Synthesis of primary amines from the renewable compound citronellal via biphasic reductive amination

Article abstract of DOI:10.1016/j.molcata.2015.04.006

The reductive amination of the natural product citronellal with ammonia is presented as a new and atom economic way to its primary amine derivatives. The aqueous ammonia phase contains the homogeneous catalytic system [Rh(cod)Cl]₂/TPPTS in a biphasic solvent system, whereas the starting material and the products remain in the apolar solvent phase. This concept supresses side reactions effectively, achieving a high yield of primary amines of up to 87%. Systematic investigations demonstrate that the cleavage of the secondary imine as an undesired by-product is necessary in achieving high selectivites, which can be controlled by the reaction conditions. Surfactants, ionic liquids or native cyclodextrins and their derivates prove to be useful phase transfer agents for optimising the interaction between the organic and the aqueous phase. The use of the ionic liquid [DecMIM]Br and the cyclodextrin derivative methyl-β-cyclodextrin provided especially fast and accurate phase separation.

Full text of DOI:10.1016/j.molcata.2015.04.006

Journal of Molecular Catalysis A: Chemical 404 (2015) 74–82
Contents lists available at ScienceDirect
Journal of Molecular Catalysis A: Chemical
journal homepage: www.elsevier.com/locate/molcata
Synthesis of primary amines from the renewable compound
citronellal via biphasic reductive amination
A. Behr∗, A. Wintzer, C. Lübke, M. Müller
Department of Biochemical and Chemical Engineering, Laboratory of Technical Chemistry – Chemical Process Development, Emil-Figgestr. 66, D-44227
Dortmund, Germany
a r t i c l e i n f o
a b s t r a c t
Article history:
The reductive amination of the natural product citronellal with ammonia is presented as a new and atom
economic way to its primary amine derivatives. The aqueous ammonia phase contains the homogeneous
catalytic system [Rh(cod)Cl]2/TPPTS in a biphasic solvent system, whereas the starting material and the
products remain in the apolar solvent phase. This concept supresses side reactions effectively, achieving
a high yield of primary amines of up to 87%. Systematic investigations demonstrate that the cleavage
of the secondary imine as an undesired by-product is necessary in achieving high selectivites, which
can be controlled by the reaction conditions. Surfactants, ionic liquids or native cyclodextrins and their
derivates prove to be useful phase transfer agents for optimising the interaction between the organic
and the aqueous phase. The use of the ionic liquid [DecMIM]Br and the cyclodextrin derivative methyl-
␤-cyclodextrin provided especially fast and accurate phase separation.
Received 12 January 2015
Received in revised form 10 April 2015
Accepted 13 April 2015
Available online 14 April 2015
Keywords:
Amination
Terpenes
Homogeneous catalysis
Renewables
Biphasic
© 2015 Elsevier B.V. All rights reserved.
1. Introduction
The reductive amination is a tandem reaction that consists of a
condensation and a hydrogenation step (Scheme 1). Reductive ami-
Citronellal (3,7-dimethyloct-6-en-1-al) is a monoterpenoid that
contains an aldehyde function. As a natural resource, it is well
known as a flavoring agent and insect repellent. Citronellal exists
in the form of two possible enantiomers: the (R)-isomer is typically
found in citronella [1] while the (S)-isomer appears mainly in the
essential oil of kaffir lime [2] (Fig. 1).
In addition to that, citronellal as a racemate is among many
essential oils that are derived from the fruits and leaves of citrus
plants. Essential oils can be extracted from these plants by means
of steam distillation. Apart from the natural way of producing cit-
ronellal, it can be synthesised from the monoterpene ␤-pinene.
␤-Pinene itself is in large part the turpentine oil from conifers. After
pyrolysis to ␤-myrcene, amination, isomerisation and hydrolysis,
citronellal is produced as a racemate (1.000 t a−1) [3a,b]. Alterna-
tive synthesis routes include the dehydrogenation of citronellol [4]
or the selective hydrogenation of citral [5a–c]. Citronellal is used
for the production of isopulegol and menthol, as well as vitamins A
and E at an industrial scale [6]. A review describes well-established
synthesis routes based on citronellal.[7]
nations are possible with ketones, aldehydes, and even alcohols as
starting materials. They react with an amine substrate in a conden-
sation reaction, which results in an aldimine, imine or enamine in
the initial step. In the second step, the intermediate is hydrogenated
catalytically under the same reaction conditions, which results in
primary, secondary or tertiary amines depending on which amine
is used.
Reductive aminations of aldehydes with ammonia were first
described in a general sense in relation to the Borch-reduction [8].
Bhattacharyya et al. investigated the reductive amination in two
steps including the combination of ammonia with titanium(IV) iso-
propoxide and a consecutive reduction with sodium borohydride
[9]. With benzaldehyde as substrate, a 77% yield of the primary
amine was achieved. Later, Timmer et al. developed an alterna-
tive process using sodium cyanohydridoborate in combination with
ammonium acetate and ammonia [10]. The best yield was reached
in the reduction of n-nonanal to its analogous primary amine at 98%.
Late transition metals have also been used for the reductive amina-
tion of aldehydes, mainly in combination with the model substrate
benzaldehyde [11–13]. Beller et al. transformed benzaldehyde to
benzylamine with a 96% yield by raising the reaction temperature
to 135 ◦C [12]. A catalytic system consisting of a rhodium precursor
and water-soluble phosphine ligand was used in a biphasic sol-
vent system. Liu et al. described an iridium-catalysed alternative
∗
Corresponding author. Tel.: +49 231 755 2310.
E-mail address: arno.behr@bci.tu-dortmund.de (A. Behr).
http://dx.doi.org/10.1016/j.molcata.2015.04.006
1381-1169/© 2015 Elsevier B.V. All rights reserved.

A. Behr et al. / Journal of Molecular Catalysis A: Chemical 404 (2015) 74–82
75
Fig. 1. The two natural isomers of citronellal.
Scheme 1. Reaction scheme of the reductive amination of an aldehyde with different amine substrates.
that required an additional reducing agent (triethylsilane) to pro-
duce a 40% yield of benzylamine [13]. Heterogeneous alternatives
have been described by Peters et al., [14] Bayer AG [15] and OXEA
[16]. Articles reviewing reductive amination in a general sense have
been written by Hartwig et al. [17] and Maschmeyer et al. [18].
Implementation on an industrial scale is known from the heteroge-
neously catalysed reductive amination of alcohols using ammonia.
It results in a product mixture of primary, secondary, and tertiary
amines. [19]
To the best of our knowledge, the functionalisation of citronel-
lal with ammonia leading directly to primary amines has not been
described until now. The main product, citronellyl amine, was syn-
thesised from the analogous oxime, [20] amide [21] and from
geranylnitrile [22]. Published examples of the transition metal
catalysed reductive amination of aldehydes with ammonia demon-
strated the need to suppress side reactions in order to achieve
high primary amine yield. The use of an efficient working biphasic
solvent system is an alternative to the subsequent cleavage of sec-
ondary amines [23]. Aqueous biphasic reaction systems like the one
used in this reaction often require phase transfer agents to reduce
transport limitations. Important examples include the use of sur-
factants, [24] cyclodextrins [25] or ionic liquids [26], predominantly
known for their use in hydroformylation and hydrogenation reac-
tions. We compared these potential means of reductive amination
of citronellal with ammonia in order to identify an optimal catalytic
system that provides ample opportunities for catalyst separation.
to cool to room temperature, the pressure was released, the two
layers were separated using a separating funnel, and the solution
was analysed by gas chromatography.
2.2. Characterisation
Citronellyl amine 4 (see Scheme 2) is known from the synthe-
sis of the analogous oxime. Analytical data is in accordance with
published data [22].
MS (70 eV, EI): m/z (%): 155 (M+, 1%), 138 (2), 123 (6), 112 (3),
109 (3), 95 (17), 81 (29), 70 (100), 55 (52).
For the characterisation of amine 5, a mixture of amines 4 and
5 was hydrogenated, which resulted in compound 5 exclusively.
1H NMR (500.13 MHz, CDCl3): ı = 0.86 ppm (d, 3J(H,H) = 6.6 Hz,
6H, 2 x -CH3), 0.89 (d, 3J(H,H) = 6.6 Hz, 3H, CH3), 1.15 (m, 6H, 3 x
-CH2), 1.37 (m, 2H, -CH2), 1.58 (m, 2H, 2 x -CH-) 2.80 (m, 2H, -CH2),
3.68 (m, 2H, NH2).
13C NMR (125.77 MHz, CDCl3): ı = 19.92, 22.87, 22.95, 25.00,
28.31, 29.85, 37.70, 39.58, 40.33, 41.58.
MS (70 eV, EI): m/z (%): 157 (M+, 6), 140 (11), 125 (4), 114 (43),
100 (26), 84 (21), 70 (80), 55 (100)
The observed side products are identified by comparing the
analytical data with the pure substance (citronellol 2) or by mass
spectrometry.
Citronellol 2
MS (70 eV, EI): m/z (%): 156 (M+, 1), 138 (2), 123 (9), 109 (7), 95
(31), 81 (50), 69 (100), 55 (92).
2. Experimental
3,7-Dimethyloct-6-en-1-imine 3
MS (70 eV, EI): m/z (%): 154 (M+, 1%), 139 (3), 121 (12), 111 (8),
95 (33), 93 (13), 91 (11), 84 (16), 79 (27), 77 (22), 69 (100), 67 (74),
65 (19), 63 (7), 55 (90), 83 (74), 51 (28).
2.1. Typical synthetic procedure
All reactants, catalysts and solvents were commercially avail-
able and were used without further purification. In a typical
experiment Chloro(1,5-cyclooctadiene) rhodium(I) (0.03 mmol
14.8 mg), triphenylphosphine trisulfonate (TPPTS, aq. saturated,
0.12 mmol, 68.2 mg) and hexadecyltrimethylammonium chloride
(CTAC, aq. 25%, 0.031 mol, 60.7 mg) were dissolved in an aqueous
ammonia solution (28%, 216.0 mmol, 13.14 g NH3). Nexy, toluene
(5.0 ml, 4.3 g) and citronellal (6.0 mmol, 925.5 mg) were added to
the solution, which was sonicated for 5 min. The solution was trans-
ferred to an evacuated stainless steel autoclave with a volume of
70 ml (Parr Instrument Company). The autoclave was pressurised
at 60 bar of H2, mechanically stirred with 800 rpm and heated up to
130 ◦C. At the end of the reaction time, the autoclave was allowed
(E)-N-(3,7-dimethyloct-6-en-1-ylidene)-3,7-dimethyloct-6-
en-1-amine 6
MS (70 eV, EI): m/z (%): 291 (M+, 1%), 276 (7), 248 (41), 234 (8),
222 (11), 208 (49), 194 (5), 180 (6), 166 (100), 152 (15), 138 (9), 124
(26), 110 (16), 98 (28), 84 (43), 69 (69), 55 (42).
Bis(3,7-dimethyloct-6-en-1-yl) amine 7
MS (70 eV, EI): m/z (%): 293 (M+, 2%), 278 (5), 250 (1), 236 (2),
224 (2), 208 (100), 168 (25), 152 (4), 138 (2), 126 (24), 112 (7), 98
(18), 93 (2), 81 (18), 69 (100), 65 (4), 55 (57), 51 (2).
(E)-N,N-Bis(3,7-dimethyloct-6-en-1-yl)-3,7-dimethylocta-1,6-
diene-1-amine 8
MS (70 eV, EI): 346 (2), 180 (2), 152 (2), 124 (2), 109 (4), 95 (6),
82 (12), 77 (2), 69 (100), 55 (57), 51 (1).

76
A. Behr et al. / Journal of Molecular Catalysis A: Chemical 404 (2015) 74–82
Fig. 2. Molecular structure of selected sulfonated phosphine ligands.
Tris(3,7-dimethyloct-6-en-1-yl) amine 9
MS (70 eV, EI): 346 (2), 304 (1), 236 (1), 180 (1), 152 (2), 138 (2),
Hydrogenation catalysts then convert the aldimine into another
primary amine molecule.
109 (4), 95 (8), 82 (11), 77 (3), 69 (100), 65 (2), 55 (53), 51 (1).
In the results, main products 4 and 5 as well as secondary (6
and 7) and tertiary (8 and 9) side products are combined. Unless
otherwise indicated, the conversion was quantitative.
Due to the complexity of the reaction network, a catalytic sys-
tem that was very active in imine hydrogenation and less active in
aldehyde hydrogenation was needed. Rhodium and iridium based
systems with water soluble phosphine ligands are described as
active for the synthesis of a primary amine via reductive amina-
tion [12,13] and were tested in combination with the ligand TPPTS
(Table 1).
The data showed that iridium complexes had a higher hydro-
genation activity, which resulted in the formation of more
citronellol 2. Especially with [Ir(cod)Cl]2 40% yield of 2 were
observed (entry 2), whereas the rhodium analogon [Rh(cod)Cl]2
leads to the formation of 11% 2 and 77% of the primary amines
4 and 5 (entry 1). Higher hydrogenation activity of rhodium pre-
cursors, for example of the Wilkinson catalyst, also resulted in the
formation of citronellol as well as the unwanted secondary amine
7 (entries 3-5).
3. Results and discussion
Apart from the condensation of starting material 1 with ammo-
nia, the hydrogenation to citronellol 2 occured as side reaction
(Scheme 2). Condensation between ammonia and citronellal 1 fol-
lowed by hydrogenation of the resulting aldimine 3 to citronellyl
amine 4 was identified as the other pathway. We could demon-
strate that over long reaction times, 4 was further hydrogenated to
form the other primary amine 5. Another undesired side reaction
was the condensation of 4 with the remaining citronellal 1 which
resulted in imine 6. This secondary product has two possibilities for
further reaction: One pathway is the hydrogenation to secondary
amine 7, which condenses further with the remaining citronellal
1 to form tertiary product 8 and 9 after subsequent hydrogena-
tion. Products 6–9 are declared as higher molecular side products.
The desired pathway is the most important step in this cascade:
imine 6 showed to be converted back to the primary products,
which will be discussed later in detail. This cleavage of secondary
imines has not been described in other research, though Beller et al.
have described a ruthenium system for the catalytic cleavage of
secondary amines.[23] After the dehydrogenation of the secondary
amine, the imine is cleaved to one molecule of primary amine and
one molecule of aldimine via a nucleophilic attack of ammonia.
Bidentate
TPPTS,
mono-dentate ligand. BINAS (sulfonated 2,2ꢀ-
bis(diphenylphosphino-methyl)-1,1ꢀ-binaphthylene) and BISBIS
(sulfonated
2,2ꢀ-bis(diphenylphosphinomethyl)-1,1ꢀ-biphenyl),
ligands
were
tested
and
compared
to
a
both chelating phosphine ligands (Fig. 2), produced desirable
results in the biphasic hydroformylation of higher alkenes [27].
Scheme 2. Complete reaction scheme from the reductive amination of citronellal with ammonia.

A. Behr et al. / Journal of Molecular Catalysis A: Chemical 404 (2015) 74–82
77
Table 1
Influence of the precursor on the product distribution in the reductive amination of citronellal.
Nr.
Precursor
Y (main products) [%]
Y (side products) [%]
2
4
5
4 + 5
6 + 7
8 + 9
1
2
3
4
5
6
[Rh(cod)Cl]2
[Ir(cod)Cl]2
[Rh(PPh3)3Cl]
[Rh(cod)2]BF4
[Rh(CO)H(PPh3)3]
[Ir(CO)H(PPh3)3]
51
36
33
29
29
25
26
3
0
1
0
77
39
33
30
29
25
11
40
21
17
12
19
9
13
42
46
45
49
3
6
0
5
11
4
0
Reaction conditions: 6 mmol citronellal, 144 mmol NH3, 0.5 mol% precursor, 4.0 mol% TPPTS, 0.5 mol% CTAB (Cetyltrimethylammonium bromide), 5 ml toluene, 60 bar H2,
130 ◦C, 800 rpm, 6 h.
Table 2
Various precursor/phosphine ratios in comparison for two different ligands.
Nr.
Y (main products) [%]
Y (side products) [%]
2
Ligand /
precursor/ligand ratio
4
5
4 + 5
6 + 7
8 + 9
1
2
3
4
TPPTS / 1:4
TPPTS / 1:8
BINAS / 1:4
BINAS / 1:8
35
51
35
48
43
26
25
15
78
77
60
63
9
11
10
10
12
9
24
23
1
3
6
5
Reaction conditions: 6 mmol citronellal, 144 mmol NH3, 0.5 mol% [Rh(cod)Cl]2, 0.5 mol% CTAB, 5 ml toluene, 60 bar H2, 130 ◦C, 800 rpm, 6 h.
Especially in terms of the n/iso ratio, the chelating ligands showed
improved yield compared to TPPTS. There was no need to improve
the n/iso ratio for the reductive amination of citronellal.
Therefore, the bidentate ligands were unable to increase the
yield of the primary amines. Apart from that, it can be said that they
provided faster phase separation. Due to its simple structure and
therefore lower cost, the TPPTS ligand was selected as the more eco-
nomic choice. Experimental investigations showed that the ratio
of the dimeric precursor and the monodentate sulfonated triph-
enylphosphine can be reduced to a molar ratio of 1:4 for a precursor
concentration of 0.5 mol% based on the starting material (Table 2).
In this way, costs can be reduced additionally whereas the ligand
excess is high enough to provide sufficient catalyst stabilisation and
activity.
The principle of micellar catalysis uses the formation of spheric
aggregates which act as amphoteric compounds for the interac-
tion between two unsoluble phases. Changing the surfactant or
using a derivative can lead to another form of micelle forma-
tion. Using CTAC (Cetyltrimethylammonium chloride) instead of
CTAB (Cetyltrimethylammonium bromide) in this investigation
improved the yield of this reductive amination slightly to 81%
(Table 3).
Fig. 3. Reaction conditions: 6 mmol citronellal, 144 mmol NH3, 0.5 mol%
[Rh(cod)Cl]2, 2.0 mol% TPPTS, 0.5 mol% CTAC, 5 ml solvent, 60 bar H2, 130 ◦C,
800 rpm, 6 h.
In addition to toluene, the organic solvents cyclohexane and 1-
decane were also suitable for this reaction. The yield dropped to 75%
using cyclohexane and to 67% using decane. When the structural
properties of the starting material citronellal were compared to the
organic solvents, it became clear that citronellal itself presented an
adequate solvent as an organic phase. In fact, using a large excess of
citronellal rather than an additional solvent led to a lower primary
amine yield of 55% at a 97% conversion rate (Fig. 3). With respect
to the rules of “green” chemistry it is desirable to minimise the
use of solvents in chemical syntheses [28]. This environmentally-
friendly method resulted in lower primary amine yield, which could
pose a suitable compromise for an upscaled process. In a continuous
process this would imply recycling the excess citronellal after the
separation step.
Further on, different reaction parameters were varied within the
established catalytic system to investigate their effects on product
distribution. In an initial step, the substrate ratio of citronellal to
ammonia proved ideal when a large excess of ammonia was used
(Table 4). Using a ratio of 1:12, the yield dropped to 68% compared
to a ratio of 1:24 (81%). When the excess of ammonia was increased
to a ratio of 1:36 the yield was improved to 87%. Increasing the ratio
even more to 1:48 showed with 82% no further improvement in
yield. Comparing the yield of the secondary products (15% at 1:12
and 8% at 1:36) indicated that a larger aqueous ammonia phase
Shortening the apolar chain of the surfactant using TTAB
(Tetradecyltrimethylammonium bromide) did not lead to changes
in the product yield. A useful way to avoid the use of sulfonated
ligands was demonstrated by using the anionic surfactant SDS
(sodium dodecyl sulfate). The coordination of the sulfate function-
ality to the active metal centre generated an amphoteric catalyst
system, which interacted in the apolar phase with the starting
material. The compound is also able to stabilise the catalytic system.
Although the yield decreased to 58%, it proved to be a highly effec-
tive method of combining the amphoteric properties of a surfactant
with the metal-coordinating functionality of a ligand. In addition,
tests were performed to determine whether the reaction could take
place without any surfactant. Still, a moderate yield of 45% of the
primary amines was achieved. The amount of imine 6 as part of the
secondary products increased to 33% (entry 5), indicating a lower
hydrogenation activity as a result of decreased phase interaction.
This showed that transport limitations were low enough to provide
proper phase interaction without the use of a phase transfer agent
in the biphasic citronellal + toluene and water + ammonia + catalyst
system.

78
A. Behr et al. / Journal of Molecular Catalysis A: Chemical 404 (2015) 74–82
Table 3
Influence of the specific surfactant on the product distribution in the reductive amination of citronellal.
Nr.
Y (main products) [%]
Y (side products) [%]
Phase transfer agent
4
5
4 + 5
2
6 + 7
8 + 9
1
2
3
4
5
CTAC
CTAB
TTAB
SDS[a]
–
36
35
45
39
40
45
43
34
19
5
81
78
79
58
45
7
9
9
8
9
11
12
13
22
38
0
1
0
12
9
Reaction conditions: 6 mmol citronellal, 144 mmol NH3, 0.5 mol% [Rh(cod)Cl]2, 2.0 mol% TPPTS, 0.5 mol% surfactant, 5 ml toluene, 60 bar H2, 130 ◦C, 800 rpm, 6 h. [a] without
the use of TPPTS.
Table 4
Comparison of different citronellal/ammonia ratios.
Nr.
Y (main products) [%]
Y (side products) [%]
citronellal/NH3 ratio
[mol/mol]
4
5
4 + 5
2
6 + 7
8 + 9
1
2
3
4
1:12
1:24
1:36
1:48
56
36
62
49
12
45
25
33
68
81
87
82
10
7
5
15
11
8
0
0
0
0
9
9
Reaction conditions: 6 mmol citronellal, 0.5 mol% [Rh(cod)Cl]2, 2.0 mol% TPPTS, 0.5 mol% CTAC, 5 ml toluene, 60 bar H2, 130 ◦C, 800 rpm, 6 h.
had a controlling effect, which was expected to work as cleavage
instrument for imine 6.
of a secondary imine to the primary compounds is more efficient
at high reaction temperatures.
The reaction temperature and hydrogen pressure were other
important process parameters in terms of the imine cleavage. The
primary amine yield was controlled with ease by varying these
factors. A maximum yield of 87% (Fig. 4), was achieved within a
temperature range of 130–150 ◦C. There were only traces of imine
6 at a reaction temperature of 130 ◦C, while a reaction temperature
of 150 ◦C provided no improvement.
Compared to the reaction temperature, the effects of varying the
hydrogen pressure showed a similar behaviour: a pressure of 60 bar
afforded a yield of 87% (Fig. 5). In general, a slight increase of pri-
mary amine yield from 80% to 87% was observed within a range
between 20 bar and 60 bar. In terms of atom economy, keeping
the hydrogen pressure at 20 bar proved to be the most reasonable
value. Further pressure drops to 10 and 2 bar resulted in higher
secondary product yields of 19% and 52%, respectively. In the lat-
ter case, the secondary products consisted of imine 6 only. High
hydrogen pressure provides the best conditions for cleavage of the
undesired secondary imine.
These experimental results demonstrated the possible ways
of reaction control quite simply, but to investigate especially the
transformation of imine 6 to the primary products in more detail,
a conversion-time plot was determined (Fig. 6). The reaction was
very fast; after one hour the conversion was quantitative. Surpris-
ingly, imine 6 was formed analogously to the conversion curve. This
indicated that the condensation of citronellal 1 with ammonia and
the following hydrogenation of aldimine 3 to amine 4, as well as
the condensation to imine 6 took place very quickly. It should be
noted that the autoclave was heating up during this period (reach-
ing 130 ◦C after 45 min), which indicated that these reactions were
carried out at even lower temperatures than 130 ◦C. After one hour,
Decreasing the temperature to 110 ◦C resulted in a higher yield
of secondary imine 6 at 20%, indicating insufficient imine cleavage.
This trend became even more clear when the reaction was per-
formed at 90 ◦C. 47% of imine 6 was produced, whereas the primary
amine yield decreased to 30%. These results clearly demonstrated
that temperatures of at least 130 ◦C are necessary for the cleavage
of the imine to the primary products. The cleavage of secondary
amines via the imine as an intermediate as described previously
was performed under similar reaction conditions.[23] For exam-
ple, dicyclohexylamine was cleaved more effectively at a reaction
temperature of 150 ◦C (87% yield) than at a temperature of 130 ◦C
(49% yield). It should be noted that the ammonia catalysed cleavage
Fig. 4. Impact of the reaction temperature on the imine cleavage; data points are
connected to result in a better comparison. Reaction conditions: 6 mmol citronellal,
216 mmol NH3, 0.5 mol% [Rh(cod)Cl]2, 2.0 mol% TPPTS, 0.5 mol% CTAC, 5 ml toluene,
60 bar H2, 800 rpm, 6 h.
Fig. 5. Impact of the hydrogen pressure on the product distribution in the reductive
amination of citronellal; data points are connected to result in a better compari-
son. Reaction conditions: 6 mmol citronellal, 216 mmol NH3, 0.5 mol% [Rh(cod)Cl]2,
2.0 mol% TPPTS, 0.5 mol% CTAC, 5 ml toluene, 130 ◦C, 800 rpm, 6 h.

A. Behr et al. / Journal of Molecular Catalysis A: Chemical 404 (2015) 74–82
79
Fig. 6. Conversion-time plot of the reductive amination of citronellal with ammonia. Reaction conditions: 6 mmol citronellal, 216 mmol NH3, 0.5 mol% [Rh(cod)Cl]2, 2.0 mol%
TPPTS, 0.5 mol% CTAC, 5 ml toluene, 60 bar H2, 130 ◦C, 800 rpm.
the yield of imine 6 reached 88%. As time passed, the imine yield
decreased while amine 4 was formed in increasing amounts. After
5 h, a yield of 58% for amine 4 and 19% for imine 6 was observed.
The highest yield of primary amine 4 was achieved after 8 h at 66%.
The internal double bound of 4 was hydrogenated as time passed,
leading to a yield of 80% of amine 5 and 3% of amine 4 after 24 h. It
has been described before that specific parameters are crucial for a
low amount of higher molecular by-products such as imine 6. While
6 results from a condensation of 1 with 4, the depletion over the
reaction time is an interesting observation. In general, imines like 6
can be hydrolysed to give an amine and an aldehyde, although this
is more likely for acidic systems.
Another possible mechanism is the cleavage by ammonia, which
would result in two amine products and does not require acidic
reaction conditions (Scheme 3). This way of reaction has been also
described by Beller et al. with and without the use of water [23].
As we cannot exclude that hydrolysis is involved in the reaction
mechanism, it is likely that ammonia is necessary to shift the equi-
librium which results from reversible hydrolysis. An exceptional
hydrolysis would not result in a decreasing, rather than in a sta-
bilising imine yield. Therefore we believe that ammonia acts as a
cleavage instrument which is responsible for a high primary amine
yield.
The use of phase transfer agents in biphasic aqueous solvent sys-
tems can enhance the interaction between the substrates and the
catalyst. Their use can also negatively impact the phase separation
of the reaction system which would lead to a lower space-time yield
and therefore to a less economical process. It is well-established
that typical surfactants like CTAB offer suitable properties in terms
of reaction rate. In this point of view, the critical micelle concentra-
tion (CMC) is responsible for the occurring micellar catalysis which
is described for several reactions by Oehme et al [29]. But the use of
surfactants can also lead to the formation of emulsions in biphasic
aqueous solvent systems [26a]. In fact, we observed the forma-
tion of an emulsion phase between the aqueous and toluenic layer
in the tested reaction system. This led to a very slow separation
process that occurred overnight. In certain cases, it even com-
plicated or interfered with the separation of the product phase
from the catalyst phase. In this case, the capacity of a produc-
tion plant would decrease dramatically. Therefore, we searched for
alternative phase transfer agents. Suitable examples from biphasic
aqueous hydroformylation reactions of alkenes are well estab-
lished. Ionic liquids, especially imidazolium derivatives, were used
by Hamilton et al. for the hydroformylation of 1-hexene, 1-octene
and 1-decene. Phase separation occurred with the matching agent
within a few minutes and resulted in a very accurate interface
[26a–b].
These imidazolium based ionic liquids can be varied in form
of the chain length and in the type of the anion. As the results
achieved by Hamilton et al. indicated a correlation between the
chain length of the ionic liquid and the starting material in order
to secure a suitable interaction between the phases,26b imida-
zolium derivatives with a side chain with a carbon number of 8,
10 and 12 were chosen for this experiment (Fig. 7). Aside from the
ionic liquids 1-methyl-3-octylimidazolium bromide ([OctMIM]Br),
1-decyl-3-methylimidazolium bromide ([DecMIM]Br) and 1-
dodecyl-3-methylimidazolium bromide ([DodMIM]Br), another
class of ionic liquids was selected to test its properties as a phase
transfer agent. These so-called Ammoeng molecules are quar-
ternary ammonium compounds with chains from fatty acids and
polyethylene glycol of different chain lengths. Their potential has
been shown in the production of biodiesel [30] and the exploitation
of crude oil [31].
Compared to the use of surfactants, all of the tested ionic liquids
provided fast and accurate phase separation. A clear phase sepa-
ration was achieved in a matter of minutes, which demonstrated
the potential of ionic liquids as efficient phase transfer agents.
Regarding the yield of the primary amines 4 and 5, the ionic liquids
generated different results (Table 5).
A maximum yield of 82% was achieved using [DecMIM]Br. Using
imidazolium bromides with the octyl and dodecyl chain, higher
amounts of citronellol 2 (20%, entry 1) and secondary products 6
Scheme 3. Potential cleavage mechanisms of imine 6 to produce primary products.

80
A. Behr et al. / Journal of Molecular Catalysis A: Chemical 404 (2015) 74–82
Fig. 7. Structures of the used ionic liquids as phase transfer agents for the reductive amination of citronellal.
Table 5
Comparison of ionic liquids as phase transfer agents in the reductive amination of citronellal.
Nr.
Y (main products) [%]
Y (side products) [%]
Phase transfer agent
4
5
4 + 5
2
3
6 + 7
8 + 9
1
2
3
4
5
6
[OctMim]Br
41
81
27
73
40
0
2
1
2
6
4
43
82
29
79
44
20
20
8
0
8
14
0
21
8
0
8
14
0
15
1
71
4
27
80
0
0
0
0
0
0
[DecMIM]Br
[DodMim]Br
Ammoeng 100
Ammoeng 101
Ammoeng 102
20
Reaction conditions: 6 mmol citronellal, 216 mmol NH3, 0.5 mol% [Rh(cod)Cl]2, 2.0 mol% TPPTS, 5 mmol ionic liquid, 5 ml toluene, 60 bar H2, 800 rpm, 130 ◦C, 6 h.
and 7 (71%, entry 3) were formed, which resulted in a lower primary
amine yield. Using the Ammoeng compounds, comparable results
to the imidazolium species were generated (entries 4–6). Ammo-
eng 100 in particular, which contains fatty acid chains from coconut
oil, led to high amounts of the primary amines 4 and 5 at 79%. For
the most part, unsaturated amine 4 was observed, with the excep-
tion of Ammoeng 102 (entry 6). Using this species, which has fatty
acid chains from tall oil, led to the selective formation of saturated
primary amine 5 and imine 6 without observing secondary amine
7. Similar results are obtained for [DodMim]Br with 70% of imine
6 (entry 3). Both species represent longer chained examples for
phase transfer agents which do not seem suitable for this biphasic
reaction system.
methylated ␤-cyclodextrin (RAME-␤-CD) in particular proved its
ability to perform homogeneously catalysed reactions in bipha-
sic solvent systems. It was tested and compared with the native
cyclodextrins ␣-cyclodextrin (␣-CD), ␤-cyclodextrin (␤-CD) and
␥-cyclodextrin (␥-CD) as well as hydroxypropyl-␤-cyclodextrin
(HP-␤-CD), another derivative of ␤-CD for the reductive amination
of citronellal with ammonia (Table 6).
The influence of the ring size of these molecules showed only
slight increases of primary amines using ␣-CD. Using the larger ␥-
CD, the primary amine yield dropped from 61% to 52%. Secondary
products were observed within a range between 39% and 48% in all
cases. The yield of the side products was supressed using RAME-␤-
CD (entry 4). Secondary product yield decreased to 4% whereas a
high yield of the primary amines was observed at 79%. In contrast
to the native cyclodextrins where no citronellol 2 was detected,
alcohol was observed at a low yield of 8%. This indicated adequate
The use of native cyclodextrines and their derivatives as
phase transfer agents in aqueous biphasic reaction systems were
described in detail by Monflier et al. and reviewed [25g]. Randomly
16-2-16
2 Br
cmc = 0.021 mmol l-1
N
N
CTAB
CTAC
cmc = 0.97 mmol l-1
cmc = 1.1 mmol l-1
Br
Cl
N
N
Fig. 8. Structural comparison of typical surfactants with the gemini surfactant 16-2-16.

A. Behr et al. / Journal of Molecular Catalysis A: Chemical 404 (2015) 74–82
81
Table 6
Comparison of cyclodextrins as phase transfer agents in the reductive amination of citronellal.
Nr.
Y (main products) [%]
Y (side products) [%]
Phase transfer agent
4
5
4 + 5
2
6 + 7
8 + 9
1
2
3
4
5
␣-CD
␤-CD
␥-CD
RAME-␤-CD
HP-␤-CD
54
52
48
73
59
7
7
4
6
5
61
58
52
79
64
0
0
0
8
0
39
42
48
4
0
0
0
0
0
36
Reaction conditions: 6 mmol citronellal, 216 mmol NH3, 0.5 mol% [Rh(cod)Cl]2, 2.0 mol% TPPTS, 0.54 mmol cyclodextrin, 5 ml Toluol, 60 bar H2, 800 rpm, 130 ◦C, 6 h.
interaction between the phases, as citronellal came into contact
with the catalyst and hydrogenated to citronellol 2. A lower pri-
mary amine yield was observed for HP-␤-CD (entry 5, 64%), which
indicated that the polarity of the molecules is the most important
factor in improving phase interaction.
References
[1] K.-G. Fahlbusch, F.-J. Hammerschmidt, J. Panten, W. Pickenhagen, D.
Schatkowski, K. Bauer, D. Garbe, H. Surburg, Flavors and Fragrances in
Ullmann’s Encyclopedia of Industrial Chemistry, Wiley-VCH, Weinheim, 2012.
[2] J.D. Connolly, R.A. Hill, Dictionary of Terpenoids, Chapman & Hall, London,
1991.
Gemini surfactants are a special type of surfactants. They dif-
fer from typical surfactants due to their specific structure: two
polar headgroups, each connected to an apolar chain, are linked
by one spacer. This can lead to much lower critical micell con-
centrations (CMC) compared to surfactants such as CTAC. [32]
Using a gemini surfactant with a low CMC in the biphasic sol-
vent system was expected to lower the surfactant concentration in
comparison to CTAC. Therefore, phase separation was expected to
improve due to decreased emulsion formation. We decided to use
the gemini surfactant hexadeca-diylbis(dimethylethylammonium)
bromide (16-2-16). With a CMC of 0.021 mmol l−1, [33] it provides
a much smaller CMC than the surfactant CTAC (1.10 mmol l−1) [34]
(Fig. 8).
[3] (a) V.N. Kraseva, A.A. Bag, L.L. Malkina, O.M. Kohl’mer, Chem. Abstr. 53 (1959)
22067h.
[4] R.L. Webb (Glidden, USA), Process for preparing organoleptic materials,
US3028431 A, 1962.
[5] P.S. Gradeff (Rhodia, USA), Process for the semihydrogenation of citral to
citronellal, US3971830 A, 1976.
[6] (b) A. Behr, L. Johnen, ChemSusChem 2 (2009) 1072–1095;
M. Friedrich, K. Ebel, N. (BASF AG), Verfahren zur Herstellung von Isopulegol,
DE102004063003 A1, 2006.
[7] E.J. Lenardao, G.V. Botteselle, F. de Azambuja, G. Perin, R.G. Jacob, Tetrahedron
63 (2007) 6671–6712.
[8] R.F. Borch, M.D. Bernstein, H.D. Durst, J. Am. Chem. Soc. 93 (1971) 2897–2904.
[9] B. Miriyala, S. Bhattacharyya, J.S. Williamson, Tetrahedron 60 (2004)
1463–1471.
[10] E.M. Dangerfield, C.H. Plunkett, A.L. Win-Mason, B.L. Stocker, M.S.M. Timmer,
J. Org. Chem. 75 (2010) 5470–5477.
[11] T. Riermeier, K.-J. Haack, U. Dingerdissen, A. Börner, V. Tararov, R. Kadyrov
(Aventis Research & Technologies GmbH & Co. KG, Germany), Method for
producing amines by homogeneously catalyzed reductive amination of
carbonyl compounds, WO 01/05741 A1, 2001.
Performing the reaction with the gemini surfactant reduced the
formation of emulsions, thus offering improved phase separation
compared to CTAC. At the same time, high primary amine yield at a
rate of 79% was achieved, indicating micellar catalysis was carried
out successfully in this catalytic system as well.
[12] T. Gross, A.M. Seayad, M. Ahmad, M. Beller, Org. Lett. 4 (2002) 2055–2058.
[13] R.Y. Lai, C.I. Lee, S.T. Liu, Tetrahedron 64 (2008) 1213–1217.
[14] A.W. Heinen, J.A. Peters, H. van Bekkum, Eur. J. Org. Chem. 2000 (2000)
2501–2506.
[15] W. Kiel (Bayer AG), Reductive amination of aldehydes, US6429335 B1, 2002.
[16] K. Schalapski, N. Nowotny, M. Eisenacher, D. Bermann, T. Kreickmann, P.
Heymanns, H. Strutz (Oxea GmbH, Germany), Method for Producing Primary
Aliphatic Amines from Aldehydes, WO2012/031672, 2012.
[17] J.L. Klinkenberg, J.F. Hartwig, Angew. Chem. Int. Ed. 50 (2011) 86–95.
[18] S. Gomez, J.A. Peters, T. Maschmeyer, Adv. Synth. Catal. 344 (2002)
1037–1057.
4. Conclusions
Reductive amination of citronellal with ammonia was inves-
tigated as a selective one-step synthesis of primary amines from
a renewable material. Using aqueous ammonia as substrate, a
two phase solvent system was created to prevent side reactions.
In addition, a conversion-time plot showed that the cleavage of
the undesired secondary imine is the most important step in
this tandem reaction, which was also indicated by temperature
and pressure variations. We compared a variety of phase transfer
agents, combining surfactants, ionic liquids and cyclodextrins with
a water soluble homogeneous rhodium catalyst. Ionic liquids and
cyclodextrins provided particularly fast and accurate phase sep-
aration, yielding the primary amines at a maximum yield of 82%
at full conversion. Using gemini surfactants as an alternative was
also investigated, which provided extremely low CMC. Therefore,
no formation of emulsion was observed. In addition, the products
were isolated from the catalyst with ease by separating the aqueous
from the organic layer.
[19] S.A. Lawrence, Amines: Synthesis, in: Properties and Applications, Cambridge
University Press, Cambridge, 2004.
[20] (a) J. Ipaktschi, Chemische Berichte 117 (1984) 856–858;
(b) D. Arigoni, O. Jeger, Helvetica Chimica Acta 37 (1954) 881–883.
[21] C.G. Overberger, T. Takekoshi, Macromolecules 1 (1968) 1–6.
[22] P. Kukula, K. Koprivova, J. Catal. 234 (2005) 161–171.
[23] S. Bahn, S. Imm, L. Neubert, M. Zhang, H. Neumann, M. Beller, Chem. Eur. J. 17
(2011) 4705–4708.
[24] T. Dwars, E. Paetzold, G. Oehme, Angew. Chem. Int. Ed. 44 (2005) 7174–7199.
[25] (a) E. Monflier, S. Tilloy, G. Fremy, Y. Castanet, A. Mortreux, Tetrahedron Lett.
36 (1995) 9481–9484;
(b) E. Monflier, S. Tilloy, Y. Castanet, A. Mortreux, Tetrahedron Lett. 39 (1998)
2959–2960;
(c) T. Mathivet, C. Meliet, Y. Castanet, A. Mortreux, L. Caron, S. Tilloy, E.
Monflier, J. Mol. Catal. A: Chem. 176 (2001) 105–116;
(d) S. Tilloy, H. Bricout, E. Monflier, Green Chem. 4 (2002) 188–193;
(e) B. Sueur, L. Leclercq, M. Sauthier, Y. Castanet, A. Mortreux, H. Bricout, S.
Tilloy, E. Monflier, Eur. J. Chem. 11 (2005) 6228–6236;
(f) F. Hapiot, L. Leclercq, N. Azaroual, S. Fourmentin, S. Tilloy, E. Monflier, Curr.
Org. Synth. 5 (2008) 162–172;
(g) F. Hapiot, A. Ponchel, S. Tilloy, E. Monflier, Cr. Chim. 14 (2011) 149–166;
(h) C. Blaszkiewicz, H. Bricout, E. Leonard, C. Len, D. Landy, C. Cezard, F.
Djedaini-Pilard, E. Monflier, S. Tilloy, Chem. Commun. 49 (2013) 6989–6991.
[26] (a) S.L. Desset, D.J. Cole-Hamilton, D.F. Foster, Chem. Commun. (2007)
1933–1935;
Acknowledgements
(b) S.L. Desset, S.W. Reader, D.J. Cole-Hamilton, Green Chem. 11 (2009)
630–637;
(c) S.G. Lee, Chem. Commun. (2006) 1049–1063.
This work was financially supported by the DFG (Deutsche
Forschungsgemeinschaft). The authors thank Umicore AG & Co. KG
for the donation of precursors.

82
A. Behr et al. / Journal of Molecular Catalysis A: Chemical 404 (2015) 74–82
[27] B. Cornils, E.G. Kuntz, Aqueous Phase Organometallic Catalysis: Concepts and
Applications, in: B. Cornils, W.A. Herrmann (Eds.), 2nd ed., Wiley-VCH,
Weinheim, 2005.
[28] A. Behr, A. Wintzer, Chem. Eng. Tech. 83 (2011) 1356–1370.
[29] T. Dwars, E. Paetzold, G. Oehme, Ang. Chem. Int. Ed. 44 (2005) 7174–7199.
[30] B.L.A.P. Devi, Z. Guo, X.B. Xu, Aiche J. 57 (2011) 1628–1637.
[31] M.S. Benzagouta, I.M. AlNashef, W. Karnanda, K. Al-Khidir, Kor. J. Chem. Eng.
30 (2013) 2108–2117.
[32] F.M. Menger, J.S. Keiper, Angew. Chem. Int. Ed. 39 (2000) 1907–1920.
[33] R. Zana, M. Benrraou, R. Rueff, Langmuir 7 (1991) 1072–1075.
[34] L. Tofani, A. Feis, R.E. Snoke, D. Berti, P. Baglioni, G. Smulevich, Biophys. J. 87
(2004) 1186–1195.