Mono- and Bifunctional Heterogeneous Catalytic Transformation of Terpenes and Terpenoids

Mono- and bifunctional heterogeneous catalytic transformation

of terpenes and terpenoids

Nicoletta Ravasioa,�, Federica Zaccheriab, Matteo Guidottia, and Rinaldo Psaroa

aCNR-Istituto di Scienze e Tecnologie Molecolari, Via Golgi 19, 20133 Milano, ItalybUniversita di Milano, Dipartimento di Chimica Inorganica Metallorganica e Analitica, Via Venezian 21, 20133 Milano, Italy

Selective reactions of terpenes, catalyzed by different heterogeneous systems, are reported. �-pinene and limonene epoxides can

be effectively isomerized to carbonyl compounds (selectivity >70%) over silica aluminas that appear to be good alternatives to

homogeneous ZnBr2. The different reactivity of geometric isomers can be used to separate them. Unsaturated ketones can be

converted through a two-step one-pot reaction in cyclic or bicyclic ethers through a bifunctional process involving a hydrogenation

and an acid-catalyzed step. However, the choice of the solvent allows to inhibit the acidic sites, thus obtaining a selective

hydrogenation reaction. On the other hand, the acidic reaction alone can be useful for alcohol epimerization and epimer separation.

Grafting of Ti in a siliceous matrix gives rise to a material with both redox and acidic properties. Solids obtained in this way are

active and selective in the epoxidation of terpenic alcohols but can also promote bifunctional reactions.

KEY WORDS: bifunctional catalysts; copper catalysts; titanium silicates; chemoselective hydrogenation catalysts; epoxide

isomerization; mixed oxide; ether formation; epoxidation; isomer resolution; MCM-41.

1. Introduction

Terpenes are natural products whose structures arebuilt up from isoprene units. The broader term ‘‘ter-penoids’’ also covers natural degradation products,such as ionones, and natural and synthetic derivatives,e.g., terpene alcohols, aldehydes, ketones, acids, esters,epoxides and hydrogenation products. They are de-graded by microorganisms such as the Pseudomonas andAspergillus species and occur everywhere and in allorganisms, in particular, in higher plants. Major sourcesof terpenes are thus balsams, natural resins and essentialoils, but they are also by-products of lemon and orangejuice production as well as of pulp and paper industries.Some of them, such as (�)-�-pinene and (þ)-limoneneare among the more readily available optically activeproducts and are therefore used for the synthesis of otheroptically active products and as reagents for cleavingracemates. Therefore, they are very cheap precursors offragrances, flavors, drugs and agrochemicals [1]. The highreactivity of these molecules often makes it difficult toachieve a selective transformation. However, the target isworth pursuing as almost all products can be useful.Here, we report some of our results in the field ofheterogeneous catalytic reactions of terpenes and ter-penoids. The paper will be organized as follows:

– acid-catalyzed rearrangements of terpenic epoxidesover different amorphous silica aluminas;

– hydrogenation and hydrogenation þ acidic reactionscatalyzed by supported copper catalysts;

– selective epoxidation of terpenic alcohols catalyzed bytitanium silicates.

2. Experimentals

2.1. Epoxide isomerization

All the mixed oxides used were obtained from GraceDavison (Worms, Germany). Their textural propertiesare reported in table 1.

They were treated at 270 �C or 450 �C for 20min inair and for 20min under reduced pressure at the sametemperature.

�-Pinene oxide (97%) and the mixture of cis and trans(þ)-limonene oxide (42% cis : 58% trans, >97%) wereobtained from Aldrich, (þ)-cis and (þ)-trans-limonene-1,2-epoxide (>99%) from Fluka.

The substrates were dissolved in toluene, dried overmolecular sieves, and the solution transferred under N2

into a glass reaction vessel in which the catalyst hadbeen previously activated. Reaction mixtures wereanalyzed by GC using a polyethylene glycol capillarycolumn (injection T ¼ 140 �C) for �-pinene oxideisomerization and a cross-linked 5% phenylmethylsilicone or a nonbonded poly (80% biscyanopropyl/20% cyanopropylphenyl siloxane) capillary column forlimonene oxide isomerization.

2.2. Hydrogenation reactions

Cu=SiO2 was prepared by adding the support (SiO2

gel from Grace Davison, N2-BET ¼ 320m2=g,PV ¼ 1:79mL=g) to a solution of Cu½ðNH3Þ4�2þ and

�To whom correspondence should be addressed.

E-mail: [email protected]

Topics in Catalysis Vol. 27, Nos. 1–4, February 2004 (# 2004) 157

1022-5528/04/0200–0157/0 # 2004 Plenum Publishing Corporation

hydrolyzing this solution by slowly diluting withwater. After filtration, drying and calcination at350 �C for 4 h, the catalyst features were 8.6% ofcopper, N2-BET ¼ 332m2=g, PV ¼ 1:06mL=g, SCuð0Þ ¼129m2=gCu.

The catalyst was activated before use by treatment at270 �C in air for 20min, under reduced pressure at thesame temperature for 20min and then under H2 at thesame temperature by removing the water formed by thereduction under reduced pressure.

�-Ionone (96%), geranylacetone (99%), 6-methyl-5-hepten-2-one (99%), dihydrocarveols (95%), and car-vone (98%) were purchased from Aldrich.

The substrates were dissolved in n-heptane, driedover molecular sieves, and the solution transferredunder N2 into a glass reaction vessel in which thecatalyst had been previously reduced. The hydrogena-tion reaction was carried out with 1 atm of hydrogen.Reaction mixtures were analyzed by GC, using anonbonded biscyanopropyl polysiloxane capillary col-umn for �-ionone, 6-methyl-5-hepten-2-one, dihydro-carveols and carvone and a (50% cyanopropylphenyl)methylpolysiloxane column for geranylacetone hydro-genation, by GC-MS (5% phenylmethyl siliconecapillary column) and by 1H NMR.

The Cu(0) surface area used for TOF calculationswas determined by the N2O dissociative adsorptionmethod using a pulse-flow technique. The Cu(0) surfacearea was computed by considering a surface coveragefactor (moles of oxygen atoms per moles of surfaceCu(0) atoms) Q ¼ 0:35 and a mean surface area of thecopper site of 7:41 A2.

2.3. Epoxidation reactions

Ti-MCM41, Ti=SiO2 and Ti/Aerosil were preparedfrom the corresponding supports by the graftingtechnique, using a solution of titanocene dichloride(Aldrich) in chloroform (Carlo Erba, RPE) andtriethylamine (Aldrich).

Four terpenic substrates were used as purchased: �-terpineol (96% Aldrich), (�)-isopulegol (puriss. Fluka),(�)-carveol (97%, Aldrich, cis/trans ratio ¼ 40 : 60) and(R)-(þ)-limonene (97%, Aldrich; 98% ee).

All of the catalysts were pretreated in air at 500 �C(10 �C=min temperature programme from room tem-perature to 500 �C) for 1 h and cooled to reactiontemperature in vacuo. Acetonitrile (Aldrich) was used assolvent (previously dried on molecular sieves), anhy-drous tert-butyl hydroperoxide (Aldrich, 5M solution indecane) as oxidant ðoxidant : substrate molar ratio ¼1 : 1Þ, and the catalyst to substrate ratio was 30%.Reaction mixtures were analyzed by GC chromatogra-phy (HP5890, (50% cyanopropylphenyl) methylpolysi-loxane column, FID and MS detectors) and 1H NMRspectroscopy.

3. Results and discussion

3.1. Epoxides isomerization

�- and �-pinene (ca. 18 000 and 12 000 tons/yearproduced respectively) are the main inexpensive mono-terpenes present in turpentine, a very important rawmaterial in the flavor and fragrances industry (F&F) [2].Epoxidation of �-pinene by peracetic acid gives �-pinene oxide 1 that rearranges to campholenic aldehyde2 in the presence of catalytic amounts of zinc salts.This aldehyde is a very important intermediate used forthe manufacture of sandalwood fragrances such asSandalore1 (Givaudan), Bacdanol1 (IFF), Brahmanol1

(Dragoco) and Polysantol1 (Firmenich).Recently, concerns about the toxicity of nitromusks

and the low biodegradability of polycyclic musks(OSPAR Commission, 1998) which are among the mostused fragrances in laundry detergents, fabric softeners,cleaning agents, and cosmetic products in Europe(ca. 2000 tonnes in 1998) and therefore ubiquitouslypresent in the aquatic environment, have increased theimportance of sandalwood fragrances. This urged us tolook for innovative catalysts for �-pinene oxideisomerization and, in particular, for heterogeneous ones.

The acidic rearrangements give several products,some of which are shown in scheme 1. HomogeneousZnCl2 is quite selective, giving 85% selectivity of thedesired product, while heterogeneous systems usuallygive lower selectivity as they contain different acidic sitesas far as both nature (Brønsted or Lewis) and strengthare concerned [3].

According to the comprehensive work of vanBekkum et al. [4], Brønsted acids, both homogeneousand heterogeneous ones, give about 50% of 2, the mainby-product being carveol 3, whereas Lewis acids give 2

with yields varying from 40 to 80%, the main by-productbeing pinocamphone 4. A selectivity of 78% wasreported by using a carefully prepared H-USY zeoliteat low reaction temperatures (around 0 �C). The authorsproposed that the high selectivity was a result of well-dispersed Lewis acid sites in a nearly all-silica matrix [5].According to this hypothesis, Ti-beta has proved to bean excellent catalyst for this reaction in both the liquid

Table 1

Textural features of different cogels used as Lewis acids

Cogel Co-oxide loading

(wt%)

N2-BET

(m2/g)

PV

(mL/g)

SiO2-TiO2 2.3 297 1.26

SiO2-ZrO2 4.7 304 1.62

SiO2-Al2O3 A 13 400 1.1

SiO2-Al2O3 B 1.5 485 0.79

SiO2-Al2O3 C 25 400 1.0

SiO2-Al2O3 D 13 475 0.8

N. Ravasio et al./Heterogeneous catalytic transformation of terpenes and terpenoids158

and the vapor phase (up to 94% selectivity), owing to thepresence of isolated well-dispersed titanium sites in aBrønsted acid-free silica matrix and to the transitionstate shape selectivity induced by the pore structure ofthe zeolite [6]. Sulfated alumina activated at 650 �C hasalso shown to be a good catalyst, giving yields up to 76%at full conversion at 0 �C [7].

Following the hypothesis of selectivity being due tothe dispersion of Lewis acid sites, we obtained very goodresults by using commercial silicas modified with smallamounts of a second oxide containing Lewis acidic sites,namely, alumina, titania and zirconia [8].

Table 2 reports some selected results; yields inproduct 3 and 4 are also reported as they are diagnostic,respectively, of the presence of Brønsted (3) and Lewis(4) acidity, as already mentioned.

All of them turned out to be active under mildconditions ð25–90 �CÞ and oligomeric products werenever detected. Silica and silica titania activated at lowtemperature showed a very low activity with a productdistribution in agreement with Brønsted acid character,whereas silica alumina 1.5 and silica zirconia gave higherselectivity towards campholenic aldehyde and pinocam-phone, according to their more significant Lewis acidity,particularly when activated at 450 �C. Preliminaryresults on surface characterization of these materialsrule out the presence of Brønsted acidic sites, whereasFT-IR spectra of SiO2-Al2O3 1.5 and SiO2-ZrO2 afteradsorption of pyridine showed the vibration modes at1596 and 1445 cm�1, diagnostic of the presence of Lewisacid sites, but no evidence of pyridinium ion formation[9]. Activation at higher temperature, namely 450 �C,

improved both activity and selectivity of these twosolids, particularly of SiO2-Al2O3 1.5wt% that could beused at room temperature giving a 72% yield incampholenic aldehyde. It is also worth noting that thislow loading silica alumina ðSi=Al ¼ 70Þ was found to bemore active than a 13% one ðSi=Al ¼ 6Þ, thus showingthat this reaction requires well-dispersed Lewis acid sitesinstead of Brønsted ones.

The selectivity found is not very high when comparedwith the best systems reported in the literature.However, the issue of productivity has recently beenaddressed for this specific reaction carried out onsulfated alumina [7]. Therefore, we would like to stressthat the simple catalyst we used, amorphous andcommercial silica alumina, activated at 450 �C andoperating at room temperature, is much more activethan both sulfated alumina and Ti-beta (table 3).

The rearrangement of limonene oxide 5 has beenstudied only occasionally. ZnBr2, which is the mostwidely used Lewis acid in terpene chemistry, was foundto give cyclopentanecarboxyaldehyde 6 and dihydrocar-vone 7 [10], BF3-Et2O gave mainly 7. Tanabe et al.,exploring the reactivity of different solid acids, foundthat silica alumina gives the same two products [11]. Wehave studied the isomerization of limonene oxide withfour different silica aluminas and we have foundcompletely different reactivities and selectivities for eachof the two isomers (scheme 2, table 4): cis limoneneoxide gave cyclopentanecarboxyaldehyde 6 (77% with13% SiO2-Al2O3 A) whereas trans reacted very muchslower to give dihydrocarvone 7 (70% with 13%SiO2-Al2O3 D) [12]. The formation of 6 and 7 appearsto be in agreement with the Lewis acid character of thecatalysts.

The difference in reactivity of the two isomers hasrarely been observed. Ring opening with water in thepresence of a Mo(VI) complex occurs only on the cis

Scheme 1.

Table 2

Isomerization of �-pinene oxide in the presence of different solid acidsa

Entry Catalyst T

(8C)t

(min)

Yield

(%)

%3 %4

1 SiO2 90 60 55 10 6

2 SiO2-TiO2 2.3 90 60 57 14 7

3 SiO2-Al2O3 1.5 90 5 63 9 13

4 SiO2-Al2O3 1.5b 25 5 72 3 14

5 SiO2-Al2O3 13 90 10 65 4 17

6 SiO2-ZrO2 4.7 90 15 64 7 11

7 SiO2-ZrO2 4.7b 60 20 72 4 12

a0.2-g �-pinene oxide, 0.2-g catalyst activated at 270 8C, toluene, time

required to reach 100% conversion.bCatalyst activated at 450 8C.

Table 3

Selectivity versus productivity in the isomerization of �-pinene oxide

Catalyst Selectivity

(%)

Productivity

(mmol/gcat/h)

References

ZnCl2 85 125 4

SiO2-Al2O3 commercial 72 57 8

Al2O3 sulfated 76 19 7

Ti-beta 94 4 6

N. Ravasio et al./Heterogeneous catalytic transformation of terpenes and terpenoids 159

isomer [13], whereas only the trans reacts with secondaryamines to give �-amino alcohol chiral auxiliaries [14].Conformational differences in the transition states canaccount for the different reactivity in the homogeneousphase. In our opinion, the very fast reaction of cis 5 oversilica aluminas may be due to both release of strain andits easier access to the catalytic site, as comparisonbetween the two more stable conformers shows theoxygen atom to be more hindered in the trans isomer[12]. This difference can be exploited to set up a kineticresolution of the two geometric isomers that can beconveniently realized by using the low loading 1.5wt%silica alumina B; almost all the trans isomers can berecovered unaffected.

Completely different reactions were observed whenusing copper supported on the same silica aluminas andreduced at 270 �C before reaction under hydrogenationconditions. Cu=SiO2-Al2O3 A and Cu=SiO2-Al2O3 C

again produced 6 and 7, whereas Cu=SiO2-Al2O3 B andCu=SiO2-Al2O3 D gave mainly carvenone 8 and p-cymene 9 (table 5). According to Tanabe et al. [11], theseproducts are due to strong Brønsted acidity. Thus, largeamounts of 8 and 9 (62 and 35% respectively) wereformed over H2SO4=SiO2 while CH3COOH � 1%H2SO4 gave 35% 8, 16% 7 and 10% 9 [15].

The genesis and nature of these strong Brønsted siteson Cu=SiO2-Al2O3 is still to be elucidated. However, it

appears to be related to the presence of metallic Cu andmolecular H2. Indeed, Cu=SiO2-Al2O3 B and Cu=SiO2-Al2O3 D also showed hydrogenation activity, whereasCu=SiO2-Al2O3 A and Cu=SiO2-Al2O3 C did not.

Only one direct transformation of limonene oxide 5

into carvenone 8 in high yield (90%) has been claimed. Itwas carried out by adding dropwise at a given speed thesubstrate to a tubular reactor of a specific dimensionfilled with CaA zeolite at 190 �C and 10-mm Hg [16].

3.2. Hydrogenation and bifunctional hydrogenation-acidic reactions

Catalytic hydrogenation has a long tradition in the areaof fine chemicals production. No other reduction methodhas proved to be as versatile and industrially important,the most frequently used catalysts being those based onpalladium on various supports, followed by nickelcatalysts (mostly Raney type) and then by platinum,rhodium and ruthenium ones. However, Raney catalystsare pyrophoric while supported noble metals can bepyrogenic; therefore, concerns about safety in handlingthem are rising. Moreover, palladium can leach undercatalytic hydrogenation conditions, thus representing arelevant problem for pharma end products.

On the other hand, someparticular aspects of selectivitystill need to be improved, e.g., the chemoselective

Scheme 2.

Table 4

Isomerization of limonene oxide in the presence of different silica

alumina catalysts activated at 270 8C at room temperature in toluene

Entry Substrate Catalyst t (h) %trans %6 %7

1 Cis 5 13% SiAl A 1 – 77 7

2 Cis 5 1.5% SiAl B 2 – 48 13

3 Cis 5 25% SiAl C 1 74 9

4 Cis 5 13% SiAl D 1.5 – 72 9

5 Trans 5 13% SiAl A >4 69

6 Trans 5 1.5% SiAl B >4 – – 67

7 Trans 5 25% SiAl C >4 20 52

8 Trans 5 13% SiAl D >4 – – 70

Table 5

O

toluene, 1 atm H2, 90˚CO

+

5 8 9

t (h) %8 %9

Cu/SiO2-Al2O3 B 4 14 80

Cu/SiO2-Al2O3 D 6 47 30


hydrogenation of an �,�-unsaturated carbonyl system tothe saturated one. The formation of even very lowamounts of allylic alcohol is undesirable in manypreparations. We have obtained very good results in thisreaction by using an 8% Cu=SiO2 catalyst supported onfumed, nonporous (Aerosil 380) silica [17] and we haveapplied this catalyst with success in the hydrogenation of�-ionone 10. We have chosen this particular substratebecause this terpenoid is the key intermediate in thesynthesis of vitamin A, but also for a series of moleculesvaluable for the F&F industry, e.g., theaspirane andtheaspirone [18] and because the conjugation of threedouble bonds makes the selective reduction of only one ofthem more difficult. The selective hydrogenation of �-ionone to dihydro-�-ionone 11 can be carried out using aheterogeneous Raney Ni–Al alloy treated by sodiumhydroxide in 89% yield [19], with a Ru–C catalyst [20] orwith Ph3SnH in benzene (94%) [21].

Cu/Aerosil allowed us to obtain dihydro-�-iononewith selectivities between 95% and 100% at very highconversions. However, a significant amount of catalystis required in order to complete the reaction in a fewhours and the use of pyrogenic silica during the catalystpreparation may represent a drawback.

We also found that by using acidic supports likeSiO2-TiO2 and SiO2-ZrO2, a two-step one-pot processcould be set up with formation of bicyclic ethers [22].

Here, we report that the use of copper catalystsupported on mesoporous silica gel not only leads tovery high activity and yield in dihydro-�-ionone but,through the tuning of reaction conditions, also allows toobtain different products with high selectivity.

Results obtained in the hydrogenation of �-iononeare reported in table 6. A significant improvement inproductivity is obtained with respect to Cu/Aerosil.

The very high activity of the Cu=SiO2 catalyst can bedue to the preparation method and requires somecomments.

The influence of the preparation method has beendeeply investigated by comparing two sets of Cu=TiO2

catalysts; one prepared by the chemisorption-hydrolysismethod (CH), and the other by the incipient wetnesstechnique (IW), using the hydrogenation of 1,3-COD asa test reaction [23]. The former series was found to bemuch more active, turnover frequencies being about 100times higher than those observed for the IW samples. Asignificant contribution can be ascribed to the highmetal dispersion of CH samples. Thus, a specific copperarea of 55m2=gCu was measured for the 8% CHCu=TiO2 whereas the 8% IW sample’s Cu(0) surfacearea was only 9m2=gCu.

However, the difference in activity is so big that someother factor has to be taken into account. In particular,the preparation method also has a strong influence onthe metallic phase morphology. FT-IR spectra ofadsorbed CO (figure 1) suggested evidence that themain difference between CH and IW catalysts is thepresence on the former of well formed crystallitesexposing a significant fraction of step and edge sites,whilst IW samples expose only almost isolated coppersites [24].

Characterization of 8% Cu=SiO2 used throughoutthis work showed that copper particles morphology onthis catalyst is very similar to that observed on Cu=TiO2

and well-formed small Cu crystallites, exposing asignificant fraction of step and edge sites, particularly(111) microfacets, very efficient in H2 dissociation, wereidentified [25]. Moreover, the use of a support withhigher surface area allows the formation of a welldispersed copper metallic phase, with a specific surfacearea, measured by N2O titration of about 130m2=gCuand mean particle diameter of around 3.5 nm.

High dispersion and morphology of the metallicphase account for the TOF, calculated per active siteand reported in table 6, as well as of the results obtainedin the hydrogenation of other functional groups [26].

Coming back to selectivity, dihydro-�-ionone can beobtained with 98% yield by hydrogenating �-iononeover a limited amount of Cu=SiO2 in n-heptane (scheme3). It is worth noting that the allylic alcohol was never

Table 6

Hydrogenation of �-ionone over 8.6% Cu/SiO2a

Catalyst weight

(mg)

%11 after

20 min

tb

(h)

Selectivity

(%)

Productivityc TOFd

(h�1)

200 80 0.5 94 9.8 4.8 � 105

100 44 1 95 9.9 5.2 � 105

50 20 2 97 10.1 4.5 � 105

25 7 6 98 6.8 3.3 � 105

Cu/Aerosil 300mg – 2.5 100 1.0 5.1 � 104e

a200mg substrate (1.04mmols), heptane (90 8C).bTime required to reach complete conversion.cMmols produced (gcat/h).dcalculated after 20min.eCalculated at the end of reaction with a SCu(0) ¼ 117m2/gCu.


detected in any step of the reaction, neither in thehydrogenation of �-ionone nor in that of other, evenmore activated, �, �-unsaturated ketones.

On the other hand, by carrying out the reaction withhigh amounts of catalyst in a hydrocarbon (n-heptane)solvent, we can obtain in one step the bicyclic etherstetrahydroedulans 12, valuable molecules for the F&Findustry [27].

The mechanism of formation of these ethers is due tothe presence on the catalyst surface of two differentsites: hydrogenation sites and Brønsted acidic sites.

By allowing the reaction to proceed, the ketoneinitially formed is reduced to the saturated alcohol(figure 2), leaving unaffected the endocyclic olefinicbond. The last double bond is activated as carbonium

ion on the acidic site and promotes the nucleophilicaddition of the hydroxy group, thus giving the bicyclicether. A yield of 65% of ethers 12 could be obtainedfrom this one-pot one-step bifunctional transformation,the main by-products are spiroderivatives and dehydra-tion products also formed through acid-catalyzedreactions. As far as stereochemistry is concerned, transisomers are strongly favored starting from �-ionone(75%). More valuable cis ones can be obtained by thesame route starting from �-ionone (60%) or from �-ionol (80%) [22]. Conditions reported here are muchmilder than those previously reported for Cu=SiO2-TiO2

and Cu=SiO2-ZrO2 requiring a low H2 pressure toreduce the ketonic group. It is also worth noting that inthe patent literature these ethers can be obtained in three

Figure 1. IR absorption spectra of the carbonyl stretching region, taken after 10mbar CO interaction on different samples. (a) Cu=TiO2 IW (4%

Cu) prereduced at 50 �C (1), 150 �C (2), 400 �C (3); (b) Cu=TiO2 CH (4% Cu) prereduced at 50 �C (1), 150 �C (2), 400 �C (3). From reference [24].

Scheme 3.


separate steps from �-ionone: (i) hydrogenation of �-ionone to dihydro-�-ionone over Raney nickel, (ii)selective reduction of dihydro-�-ionone to the alcoholwith LiAlH4 in Et2O at 0 �C, and (iii) formation of theether catalyzed by concentrated H2SO4 in toluene at2–3 �C [28]. From the green chemistry point of view, theadvantages offered by the use of Cu=SiO2 are evident.

However, the surface acidity of the catalyst can beeasily inhibited by only using an oxygenated solvent.Ethereal solvents like dioxane strongly reduce thereaction rate, but 2-propanol can be very effectivelyused whenever reduction of a carbonyl group isrequired. In the case of �-ionone, dihydro-�-ionol 13can be obtained in 95% yield.

This ether formation reaction is not limited to theparticular structure of ionones, but it can always takeplace when hydrogenating an unsaturated ketone. In thecase of 6-methyl-5-hepten-2-one 14, we can easily switchfrom the unsaturated alcohol 15 to the pyranicderivative 16 only by changing the solvent (scheme 4)[29].

It is worth noting that acid-catalyzed cyclization ofalkenols is the most general method used to synthesizetetrahydropyranoids and tetrahydrofuranoids, valuablemolecules in fragrance chemistry [30].

On the other hand, in the case of geranylacetone 17, itis almost impossible to obtain a selective acid-catalyzedreaction, owing to the many possible pathways, but byusing 2-propanol as the solvent, a very high selectivity in

the alcohol 18 can be obtained (scheme 5), comparableonly with that found through reduction of 17 withstoichiometric amounts of polymethylhydrosiloxaneand catalytic amounts of an active zinc compound [31].

The formation of ethers can also be used to separatedifferent epimers of an alcohol. Thus, the reactionof a mixture of dihydrocarveols 19 reacts overCu=SiO2-TiO2 under H2 to give the two ethers, cis-and trans-dihydropinols 20 (scheme 6). However, onlytwo of the epimers can assume a conformation suitablefor the cyclization reaction, that is, the two with thehydroxy group in an equatorial position. In particular,the e,a would give cis-dihydropinol, whereas the e,ewould give the trans ether (scheme 7). By treating acommercial mixture of 19 (65% of e,e) under catalytichydrogenation conditions, the two ‘‘right’’ isomers reactto give the two ethers. However, reaction of e,a is veryfast whereas reaction of e,e is not, as it requires a three-axial conformation.

A parallel copper-catalyzed dehydrogenation–hydro-genation process results in epimerization at C1 (thecorresponding ketones are formed when working underN2). Thus, a,a converts into e,a that in turn gives the cis-ether, while epimerization of e,e gives a,e (figure 3). Thefinal result is the elimination of two stereoisomers andthe partial conversion of the more stable e,e isomer intothe corresponding axial alcohol.

On the other hand, starting from the unsaturatedketone carvone 21, a different distribution of 19 epimers

Figure 2. Hydrogenation of �-ionone followed by acid catalyzed cyclization carried out with H2 (1 atm) and 200mg of Cu=SiO2.

Scheme 4.


was obtained through hydrogenation over coppercatalysts favoring the e,a isomer (41%). The highportion of this ‘‘right’’ isomer gives a higher yield in20 through in situ cyclization, with a high stereo-selectivity toward the cis one. It is worth underliningthat by carrying out bifunctional reactions in one step,we can not only reduce the number of steps in asynthetic pathway but also improve the stereochemistryof the process with respect to the two separate steps, asshown in the synthesis of (–)-menthol starting from (þ)-citronellal [32].

Finally, in order to link the results described in thefirst and second parts of this paper and to have a deeper

insight into the potential of Cu=SiO2, we have carried

out preliminary tests on the isomerization of limonene

oxide 5 at 90 �C under hydrogenation conditions. Main

products were dihydropinols 20 (ca. 25%) and dihydro-

carveols 19 (ca. 25%). Though yield in ethers is low, by

following the products distribution versus time, we

could show that the ether is formed through a three-step

process: acid-catalyzed opening of the epoxide, hydro-

genation of the carbonyl group and acid-catalyzed ether

formation (scheme 8).

Scheme 5.

Scheme 6.

Scheme 7.


3.3. Epoxidation reactions

Selective epoxidation of C¼C double bonds is ahighly relevant transformation in the synthesis ofintermediates in fine and speciality chemistry. Duringthe last years, particular efforts have been focused onthe development of heterogeneous catalysts that can beused under mild conditions with aqueous hydrogenperoxide or organic hydroperoxides as the oxidants. Wehave investigated the epoxidation of terpenic alcoholswith TBHP over Ti-containing silicates by focusingmainly on three points:

– the preparation method;– the influence of structural features of the silicate;– the synthetic potential of acidic sites.

As far as the first point is concerned, we first took intoaccount two MCM-41 based catalysts with comparablemetal loading: in the first one, hereafter referred to as[Ti]-MCM41, Ti was incorporated in the matrix duringthe sol-gel synthesis whereas in the second, referred to asTi-MCM41, the metal was grafted from a titanocenesolution as described for the first time by Thomas et al.

[33]. Selected results of the catalytic performances ofthese two catalysts in the epoxidation of substrates 22,23, 3 and 24 are reported in table 7 [34].

The grafted catalyst appears to be more active thanthe in-framework one by a factor of 10 for the case of �-terpineol 22, the main constituent of pine oil. Thisbehavior is in agreement with the supposed higherexposure of active titanium sites obtained by graftingtitanocene compounds on the surface of the material.However, the gap in activity is much smaller when theOH group approaches the C¼C bond on the molecular

Figure 3. Kinetic resolution of the two couples of epimers 20 through

acid-catalyzed ether formation over Cu=SiO2 under H2.

Scheme 8.

Table 7

TOF of terpene epoxidation on [Ti]-MCM-41 and

Ti-MCM-41a

Substrate TOF (h�1)b

[Ti] Ti

OH22

2 20

OH

23

9 15

OH

3

15 33

24

4 20

aReaction conditions: CH3CN solvent; 30% wt%

catalyst; TBHP : terpene molar ratio ¼ 1.1; 85 8C.


skeleton, due to framework titanium being much moresensitive to the activating effect of hydroxy group. Thesolvent choice is crucial for selectivity. Thus, epoxida-tions performed in AcOEt over the more active graftedcatalyst show that the closer the C¼C bond to the OHgroup, the higher the activity in the first 6 h of reaction,but that as time progresses, secondary reactions lead toby-products and lower the selectivity to epoxides (after areaction time of 24 h) (table 8 ).

Most of the by-products are due to acid-catalyzedintramolecular ring closure on the already formed epoxyring for substrates such as 22 and to oxidation of theOH group for secondary alcohols. The acidic characterhas to be attributed mainly to the Lewis acidity of Ti(IV) sites grafted onto the siliceous matrix [35]. The useof CH3CN with a more significant Lewis basic characterkeeps the selectivity value constant during the wholereaction time, as already observed in the presence of Ti-beta [36].

The role of a OH group in activating the epoxidationreaction should not be overestimated. It can enhancethe reaction rate by binding in some way to thehydrophilic surface, but the overall nucleophilic char-acter of the C¼C bond is ruled by the number ofsubstituents. Thus, a very similar regioselectivity wasobserved on carveol 3 and limonene 24, the electronricher endocyclic bond being epoxidized selectively withrespect to the exocyclic one (83 : 17 for 3, 80 : 20 for 24)both in the presence and in the absence of an alcoholicfunction. On the contrary, a preferential oxidation ofthe side chain double bond occurs on narrower poremolecular sieves as Ti-beta, where steric effects prevailover electronic ones [37].

The second point regards the influence of the orderedstructure of MCM-41 and of mesoporosity on theeffectiveness of Ti-based heterogeneous catalysts. Wehave addressed this point in a systematic way bycomparing Ti-MCM-41 with two catalysts obtained bygrafting titanium through the same technique on anonordered, but mesoporous, silica gel, Ti=SiO2, and on

a nonordered, fumed and nonporous silica, Ti/Aerosil[38].

Under the conditions used, the porosity features donot appear to have any significant influence on catalyticperformances. Results reported in table 9 show that allthe catalysts are good ones for the epoxidation of thesubstrates considered here as far as both activity andselectivity are concerned. The similar reaction ratesobserved show that activity is determined by thepreparation method, as in Ti-grafted silicates virtuallyall titanium atoms are exposed and accessible, the highsurface area of MCM-41 playing only a secondary role.

On the other hand, very similar selectivity valuesshow that the active sites work in the same way in allcatalysts, notwithstanding the structural surroundings.It is worth noting that a commercial TiO2-SiO2 (GraceDavison, 2.3% TiO2) turned out to be the most activeand selective in most cases, particularly when consider-ing TON as a parameter [38].

Only in the epoxidation of 22 did we observe a higheractivity, together with a lower selectivity, in the presenceof Ti-MCM-41. This may be due to a particular affinityof this substrate with the catalyst surface, as also shownby the comparison with in-framework [Ti]MCM-41. Asalready mentioned, the low selectivity is due to thecatalyst surface acidity. A qualitative evaluation of thisproperty can be obtained from comparison of theconversion rates in the cyclization of citronellal 25 toisopulegol 23 (scheme 9). This reaction involves theformation of a C–C bond between one of the atomsinvolved in the olefinic bond and the carbon atom of thecarbonyl group. It can be activated thermally but asthe enophile should be electron deficient, complexationof the C¼O bond with a Lewis acid or protonationwith a Brønsted acid can increase the reaction rate.

Table 8

Terpene epoxidation in acetonitrile and ethyl acetate on Ti-MCM-41a

Substrate Solvent

CH3CN AcOEt

C (%)b S (%)c C (%)b S (%)c

22 90 51 95 17

23 73 80 78 41

3 82 73 82 69

24 62 79 68 62

aReaction conditions: 30wt% catalyst; TBHP : terpene molar ratio ¼1 : 1; 85 8C.

bTerpene conversion after 24 h.c1,2-epoxide selectivity after 24 h.

Table 9

Conversion and selectivity in the epoxidation of terpenes over three Ti-

containing silicatesa

Substrate Ti-MCM-41 Ti/SiO2 Ti/Aerosil

C (%) S (%) C (%) S (%) C (%) S (%)

22 90 51 71 60 55 57

23 73 80 71 82 61 88

3 82 73 86 84 83 83

24 62 90 63 89 63 92

aReaction conditions: mcatalyst ¼ 50mg; 1 mmol substrate; TBHP:

terpene molar ratio ¼ 1 : 1; CH3CN; 90 8C, 24 h; vtot mix ¼ 10mL.

Scheme 9.


Industrially, the reaction is catalyzed by ZnBr2. Solidalternatives are zeolites [39, 40] or amorphous silicaaluminas [41]. By carrying out the cyclization ofcitronellal in hydrocarbon solvent over the titaniumsilicates used for the epoxidation of terpenic alcohols,we obtained the following order of acidity: Ti-MCM-41 > Ti/Aerosil > Ti/SiO2. According to this order, thelowest epoxidation selectivities, particularly when theformation of a bicyclic ether is possible, were obtainedon the ordered catalyst.

The remarkable acidity of Ti-MCM-41 has beenexploited for the one-pot synthesis of isopulegol epoxidestarting from citronellal. Here again the choice of thesolvent was not trivial. The cyclization reactions, like allacidic reactions as already mentioned, are easily carriedout in apolar solvents, typically hydrocarbons, like n-heptane or toluene, whereas best results in epoxidationreactions are achieved in aprotic polar solvents, there-fore the use of a unique solvent is hardly feasible.Moreover, to avoid oxidation of unreacted citronellal byTBHP, the oxidant should be added after completeconversion of the starting material. Best results wereobtained by carrying out the first step in toluene andlater adding TBHP in acetonitrile. Conversion andselectivity were >98% for the cyclization reaction,whereas in the epoxidation, one 76% isopulegol wasconverted with 90% selectivity, thus giving an overallyield of 67% of isopulegol epoxide [42].

4. Conclusions

Since the study of organic chemistry began, it hasbeen intensively concerned with terpenes. Still, manytransformations in this field are carried out today byconventional, stoichiometric organic reactions. Here, wehave shown that selective reactions of terpenes can bepromoted by heterogeneous systems, often by verysimple ones.

The identification and quantification of by-products,even if formed in very low amounts, help understand theproperties of the catalyst and plan a more effective andselective transformation. Thus, acidic sites on thecatalyst surface can be selectively poisoned or can beexploited for synthetic purposes, to set up bifunctionalprocesses or to separate isomeric mixtures. Chemicalintuition is still needed to design and develop newsynthetic pathways and improve the selectivity ofreactions.

Acknowledgments

We acknowledge Consorzio InteruniversitarioINSTM for a fellowship to Federica Zaccheria. CNRAgenzia 2000 is acknowledged for financial support.

References

[1] K. Bauer, D. Garbe and H. Surburg, Common Fragrance and

Flavours Materials (Wiley-VCH, New York, 1997).

[2] H. Mimoun, Chimia 50 (1996) 620.

[3] K. Arata and K. Tanabe, Chem. Lett. (1979) 1017.

[4] J. Kaminska, M.A. Schwegler, A.J. Hoefnagel and H. van

Bekkum, Recl. Trav. Chim. Pays-Bas 111 (1992) 432.

[5] A.T. Liebens, C. Mahaim and W.F. Holderich, Stud. Surf. Sci.

Catal. 108 (1997) 587.

[6] P.J. Kunkeler, J.C. van der Waal, J. Bremmer, B.J. Zuurdeeg,

R.S. Downing and H. van Bekkum, Catal. Lett. 53 (1998) 135.

[7] J.L. Flores Moreno, L. Barakat and F. Figueras, Catal. Lett. 77

(2001) 113.

[8] N. Ravasio, M. Finiguerra, M. Gargano, Chem. Ind. Series,

Catalysis of Organic Reactions, F.E. Herkes (ed.) (Marcel Dekker,

New York, 1998) p. 513.

[9] G. Ramis, L. Yi, G. Busca and N. Ravasio, II Conv. Sc. Cons.

Chim. Mater., Firenze (I), 13–15 Feb. 1995, Book of Abstracts,

C23.

[10] R.L. Settine, G.L. Parks and G.L.K. Hunter, J. Org. Chem. 29

(1964) 616.

[11] K. Arata, S. Akutagawa and K. Tanabe, J. Catal. 41 (1976)

173.

[12] F. Zaccheria, R. Psaro, N. Ravasio and L. De Gioia, Chem. Ind.

Series, Catalysis of Organic Reactions, M. Ford (ed.) (Marcel

Dekker, New York, 2001) p. 601.

[13] L. Salles, A.F. Nixon, N.C. Russell, R. Clarke, P. Pogorzelec and

D.J. Cole-Hamilton, Tetrahedron: Asymmetry 10 (1999) 1471.

[14] W. Chrisman, J.N. Camara, K. Marcellini, B. Singaram, C.T.

Goralski, D.L. Hasha, P.R. Rudolf, L.W. Nicholson and K.K.

Borodychuk, Tetrahedron Lett. 42 (2001) 5805.

[15] E.E. Royals and L.L. Harrell, J. Am. Chem. Soc. 77 (1955) 3405.

[16] Y. Fujiwara, N. Masato and K. Igawa, JP 85-253708 to Toyo

Soda.

[17] N. Ravasio, M. Antenori, M. Gargano and P. Mastrorilli,

Tetrahedron Lett. 37 (1996) 3529.

[18] K.H. Schulte-Elte, P. Fankhauser and G. Ohloff, JP 77-124356 to

Firmenich.

[19] H. Masuda and S. Mihara, JP 84-155438 to Ogawa and Co.

[20] C. Chapuis and D. Jacoby, Appl. Catal. A 221 (2001) 93.

[21] J-j. Young, L-j. Jung and K-m. Cheng, Tetrahedron Lett. 41

(2000) 3415.

[22] N. Ravasio, V. Leo, F. Babudri and M. Gargano, Tetrahedron

Lett. 38 (1997) 7103.

[23] F. Boccuzzi, A. Chiorino, M. Gargano and N. Ravasio, J. Catal.

165 (1997) 140.

[24] F. Boccuzzi, A. Chiorino, G. Martra, M. Gargano, N. Ravasio

and B. Carrozzini, J. Catal. 165 (1997) 129.

[25] F. Boccuzzi, S. Coluccia, G. Martra and N. Ravasio, J. Catal. 184

(1999) 316.

[26] F. Zaccheria, R. Psaro, N. Ravasio, Recent Research and

Developments in Catalysis, G. Pandalai (ed.) (Research Signpost,

Trivandrum) 2 (2003) 23.

[27] G. Ohloff, W. Giersch, K.H. Schulte-Elte and C. Vial, Helv.

Chim. Acta 59 (1976) 1140.

[28] H. Masuda and S. Mihara, JP 85 65,879 to Ogawa and Co.

[29] N. Ravasio, R. Psaro, F. Zaccheria and S. Recchia, Chem. Ind.

Series, Catalysis of Organic Reactions, D. Morrel (ed.) (Marcel

Dekker, New York, 2003) p. 263.

[30] P. Linares, S. Salido, J. Altarejos, M. Nogueras and A. Sanchez,

Flavour and Fragrance Chemistry, V. Lanzotti and O. Tagliatela

Scafati (eds) (Kluwer Ac. Publ., Dordrecht, 2000) p.101.

[31] H. Mimoun, U.S. Patent 6,245,952 to Firmenich SA (2001).

[32] N. Ravasio, N. Poli, R. Psaro, M. Saba and F. Zaccheria, Top.

Catal. 13 (2000) 195.

[33] T. Maschmeyer, F. Rey, G. Sankar and J.M. Thomas, Nature 378

(1995) 159.


[34] C. Berlini, M. Guidotti, G. Moretti, R. Psaro and N. Ravasio,

Catal. Today 60 (2000) 219.

[35] A. Bhaumik and T. Tatsumi, J. Catal. 182 (1999) 349.

[36] J.C. van der Waal, P. Lin, M.S. Rigutto and H. van Bekkum,

Stud. Surf. Sci. Catal. 105 (1997) 1093.

[37] J.C. van der Waal, P. Lin, M.S. Rigutto and H. van Bekkum,

Appl. Catal. A: Gen. 167 (1998) 331.

[38] M. Guidotti, N. Ravasio, R. Psaro, G. Ferraris and G. Moretti,

J. Catal. 214 (2003) 247.

[39] J. Shabtai, R. Lazar and E. Biron, J. Mol. Catal. 27 (1984)

35.

[40] M. Fuentes, J. Magraner, C. De Las Pozas, R. Roque-Malherbe,

J. Perez Pariente and A. Corma, Appl. Catal. 47 (1989) 367.

[41] N. Ravasio, M. Antenori, F. Babudri and M. Gargano, Stud.

Surf. Sci. Catal. 108 (1997) 625.

[42] M. Guidotti, R. Psaro and N. Ravasio, Chem. Commun. (2000)

1789.


Mono- and Bifunctional Heterogeneous Catalytic Transformation of Terpenes and Terpenoids

Documents