A DFT Study on the Mechanism of the Cycloaddition Reaction of CO 2 to Epoxides Catalyzed by Zn(Salphen) Complexes

DOI: 10.1002/chem.201203985

A DFT Study on the Mechanism of the Cycloaddition Reaction of CO2 toEpoxides Catalyzed by ZnACHTUNGTRENNUNG(Salphen) Complexes

Fernando Castro-G�mez,[a] Giovanni Salassa,[a] Arjan W. Kleij,*[a, b] and Carles Bo*[a, c]

Introduction

The growing concerns about climate change and the need tofind suitable alternatives for our depleting fossil-fuel-basedfeed stocks have resulted in extensive research to find newrenewable carbon sources. Carbon dioxide (CO2) can beconsidered as an abundant and renewable carbon source. Inlight of this, it has become a popular building block from asustainability point of view[1] and can thus be seen as an in-teresting starting point for the synthesis of various organicmolecules.[2] One reaction that utilizes CO2 as a carbonsource and which is currently attracting a great deal of inter-est, is the atom-efficient cycloaddition of CO2 to epoxides,resulting in useful cyclic carbonate products (Scheme 1).[3]

The conventional industrial way of organic carbonate forma-tion uses phosgene as a (toxic) reagent and results in haz-ardous waste streams.[4] Therefore, new and greener method-ologies able to mediate this transformation have become in-

creasingly interesting. Cyclic carbonates are employed aspolar aprotic solvents and electrolytes in rechargeable bat-teries, intermediates for organic polycarbonate synthesis,and the production of pharmaceuticals and fine chemi-cals.[1,5] The cycloaddition reaction has been widely studiedand can be mediated by a variety of catalysts, such as qua-ternary ammonium salts,[6] alkali metal halides,[7] ionic liq-uids,[8] functional polymers,[9] and transition-metal com-plexes.[10] However, there are still disadvantages to over-come upon using these catalyst systems, such as low catalystreactivity/stability, use of high pressures and/or tempera-tures, high catalyst loadings required for efficient turnover,toxicity issues, cost effectiveness, availability of the catalystsystem, or a combination of these features. Hence, it is cru-cial to know how the performance of any given catalyst canbe improved, and mechanistic understanding provides ameans to unravel the obstacles associated with the observa-tion of low reactivity and/or selectivity.

To date, a limited series of theoretical studies on themechanism of the catalytic cycloaddition of CO2 to epoxideshave been reported involving either heterobimetallic Ru–Mn complexes,[11] ionic liquids,[12] N-heterocyclic carbenes,[13]

polyoxometalates,[14] polyphenolic compounds,[15] or carbox-ylate-tagged biopolymers.[16] Zhang and co-workers[17] havethoroughly elucidated the mechanism of conversion of ethyl-ene oxide with CO2 catalyzed by quaternary ammonium

Abstract: The reaction mechanism forthe ZnACHTUNGTRENNUNG(salphen)/NBu4X (X= Br, I)mediated cycloaddition of CO2 to aseries of epoxides, affording five-mem-bered cyclic carbonate products hasbeen investigated in detail by usingDFT methods. The ring-opening stepof the process was examined and thepreference for opening at the methyl-ene (Cb) or methine carbon (Ca) was

established. Furthermore, calculationswere performed to clarify the reasonsfor the lethargic behavior of internalepoxides in the presence of the binarycatalyst. Also, the CO2 insertion and

the ring-closing steps have been ex-plored for six differently substitutedepoxides and proved to be significantlymore challenging compared with thering-opening step. The computationalfindings should allow the design andapplication of more efficient catalystsfor organic carbonate formation.

Keywords: carbon dioxide · cyclo-addition · density functional calcu-lations · epoxides · salen complexes

[a] F. Castro-G�mez, G. Salassa, Prof. Dr. A. W. Kleij, Prof. Dr. C. BoInstitute of Chemical Research of Catalonia (ICIQ)Av. Pa�sos Catalans 16, 43007, Tarragona (Spain)Fax: (+34) 977920828E-mail : [email protected]

[email protected]

[b] Prof. Dr. A. W. KleijCatalan Institute for Research andAdvanced Studies (ICREA)Pg. Lluis Companys 2308010 Barcelona (Spain)

[c] Prof. Dr. C. BoDepartament de Qu�mica F�sica i Inorg�nicaUniversitat Rovira i Virgili, Marcel·l� Domingos/n. 43007, Tarragona (Spain)

Supporting information for this article is available on the WWWunder http://dx.doi.org/10.1002/chem.201203985.

Scheme 1. The cycloaddition reaction of CO2 with epoxides generatingfive-membered cyclic carbonates using a Lewis acidic metal catalyst(Cat) and a co-catalyst (Nu; nucleophile).

Chem. Eur. J. 2013, 19, 6289 – 6298 � 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim 6289

FULL PAPER

salts. In these studies, they obtained structural and energeticinformation concerning each step of the catalytic cycle andalso evaluated the effect of the N-alkyl chain length and thetype of anion. In addition, Han and co-workers[18] studied indetail the KI-catalyzed cycloaddition of CO2 to propyleneoxide, and also investigated the (co)catalytic role of glyceroland propylene glycol.

We recently reported on the use of ZnACHTUNGTRENNUNG(salphen) com-plexes (salphen=N,N�-bis(salicylidene)-1,2-phenylenedi-ACHTUNGTRENNUNGamine, see Scheme 2 a) complexes in conjunction with

NBu4I as efficient binary catalysts for the formation ofcyclic carbonates under mild reaction conditions (pCO2 = 2–10 bar; T=25–45 8C).[19] From these previous studies it hasbecome clear that terminal epoxides are conveniently con-verted into their carbonates in high yield, whereas the samebinary catalyst system proved to be rather ineffective for theconversion of internal and more sterically congested sub-strates. At a later stage, we found that more efficient cataly-sis of these latter substrates can be achieved when workingunder solvent-free, supercritical CO2 conditions,[20] givinghighly improved conversion levels in much shorter timeframes. As reported extensively in literature, there are a va-riety of well-defined binary catalytic systems based on met-allosalens[21] and halide salts that provide efficient CO2/ep-oxide cycloaddition catalysts under mild reaction conditions.Thus, the metallosalen family of catalysts could be regardedas privileged systems in the context of (cyclic) organic car-bonate formation. Though various mechanistic proposalshave been put forward throughout the years, and in somecases verified by experimental data,[22] there is surprisinglylimited information available regarding the major obstaclesassociated with the use of relatively more challenging sub-strates including those based on sterically congested oxir-anes and internal epoxides. This in combination with ourprevious findings[19, 20] using ZnACHTUNGTRENNUNG(salphen)s as catalyst systems,prompted us to investigate these more challenging conver-sions in more detail by using computational methods.

Herein, we report a detailed DFT study on the mecha-nism of the cycloaddition reaction with a series of epoxides(Scheme 2b, compounds 1–6) catalyzed by the binary systemZnACHTUNGTRENNUNG(salphen)/NBu4X (X= Br, I). We have calculated the fullenergy profiles for various substrates and particularly fo-

cused on those aspects that help to explain the lower reac-tivities found for the more challenging substrates, providingnew insights that are potentially useful for the developmentof more powerful catalyst systems.

Results and Discussion

The uncatalyzed cycloaddition reaction : The cycloadditionof CO2 to propylene oxide 1 can be achieved through asingle step, leading to the formation of propylene carbonate(PC). The reaction in the absence of catalyst proceedsthrough nucleophilic attack from an oxygen atom of CO2 onthe a carbon (most substituted carbon) or the b carbon(least substituted carbon) atom of 1. The two calculated re-action pathways are shown in Figure 1, together with the op-

timized structures of the two transition-states (TS). Theunique imaginary vibrational frequencies of TS-a and TS-bcorrespond to the simultaneous breaking of the Ca

�O orCb�O bond of the epoxide and the simultaneous formation

of two new C�O bonds that originate from the insertion ofCO2. High energy-barriers of 48.6 (a pathway) and 53.7 kcalmol�1 (b pathway) are involved to form the cyclic carbonateproduct. The a pathway is favored by 5.1 kcal mol�1 com-pared with the b pathway. Previous DFT calculations report-ed by Zhang and co-workers[12a] showed somewhat higherenergy-barriers of 53.4 kcal mol�1 for the a pathway and58.1 kcal mol�1 for the b pathway. Similar to the work fromZhang, Han and co-workers[18] reported energy barriers of55.3 and 61.4 kcal mol�1 for the a- and b pathways, respec-tively.

Scheme 2. a) Schematic drawing of the ZnACHTUNGTRENNUNG(salphen) catalyst ; b) substrates1–6 used in this investigation.

Figure 1. Potential energy profile for the uncatalyzed cycloaddition ofCO2 with propylene oxide 1 to give propylene carbonate (PC). IC standsfor initial complex formation. Note that potential energy is here used tomake a possible comparison with the literature data.

www.chemeurj.org � 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Chem. Eur. J. 2013, 19, 6289 – 62986290

The Zn ACHTUNGTRENNUNG(salphen) catalyzed cycloaddition reaction : Similarto the uncatalyzed reaction, the cycloaddition reaction ofCO2 to propylene oxide 1 catalyzed by the Zn ACHTUNGTRENNUNG(salphen)complex of Scheme 2 can involve two possible reactionpathways. The relative free-energy profile of the a and b

pathways are depicted in Figure 2, in which the sum of ener-

gies of the isolated reactants (1 + CO2) and ZnACHTUNGTRENNUNG(salphen)were set to zero. The initial step is the coordination of theepoxide to the Zn ACHTUNGTRENNUNG(salphen) complex, which forms an initialcomplex (IC), followed by a concerted ring-opening andCO2 insertion step. For both pathways the only transition-state relates to the simultaneous stretching/breaking of theCa�O or Cb

�O bond of the epoxide, and the bending of theCO2 molecule leading to the formation of two new C�Obonds.

Once the coordinated cyclic carbonate is formed (repre-sented by FC, which is the intermediate complex having thecarbonate product coordinated) it is released from the Zn-ACHTUNGTRENNUNG(salphen) complex allowing for further epoxide turnover.The calculated barriers for the exothermic reaction showthat the a pathway is favored by 15.2 kcal mol�1 comparedwith the b pathway. The relative potential energy barrier forthe Zn ACHTUNGTRENNUNG(salphen)-catalyzed cycloaddition (Figure 2, energyvalues in brackets) is reduced by 19.9 kcal mol�1 comparedwith the uncatalyzed reaction, which is likely as a result ofthe formation of a ZnACHTUNGTRENNUNG(salphen)/epoxide complex polarizingthe C�O bond of the substrate and thus playing an impor-tant role in its activation.

The Zn ACHTUNGTRENNUNG(salphen)/NBu4I-catalyzed cycloaddition reaction : Inthe previous discussion, the investigation of the uncatalyzed

and the Zn ACHTUNGTRENNUNG(salphen)-catalyzed mechanisms highlight thatthe ring-opening step is initiated by CO2 leading to thedirect formation of the cyclic carbonate. The high energy-barriers found for this ring-opening process suggest thatthese reactions can thus only occur under harsh reactionconditions, that is, high temperatures and/or pressures. Asreported extensively in the literature,[21] generally a binarycatalytic system is needed to obtain good conversions/yieldsunder mild reaction conditions. The binary catalytic systemusually combines a Lewis acid and a suitable nucleophile(most often a halide), which make the ring-opening proce-dure less energetically demanding and the subsequent CO2

insertion easier. It should be noted that the nucleophilesthemselves are able to catalyze the CO2 addition to epox-ACHTUNGTRENNUNGides.[6a,15,17, 18] The most relevant result to the present study isthe KI-catalyzed ring-opening of propylene oxide reportedby Han,[18] which involves gas phase barriers in the range of36.3–37.6 kcal mol�1 depending on the reaction pathway. Werecently found a fairly similar barrier (38.9 kcal mol�1) whenusing NBu4I as a catalyst.[15a] Thus, halide nucleophiles areable to significantly lower the transition state related to thering-opening of the epoxide (cf., values reported for the un-catalyzed reaction in Figure 1).

ZnACHTUNGTRENNUNG(salphen) complexes have recently been shown to effi-ciently catalyze the cycloaddition reaction under mild reac-tion conditions, by using quaternary ammonium halide saltsas nucleophiles.[19,20] Next we evaluated computationally theeffect of combining both the ZnACHTUNGTRENNUNG(salphen) as well as halidenucleophile (NBu4X; X=Br, I) in the DFT analysis. Uponevaluation of this binary system, the potential for two differ-ent catalytic pathways (a or b attack on the epoxide) wasidentified similar to the uncatalyzed and Zn ACHTUNGTRENNUNG(salphen)-cata-lyzed cycloaddition reaction (Figure 1 and 2). DFT calcula-tions were performed in the first instance on the simplestsubstrate, propylene oxide 1.

It should be noted that for the envisioned mechanisms, wehave only considered penta-coordinated rather than hexa-coordinated reaction intermediates. Hexa-coordinated spe-cies have been frequently observed and proposed for othermetallosalen catalyst comprising of Al,[23] Mn,[24] Co,[25] andCr[26] metal ions, and in some of these cases (preliminary)mechanistic work has revealed that bimetallic pathways leadto more efficient catalytic processes. However, in the case ofZnACHTUNGTRENNUNG(salphen)s, hexa-coordination is an extremely rare phe-nomenon and is only observed in condensed solid phases.[27]

Furthermore, we[28] and others[29] have clearly demonstratedthat penta-coordination in ZnACHTUNGTRENNUNG(salphen)s is highly favoredboth in the solid state as well as in solution phases as sup-ported by crystallographic evidence, UV/Vis titration data,Job plot analyses and fitting 1:1 complex–ligand bindingmodels by using multivariate data analysis.[30] Therefore, theconsideration of penta-coordination in the ZnACHTUNGTRENNUNG(salphen) caseseems to give a reasonable starting point in the mechanismsdiscussed below.

The resulting energy profiles[31] of the a and b pathwaysare shown in Figures 3 and 4 with schematic representationsof the involved intermediates and transitions states; in both

Figure 2. Free-energy profile for the ZnACHTUNGTRENNUNG(salphen)-catalyzed cycloadditionof CO2 to propylene oxide 1; the relative potential energy values are re-ported in brackets. IC stands for initial complex formation; FC stands forthe intermediate complex having the carbonate product coordinated.

Chem. Eur. J. 2013, 19, 6289 – 6298 � 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chemeurj.org 6291

FULL PAPERCycloaddition Reaction of CO2 to Epoxides

cases, the first step involves the coordination of 1 (IC) tothe ZnACHTUNGTRENNUNG(salphen) which, as mentioned before, polarizes theC�O epoxide bond thereby facilitating the ring-openingstep. It should be noted that the higher free-energy for ICcompared with the separate components is mainly due to anentropic cost for bringing together the Zn ACHTUNGTRENNUNG(salphen) and theepoxide. The C�O-bond polarization is supported by chargeanalysis of the carbon atoms of the epoxide revealing thatboth Ca and Cb become more electron-deficient (see theSupporting Information) compared with a non-coordinatedepoxide. After initial coordination of the epoxide to the Zn-ACHTUNGTRENNUNG(salphen) complex (cf. , IC), the ring-opening step occursthrough nucleophilic attack of the iodide. The first transi-tion-state (TS1) is characterized by the breaking of the Ca/b�O bond and the simultaneous formation of a Ca/b�I bond asconfirmed by the unique imaginary frequency for the attackat Ca and for Cb (Figure 5). In the case of the b pathway, theepoxide ring-opening is energetically more favorable by

almost 2 kcal mol�1 compared with the a pathway. In thesubsequent step, a molecule of CO2 reacts with the negative-ly charged oxygen atom (insertion) of the intermediate Int-1leading to the formation of linear carbonate Int-2. Thesecond transition state TS2 involves the formation of newC�O and Zn�O bonds that involve the CO2 molecule asshown in the relative imaginary frequencies at 58.3 i cm�1

for the a pathway and 51.7 i cm�1 for the b route (Figure 5).In the pathway related to the b attack, this step is rate-deter-mining with an energy value of 34.4 kcal mol�1, whereas inthe a pathway this step has an energy of 30.9 kcal mol�1 andis thus more facile. The reason for this difference betweenthe two pathways has mainly a steric origin; the position ofthe methyl group in the b pathway is closer to the O atomof the epoxide making the insertion of CO2 more difficult(Figure 5).

The linear carbonate intermediate Int-2 undergoes an in-tramolecular ring-closing with the concomitant release of

Figure 3. Free-energy profile for the cycloaddition of CO2 to propylene oxide 1 catalyzed by ZnACHTUNGTRENNUNG(salphen)/iodide considering the a pathway. IC stands forinitial complex formation; FC stands for the intermediate complex having the carbonate product coordinated.

Figure 4. Free-energy profile for the cycloaddition of CO2 to propylene oxide 1 catalyzed by Zn ACHTUNGTRENNUNG(salphen)/iodide considering the b pathway. IC stands forinitial complex formation; FC stands for the intermediate complex having the carbonate product coordinated.

www.chemeurj.org � 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Chem. Eur. J. 2013, 19, 6289 – 62986292

A. W. Kleij, C. Bo et al.

the iodide nucleophile and formation of the FC. In thislatter step, the carbon atom bound to the iodide binds tothe nearest oxygen atom forming a new C�O bond; at thesame time, the iodide bond elongates until it breaks(Figure 5). The ring-closing step is found to be the rate-de-termining step in the a pathway with an energy of 35.4 kcalmol�1. Similar to the CO2 insertion step, the methyl groupon the Ca makes the ring-closing step more difficult in the a

pathway, whereas in the b pathway the intermediate linearcarbonate is in a more suitable conformation for the ring-closing event.

Once formed, the cyclic carbonate is released from theZnACHTUNGTRENNUNG(salphen) complex allowing further epoxide turnover.The overall reaction is exergonic with a release of 5.0 kcalmol�1. The synergistic effect of the binary ZnACHTUNGTRENNUNG(salphen)/

NBu4I catalyst system makes the synthesis of the cyclic car-bonate much more accessible. In particular, the high Lewisacidity of the ZnACHTUNGTRENNUNG(salphen) complex[32] strongly reduces theenergy barrier for the ring-opening step through a polariza-tion of the epoxide carbon atoms (48.6 kcal mol�1 for the un-catalyzed reaction to 7.03 kcal mol�1 for the binary catalyst;both energy values refer to potential energy, see the Sup-porting Information), allowing the nucleophile to attackmore easily. As shown in the energetic profiles in Figures 3and 4, there are some important differences between the a

and b pathway using propylene oxide 1 as substrate. First, inthe ring-opening step, in contrast with the non-catalyzed re-action, the attack at the b carbon is favored. Second, therate-determining steps of the a and b pathways are at differ-ent points during the catalytic cycle; the ring-closing step inthe a pathway and the CO2-insertion step in the b pathway.This is a consequence of the different position in which themethyl substituent is located during the two possible path-ways.

Ring-closing mechanism : As already shown in Figures 3 and4, the ring-closing step involves the formation of a new C�O

bond in both pathways, but when considering the linear in-termediate Int-2 there are two possible routes leading to thecyclic carbonate product through either O-atom of the car-bonate (Figure 6 a, O1 and O2). To establish which ring-clos-ing mechanism is more preferred, the energy profiles forboth possible intramolecular nucleophilic pathways werecalculated. It can be seen in Figure 6 a that when ring-clos-ing occurs through the non-coordinating O-atom O2 (RC1)only one step is required evolving through TS3 giving thefinal intermediate FC. When the nucleophilic attack is madeby O1 (i.e., through pathway RC2), a multi-step mechanismwould be operative going first through TS3’, which, afterskeletal rearrangement, affords Int-3 followed by ring-clos-ing in TS4 giving FC’. The difference in product complexesFC and FC’ is related to which O atom of the former CO2

molecule coordinates to the Zn metal center before release.The comparison between the two pathways indicates that

RC1 is favored over RC2 as a result of the fewer steps re-quired and also the lower energy requirement for the rate-determining step (35.4 kcal mol�1 for RC1 compared to37.9 kcal mol�1 for RC2). The energy profiles for the twopossible ring-closing events in the a pathway are shown inFigure 6 b and a similar behavior is observed for the b path-way (see the Supporting Information).

The Zn ACHTUNGTRENNUNG(salphen)/NBu4Br-catalyzed cycloaddition reaction :From our initial results[19a] we observed that iodide provedto be a better co-catalyst than bromide giving higher conver-sion levels when using 1,2-epoxyhexane as substrate. There-fore, the effect of using bromide as the nucleophile insteadof iodide was evaluated for both the a and b pathwayduring the cycloaddition of CO2 to propylene oxide 1.

The first two entries of Table 1 show that the a pathway isless energetically demanding compared with the b pathwayand the rate-determining step in the latter is (as calculated

Figure 5. Optimized structures for the transition states TS1–TS3 for boththe a and b pathway together with the most relevant calculated distances(in �) and value of the negative (imaginary) vibrational frequenciesusing propylene oxide 1 as substrate.

Chem. Eur. J. 2013, 19, 6289 – 6298 � 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chemeurj.org 6293

FULL PAPERCycloaddition Reaction of CO2 to Epoxides

for the same pathway using iodide as nucleophile, Figure 4),the CO2 insertion step (36.5 kcal mol�1). In the ring-openingstep, bromide favors the nucleophilic attack at the b carbonby 2.6 kcal mol�1 compared with the a carbon. When com-paring both nucleophiles in the b pathway, the use of bro-mide (19.6 kcal mol�1) thus lowers the barrier for the ring-opening (TS1) of propylene oxide compared to iodide(22.4 kcal mol�1) but higher barriers are found for the other

two steps, that is, TS2 and TS3. Thus, these results are inline with the experimental data obtained for the binary cata-lyst ZnACHTUNGTRENNUNG(salphen)/NBu4X (X= Br, I) showing generally some-what higher activity when iodide is used as nucleophile, andin particular in the conversion of terminal epoxides throughthe b pathway that shows less dependency on the steric re-quirements of the catalyst.[19,20]

Effect of the epoxide substituent : So far the focus has beenon the cycloaddition of CO2 using propylene oxide 1 as sub-strate. To study the influence of the steric bulk/substitutionpattern in the epoxide, further substrates 2–6 (Scheme 2)were considered. When the steric bulk of the Ca substituentis increased (methyl!isopropyl!tert-butyl), the barriers re-lated to the transition states TS1–TS3 significantly increaseand thus reveal that the conversion of more sterically en-cumbered substrates is more difficult in line with our previ-ous experimental findings using ZnACHTUNGTRENNUNG(salphen)s as cata-lysts.[19, 20] Furthermore, for the ring-opening of the epoxides1–3, a clear increase in the difference in energy for TS1 canbe noted when comparing the a and b pathway in the series1!2!3 (Figure 7), and overall the b pathway becomesmore favorable. Similar to the calculated mechanism of 1(Figures 3 and 4), we observe that with epoxides 2 and 3 therate-limiting step does not change and remains the CO2 in-sertion for the b pathway and the ring-closing step for the a

pathway.As an example of an internal epoxide, the reaction mech-

anism of trans-2,3-epoxybutane 4 has also been calculated asexperimentally 4 has been shown to be a significantly lessreactive substrate[19] and as a result it was necessary toemploy much harsher reaction conditions (i.e., higher tem-

Figure 6. a) Two possible ring-closing pathways (RC1 and RC2) for the intermediate Int-2 involving different O atoms of the carbonate fragment (O1and O2) in the formation of propylene carbonate; b) Free-energy profile for the ring-closing step in the a pathway for the conversion of propylene oxide1 catalyzed by the binary ZnACHTUNGTRENNUNG(salphen)/NBu4I catalyst.

Table 1. Calculated free-energy values for the three transition-states inthe cycloaddition of CO2 to epoxides 1–6 at 25 8C.[a]

Substrate[b] Nucleophile TS1[c]

[kcal mol�1]TS2[c]

[kcal mol�1]TS3[c]

[kcal mol�1]

1a Br 22.2 33.5 33.91b Br 19.6 36.5 32.01a I 24.3 30.9 35.41b I 22.4 34.4 31.52a I 26.9 31.2 33.72b I 23.1 38.4 34.5

3a I 31.0 41.1 41.23b I 25.8 44.4 36.3

4 I 29.5 38.5 40.0

5a I 20.0 35.1 32.25b I 23.5 37.1 35.7

6a I 26.0 36.0 36.36b I 27.5 42.9 40.0

[a] Transition states: TS1 refers to the ring-opening of the epoxide, TS2to the CO2-insertion step, and TS3 to the energetically most favored ring-closing pathway. [b] The a and b designations relate to their respectivepathways. [c] The energies indicate a system in which the CO2 pressure is10 bar as previously used.[19]

www.chemeurj.org � 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Chem. Eur. J. 2013, 19, 6289 – 62986294

A. W. Kleij, C. Bo et al.

peratures and pressures) for successful conversion comparedwith terminal epoxides.[20] As reported in Table 1, all thetransition states of 4 have a relatively high energy comparedwith propylene oxide 1, and thus the presence of two methylsubstituents creates significant steric hindrance in the epox-ide to raise the energy barriers for all three steps. These bar-riers could be overcome by using supercritical CO2 (condi-tions used: 80 8C, pCO2 =80 bar) as reaction medium[20] al-lowing for good conversion of 4 by using Zn ACHTUNGTRENNUNG(salphen)/NBu4Br as a binary catalyst. From the computational datapresented thus far, it seems reasonable to assume that forepoxide substrates 1–4 steric effects are dominant and elec-tronic effects play a less important role.

Whereas most terminal epoxides are easily converted intotheir respective cyclic carbonates, in general substrates 5and 6 (Scheme 2) are slightly more sluggish and show lowerconversion levels with various catalyst systems. Therefore,the conversion of these substrates with the binary Zn-ACHTUNGTRENNUNG(salphen)/NBu4I catalyst was also investigated in detailusing DFT calculations (Table 1). In the case of 2-vinyloxi-ACHTUNGTRENNUNGrane 5 (Scheme 2), the presence of the vinyl functionalitymakes the ring-opening step more favorable on Ca unlikethe clear preference for Cb observed with epoxides 1–3. Thisis a result of the larger stabilization of TS1 through delocali-zation of the charge in the linear alkoxide through the vinylfragment; this is likely also the case in Int-1 making the co-ordinated alkoxide less nucleophilic. In line with the lowerreactivity is the higher barrier found for the CO2 insertionstep (TS2, 35.1 kcal mol�1) compared with 30.9 kcal mol�1

using 1 as substrate. Figure 8 a shows the difference in thefirst transition-states TS1 for both pathways using epoxide5. In the a pathway, the four carbon atoms are lying in thesame plane allowing p-conjugation and consequently higherstabilization of the system. In contrast with the trend seenfor substrates 1–3, for 5 the a pathway is more favorablecompared with the b pathway and the CO2 insertion step israte-limiting.

For styrene oxide 6 (Figure 8 b), the presence of thephenyl substituent also favors the a pathway, and in particu-lar the transition-state TS2 related to the CO2 insertion step

is markedly lower (36.0 kcal mol�1) compared with the b

pathway (42.6 kcal mol�1) but again higher than observedwith propylene oxide 1 (33.5 kcal mol�1). In the first transi-tion state TS1, similar to 5, a planar conformation allowsthe aromatic ring to stabilize the charge in the coordinatedalkoxide. In the transition state describing the ring-closingstep (see the Supporting Information, TS3-6a) again the in-ductive effect of the phenyl group is responsible for stabili-zation of the charge resulting from the release of iodide. Ex-perimental data with the binary ZnACHTUNGTRENNUNG(salphen)/NBu4I catalystsystem[19] are fully in line with these computational findingsfor substrates 5 and 6 and have shown that these latter epox-ides have lower conversion levels compared with other ter-minal epoxides such as propylene oxide 1. For 6 both TS2(CO2 insertion) and TS3 (ring-closing step) have similar(rate-determining) barriers of around 36 kcal mol�1.

The preference for ring-opening at Ca in both 5 as well as6 (i.e., the methine carbon) follows the observations fromDarensbourg, Lu, and co-workers who studied the CoIII-salen-catalyzed formation of polycarbonates based on CO2

and epichlorohydrin[33] or styrene carbonate;[34] in thesecases a higher preference for the a pathway in these epox-ides with electron-withdrawing groups was noted comparedto propylene oxide 1.

The DFT analyses suggest that for catalyst improvement,and specifically for those substrates that are sterically moredemanding such as internal epoxides, metallosalen systemshaving a planar coordination environment cannot be effec-

Figure 7. Free-energy difference between a pathway and b-pathway inthe ring-opening step for epoxides 1, 2 and 3.

Figure 8. Optimized structures of TS1 in a and b pathways for epoxidesa) 5 and b) 6 together with the most relevant calculated bond distances(in �), and value of the negative (imaginary) vibrational frequencies.

Chem. Eur. J. 2013, 19, 6289 – 6298 � 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chemeurj.org 6295

FULL PAPERCycloaddition Reaction of CO2 to Epoxides

tive mediators under mild reaction conditions. To createmore space to facilitate both the sterically demanding CO2

insertion and ring-closing steps (which have shown to bemore challenging than the ring-opening of the epoxide, seeTable 1) other ligand geometries (Figure 9) should be con-

sidered such as metal complexes comprising of a trigonal-bipyrimidal (TBP) coordination environment. Recent workfrom our group concentrating on FeIII-based amino tri ACHTUNGTRENNUNGphen-ACHTUNGTRENNUNGolate complexes[3c,35] with these NO3-chelating ligands incombination with an external ligand such as pyridine, THF,or an epoxide (Figure 9) provide a TBP coordination ar-rangement; these complexes have indeed proven to be moreeffective catalyst systems with lower metal loadings (0.1–0.5 mol % Fe vs. 2.5 mol% Zn) and able to convert internalepoxides and oxetanes. Furthermore, whereas the Zn-ACHTUNGTRENNUNG(salphen) complexes were hardly effective for conversion ofinternal epoxides except when using very high pressures andelevated temperatures (i.e., by using sc-CO2 as medium),[20]

these Fe-amino trisphenolate complexes (using 0.5 mol %Fe) showed good conversion levels and yields for several in-ternal epoxides using much lower pressures (pCO2 =10 bar)and similar temperatures (80 8C). Thus, the presence of lessdonor atoms in the plane of the metal center (O3) comparedwith the family of metallosalens (N2O2) seems to be benefi-cial in the creation of more powerful catalyst systems forthe formation of cyclic carbonates and possibly also for thesynthesis of polycarbonates.

Conclusion

The mechanism for the cycloaddition of CO2 to epoxidescatalyzed by the binary Zn ACHTUNGTRENNUNG(salphen)/NBu4X (X=Br, I) cata-lyst system has been investigated and elucidated in detail byDFT methods. Computational studies clearly explain thereasons why the cycloaddition of CO2 to the benchmarksubstrate propylene oxide 1 proceeds in a much easier wayin presence of Zn ACHTUNGTRENNUNG(salphen)/NBu4I compared with the un ACHTUNGTRENNUNGcat-ACHTUNGTRENNUNGalyzed or Zn ACHTUNGTRENNUNG(salphen)-catalyzed route. More importantly,this work has also revealed a number of interesting and im-portant observations that may help to design better andmore broadly applicable catalysts for organic carbonate for-mation.

For alkyl-substituted terminal epoxides 1–4 the reaction ispredominantly controlled by steric factors, whereas for thevinyloxirane 5 and styrene oxide 6 electronic factors aremore dominant. For substrates 1–3 ring-opening at the un-substituted carbon atom (b pathway) is favored and thispreference becomes more pronounced upon increasing thesteric bulk at Ca. In general, the b pathway is favored andthe rate-limiting step for these substrates is the CO2 inser-tion step in the coordinated linear alkoxide complex. In con-trast to 1–3, for substrates 5 and 6 the a pathway is favoredwith the CO2 insertion step also being rate-determining. Thecurrent results also explain why internal epoxides such as2,3-dimethyl-oxirane 4 are more sluggish as all the calculat-ed transition states (TS1–TS3) are significantly higher com-pared with those computed for propylene oxide 1. In sum-mary, we believe that this comprehensive study based on thecycloaddition of CO2 to various epoxides gives useful in-sights in the limitations of a very important family of cata-lysts (metallosalens)[21] and provides a deeper understandingof this well-known CO2 fixation reaction. Based on recentpromising findings with other types of binary catalyst sys-tems (amino trisphenolate complexes)[3c,35] having differentcoordination environments around the active metal center,the development of more active/selective catalysts for or-ganic carbonate synthesis from epoxides and carbon dioxideshould be feasible.

Experimental Section

Computational details : All calculations were carried out by using theAmsterdam Density Functional (ADF v2010.01) package.[36] DFT-basedmethods was employed at the generalized-gradient-approximation(GGA) level, with the Becke exchange[37] and the Perdew correlation[38]

functionals (BP86). A triple-x plus polarization Slater basis set was usedon all atoms. Relativistic corrections were introduced by scalar-relativisticzero-order regular approximation (ZORA).[39] Full geometry optimiza-tions were performed without constraints, and the nature of the station-ary points in the potential energy hypersurface, were characterized eitheras minima or transition states by means of harmonic vibrational frequen-cies analysis. To match with the experimental conditions reported previ-ously,[19] standard corrections to Gibbs free energy were evaluated at298 K and pressure of 10 bar, according to the expression of an ideal gas:Gm(P)=G0

m +RT ln PP0.[40] Solvent effects were introduced by using the

continuous solvent model COSMO.[41]

Acknowledgements

This work was supported by the ICIQ Foundation, ICREA of the Gener-alitat de Catalunya and the Spanish Ministry of Science and Innovationthrough projects CTQ2011-27385, CTQ2011-29054-C02-02, 2009-SGR-259 and a FPU fellowship (G.S.).

[1] a) Carbon Dioxide as Chemical Feedstock (Ed.: M. Aresta), Wiley-VCH, Weinheim, 2010 ; b) P. Markewitz, W. Kuckshinrichs, W. Leit-ner, J. Linssen, P. Zapp, R. Bongartz, A. Schreiber, T. E. Mller,Energy Environ. Sci. 2012, 5, 7281 –7305; c) T. Sakakura, J. C. Choi,H. Yasuda, Chem. Rev. 2007, 107, 2365 – 2387.

Figure 9. Schematic representation and comparison of the coordinationgeometry in the plane of the metal centers in Zn ACHTUNGTRENNUNG(salphen) complexes andFe-based amino triphenolates.

www.chemeurj.org � 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Chem. Eur. J. 2013, 19, 6289 – 62986296

A. W. Kleij, C. Bo et al.

[2] a) R. Mart�n, A. W. Kleij, ChemSusChem 2011, 4, 1259 – 1263; b) M.Peters, B. Kçhler, W. Kuckshinrichs, W. Leitner, P. Markewitz, T. E.Mller, ChemSusChem 2011, 4, 1216 –1240; c) M. Cokoja, C. Bruck-meier, B. Rieger, W. A. Herrmann, F. E. Khn, Angew. Chem. 2011,123, 8662 –8690; Angew. Chem. Int. Ed. 2011, 50, 8510 – 8537.

[3] For recent original contributions: a) M. North, C. Young, Catal. Sci.Technol. 2011, 1, 93– 99; b) A. Coletti, C. J. Whiteoak, V. Conte,A. W. Kleij, ChemCatChem 2012, 4, 1190 –1196; c) C. J. Whiteoak,E. Martin, M. Mart�nez Belmonte, J. Benet-Buchholz, A. W. Kleij,Adv. Synth. Catal. 2012, 354, 469 –473; d) T. Ohshima, J. Okudac, K.Mashima, Catal. Sci. Technol. 2012, 2, 509 – 513; e) T. Ema, Y. Miya-zaki, S. Koyama, Y. Yano, T. Sakai, Chem. Commun. 2012, 4489 –4491; f) J. Qu, C.-Y. Cao, Z.-F. Dou, H. Liu, Y. Yu, P. Li, W.-G.Song, ChemSusChem 2012, 5, 652 –655; g) C. J. Whiteoak, A. Nova,F. Maseras, A. W. Kleij, ChemSusChem 2012, 5, 2032 –2038; h) Q.-W. Song, L.-N. He, J.-Q. Wang, H. Yasuda, T. Sakakura, GreenChem. 2013, 15, 110 –115; i) C. Qi, J. Ye, W. Zeng, H. Jiang, Adv.Synth. Catal. 2010, 352, 1925 –1933; j) J. Sun, L. Han, W. Cheng, J.Wang, X. Zhang, S. Zhang, ChemSusChem 2011, 4, 502 –507.

[4] For a biography on phosgene and its derivatives, see: a) D. D. Ber-olzheimer, Ind. Eng. Chem. 1919, 11, 263 – 266; b) A.-A. G. Shaikh,S. Sivaram, Chem. Rev. 1996, 96, 951 – 976.

[5] a) T. Sakakura, K. Kohno, Chem. Commun. 2009, 1312 –1330;b) A. A. G. Shaikh, S. Sivaram, Chem. Rev. 1996, 96, 951 –976; c) B.Schffner, F. Schffner, S. P. Verevkin, A. Bçrner, Chem. Rev. 2010,110, 4554 –4581; d) D. J. Darensbourg, S. J. Wilson, Green Chem.2012, 14, 2665 –2671.

[6] a) V. Cal�, A. Nacci, A. Monopoli, A. Fanizzi, Org. Lett. 2002, 4,2561 – 2563; b) B. R. Buckley, A. P. Patel, K. G. Wijayantha, Chem.Commun. 2011, 47, 11888 –11890.

[7] a) N. Kihara, N. Hara, T. Endo, J. Org. Chem. 1993, 58, 6198 –6202;b) L. Guo, C. Wang, X. Luo, G. Cui, H. Li, Chem. Commun. 2010,46, 5960 – 5962; c) M. Aresta, A. Dibenedetto, J. Mol. Catal. A-Chem. 2001, 182 –183, 399 –409.

[8] a) J. J. Peng, Y. Q. Deng, New J. Chem. 2001, 25, 639 – 641; b) H.Yang, Y. Gu, Y. Deng, F. Shi, Chem. Commun. 2002, 274 – 275; c) H.Kawanami, A. Sasaki, K. Matsui, Y. Ikushima, Chem. Commun.2003, 896 –897; d) J. Sun, S. Fujita, M. Arai, J. Organomet. Chem.2005, 690, 3490 –3497.

[9] a) Y. Xie, Z. Zhang, T. Jiang, J. He, B. Han, T. Wu, K. L. Ding,Angew. Chem. 2007, 119, 7393 –7396; Angew. Chem. Int. Ed. 2007,46, 7255 –7258; b) Y. B. Xiong, H. Wang, R. M. Wang, Y. F. Yan, B.Zheng, Y. P. Wang, Chem. Commun. 2010, 46, 3399 –3401.

[10] a) W. J. Kruper, D. V. Dellar, J. Org. Chem. 1995, 60, 725 –727;b) F. W. Li, C. G. Xia, L. W. Xu, W. Sun, G. X. Chen, Chem.Commun. 2003, 2042 – 2043; c) W. Clegg, R. W. Harrington, M.North, R. Pasquale, Chem. Eur. J. 2010, 16, 6828 –6843; d) R. L.Paddock, S. T. Nguyen, Chem. Commun. 2004, 1622 –1623.

[11] M. L. Man, K. C. Lam, W. N. Sit, S. M. Ng, Z. Zhou, Z. Lin, C. P.Lau, Chem. Eur. J. 2006, 12, 1004 – 1015.

[12] a) H. Sun, D. Zhang, J. Phys. Chem. A 2007, 111, 8036 – 8043; b) Y.Ren, T. T. Meng, J. Jia, H. S. Wu, Comput. Theor. Chem. 2011, 978,47– 56; c) Z.-Z. Yang, Y.-N. Zhao, L.-N. He, J. Gao, Z.-S. Yin, GreenChem. 2012, 14, 519 –527; d) R. A. Watile, D. B. Bagal, K. M. Desh-mukh, K. P. Dhake, B. M. Bhanage, J. Mol. Catal. A-Chem. 2011,351, 196 –203; e) Z.-Z. Yang, L.-N. He, C.-X. Miao, S. Chanfreau,Adv. Synth. Catal. 2010, 352, 2233 – 2240.

[13] M. J. Ajitha, C. H. Suresh, Tetrahedron Lett. 2011, 52, 5403 – 5406.[14] F. Chen, X. Li, B. Wang, T. Xu, S. L. Chen, P. Liu, C. Hu, Chem.

Eur. J. 2012, 18, 9870 –9876.[15] a) C. J. Whiteoak, A. Nova, F. Maseras, A. W. Kleij, ChemSusChem

2012, 5, 2032 – 2037; b) J.-Q. Wang, J. Sun, W.-G. Cheng, K. Dong,X.-P. Zhang, S.-J. Zhang, Phys. Chem. Chem. Phys. 2012, 14, 11021 –11026.

[16] K. Roshith Roshan, G. Mathai, J. Kim, J. Tharun, G.-A. Park, D.-W.Park, Green Chem. 2012, 14, 2933 –2940.

[17] J.-Q. Wang, K. Dong, W.-G. Cheng, J. Sun, S. J. Zhang, Catal. Sci.Technol. 2012, 2, 1480 –1484.

[18] J. Ma, J. Liu, Z. Zhang, B. Han, Green Chem. 2012, 14, 2410 –2420.

[19] a) A. Decortes, M. Mart�nez Belmonte, J. Benet-Buchholz, A. W.Kleij, Chem. Commun. 2010, 46, 4580 –4582; b) A. Decortes, A. W.Kleij, ChemCatChem 2011, 3, 831 –834.

[20] M. Taherimehr, A. Decortes, S. M. Al-Amsyar, W. Lueangchaicha-weng, C. J. Whiteoak, E. C. Escudero-Ad�n, A. W. Kleij, P. P. Pes-carmona, Catal. Sci. Technol. 2012, 2, 2231 –2237.

[21] a) A. Decortes, A. M. Castilla, A. W. Kleij, Angew. Chem. 2010, 122,10016 – 10032; Angew. Chem. Int. Ed. 2010, 49, 9822 –9837. For someother reviews, see: b) P. P. Pescarmona, M. Taherimehr, Catal. Sci.Technol. 2012, 2, 2169 –2187; c) M. North, R. Pasquale, C. Young,Green Chem. 2010, 12, 1514 – 1539. For a computational review onCO2 conversion, see: d) M. Drees, M. Cokoja, F. E. Khn, Chem-CatChem 2012, 4, 1703 –1712.

[22] For a detailed study on the mechanism using bimetallic Al ACHTUNGTRENNUNG(salen)complexes, see: M. North, R. Pasquale, Angew. Chem. 2009, 121,2990 – 2992; Angew. Chem. Int. Ed. 2009, 48, 2946 –2948.

[23] D. Tian, B. Liu, Q. Gan, H. Li, D. J. Darensbourg, ACS Catal. 2012,2, 2029 –2035; see also ref. [22].

[24] F. Jutz, J.-D. Grunwaldt, A. Baiker, J. Mol. Catal. A 2008, 279, 94 –103.

[25] X.-B. Lu, B. Liang, Y.-J. Zhang, Y.-Z. Tian, Y.-M. Wang, C.-X. Bai,H. Wang, R. Zhang, J. Am. Chem. Soc. 2004, 126, 3732 –3733.

[26] D. J. Darensbourg, A. I. Moncada, W. Choi, J. H. Reibenspies, J.Am. Chem. Soc. 2008, 130, 6523 – 6533.

[27] S. J. Wezenberg, G. A. Metselaar, E. C. Escudero-Ad�n, J. Benet-Buchholz, A. W. Kleij, Inorg. Chim. Acta 2009, 362, 1053 – 1057.

[28] For illustrative examples, see: a) R. M. Haak, A. Decortes, E. C. Es-cudero-Ad�n, M. Mart�nez Belmonte, E. Martin, J. Benet-Buchholz,A. W. Kleij, Inorg. Chem. 2011, 50, 7934 –7936; b) J. A. A. W. Ele-mans, S. J. Wezenberg, E. C. Escudero-Ad�n, J. Benet-Buchholz, D.den Boer, M. J. J. Coenen, S. Speller, A. W. Kleij, S. De Feyter,Chem. Commun. 2010, 46, 2548 – 2550; c) E. C. Escudero-Ad�n, J.Benet-Buchholz, A. W. Kleij, Chem. Eur. J. 2009, 15, 4233 –4237;d) S. J. Wezenberg, E. C. Escudero-Ad�n, J. Benet-Buchholz, A. W.Kleij, Chem. Eur. J. 2009, 15, 5695 – 5700; e) E. C. Escudero-Ad�n, J.Benet-Buchholz, A. W. Kleij, Eur. J. Inorg. Chem. 2009, 3562 – 3568;see also ref. [31].

[29] For some examples, see: a) A. L. Singer, D. A. Atwood, Inorg.Chim. Acta 1998, 277, 157 –162; b) P. D. Frischmann, A. J. Gallant,J. H. Chong, M. J. MacLachlan, Inorg. Chem. 2008, 47, 101 –112;c) M. Kuil, P. E. Goudriaan, A. W. Kleij, D. M. Tooke, A. L. Spek,P. W. N. M. van Leeuwen, J. N. H. Reek, Dalton Trans. 2007, 2311 –2320; see also: d) C. J. Whiteoak, G. Salassa, A. W. Kleij, Chem.Soc. Rev. 2012, 41, 622 –631.

[30] G. Salassa, M. J. J. Coenen, S. J. Wezenberg, B. L. M. Hendriksen, S.Speller, J. A. A. W. Elemans, A. W. Kleij, J. Am. Chem. Soc. 2012,134, 7186 –7192; see also refs. [28 b] and [31 c].

[31] We also computed the a and b pathways for propylene oxide 1 cata-lyzed by the Zn ACHTUNGTRENNUNG(salphen)/NBu4I binary couple using the M06 levelof theory by the meta-hybrid GGA M06 DFT method (see the Sup-porting Information for details). The relative energies of the differ-ent TS proved to lower and markedly different (comparing TS1 withTS2/TS3), though the TS1 (i.e., ring-opening step) showed to be rel-atively more demanding. A closer inspection of the TS1 conforma-tion showed that the M06 method favors the iodide attacking the co-ordinated epoxide from the side where the tBu groups of the Zn-ACHTUNGTRENNUNG(salphen) are located (a pathway). In the b pathway this observationis less pronounced. Thus, the use of M06 favors the steric bulk ofthe ZnACHTUNGTRENNUNG(salphen), the coordinated epoxide and the incoming (large)iodide nucleophile to reside in the same area of space. Previous co-ordination studies carried out by us using substituted pyridines andZn ACHTUNGTRENNUNG(salphen)s revealed a preferred positioning of the ligand both inthe solid state (X-ray data) and in solution (NMR experiments) sup-ported by data consistent with the ortho-pyridine groups pointingaway from the tBu fragments of the ZnACHTUNGTRENNUNG(salphen)complex. See:a) E. C. Escudero-Ad�n, J. Benet-Buchholz, A. W. Kleij, Eur. J.Inorg. Chem. 2009, 3562; b) E. C. Escudero-Ad�n, J. Benet-Buch-holz, A. W. Kleij, Inorg. Chem. 2008, 47, 4256. We then carried outcoordination studies with propylene oxide (PO 1) and the Zn-

Chem. Eur. J. 2013, 19, 6289 – 6298 � 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chemeurj.org 6297

FULL PAPERCycloaddition Reaction of CO2 to Epoxides

ACHTUNGTRENNUNG(salphen)complex, and evaluated the chemical-shift changes uponcomplexation. Interestingly, upon coordination the epoxide hydro-gens all undergo a downfield shift (inductive effect) except for theepoxide-Me group, which shows a small upfield shift (the SupportingInformation). This seems to indicate that also in this case the coordi-nated ligand (epoxide) may have a preferred positioning to releasesteric strain. Therefore, some caution may be needed with the M06computational data.

[32] The high Lewis acidity is provoked by the rigid, planar coordinationenvironment around the Zn metal center. For applications, see:a) A. W. Kleij, D. M. Tooke, M. Kuil, M. Lutz, A. L. Spek, J. N. H.Reek, Chem. Eur. J. 2005, 11, 4743 – 4750; b) S. J. Wezenberg, G. Sal-assa, E. C. Escudero-Ad�n, J. Benet-Buchholz, A. W. Kleij, Angew.Chem. 2011, 123, 739 – 742; Angew. Chem. Int. Ed. 2011, 50, 713 –71.

[33] G.-P. Wu, S.-H. Wei, W.-M. Ren, X.-B. Lu, T.-Q. Xu, D. J. Dare-nsbourg, J. Am. Chem. Soc. 2011, 133, 15191 – 15199.

[34] a) G.-P. Wu, S.-H. Wei, W.-M. Ren, X.-B. Lu, B. Li, Y.-P. Zua, D. J.Darensbourg, Energy Environ. Sci. 2011, 4, 5084 – 5092; b) G.-P. Wu,S.-H. Wei, X.-B. Lu, W.-M. Ren, D. J. Darensbourg, Macromolecules2010, 43, 9202 –9204.

[35] a) C. J. Whiteoak, B. Gjoka, E. Martin, M. Mart�nez Belmonte, E. C.Escudero-Ad�n, C. Zonta, G. Licini, A. W. Kleij, Inorg. Chem. 2012,51, 10639 –10649; b) C. J. Whiteoak, N. Kielland, V. Laserna, E. C.Escudero-Ad�n, E. Martin, A. W. Kleij, J. Am. Chem. Soc. 2013,135, 1228 –1231.

[36] ADF2010, SCM, Theoretical Chemistry, Vrije Universiteit, Amster-dam, The Netherlands, http://www.scm.com: E. J. Baerends, T. Zie-gler, J. Autschbach, D. Bashford, A. B�rces, F. M. Bickelhaupt, C.

Bo, P. M. Boerrigter, L. Cavallo, D. P. Chong, L. Deng, R. M. Dick-son, D. E. Ellis, M. van Faassen, L. Fan, T. H. Fischer, C. FonsecaGuerra, A. Ghysels, A. Giammona, S. J. A. van Gisbergen, A. W.Gçtz, J. A. Groeneveld, O. V. Gritsenko, M. Grning, S. Gusarov,F. E. Harris, P. van den Hoek, C. R. Jacob, H. Jacobsen, L. Jensen,J. W. Kaminski, G. van Kessel, F. Kootstra, A. Kovalenko, M. V.Krykunov, E. van Lenthe, D. A. McCormack, A. Michalak, M. Mi-toraj, J. Neugebauer, V. P. Nicu, L. Noodleman, V. P. Osinga, S.Patchkovskii, P. H. T. Philipsen, D. Post, C. C. Pye, W. Ravenek, J. I.Rodr�guez, P. Ros, P. R. T. Schipper, G. Schreckenbach, J. S. Selden-thuis, M. Seth, J. G. Snijders, M. Sol�, M. Swart, D. Swerhone, G. teVelde, P. Vernooijs, L. Versluis, L. Visscher, O. Visser, F. Wang,T. A. Wesolowski, E. M. van Wezenbeek, G. Wiesenekker, S. K.Wolff, T. K. Woo, A. L. Yakovlev.

[37] A. D. Becke, Phys. Rev. A 1988, 38, 3098 –3100.[38] a) J. P. Perdew, Phys. Rev. B 1986, 33, 8822 – 8824; b) J. P. Perdew,

Phys. Rev. B 1986, 34, 7406 – 7406.[39] a) E. van Lenthe, E. J. Baerends, J. G. Snijders, J. Chem. Phys. 1993,

99, 4597 –4610; b) E. van Lenthe, E. J. Baerends, J. G. Snijders, J.Chem. Phys. 1994, 101, 9783 – 9792.

[40] Atkins� Physical Chemistry (Eds.: P. Atkins, J. de Paula), 8th ed.,Oxford University Press, Oxford, 2006, pp. 108 –109.

[41] A. Klamt, J. Phys. Chem. 1995, 99, 2224 –2235.

Received: November 7, 2012Revised: February 15, 2013

Published online: March 19, 2013

www.chemeurj.org � 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Chem. Eur. J. 2013, 19, 6289 – 62986298

A. W. Kleij, C. Bo et al.