amine α-H acidity, together with a sufficiently accessiblerare earth, thus favouring an σ-insertive pathway, offer an avenuetowards highly active HA catalysts.

Conclusion

The present study aims to contribute towards a more detailedmechanistic understanding of the rare earth-mediated HA ofaminoalkenes. It employs DFT as an established and predictivemethod to scrutinise rival mechanistic pathways for intramolecu-lar HA of prototype 2,2-dimethyl-4-penten-1-amine 1 by ahighly active [{CpoM}YCH2SiMe3] precatalyst11 2 (CpoM =bis(4,4-dimethyl-2-oxazolinyl)(cyclopentadienyl)phenylborate).Two plausible mechanistic scenarios are consistent with theobserved distinct process features, which include a second-orderrate law, significant primary KIEs and marked isotope effects onenantioselectivity (observed for a chiral analogue 2*). Firstly, aconcerted amidoalkene → cycloamine conversion through N–Cring closure concurrent with amino proton delivery at the adja-cent olefin–C centre and secondly, a stepwise σ-insertivepathway that involves a fast and reversible migratory olefin inser-tion into the Y–N amido σ-bond linked to irreversible and lessrapid Y–C bond protonolysis. A recent experimental studyfavours a concerted C–N and C–H bond formation.11 The com-putational examination presented herein reveals non-competitivekinetic demands for a pathway describing N–C ring closure trig-gered by concurrent amino proton transfer at the CvC linkageevolving through a six-centre TS structure, which militatesagainst the operation of a non-insertive cyclisation pathway. DFTfavours a stepwise mechanism as the energetically preferredpathway that entails: (1) a highly facile and reversible 3′ ⇆ 4σ-insertive cyclisation at a Y(NHR) catalyst species 3′ with theequilibrium is shifted towards the reactant; 3′ corresponds to thecatalyst resting state and is in rapid equilibrium with its

substrate-adducted species 3·S and 3·(S);2 (2) subsequent irre-versible and turnover-limiting 4 + S → 5 Y–C bond aminolysisto be followed by (3) rapid and exergonic 5 → 3′ + P liberationof the cycloamine product to regenerate the active catalystspecies. The effective barrier assessed by DFT matches theempirically determined Eyring parameter well11b and the compu-tationally estimated primary KIE is close to observed values. Arecent DFT study revealed a similar scenario for an analogoustris(oxazolinyl)borate magnesium system.8 The {CpoM}-Y(NHR)- and {ToM}Mg(NHR)-catalysed hydroaminationsexhibit similar features, including large primary KIEs andempirical formal second-order rate laws. A comparison of theseprocesses revealed valuable insights into catalytic structure–reactivity relationships that will likely direct the rational designof rare earth-based catalysts.

Computational methodology

All calculations based on Kohn–Sham density functional theory(DFT)21 were performed by means of the program packageTURBOMOLE22 using the almost nonempirical meta-GGATao–Perdew–Staroverov–Scuseria (TPSS) functional23 withinthe RI-J integral approximation24 in conjunction with flexiblebasis sets of triple-ζ quality. For Y we used the Stuttgart–Dresden scalar-relativistic pseudopotential (28MWB, 28 coreelectrons)25 in combination with the associate (7s7p5d1f)/[6s4p3d1f] valence basis set.26 All remaining elements were rep-resented by Ahlrich’s valence triple-ζ TZVP basis set27 withpolarization functions on all atoms. The good to excellent per-formance of the TPSS functional for a wide range of appli-cations,28 and alkaline earth8 and rare earth-catalysed HA3d,g inparticular, has been demonstrated previously. The free-energylandscape of the entire HA course was assessed for 2,2-dimethyl-4-penten-1-amine (S) together with [{CpoM}-Y(NHR)·(NH2R)n] as the catalyst complex. No structural sim-plification of any of the key species involved was imposed. TheDFT calculations have simulated the authentic reaction con-ditions by treating the bulk effects of the benzene solvent by aconsistent continuum model in the form of the conductor-likescreening model (COSMO).29 All the stationary points werefully located at the TPSS level with inclusion of solvation. Fre-quency calculations were performed to confirm the nature of alloptimised key structures and to determine thermodynamic par-ameters (298 K, 1 atm) under the rigid rotor and harmonic oscil-lator approximations. The mechanistic conclusions drawn in thisstudy were based on the Gibbs free-energy profile of the entirecatalytic cycle assessed at the TPSS(COSMO)/SDD + TZVPlevel of approximation for experimental condensed phase con-ditions. Further details are given in the ESI.† Calculated struc-tures were visualised by employing the StrukEd program,30

which was also used for the preparation of 3D moleculedrawings.

Notes and references

1 For reviews of catalytic hydroamination, see: (a) L. S. Hegedus, Angew.Chem., 1988, 100, 1147, (Angew. Chem., Int. Ed., 1988, 27, 1113);(b) D. M. Roundhill, Catal. Today, 1997, 37, 155; (c) T. E. Müller andM. Beller, Chem. Rev., 1988, 98, 675; (d) M. Nobis and B. Drießen-

9190 | Dalton Trans., 2012, 41, 9182–9191 This journal is © The Royal Society of Chemistry 2012

Publ

ishe

d on

19

June

201

2. D

ownl

oade

d by

Uni

vers

ity o

f C

alif

orni

a -

Sant

a C

ruz

on 2

2/10

/201

4 12

:56:

46.

View Article Online

Hölscher, Angew. Chem., 2001, 113, 4105, (Angew. Chem., Int. Ed.,2001, 40, 3983); (e) R. Taube, in Applied Homogeneous Catalysis withOrganometallic Complexes, ed. B. Cornils and W. A. Herrmann, Wiley-VCH, Weinheim, 2002, pp. 513–524; (f ) F. Seayad, A. Tillack,C. G. Hartung and M. Beller, Adv. Synth. Catal., 2002, 344, 795;(g) G. A. Molander and J. A. C. Romero, Chem. Rev., 2002, 102, 2161;(h) F. Pohlki and S. Doye, Chem. Soc. Rev., 2003, 32, 104;(i) P. W. Roesky and T. E. Müller, Angew. Chem., 2003, 115, 2812,(Angew. Chem., Int. Ed., 2003, 42, 2708); ( j) J. F. Hartwig, Pure Appl.Chem., 2004, 76, 507; (k) S. Hong and T. J. Marks, Acc. Chem. Res.,2004, 37, 673; (l) K. C. Hultzsch, Adv. Synth. Catal., 2005, 347, 367;(m) A. L. Odom, Dalton Trans., 2005, 225; (n) R. Severin and S. Doye,Chem. Soc. Rev., 2007, 36, 1407; (o) I. Aillaud, J. Collin,J. Hannedouche and E. Schulz, Dalton Trans., 2007, 5105;(p) J. F. Hartwig, Nature, 2008, 455, 314; (q) T. E. Müller,K. C. Hultzsch, M. Yus, F. Foubelo and M. Tada, Chem. Rev., 2008, 108,3795; (r) A. L. Reznichenko and K. C. Hultzsch, Top. Organmet. Chem.,2012, in press

2 For rare earth-catalysed intramolecular hydroamination, see:(a) M. R. Gagné and T. J. Marks, J. Am. Chem. Soc., 1989, 111, 4108;(b) M. R. Gagné and T. J. Marks, J. Am. Chem. Soc., 1992, 114, 275;(c) Y. Li, P.-F. Fu and T. J. Marks, Organometallics, 1994, 13, 439;(d) Y. Li and T. J. Marks, J. Am. Chem. Soc., 1996, 118, 9295;(e) G. A. Molander and E. D. Dowdy, J. Org. Chem., 1998, 63, 8983;(f ) M. R. Bürgstein, H. Berberich and P. W. Roesky, Organometallics,1998, 17, 1452; (g) A. T. Gilbert, B. L. Davis, T. J. Emge andR. D. Broene, Organometallics, 1999, 18, 2125; (h) G. A. Molander andE. D. Dowdy, J. Org. Chem., 1999, 64, 6515; (i) Y. K. Kim,T. Livinghouse and J. E. Bercaw, Tetrahedron Lett., 2001, 42, 2933;( j) G. A. Molander, E. D. Dowdy and S. K. Pack, J. Org. Chem., 2001,66, 4344; (k) M. R. Bürgstein, H. Berberich and P. W. Roesky, Chem.–Eur. J., 2001, 7, 7078; (l) Y. K. Kim and T. Livinghouse, Angew. Chem.,2002, 114, 3797, (Angew. Chem., Int. Ed., 2002, 41, 3645); (m) S. Hongand T. J. Marks, J. Am. Chem. Soc., 2002, 124, 7886; (n) S. Hong,S. Tian, M. V. Metz and T. J. Marks, J. Am. Chem. Soc., 2003, 125,14768; (o) P. N. O’Shaughnessy, P. D. Knight, C. Morton and K. Horino,J. Am. Chem. Soc., 2003, 125, 9560; (p) F. Lauterwasser, P. G. Hayes,St. Bräse, W. E. Piers and L. L. Schafer, Organometallics, 2004, 23,2234; (q) A. Zulys, T. K. Panda, M. T. Gamer and P. W. Roesky, Chem.Commun., 2004, 2584; (r) J.-S. Ryu, T. J. Marks and F. E. McDonald,J. Org. Chem., 2004, 69, 1028; (s) A. M. Seyam, B. D. Stubbert,T. R. Jensen, J. J. O’Donnell III, C. L. Stern and T. J. Marks, Inorg.Chim. Acta, 2004, 357, 4029; (t) J. Collin, J.-C. Daran, O. Jacquet,E. Schulz and A. Trifonov, Chem.–Eur. J., 2005, 11, 3455;(u) D. V. Gribkov, K. C. Hultzsch and F. Hampel, J. Am. Chem. Soc.,2006, 128, 3748; (v) S. Bambirra, H. Tsurugi, D. van Leusen andB. Hessen, Dalton Trans., 2006, 1157; (w) S. Datta, M. T. Garner andP. W. Roesky, Organometallics, 2008, 27, 1207; (x) L. J. E. Stanlake andL. L. Schafer, Organometallics, 2009, 28, 3990.

3 For computational studies of organolanthanide-assisted intramolecularHA of various substrate classes, see: aminoalkene substrates;(a) A. Motta, G. Lanza, I. L. Fragala and T. J. Marks, Organometallics,2004, 23, 4097; aminoalkyne substrates (b) A. Motta, G. Lanza,I. L. Fragala and T. J. Marks, Organometallics, 2006, 25, 5533; amino-allene substrate (c) S. Tobisch, Chem.–Eur. J., 2006, 12, 2520;(d) S. Tobisch, Dalton Trans., 2011, 40, 249; aminodiene substrates(e) S. Tobisch, J. Am. Chem. Soc., 2005, 127, 11979; (f ) S. Tobisch,Chem.–Eur. J., 2007, 13, 9127; (g) S. Tobisch, Chem.–Eur. J., 2010, 16,13814.

4 B. D. Stubbert and T. J. Marks, J. Am. Chem. Soc., 2007, 129, 6149.5 L. E. N. Allan, D. J. Fox, R. J. Deeth, A. L. Gott and P. Scott, Proceed-ings of the 42nd IUPAC Congress, Glasgow, 2–7 August 2009.

6 N. K. Hangaly, A. R. Petrov, K. A. Rufanov, K. Harms, M. Elfferdingand J. Sundermeyer, Organometallics, 2011, 30, 4544.

7 (a) J. F. Dunne, D. B. Fulton, A. Ellern and A. D. Sadow, J. Am. Chem.Soc., 2010, 132, 17680; (b) M. Arrowsmith, M. R. Crimmin,A. G. M. Barrett, M. S. Hill, G. Kociok-Köhn and P. A. Procopiou, Orga-nometallics, 2011, 30, 1493; (c) X. Zhang, T. J. Emge andK. C. Hultzsch, Angew. Chem., Int. Ed., 2012, 51, 394.

8 S. Tobisch, Chem.–Eur. J., 2011, 17, 14974.9 (a) L. E. N. Allan, G. J. Clarkson, D. J. Fox, A. L. Gott and P. Scott,J. Am. Chem. Soc., 2010, 132, 15308; (b) K. Manna, S. Xu andA. D. Sadow, Angew. Chem., Int. Ed., 2011, 50, 1865; (c) D. C. Leitch,R. H. Platel and L. L. Schafer, J. Am. Chem. Soc., 2011, 133, 15443.

10 S. Tobisch, Inorg. Chem., 2012, 51, 3786.

11 (a) K. Manna, M. L. Kruse and A. D. Sadow, ACS Catal., 2011, 1, 1637;(b) Eyring analysis for the cyclisation of 2,2-cyclohexyl substituted4-aminoalkene S′ by chiral 2* (Scheme 2) afforded ΔH‡ = 7.5 kcal mol−1

and ΔS‡ = −38 cal K mol−1, hence ΔG‡ = 18.8 kcal mol−1 (298 K).Unfortunately, Eyring data have not been reported for the cyclisation of2,2-dimethyl substituted 4-aminoalkene S by achiral 2

12 A. L. Reznichenko and K. C. Hultzsch, J. Am. Chem. Soc., 2012, 134, 3300.13 P. A. Hunt, Dalton Trans., 2007, 1743.14 S. Tobisch, Chem.–Eur. J., 2010, 16, 3441.15 (a) Amine association and dissociation at the various intermediates is pre-

sumably facile and rapid equilibria are expected (b) Examination bya linear-transit approach gave no indication that such processes are associ-ated with a significant enthalpic barrier

16 A {CpoM}Y ligation through bonding interactions between the yttriumcentre and two N(oxazoline) donor centres and a multihapto bound cyclo-pentadienyl unit, as established in starting material 2, is denoted as (κ2-Ox2)Y throughout this study. The notations (κ1-Ox2)Y and (κ0-Ox2)Y areused to designate species with one or two dissociated oxazoline rings,respectively

17 (a) The [{CpoM}YCH2SiMe3] compound 2 (with the appropriate numberof substrate molecules) has been chosen as a reference for relative freeenergies (kcal mol−1) (b) The prevalent {CpoM}Y(NHR) form 3′, featur-ing a chelating amidoalkene unit, of the catalytically relevant catalystcomplex (with the appropriate number of substrate or cycloamine mole-cules) has been chosen as reference for relative free energies (kcal mol−1)

18 The methyl and phenyl groups of the {CpoM}Y ligand backbone are dis-played in a truncated fashion and also amino-/amidoalkene moieties

19 It is worth noting that DFT assessed total turnover-limiting barriers matchexperimental values well, by approximately 2 kcal mol−1 for aminoalkeneHA by tris(oxazolinyl)borate magnesium (see ref. 7a, 8) and bis(ureate)zirconium (see ref. 9c, 10) systems

20 (a) The primary KIE has been estimated from the deviation of zero-point-vibrational energies and partition function values for 1 and 1-d2 sub-strates (b) Zero curvature tunneling has been accounted for by the Skodjeand Truhlar approximation (ref. 20c). See the ESI† (c) R. T. Skodje, D.G. Truhlar and B. C. Garrett, J. Phys. Chem., 1981, 85, 3019; (d) Theaccuracy of computationally predicted kH/kD values for hydrogen transferprocesses is limited by transition-state theory approximation (ref. 20e),semiclassical tunnelling corrections and rigid rotator harmonic oscillatorapproximation for vibrational frequencies (e) D. H. Lu, D. Maurice andD. G. Truhlar, J. Am. Chem. Soc., 1990, 112, 6206.

21 R. G. Parr and W. Yang, Density-Functional Theory of Atoms and Mole-cules, Oxford University Press, 1989.

22 (a) R. Ahlrichs, M. Bär, M. Häser, H. Horn and C. Kölmel, Chem. Phys.Lett., 1989, 162, 165; (b) O. Treutler and R. Ahlrichs, J. Chem. Phys.,1995, 102, 346; (c) R. Ahlrichs, F. Furche, C. Hättig, W. Klopper,M. Sierka and F. Weigend, TURBOMOLE, version 6.0, University ofKarlsruhe, Karlsruhe, Germany, 2009, http://www.turbomole.com

23 (a) P. A. M. Dirac, Proc. R. Soc. London, Ser. A, 1929, 123, 714;(b) J. C. Slater, Phys. Rev., 1951, 81, 385; (c) J. P. Perdew andY. Wang, Phys. Rev., 1992, B45, 13244; (d) J. Tao, J. P. Perdew,V. N. Staroverov and G. E. Scuseria, Phys. Rev. Lett., 2003, 91,146401; (e) J. P. Perdew, J. Tao, V. N. Staroverov and G. E. Scuseria,J. Chem. Phys., 2004, 120, 6898.

24 (a) O. Vahtras, J. Almlöf and M. W. Feyereisen, Chem. Phys. Lett., 1993,213, 514; (b) K. Eichkorn, O. Treutler, H. Öhm, M. Häser andR. Ahlrichs, Chem. Phys. Lett., 1995, 242, 652.

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

26 (a) F. Weigend and R. Ahlrichs, Phys. Chem. Chem. Phys., 2005, 7,3297; (b) F. Weigend, Phys. Chem. Chem. Phys., 2006, 8, 1057.

27 (a) A. Schäfer, C. Huber and R. Ahlrichs, J. Chem. Phys., 1994, 100,5829; (b) K. Eichkorn, F. Weigend, O. Treutler and R. Ahlrichs, Theor.Chem. Acc., 1997, 97, 119.

28 (a) V. N. Staroveroc, G. E. Scuseria, J. Tao and J. P. Perdew, J. Chem.Phys., 2003, 119, 12129; (b) I. Hyla-Kryspin and S. Grimme, Organome-tallics, 2004, 23, 5581; (c) Y. Zhao and D. G. Truhlar, J. Chem. TheoryComput., 2005, 1, 415; (d) F. Furche and J. P. Perdew, J. Chem. Phys.,2006, 124, 044103; (e) F. Neese, T. Schwabe and S. Grimme, J. Chem.Phys., 2007, 126, 124115; (f ) M. Bühl, C. Reimann, D. A. Pantazis,T. Bredow and F. Neese, J. Chem. Theor. Comput., 2008, 4, 1449.

29 (a) A. Klamt and G. Schüürmann, J. Chem. Soc., Perkin Trans. 2, 1993,799; (b) A. Klamt, in Encyclopaedia of Computational Chemistry, Vol. 1,ed. P. v. R. Schleyer, John Wiley & Sons, Chichester, 1998, pp. 604–615.

30 For further details, see http://www.struked.de

This journal is © The Royal Society of Chemistry 2012 Dalton Trans., 2012, 41, 9182–9191 | 9191

Publ

ishe

d on

19

June

201

2. D

ownl

oade

d by

Uni

vers

ity o

f C

alif

orni

a -

Sant

a C

ruz

on 2

2/10

/201

4 12

:56:

46.

View Article Online