DOI: 10.1002/cssc.

200700140

Oxidation of Citronellal to Citronellic Acid by Molecular

Oxygen Using Supported Gold Catalysts
A. Martin,*[a] U. Armbruster,[a] D. Decker,[b] T. Gedig,[c] and A. Kçckritz[a]

The oxidation of citronellal to citronellic acid was studied using is, the C=C bond is attacked and a diol is primarily formed prob-
molecular oxygen as oxidant and gold-containing supported cat- ably via an epoxide intermediate. This side reaction can be sup-
alysts under aqueous conditions. The reactions were carried out pressed to a large extent by increasing the pH to 12 and by rais-
at 60–90 8C, with 200 Nml min 1 O2 and at pH values from 9 to ing the catalyst/oxygen ratio. Furthermore, detailed studies on
12. The alumina- or titania-supported catalysts were synthesized the influence of reaction time, pH value, reactant concentration
according to the deposition–precipitation procedure using urea and amount of catalyst show that citronellic acid can be ob-
or NaOH. Mechanistic studies have revealed that radical-initiated tained in over 90 % yield with total conversion of citronellal at
reactions lead to undesired by-products especially at pH < 9, that pH 12 and a temperature of 80 8C.

Introduction

Carboxylic acids are widely found in nature and are also impor- Catalysis by gold has rapidly developed over the last two de-
tant intermediates as well as final products on different scales, cades, especially when the guiding work of Haruta revealed its
from megatons of products in the petrochemicals industry to surprising catalytic activity for the low-temperature oxidation
several hundred kilograms in the fine chemical industry (for ex- of CO.[5–7] Recently, it was shown by various groups that gold is
ample, see Refs. [1, 2]). A wide range of C numbers and linear an attractive catalyst for many other reactions when supported
or branched compounds are commercially available. Various nanoparticles are used, for example, in the oxidation of glu-
production methods are used: for example, acetic acid is cose.[8–10] , other aldoses,[11] alcohols[12–15] or glyoxal.[16] Numer-
mainly produced by carbonylation of methanol, and salicylic ous reports deal with the preparation methods and characteri-
acid is manufactured from sodium phenolate and carbon diox- sation of supported gold catalysts,[12, 17, 18] however, the synthe-
ide, whereas long-chain carboxylic acids are often obtained by ses are not easy to reproduce sometimes. The size of the gold
hydrogenolysis of triglycerides from plant or animal oils and particle and the nature of the used support appear to be the
fats. Furthermore, alcohols or aldehydes can be oxidized to most important parameters with respect to the activity and se-
carboxylic acids, but present processes often suffer from the lectivity in the above-mentioned reactions. Typically, a low
application of environmentally harmful, stoichiometrically used gold content in the range of 0.5–1 % is necessary to obtain
oxidants such as permanganate, nitric acid or chromium active and selective gold catalysts. A higher gold content
oxides. However, modern environmentally benign oxidation would mostly promote the formation of large but inactive
processes use catalysts and hydrogen peroxide, molecular metal clusters. Further investigations have shown that the par-
oxygen or air as oxidant. ticle size can be controlled by choosing an appropriate prepa-
Citronellic acid (3,7-dimethyl-6-octenoic acid) is an acyclic ration method for the catalyst as well as a suitable support.[7]
monoterpene carboxylic acid. It is used in the cosmetic indus- The current challenging task of the production of fine chemi-
try as a valuable ingredient of various products, but nowadays cals using renewable feedstocks could benefit in particular
it is also known as a nature-identical preserving agent as a from the development of new catalytic systems containing bi-
result of its antimicrobial effect.[3] Terpenes are produced pri- metallic gold catalysts with, for example, platinum[19] or
marily from a wide variety of plants, particularly conifers, indium[20] as a second metal.
which were the original supplier of so-called turpentine oil[4]
containing mainly a- and b-pinene, but also by some insects, [a] Dr. A. Martin, Dr. U. Armbruster, Dr. A. Kçckritz
Leibniz-Institut fr Katalyse e.V. an der Universit t Rostock,
sponges and other animal organisms. The structure of citronel-
Außenstelle Berlin
lic acid is derived from that of nerol (3,7-dimethyl-2,6-octadien- Richard-Willst tter-Strasse 12, 12489 Berlin (Germany)
1-ol), a natural acyclic monoterpene found in lemongrass and Fax: (+ 49) 30-6392 4454
many essential oils. Citronellol (3,7-dimethyl-6-octenol) and cit- E-mail: [email protected]
ronellal (3,7-dimethyl-6-octenal) are the main ingredients of cit- [b] Dr. D. Decker
Clariant Produkte (Deutschland) GmbH
ronella oil (up to 60 %) besides geraniol (trans conformer of
Am Unisys Park 1, 65843 Sulzbach (Germany)
nerol). These chemicals are used extensively as basic com-
[c] Dr. T. Gedig
pounds in soap and the perfumery, cosmetics and flavourings Clariant Produkte (Deutschland) GmbH
industries. Industriestrasse, 50354 Hrth-Knapsack (Germany)

242 www.chemsuschem.org > 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim ChemSusChem 2008, 1, 242 – 248
Gold-Catalysed Oxidation of Citronellal

Herein, we report the investigation of the environmentally

Table 2. Oxidation runs using citronellal (1) as substrate.[a]
benign liquid-phase oxidation of citronellal (1) to citronellic
acid (2) using molecular oxygen as oxidant in the presence of Run Catalyst VACHTUNGRE(water) 1 t Conv. Yield
nanostructured gold catalysts. By such a route, citronellic acid [ml] ACHTUNGRE[mmol] [h] [%][b] [%][c]
can be synthesized using a sustainable oxidant and used in 1 1 % Au/TiO2ACHTUNGRE(urea) 400 50 8 97 45
the above-mentioned applications without the concern of 2 1 % Au/Al2O3CHTUNGRE(A urea) 400 50 7 100 58
there being present traces of transition-metal impurities in the [a] Reaction conditions: T = 80 8C, ambient pressure, 200 Nml O2/min,
final products as found for common manufacturing processes. pH 9.0–9.1, 1 g catalyst. [b] Conversion of citronellal (1) determined by
GC. [c] Yield of citronellic acid (2) determined by GC (after derivatisation
with trimethylsulfonium hydroxide).

Results and Discussion

Catalyst Characterisation
Transmission electron microscopy (TEM) analysis of the Au par- the yield of the desired product 2 was rather low and did not
ticles revealed a particle size of 1–5 nm in general. Figure 1 de- reach values above 40–60 %. Surprisingly, consumption of KOH
picts a TEM image for a 1 % Au/Al2O3ACHTUNGRE(urea) catalyst. All other was significantly higher than necessary for such a quantity of
acid. No other products were detected using HPLC analysis.
Some further tests carried out at different reaction times and
reaction temperature did not show any improvement in the
yield of 2. To the contrary, in particular at pH < 9 no product
was observed although the starting material was completely
converted. The same observation was made in tests in the ab-
sence of any catalyst even at pH 9, that is, HPLC analysis re-
vealed total conversion of 1 but no trace of the desired prod-
uct 2 or other detectable products at all.
Results of intensive NMR studies on reaction solutions from
catalysed and non-catalysed runs, especially at different pH
Figure 1. TEM image of fresh 1 % Au/Al2O3ACHTUNGRE(urea) catalyst showing Au parti-
cles with sizes of about 2–5 nm.
values, that were made after various reaction times help to
throw more light on parallel and consecutively running reac-
tions. Scheme 1 summarises the desired oxidation of citronellal
synthesized catalysts show similar particle sizes. TEM investiga- (1) to citronellic acid (2), as well as possible side-reaction steps
tions on spent catalysts did not show significant alterations in leading to undesired by-products. Route A indicates a hetero-
particle size. Additionally, the gold content of spent catalysts geneous catalytic oxidation in the presence of a catalyst,
after repeated use was studied by inductively coupled plasma where oxygen and 1 are adsorbed onto the gold-containing
(ICP), but no changes were recognised; that is, no significant catalyst, whereas route B corresponds to a radical mechanism
leaching appeared to occur during tests. Details of the cata- involving peroxo radicals 3 and peracid 4 in the absence of a
lysts and some characterisation data (Au content, BET surface heterogeneous catalyst. The mechanism for route A is assumed
area) are summarised in Table 1. to proceed similarly via peroxidic intermediates as described
for the oxidation of glucose to gluconic acid by nanoscale
gold catalysts.[9, 25] In the first step, electron-rich gold species
Table 1. Details of the studied catalysts. are formed by surface gold atoms and hydrated glucose
anions are generated in the alkaline reaction solution. These
Catalyst Synthesis Surface area Au content
route[a] ACHTUNGRE[m2 g 1][b] ACHTUNGRE[wt %][c] intermediates add molecular oxygen to create peroxidic-like
species, and a two-electron transfer from glucose to dioxygen
1 % Au/TiO2ACHTUNGRE(urea) urea 48 1
1 % Au/Al2O3CHTUNGRE(A urea) urea 164 1
is enabled. Finally, the formation of the acid is achieved by the
1 % Au/Al2O3CHTUNGRE(A NaOH) NaOH n.d. 0.6–0.7 cleavage of these species under release of hydrogen peroxide.
The latter is decomposed in the basic medium and contributes
[a] See Experimental Section for details. [b] BET surface area (n.d. = not de-
termined). [c] Au content determined by ICP. only marginally to the oxidation of the substrate.
However, oxygen radicals as well as intermediate radicals
generated via route B could also attack the C=C bond of the
reactant and also that of the formed acid. This might explain
Mechanistic Investigations
the total conversion of 1 and the relatively low yield of 2 ob-
Initial tests at pH 9 showed that citronellal (1) can be oxidized served at KOH consumption that seems to correspond to more
to citronellic acid (2) on gold-containing catalysts (e.g. 1 % Au/ or less complete acid stage formation. Such attack may proba-
TiO2ACHTUNGRE(urea) or 1 % Au/Al2O3ACHTUNGRE(urea)) using molecular oxygen as ox- bly lead to epoxide-like intermediates 5 and 6. The epoxida-
idant (Table 2). However, these initial tests with 1 as substrate tion of double bonds by molecular oxygen in the presence of
also showed a surprisingly rapid and complete conversion, but an aldehyde as co-oxidant has been reported both in the pres-

ChemSusChem 2008, 1, 242 – 248 > 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chemsuschem.org 243
 A. Martin et al.

tioned above. Alternatively, the

HPLC result could also point to
a keto–enol (9–10) tautomerism
as shown in Scheme 2. Howev-
er, it becomes clear from these
investigations that the unde-
sired by-products are oxygen-
ates formed by C=C bond reac-
tions and consecutive reactions
such as condensation or oligo-
merisation, and it is proven that
low pH values push the oxida-
tion reaction in this undesired
direction.
Further investigations on the
reaction mechanism using other
reactants with similar structures
gave interesting results and
support the deliberations made
earlier. Two catalytic oxidation
tests were conducted using 7-
hydroxycitronellal (11) and ter-
pinolene (12) as reactants at
Scheme 1. Oxidation of citronellal (1) to citronellic acid (2) with molecular oxygen in the presence of a gold cata- 80 8C, pH 9 and a reaction time
lyst (route A) or by radical reaction (route B) leading to undesired by-products. of 5 h; substrate 11 does not
possess a C=C bond, while com-

ence and absence of catalysts.[21, 22] Such intermediate generat-

ed epoxide-like structures (5, 6) can be converted by base-cat-
alysed solvolytic cleavage into dihydroxy species 7 and 8 that
can undergo consecutive reactions such as oligomerisation
through nucleophilic ring-opening of the epoxides. Further-
more, the formation of cyclic esters, that is, lactones, seems
also to be conceivable but no evidence of their presence was
found.
Scheme 2. Keto–enol tautomerism of double-bond-oxidized products.
Initial trials to identify possible side products in the reaction
mixtures from the oxidation runs by LC-MS were not success-
ful; it is thought that inorganic ions present in the mixtures pound 12 lacks an aldehyde group but contains C=C bonds
showed a disturbing effect. However, the analysis of a pur- (Scheme 3). The reaction mixtures were again analyzed by
chased sample of 2 (98 % purity) revealed impurities that were HPLC and NMR spectroscopy (Table 3, runs 1 and 2). Com-
identified as epoxide (6; tr = 2.9 min, Mr = 186 g mol 1) and diol pound 11 reacted very selectively to give the corresponding 7-
(8; tr = 3.54 min, Mr = 204 g mol 1; see Scheme 1). It seems hydroxycitronellic acid as confirmed by NMR spectroscopy and
likely that these compounds were formed as a result of oxida- comparison of HPLC data with that of an authentic sample. For
tion by ambient air and storage, even in the absence of any comparison, a further test was carried out in the absence of
catalyst. These two oxygen-containing compounds showed re- catalyst whereby low conversion of 11 and low yield of the
tention times similar to those of the unknown by-products acid were obtained as compared to the catalyzed run, howev-
from reaction mixtures, that is, the formation of such oxygen- er, no further by-products were found. This result shows that
ates is very likely.
Additionally, a separate synthesis of the diol 8 starting from
2 for comparison was attempted according to reference [23]
using diluted permanganate solution as oxidant. However, the
isolated reaction product was neither the epoxide 6 nor the
expected diol 8, but instead a consecutively formed ketone 9
was revealed by NMR spectroscopy. Whereas the NMR signals
suggested a pure compound, HPLC analysis showed two peaks
with tr = 2.82 min and 3.45 min, which were very similar to
those found during investigation of commercial 2 as men- Scheme 3. Terpenes used for further mechanistic experiments.

244 www.chemsuschem.org > 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim ChemSusChem 2008, 1, 242 – 248
Gold-Catalysed Oxidation of Citronellal

Variation of pH
Table 3. Oxidation reactions carried out under various conditions with
1 % Au/TiO2ACHTUNGRE(urea) as catalyst.[a] The pH values in the reaction mixture were adjusted in the
Run pH Cat. VACHTUNGRE(water) Substrate t Conv. Yield above-described manner by adding continuously KOH and
[g] [ml] ACHTUNGRE[mmol] [h] [%] [%] keeping the pH constant. The catalytic results on the influence
1 9 1 400 11 (50) 5 100 [b]
of the pH value of the aqueous reaction mixture on the oxida-
2 9 1 400 12 (50) 5 100 [c] tion of 1 using a 1 % Au/Al2O3ACHTUNGRE(urea) catalyst are summarised in
3 9 1 400 1 (50) 5 100 48.4 Table 4. The tests were carried out at 80 8C and 200 Nml min 1
4[d] 9 1 400 1 (50) 5 100 62.3
5[e] 9 0.1 40 1 (5) 5 99 3.3
6[f] 9 0.1 40 1 (5) 5 99 3.1
Table 4. Influence of pH on the conversion of citronellal (1) into citronel-
[a] Reaction conditions: T = 80 8C, ambient pressure, 200 Nml min 1 O2,
lic acid (2) over 1 % Au/Al2O3ACHTUNGRE(urea).[a]
1 % Au/TiO2ACHTUNGRE(urea). [b] Product: 7-hydroxycitronellic acid (no quantification
carried out). [c] Various oxidation products (no quantification carried out). Run pH Conv. [%] Yield [%]
[d] With addition of vitamin E (2 mmol). [e] Carried out in a 100-mL batch
reactor; the autoclave was additionally secured with Ar. [f] Carried out in 1 9 100 48.4
a 100-mL batch reactor. 2 10 100 64.4
3 12 100 84.0
1
[a] Reaction conditions: T = 80 8C, ambient pressure, 200 Nml min O2, 1
(50 mmol), H2O (400 mL), t = 5 h, 1 % Au/Al2O3ACHTUNGRE(urea) (1 g).

the oxidation also proceeds in the absence of a catalyst al-

though the reaction is much slower. Furthermore, the radical O2, at a concentration of 1 of 50 mmol in 400 mL distilled
double-bond oxidation via the epoxide and diol proceeds water and a reaction time of 5 h. It can be seen that an in-
faster than oxidation of the carbonyl group in 1 in the absence creasing pH value up to 12 is advantageous for the yield of 2,
of catalysts as mentioned earlier. The conversion of terpinolene which increases to 84 %. However, as discussed above, at lower
showed a variety of oxidation products containing COC- and pH values the yield of 2 is less then 50 % and larger amounts
COH structures as revealed by 13C NMR spectra (spectral details of by-products may be formed. As the alumina support does
are given in the Experimental Section). For comparison, citro- not exhibit long-term stability at pH 12, we plan to use the
nellal (1) underwent complete conversion also, but acid 2 was Au/TiO2 catalyst in further oxidation experiments at this pH
formed in 48 % yield only; neither known nor other detectable value.
products were further observed by HPLC (Table 3, run 3). These
results again clearly support the assumption that oxidation of
the C=C bond is responsible for the formation of non-selective Variation of Reactant Concentration and Amount of Catalyst
products probably by autoxidation and/or a radical oxidation Table 5 summarises the results from the oxidation of 1 with
mechanism. changing reactant concentration and increasing amount of cat-
To confirm that the reaction on the C=C bond proceeds by a alyst under similar conditions as applied in the tests discussed
radical mechanism, vitamin E (a-tocopherole) was applied as a above. The increase in the amount of catalyst used from 1 g to
radical scavenger in a separate catalyzed reaction using 1 as 3 g (Table 5, run 1 vs run 2) lead at pH 9 to an increase in the
reactant. Addition of vitamin E to the reaction mixture at pH 9 yield of 2 by approximately 20 %, demonstrating the catalytic
lead to the expected increase in yield of 2 (62 %; see Table 3, effect of the used Au-containing solid that suppressed radically
run 4) as compared to the reaction in the absence of such a steered side reactions. On the other hand, an increase in the
radical scavenger (run 3). The test proves the responsibility of reactant concentration by a factor of four (Table 4, run 3 vs
radicals for the C=C bond oxidation and the undesired side re- Table 5, run 3) revealed the expected drop in the yield of 2
actions. Otherwise, the formation of the desired acid (2) was over a similar reaction time; for example, with 200 mmol 1 at
only observed in the presence of the catalyst, that is, its forma- pH 12, the product 2 was formed with only 52 % yield. This
tion does not follow a radical mechanism. It seems to be possi-
ble that the oxidation may proceed by transfer of an electro-
philic oxygen atom in the presence of the gold catalyst. Table 5. Influence of the concentration of citronellal (1) and amount of
1 % Au/Al2O3ACHTUNGRE(urea) on the yield of citronellic acid (2).[a]
Furthermore, two additional tests were carried out in smaller
reactors in the absence of oxygen (Table 3, runs 5 and 6). Addi- Run pH Cat. VACHTUNGRE(water) 1 t Conv. Yield
tionally, the autoclave was secured with argon in one of the [g] [ml] ACHTUNGRE[mmol] [h] [%] [%]
tests (run 5). Despite a complete conversion of 1, only traces of 1 9 1 400 50 5 100 48.4
2 were found probably as a result of a contaminant in the re- 2 9 3 400 50 5 100 70.6
actant. No other products were observed by HPLC. It can be 3 12 1 400 200 4 100 52.3
4 12 4 400 200 4 100 84.8
assumed that radical polymerisation and/or aldol condensa-
5[b] 12 10.8 269 500 4 100 92.0
tions occur in these tests.
1
[a] Reaction conditions: T = 80 8C, ambient pressure, 200 Nml min O2.
[b] 1 % Au/Al2O3ACHTUNGRE(NaOH) used as catalyst.

ChemSusChem 2008, 1, 242 – 248 > 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chemsuschem.org 245
 A. Martin et al.

result points in the same direction as revealed earlier. Further this time consecutive reactions as discussed above become im-
increases in the amount of reactant (500 mmol) or catalyst (up portant, that is, the formed product 2 undergoes further oxida-
to 10.8 g) resulted in a yield of 2 above 90 %, and simultane- tions, in particular, at the C=C bond. Moreover, and as dis-
ously the yield of 2 per batch could be increased by a factor of cussed above, by running the experiment at pH < 9 such unde-
20. sired reaction paths dominate the reaction.
A reasonable explanation for these observations is based on
reaction kinetics as expressed by common power laws (rate =
keff P ci ; where keff corresponds to the effective rate constant, Variation of the Catalyst
and P ci corresponds to the product term): considering con-
Mainly two types of Au/Al2O3 catalysts were used that differed
centration dependency, an increase in the substrate/catalyst
in the procedure by which they were prepared, by applying
ratio is expected to slow down the overall conversion. Similarly,
either urea or NaOH. In general, catalysts prepared with urea
if the substrate and catalyst concentrations increase while the
contain more or less the offered amount of Au, while those
oxygen flow rate is fixed at 200 Nml min 1, the probability for
prepared with NaOH have a lower Au content. The following
the desired pathway is favoured at the cost of concurrent radi-
conditions were chosen for a comparison of the two catalysts:
cal oxidation. Indeed, the radical oxidation may only play a sig-
50 mmol 1, 80 8C, 200 mL min 1 O2, pH 12. The results (data
nificant role in the presence of an excess of oxygen and so cit-
normalised to 10 mg of Au) reveal that the catalyst prepared
ronellal (1) probably should be oxidized in the desired manner
by the urea route leads to significantly higher yields of 2
at lower oxygen concentrations and high catalyst amounts.
(92 %) as compared to the catalyst obtained by the NaOH
Further optimisation of the substrate/oxygen/catalyst ratio
route (67 % yield). However, this effect is not completely under-
should lead to an improved product batch yield.
stood and has to be investigated in more detail in future. As
Besides the result from HPLC analysis (92 % yield of 2), the
recently reported,[24] NaOH-based synthesis may lead to smaller
products of this batch (Table 5, run 5) were quantitatively re-
particles as compared to the urea route and could explain the
covered by concentrating the product mixture under vacuum,
easier leaching of Au from the support.
extracting the organics using dichloromethane and further re-
moval of the solvent. The crude yield amounted to around
94 %. The amount of 2 in the crude product was additionally Conclusions
determined by 1H NMR spectroscopy using 2.6-dimethylnaph-
thalene as internal standard. The signals at d = 5.06 ppm (2) The investigations have shown that citronellal (1) can be oxi-
and 7.29 ppm (standard) were selected for quantitative evalua- dized to citronellic acid (2) using oxygen as the oxidant in the
tion. The yield of 2 determined by this method was about 87– presence of Au-containing catalysts in a batch reactor. Citronel-
90 %. This comparison of different analysis methods showed a lal can be completely converted, and the desired carboxylic
good reliability of the results obtained by HPLC. acid is formed in high selectivity. The used Au catalysts show a
significant shift of selectivity towards the desired product citro-
nellic acid. Intensive studies on the reaction mechanism re-
vealed a competing side reaction, that is, radical epoxidation
Variation of Reaction Time
at the C=C bond of citronellal leading to by-products. The un-
The variation of reaction time was studied in a test series at desired side reaction can be suppressed by adding a radical
pH 12 with 50 mmol 1 using 1 g of 1 % Au/Al2O3ACHTUNGRE(urea) as cata- scavenger to the reaction mixture, increasing the amount of
lyst. The plotted data (Figure 2) were obtained from individual catalyst and pH value. In particular, the results from tests in
runs that were stopped and worked up at the specified reac- which the initial concentrations of reactants were varied show
tion time. The yield of 2 increases up to a reaction time of 4 h, a strong influence of the ratio of catalyst and oxygen on the
then the yield decreases again. It can be concluded that after selectivity. The target reaction on the aldehyde group is pro-
moted in the presence of Au catalyst, whereas the side reac-
tion on the C=C bond seems to proceed non-catalytically and
becomes dominant when a sufficient surplus of oxygen is
available in the liquid phase. These observations show that oxi-
dation at low oxygen concentrations with high catalyst con-
centration might be favourable in terms of the yield of citro-
nellic acid. The highest yield of isolated citronellic acid, up to
90 % (HPLC yield 92 %), was obtained at complete conversion
of 1, at pH 12 and a reaction temperature of 80 8C.

Experimental Section
Figure 2. Yield of citronellic acid (2) as a function of reaction time on 1 % Preparation of the catalyst: Catalytic runs were mainly carried out
Au/Al2O3ACHTUNGRE(urea). Reaction conditions: 1 (50 mmol), 80 8C, 200 mL min 1 O2, using 1 wt % Au-containing catalysts on different supports.[12, 13]
pH 12, 1 g of catalyst. Such catalysts were prepared as follows:

246 www.chemsuschem.org > 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim ChemSusChem 2008, 1, 242 – 248
Gold-Catalysed Oxidation of Citronellal

1) 1 % Au/Al2O3ACHTUNGRE(NaOH): Al2O3 (Puralox HP 14/150, Sasol; 30 g) was mation. The initial volume of liquid was about 400 mL in each run.
added to distilled water (600 mL) in a 2-L beaker equipped with In general, the tests were carried out over a temperature range
pH control, and the mixture was stirred for 15 min at ambient tem- from 60 to 90 8C under ambient pressure.
perature. The suspension was then heated to 70 8C during 20 min
under stirring. A second solution containing HAuCl4·3 H2O (900 mg) The reactor was first filled with distilled water (200 mL), and the
in H2O (450 mL) was added dropwise to the suspension at pH 7 calculated amount of citronellal (1; Alfa Aesar) and water (200 mL)
over approximately 2 h. The pH value of the mixture was adjusted were subsequently added. The emulsion was heated to reaction
by using 0.1 n NaOH, then the mixture was stirred at 70 8C for 1 h temperature (in general, 80 8C) under stirring (1200 rpm), and KOH
and cooled down to room temperature under stirring. Finally, mag- was added to adjust the pH (in general, from pH 9 to 12). The cata-
nesium hydrogen citrate (MgHC6H5O7·5 H2O; 4.172 g) was dissolved lyst was added at t = 0, and molecular oxygen was metered (typi-
in distilled water (100 mL) and the pH of the solution was adjusted cally 200 Nml min 1) at atmospheric pressure. Aqueous KOH (1 n in
to pH 7 using 0.1 n NaOH. This solution was then added to the sus- general) was fed automatically as mentioned above to maintain
pension at once under stirring, and the reaction mixture was the desired pH (accuracy of  0.1). The consumption of KOH was
cooled down under stirring for 1 h. Centrifugation, washing with registered continuously by means of a balance throughout the run
water and drying first at room temperature and later at 50 8C for and indicated the generation of acidic compounds including the
4 h led to the precursor material, which was calcined under air at desired carboxylic acid. The reaction was stopped by shutting off
250 8C for 3 h (heating rate 1 K min 1) to give 1 % Au/Al2O3ACHTUNGRE(NaOH). the stirrer and the oxygen supply after the defined reaction time.
In most runs, HPLC was used as the analytical tool for quantifica-
2) 1 % Au/Al2O3ACHTUNGRE(urea): Al2O3 (Puralox HP 14/150, Sasol; 10 g) was tion (see below). For this purpose, the content was discharged and
added to distilled water (200 mL) in a 1-L triple-necked round-bot- the reactor was flushed with ethanol to dissolve any remnants. The
tomed flask equipped with pH control, and a solution of catalyst was separated from the liquid by filtering and washed
HAuCl4·3 H2O (200 mg) in H2O (200 mL) was added under stirring at again with ethanol. The ethanolic phases (ca. 600 mL) were com-
room temperature. A solution of urea (750 mg) in distilled water bined and acidified with diluted HCl to pH 3 to give the free form
(2 mL) was then added to the reaction mixture, and the suspension of citronellic acid (2) from its potassium salt, and then the solution
was heated to 70 8C and stirred at this temperature overnight. was completed to 1000 mL with ethanol and used for HPLC analy-
After approximately 24 h, a solution of urea (330 mg) in distilled sis.
water (2 mL) was added to the suspension, and the mixture was
Quantitative isolation of the formed citronellic acid was carried out
stirred again overnight at 70 8C. The procedure was repeated again
according to the following procedure: The aqueous reaction solu-
after about 24 h. The suspension was then cooled down to room
tion of a run at pH 12 (see Table 4, run 3, for reaction conditions),
temperature under stirring, and a solution of MgHC6H5O7·5 H2O
which was removed from the reactor, was extracted with CH2Cl2
(927 mg) in distilled water (50 mL) at pH 7 adjusted using 0.1 n
(3 L 50 mL). Then, the aqueous phase was acidified with 2 n HCl to
NaOH was added to the suspension at once under stirring. Centri-
pH 3 and again extracted with CH2Cl2 (1 L 100 mL and 2 L 50 mL).
fugation, washing with water (3 L 200 mL) and drying first at room
The organic phase was treated with saturated NaHCO3 solution,
temperature (17 h) and later at 50 8C for 4 h led to the precursor
then H2O (2 L 50 mL), and dried with Na2SO4. The solvent was
material, which was calcined under air at 250 8C for 3 h (heating
evaporated, and the crude citronellic acid was purified by column
rate 1 K min 1) to give 1 % Au/Al2O3ACHTUNGRE(urea).
chromatography (silica gel, hexane/ethyl acetate, gradient 5:1 to
3) 1 % Au/TiO2ACHTUNGRE(urea): A titania-supported Au catalyst was synthe- 1:1) to give 2 (6.8 g, 80 % yield).
sized in the same way as described above for urea route (2), Besides the runs with citronellal (1), also 7-hydroxycitronellal (3,7-
except that TiO2 (P25, Degussa; 10 g) was used instead of alumina. dimethyl-7-hydroxyoctanal) and terpinolene (1-methyl-4-propan-2-
ylidenecyclohexene) were included in the set of investigations to
Characterisation of the catalyst: Inductively coupled plasma optical
investigate their reactivity under similar reaction conditions.
emission spectroscopy (ICP-OES; Optima 3000 XL, Perkin–Elmer)
was used for quantification of deposited gold in the catalyst sam- Product analysis: A HPLC routine was elaborated using a HP Series
ples. From experience, the deposition of gold on alumina using 1050 chromatograph with a UV detector and a YMC Hydrosphere
the urea method leads to a more or less complete recovery of the C18 column (150 mm L 4.6 mm) at 30 8C. A mixture of acetonitrile
added gold amount on the support. The application of NaOH, on (60 %) and 0.1 % aqueous trifluoroacetic acid solution (40 %) was
the other hand, produces materials with only 60–65 % of the of- used as the mobile phase (flow rate 0.5 mL min 1). The starting ma-
fered gold amount. Furthermore, some recovered used catalysts terial 1 and the product 2 were well separated, and 1-methylcyclo-
were analyzed for possible loss of gold during catalytic runs. BET hexene was used as internal standard for quantification. Calibration
surface areas were determined using a Micromeritics Gemini 2375 data of mixtures with known concentrations of reactant (1) and
III apparatus at liquid-nitrogen temperature. Parent catalyst sam- the desired product (2) served as the basis for evaluation of chro-
ples were characterised by TEM using a Philips CM20 ST instrument matograms and calculation of conversion and yield. Product selec-
equipped with an EDXS (EDAX, PV9900) operating at 200 kV, which tivity is defined as the product yield divided by the conversion.
showed an Au particle size of less than 10 nm in general. The sam-
ples were prepared by depositing the catalysts directly onto References samples of 2 for analytical use were obtained from Al-
copper grids with lacey carbon. drich. In some tests, small amounts of by-products were observed
by HPLC that were not investigated further. Additionally, a main
Catalytic setup and runs: In general, all the catalytic test runs were side reaction was observed which ran consecutively and also in
carried out in aqueous phase using a 1-L glass autoclave (BNCHI) parallel especially at low pH values leading to C=C bond oxidation
equipped with double jacket heater, stirrer, oxygen inlet dipping and non-selective oxidations. These transformations yielded con-
into the liquid, pH electrode and base (KOH) inlet to maintain a densation products and probably oligomers, as discussed in detail
constant pH. The pH of the reaction mixture was automatically below. Some analyses were also carried out using LC-MS (API 4000,
controlled using a closed-loop system with a HPLC pump for me- Applied Biosystems), especially to detect and identify side prod-
tering KOH to compensate for the pH decrease caused by acid for- ucts.

ChemSusChem 2008, 1, 242 – 248 > 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chemsuschem.org 247
 A. Martin et al.

In addition, off-gas measurements were carried out during some Acknowledgement

catalytic tests to check the possible formation of carbon oxides.
Gaseous samples were repeatedly collected with a glass cylinder
The authors thank Mrs. R. Bienert and Mrs. J. Kubias for experi-
(volume 250 mL) that was connected to the outlet of the auto-
mental assistance, Dr. A. Lehmann (BAM) for LC-MS measure-
clave. These samples were analyzed by an offline GC (Shimadzu
GC-14B; combination of two packed columns (molecular sieve 13X, ments and Dr. M. Kant for helpful discussions in the interpreta-
3 m L 0.75 mm i.d.; Porapak N, 1 m L 0.75 mm i.d.) with a valve- tion of NMR spectra.
switching program, TCD). However, the registered and accumulat-
ed amount of CO and CO2 was rather less, that is, selectivity was
Keywords: gold · heterogeneous catalysis · oxidation ·
below 1–2 %. These values were confirmed by precipitation of the
formed CO2 as BaCO3. The gas stream from the reactor outlet was
supported catalysts · terpenoids
passed through a solution of 0.2 m Ba(OH)2, and the molar amount
[1] H. A. Wittcoff, B. G. Reuben, Industrial Organic Chemicals, Wiley, New
of the formed barium carbonate was determined gravimetrically.
York, 1996.
Owing to such low amounts of carbon oxides, further data is nei- [2] K. Weissermel, H.-J. Arpe, Industrielle Organische Chemie, VCH, Wein-
ther given below nor discussed. heim, 1994.
1 [3] Dr. A. Rieks GmbH, DE 10335634, 2003.
H and 13C NMR spectroscopy (Varian UNITYplus-300) measure-
[4] Ullmann’s Encyclopedia of Industrial Chemistry, 4th ed., Vol. 22 (Eds.: F.
ments were carried out to identify the target product as well as Ullmann et al.), Verlag Chemie, Weinheim, 1982, p. 559.
possible by-products. The solvent was carefully removed from 10- [5] M. Haruta, Catal. Today 1997, 36, 153.
mL samples of the reaction solutions pre-treated for HPLC analysis, [6] M. Haruta, M. Date, Appl. Catal. A 2001, 222, 427.
and the residue was examined by NMR spectroscopy. Chemical [7] M. Haruta, Gold Bull. 2004, 37, 27.
shift values are given in ppm; coupling constants are given in Hz. [8] L. Prati, M. Rossi, J. Catal. 1998, 176, 552.
[9] M. Comotti, C. Della Pina, E. Falletta, M. Rossi, Adv. Synth. Catal. 2006,
Oxidation of citronellal at pH 9: In 13C NMR spectra, beside the sig- 348, 313.
nals for citronellic acid (compared with an authentic sample), a [10] A. Mirescu, H. Berndt, A. Martin, U. PrQsse, Appl. Catal. A 2007, 317, 204.
number of signals were observed in the range of aliphatic carbons [11] A. Mirescu, U. PrQße, Appl. Catal. B 2007, 70, 644.
at d = 47–42 ppm, in the range of COC and COH groups at d = 56– [12] H. Berndt, I. Pitsch, S. Evert, K. Struve, M.-M. Pohl, J. Radnik, A. Martin,
78 ppm, and additional signals for COOH groups at d = 174.3, Appl. Catal. A 2003, 244, 169.
[13] H. Berndt, A. Martin, I. Pitsch, U. PrQsse, K. D. Vorlop, Catal. Today 2004,
174.33 and 178.0 ppm.
91–92, 191.
Oxidation of citronellic acid with KMnO4 :[23] 1H NMR ([D6]DMSO): [14] S. Demirel-GQlen, M. Lucas, P. Claus, Catal. Today 2005, 102–103, 166.
d = 0.87 (d, 3 H, J = 6.0; CH3CH) 1.16 (2 singlets, 6 H; 2 CH3), 1.31 (m, [15] L. Prati, F. Porta, Appl. Catal. A 2005, 291, 199.
[16] S. Hermans, M. Devillers, Catal. Lett. 2005, 99, 55.
1 H; CH2), 1.47 (m, 1 H; CH2), 1.81 (m, 1 H; CH), 1.99 (dd, 1 H, J =
[17] R. Zanella, S. Giorgio, C. R. Henry, C. Louis, J. Phys. Chem. B 2002, 106,
14.9; CH2COOH) 2.20 (dd, 1 H, J = 14.9; CH2COOH), 2.64 (m, 2 H; 7634.
CH2C=O), 5.20 (br s, 1 H; OH), 11.99 ppm (br, 1 H; COOH). 13C NMR [18] R. Zanella, L. Delannoy, C. Louis, Appl. Catal. A 2005, 291, 62.
([D6]DMSO): d = 19.5, 26.6 (two signals), 29.5, 29.9, 33.0, 41.3 (7 C: [19] S. Demirel, K. Lehnert, M. Lucas, P. Claus, Appl. Catal. B 2007, 70, 637.
CH3, CH2, CH), 76.0 (COH), 174.0 (COOH), 215.7 ppm (C=O). [20] P. Claus, Appl. Catal. A 2005, 291, 222.
[21] N. Mizuno, H. Weiner, R. G. Finke, J. Mol. Catal. A 1996, 114, 15.
Oxidation of 7-hydroxycitronellal: 1H NMR ([D6]DMSO): d = 0.87 (d, [22] M. G. Clerici, P. Ingallina, Catal. Today 1998, 41, 351.
3 H, J = 6.6; CH3CH) 1.05 (m, 6 H; 2 CH3), 1.10–1.34 (m, 6 H; 3 CH2), [23] F. W. Semmler, Ber. Dtsch. Chem. Ges. 1893, 26, 2256.
1.81 (m, 1 H; CH), 1.98 (dd, 1 H, J = 14.9; CH2COOH), 2.19 (dd, 1 H, [24] J. Radnik, L. Wilde, M. Schneider, M.-M. Pohl, D. Herein, J. Phys. Chem. B
J = 14.9; CH2COOH), 4.04, (br s, 1 H; OH), 11.96 ppm (br, 1 H; 2006, 110, 23688.
COOH). 13C NMR ([D6]DMSO): d = 19.6, 21.3, 29.29, 29.31, 29.8, 36.9, [25] P. Beltrame, M. Comotti, C. Della Pina, M. Rossi, Appl. Catal. A 2006, 297,
41.5, 43.8 (8 C: CH3, CH2, CH), 68.8 (COH), 174.0 ppm (COOH). 1.

Oxidation of terpinolene: In 13C NMR spectra ([D6]DMSO), multiple Received: November 9, 2007
signals were observed in the range of d = 20–36 (CCC), 65–77 Revised: December 12, 2007
(COC, COH), 112–147 ppm (C=C). Published online on February 12, 2008

248 www.chemsuschem.org > 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim ChemSusChem 2008, 1, 242 – 248