A New Heterogeneously Catalytic Pathway for Isomerization of Linoleic Acid over Ru/C and Ni/H-MCM-41 Catalysts

Article abstract of DOI:10.1006/jcat.2002.3690

The isomerization reaction of linoleic acid (cis-9,cis-12-octadecadienoic acid) to conjugated linoleic acids (CLA) was studied. A new heterogeneously catalytic pathway was developed for synthesis of CLA at mild reaction conditions (80-120 deg C) over Ru/C and Ni/H-MCM-41 catalysts in a diluted system. The isomerization reaction was tested in 1-octanol and n-decane. Solvents with low polarity were selected to afford good reactant adsorption on the catalyst surface and a high solvent-to-reactant ratio was used to prevent potential side reactions. Catalyst characterization was done by X-ray powder diffraction, X-ray fluorescence, X-ray photoelectron spectroscopy, hydrogen temperature-programmed desorption, direct current plasma atomic emission spectrometry, and BET techniques. The reactions taking place were isomerization of linoleic acid C18:2, hydrogenation of C18:2 acids to C18:1 monounsaturated octadecenoic acids (oleic acid, elaidic acid, cis-vaccenic acid and trans-vaccenic acid), and further hydrogenation of C18:1 acids to C18:0 stearic acid (n-octadecanoic acid) where isomerization and hydrogenation were two competing parallel reactions. The isomerization reaction was enhanced by preactivation of the catalysts under hydrogen, but such a treatment also increased the side reaction double-bond hydrogenation.

Full text of DOI:10.1006/jcat.2002.3690

Journal of Catalysis 210, 354–366 (2002)
doi:10.1006/jcat.2002.3690
A New Heterogeneously Catalytic Pathway for Isomerization of Linoleic
Acid over Ru/C and Ni/H–MCM-41 Catalysts
∗
∗
∗
Andreas Bernas, Pekka Laukkanen,† Narendra Kumar, P a¨ ivi M a¨ ki-Arvela, Juhani V a¨ yrynen,†
∗
∗,1
Ensio Laine,‡ Bjarne Holmbom,§ Tapio Salmi, and Dmitry Yu. Murzin
∗
˚
˚
Laboratory of Industrial Chemistry, Process Chemistry Group, Abo Akademi University, FIN-20500 Abo/Turku, Finland; †Laboratory
˚
of Surface Physics, Department of Physics, University of Turku, FIN-20014 Abo/Turku, Finland; ‡Laboratory of Industrial Physics,
˚
Department of Physics, University of Turku, FIN-20014 Abo/Turku, Finland; and §Laboratory of Wood Chemistry,
˚
˚
Process Chemistry Group, Abo Akademi University, FIN-20500 Abo/Turku, Finland
Received February 5, 2002; revised April 16, 2002; accepted May 24, 2002
or both located along the carbon chain (1). Useful analytical
1
3
The isomerization reaction of linoleic acid (cis-9,cis-12- data can be obtained through C NMR of the olefinic car-
octadecadienoic acid) to conjugated linoleic acids (CLA) was stud- bon atoms. Each CLA isomer has signals for its four olefinic
ied. A new heterogeneously catalytic pathway was developed for
synthesis of CLA at mild reaction conditions (80–120 C) over Ru/C
carbon atoms and chemical shifts have been identified for
◦
20 different CLA isomers. These are cis,cis; trans,trans;
and Ni/H–MCM-41 catalysts in a diluted system. The isomeriza-
tion reaction was tested in 1-octanol and n-decane. Solvents with
low polarity were selected to afford good reactant adsorption on
the catalyst surface and a high solvent-to-reactant ratio was used
to prevent potential side reactions. Catalyst characterization was
done by X-ray powder diffraction, X-ray fluorescence, X-ray pho-
toelectron spectroscopy, hydrogen temperature-programmed des-
orption, direct current plasma atomic emission spectrometry, and
cis,trans; and trans,cis isomers of the 7,9; 8,10; 9,11; 10,12;
and 11,13 C18 diene acids. Structures of linoleic acid, mo-
nounsaturated octadecenoic acids (oleic acid, elaidic acid,
cis-vaccenic acid, and trans-vaccenic acid), saturated stearic
acid (n-octadecanoic acid), and CLA isomers are given in
Fig. 1.
CLA is naturally present in meat and dairy products. It
BET techniques. The reactions taking place were isomerization of is found in many animal products, especially those from ru-
linoleic acid C18:2, hydrogenation of C18:2 acids to C18:1 monoun- minant sources where it is synthesized from linoleic acid by
saturated octadecenoic acids (oleic acid, elaidic acid, cis-vaccenic
acid and trans-vaccenic acid), and further hydrogenation of C18:1
acids to C18:0 stearic acid (n-octadecanoic acid) where isomeriza-
tion and hydrogenation were two competing parallel reactions. The
isomerization reaction was enhanced by preactivation of the cata-
lysts under hydrogen, but such a treatment also increased the side
rumen bacteria (2). The total content of CLA in food varies
widely, but the cis-9,trans-11 isomer has been found to be
the predominant form (as much as 80 to 90% of total CLA)
(
3). Increased dairy fat consumption has been shown to
be associated with increased CLA levels in human adipose
tissue and human milk (2).
reaction double-bond hydrogenation.
ꢀc 2002 Elsevier Science (USA)
CLA has recently been found to be both anticarcinogenic
and antioxidative, hence these fatty acids are of great inter-
est in food and health research. They have been found to af-
fect immunomodulation and body composition alteration,
and they can prevent or cure cancer and atherosclerosis (2).
Several studies have demonstrated that the cis-9,trans-11
isomer is responsible for the anticarcinogenic effects of
CLA (1, 4). It has also been concluded that the trans-10,
cis-12 isomer produces the changes in body composition
that have been observed routinely in mice, rats, and pigs.
CLAs have also been linked directly to increased insulin
sensitivity, normalized glucose tolerance, improved hyper-
insulinemia, and lowered levels of circulating free fatty
acids (1).
Key Words: functional food; conjugated linoleic acids; Ni/H–
MCM-41; double-bond migration; isomerization; hydrogenation.
1
. INTRODUCTION
In this paper, a new heterogeneously catalytic path-
way is described for the synthesis of conjugated linoleic
acids over supported metal catalysts. Conjugated linoleic
acids (CLAs), first positively identified in 1987, is a
collective term describing the positional and geometric
conjugated dienoic isomers of linoleic acid (cis-9,cis-12-
octadecadienoic acid). Linoleic acid (C18:2) has double
bonds located on carbons 9 and 12, both in the cis configura-
tion, whereas CLA has either the cis or trans configuration
CLA can be prepared by isomerization of linoleic acid
under alkaline conditions. It has been reported that when
CLAwassynthesizedfromlinoleicacidinahomogeneously
catalyzed isomerization reaction, 90% of the product
1
To whom correspondence should be addressed. Fax: 358 2 215 4479.
E-mail: dmurzin@abo.fi.
354
0
ꢀ
021-9517/02 $35.00
c 2002 Elsevier Science (USA)
All rights reserved.

ISOMERIZATION OF LINOLEIC ACID
355
COOH
COOH
COOH
II cis-9-octadecenoic acid
(oleic acid)
COOH
III trans-9-octadecenoic acid
I cis-9,cis-12-octadecadienoic acid
(linoleic acid)
(elaidic acid)
COOH
V trans-11-octadecenoic acid
(trans-vaccenic acid)
COOH
IV cis-11-octadecenoic acid
cis-vaccenic acid)
VI n-octadecanoic acid
(stearic acid)
(
COOH
COOH
VII cis-7,cis-9-octadecadienoic acid
COOH
VIII trans-7,trans-9-octadecadienoic acid
IX cis-7,trans-9-octadecadienoic acid
COOH
COOH
X trans-7,cis-9-octadecadienoic acid
COOH
XI cis-8,cis-10-octadecadienoic acid
XII trans-8,trans-10-octadecadienoic acid
COOH
COOH
COOH
XIII cis-8,trans-10-octadecadienoic acid
XIV trans-8,cis-10-octadecadienoic acid
XV cis-9,cis-11-octadecadienoic acid
COOH
COOH
COOH
XVI trans-9,trans-11-octadecadienoic acid
XVII cis-9,trans-11-octadecadienoic acid
XVIII trans-9,cis-11-octadecadienoic acid
COOH
COOH
COOH
XIX cis-10, cis-12-octadecadienoic acid
XX trans-10, trans-12-octadecadienoic acid
XXI cis-10, trans-12-octadecadienoic acid
COOH
COOH
COOH
XXII trans-10, cis-12-octadecadienoic acid
XXIII cis-11, cis-13-octadecadienoic acid XXIV trans-11, trans-13-octadecadienoic acid
COOH
COOH
XXV cis-11, trans-13-octadecadienoic acid
XXVI trans-11, cis-13-octadecadienoic acid
FIG. 1. Structures of linoleic acid (C18H32O2), monounsaturated acids (C18H34O2), saturated stearic acid (C18H36O2), and CLA isomers
(C18H32O2).
composition consisted of cis-9,trans-11, trans-10,cis-12, isomerization of methyl linoleate, but not of linoleic acid.
trans-9,trans-11, and trans-10,trans-12 isomers, and the The reason for this multistep route could be the slower
remaining 10% of cis-9,cis-11, trans-9,cis-11, cis-10,cis-12, isomerization rate for carboxylic acids (8). These reactions
and cis-10,trans-12 isomers, including a minor fraction with methyl linoleate were carried out over rhodium and
of the cis-11,cis-13 isomer (1). In the conventional alka- ruthenium supported by activated carbon between 200 and
◦
line isomerization methods for producing conjugated fatty 270 C (9, 10). Bimetallic ruthenium–nickel catalysts sup-
acids, ethylene glycol, dimethyl sulfoxide, or dimethylfor- ported by γ -Al2O3, SiO2Al2O3, C, and MgO have also been
◦
mamide are used as a solvent (5). When the end product tested at 200 C (11).
is used in the food industry, the choice of the solvent is
The use of linoleic acid as raw material is an attractive
limited to the ones that are harmless to the human body. alternative to methyl linoleate. Reactions with carboxylic
The main disadvantage of alkali isomerization of linoleic acids over heterogeneous catalysts are difficult to perform,
acid is the use of excess of strong basic potassium hy- so the question remains if it is possible to carry out the
droxide or sodium methoxide. Homogeneous catalysts double-bond migration reaction using linoleic acid as a raw
are tris(triphenylphosphine)chlororhodium (6) and arene material instead of methyl linoleate.
chromiumcarbonyl complexes (7), which are difficult to re-
use and not environmentally friendly.
2
. EXPERIMENTAL
Instead, a heterogeneous catalyst would be easy to
filter and reuse. Several possibilities exist to isomerize
methyl linoleate, the ester of linoleic acid, over heteroge-
neous catalysts. Isomerization reactions can be catalyzed
2
.1. Catalyst Synthesis and Characterization Methods
Fluka (Switzerland) supplied 5 wt% Ru/C catalyst. Na–
either by acidic, basic, or supported noble-metal catalyst. MCM-41 mesoporous molecular sieve was synthesized
The double-bond migration reactions over noble-metal- using a method described elsewhere (12) with some
supported catalysts reported in the literature consider the modifications. The reagents tetradecyltrimethylammonium

3
56
BERNAS ET AL.
bromide (CH3(CH2)13N(CH3)3Br), aluminium isopropox- ergy shift for all spectral lines, and the binding energy scale
ide ([(CH3)2CHO]3Al), and fumed silica (SiO2) were sup- was calibrated using the C 1s (284.5-eV) line as a reference.
pliedbyAldrich. Tetramethylammoniumsilicate(TMSiO2) In the analysis of the spectra the background shape was
and sodium silicate (Na4O4Si) were supplied by Sachem subtracted by Shirley algorithm prior to fitting of the Voigt
and Merck, respectively. A gel mixture was prepared and line shape for each peak in the spectra.
introduced in a 300-ml autoclave (Parr). The synthesis was
The Si/Al ratio of MCM was investigated by an X-ray
◦
carried out in an oven at 200 C for 24 h. After comple- fluorescence spectrometer (Siemens SRS 303) and the BET
tion of the synthesis, the autoclave was quenched and the specific surface areas were measured with an automa-
mesoporous material was filtered and washed with distilled tic physisorption–chemisorption apparatus (Sorptometer
◦
◦
water. The sample was dried at 110 C and calcined at 550 C. 1900, Carlo Erba Instruments).
The Na–MCM-41 was ion-exchanged with 1-M NH4Cl so-
Hydrogen storage capacities of the catalysts were inves-
lution, washed with distilled water to remove chloride tigated by performing hydrogen temperature-programmed
◦
ions, and dried at 100 C. The H–MCM-41 was obtained desorption (H2-TPD) measurements using a volumet-
◦
by calcination of NH4–MCM-41 at 550 C. H–MCM-41 ric flow apparatus (Autochem 2910, Micrometrics) with
was modified with Ni by an incipient wetness impregnation nitrogen–argon (N2 99.5%, Ar 0.5%, AGA) as a carrier
method using an aqueous solution of nickel nitrate. The gas. In the H2-TPD measurements, the Ni was reduced
◦
H–MCM-41 samples with varied Ni loadings were dried in situ by heating from room temperature to 450 C with
◦
◦
◦
at 100 C and calcined at 450 C for 4 h. The catalytic ma- a heating rate of 10 C/min in a 50 ml/min 100% hydrogen
terials were thereafter sieved to the particle size interval flow (AGA). The catalyst samples were kept at the reduc-
0
–180 µm. The particle size distribution of Ni/H–MCM-41 tion temperature in flowing hydrogen for 4 h, cooled to the
◦
was measured with sieves.
hydrogen adsorption temperature at a rate of 10 C/min,
The calcined Ni/H–MCM-41 catalyst samples were re- kept at the adsorption temperature for 1 h, and cooled to
◦
◦
◦
duced ex situ by heating from room temperature to 450 C 40 C at a rate of 3 C/min under flowing hydrogen. The hy-
◦
in flowing hydrogen (100 ml/min) with a heating rate of drogen adsorption temperatures were 100 and 180 C for
0 C/min, and keeping the catalyst at 450 C for 4 h. There- Ru and Ni, respectively. Thereafter the catalyst was flushed
◦
◦
1
after the catalysts were slowly cooled to room temperature with 50 ml/min N2/Ar for 45 min. The desorption heating
◦
◦
before directly admitting exposure to pure air.
rate was 10 C/min from 40 to 650 C and the catalyst was
◦
X-ray diffraction (XRD) measurements on Ru/C and kept at 650 C for 30 min. The analyses of the desorbed gases
Ni/H–MCM-41 were performed using a diffractometer were carried out continuously with a quadrupole mass spec-
(
Philips PW 1800) to determine structure, phase purity, and trometer (QTMD, Carlo Erba Instruments). For quantita-
crystallite size. The measuring conditions were as follows: tive measurements, the mass spectrometer was calibrated
generator voltage 50 kV, generator current 40 mA, Cu- for the hydrogen signal.
tube anode, automatic divergence slit, irradiated sample
◦
2.2. Isomerization Experiment
length 12 mm, receiving slit 0.2 mm, step size 0.025 , and
counting time 5.0 s per step. The effective crystallite size
Linoleic acid (C18H32O2) of 99% purity was supplied
was measured using the single-line and the approximate by Fluka (Switzerland). The solvents 1-octanol (C8H18O)
methods. of 99.5% purity and n-decane (C10H22) of 95% purity
X-ray photoelectron spectroscopy (XPS) measurements were supplied by Sigma-Aldrich (Germany) and Merck
on Ni/H–MCM-41 were utilized to analyze the chemical (Germany), respectively. The trans-10,cis-12, cis-9,trans-11,
composition of the catalyst surface. Before performing cis-11,trans-13, cis-9,cis-11, and trans-9,trans-11 isomers of
XPS, a calcined catalyst sample and a reduced catalyst sam- CLA (C18H32O2) were supplied by Matreya, Inc. (USA).
◦
ple were preactivated ex situ at 180 C under hydrogen in In a typical experiment, the catalyst was charged into a
the test reactor to get the same state of reduction as in the 200-ml stirred batch reactor, which was provided with a re-
catalytic experiments. Thereafter, n-decane was fed into the flux condenser and a heating jacket. The catalyst amount
reactor under stirring. The samples were filtered and stored was typically 200 mg, but the isomerization reaction was
in a special sample holder to avoid exposure to air. The ex- also performed with other catalyst quantities. Silicone oil
periments were carried out with a Perkin Elmer PHI 5400 served as a heat transfer fluid and the reactor was operated
ESCA system. The spectra were excited by the 1253.6-eV at atmospheric pressure. Catalyst preactivation by heating
unmonochromatized MgKα X rays with a total instrumen- the reactor to the activation temperature in a hydrogen at-
tal resolution of about 1 eV at the analyzer pass energy of mosphere started as soon as thecatalystentered the reactor.
◦
3
5.75 eV. The photoelectron emission angle was 30 rela- Isomerization reactions over a catalyst preheated in a ni-
tive to the sample surface normal. Chamber pressure was trogen atmosphere were also performed. The preactivation
−8
below 2 × 10 torr during the measurements. The sample was carried out in situ for 1 h (including heating time) at
charging during the measurements caused a constant en- a hydrogen or nitrogen flow of approximately 100 ml/min.

ISOMERIZATION OF LINOLEIC ACID
357
◦
Ru and Ni catalysts were preactivated at 100 and 180 C, re-
3. RESULTS
spectively. Linoleic acid ( 200 mg )was mixed with 70 ml of
solvent in a glass tube. The initial concentration of linoleic
acid was 0.010 mol/dm . Air above the reaction mixture and
3
.1. Catalyst Characterization
3
XRD measurements on Na–MCM-41 confirmed that the
oxygen dissolved in the liquid phase were purged out by a synthesized mesoporous material had MCM-41 structure.
nitrogen flow of 100 ml/min through the reactant solution. The XRD pattern demonstrated in Fig. 2 exhibits the pres-
A nitrogen flow at the same rate was also fed through the ence of four peaks, indicating the presence of a hexago-
reactor to produce an inert atmosphere while cooling the nal MCM-41 phase (13). XRD measurements on 5 wt%
◦
reactor to the reaction temperature of 80, 100, or 120 C. Ni/H–MCM-41 indicated a reduction of NiO. The crystal-
The liquid phase containing linoleic acid and the solvent lite size of the Ni oxide was determined to be 9.89 nm
was fed into the reactor after a purging time of 20 min and by the approximate method and 9.47 nm by the single-
the reaction time was initialized to zero. The temperature line method. The crystallite size of the metallic Ni was de-
◦
of the reflux condensers cooling medium was set to −20 C. termined to be 5.77 nm by the approximate method and
Stirring baffles were used inside the reactor. The typical 6.95 nm by the single-line method. The metal particle size
stirring rate was 800 rpm but the isomerization reaction of the 5 wt% Ru/C catalyst was higher. According to the ap-
was also conducted at other stirring rates. For calculation proximate method, the Ru crystallite size was determined
of the concentration versus time dependence, samples were to be 50.2 nm.
withdrawn from the reactor at certain intervals through a
sampling valve.
Ni 2p photoelectron spectra of H–MCM-41-supported Ni
catalysts from both reduced and calcined states are shown
in Fig. 3. The satellite structure in both spectra at the bind-
ing energy 861–862 eV is clear evidence that Ni oxides are
2
.3. Analytical Procedure
The samples were silylated and analyzed by a present. The Ni 2p spectrum of the calcined catalyst is char-
temperature-programmedgaschromatograph(GC)usinga acteristic of the oxides because no spectral structures are
2
5-m HP-5 column ( 0.20-mm inner diameter, 0.11-µm film present below 855 eV (14). The satellite line positions are
layer) and a flame ionization detector (FID) operating at about 6 eV to higher binding energies from the main lines,
◦
2
90 C. In the silylation procedure, 10 µl of samples from the and they can be used as references in the interpretation of
reactor were added into glass tubes that had been washed the main lines originating from different Ni oxide states,
with methyl tert-butyl ether (MTBE, C5H12O). 50 µl of and to estimate the presence of metallic Ni in the catalyst.
0
.5-mg/ml C17:0 fatty acid solution was added to each sam-
ple to serve as an internal standard for the GC. MTBE and
solvent were evaporated in a stream of nitrogen with water
as heating medium and the samples were further dried in a
◦
vacuumdesiccatorat40 Cfor20min. Thesamplesweredis-
solved in 20 µl pyridine. Acros Organics supplied 80 µl N,O-
bis(trimethylsilyl)trifluoroacetamide (BSTFA) of 98% pu-
rity and 40 µl trimethylchlorosilane (TMCS) of 98% purity,
which were added to the residue. The solutions were kept
◦
in an oven at 70 C for 45 min and were thereafter ready for
analysis by GC. The evaporation and silylation operations
were conveniently performed with the entire series of sam-
ples from one isomerization reaction in parallel to eliminate
effects of decomposition. 1-µl samples were injected into
the GC with an autosampler unit using the injector temper-
◦
ature 260 C. Helium served as a carrier gas with a flowrate
of 0.9 ml/min and a split ratio of 1 : 20. The column temper-
◦
ature was initially 150 C. The temperature was increased
◦
from 150 to 230 C (0.5 min after the injection) with a rate
of 7 C/min, and thereafter from 230 to 290 C with a rate of
0 C/min. Thereafter the column was purged at 290 C for
0 min. Peaks corresponding to conjugated dienoic isomers
◦
◦
◦
◦
1
1
of linoleic acid were identified by injecting the authentic
samples. Peak identities were verified by analysis with a gas
chromatograph-mass spectrometer (GC/MS) applying the
same GC conditions.
FIG. 2. XRD pattern of (I) Na–MCM-41, (II) calcined 5 wt% Ni/H–
MCM-41, (III) calcined and reduced 5 wt% Ni/H–MCM-41, (IV) used
5 wt% Ni/H–MCM-41, and (V) 5 wt% Ru/C.

3
58
BERNAS ET AL.
TABLE 2
Pore Size Distributions of Ru/C and Ni/H–MCM-41
Relative volume (%)
Pore ranges (nm)
5 wt% Ru/C
5 wt% Ni/H–MCM-41
0–0.9
0
0
0
1
1
2
5
.9–1.0
.0–1.5
.5–2.0
.0–5.0
.0–10.0
2.6
5.9
4.2
20.8
18.1
48.4
5.2
57.2
1.5
5.3
1.3
29.6
1
0.0–100.0
FIG. 3. Ni 2p photoelectron spectra of calcined and reduced 5 wt%
Ni/H–MCM-41 samples after background subtracting by Shirley algo-
rithm. The peak originating from metallic nickel is marked by M.
TPD measurements on reduced 5 wt% Ni/H–MCM-41
indicated that as much as 74% of the total amount of ad-
sorbed hydrogen remained on the catalyst surface after a
temperature increase to 180 C, which was used as cata-
lyst preactivation temperature for the Ni catalysts. For
◦
The 2p main line of metallic nickel (marked by M in Fig. 3)
has a binding energy which is more than 2 eV lower than
that of the oxides. The satellite structure associated with
the metallic nickel spectrum is quite broad and is located
in the region between the main and satellite lines of the ox-
ides. In the analysis of the spectra the curve-fitting proce-
dure with the appropriate line shapes was used. The amount
of metallic Ni in the reduced catalyst was estimated to be
about 30%.
The Si/Al ratio of Na–MCM-41 was determined to be 20
as measured by X-ray fluorescence. BET specific surface
area measurements, the pore size distribution, and parti-
cle size distribution of the catalytic materials are presented
in Tables 1–3. Ni/H–MCM-41 has slightly larger BET area
5
wt% Ru/C a fraction of 90% of the total hydrogen amount
remained on the surface after a temperature increase to
◦
1
00 C, which was used as preactivation temperature in the
catalytic experiments over Ru. The TPD patterns in Fig. 4
indicate spillover hydrogen for both Ru/C and Ni/H–MCM-
41. Generally, the amount of spillover hydrogen on the
surface of the catalyst support material is proportional to
the square root of the catalyst reducing time (15), and its
amount increases with reduction temperature.
3.2. Catalytic Activity and Selectivity
The isomerization properties of several supported metal
than Ru/C. Ni/H–MCM-41 has a major fraction of pores in catalysts were investigated. Investigated catalysts were Ru,
the range of 1.0–1.5 nm while Ru/C has larger pores and a Ni, Pt, Pd, Rh, Ir, Os, and bimetallic Pt/Rh supported by ac-
fraction of the same magnitude between 10 and 100 nm. tivated carbon, Al O , Al/Si, SiO , MCM-22, H–MCM-41,
2
3
2
It should be noted that approximately only 30% of the and zeolites Y and β. After some catalyst screening experi-
MCM-41 pores are larger than 10 nm, which is close to ments, it was concluded that Ru, Ni, and Pt have good prop-
the crystallite sizes of metallic Ni and NiO, as measured erties for isomerization of linoleic acid. It became appar-
by XRD. Hence most of the nickel particles are located on ent that the activity and the isomerization selectivity were
the outer surface of the MCM pores. Table 3 indicates that sensitive to the surface structure and hydrogen adsorption
the particle sizes of 5 wt% Ni/H–MCM-41 are distributed capacity of the metal-support combinations. Metals with
around 71–90 µm but that a large fraction is also present in high hydrogen storage capacity such as Pd showed high
the interval 0–45 µm. Ru/C has a large fraction of particles
in the size intervals 45–63 and 71–90 µm.
TABLE 3
TABLE 1
Particle Size Distributions of Ru/C and Ni/H–MCM-41
BET Specific Surface Area Measurements
Mass fraction (%)
2
Catalyst
BET (m /g)
Size ranges (µm)
5 wt% Ru/C
5 wt% Ni/H–MCM-41
5
5
wt% Ru/C
wt% Ru/C, used catalyst
841
831
1214
913
1051
1021
77
0–45
45–63
63–71
6.2
20.5
3.3
31.9
16.7
10.6
10.8
22.7
5.9
13.7
23.7
12.2
8.7
Na–MCM-41
5 wt% Ni/H–MCM-41, not reduced
5 wt% Ni/H–MCM-41, reduced
15 wt% Ni/H–MCM-41, reduced
5 wt% Ni/H–MCM-41, used catalyst
71–90
90–125
125–150
150–180
13.2

ISOMERIZATION OF LINOLEIC ACID
359
sented in Fig. 5. The mole fraction in the figures refers to
the concentration of the specific fatty acid compound rel-
ative to the total concentration of fatty acids. The concen-
tration of linoleic acid decreased until the raw material was
totally converted. The concentrations of one conjugated
dienoic isomer of linoleic acid and monounsaturated acids
initially increased with time and started to decrease after
reaching a maximum value. The concentration of the conju-
gated isomer started to decrease earlier than the concentra-
tions of the monounsaturated compounds. The concentra-
tion of stearic acid increased monotonically but this product
started to build up later than the monounsaturated prod-
ucts. Hydrogenation of linoleic acid to monounsaturated
acids and isomerization of linoleic acid are two competing
reactions. It can also be concluded that hydrogenation of
I
II
0
100
200
300
400
500
600
700
temperature (C)
FIG. 4. H2-TPD profiles: I, 5 wt% Ru/C (H2 adsorption tempera- linoleic acid proceeds consecutively via monounsaturated
◦
◦
ture = 100 C); II, reduced 5 wt% Ni/H–MCM-41 (H2 adsorption tempera-
acids to stearic acid. When the reaction was carried out un-
der hydrogen, no other reactions were detected except for
isomerization and hydrogenation.
ture = 180 C).
Despite the fact that the presence of hydrogen leads to
hydrogenated products, the isomerization reaction in a ni-
trogen atmosphere was dramatically enhanced when the
catalyst was preactivated with hydrogen. When the cata-
lyst was preactivated under hydrogen and the reaction was
carried out in 1-octanol over 50 mg of 5 wt% Ru/C in an
inert atmosphere (e.g., in the absence of hydrogen) with a
nitrogen flow through the reactor, otherwise using the same
reaction conditions, the main reaction that took place was
isomerization. All reactions in a nitrogen atmosphere were
carried out for a specific duration of time (6 h). The overall
reaction rate was lower than the rate in the previous ex-
periment in a hydrogen atmosphere, but the isomerization
selectivity was higher than the hydrogenation selectivity
and only a small trace of stearic acid was formed. The to-
tal mole fraction of CLA after 6 h was 33%. The reaction
temperature affected the total conversion and the catalyst
activityandhighselectivityforthedouble-bondhydrogena-
tion reaction, whereas most of the other tested metals, espe-
cially Ru, Ni, and Pt, favored double-bond migration. Very
few catalysts, including Ru/C and Ni/H–MCM-41, exhib-
ited feasible conversion. For Ni/Al2O3 and Ni/SiO2, which
have much wider pores than MCM-41, more of the metal
particles are located inside these pores; however, conver-
sion and isomerization selectivity at the applied standard
test conditions were only 10 and 25% for Ni/Al2O3 and 43
and 37% for Ni/SiO2, respectively, after 6 h reaction time.
Therefore the Ru/C and Ni/H–MCM-41 catalysts were se-
lected for further studies.
◦
Preliminary isomerization experiments at 120 C in a ni-
trogen atmosphere over 5 wt% Ru/C, preheated for 1 h at
◦
1
00 C in pure nitrogen, resulted in 89% selectivity for iso-
merization. The remaining reaction was hydrogenation (in-
cluding both monounsaturated and stearic acid). However,
the total converted amount of linoleic acid after 6 h was
only 11%. In general, protic solvents generate more hydro-
gen chemisorbed sites on the catalyst surface than nonpo-
lar solvents. In these experiments, n-decane was used as the
solvent to have no hydrogen in the system and the catalyst
quantitywas200mg. Polymerization, cracking, andgeomet-
ric or skeletal isomerization of fatty acids were not observed
at these conditions. No other reactions were detected ex-
cept for isomerization of linoleic acid to CLAs and hydro-
genation of C18:2 fatty acids to C18:1 and C18:0 fatty acids.
Hence, in the following discussion, the term isomerization
refers to double-bond migration to form a conjugated sys-
tem and the sum of the isomerization and hydrogenation
selectivities is 1.
1
0.8
0.6
0.4
0.2
0
0
10
20
30
40
50
60
70
80
reaction time (min)
◦
FIG. 5. Kinetic experiment in a hydrogen atmosphere: ꢀ, linoleic
acid; ꢁ, monounsaturated acids; ꢂ, stearic acid; ꢃ, conjugated dienoic
Experiments at 100 C in a hydrogen atmosphere over
the same Ru/C, preactivated in pure hydrogen, resulted in
hydrogenated products. In this case 1-octanol was used as
solvent and the catalyst quantity was 50 mg. The concen-
◦
isomer of linoleic acid. Conditions: reaction temperature, 100 C; stir-
ring rate, 800 rpm; solvent, 70 ml 1-octanol; raw material, 200 mg linoleic
◦
acid; catalyst, 50 mg 5 wt% Ru/C (preactivated at 100 C for 1 h under
tration versus time dependence for this experiment is pre- hydrogen).

3
60
BERNAS ET AL.
1
The competing isomerization and hydrogenation reac-
(
a)
tions are influenced by the concentration of chemisorbed
hydrogen. If the hydrogen concentration in the solvent is
high (as when the reaction is carried out in a hydrogen at-
mosphere over a catalyst preactivated under hydrogen), the
concentration of hydrogen-chemisorbed sites on the cata-
lyst surface will also be high. This increases the rate for
the hydrogenation of linoleic acid, consecutively proceed-
ing via monounsaturated acids to stearic acid. If, on the
other hand, the concentration of chemisorbed hydrogen is
low (as when hydrogen is chemisorbed during the preacti-
vation and consumed during the reaction under a nitrogen
pressure), isomerization is preferred over hydrogenation.
The fact that isomerization can occur in absence of hydro-
gen preadsorption is also an important observation.
0
.8
.6
.4
.2
0
0
0
0
0
60
120
180
240
300
360
reaction time (min)
1
(
b)
0
.8
.6
.4
◦
For comparison, 5 wt% Ni/H–MCM-41 reduced at 450 C
◦
0
in hydrogen for 4 h and preactivated in hydrogen at 180 C
for 1 h was tested. Also in this case the reaction was car-
0
◦
ried out at 120 C in both 1-octanol and n-decane to evalu-
ate the effects of a slightly protic and a nonpolar solvent.
Figure 7 demonstrates the general trend of results obtained
0.2
0
0
0.2
0.4
0.6
conversion
0.8
1
1
(a)
FIG. 6. Kinetic experiment in a nitrogen atmosphere: (a) concentra-
0.8
.6
.4
.2
tion versus time dependence and (b) selectivity versus conversion depen-
dence. ꢀ, linoleic acid; ꢁ, monounsaturated acids; ꢂ, stearic acid; ꢃ, con-
jugated linoleic acids; ꢄ, selectivity for isomerization. Conditions: reaction
0
◦
temperature, 120 C; stirring rate, 800 rpm; solvent, 70 ml n-decane; raw
0
material, 200 mg linoleic acid; catalyst, 200 mg 5 wt% Ru/C (preactivated
◦
at 100 C for 1 h under hydrogen).
0
quantity affected both total conversion and selectivity. The
conversion increased from 46 to 53% when the reaction
temperature was increased from 100 to 120 C, but the iso-
0
0
60
120
180
240
300
360
◦
reaction time (min)
merization selectivity was still 74%, resulting in 39 mol%
CLA. The isomerization selectivity decreased from 74 to
1
(
b)
4
2% when 200-mg catalyst was used instead of 50 mg, due
0
.8
0.6
0.4
to the higher hydrogen content of the catalyst. With this
larger catalyst quantity the conversion and mole fraction
CLA were 78 and 33% respectively. Using n-decane in-
stead of 1-octanol, the conversion stayed at 77%, but the
isomerization selectivity increased from 42 to 69%, result-
ing in a yield of 53% conjugation and 24% hydrogenation.
The concentration versus reaction time and selectivity ver-
sus conversion dependence for this experiment are demon-
strated in Fig. 6. The hydrogenation selectivity increases
with conversion since the chemisorbed hydrogen gener-
ated in the preactivation step is being consumed. In this ex-
0
.2
0
0
0.1
0.2
0.3
0.4
0.5
0.6
conversion
FIG. 7. (a) Concentration versus time dependence and (b) selectiv-
periment, the yields of monounsaturated and stearic acids ity versus conversion dependence for isomerization of linoleic acid over
5
wt% Ni/H–MCM-41 in n-decane. ꢀ, linoleic acid; ꢁ, monounsaturated
were 23 and 1%, respectively. The H2 required for the ob-
served hydrogenation is 17.8 mmol/gRu. The H2 quantity
chemisorbed on the Ru surface according to monolayer
adsorption is 9.89 mmol/gRu, indicating that the spillover
hydrogen (Fig. 4) is involved.
acids; ꢂ, stearic acid; ꢃ, conjugated linoleic acids; ꢄ, selectivity for isomer-
ization. Conditions: reaction temperature, 120 C; stirring rate, 800 rpm;
solvent, 70 ml; raw material, 200 mg linoleic acid; catalyst quantity, 200 mg;
catalyst particle size interval, 0–180 µm (catalyst reduced and preactivated
◦
◦
at 180 C for 1 h under hydrogen).

ISOMERIZATION OF LINOLEIC ACID
361
for 5 wt% Ni/H–MCM-41. The overall reaction rate was tivity was 71 and 29% respectively, resulting in 37% con-
high in the beginning of the reaction but decreased with jugation. The yields of monounsaturated and stearic acids
the conversion. After 2 h the concentrations of the fatty were 12 and 3%, respectively, indicating that the required
acids reached approximately constant values and the con- amount of molecular H2 for the observed double-bond hy-
sumption rate of linoleic acid was almost zero after a certain drogenation is 12.8 mmol/gNi. The H2 quantity chemisorbed
period of time. Conversion and selectivities at this point are on the Ni metal surface according to the monolayer ad-
influenced through a complex connection by (a) the total sorption theory is 17.0 mmol/gNi, but TPD measurements
amount of chemisorbed hydrogen generated in the cata- showed that 70% of the total amount of chemisorbed hy-
lyst preactivation; (b) reaction conditions; (c) deactivation drogen remained on the surface after temperature increase
◦
of the catalyst; (d) availability of active vacant sites on the to 180 C, which was used as preactivation temperature, re-
catalyst surface; and (e) competitive adsorption between sultingin11.9mmol/gNi formonolayeradsorption. Itisclear
linoleic acid, solvent, and reaction products. When all hy- that the hydrogenation is limited to the amount of hydro-
drogen preadsorbed on the catalyst surface is consumed, gen that is chemisorbed in the hydrogen preactivation step.
the consumption rate of linoleic acid is limited to hydrogen For this particular catalyst, n-decane gave higher conver-
derived from the solvent. In essence, the ease of availability sion and higher isomerization selectivity than 1-octanol.
of hydrogen at the catalyst preactivation influences the hy- This may not be a general rule for other catalysts since
drogen coverage on the catalyst surface. This in turn affects experiments over bimetallic platinum–rhodium on carbon
the activity and selectivity pattern observed.
in 1-octanol and n-decane resulted in higher conversion for
No significant leaching of the impregnated Ni salt by the 1-octanol. When the reaction was carried out in a nitro-
fatty acids or by the solvent was detected, which indicates gen atmosphere over a hydrogen preactivated catalyst, no
that the isomerization and hydrogenation reactions do not other reactions were detected except for isomerization and
occur in the homogeneous phase. The Ni content of the hydrogenation of linoleic acid.
fatty acid solution after a reaction time of 6 h was 10.6 µg/l
as determined by direct current plasma atomic emission gations. For n-decane, the overall isomerization rate was
spectrometry (DCP-AES) (16). higher than the overall hydrogenation rate. A typical pat-
n-decane was chosen as a solvent for further investi-
In a catalyst regeneration study, both Ru and Ni cata- tern for these isomerization reactions in an inert atmo-
lysts were separated from the fatty acid solution by filtra- sphere over a catalyst preactivated under a pressure of
tion, washed with acetone in a stirred tank, dried overnight hydrogen is that the yields of monounsaturated acids and
◦
at 100 C, and re-reduced in flowing hydrogen using 400 stearic acid increase at the beginning of the reaction, and
◦
and 450 C as reduction temperatures for Ru/C and Ni/H– thereafter reach a stable value. The yield of CLA, in turn,
MCM-41, respectively. In the isomerization experiment typically increases slightly and continuously with the re-
demonstrated in Fig. 7, the BET area of the Ni/H–MCM-41 action time (as shown in Figs. 6 and 7). The isomeriza-
2
catalyst decreased from 1051 to 77 m /g and the catalyst tion selectivity, in other words, increases with the con-
was no longer active in a second consecutive experiment. version and the hydrogenation selectivity decreases with
XRD measurements on used Ni/H–MCM-41 (Fig. 2) did the conversion, once again indicating that the coverage of
not indicate any changes in the MCM structure. It can thus chemisorbed hydrogen on the catalyst surface affects the
be concluded that a collapse of the structure is not respon- isomerization and hydrogenation rates.
sible for the deactivation, but blocking of pores and coke
formation. The chemical nature of the residue requires a
separate study, which is currently being performed. In con-
3.3. Isomer Distribution
trast to Ni/H–MCM-41, Ru/C did not deactivate. When the
It is clear that the nature of the used solvent affects
experiment was carried out over Ru/C, the surface area the isomer distribution. When the isomerization reaction
2
decreased slightly from 841 to 831 m /g, but the catalytic over 5 wt% Ni/H–MCM-41 was performed in n-decane as
performance did not indicate any deactivation.
demonstrated in Fig. 7, the 6-h sample included linoleic
When 1-octanol was used as a solvent, the total converted acid, monounsaturated acids, stearic acid, trans-9,trans-11,
amount of linoleic acid was only 17%, and the isomeriza- cis-9,trans-11, trans-10,cis-12, cis-9,cis-11 CLA isomers, and
tion selectivity was only 13%, resulting in 2% conjugation. six unknown CLA isomers. The composition of fatty acids
Although the concentration decrease of linoleic acid was comprised 49 mol% linoleic acid, 12% monounsaturated
very low, as much as 120 mg of a branched C20:0 compound acids, 3% stearic acid, and 37% conjugated dienoic isomers
was formed as a by-product, indicating that 1-octanol is very of linoleic acid. The distribution of conjugated dienoic iso-
reactive and not suitable as a solvent in the isomerization mers was 38% trans-9,trans-11-CLA, 10% cis-9,trans-11-
reaction of linoleic acid over this particular catalyst. Traces CLA, 4% trans-10,cis-12-CLA, 2% cis-9,cis-11-CLA, and
of 3,6-dimethyl octane were also found in the 6-h sample.
the remaining 45% unknown conjugated dienoic isomers.
When n-decane was used as a solvent, the total converted OftheCLAobtained, 39%consistedofonesingleunknown
amount of linoleic acid was 51%. The isomerization selec- isomer.

3
62
BERNAS ET AL.
When, on the other hand, the isomerization reaction linoleic acid increased with temperature, but the selectivi-
over 5 wt% Ni/H–MCM-41 was performed in 1-octanol, the ties for isomerization and hydrogenation were not strongly
-h sample included linoleic acid, monounsaturated acids, affected by the temperature. The selectivities alternated
6
stearic acid, and cis-9,trans-11, trans-9,trans-11, trans-10, around 70% for isomerization and 30% for hydrogenation
cis-12 CLA isomers. The composition of fatty acids con- at varied reaction temperature. The total conversion and
sisted of 83 mol% linoleic acid, 13% monounsaturated conjugation increased from 11 and 7% to 51 and 37%, re-
acids, 2% stearic acid, and 2% conjugated isomers of spectively, when the temperature was increased from 80 to
◦
linoleic acid. The distribution of conjugated dienoic iso- 120 C. The reaction temperature did not affect the selec-
mers was 45% cis-9,trans-11-CLA, 40% trans-9,trans-11- tivities for specific isomers of CLA.
CLA, and 15% cis-10,cis-12-CLA.
Ru/C showed higher selectivity toward the cis-9,trans-11 3.5. Effect of Catalyst Reduction
and trans-10,cis-12 isomers of CLA (desired products) than
Ni/H–MCM-41. When the experiment was conducted over
The surface area of calcined 5 wt% Ni/H–MCM-41 in-
2
creased from 913 to 1051 m /g when the catalyst was re-
5
wt% Ru/C in n-decane as shown in Fig. 6, the 6-h sample
◦
duced in a hydrogen flow at 450 C for 4 h. Conversion
included linoleic acid, monounsaturated acids, stearic acid,
cis-9,trans-11, trans-10,cis-12, cis-9,cis-11, trans-9,trans-11
CLA isomers, and two unknown CLA isomers. The fatty
acid composition comprised 23 mol% linoleic acid, 23%
monounsaturated acids, 1% stearic acid, and 53% CLA
isomers. The CLA isomer distribution was 29% trans-9,
trans-11-CLA, 24% cis-9,trans-11-CLA, 16% trans-10,cis-
and selectivities were not strongly affected by the Ni re-
duction. TPR studies on Ni/Al2O3, Ru/Al2O3, Ru/SiO2,
and Ru/Na–Y (17, 18) have shown that after oxygen treat-
◦
ment, Ru/Na–Y already started to consume H2 at 100 C,
◦
whereas the hydrogen uptake started at 180 to 200 C for
Ru/Al2O3 and Ru/SiO2, and the catalysts were easily fully
reduced. Oxidic Ni catalysts with varied Ni loading showed
1
2-CLA, 4% cis-9,cis-11-CLA, and the remaining 27% un-
◦
a broad H2 consumption range of 230 to 580 C with a
known CLA isomers. Of the obtained CLA 26% consisted
of a single unknown isomer. The selectivities toward spe-
cific isomers of CLA did not depend on the reaction time
or the conversion.
◦
maximum at 380 C, but TPR profiles on Ni catalysts pre-
treated with high-temperature reduction and passivation in
nitrogen-diluted oxygen showed a maximum H2 consump-
◦
tion at 270 C. The integrated hydrogen consumption was
quantified and normalized to the Ni content. Assuming that
the stoichiometry of nickel oxide is NiO, the reduction de-
3
.4. Effect of Reaction Temperature
Activity and selectivity data on Ni/H–MCM-41 catalyst gree of the reduced and passivated Ni catalysts was in the
are reported in Table 4. The total converted amount of range of 28–40%.
TABLE 4
Activity and Selectivity Data on Ni/H–MCM-41 for Isomerization of Linoleic Acid
f
Reaction
temp.
( C)
Metal
loading
(wt%)
Cat.
mass
(mg)
Particle
size
(µm)
Initial
conc.
(mol dm
Initial rate
Conv.a
(%)
Conj.b
(%)
Hydr.c
(%)
(×10
−7
Catalyst
reduction
◦
−3
SId
SHe
−1 −1
Solvent
)
mol s
g )
n-decane
n-decane
n-decane
n-decane
n-decane
n-decane
n-decane
n-decane
n-decane
120
120
120
120
120
120
100
80
5
5
5
0
15
5
5
5
5
5
200
200
200
200
200
400
200
200
200
200
0–180
0–180
45–63
45–63
45–63
0–180
0–180
0–180
0–180
0–180
No
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.005
0.01
56
51
34
22
64
72
22
11
70
17
40
37
25
16
45
41
15
7
16
15
9
0.71
0.71
0.73
0.71
0.70
0.57
0.70
0.64
0.60
0.13
0.29
0.29
0.27
0.29
0.30
0.43
0.30
0.36
0.40
0.87
3.8
3.2
1.4
0.9
3.9
2.6
1.0
0.6
2.6
1.4
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
6
19
25
7
4
28
15
120
120
42
2
1
-octanol
Note. Conditions: raw material, linoleic acid; solvent quantity, 70 ml; stirring rate, 800 rpm; reaction pressure, 1 atm of nitrogen; hydrogen adsorption
◦
time, 1 h; hydrogen adsorption temperature, 180 C; reaction time, 6 h.
a
All percentages refer to mole percentages.
Includes all CLA isomers.
Includes both monounsaturated acids and stearic acids.
b
c
d
Selectivity for isomerization: SI = conjugation/total conversion.
e
Selectivity for hydrogenation: SH = hydrogenation/total conversion.
f
Based on total converted amount of linoleic acid after 60 min and catalyst mass.

ISOMERIZATION OF LINOLEIC ACID
363
1
how conversion and yield of CLA increased when the cata-
lyst metal loading was increased. Selectivities for isomeriza-
tion and hydrogenation were not strongly affected by the
loading, but the linoleic acid converted after 6 h is a linear
function of the Ni loading with an R-squared value of 0.99.
No changes in selectivities for specific isomers of CLA were
detected as the metal loading was varied.
0
.8
0
.6
3.8. Effects of Catalyst Quantity
0.4
The total converted amount of linoleic acid over reduced
5
wt% Ni/H–MCM-41 at 6 h increased from 51 to 72%
0
.2
when 400-mg catalyst was used instead of 200 mg at the
same reaction conditions, but the isomerization selectivity
decreasedfrom71to57%becausetherewasmoreadsorbed
hydrogen available for the competing hydrogenation re-
action, resulting in an increase of the yield of conjugated
linoleic acids from 37 to 41% as demonstrated in Fig. 8.
The initial overall reaction rate was practically the same
0
0
60
120
180
240
300
360
reaction time (min)
FIG. 8. Influence on concentration decrease of linoleic acid and yield
−
7
−1 −1
−7
−1 −1
of CLA of varied Ni loading, catalyst particle size, catalyst quantity, initial (3.2 × 10 mols
g
and2.7 × 10 mols
g )as200mg
concentration of linoleic acid, and reduction state of Ni. ꢀ, 5 wt% Ni, 0– reduced 5 wt% Ni/H–MCM-41 was changed to 400 mg, but
3
1
0
6
6
4
0
1
80 µm, 200-mg not-reduced catalyst, c0 = 0.01 mol/dm ; ꢁ, 5 wt% Ni,
the concentration of linoleic acid as a function of the re-
action time decreased more rapidly for the higher catalyst
quantity. The catalyst mass did not affect the selectivities
3
–180 µm, 200-mg reduced catalyst, c0 = 0.01 mol/dm ; ꢂ, no metal, 45–
3
3 µm, 200-mg reduced catalyst, c0 = 0.01 mol/dm ; ꢃ, 5 wt% Ni, 45–
3
3 µm, 200-mg reduced catalyst, c0 = 0.01 mol/dm ; ꢄ, 15 wt% Ni,
3
5–63 µm, 200-mg reduced catalyst, c0 = 0.01 mol/dm ; +, 5 wt% Ni, for specific isomers of CLA. The dissolubility of Ni in the
3
–180 µm, 400-mg reduced catalyst, c0 = 0.01 mol/dm ; ×, 5 wt% Ni, 0– fatty acid solution after the 6-h reaction time was still low at
3
80 µm, 200-mg reduced catalyst, c0 = 0.005 mol/dm . Conditions: reac-
the 400-mg catalyst mass level. The concentration of Ni in
the liquid phase was only 8.7 µg/l as indicated by DCP-AES
measurements.
◦
tion temperature, 120 C; stirring rate, 800 rpm; solvent, 70 ml n-decane
◦
(
catalyst preactivated at 180 C for 1 h under hydrogen).
3
.6. Effect of Catalyst Particle Size
3.9. Effect of Initial Linoleic Acid Concentration
The selectivities for isomerization and hydrogenation
The reactant mass to catalyst mass ratio was decreased by
were also not strongly affected by the catalyst particle size 50% because 400-mg catalyst was used instead of 200 mg.
but the total conversion decreased from 51 to 34% when This ratio was also decreased by 50% because the initial
the particle size interval of the 5 wt% Ni/H–MCM-41 was linoleic acid concentration was decreased from 0.010 to
3
narrowed from 0–180 µm to 45–63 µm as demonstrated 0.005 mol/dm . The converted linoleic acid over reduced
in Fig. 8. The initial overall reaction rate based on con- 5 wt% Ni/H–MCM-41 at 6 h increased from 51 to 70%
version at 60 min and catalyst mass decreases dramatically when 100 mg of linoleic acid was used instead of 200 mg at
−
7
−7
−1 −1
from 3.2 × 10 to 1.4 × 10 mol s
g
. The reason for the same reaction conditions, but the isomerization selec-
this decrease in conversion and initial rate might be that tivity decreased from 71 to 60% due to the higher amount
the fraction with particle sizes less than 45 µm (as much as of hydrogen in the system relative to the amount of the re-
2
3 mass%, as reported in Table 3) was removed, indicating actant, resulting in an increase of the yield of conjugated
that internal diffusional limitations exist. The particle size linoleic acids from 37 to 42%. It should be noted that the
does not affect selectivities for specific CLA isomers.
same conversion and selectivities could be obtained by ei-
ther increasing the catalyst mass or decreasing the reactant
mass by a factor of 2, as expected.
3
.7. Effect of Catalyst Metal Loading
The BET area decreased as the metal loading was in-
creased. Considering 0, 5, and 15 wt% Ni, the initial overall
reaction rate was proportional to the catalyst metal load-
3
.10. Effects of Agitation
Influence of agitation was experimentally determined by
ing. The effect of loading of Ni on the support material was varying the stirring rate. The isomerization reaction was
studied by using a very narrow catalyst particle size interval carried out at 400, 600, 800, and 1000 rpm, otherwise using
(
45–63 µm). A special experiment was carried out with the the same reaction conditions. Catalytic activity, selectivi-
proton form, H–MCM-41, to investigate the catalytic prop- ties toward isomerization and hydrogenation, and the initial
erties of the support material. Figure 8 also demonstrates overall reaction rate at these stirring speeds are presented

3
64
BERNAS ET AL.
TABLE 5
derived from monounsaturated acids, respectively. (Ci )ads
denotes chemisorbed organic molecule i, etc. After pre-
activation under hydrogen (step 1) initially the compound
containing one or two double bonds is chemisorbed on the
surface of the metal or on the surface of the support mate-
rial for the case with no metal loading (step 2). Thereafter, a
hydrogen atom derived from a hydrogen chemisorbed site
is added to the chemisorbed acid to give a chemisorbed
half-hydrogenated intermediate (step 3). If the hydrogen
coverage on the catalyst surface is low (as at the end of the
reaction when conjugation is preferred), hydrogen abstrac-
Activity and Selectivity Data on Reduced Ni/H–MCM-41
at Varied Stirring Rate
Stirring
rate
rpm)
Initial overall
reaction rate
Conv. Conj. Hydr.
7
−1 −1
g )
(
(%)
(%)
(%)
SC
SH
(×10− mol s
4
6
8
00
00
00
45
54
45
52
32
34
32
34
13
20
12
18
0.72 0.28
0.64 0.36
0.72 0.28
0.65 0.35
3.4
4.5
3.1
4.1
1
000
Note. Conditions: raw material, 200 mg linoleic acid; solvent, 70 ml n- tion by the metal takes place from an adjacent carbon atom
◦
decane; reaction temperature, 120 C; Ni loading, 5 wt%; catalyst mass, (step 4). This could lead to a double-bond migration de-
2
00 mg; catalyst particle size interval, 0–180 µm; reaction pressure, 1 atm
pending on which carbon atom the hydrogen is abstracted
from. Thereafter the chemisorbed product is desorbed from
the surface of the catalyst metal or the support material
of nitrogen; hydrogen adsorption time, 1 h; hydrogen adsorption temper-
ature, 180 C; reaction time, 3 h.
◦
(
step 5). On the other hand, if the coverage of hydrogen is
in Table 5. It is clear from these results that the diluted sys-
tem is well mixed and it is obvious that all isomerization
experiments were performed on the plateau of the initial
rate against the stirring rate.
high (as in the beginning of the reaction when hydrogena-
tion is preferred over isomerization), a second hydrogen
atom is added to the chemisorbed half-hydrogenated in-
termediate (step 6), leading to a hydrogenated product. In
essence the addition of hydrogen to olefins are surface re-
actions involving the addition of chemisorbed hydrogen to
chemisorbed olefins. Steps 3 and 6 correspond to addition
of two hydrogen atoms to the same side of the chemisorbed
olefin (cis-addition). We have chosen to model step 6 as an
irreversible reaction since this step is almost irreversible for
hydrogenation of olefins at low temperatures.
An explanation of the fact that double-bond migration
also occurs in total absence of hydrogen must be given. As
much as 89% of the converted linoleic acid contributed
to the yield of CLA at a conversion of 11% when the
isomerization reaction was carried out in absence of hy-
drogen. In this case the solvent is assumed to generate
hydrogen chemisorbed sites. The reason for the high se-
lectivity toward isomerization in absence of preactivation
is the low concentration of chemisorbed hydrogen. For a
catalyst preactivated under hydrogen, we can assume that
an equilibrium concentration of chemisorbed hydrogen on
the catalyst surface is reached during the activation, hence
the chemisorbed hydrogen generated during the preactiva-
tion is the hydrogen source in the beginning of the reaction.
A negligible noncatalytic direct hydrogenation of linoleic
acid proceeding consecutively via monounsaturated acids
to stearic acids also occurs. The hydrogen source for this di-
rect hydrogenation is the solvent and the rate is influenced
by the nature of the solvent.
4
. DISCUSSION
The double-bond migration reaction over a hydrogen
preactivated catalyst is thought to occur via the Horiuti–
Polanyi mechanism (19) describing hydrogenation and iso-
merization of olefins. A schematic representation of a pos-
sible reaction network for a catalyst preactivated under
ꢁ
ꢁ
hydrogen is demonstrated in Fig. 9. H2, Z, Z , Z H, C2,
C2i , C1, C0, C2H, and C1H denote hydrogen, active va-
cant surface site for chemisorption of organic molecule,
active vacant surface site for chemisorption of hydro-
gen, hydrogen chemisorbed site, linoleic acid, conjugated
linoleic acids, monounsaturated acids, stearic acid, half-
hydrogenated intermediate species derived from diunsat-
urated acids, and half-hydrogenated intermediate species
chemisorption of hydrogen
K1
+
2Z'
H2
2(H)ads
-2Z'
conversion of linoleic acid
K2
K3
k6
K7
K8
k9
+
Z
+Z'H-Z'
+Z'H-Z-Z'
+Z
-Z
+Z'H-Z'
+Z'-Z'H
+Z'H-Z-Z'
C2
(C2)ads
(C2H)ads
C1
(C1)ads
(C1H)ads
C0
-Z
+Z'-Z'H
+
Z'H-Z' +Z'-Z'H K4
C2i)ads
No changes in the distribution between specific CLA iso-
mers were detected by varying reaction temperature, cata-
lyst reduction, catalyst particle size interval, metal loading,
catalyst quantity, initial linoleic acid concentration, or stir-
ring rate. Based on the obtained data, activity and selectiv-
ity can be treated separately. The total converted amount of
linoleic acid was affected by reaction temperature, catalyst
(
+
Z
-Z
K5
C2i
FIG. 9. Schematic representation of a possible reaction network for
a catalyst preactivated under hydrogen.

ISOMERIZATION OF LINOLEIC ACID
365
metal loading, catalyst quantity, initial linoleic acid concen-
Infact, thecatalystmetalloadingaffectstheinitialoverall
tration, catalyst particle size, and nature of the solvent. An reaction rate but not the selectivities. It is thus understand-
increase of reaction temperature, metal loading, and cata- able that the conversion at a reaction time of 6 h is a linear
lyst quantity, and a decrease of initial linoleic acid concen- function of the metal loading, since an increased metal load-
tration and particle size resulted in higher total conversion, ing increases the total amount of hydrogen generated in the
and n-decane showed higher conversion than 1-octanol. catalyst preactivation and the total amount of chemisorbed
The selectivities toward isomerization and hydrogenation half-hydrogenated intermediate species. The reason for the
on the other hand were affected only by catalyst quantity, constant selectivities might be that the ratio between cover-
initial linoleic acid concentration, and nature of the solvent. ageofintermediatesandcoverageofhydrogenisnotgreatly
An increase of the catalyst mass or a decrease of the initial affected by the metal loading, and that the total amount
linoleic acid concentration resulted in a decrease of the iso- of chemisorbed hydrogen generated in the preactivation
merization selectivity due to the hydrogen storage capacity is not dramatically changed. The hydrogen rate increases
of the catalyst, and n-decane showed very high isomeriza- dramatically compared to the isomerization rate when the
tion selectivity compared to 1-octanol. Also, selectivities catalyst mass is increased. The reason for this might be
for specific CLA isomers were affected by the solvent.
that the total amount of chemisorbed hydrogen gener-
Considering the second step in Fig. 9, the nature of the ated in the preactivation increases more when the cata-
solvent might be affecting the adsorption equilibrium and lyst mass is increased compared to the case with increased
thereby the coverage of linoleic acid and the total amount metal loading, since not only the catalyst metal is involved
of adsorbed linoleic acid. A nonpolar solvent might be pre- in the hydrogen storage, but also the support material.
ferred for having a maximum adsorption of linoleic acid
The Horiuti–Polanyi mechanism was suggested earlier
on the catalyst surface due to less strong competitive ad- (9) in the case of isomerization of methyl linoleate over
sorption between linoleic acid and solvent. Poor adsorption carbon-supported Rh and Ru catalysts and the mechanism
of linoleic acid leads to a lower coverage of chemisorbed is supported by our results. The kinetics of the isomerization
◦
half-hydrogenated intermediate species. A lower interme- of methyl linoleate were studied between 200 and 270 C un-
diate coverage in turn favors hydrogenation, since the ratio der nitrogen (9). Reactions taking place were double-bond
between intermediates and chemisorbed hydrogen in this migration, hydrogenation, and polymerization. It was con-
case is lower. It is possible that this fact contributed to the cluded that the hydrogenation was caused essentially by hy-
changes in the selectivities toward isomerization and hydro- drogen derived from the solvent, and that activity and selec-
genation when the isomerization reaction was performed tivities were dependent on the solvent used. Highly protic
in 1-octanol and n-decane. A nonpolar solvent might also solvents like methanol or isopropyl alcohol exhibited high
provide a better solvation of linoleic acid.
activity and high hydrogenation selectivity, whereas aprotic
Another parameter that has a definite influence on the solventslikehexaneorcyclohexaneshowedhighisomeriza-
second step in Fig. 9 is the catalyst metal loading. An in- tion selectivity. It was also concluded that low solvent con-
crease of the metal loading results in an increase of the centration, high catalyst quantity, and high reaction temper-
total amount of chemisorbed linoleic acid. Even the reac- ature favor polymer formation, that the isomer distribution
tion temperature and the catalyst mass affect the amount of was affected by both reaction time and reaction tempera-
chemisorbed linoleic acid. The total amount of active sites ture, and that activity and selectivitiescould bemanipulated
is increased by an increase of the catalyst mass, resulting by varying the solvent compositions. But in Deshpande
in an increase of the amount of chemisorbed linoleic acid. et al. (9) and in similar studies of the isomerization reaction
The reaction temperature, in turn, affects the adsorption of methyl linoleate (10, 11), the hydrogenation reaction of
isotherm. However, the fractional surface coverage is not methyl linoleate was believed to be consecutive proceed-
affected by catalyst metal loading or catalyst mass.
ing via the conjugated isomer. (Conjugation was followed
Considering the third step in Fig. 9, where a hydrogen by hydrogenation to monounsaturated species with no par-
atom derived from a hydrogen chemisorbed site is added allel hydrogenation of methyl linoleate.)
to the adsorbed organic molecule, the rate is determined by
The isomerization reaction of linoleic acid to conjugated
the coverage of hydrogen and the coverage of linoleic acid. linoleic acids can also take place through several other
The total amount of hydrogen is increased by an increase of routes. If the first step involves C–H bond cleavage, an al-
metal loading and catalyst mass, but the fractional surface lylic intermediate is formed on a supported metal atom or
coverage of hydrogen is not affected by these parameters. on an acidic site. Subsequent hydrogenation at a different
The hydrogen addition and hydrogen abstraction in the carbon atom results in double-bond migration. If linoleic
fourth and the sixth steps in Fig. 9 are affected by the cov- acid adsorbs molecularly, it forms a π complex on the sur-
erage of hydrogen and the coverage of half-hydrogenated face, with a C=C bond coordinated to a Lewis acid site. If
intermediate species. Conjugation is favored by high inter- Brønsted acid sites are present, protonation of linoleic acid
mediate coverage and low hydrogen coverage.
can occur, resulting in a carbenium ion intermediate where

3
66
BERNAS ET AL.
the C=C character has been lost. Subsequent loss of a pro-
ton from a different carbon atom results in double-bond
migration (20). Acid-catalyzed isomerization may also oc-
cur on NiO sites.
ACKNOWLEDGMENTS
This work is part of the activities at the A˚ bo Akademi Process Chemistry
Group within the Finnish Centre of Excellence Programme (2000-2005)
by the Academy of Finland. Financial support from the Raisio Group
Research Foundation is gratefully acknowledged. The authors express
their gratitude to Markku Reunanen for his contribution to the GC/MS
analyses, Kari Er a¨ nen for help in construction of the equipment, Sofia
Br a¨ nnbacka for her contribution to the DCP-AES analyses, and Nina
5
. CONCLUSIONS
A new heterogeneously catalytic pathway was developed
for isomerization of linoleic acid to CLAs at mild reaction Granholm for assistance in the silylation procedure.
conditions over Ru/C and Ni/H–MCM-41 catalysts, which
is the first report on the isomerization of this acid but not
its ester, over heterogeneous catalysts. The major finding is
summarized as follows. The isomerization of linoleic acid to
CLA and the hydrogenation of linoleic acid and CLA are
two competing parallel reactions. Hydrogenation of linoleic
acid and CLA proceeds consecutively via monounsaturated
octadecenoic acids to stearic acid. Isomerization can occur
through a Brønsted acid catalyzed pathway, an allylic de-
hydrogenation pathway, or through half-hydrogenated in-
termediates. The concentrations of chemisorbed hydrogen
and key intermediates on the catalyst surface have a definite
influence on catalytic activity and selectivities toward iso-
merization and hydrogenation through a complex relation.
A low concentration of chemisorbed hydrogen on the cata-
lyst surface favors isomerization while a high concentration
favors hydrogenation; however, chemisorbed hydrogen in-
creases the isomerization reaction rate. Therefore the iso-
merization can be enhanced if the reaction is carried out
REFERENCES
1
2
3
4
5
6
. O’Quinn, P. R., Nelssen, J. L., Goodband, R. D., and Tokach, M. D.,
Anim. Health Res. Rev. 1, 35 (2000).
. Whigham, L. D., Cook, M. E., and Atkinson, R. L., Pharmacol. Res.
4
2, 503 (2000).
. Leung, Y. H., and Liu, R. H., J. Agric. Food Chem. 48, 5469
(2000).
. Bauman, D. E., Baumgard, L. H., Corl, B. A., and Griinari, J. M., Proc.
Amer. Soc. Anim. Sci. (1999).
. Iwata, T., Kamegai, T., Sato, Y., Watanabe, K., and Kasai, M., EP
0839897A1 (1998).
. DeJarlais, W., and Gast, L., J. Am. Oil Chem. Soc. 48, 21 (1971).
7. Frankel, E., J. Am. Oil Chem. Soc. 47, 33 (1970).
8. Smith, G., and Roth, J., in “Proceedings of the 3rd International
Congress on Catalysis,” No. 1, p. 379, Amsterdam, 1964.
. Deshpande, V. M., Gadkari, R. G., Mukesh, D., and Narasimhan, C. S.,
J. Am. Oil Chem. Soc. 62, 734 (1985).
0. Mukesh, D., Narasimhan, S., Gadkari, R., and Deshpande, V. M., Ind.
Eng. Chem. Prod. Res. Dev. 24, 318 (1985).
9
1
in a nitrogen atmosphere over a catalyst preactivated un- 11. Mukesh, D., Narasimhan, S., Deshpande, V. M., and Ramnarayan, K.,
Ind. Eng. Chem. Res. 27, 409 (1988).
2. Reddy, K. M., and Song, C., Catal. Lett. 36, 103 (1996).
3. Yu, J. S., Kim, J. Y., and Kevan, L., Microporous Mater. 40, 135
der a pressure of hydrogen compared to the case when
1
1
chemisorbed hydrogen is derived from the solvent. Ru/C
exhibits higher selectivity toward cis-9,trans-11-CLA and
trans-10,cis-12-CLA than Ni/H–MCM-41. When the reac-
tion is carried out in an inert atmosphere over a catalyst
preactivated in hydrogen, the total conversion is affected
by several parameters such as reaction temperature, cata-
lyst metal loading, catalyst quantity, initial linoleic acid con-
centration, catalyst particle size, and nature of the solvent,
but the selectivities toward isomerization and hydrogena-
tion are affected only by catalyst quantity, initial linoleic
acid concentration, and nature of the solvent. Formation of
trans bonds is preferred over formation of cis bonds, but se-
lectivities toward individual CLA isomers can be affected
by using different solvents.
(
2000).
1
4. Moulder, J. F., Stickle, W. F., Sobol, P. E., and Bomben, K. D., “Hand-
book of X-ray Photoelectron Spectroscopy.” Perkin Elmer Corpora-
tion, Physical Electronic Division, Eden Prairie, Minnesota, 1992.
5. Kramer, R., and Andre, M., J. Catal. 58, 287 (1979).
1
1
6. Ek, P., Huld e´ n, S. G., and Ivaska, A., J. Anal. At. Spectrom. 10, 121
(
1995).
1
1
1
2
7. Guczi, L., Sundararajan, R., Kopp a´ ny, Zs., Zsoldos, Z., Schay, Z.,
Mizaukami, F., and Niwa, S., J. Catal. 167, 482 (1997).
8. Hoffer, B. W., van Langeweld, A. D., Janssens, J. P., Bonn e´ , L. C., Lok,
C. M., and Moulijn, J. A., J. Catal. 192, 432 (2000).
9. Pines, H., “The Chemistry of Catalytic Hydrocarbon Conversions.”
Academic Press, New York, 1981.
0. Ramani, N. C., Sullivan, D. L., and Ekerdt, J. G., J. Catal. 173, 105
(1998).