A new, alternative, halogen-free synthesis for the fragrance compound Melonal using zeolites and mesoporous materials as oxidation catalysts

Article abstract of DOI:10.1016/j.jcat.2005.06.006

At present Melonal (2,6-dimethyl-5-hepten-1-al) fragrance is industrially produced by a Darzens reaction from 6-methyl-5-hepten-2-one, with ethylchloroacetate as reagent. We present here a novel halogen-free synthesis strategy that involves the chemoselective oxidation of Citral (3,7-dimethyl-6-octen-1-al), a common compound in the fragrance industry, with H₂O₂ and Sn-Beta or Sn-MCM-41 as catalysts. The performance of tin Lewis acid sites is compared with other potential catalytic sites, and it is found that aluminium Bronsted acid sites and zirconium or titanium Lewis acid sites are less efficient and selective than Sn. In the case of Ti, epoxidation by-products could be found. The re-usability of Sn-Beta zeolite is discussed, as is the heterogeneity of the reaction.

Full text of DOI:10.1016/j.jcat.2005.06.006

Journal of Catalysis 234 (2005) 96–100
www.elsevier.com/locate/jcat
A new, alternative, halogen-free synthesis for the fragrance compound
Melonal using zeolites and mesoporous materials
as oxidation catalysts
Avelino Corma ∗, Sara Iborra, María Mifsud, Michael Renz
Instituto de Tecnología Química, UPV-CSIC, Avda. los Naranjos s/n, 46022 Valencia, Spain
Received 27 April 2005; revised 7 June 2005; accepted 10 June 2005
Available online 14 July 2005
Abstract
At present Melonal (2,6-dimethyl-5-hepten-1-al) fragrance is industrially produced by a Darzens reaction from 6-methyl-5-hepten-2-one,
with ethylchloroacetate as reagent. We present here a novel halogen-free synthesis strategy that involves the chemoselective oxidation of
Citral (3,7-dimethyl-6-octen-1-al), a common compound in the fragrance industry, with H O and Sn-Beta or Sn-MCM-41 as catalysts. The
2
2
performance of tin Lewis acid sites is compared with other potential catalytic sites, and it is found that aluminium Brønsted acid sites and
zirconium or titanium Lewis acid sites are less efficient and selective than Sn. In the case of Ti, epoxidation by-products could be found. The
re-usability of Sn-Beta zeolite is discussed, as is the heterogeneity of the reaction.
2005 Elsevier Inc. All rights reserved.
Keywords: Baeyer–Villiger oxidation; Citral; Melonal; Solid Lewis acid; Tin centre
1. Introduction
We thought of a catalytic and halogen-free alternative
synthesis strategy (Scheme 2), where the one carbon ho-
mologation can be achieved by transformation of 6-methyl-
5-heptene-2-one (2) into 3,7-dimethyl-6-octen-1-al (3, Cit-
ral) by aldol condensation of ketone 2 with acetaldehyde.
Subsequent Baeyer–Villiger oxidation of aldehyde 3 affords
Melonal (Scheme 2). This new route is quite attractive from
a synthetic and industrial point of view, not only because it
does not involve any halogenated product, but also because
Citral (3) is inexpensive and readily available by synthe-
Melonal (2,6-dimethyl-5-hepten-1-al; 1) is used to intro-
duce melon and cucumber notes into fragrances because of
its powerful green melon- and cucumber-like odour [1]. In-
dustrially Melonal is synthesised by a Darzens condensation
of 6-methyl-5-hepten-2-one with ethylchloroacetate to yield
an α,β-epoxy ester (Scheme 1). This ester is then saponified
and decarboxylated to produce the desired aldehyde 1. In
general, the homologation of ketones or aldehydes to yield
an aldehyde with one additional carbon atom is a challenging
process that cannot be carried out in one step. For this rea-
son, a Darzens reaction using halogenated esters, followed
by saponification and decarboxylation, is used for this pur-
pose [2].
*
Scheme 1. Synthesis of Melonal (1) from 6-methyl-5-hepten-2-one (2) by
Darzens reaction with subsequent decarboxylation.
Corresponding author. Fax: +34-96-3877809.
E-mail address: acorma@itq.upv.es (A. Corma).
0021-9517/$ – see front matter
doi:10.1016/j.jcat.2005.06.006
2005 Elsevier Inc. All rights reserved.

A. Corma et al. / Journal of Catalysis 234 (2005) 96–100
97
cropore volume of 0.21 cm3 g−1 and BET surface areas of
450–475 m2 g−1
.
The two Sn-MCM-41 sample were synthesised accord-
ing to the following procedure [9]. An aqueous solu-
tion of hexadecyltrimethylammonium hydroxide/bromide
(C16TABr/OH) was mixed with a tetramethylammonium hy-
droxide solution (25%, Aldrich) and an aqueous solution of
SnCl4·5H2O (98%, Aldrich). After homogenisation, the sil-
ica (Aerosil, Degussa) was added with continuous stirring.
The final composition was the following: 1 SiO2:(0.16−4x)
C16TABr:4x C16TAOH:0.26 TMAOH:x SnCl4:24.3 H2O,
where x is either 0.040 or 0.008. C16TABr was partially ex-
changed in order to compensate for the OH− depletion pro-
duced by the incorporation of SnCl4 into the synthesis gel.
The homogeneous gel was sealed in Teflon-lined stainless-
steel autoclaves and heated at 135 ◦C under static conditions
for 24 h. The resulting solid product was recovered by fil-
tration, washed, and dried at 60 ◦C for 24 h. We removed
the occluded organic by heating the solid at 813 K for 1 h
in a flow of N2, followed by 6 h in air. The solid obtained
presents an XRD pattern typical for MCM-41 structure.
Scheme 2. Synthesis of Melonal (1) from 6-methyl-5-hepten-2-one (2) by
Baeyer–Villiger oxidation of the aldol condensation product with acetalde-
hyde.
sis [3] and by distillation from essential oils (lemongrass
oil or Litsea cubeba oil) [4]. However, it has to be pointed
out that this route presents a complication derived from the
fact that a tri-substituted double bond is present in the alde-
hyde chain, and industrial processes for carrying out the
Baeyer–Villiger oxidations use peracids that are also good
epoxidation agents and do not allow Melonal to be obtained
selectively [5,6]. Indeed, when we reacted Citral with meta-
chloroperbenzoic acid (m-CPBA), the selectivity for Mel-
onal or for its precursor 4 is only 33%.
Recently we presented Sn-Beta [7,8] and Sn-MCM-41 [9]
materials for the chemoselective Baeyer–Villiger oxidation
of cyclic ketones and aromatic aldehydes, with hydrogen
peroxide as oxidant. The feasibility of industrial processes
with Sn-Beta has been demonstrated successfully in the syn-
thesis of delta-decalactone, which is also a fragrance com-
pound. The zeolite catalyst was used in solvent-free condi-
tions with high substrate/catalyst ratios (up to 200 wt/wt),
and high turnover numbers (TONs) were achieved [10]. Here
we show that these materials can afford the syntheses of
Melonal by this new route, with high selectivities and with
hydrogen peroxide as oxidant. It is further demonstrated that
the tin Lewis acid sites are better catalytic sites than other
Lewis or Brønsted acid sites.
2.2. General procedure for the Baeyer–Villiger oxidation
Citral (0.5 g), aqueous hydrogen peroxide (50%, 1.5
equiv.), 3.0 g of solvent, and Sn-Beta catalyst (normally
50 mg) were stirred magnetically and heated to the de-
sired reaction temperature. The reaction was followed by gas
chromatography, and the products were identified by com-
parison with reference samples, by GC-MS spectroscopy, or
after purification by 1H NMR spectroscopy. For the isolation
of the saponified product, the unreacted hydrogen peroxide
was first decomposed with manganese(IV) oxide, followed
by treatment with catalytic amounts of sodium hydroxide.
2. Experimental
Hydrogen peroxide (50%), Melonal (1), 6-methyl-5-
hepten-2-one (2), and 3,7-dimethyl-2,6-octadienal (3) were
purchased from Aldrich. GC analyses were carried out with
a HP 5890 gas chromatograph equipped with a 25-m HP-5
column. GC-MS analyses for the identification of products
were carried out with an Agilent Technologies 6890N ap-
paratus coupled with a Mass Selective Detector Network.
1H NMR spectra were recorded with a Bruker spectrometer
at a frequency of 300 MHz and 13C spectra at a frequency of
75 MHz.
3. Results and discussion
Citral (3) was oxidised with hydrogen peroxide with a
series of solvents, such as acetonitrile, iso-propanol, tert-
butanol, tert-amylalcohol, and cyclohexanol (3.00 g). In all
cases the formate ester 4, which is the Baeyer–Villiger oxi-
dation product of Citral, and Melonal (1) are the main prod-
ucts obtained, with added selectivities of ꢀ95% when alco-
hols are used as solvent (Table 1). Since product 4 is easily
and quantitatively transformed into Melonal in the presence
of catalytic amounts of NaOH, we can say that it is possible
to produce with high selectivity (ꢀ95%) the homologation
of aldehydes through this new halogen-free chemical route.
When the yield of the different products is plotted versus
total conversion (Fig. 1), it can be seen that the enol ester
4 is a primary unstable product, whereas Melonal (1) is a
secondary stable product. This behaviour is consistent with
the formation of the formate enol ester of Melonal (4) by the
Baeyer–Villiger oxidation of Citral, whereas Melonal will be
the secondary product generated through its hydrolysis. The
2.1. Synthesis of the molecular sieves
Sn-Beta was synthesised with a procedure described in
the literature [8,11]. The Sn content (2.0 wt% as SnO2) was
determined by atomic absorption. The Sn-Beta zeolite was
calcined at 853 K for 3 h. The zeolite was highly crystalline,
and no peaks of SnO2 were found by XRD. Nitrogen ad-
sorption experiments on the calcined Beta samples gave an
isotherm very similar to that of pure silica Beta with a mi-

98
A. Corma et al. / Journal of Catalysis 234 (2005) 96–100
Table 1
Table 2
Baeyer–Villiger oxidation of aldehyde 3 in different solvents. Reaction con-
ditions: 0.5 g of aldehyde 3 (mixture of isomers, 3.7 mmol), 1.5 equiv. of
aqueous hydrogen peroxide (50%), 50 mg of Sn-Beta, 3.0 g of solvent,
Baeyer–Villiger oxidation of aldehyde 3 at different temperatures. Reaction
conditions: 0.5 g of aldehyde 3 (mixture of isomers, 3.7 mmol), 1.5 equiv.
of aqueous hydrogen peroxide (50%), 50 mg of Sn-Beta, 3.0 g of solvent,
and 1 h reaction time
◦
80 C reaction temperature, and 1 h reaction time
Entry
Solvent
Conversion
(%)
Product distribution
Entry
T
Solvent
Conver-
sion (%)
Product distribution
◦
1
4
2
Other
( C)
1
4
2
Other
1
2
3
4
5
Acetonitrile
iso-Propanol
tert-Butanol
tert-Amylalcohol
Cyclohexanol
40
33
40
29
49
13
17
10
1
78
79
85
95
75
0
0
2
0
5
9
4
3
4
0
1
2
3
4
5
6
7
8
80
90
100
80
100
100
120
tert-Amylalcohol
tert-Amylalcohol
tert-Amylalcohol
Cyclohexanol
29
36
53
49
52
1
95
0
0
0
5
5
2
0
3
4
4
5
0
5
7
6
11 85
10 85
20 75
18 72
8
5
20
Cyclohexanol
1-Methylcyclohexanol 52
1-Methylcyclohexanol 52
1-Methylcyclohexanol 53
83
89
a
120
12 71
14
a
5 min reaction time.
dation catalysed by the Sn-Beta catalyst. It has been estab-
lished by in situ IR spectroscopy that the carbonyl oxygen
coordinates with the Lewis-acid tin centre, causing a posi-
tive partial charge at the carbon atom of the carbonyl group.
Then, in the case of the α,β-unsaturated aldehydes, this par-
tial charge can be delocalised to the double bond (β posi-
tion), and consequently the initial step of the Weitz–Scheffer
epoxidation, a nucleophilic attack of the double bond, should
be facilitated. Nevertheless, the rate of this reaction is much
lower than the corresponding nucleophilic attack by hydro-
gen peroxide of the carbon atom in the carbonyl group, and
the selectivity for the epoxide is rather low (ꢁ5%).
Fig. 1. Yields of Melonal (1) (2), ester 4 (") and ketone 2 (Q) for the
Baeyer–Villiger oxidation over the conversion of Citral (3) in cyclohexanol.
In the different alcohol solvents, conversions in the range
of 30 to 50% have been observed. Regardless of the solvent
used, conversion was slightly increased, whereas selectiv-
ity remained at ∼95%, when we increased the temperature
to 100 ◦C. At 120 ◦C, 52% conversion was already achieved
after 5 min of reaction time (Table 2, entry 7), what corre-
sponds to a turnover number (moles converted per mole Sn)
of 257 and a turnover frequency (moles converted per mole
Sn per hour) of 3060. When the reaction time was prolonged
to 1 h at this temperature, no further conversion could be de-
tected, but a small decrease in selectivity could.
Selectivity for hydrogen peroxide was always above 98%,
and therefore the controlling reactant was always the Cit-
ral (3). Thus, the fact that total conversion of Citral was
not obtained could be ascribed to catalyst deactivation or
to catalyst inhibition due to the adsorption of products at
the catalytic sites in the Langmuir–Hinshelwood sense of
competing adsorption. This type of product inhibition has
been observed in other homogeneously [12] and heteroge-
neously [8] Lewis-acid-catalysed reactions.
Scheme 3. Product formation in the Baeyer–Villiger oxidation of alde-
hyde 3.
main by-product, ketone 2, is formed only in small amounts
and appears as a secondary product. This could be formed
(see Scheme 3) from the 2,3-epoxide 5 after opening with
hydrogen peroxide and the Grob fragmentation to give the
ketone 2.
We are saying here that the formation of the ketone 2 re-
quires the formation of an epoxide 5. However, it has been
proposed before that Sn-Beta does not catalyse the epoxida-
tion of olefins by hydrogen peroxide [7,8], and this is indeed
confirmed by the fact that we have not detected epoxidation
of the 6,7 double bond in the Citral molecule. However, for
the double bond in α,β position to the carbonyl group, as it
occurs in the case of Citral (3), the epoxide can be formed
(though in small amounts) by a Weitz–Scheffer-type epoxi-
As we have observed in the oxidation of cyclohexanone
with hydrogen peroxide in the presence of Sn-Beta, there
is a negative effect of water on the reaction rate by com-
petitive adsorption on Sn sites. In any case, a competitive
adsorption effect is more important in the case of the lac-
tone product [13]. When this competitive adsorption occurs,
the catalyst should not suffer a permanent deactivation, and
its activity should be restored simply by desorption of the

A. Corma et al. / Journal of Catalysis 234 (2005) 96–100
99
Table 3
Re-use of the catalyst for the Baeyer–Villiger oxidation of aldehyde 3. Re-
action conditions: 0.5 g of aldehyde 3 (mixture of isomers, 3.7 mmol),
1.5 equiv. of aqueous hydrogen peroxide (50%), 50 mg of Sn-Beta, 3.0 g
◦
of tert-amylalcohol as solvent, 100 C reaction temperature, and 1 h reac-
tion time in the first cycle. For the subsequent runs the amount of substrates
and solvent were scaled down with respect to the available amount of cata-
lyst
Scheme 4. Products formed with Ti-Beta/hydrogen peroxide and m-CPBA
in the oxidation of Citral (3).
Entry
Cycle
Conversion
(%)
Product distribution
1
4
2
Other
1
2
3
4
5
6
1
2
3
4
5
53
58
56
56
48
62
10
9
11
11
9
85
87
85
84
85
77
0
0
0
0
0
0
5
4
4
5
6
a
6
12
11
a
After activation by calcination.
Table 4
Scheme 5. Baeyer–Villiger oxidation of aldehyde 8. Reaction conditions:
0.5 g of Cyclocitral (8; 3.7 mmol), 1.5 equiv. of aqueous hydrogen peroxide
Baeyer–Villiger oxidation of aldehyde 3 by different oxidation systems or
oxidants. Reaction conditions: 0.5 g of aldehyde 3 (mixture of isomers,
3.7 mmol), 1.5 equiv. of aqueous hydrogen peroxide (50%), 50 mg of Beta
(50%), 50 mg of Sn-Beta or Sn-MCM-41 (each 2 wt% in SnO ), 3.0 g of
2
◦
tert-amylalcohol, 100 C reaction temperature, and a reaction time of 1 h.
◦
zeolite or Sn-MCM-41, 3.0 g of tert-amylalcohol, 100 C reaction tem-
perature, and 1 h reaction time. The reactions with meta-chloroperbenzoic
◦
double bonds. It has been proved there that when olefinic
groups are present in the reactant molecule, Al-Beta catalyst
deactivates readily either by formation of fairly stable unre-
active carbocations and/or by blocking of the pores by bulky
oligomers. This is indeed what can also occur in the case of
Citral/Melonal.
The performance of Zr-Beta is similar to or even better
than that of Sn-Beta in the Meerwein–Ponndorf–Verley re-
action [15,16]. However, in the Baeyer–Villiger oxidation of
Citral (3), Zr-Beta was less active and less selective than
Sn-Beta. It gives 28% conversion with a selectivity for the
desired 1 + 4 products of only 59% (Table 4, entry 3).
Titanium incorporated into the Beta structure resulted in
low conversion and low selectivity (Table 4, entry 4). Apart
from the Baeyer–Villiger oxidation, epoxidation is observed,
yielding products 6 and 7 (Scheme 4). The Si-Beta parent
system without metal incorporation showed no catalytic ac-
tivity (entry 5), and SnCl4 gave a very low conversion (en-
try 6).
acid (m-CPBA) were carried out at 100 C and conversion and selectivity
is given for 30 min reaction time
Entry Oxidant
Conver-
sion (%)
Product distribution
1 + 4
2
6
7
Other
1
2
Sn-Beta/H O
Al-Beta /H O
53
17
28
16
0
95
66
59
48
0
0
0
0
0
0
0
5
0
13
0
2
2
a
34
28
0
2
2
3
Zr-Beta/H O
2
2
2
2
4
Ti-Beta/H O
39 13
2
5
Si-Beta/H O
2
6
7
8
9
SnCl /H O
2 2
Sn-MCM-41 /H O
Sn-MCM-41 /H O
m-CPBA (0.3 equiv.) 19
m-CPBA (1.0 equiv.) 49
4
46
40
100
90
96
48
0
7
4
0
0
0
0
0
0
0
0
0
3
0
0
0
4
b
2
2
2
c
2
39 13
33 34
10
33
a
b
c
Si/Al ratio = 15.
9 wt% of SnO .
2
2 wt% of SnO .
2
product. Indeed, in the case of Melonal, the catalyst can be
re-used after each run with the usual simple activation at
200 ◦C and 2 Torr for 2 h (Table 3). A final re-calcination
provides a more active catalyst at the cost of some selectiv-
ity. The fact that the conversion per pass is 50% requires the
introduction of a separation and recycling step in the overall
industrial process.
We have compared the oxidation activity and selectivity
of Sn-Beta with other potential catalysts [14,15]. For in-
stance, after 1 h of reaction time at 100 ◦C (Table 4), an
Al-Beta catalyst gives only 30% of the conversion that was
obtained with Sn-Beta, namely 17% versus 53% (cf. Table 4,
entries 1 and 2), whereas the undesired ketone 2 was ob-
served in significant yield.
3.1. Sn-mesoporous material as catalyst
Interesting results were obtained with Sn-MCM-41, with
Citral conversions of 46 and 40% for tin contents of 9 and
2 wt%, respectively (Table 4, entries 7 and 8), with high
selectivities for the desired products. As already observed
for the Baeyer–Villiger oxidation of cyclic ketones [10], Sn-
Beta is intrinsically more active than Sn-MCM-41 (entries
1 and 8), provided that the substrate can diffuse through
the Beta channels without any steric restriction. On the
other hand, when the pore dimensions of the zeolite present
diffusional restrictions for the reactant, then Sn-substituted
mesoporous molecular sieves can be an alternative to ze-
olites [17]. We can see that when Cyclocitral (8; 2,6,6-
In the Baeyer–Villiger oxidation of aromatic aldehydes
with Al-Beta, an activity similar to that observed for Sn-
Beta has been found [14], except for substrates with isolated

100
A. Corma et al. / Journal of Catalysis 234 (2005) 96–100
trimethyl-1-cyclohexene-1-carboxaldehyde) is reacted with
Sn-Beta, the conversion decreases to 43%; however, the se-
lectivity for the desired products is only 36% (the missing
64% is the unusual regioisomer of the Baeyer–Villiger oxi-
dation). In contrast, Sn-MCM-41 (2 wt% in SnO2) converts
90% of the α,β-unsaturated aldehyde 8 into the correspond-
ing formate ester without any saponification to the ketone 10
(Scheme 5).
nancing this work. M.R. is grateful to the Spanish Ministry
of Science and Technology for a “Ramón y Cajal” Fellow-
ship.
References
[1] K. Bauer, D. Garbe, H. Suburg, Common Fragrance and Flavor Mate-
rials, Wiley–VCH, Weinheim, 1997, p. 15.
[2] J. March, Advanced Organic Chemistry: Reactions, Mechanisms and
Structure, fourth ed., Wiley, New York, 1992, p. 954.
[3] P. Short, Chem. Eng. News 82 (2004) Sep 6.
4. Conclusions
[4] K. Bauer, D. Garbe, H. Suburg, Common Fragrance and Flavor Mate-
rials, Wiley–VCH, Weinheim, 1997, p. 35.
[5] M. Renz, B. Meunier, Eur. J. Org. Chem. (1999) 737.
[6] G.R. Krow, Org. React. 43 (1993) 251.
[7] A. Corma, L.T. Nemeth, M. Renz, S. Valencia, Nature 412 (2001) 423.
[8] M. Renz, T. Blasco, A. Corma, V. Fornés, R. Jensen, L. Nemeth,
Chem. Eur. J. 8 (2002) 4708.
[9] A. Corma, M.T. Navarro, M. Renz, Chem. Commun. (2001) 2190.
[10] A. Corma, S. Iborra, M. Mifsud, M. Renz, M. Susarte, Adv. Synth.
Catal. 346 (2004) 257.
[11] S. Valencia, A. Corma, US Patent 5 968 473 (1999), to UOP.
[12] A. Behr, M. Fiene, Eur. J. Lipid Sci. Technol. (2000) 212.
[13] M. Boronat, A. Corma, M. Renz, G. Sastre, P.M. Viruela, Chem.
Eur. J., in press.
[14] A. Corma, V. Fornés, S. Iborra, M. Mifsud, M. Renz, J. Catal. 221
(2004) 67.
Here we have demonstrated a new strategy for the synthe-
sis of the fragrance compound Melonal. In our alternative
route from a different substrate, namely Citral, we avoid
halogenated elaborated reagents and use a green oxidant in
combination with a heterogeneous catalyst. The most ap-
propriate catalytic sites are Sn Lewis incorporated into a
siliceous material. They perform better than Brønsted or
other Lewis acid sites like titanium or zirconium. From an
industrial point of view, this new synthesis route is inter-
esting, since it has been demonstrated that the catalyst can
easily be separated and re-used in several reaction cycles.
[15] Y. Zhu, G. Chuah, S. Jaenicke, J. Catal. 227 (2004) 1.
[16] A. Corma, M.E. Domine, S. Valencia, J. Catal. 215 (2003) 294.
[17] A. Corma, S. Iborra, M. Mifsud, M. Renz, Arkivoc 9 (2005), ME-
1342H; http://www.arkivoc.org/ark/journal/2005/I09_Molina-Elguero/
1342/ME-1342H.pdf.
Acknowledgments
The authors thank Acedesa-Tagasako, CICYT (MAT
2003-07945-C02-01), and the Generalitat Valenciana for fi-