Heterogeneous catalytic production of conjugated linoleic acid

Article abstract of DOI:10.1021/op034127v

Kinetic experiments in selective isomerization of technical grade (～55%) linoleic acid to cis-9,trans-11-conjugated linoleic acid (B) and trans-10,cis-12-conjugated linoleic (E) acid isomers were performed batchwise at 165°C over two series of supported metal catalysts, i.e., hydrogen preactivated and nonpreactivated. Activated carbon- and aluminium oxide-supported Ru, Pd, Os, Ir, and Pt-Rh catalysts with 5 wt % metal loading were screened. Catalyst characterization was done by X-ray powder diffraction (XRD), X-ray photoelectron spectroscopy (XPS), scanning electron microscopy-energy-dispersive X-ray analysis (SEM-EDX), temperature-programmed desorption of hydrogen (H₂-TPD), direct current plasma atomic emission spectrometry (DCP-AES), and nitrogen adsorption techniques. Over such catalysts, the reaction scheme involves six steps: (1) double bond migration of linoleic acid to conjugated linoleic acid, (2) positional and geometric isomerization of conjugated linoleic acid, (3) double bond hydrogenation of linoleic acid to monoenoic acids, (4) double bond hydrogenation of conjugated linoleic acid to monoenoic acids, (5) positional and geometric isomerization of monoenoic acids, and (6) double bond hydrogenation of monoenoic acids to stearic acid. Over Ru/C catalyst, chemisorbed hydrogen on the metal surface dramatically increased the linoleic acid isomerization rate in a diluted system as an astoichiometric component-enhancing double bond migration but decreased the isomerization rate by promoting deactivation in a solvent-free system. Over Ru/Al₂O₃ catalyst, on the other hand, dissociated hydrogen increased the isomerization rate for both the diluted and nondiluted systems. The effect of eliminating the solvent was an increase of the turnover frequency (TOF) at 165°C by a factor of 12 with respect to that shown in a diluted system. At the same conversion carbon-and aluminium oxide-supported Ru catalyst showed higher selectivity toward B and E than carbon-supported Pd, Os, Ir, and Pt-Rh catalysts.

Full text of DOI:10.1021/op034127v

Organic Process Research & Development 2004, 8, 341−352
Heterogeneous Catalytic Production of Conjugated Linoleic Acid
†
†
†
‡
†
,†
Andreas Bernas, Narendra Kumar, P a¨ ivi M a¨ ki-Arvela, Bjarne Holmbom, Tapio Salmi, and Dmitry Yu. Murzin*
Laboratory of Industrial Chemistry and Laboratory of Wood and Paper Chemistry, Process Chemistry Centre,
A° bo Akademi UniVersity, FIN-20500 A° bo/Turku, Finland
Abstract:
on animals and humans for the past decade; the term
functional food is increasingly used as a generic description
for the beneficial effects of ingested foods that go beyond
their traditional nutritive value and CLA is of huge interest
for food and health research due to its anticarcinogenic and
Kinetic experiments in selective isomerization of technical grade
(
∼55%) linoleic acid to cis-9,trans-11-conjugated linoleic acid
(B) and trans-10,cis-12-conjugated linoleic (E) acid isomers were
performed batchwise at 165 °C over two series of supported
metal catalysts, i.e., hydrogen preactivated and nonpreactivated.
Activated carbon- and aluminium oxide-supported Ru, Pd, Os,
Ir, and Pt-Rh catalysts with 5 wt % metal loading were
screened. Catalyst characterization was done by X-ray powder
diffraction (XRD), X-ray photoelectron spectroscopy (XPS),
scanning electron microscopy-energy-dispersive X-ray analysis
1
-9
antioxidative properties .
CLA (naturally present in meat and dairy products, where
it is synthesized from linoleic acid by rumen bacteria) is a
collective term for describing the positional and geometric
conjugated dienoic isomers of linoleic acid (cis-9,cis-12-
octadecadienoic acid). Linoleic acid (the main constituent
of vegetable oils) has double bonds located on carbons 9
and 12, both in the cis configuration, whereas CLA has either
the cis or trans configuration or both located along the carbon
chain. Recent findings suggest that not only is CLA effective
in many different pathways, but that individual isomers of
CLA act differently. Positive health benefits are ascribed to
the consumption of the cis-9,trans-11- and trans-10,cis-12-
CLA isomers. The dietary significance of the other isomers
is unknown.
(SEM-EDX), temperature-programmed desorption of hydro-
gen (H -TPD), direct current plasma atomic emission spec-
2
trometry (DCP-AES), and nitrogen adsorption techniques. Over
such catalysts, the reaction scheme involves six steps: (1) double
bond migration of linoleic acid to conjugated linoleic acid, (2)
positional and geometric isomerization of conjugated linoleic
acid, (3) double bond hydrogenation of linoleic acid to mo-
noenoic acids, (4) double bond hydrogenation of conjugated
linoleic acid to monoenoic acids, (5) positional and geometric
isomerization of monoenoic acids, and (6) double bond hydro-
genation of monoenoic acids to stearic acid. Over Ru/C catalyst,
chemisorbed hydrogen on the metal surface dramatically
increased the linoleic acid isomerization rate in a diluted system
as an astoichiometric component-enhancing double bond mi-
gration but decreased the isomerization rate by promoting
10
It has recently been concluded that CLA can be prepared
by isomerization of linoleic acid over heterogeneous sup-
ported metal catalysts, which due to the cleaner chemistry
(
(
(
1) O’Quinn, P. R.; Nelssen, J. L.; Goodband, R. D.; Tokach, M. D. Conjugated
Linoleic Acid. Anim. Health Res. ReV. 2000, 1, 35-46.
2) Whigham, L. D.; Cook, M. E.; Atkinson, R. L. Conjugated Linoleic Acid:
Implications for Human Health. Pharmacol. Res. 2000, 42, 503-510.
3) Bauman, D. E.; Baumgard, L. H.; Corl, B. A.; Griinari, J. M. Biosynthesis
of Conjugated Linoleic Acid in Ruminants. Proc. Am. Soc. Anim. Sci. 1999,
E29; http://www.asas.org/jas/symposia/proceedings/0937.pdf.
deactivation in a solvent-free system. Over Ru/Al
2 3
O catalyst,
on the other hand, dissociated hydrogen increased the isomer-
ization rate for both the diluted and nondiluted systems. The
effect of eliminating the solvent was an increase of the turnover
frequency (TOF) at 165 °C by a factor of 12 with respect to
that shown in a diluted system. At the same conversion carbon-
and aluminium oxide-supported Ru catalyst showed higher
selectivity toward B and E than carbon-supported Pd, Os, Ir,
and Pt-Rh catalysts.
(
4) Chin, S. F.; Liu, W.; Storkson, M.; Ha, Y. L.; Pariza, M. W. Dietary Sources
of Conjugated Dienoic Isomers of Linoleic Acid, A Newly Recognized Class
of Anticarcinogens. J. Food Compos. Anal. 1992, 5, 185-197.
(
5) Leung, Y. H.; Liu, R. H. trans-10,cis-12-Conjugated Linoleic Acid Isomer
Exhibits Stronger Oxyradical Scavenging Capacity than cis-9,trans-11-
Conjugated Linoleic Acid Isomer. J. Agric. Food Chem. 2000, 48, 5469-
5475.
(
6) Igarashi, M.; Miyazawa, T. The Growth Inhibitory Effect of Conjugated
Linoleic Acid on a Human Hepatoma Cell Line, HepG2, Is Induced by a
Change in Fatty Acid Metabolism, but Not the Facilitation of Lipid
Peroxidation in the Cells. Biochim. Biophys. Acta 2001, 1530, 162-171.
1. Introduction
(7) Basu, S.; Smedman, A.; Vessby, B. Isomer Specific Effects of Conjugated
Linoleic Acid (CLA) on Lipid Peroxidation and Its Regulation by COX 2
Inhibitor and Vitamin E in Humans. Free Radical Biol. Med. 2001, 31,
Suppl. 1, 333.
(8) Mougios, V.; Matsakas, A.; Petridou, A.; Ring, S.; Sagredos, A.; Melisso-
poulou, A.; Tsigilis, N.; Nikolaidis, M. Effects of Supplementation with
Conjugated Linoleic Acid on Human Serum Lipids and Body Fat. J. Nutr.
Biochem. 2001, 12, 585-594.
Conjugated linoleic acid (CLA), which was identified as
late as 1987, is being sold as a panacea that has the capability
of reducing or even eliminating atherosclerosis and cancer,
preventing heart disease, improving immune function, in-
creasing insulin sensitivity, and altering the body composition
in order to treat obesity or build lean body mass. Much
attention has been paid to the physiological effects of CLA
(
9) Palombo, J. D.; Ganguly, A.; Bistrian, B. R.; Menard, M. P. The
Antiproliferative Effects of Biologically Active Isomers of Conjugated
Linoleic Acid on Human Colorectal and Prostatic Cancer Cells. Cancer
Lett. 2002, 177, 163-172.
*
To whom correspondence should be addressed. Fax: +358 2 215 4479.
(10) Bernas, A.; Kumar, N.; M a¨ ki-Arvela, P.; Laine, E.; Holmbom, B.; Salmi,
T.; Murzin, D. Yu. Conjugation of Linoleic Acid over a Hydrogen
Preactivated Heterogeneous Catalyst. Chem. Commun. 2002, 10, 1142-
1143.
Telephone: +358 2 215 4985. E-mail: dmurzin@abo.fi.
†
Laboratory of Industrial Chemistry.
Laboratory of Wood and Paper Chemistry.
‡
1
0.1021/op034127v CCC: $27.50 © 2004 American Chemical Society
Vol. 8, No. 3, 2004 / Organic Process Research & Development
•
341
Published on Web 03/17/2004

Figure 1. Main reaction network in isomerization and hydrogenation of linoleic acid over supported metal catalysts. (A) linoleic
acid; (B) cis-9,trans-11-conjugated linoleic acid; (C) cis-9,cis-11-conjugated linoleic acid; (D) trans-9,trans-11-conjugated linoleic
acid; (E) trans-10,cis-12-conjugated linoleic acid; (F) oleic acid; (G) elaidic acid; (H) trans-vaccenic acid; (I) cis-vaccenic acid; (J)
stearic acid.
is an attractive alternative to the traditional homogeneously11
catalyzed linoleic acid isomerizations. A new heteroge-
neously catalytic pathway has been developed for isomer-
ization of linoleic acid to CLA at mild reaction conditions
in a diluted system under a nitrogen atmosphere with
Moreover, the isomerization properties of several sup-
ported metal catalysts have been evaluated with reagent grade
(99%) linoleic acid. Investigated catalysts were Ru, Ni, Pt,
Pd, Rh, Ir, Os, and bimetallic Pt-Rh supported by activated
2 3 2 2 2 3
carbon, Al O , SiO , SiO Al O , MCM-22, H-MCM-41, and
14
12
hydrogen preadsorbed on the catalyst surface. It was shown
that isomerization and hydrogenation of linoleic acid are two
competing parallel reactions and that these reactions are
influenced by the concentration of chemisorbed hydrogen.
The pathway for the isomerization and hydrogenation of
linoleic acid is demonstrated in Figure 1. Linoleic acid does
not only react to the cis-9,trans-11-, trans-10,cis-12-, cis-
Y and â zeolites. It was concluded that Ru, Ni, and Pt have
good properties for isomerization of linoleic acid. The activity
and the isomerization selectivity were sensitive to the surface
structure and hydrogen adsorption capacity of the metal-
support combination. Metals with high hydrogen storage
capacity such as Pd showed high activity and high selectivity
for the double bond hydrogenation reaction, whereas most
of the other metals, especially Ru, favored double bond
migration. The Ru metal’s activity for isomerization is to
be expected from the fact that it has vacant d-orbitals, which
can interact with π bonds of fatty acids as well as activating
an adjacent C-H bond, which is a necessary step for a
double bond migration. Very few catalysts, including Ru/C
9,cis-11-, and trans-9,trans-11-CLA isomers and monoenoic
acids, but also to several other CLA isomers (the cis,cis,
trans,trans, cis,trans, and trans,cis isomers of the 7,9-, 8,10-,
9,11-, 10,12-, and 11,13-C18 diene acids). Chemical com-
pounds linoleic acid, cis-9,trans-11-conjugated linoleic acid,
cis-9,cis-11-conjugated linoleic acid, trans-9,trans-11-
conjugated linoleic acid, trans-10,cis-12-conjugated linoleic
acid, oleic acid, elaidic acid, trans-vaccenic acid, cis-vaccenic
acid, and stearic acid are hereafter in text referred to as A,
B, C, D, E, F, G, H, I, and J, respectively. Isomerization
experiments under hydrogen resulted in hydrogenated end
products, and experiments under nitrogen resulted in low
catalytic activity; however, the isomerization reaction in a
nitrogen atmosphere could be enhanced by catalyst preac-
tivation under hydrogen with remaining high isomerization
1
4
and Ru/Al O , exhibit feasible conversion and reasonable
2
3
selectivity toward functional foods B and E isomers and,
which also is very important, low selectivity toward several
undesired conjugated dienoic isomers, especially D. In
addition, the deactivation of Ru/Al O and Ru/C was very
2
3
slow compared to the rapid deactivation of zeolites, giving
a reason for selecting aluminium oxide- and carbon-supported
Ru catalysts for further studies. Moreover, the isomerization
selectivity was optimized by the use of nonpolar solvents
1
3
15
selectivity.
such as n-nonane and n-decane.
In the present contribution comparative experiments in
the selective isomerization of technical grade (∼55%) linoleic
(
11) Reaney, M. J. T.; Liu, Y. D.; Westcott, N. D. Commercial Production of
Conjugated Linoleic Acids; Agriculture and Agri-Food Canada: Saskatoon,
SK, Canada, 1999.
(
12) Bernas, A.; Laukkanen, P.; Kumar, N.; M a¨ ki-Arvela, P.; V a¨ yrynen, J.; Laine,
E.; Holmbom, B.; Salmi, T.; Murzin, D. Yu. A New Heterogeneously
Catalytic Pathway for Isomerization of Linoleic Acid over Ru/C and Ni/
H-MCM-41 Catalysts. J. Catal. 2002, 210, 354-366.
(14) Bernas, A.; Kumar, N.; M a¨ ki-Arvela, P.; Kul’kova, N. V.; Holmbom, B.;
Salmi, T.; Murzin, D. Yu. Isomerization of Linoleic Acid over Supported
Metal Catalysts. Appl. Catal., A 2003, 245, 257-275.
(15) Bernas, A.; M a¨ ki-Arvela, P.; Kumar, N.; Holmbom, B.; Salmi, T.; Murzin,
D. Yu. Heterogeneously Catalytic Isomerization of Linoleic Acid over
Supported Ruthenium Catalysts for Production of Anticarcinogenic Food
Constituents. Ind. Eng. Chem. Res. 2003, 42, 718-727.
(
13) Bernas, A.; Murzin, D. Yu. Influence of Hydrogen Preactivation on the
Linoleic Acid Isomerization Properties of Supported Ruthenium Catalysts.
React. Kinet. Catal. Lett. 2003, 78, 3-10.
342
•
Vol. 8, No. 3, 2004 / Organic Process Research & Development

acid to isomers B and E over aluminium oxide- and carbon-
supported metal catalysts are conducted batchwise at mild
reaction conditions to evaluate the influence of factors such
as reactant purity, presence of solvent, preadsorption of
hydrogen on the metal surface, and catalyst deactivation on
catalytic activity and isomerization selectivities.
2
H -preactivation temperature in the isomerization experi-
ments) with a heating rate of 10 °C/min under a 50 mL/min
100% H flow (AGA), kept at 100 °C for 1 h, cooled to 40
2
°C with a rate of 3 °C/min under flowing H , and flushed
2
with 50 mL/min N /Ar for 45 min. The H desorption heating
2
2
rate was 10 °C/min from 40 to 650 °C, and the catalysts
were kept at 650 °C for 30 min. The analyses of the desorbed
gases were performed continuously with a quadrupole mass
spectrometer (QTMD, Carlo Erba Instruments). For quantita-
tive measurements, the mass spectrometer was calibrated for
2
. Experimental Section
.1. Materials. Linoleic acid of 99% purity, technical
grade linoleic acid of 55% purity (impurities: 35% oleic and
0% palmitic acids), Ru/Al , and Ru/C (both with 5 wt
metal loading) were supplied by Fluka (Switzerland). Pd/C
2
1
%
2 3
O
2
the H signal.
2
.6. Isomerization Experiments. The isomerization
catalyst (5 wt %) and the solvent n-decane of 95% purity
were supplied by Sigma-Aldrich (Germany) and Merck
experiments were conducted in a 200-mL stirred batch
reactor, which was provided with a reflux condenser system
and a heating jacket. Silicone oil served as a heat-transfer
fluid, and the reactor was operating at atmospheric pressure.
A measured amount of catalyst was charged into the reactor
and preactivated in situ at 100 °C for 1 h including heating
(
Germany), respectively. The B, C, D, E, and cis-11,trans-
3-CLA isomers of high purity were supplied by Matreya,
Inc. (U.S.A.). Synthesis of catalysts Os, Ir, and Pt-Rh on
1
14
activated carbon was reported in a previous study. The BET
specific surface areas and pore size distributions were
measured with an automatic physisorption-chemisorption
apparatus (Sorptometer 1900, Carlo Erba Instruments), and
the pore size distributions were calculated according to the
Dollimore-Heal method.
2
time under a H flow of 100 mL/min, because catalyst
preactivation under hydrogen dramatically enhances the
1
0,12-15
isomerization rate.
atmosphere was also tested for the sake of comparison. After
the preactivation, a 100 mL/min N flow was fed through
Catalyst preactivation under a N
2
2
2.2. X-ray Powder Diffraction. X-ray powder diffraction
the reactor for 20 min to achieve an inert atmosphere while
heating the system to the reaction temperature 165 °C. The
reactant solution comprising (i) 0.2 g of technical grade
linoleic acid and 70 mL of the solvent n-decane or (ii) 70
mL of technical grade linoleic acid was de-oxygenized by a
N flow of the same flow rate and introduced into the reactor
2
under stirring. The reaction time was initialized to zero as
soon as the liquid phase came in contact with the catalyst.
(XRD) measurements on Ru/C were performed using a
diffractometer (Philips PW 1800) to determine the Ru
crystallite size. The measuring conditions were as follows:
generator voltage 50 kV, generator current 40 mA, Cu tube
anode, automatic divergence slit, irradiated sample length
1
2 mm, receiving slit 0.2 mm, step size 0.025°, counting
time 5.0 s per step.
.3. X-ray Photoelectron Spectroscopy. Analysis by
X-ray photoelectron spectroscopy (XPS) on Ru/Al was
2
2
A 100 mL/min N flow was fed through the reactor during
2
O
3
the isomerizations to have an inert atmosphere, and the
reactor outlet was locked by a fluid to prevent back-diffusion
of oxygen into the mixture. The course of the reaction was
followed by withdrawing 1 mL samples from the mixture
periodically through a 7-µm catalyst filter to await analysis
by gas chromatography. The temperature of the cooling
medium for the reflux condensers was set to -20 °C, and
stirring baffles were used inside the reactor.
The system was typically stirred at 800 rpm to keep the
catalyst uniformly dispersed in the reaction medium and to
eliminate effects of external mass transfer, which was verified
by comparing overall reaction rates from isomerizations of
both reagent and technical grade linoleic acid at varied
stirring rates and catalyst quantities but otherwise using the
same reaction conditions.
utilized to analyze the chemical composition of the “work-
ing” catalyst surface. Before performing XPS, a catalyst
sample was preactivated ex situ at 100 °C for 1 h including
2
heating time under a H flow of 100 mL/min in the test
reactor to obtain the same state of reduction as in the catalytic
experiments. Thereafter, n-decane was fed into the reactor
under stirring, and the wet sample was charged into a special
sample holder to avoid exposure to air. The measurements
were carried out with a Perkin-Elmer PHI 5400 ESCA
system equipped with PE software. A detailed description
of the XPS measurements were reported in a previous
paper.12
2.4. Scanning Electron Microscopy. Chemical composi-
tion as well as metal particle size of Ru/Al was analyzed
2
O
3
by scanning electron microscopy-energy-dispersive X-ray
analysis (SEM-EDX) measurements using a LEO 1530
system equipped with a ThermoNORAN Vantage analyzer
using an acceleration voltage of 15 keV and a take off angle
of 34.4862°.
Deactivation of the catalysts was investigated by perform-
ing consecutive experiments over the same catalyst sample.
After the experiment, the catalyst was separated from the
fatty acid solution by vacuum filtration, washed with acetone
by pouring or stirring in a tank, dried overnight at 100 °C,
and rereduced at 400 °C for 4 h under 100 mL/min flowing
2
.5. Temperature-Programmed Desorption of Hydro-
gen. Hydrogen adsorption capacities of Ru/Al and Ru/C
were investigated by temperature-programmed desorption of
hydrogen (H -TPD) using a volumetric flow apparatus
Autochem 2910, Micrometrics) with nitrogen-argon (N
2 3
O
2
100% H (AGA).
2.7. Analysis. The samples from the reactor were silylated
by using the reagents N,O-bis(trimethylsilyl)trifluoroacet-
amide (BSTFA) and trimethylchlorosilane (TMCS), both
supplied by Acros Organics, and analyzed by a gas chro-
2
(
2
99.5%, Ar 0.5%, AGA) as a carrier gas. The Ru catalysts
were heated in situ to 100 °C (since 100 °C was used as
Vol. 8, No. 3, 2004 / Organic Process Research & Development
•
343

Table 1. Pore size distribution of the aluminium oxide- and
carbon-supported Ru (5 wt %) catalysts
3.1.2. Hydrogen-Adsorption Capacity of Ruthenium. The
-TPD patterns of Al - and C-supported Ru are demon-
strated in Figure 3. Ru/C had 2.3 times higher H -adsorption
capacity than Ru/Al , and the TPD measurements indicated
presence of some spillover H on Ru/Al . For both
catalysts, as much as 90% of the total amount of chemisorbed
remained on the metal surface after a temperature increase
H
2
2 3
O
relative volume (%)
2
2 3
O
pore ranges (nm)
5 wt % Ru/Al
2
O
3
5 wt % Ru/C
2
2 3
O
0
0
1
1
2
5
-0.9
0
0
0
0
2.6
5.9
4.2
20.8
18.1
48.4
.9-1.0
.0-1.5
.5-2.0
.0-5.0
.0-10.0
H
2
to 100 °C, which was used as preactivation temperature in
the isomerization experiments, and both catalysts exhibited
0
3.2
25.0
50.6
a maximum H
2
desorption rate at 125 °C. For Ru/Al
2 3
O ,
1
0.0-100.0
49% of the total amount of chemisorbed H
remained on
2
the catalyst surface after a temperature increase to 165 °C,
which was used as reaction temperature in the isomerization
experiments. The corresponding value for Ru/C was 55%.
matograph (GC, Hewlett-Packard 6890 Series) equipped with
a 25-m HP-5 column (inner diameter: 0.20 mm, film
thickness: 0.11 µm), flame ionization detector (FID) unit,
and an autosampler injector. The silylation operations were
conveniently performed with the entire series of samples
from one isomerization reaction in parallel. A detailed
description of the silylation procedure is given in previous
3
.2. External and Internal Diffusion. To eliminate
effects of external transport of linoleic acid to the outer
surface of the catalyst particles on reaction rates and to keep
the catalyst uniformly dispersed in the reaction medium, the
system was agitated at 800 rpm in all experiments. To
confirm the negligibility of resistance on the isomerization
rate by external transport, the reaction was conducted at 165
1
2,14,15
reports.
Identities of peaks were verified by a gas
chromatograph/mass spectrometer system (GC/MS, Hewlett-
Packard) applying the same GC conditions.
°
C with varied stirring rates and catalyst quantities but
otherwise using the same reaction conditions for both the
systems: (a) n-decane diluted reagent grade linoleic acid
catalyzed by Ru/Al O and (b) solvent-free technical grade
2 3
3
. Results & Discussion
.1. Catalyst Characterization. 3.1.1. Physio-Chemical
Properties of Supported Ruthenium Catalysts. The BET
specific surface areas of the Ru/C and Ru/Al (5 wt %)
3
linoleic acid catalyzed by Ru/C. It was visually confirmed
that the catalyst particles were lifted from the bottom of the
glass reactor at the stirring rate 500 rpm and no changes in
the dark color of the reaction medium were noted with further
increase of the agitation. These observations were consistent
with the obtained reaction rate data (Figure 4). No selectivity
changes were observed when the stirring rate was varied. It
can thus be concluded that external diffusion did not affect
catalytic activity or selectivity.
2
O
3
2
catalysts were determined to be 841 and 156 m /g, respec-
tively. All catalyst powders consisted of particles smaller
than 180 µm. The catalyst particle sizes of Ru/Al O were
2 3
within the range 0-63 µm. The particle sizes of Ru/C were
distributed around 90 µm, with the finest particles less than
2 3
45 µm. The pore size distribution of Ru/Al O and Ru/C is
given in Table 1. The supports are porous materials with
majority of the pore sizes in the range 10-100 nm. Ru/C
also has a small relative pore volume fraction less than 2
nm.
The average Ru crystallite size of the Ru/C catalyst was
determined to be 50 nm, as measured by XRD, indicating
that most of the Ru crystallites are located on the outside of
the micropores of the carbon support. The Ru/Al
had a rather wide metal particle size distribution. In the
scanning electron micrographs of Ru/Al that are presented
in Figure 2, it is clearly visible that the average Ru crystallite
size is close to that of Ru/C, but some larger agglomerates
were found. The electron diffraction indicated a chemical
compositions of 27.7 atom % of Al, 71.07% of O, and 1.23%
of Ru; hence, the mass-related fraction of Ru was, according
to the SEM-EDX measurements, 6.19 wt %. The ratio
between the chemical composition of elements Al and O
analyzed by XPS is identical to that analyzed by SEM-
EDX, while XPS results show only 0.5 atom % of Ru.
Experiments with catalysts of varied particle size dem-
onstrated that with particle diameter decrease, the initial rate
was first inversely proportional to the particle size and
thereafter approached a constant value and a plateau of initial
rate against particle diameter dependence. It was concluded
15
16
that for Ru/C the chemical regime with an effectiveness
factor larger than 0.95, could be attained at the particle
diameter 25 µm, while the aluminium oxide-supported
ruthenium catalyst with particle diameters less than 63 µm
showed negligible internal diffusion.
2 3
O catalyst
2 3
O
3
.3. From Reagent Grade Raw Material to Technical
Grade. 3.3.1. Diluted System. 3.3.1.1. Catalytic ActiVity. For
the sake of comparison, technical grade linoleic acid was
10,12-15
tested as a raw material instead of previously reported
reagent grade linoleic acid; otherwise identical reaction
conditions were used. Polymerization, cracking, or skeletal
isomerizations of fatty acids were not observed. No other
reactions were detected except for the following six steps:
2 3
Moreover, as analyzed by XPS, the Ru/Al O sample
(1) double bond migration of linoleic acid to conjugated
included a large amount of C, probably because of contami-
nation. The metallic state of Ru dominates, but the Ru 3d5/2
spectra of the samples indicated that Ru species is not only
metallic, but a part of Ru may also contain some higher
oxidation states which was indicated by a slight broadening
of the spectrum to higher binding energy side.
linoleic acid, (2) positional and geometric isomerization of
conjugated linoleic acid, (3) double bond hydrogenation of
linoleic acid to monoenoic acids, (4) double bond hydroge-
(16) Santacesaria, E. Kinetics and Transport Phenomena in Heterogeneous Gas-
Solid and Gas-Liquid-Solid Systems. Catal. Today 1997, 34, 411-420.
344
•
Vol. 8, No. 3, 2004 / Organic Process Research & Development

Figure 2. Scanning electron micrographs of Ru/Al
2
O
3
, ×30000.
in Figure 5. The mole fractions in the figures refer to the
concentrations of the specific C18 compounds relative to the
total concentration of C18 compounds, where C18 denotes 18
carbon fatty acids. As demonstrated in Figure 5, the
concentration of reagent grade linoleic acid decreases more
slowly and continuously with time than the concentration
of technical grade linoleic acid, which decreases more rapidly
in the beginning of the reaction and thereafter approaches a
constant value after a reaction time of 30 min. In both cases
the desired B and E CLA isomers were the main reaction
products. In fact, the most thermodynamically stable form
is the D CLA isomer, which was concluded in isomerization
experiments with reagent grade linoleic acid, B, D, and E
2 3
isomers as reactants over hydrogen preactivated Ru/Al O
1
6
and Ru/C samples. Several positional and geometric
conjugated dienoic isomers are formed as intermediates in
the consecutive isomerization of linoleic acid to trans-9,trans-
1
1
1-CLA due to the geometry of linoleic acid, the cis-9,cis-
2 form. When CLA isomers were used as reactants, the
equilibrium sample did not contain any trace of linoleic acid.
When technical grade linoleic acid was used as a raw
material, the consumption rate of linoleic acid decreased to
zero after a certain period of time. However, the higher initial
consumption rate of technical grade linoleic acid in a diluted
system is a promising indication in the prospect of a solvent
elimination.
Figure 3. Hydrogen temperature-programmed desorption
patterns of 5 wt % loaded (I) activated carbon- and (II)
aluminium oxide-supported ruthenium catalysts.
nation of conjugated linoleic acid to monoenoic acids, (5)
positional and geometric isomerization of monoenoic acids,
and (6) double bond hydrogenation of monoenoic acids to
stearic acid. Hence, in the following discussion the term
isomerization refers to double bond migration yielding a
conjugated system. The sum of the overall isomerization
selectivity and the hydrogenation selectivity is equal to unity.
The total conversion of linoleic acid after 3 h, the
isomerization selectivity at 30% conversion, the selectivity
toward B and E CLA isomers (desired products) at 30%
conversion, and the turnover frequency (TOF) after 3 h
3.3.1.2. OVerall Isomerization SelectiVity. When isomer-
ization is carried out over H -preactivated Ru catalysts, the
2
isomerization selectivity increases while the hydrogenation
selectivity decreases with conversion, since the initially rich
layer of chemisorbed hydrogen on the ruthenium surface is
consumed in hydrogenation steps during the course of the
reaction, indicating that the coverage of hydrogen on the
metal surface affects the isomerization and hydrogenation
rates. In the beginning of the reaction when the hydrogen
coverage on the Ru surface is high, the linoleic acid
hydrogenation consecutively proceeding via monoenoic acids
to stearic acid is preferred. When, on the other hand, the
concentration of chemisorbed hydrogen is low (after some
conversion when a majority of the preadsorbed hydrogen
has been consumed), isomerization is preferred over hydro-
genation.
(based on mol of converted linoleic acid after 3 h/(mol of
Ru × 3 h)) kept approximately the same values of 49%,
-
4
-1
7
2%, 25%, and 3.49 × 10 s , correspondingly, and the
initial rate (based on total linoleic acid conversion after 30
min and ruthenium mass) increased with a factor of 2, as
reagent grade linoleic acid was changed to technical grade.
The concentration versus time dependencies are presented
Vol. 8, No. 3, 2004 / Organic Process Research & Development
•
345

Figure 4. Conversion versus time dependence at varied agitation rate and catalyst quantity for (a) n-decane diluted reagent grade
linoleic acid catalyzed by Ru/Al and (b) solvent-free technical grade linoleic acid catalyzed by Ru/C. 4 ) 0/0.8, ] ) 200/0.8, 0
500/0.8, × ) 800/0.8, O ) 1600/0.8, + ) 800/0.4, - ) 800/0.2 [ ) 800/3, 2 ) 1600/3 rpm/g.
2 3
O
)
2
Figure 5. Isomerization of (a) reagent grade and (b) technical grade linoleic acid over H -preactivated Ru/C in a diluted system.
9
) A; [ ) F; 2 ) G, H, and I; b ) J; 0 ) B and E; ] ) D; 4 ) C and other CLA isomers. Conditions: raw material, 200 mg
of linoleic acid; reaction temperature, 165 °C; solvent, 70 mL of n-decane; stirring rate, 800 rpm; catalyst mass, 200 mg; metal
loading, 5 wt %; preactivation temperature, 100 °C; preactivation time, 1 h; reaction pressure, 1 atm of nitrogen.
3
.3.1.3. Temperature Effects. The reaction temperature
diminished by hydrogenation at this temperature level, that
affects activity and selectivity by adjusting the coverage of
dissociated hydrogen on the ruthenium metal surface. An
increase of the reaction temperature results in a decrease of
the catalytic activity and an increase of the isomerization
selectivity. The influence of the reaction temperature on
reaction rate and selectivities in isomerization of reagent
is 40 °C higher than the temperature corresponding to the
maximum hydrogen desorption rate indicated by the H -TPD
2
measurements on Ru/C as demonstrated in Figure 3. There-
fore 165 °C was selected as a reaction temperature for further
studies.
Ru/C is a more active catalyst than Ru/Al
ization of linoleic acid. Experiments at 120 °C reaction
temperature in n-decane over Ru/Al (5 wt %, 200 mg)
2 3
O in isomer-
2 3
grade linoleic acid over Ru/C and Ru/Al O catalysts was
15
reported in our previous study. When the isomerization was
conducted over hydrogen-preactivated Ru/C (5 wt %, 200
mg) using n-decane as a solvent, 3 h conversion and initial
rate (based on total converted amount of linoleic acid after
2 3
O
previously preactivated at 100 °C under hydrogen resulted
in 44% conversion at a reaction time of 6 h, 65% overall
isomerization selectivity, and 28% selectivity toward B and
15
3
4
0 min and total catalyst mass) decreased from 72% and
E. However, similarly as for the Ru/C catalyst, conversion
-
7
-1 -1
-7
-1
-7
.06 × 10 mol s
g
to 49% and 2.18 × 10 mol s
and initial rate decreased from 44% and 3.27 × 10 mol
-1
-1 -1
-7
-1 -1
g , respectively, as the reaction temperature was elevated
from 120 to 165 °C. Moreover, the isomerization selectivity
increased from 66 to 85%, and the selectivity toward the
desired conjugated linoleic acid isomers increased from 24
to 37%. The reason for decreased rate and conversion and
increased isomerization selectivity is the lower amount of
chemisorbed hydrogen on the surface of the ruthenium metal.
It appears that the double bond migration is no longer
s
g
to 31% and 1.29 × 10 mol s g , respectively,
as the reaction temperature was elevated from 120 to 165
°C, and the overall isomerization selectivity increased from
65 to 89%, due to the lower amount of chemisorbed
hydrogen. The selectivity toward the desired products B and
E increased from 28 to 58%. Here, the 3 h TOF was 1.53 ×
-
4
-1
10 s . At similar conditions, the corresponding value for
-
4
-1
Ru/C was 3.27 × 10 s .
346
•
Vol. 8, No. 3, 2004 / Organic Process Research & Development

Additional experiments were conducted (at the conditions
above) over 800 mg of hydrogen-preactivated Ru/Al
3.3.2. Reaction Mechanism. Speaking in terms of molec-
ular mechanisms, the double bond migration reaction of
linoleic acid to CLA over supported metal catalysts takes
place through several routes besides the main pathway over
metal sites. Acid-catalyzed isomerization may also occur over
oxidic metal sites. Moreover, if the first step involves C-H
bond cleavage, an allylic intermediate is formed on a
supported metal atom or on an acidic site. Subsequent
hydrogenation at a different carbon atom results in double
bond migration. If linoleic acid adsorbs molecularly, it forms
a π complex on the surface, with a CdC bond coordinated
to a Lewis acid site. If Brønsted acid sites are present,
protonation of linoleic acid can occur, resulting in a
carbenium ion intermediate where the CdC character has
been lost. Subsequent loss of a proton from a different carbon
2
O
3
samples in n-decane at varied temperature levels in order to
evaluate the required reaction temperature for keeping a
feasible isomerization selectivity as rate and selectivity are
dependent on the coverage on the Ru surface of chemisorbed
hydrogen, which in turn is strongly influenced by the
temperature. Initial rate and TOF after 3 h were first slightly
increasing with temperature. As the reaction temperature was
elevated from 135 °C via 145 °C to 155 °C, the TOF
-
4
-1
-4 -1
increased from 0.38 × 10
s
via 0.43 × 10
s
to 0.83
-
4
-1
×
10 s , correspondingly, and stayed at this level after a
further increase of the temperature to 165 °C. At 135 °C,
the overall isomerization selectivity at 30% conversion was
lower, i.e. 80%, than at temperatures over 145 °C where the
isomerization selectivity remained at 87% in all experiments,
which indicates that 145 °C is the lower temperature limit
for an optimization of rate and selectivity by manipulating
the hydrogen coverage of the metal surface. The overall
isomerization selectivity and the selectivity toward desired
products increased with conversion.
1
7
atom results in double bond migration.
1
0,12-15
The observed
enhancement of TOF and initial rate
by H preactivation of the ruthenium catalysts can be
2
1
8
explained by the Horiuti-Polanyi mechanism describing
hydrogenation and isomerization of olefins. After the initial
H preactivation, linoleic acid is chemisorbed on the Ru
2
surface. Thereafter, a hydrogen atom derived from a hydrogen-
chemisorbed site is added to the chemisorbed acid to give a
chemisorbed half-hydrogenated intermediate. If the hydrogen
coverage on the catalyst surface is rather low, hydrogen
abstraction by the metal takes place from an adjacent carbon
atom, which could lead to a double bond migration, depend-
ing on which carbon atom the hydrogen is abstracted from.
If, on the other hand, the coverage of hydrogen is high, a
second hydrogen atom is added to the chemisorbed half-
hydrogenated intermediate to result in a double bond
hydrogenation. Free rotation of the half-hydrogenated inter-
mediate, hydrogen abstraction, and desorption of the olefin
results in cis/trans-isomerizations. The role of the Horiuti-
Polanyi mechanism in isomerization of butenes was inves-
tigated already in 1974 by studying the deuterium distri-
bution of exchanged and isomerized olefins obtained from
cis-butene, trans-butene, and but-1-ene on iron films in the
presence of perdeuteriopropene. As it is in line with our
results, the Horiuti-Polanyi mechanism is valid independent
of the length of the carbon chain of the olefin molecule.
3
.3.1.4. Catalyst DeactiVation. As demonstrated in Figure
a, the conversion versus time and catalyst mass dependence
in reagent grade linoleic acid isomerization falls on the same
curve for three different quantities of Ru/Al , which
4
2 3
O
indicates that not only is the reaction conducted in the
absence of external diffusion but also that the extent of the
catalyst being subject to a reduction in activity with time is
either the same independent catalyst mass or negligible. At
2 3
the same time, the selective Ru/Al O catalyst showed lower
activity in a second consecutive experiment over the same
catalyst sample that in the regeneration step had been
separated from the fatty acid solution by vacuum filtration,
washed with acetone in an agitated tank, dried overnight,
2
and rereduced under flowing H using the conditions
1
9
described in the Experimental Section. In the second run,
the isomerization selectivity at 30% conversion remained at
approximately 90%, and with respect to conversion, initial
rate, and TOF, the activity decreased by a factor of more
than 2.3. The BET specific surface area kept a constant value
of 156 m /g as the Ru/Al
regenerated twice, and the pore volume distribution presented
in Table 1 was not affected; hence, the reason for the
observed deactivation could lie within fouling on the active
ruthenium sites or poisoning rather than irreversible plugging
of pores by the heavy fatty acid molecules. In general, acidic
supports or catalysts are especially prone to coking, whereas
very little coke forms on activated carbon supports. While
consecutively repeating isomerizations and regenerations on
the same Ru/C catalyst sample no significant loss in activity
2
O
2 3
catalyst was being used and
3
.3.3. Nondiluted System. 3.3.3.1. H
Catalysts. Table 2 summarizes activity and selectivity data
on nonpreactivated and H -preactivated aluminium oxide-
2
-PreactiVated Metal
2
and carbon-supported 5 wt % ruthenium, palladium, osmium,
iridium, and bimetallic platinum-rhodium catalysts in
isomerization of nondiluted technical grade linoleic acid. The
effect of eliminating the solvent was to increase the TOF at
1
5
3
h by a factor of 12 compared to that shown in an n-decane
diluted system. This increase of the TOF is not only caused
by the observed solvent effect but also by the use a drastically
higher reactant-to-catalyst mass ratio, which in turn contrib-
utes to an increased TOF.
was observed, and the BET area decreased only slightly from
40 to 749 m /g after five uses. Hence, it appears that the
2
8
observed deactivation is not due to blocking of the pores of
the activated carbon and aluminium oxide support materials
by fatty acid compounds. Identical pore volume distributions
for carbon- as well as aluminium oxide-supported ruthenium
were obtained from nitrogen adsorption measurements on
fresh and used catalysts.
(
17) Ramani, N. C.; Sullivan, D. L.; Ekerdt, J. G. Isomerization of 1-Butene
over Silica-Supported Mo(VI), W(VI), and Cr(VI). J. Catal. 1998, 173,
105-114.
(18) Pines, H. The Chemistry of Catalytic Hydrocarbon ConVersions; Academic
Press: New York, 1981.
(19) Touroude, R.; Gault, F. G. The Mechanism of Isomerization and Exchange
of Olefins Over Metal Catalysts. J. Catal. 1974, 32, 294-303.
Vol. 8, No. 3, 2004 / Organic Process Research & Development
•
347

Table 2. Activity and selectivity data on ruthenium, palladium, osmium, iridium, and platinum-rhodium catalysts supported
by aluminium oxide and activated carbon in technical grade linoleic acid isomerizationa
initial overall
BET area BET area
before use after use
e
f
catalyst
linoleic acid
reaction rate
TOF
no.
catalyst
Ru/C
preactivation mass/catalyst mass conv.b
S
I
c
S
D
d
(10 mol s g-1
-5
-1
)
(10 s-1
-4
)
(m /g)
2
(m /g)
2
1
2
3
yes
yes
yes
14.1
16.8
18.1
0.72
0.50
0.23
0.98 0.63
0.97 0.67
0.94 0.58
2.70
2.21
0.89
8.46
6.99
3.46
841
841
604g
604g
Ru/C
-
g
Ru/C,
700
2
nd rung
g
4
5
2
O
O
3
yes
yes
15.0
18.3
0.81
0.08
0.95 0.59
2.32
0.48
10.12
1.22
156
167
147
,
-
-
167g
g
2
3
g
6
7
8
no
no
no
15.0
17.1
17.8
0.95
0.85
0.68
0.98 0.56
0.96 0.56
0.76 0.45
3.61
2.99
3.00
11.87
12.12
10.11
841
841
767g
767g
-
808
g
9
no
15.8
0.81
0.93 0.54
0.96 0.61
2.96
10.64
643h
577h
10
11
12
13
14
15
no
no
no
no
no
no
13.7
15.2
13.5
13.4
15.1
15.2
0.89
0.03
0.61
0.24
0.65
0.51
4.02
0.10
1.69
0.49
2.39
1.60
10.15
0.38
7.22
5.07
15.55
10.84
841
156
804
610
784
690
643h
2
O
3
-
-
-
h
0.68 0.30
0.68 0.34
0.46 0.21
0.66 0.31
550
-
-
-
j
a
Conditions: raw material, 70 mL of technical grade linoleic acid; reaction temperature, 165 °C; stirring rate, 800 rpm; catalyst metal loading, 5 wt %; catalyst
particle diameter interval, 0-180 µm; catalyst preactivation temperature, 100 °C; catalyst preactivation time, 1 h; reaction pressure, 1 atm of nitrogen; reaction time,
b
c
2
I
4 h. Total conversion of linoleic acid (A) after 24 h. Selectivity toward isomerization at 30% conversion. S ) mole fraction of all CLA isomers at 30% conversion
d
/
0.3.e Selectivity toward desired CLA isomers at 30% conversion. S
D
) mole fraction of cis-9,trans-11- and trans-10,cis-12-CLA isomers (B and E) at 30% conversion/
f
0
.3. Based on total linoleic acid conversion after 6 h and metal mass. Turnover frequency. TOF ) mol of converted linoleic acid after 24 h/(mol of metal × 24 h).
g
Catalyst sample had been filtered, washed by pouring 200 mL of acetone through the filter cake, dried overnight at 100 °C, reduced under a 100 mL/min hydrogen
h
flow at 400 °C for 4 h, and passivated after reaching room temperature by exposure to air. Catalyst sample had been filtered, washed by pouring 200 mL of acetone
through the filter cake, additionally washed in 200 mL of acetone at its boiling point (57 °C) in an 800 rpm stirred tank for 2 h, filtered, and washed again by pouring
2
00 mL of acetone through the filter cake, dried overnight at 100 °C, reduced under a 100 mL/min hydrogen flow at 400 °C for 4 h, and passivated after reaching room
i
j
temperature by exposure to air. Stirring rate: 1600 rpm. Metal loading: 5 wt % Pt + 0.05 wt % Rh.
2
Figure 6. Study of deactivation of (a) H -preactivated and (b) nonpreactivated Ru/C catalyst in isomerization of technical grade
linoleic acid in a solvent-free system. 9 ) 3 g of Ru/C, first run; [ ) 2 g of Ru/C, first run; 2 ) 2 g of Ru/C, second run; b ) 2
g of Ru/C, second run (catalyst was washed under agitation). Conditions: raw material, 70 mL of technical grade linoleic acid;
reaction temperature, 165 °C; stirring rate, 800 rpm; metal loading, 5 wt %; preactivation temperature, 100 °C; preactivation
time, 1 h; reaction pressure, 1 atm of nitrogen.
2
In technical grade linoleic acid isomerization in a non-
diluted system, an increase of the reactant-to-catalyst mass
ratio had a significant effect on the course of the reaction
decreased from 841 to 604 m /g as the Ru/C catalyst was
being used and regenerated by filtration, washing, and high-
temperature reduction under H
affected by the reactant-to-catalyst mass ratio. In the isomer-
ization experiment conducted over fresh H -preactivated
2
. Selectivities were not
(
Table 2, entries 1 and 2) compared to the case of reagent
1
2-15
grade linoleic acid isomerization in a diluted system.
2
Figure 6 a demonstrates that observed activity depends on
catalyst mass, which is an indication of differences in
available surface coverage as well as of some catalyst
deactivation. In addition, the BET specific surface area
Ru/C catalyst, no significant leaching of the Ru metal by
the fatty acids was detected, which indicates that the
isomerization and hydrogenation reactions do not occur in
the homogeneous phase. The Ru content of the fatty acid
348
•
Vol. 8, No. 3, 2004 / Organic Process Research & Development

Figure 7. Isomerization of solvent-free technical grade linoleic acid over (a) H
A; [ ) F; 2 ) G, H, and I; b ) J; 0 ) B and E; ] ) D; 4 ) C and other CLA isomers. Conditions same as Figure 6.
2
-preactivated and (b) nonpreactivated Ru/C. 9 )
solution after a reaction time of 24 h was 8 µg/mL as
determined by direct current plasma atomic emission spec-
trometry (DCP-AES).20
Consecutive experiments with regenerated Ru/C also
confirmed nonnegligible deactivation (Figure 6a and Table
Interestingly, conversion, initial rate, and TOF increased
by a factor of more than 30 when nonpreactivated Ru/Al O
3
2
under identical conditions was changed to nonpreactivated
Ru/C (Table 2, entries 11 and 6). The nonpreactivated Ru/C
catalyst did also show higher conversion, initial rate, and
TOF than the H2-preactivated Ru/C catalyst (Table 2, entries
1 and 6), at the same time keeping high selectivities, which
leaves nonpreactivated Ru/C as the most active and selective
catalyst in technical grade linoleic acid isomerization. In
addition the surface area remained high for used and
regenerated catalyst at 400 °C. The influence of catalyst
2
, entries 2 and 3); moreover a lowered value of the BET
2
area after use was obtained (700 m /g). The isomerization
selectivity and the selectivity toward the B and E remained
at a very high level in the second run over the regenerated
catalyst.
2 2 3
The H -preactivated Ru/Al O catalyst showed catalytic
activity and isomerization selectivity as high as the H
2
-
preactivation under H on the course of the isomerization
2
preactivated Ru/C catalyst, and the BET area of the support
material kept an approximately constant value of 160 m /g
reaction over Ru/C is illustrated in Figure 7. As seen in
2
Figure 7a, the consumption rate of A over H -preactivated
2
in the isomerization (Table 2, entry 4). Independent of the
support material, the hydrogen-preactivated ruthenium cata-
lysts showed approximately the same selectivity toward
double bond migration and the desired conjugated linoleic
Ru/C is close to zero after a reaction time of 12 h. Fatty
acid composition at this point, which is not directly related
to chemical equilibrium between fatty acids, is through a
complex connection influenced by the total amount of
chemisorbed hydrogen generated in the catalyst preactivation
step, reaction conditions, availability of active vacant sites
on the catalyst surface, competitive adsorption between
linoleic acid, the solvent, reaction products, and most
predominantly by catalyst deactivation. In Figure 7b the
dominating consecutive steps, isomerization of linoleic acid
via B and E to D, are clearly visible. Initially, the concentra-
tion of linoleic acid decreased rapidly according to zero-
order kinetics and stayed at a level of 7 mol %. Although
oleic acid is the main hydrogenation product of linoleic acid,
the concentration of F decreased continuously with time, not
only due to hydrogenation of F to J but also due to the
geometric isomerization reactions of F to G, H, and I, which
in turn also contributed to the modest increase of the
concentration of stearic acid. The concentrations of B and
E initially increased (as in a typical consecutive reaction)
and started to decrease after reaching a maximum value,
whereas concentrations of not only D but also C and other
unknown CLA isomers started to build up later than those
of the desired isomers. The fatty acid composition at
maximum yields of B and E conjugated linoleic acid isomers
is most conveniently demonstrated by the corresponding gas
chromatogram (Figure 8).
2 3
acid isomers, as well as TOF. However, the Ru/Al O catalyst
was not longer active in a second run (Table 2, entry 5), in
which the BET area changed only slightly from 167 to 147
2
m /g. Here the initial rate and TOF decreased by a factor of
10.
The reason for the observed deactivation over Ru/Al
2
O
3
could lie within fouling on the active ruthenium sites or
poisoning rather than irreversible plugging of pores by the
heavy fatty acid molecules since the decrease of BET was
minor. Substantial decrease of surface area for Ru/C could
be an indication for coking.
3.3.3.2. NonpreactiVated Metal Catalysts. As expected,
24 h conversion, initial rate (based on total linoleic acid
conversion after 6 h and metal mass), and TOF (mol of
converted linoleic acid after 24 h/(mol of metal × 24 h)) of
2 3
technical grade linoleic acid over nonpreactivated Ru/Al O
were almost negligible (Table 2, entry 11), since no chemi-
sorbed hydrogen on the catalyst surface is available as an
astoichiometric compound for promoting double bond migra-
tion.
(
20) Ek, P.; Huld e´ n, S. G.; Ivaska, A. Sequential Injection Analysis System for
the Determination of Hydride Forming Elements by Direct Current Plasma
Atomic Emission Spectrometry. J. Anal. At. Spectrom. 1995, 10, 121-126.
Vol. 8, No. 3, 2004 / Organic Process Research & Development
•
349

and Ru/C catalysts followed by isomerization under an inert
atmosphere enhanced the isomerization rate compared to that
shown in the case of absence of preactivation. H -preacti-
2
vated ruthenium metal was more resistant to deactivation
when it was supported by carbon than when it was supported
by Al
reactant-to-catalyst mass ratio, H
similarly enhanced the isomerization rate, contrary to diluted
systems, H -preactivation of Ru/C suppressed the isomer-
2
O
3
. Although, in a nondiluted system with high
2
preactivation of Ru/Al
2 3
O
2
ization rate by increasing the catalyst deactivation. The TOF
in a nondiluted system with high reactant-to-catalyst mass
ratio increased by a factor of more than 30 when nonpre-
activated Ru/Al O under identical conditions was changed
2 3
to nonpreactivated Ru/C. At the present time it can be only
speculated that oleic and palmitic acids present in technical
grade linoleic acid behave very specifically over H
preactivated Ru/C, promoting deactivation.
2
-
Figure 8. Separation of fatty acids by gas chromatography. I
)
A; II ) F; III ) G, H, and I; IV ) J; V ) B; VI ) unknown
Nonpreactivated palladium, osmium, iridium, and plati-
num-rhodium catalysts (Table 2, entries 12-15) supported
by activated carbon originating from birch wood showed
feasible technical grade linoleic acid conversions and turn-
over frequencies, especially Pd/C, Ir/C, and bimetallic Pt-
Rh/C. However, Pd, Os, Ir, and Pt-Rh supported by carbon
could not compete with Ru/C with respect to the isomer-
ization selectivity or the selectivity toward B and E,
predominantly due to higher double bond hydrogenation. The
fact that the carbon-supported Os, Ir, and Pt-Rh catalysts
were synthesized from chlorine-containing precursors might
also play a role in the lowered selectivity toward the desired
isomers over these catalysts. In addition, the Pd/C catalyst
exhibited a higher loss of surface area than the Ru/C catalyst
under similar conditions (Table 2, entries 6 and 12). The
unique linoleic acid isomerization property of the ruthenium
metal is demonstrated in Figure 9. Independent of the support
material, the ruthenium catalysts followed the very same
selectivity to desired isomers versus conversion, which was
higher than the corresponding selectivity pattern of all other
investigated metal catalysts. In contrast to isomerizations in
CLA isomer; VII ) E; VIII and IX ) unknown CLA isomers;
X ) C; XI ) unknown CLA isomer; XII ) D.
In isomerization of technical grade linoleic acid over
nonpreactivated Ru/C, a decrease of the catalyst mass from
3
to 2 g (Table 2, entries 6 and 7) indicated some
deactivation, similar to the case in isomerization over H2-
preactivated Ru/C catalyst. As seen in Figure 6b, the
conversion profile decreases slightly when the Ru/C quantity
was decreased (curves 1 and 2). As before, selectivities were
not affected by the reactant-to-catalyst mass ratio.
2
In contrast to the case of H -preactivated Ru/C, identical
values of initial rate (6 h) and TOF (24 h) were obtained
when the nonpreactivated Ru/C catalyst was regenerated and
tested in a second consecutive isomerization experiment over
the same catalyst sample. The BET surface area did not
change as the catalyst was being used (Table 2, entries 7
and 8), although the conversion after 24 h decreased slightly.
In Figure 6b the conversion profiles for the first (curve 2)
and the second (curve 3) runs were initially matching but
showed a discrepancy at higher conversions.
Two different washing procedures were applied in the
catalyst regeneration step to improve the catalyst durability.
In the experiment presented in entry 8 (Table 2), the catalyst
sample was filtered in the regeneration step, then washed
by pouring acetone over the filtrate, dried overnight, and
2 3
a diluted system, where Ru/Al O showed constant selectivi-
1
5
ties, all catalysts showed a decreasing selectivity to the
desired CLA isomers versus conversion dependence in
nondiluted technical grade linoleic acid isomerizations;
independent of catalyst pretreatment (catalyst preheating
reduced under flowing H
2
at 400 °C. In the second regenera-
2 2
under H or N ), the selectivity toward overall conjugated
tion technique, the catalyst was filtered, washed by pouring
acetone through the filter cake, stirred in an acetone-filled
tank, washed again by acetone pouring, dried overnight, and
rereduced. Thus, the main difference in the second procedure
was the stirring in acetone. The conditions for both proce-
dures are given in the footnote of Table 2. When a second
run over the nonpreactivated Ru/C catalyst was made, it
showed catalytic activity and selectivities equal to those of
the corresponding isomerization experiment over the fresh
catalyst (Table 2, entries 7 and 9). The conversion profiles
for both experiments follow the same pattern, as shown in
Figure 6b (curves 2 and 4).
linoleic acid and the selectivity toward B and E decreases
with conversion, while the selectivity toward double bond
hydrogenation of linoleic acid and the selectivity for D
increases with conversion.
The mechanism of selective isomerization of linoleic acid
into B and E over metal catalysts supported by activated
carbon and aluminium oxide is not fully understood. Ap-
parently the double bond isomerization predominantly occurs
both via the Horiuti-Polanyi mechanism, where hydrogen
takes part as an astoichiometric component by hydrogen
addition followed by hydrogen abstraction, and via an allylic
mechanism, where hydrogen abstraction is followed by
hydrogen addition.
In summary, in a solvent-diluted system15 with low
2 2 3
reactant-to-catalyst mass ratio, H preactivation of Ru/Al O
350
•
Vol. 8, No. 3, 2004 / Organic Process Research & Development

Figure 9. Selectivity toward B and E in isomerization of technical grade linoleic acid over (9) H
preactivated Ru/Al , (2) nonpreactivated Ru/C, (b) nonpreactivated Ru/C (second run over regenerated catalyst), and
nonpreactivated catalysts (0) Pd/C, (]) Ir/C, and (4) Pt-Rh/C. Conditions same as Figure 6.
2 2
-preactivated Ru/C, ([) H -
2
O
3
Formation of conjugated linoleic acid isomers is thought
to occur via several pathways. Most probably, the dienoic
acids are chemisorbed by the diene adsorption mode followed
by positional or cis/trans isomerization of one double bond
of the carbon chain (whereas the other double bond is locked
in its original position) and desorption of the products. A
second possibility, that the acids are chemisorbed via the
diene adsorption mode followed by positional or cis/trans
isomerization of both double bonds and desorption cannot
be ruled out. It is reasonable to assume that isomerization
of the cis-12-double bond to the trans-11-bond (keeping the
cis-9-bond locked) proceeds to B from cis-9,cis-12-linoleic
acid in one step. Similarly, the reaction of cis-9,cis-12-
linoleic acid to E by isomerization of the cis-9-bond to the
trans-10-bond, while keeping the cis-12-bond locked, occurs
in one step. When the concentration of A decreases and the
concentration of B form increases, the latter one reacts to
the most thermodynamically stable D form in a single step
by geometric isomerization of the cis-9-bond, whereas the
trans-11-bond is locked. It is clearly seen in Figure 7 that
isomerization of A to the D isomer occurs consecutively in
two steps via B, and in addition, no trace of the trans-9,cis-
reactions, in which the olefin itself acts as a hydrogen donor,
might take place on acidic catalysts in the case of hydro-
2
2
carbons. In the literature, such a step, where chemisorbed
methyl linoleate reacts to two hydrogen-chemisorbed sites
plus coke, has been proposed to explain the isomerization
kinetics of methyl linoleate on aluminium oxide-supported
2
3
bimetallic ruthenium-nickel catalyst. Hydride transfer
might be one of the explanations for linoleic acid to undergo
hydrogenation in isomerizations in the absence of solvent
and catalyst preactivation under hydrogen.
4. Conclusions
A new and clean process for catalytic isomerization of
linoleic acid to biologically active cis-9,trans-11-conjugated
linoleic acid and trans-10,cis-12-conjugated linoleic acid over
solid catalysts has been developed. Experiments were
conducted batchwise over carbon- and aluminium oxide-
supported Ru, Pd, Os, Ir, and Pt-Rh catalysts using technical
grade linoleic acid. Catalyst characterization was done by
X-ray powder diffraction, X-ray photoelectron spectroscopy,
scanning electron microscopy-energy-dispersive X-ray analy-
sis, temperature-programmed desorption of hydrogen, and
11-CLA isomer was detected. The contribution of the second
nitrogen adsorption techniques. In the present study as well
mechanism, in which the olefin undergoes adsorption in the
diene adsorption mode, positional isomerization of both the
double bonds, and desorption, was unmasked as the E isomer
mainly reacted to the B, C, and D CLA isomers.1
10,12-15
as in some previous publications
the influence of
several process parameters such as choice of metal and
support material, purity of the raw material, presence of
solvent, preadsorption of hydrogen on the metal surface and
hydrogen adsorption capacity of the catalyst, linoleic acid
and catalyst quantity (i.e. reactant-to-catalyst mass ratio),
agitation rate, size of support pores and catalyst particles
5
A mechanism involving hydride transfer could also be
operative. Hydride transfer from the solvent has been
discussed in the literature21 in the case of isomerization of
methyl linoleate (the methyl ester of linoleic acid) on carbon-
supported ruthenium catalyst. Moreover, hydride-transfer
(i.e., external and internal diffusion), and catalyst deactivation
(22) Poutsma, M. L. In Zeolite Chemistry and Catalysis; Rabo, J. A., Ed.; ACS
(
21) Mukesh, D.; Narasimhan, S.; Gadkari, R.; Deshpande, V. M. Kinetics and
Mathematical Modeling of Isomerization of Methyl Linoleate on Ruthenium
Catalyst. 1. Conjugation and Hydrogenation. Ind. Eng. Chem. Prod. Res.
DeV. 1985, 24, 318-323.
Monograph 171; American Chemical Society: Washington DC, 1976.
(23) Mukesh, D.; Narasimhan, C. S.; Deshpande, V. M.; Ramnarayan, K.
Isomerization of Methyl Linoleate on Supported Ruthenium-Nickel Catalyst.
Ind. Eng. Chem. Res. 1988, 27, 409-414.
Vol. 8, No. 3, 2004 / Organic Process Research & Development
•
351

on catalytic activity and selectivity was investigated. Migra-
tion of the double bond in linoleic acid to yield conjugated
linoleic acid occurs simultaneously with positional and
geometric isomerization of conjugated linoleic acid, hydro-
genation of double bonds in dienoic acids to monoenoic
acids, and further to stearic acid as well as positional and
geometric isomerization of monoenoic acids. On ruthenium
catalysts, the surface coverage of dissociated H and key
2
intermediates have a definite control over catalytic activity
and selectivities toward isomerization and hydrogenation
(d) The isomerization rate is higher in a nondiluted system
compared to that shown in a solvent-diluted system. This
increase of the rate might be caused not only by solvent
effects but also by a higher reactant-to-catalyst mass ratio.
(e) Under identical conditions, carbon- and aluminium
oxide-supported Ru catalyst showed higher overall isomer-
ization selectivity (at the same conversion) as well as higher
selectivity toward the desired cis-9,trans-11- and trans-10,cis-
1
2-conjugated linoleic acid isomers (B and E) than carbon-
supported Pd, Os, Ir, and Pt-Rh catalysts.
f) When isomerization of technical grade linoleic acid
products through a complex relationship. A modest H
coverage on the metal favors the double bond migration
steps, while a high H coverage favors the side reaction i.e.
2
(
is conducted over ruthenium catalysts in a nondiluted system,
selectivities are not affected by catalyst deactivation, influ-
ence of mass transport, or choice of carbon or aluminium
oxide supports.
2
double bond hydrogenation. However, the isomerization rate
can be increased by preactivation of the catalyst under a
hydrogen atmosphere. By applying a technique of catalyst
preactivation, one is balancing on the edge between increased
activity and decreased selectivity so that a compromise has
to be made to guarantee optimal reaction conditions. The
major findings in this contribution can be summarized as
follows.
(g) Independent of catalyst pretreatment (catalyst preheat-
ing under H or N ), the selectivity toward overall conjugated
2
2
linoleic acid and the selectivity toward cis-9,trans-11- and
trans-10,cis-12-conjugated linoleic acid (B and E) decrease
with conversion, while the selectivity toward double bond
hydrogenation of linoleic acid and the selectivity for trans-
9,trans-11-conjugated linoleic acid (D) increase with conver-
sion, when nondiluted technical grade linoleic acid is
isomerized over any of the investigated catalysts.
(a) At reaction temperatures higher than 145 °C, chemi-
sorbed hydrogen on the ruthenium surface increases the
linoleic acid isomerization rate in a solvent-diluted system
(with low reactant-to-catalyst mass ratio). Hydrogen acts as
an astoichiometric component enhancing metal-catalyzed
double bond migration without any major loss of isomer-
ization selectivity due to metal-catalyzed double bond
hydrogenation. This effect was observed over catalysts with
both carbon and aluminium oxide supports.
Acknowledgment
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. We express our
gratitude to Markku Reunanen for his contribution to GC/
MS analyses, Kari Er a¨ nen for help in construction of
equipment, Virpi V a¨ a¨ n a¨ nen for her assistance in DCP-AES
analyses, Clifford Ekholm for SEM-EDX analyses, and
Ensio Laine for XRD measurements as well as Pekka
Laukkanen and Juhani V a¨ yrynen for performing XPS
analyses.
(b) In a nondiluted system (with high reactant-to-catalyst
mass ratio), chemisorbed hydrogen similarly increases the
isomerization rate without loss of isomerization selectivity
over aluminium oxide-supported ruthenium catalysts, but
decreases the isomerization rate dramatically by promoting
deactivation over carbon-supported ruthenium.
(
c) In a solvent-diluted system, H
catalyst is very resistant to deactiVation while H
vated Ru/Al undergoes some deactivation. In a nondiluted
system, on the other hand, both Ru/C and Ru/Al catalysts
deactiVate considerably if these catalysts are preactivated
under H , while nonpreactivated Ru/C catalysts interestingly
2
-preactivated Ru/C
2
-preacti-
2 3
O
2 3
O
2
Received for review September 9, 2003.
OP034127V
demonstrate a higher activity with negligible catalyst deac-
tiVation.
352
•
Vol. 8, No. 3, 2004 / Organic Process Research & Development