Kinetics and mechanism of the production of higher olefins from stearic acid in the presence of an alumina-supported nickel sulfide catalyst

Article abstract of DOI:10.1134/S0023158417020069

A nickel sulfide catalyst which efficient in the decarbonylation of fatty acids to olefins and dienes has been obtained for the first time by treating alumina-supported nickel sulfate with hydrogen, and its properties have been studied. In its presence, the olefin selectivity of the reaction can exceed 90%. The kinetics of stearic acid deoxygenation to heptadecenes has been investigated, a kinetic model has been constructed, and a mechanism has been proposed for the reaction over this catalyst. Olefin oligomerization is the dominant side reaction. Kinetic evidence for the catalytic inhibition of oligomerization by nickel hydrides formed on the catalyst has been obtained. The compositions of active site–reactant adsorption complexes have been discussed.

Full text of DOI:10.1134/S0023158417020069

ISSN 0023-1584, Kinetics and Catalysis, 2017, Vol. 58, No. 2, pp. 147–155. © Pleiades Publishing, Ltd., 2017.
Original Russian Text © E.A. Katsman, V.Ya. Danyushevsky, P.S. Kuznetsov, R.S. Shamsiev, A.S. Berenblyum, 2017, published in Kinetika i Kataliz, 2017, Vol. 58, No. 2, pp. 159–169.
Kinetics and Mechanism of the Production
of Higher Olefins from Stearic Acid in the Presence
of an Alumina-Supported Nickel Sulfide Catalyst
E. A. Katsman, V. Ya. Danyushevsky, P. S. Kuznetsov, R. S. Shamsiev, and A. S. Berenblyum
Syrkin Department of Physical Chemistry, Moscow Technological University, Moscow, 119571 Russia
e-mail: berenblyum@gmail.com
Received August 31, 2016
Abstract—A nickel sulfide catalyst which efficient in the decarbonylation of fatty acids to olefins and dienes has
been obtained for the first time by treating alumina-supported nickel sulfate with hydrogen, and its properties
have been studied. In its presence, the olefin selectivity of the reaction can exceed 90%. The kinetics of stearic
acid deoxygenation to heptadecenes has been investigated, a kinetic model has been constructed, and a mecha-
nism has been proposed for the reaction over this catalyst. Olefin oligomerization is the dominant side reaction.
Kinetic evidence for the catalytic inhibition of oligomerization by nickel hydrides formed on the catalyst has
been obtained. The compositions of active site–reactant adsorption complexes have been discussed.
Keywords: stearic acid deoxygenation, nickel sulfide catalyst, kinetics, mechanism, adsorption complexes of
the active site
DOI: 10.1134/S0023158417020069
INTRODUCTION
lyzes deoxygenation and hydrodeoxygenation reac-
tions as well [17].
Despite the rapidly changing oil prices, interest in
renewable resources persists [1]. Inedible fats and veg-
etable oils, which are fatty acid triglycerides often con-
taining higher fatty acids, are one of the promising
types of such feedstock. The deoxygenation and
hydrodeoxygenation of such products yield paraffins
of fuel composition, and their catalytic partial hydro-
genation affords higher fatty alcohols [2]. Metals such
as Ru, Pd, Pt, Ir, Os, Rh, Ni, Cu, and Cr, generally on
supports, are used as catalysts [3–11]. The kinetics and
mechanism of stearic acid (SA) deoxygenation over a
Pd/Al O catalyst have been studied in detail [12].
In addition to paraffins of fuel composition and
higher alcohols, higher olefins, which are important
products serving as a feedstock for the preparation of
detergents, synthetic oils, additives, etc., can be
obtained from oils and fats [2]. Works on the selective
preparation of higher olefins from renewable raw
materials using heterogeneous catalysis are scarce [2].
Recently, we have shown that catalysts obtained from
nickel sulfate supported on alumina and treated with
hydrogen are promising for such a process [2]. The
aim of this work was to study the nature of the active
sites in such catalysts and the kinetics and mechanism
2
3
In industry, hydrodeoxygenation for obtaining par- of selective higher fatty acid deoxygenation to higher
affins of fuel composition is performed in the presence olefins in their presence. (There is no such informa-
of metal sulfides that are known as catalysts for tion in the literature.)
hydrodesulfurization. These catalysts contain Ni, Co,
Mo, and W compounds usually supported on γ-Al O .
2
3
EXPERIMENTAL
Chemicals, Solvents, and Materials
The following compounds were used: nickel sulfate
Such catalysts are generally obtained via the treatment
of preliminarily supported metal oxides with hydrogen
sulfide or other sulfur-containing compounds in a
hydrogen atmosphere. The hydrodeoxygenation is NiSO · 7H O (99.5%, reagent grade), nickel nitrate
4
2
accompanied by the carryover of sulfur from the cata- Ni(NO ) · 6H O (98.0%, pure grade), sodium
3
2
2
lyst in the form of H S. Because of this, additional sul-
2
hydroxide (98.0%, analytical grade), sulfuric acid
furization of the catalyst with sulfur-containing com- (94%, reagent grade), dodecane (Khimmed, Russia,
pounds contained in the feedstock or purposefully 99%, pure grade), tridecane (pure grade), chloroform
introduced into it is performed [13–16]. Recently, it (reagent grade), ethanol (distilled, 96%), methanol
has been shown that nickel sulfide Ni S powder cata- (analytical reagent for liquid chromatography), stearic
3
2
147

1
48
KATSMAN et al.
conditions and ~12.5 mL under experimental condi-
Concentration, mol/L
tions, the ratio of the gas and liquid phase volumes
under the reaction conditions was ~4.5, and the stirrer
rotation speed was 1000 rpm. The catalyst particle size
was 0.08–0.2 mm. About 20% of the dodecane was in
the gas phase.
0.5
0.4
0.3
0.2
1
2
Reaction Products Analysis and Catalyst
Characterization Methods
The procedures for the analysis of the liquid and
gaseous products of stearic acid conversion are
described earlier [12].
0
.1
3
0
20
40
60
80
100
Time, min
120
The olefins and dienes resulting from the deoxygen-
ation of SA and oleic acid (OA) were identified by the
GC–MS method on an Agilent Technology 6890N
system (United States) with a Model 5973 mass detector
using a electron impact ionization (70 eV). An HP-1
column with a length of 30 m was used for separation
Fig. 1. Concentrations of (1) CO, (2) heptadecenes, and
(
3) heptadecane as a function of time under the experi-
mental conditions of series 1 (Table 1). The initial hepta-
decene formation rate is close to the CO formation rate.
(
injection port temperature, 250°C; carrier gas flow
rate, 1 mL/min; temperature program: 5 min at 50°C,
heating to 250°C at a rate of 15 deg/min, and 5 min at
acid (99.2%, analytical grade), hydrogen (A brand,
9.99%), nitrogen (99.999%, special-purity grade),
and carbon monoxide (99.8%, pure grade). γ-Alumina
9
2
50°C).
The mass spectrum of olefin dimers (DOs) was
2
(
molded, A-0-A-1 brand, S = 254 m /g (our data),
BET
obtained on a Finnigan MAT Incos 50 spectrometer
Ukraine) was used as the support. The chemicals and
solvents were used as received.
(
United States) with direct input of the sample into the
ion source (electron impact ionization; electron
energy, 70 eV; ion source temperature, 240°C; sample
heating from 50 to 280°C at a rate of 10 deg/min).
Catalyst Synthesis
The catalysts were studied using transmission elec-
tron microscopy (TEM) on an LEO 912 AB OMEGA
microscope (Germany) at an accelerating voltage of
The support fraction with a particle size of 0.08–
0
.2 mm was used. To prepare a catalyst with a Ni con-
tent of 3.24%, 1.60 g of NiSO · 7H O and 12 mL of dis-
4
2
1
00 kV. The BET surface area was determined using an
tilled water for salt dissolution were taken per each 10 g
of the support. The support was placed into a plastic
vessel, the salt solution was added, and the contents
were slowly stirred for 1 h. The mixture was evaporated
in a porcelain bowl at 80–90°C for 3 h and dried at
ASAP 2020 instrument (Micromeritics, United
States). Sulfur was quantified on a Vario EL cube
CHNS microelemental analyzer (Elementar, Ger-
many), the weighed amount was 3–5 mg, and the
measurement accuracy was ±0.05%. Nickel was deter-
mined on a Spectroscan-LF X-ray fluorescence ana-
lyzer (OOO NPO SPEKTRON, Russia); the instru-
mental error was 0.5%.
1
20–130°C for 3 h with periodic stirring. The catalyst in
the reactor was pre-reduced with flowing hydrogen
(
50–60 mL/min) at 350°C, a pressure of 10 atm, and a
stirrer rotation speed of 300 rpm for 3 h. The BET sur-
2
face area of the catalyst was 233 m /g. In special-pur-
pose experiments, hydrogen treatment time was varied.
The catalyst from nickel nitrate was prepared in a similar
way. 1.56 g of Ni(NO ) · 6H O and 12 mL of distilled
Experimental Design and Kinetic Modeling
The reaction kinetics was studied according to a
special-purpose plan [18] developed to minimize the
number of experiments without sacrificing the maxi-
mum reliability of the results of modeling [12]. The
experimental design is presented in Table 1. To take
into account the transformations of SA during the
3
2
2
water were taken per each 10 g of alumina.
Catalytic Tests
The experiments were performed in a 50-mL auto- heating of reactor to the preset temperature, a “zero
clave equipped with a stirrer (Autoclave Engineers, experiment” was performed in each series as follows:
United States) according to the procedure reported in once this temperature had been reached, the reaction
our earlier work [12]. The catalytic experiments were was terminated and an analysis was performed. Based
performed under practically negligible diffusion lim- on its results, the initial concentrations were calcu-
itations at a constant catalytic activity [12]. Dodecane lated, and the corresponding reaction time was taken
served as the solvent.
The total autoclave volume was 73.2 mL, the reac-
tion mixture volume was ~10 mL under laboratory
to be zero. An example of the data obtained in this way
is graphically represented in Fig. 1.
The kinetic data are presented in Table 2.
KINETICS AND CATALYSIS
Vol. 58
No. 2
2017

KINETICS AND MECHANISM OF THE PRODUCTION
149
Table 1. Design of the experiments on stearic acid deoxygenation in the presence of the nickel sulfide catalyst
CSA
CNi
C
2
H O
PCO
atm
,
Time,
min
Series
T, °C
PH2 , atm
mol/L
0
, 30, 60, and 120
1
2
3
4
5
6
7
350
350
350
350
350
350
350
350
325
325
15
5
15
15
15
15
7
2
15
5
0.6
0.6
1.2
0.3
0.6
0.6
0.6
0.6
0.6
0.6
0.02
0.02
0.02
0.02
0.04
0.01
0.02
0.02
0.02
0.02
–
–
–
–
–
–
–
1.1
–
–
–
–
–
–
–
–
6
–
–
–
0 and 60
0 and 60
0 and 60
0 and 60
0 and 60
0 and 60
0 and 60
0 and 120
0 and 120
8*
9
0
1
The initial values of the reactant concentrations and gas pressures are given (specified at room temperature).
The experiments were conducted at three SA concentrations, three amounts of catalyst in terms of nickel, and four hydrogen pressure values.
Hydrogen was diluted with nitrogen to a total pressure of 15 atm.
*
Table 2. Experimentally measured concentrations of the main components of the stearic acid deoxygenation reaction over
the nickel sulfide catalyst and the corresponding values calculated using kinetic model constructed
Cexp(Ccalc), mol/L
Series
Time, min
SA
HDe
CO
HD
CO2
1
1
1
2
3
4
5
6
7
8
9
30
60
0.332(0.385)
0.233(0.296)
0.154(0.181)
0.302(0.289)
0.598(0.585)
0.123(0.081)
0.163(0.151)
0.349(0.321)
0.342(0.328)
0.434(0.410)
0.294(0.296)
0.355(0.304)
0.145(0.161)
0.181(0.182)
0.185(0.171)
0.114(0.115)
0.166(0.161)
0.107(0.117)
0.246(0.253)
0.116(0.117)
0.077(0.097)
0.071(0.060)
0.166(0.171)
0.132(0.128)
0.243(0.234)
0.356(0.315)
0.433(0.420)
0.212(0.233)
0.289(0.314)
0.200(0.204)
0.436(0.403)
0.216(0.214)
1.416(1.408)
0.113(0.141)
0.257(0.230)
0.200(0.193)
0.022(0.022)
0.032(0.036)
0.050(0.052)
0.015(0.018)
0.039(0.038)
0.025(0.028)
0.054(0.051)
0.023(0.015)
0.014(0.011)
0.007(0.003)
0.026(0.026)
0.017(0.016)
0.008(0.011)
0.024(0.019)
0.028(0.030)
0.023(0.025)
0.045(0.045)
0.008(0.007)
0.022(0.024)
0.019(0.018)
0.040(0.034)
0.011(0.009)
0.015(0.014)
0.016(0.016)
120
60
60
60
60
60
60
60
120
120
10
The experimental series are numbered in the same way as in Table 1.
The concentrations refer to the liquid phase volume under the kinetic experiment conditions.
HDe = heptadecenes,; HD = heptadecanes.
Quantum Chemical Modeling
First, the sulfate ion is reduced to sulfide, while the
nickel oxidation state remain unchanged; next, Ni(II)
reduction via an intermediate sulfide in which the
average (formal) nickel oxidation state is 4/3 begins.
Note that the reduction of palladium(II) to palla-
dium(0) on a support also proceeds via the formation
of compounds in an intermediate oxidation state [22].
Quantum chemical calculations were preformed
within the Priroda program [19] using the scalar rela-
tivistic approximation of the density functional
method. The PBE functional and the L11 all-electron
basis set were used [20]. The calculation procedure is
described in detail in Ref. [6].
It was shown [23] that, in the case of the reduction
of unsupported NiSO in excess hydrogen at 340–
4
RESULTS AND DISCUSSION
5
50°C, the main product is Ni S . Probably, under
3 2
The catalyst was obtained via the reduction of alu- these conditions, reaction (I) proceeds so rapidly that
mina-supported nickel sulfate with hydrogen. The fol- NiS formation is not detected.
lowing reactions take place in this case [21]:
NiSO + 4H = NiS + 4H O,
In our experiments, in the case of NiSO reduction
4
(I) on γ-Al O with hydrogen, water formation was
4
2
2
2 3
observed (reaction (I)) and then H S formation was
2
3
NiS + H = Ni S + H S,
(II)
2
3
2
2
detected using lead nitrate. This provides evidence for
Ni S + 2H = 3Ni + 2H S.
(III) reaction (II) and, possibly, reaction (III) taking place.
3
2
2
2
KINETICS AND CATALYSIS Vol. 58 No. 2 2017

1
50
KATSMAN et al.
2.61
2.50
3.54
Fig. 2. Structures of the Ni , Ni S , and Ni S clusters optimized using the PBE/L11 method (from left to right). The inter-
15
13
2
11 4
nuclear distances are specified in Å.
This is in agreement with the above scheme, according
Note that, when the catalyst reduction with hydro-
to which nickel sulfate is first reduced to nickel sulfide, gen is continued, the H S release does not stop, which
2
releasing water, and then, with the accumulation of is indicative of a further decrease in the sulfur content
sulfides, their reduction with H S release occurs.
of the catalyst. In the case of reduction for 8 h, the a. f.
of S in the catalyst decreases to 10.7%. Probably, this
is the reason why the performance of the catalyst
worsens. Hence, the catalyst in which the a. f. of S is
2
According to TEM data, particles of nickel com-
pounds look like round, substantially darkened
regions against the background of alumina. According
to the data for 48 particles, they are nanosized and
their average diameter is 4.7 ± 0.7 nm. From the data
acquired in the electron diffraction mode, we reliably
determined four interplanar spacings (IPSs) for the
synthesized catalyst, fewer than for crystalline sul-
fides. The data obtained allow identifying nickel sul-
fides Ni S (IPS: 2.012 and 1.374 Å, JCPDS card
~
30% is the optimal one for obtaining the maximum
higher-olefin yield. In the kinetic experiments, the
indicators of catalyst performance remained quite sta-
ble. For example, when repeating the experiment of
series 1 (120 min) on a used catalyst portion, the reac-
tant concentrations are reproduced within a typical
error of ±5%, according to the results of analysis.
3
2
Interesting results were obtained by the quantum
chemical modeling of the effect of sulfur introduction
on the structure and reactivity of the active sites of the
nickel catalysts. Fifteen-nuclei clusters containing
various numbers of sulfur atoms were chosen to be the
models of the catalyst’s active site (Fig. 2).
no. 44-1418) and NiS (IPS: 2.380 and 1.119 Å, JCPDS
card no. 47-1739) in the catalyst. Ni IPSs were not
detected, probably, because of the too small sizes of
the nickel particles. The presence of only a limited
number of IPSs may be due both to the structural fea-
tures of the nanoparticles and to their interaction with
each other and with the support surface [24].
Since the R–COOH carbon–carbon bond breaking
in the acid molecule is the key step of its deoxygenation
The atomic fraction (a. f.) of sulfur in our Ni–S
system calculated from the results of the elemental
analysis of the catalyst is 30.3%. For a homogeneous
phase, this corresponds to a solid solutions of S in Ni
[
7, 12], the activation parameters of this step proceeding
over the site models containing 0 to 4 S atoms were cal-
culated and compared. For this purpose, for each of
these sites we carried out a quantum chemical simula-
tion of C–C bond breaking in the propanoic acid
molecule (model compound) that results in the for-
mation of –COOH and –C H particles bound to the
[
25]. If, however, the sulfides NiS (a. f. S = 50%) and
Ni S (a. f. S = 40%) are present, then the a. f. of S in
3
2
the homogeneous phase will be lower.
2
5
active site.
The catalytic properties of the nickel sulfide cata-
lyst substantially excel the properties of the sulfur-free
nickel catalyst (synthesized from nickel nitrate). The
stearic acid conversion and heptadecene (HDe) yield
in the presence of the nickel sulfide catalyst are 76.9
and 35.5%, respectively, while in the presence of the
metal catalyst, they are 41.3 and 7.71%, respectively
Table 3 shows the results of the calculations of the
≠
dependence of the free activation energy ΔG
of
(
623K
)
the aforementioned step on the number of sulfur atoms
on the surface of the active site model. Introducing S
atoms successively lowers the activation energy.
(
see experimental conditions in Table 1, series 1, reac-
It can be seen in Fig. 2 that introducing S atoms
tion time of 2 h). The catalyst was usually reduced with into the nickel active site model induces metal–metal
hydrogen for 3 h. Shortening the reduction time to 1 h bond elongation, thus “distorting” the nickel surface.
did not lead to any substantial change in the HDe In particular, almost complete Ni–Ni bond breaking
yield. However, extending the reduction time from 3 to occurs in the case of Ni11
S
, which leads to an increase
4
8
h decreased the stearic acid conversion and olefin in the coordinative unsaturation of each of these two
yield to 61.1 and 20.4%, respectively.
nickel atoms or, in other words, to a growth in the
KINETICS AND CATALYSIS
Vol. 58
No. 2
2017

KINETICS AND MECHANISM OF THE PRODUCTION
151
number of vacancies for binding stearic acid that Table 3. Gibbs activation energies of C–C bond breaking in
propanoic acid over the surface of the Ni , Ni S , and
undergoes deoxygenation and the fragments of its
molecule being formed. According to the performed
calculations, the acid molecule binds initially to nickel
atoms in Ni S in the same way as in Ni , through the
carbonyl oxygen and methyl or methylene hydrogen.
It was found [26] that, in the case of a change in the
S/Ni ratio, even the electric conductivity of sulfides
changes sharply.
15
13 2
Ni S clusters (350°C)
11
4
ΔG 6# 23К*,
kcal/mol
Cluster
Symmetry
11
4
15
Ni15
C2v
C2
C2v
13.0
11.3
Ni S
13 2
9.8
Ni S
1 4
1
Calculated data (Table 3) help to explain our exper-
imental evidence that the presence of a certain amount
*
The ΔG6≠
values are relative to the energy of noninteracting
23К
of S in nickel catalysts improves their performance in cluster and propanoic acid.
the deoxygenation of fatty acids to higher olefins. A
decrease in the activation barrier of the R–COOH
bond breaking step (Table 3) under the assumption tion mechanism and kinetics were based on the
that the energy of the acid adsorption on the consid- extended Langmuir–Hinshelwood approach [12, 29].
ered clusters is almost the same may be a significant It includes fast reversible adsorption of the reactants
cause of the increase in the activity.
on catalyst active sites and their slow irreversible trans-
formation into the products. Having supposed various
compositions of adsorption complexes forming on the
nickel sulfide catalyst surface, we assumed that the
active site, which is probably a nanocluster, may bind
one or more reactant molecules. The study was based
on the chemistry of SA conversion in the presence of a
Obviously, increasing the amount of sulfur only
finitely improves rthe performance of the catalyst.
Probably, Ni–S–Ni fragments form [27] instead of Ni–
Ni bonds in the case of a large S/Ni ratio, which may
complicate the interaction of nickel atoms with an acid
molecule. Our attempts to optimize the structure of the
Pd/Al O catalyst, which includes 17 routes [12].
1
5-nuclear cluster with adjacent S atoms with respect to
2
3
energy resulted in its substantial deformation with the
We analyzed a total of ~100 hypotheses. The math-
formation of Ni–S–Ni fragments. It is known that ematical model obtained is adequate to the experi-
introducing sulfur compounds may cause complete mental data: the mean square error of the fit is ±4.3%
deactivation of a Ni catalyst in the hydrodeoxygenation (by way of illustration, see Table 2 and Fig. 3). The
reaction because of its saturation with sulfur [28].
For the understanding of the specific character of
model is described in detail below.
The modeling results made it possible to find the
the properties of the nickel sulfide catalyst, a detailed main routes (steps) of SA conversion over the nickel
kinetic model was developed. Hypotheses on the reac- sulfide catalyst:
(
(
(
(
(
(
(
(
(
1) SA = HDe + CO + H O
SA decarbonylation,
HDe hydrogenation to heptadecanes (HD),
SA decarbonylation,
2
2) HDe + H ⇄ HD
2
3′) SA = HDe + CO + H O
2
3′′) SA = HD + CO2
SA decarboxylation,
4) 2SA ⇄ Ah + H O
SA anhydride (Ah) formation,
heptadecyl stearate (Es) formation,
Ah decarboxylation to diheptadecyl ketone (Kn),
CO steam reforming,
2
5) SA + HDe ⇄ Es
6) Ah = Kn + CO2
7) CO + H O ⇄ CO + H
2
2
2
8) 2HDe = DO
HDe dimerization.
As distinct from what is observed for the palladium
While in the presence of the palladium catalyst
catalyst, the number of routes of SA conversion over heptadecene forms only at a low hydrogen pressure
the nickel sulfide catalyst under study is smaller. For and a low SA conversion, it is heptadecene that is the
example, SA hydrodeoxygenation to octadecane and main product in the case of the nickel sulfide cata-
the reduction of carbon oxides to methane do not lyst. Its oligomers (represented by DOs) and hepta-
occur. Diheptadecyl ketone Kn forms only from stea- decane form in smaller amounts (see, e.g., experi-
ric acid anhydride Ah.
mental series 1, reaction time of 1 h; product concen-
Routes (4)–(6), and (8) are noncatalytic. The for- trations are 0.182 mol/L, 0.046 mol/L (calculated
mation of DOs is confirmed by direct-input mass using the kinetic model), and 0.036 mol/L, respec-
spectrometry. Its spectrum contains a line at m/e 761, tively). The heptadecene selectivity of the reaction
which is due to the ester formed from SA and olefin increases with an increasing amount of catalyst, which
dimers.
is quite expected since the olefin synthesis step is
KINETICS AND CATALYSIS Vol. 58 No. 2 2017

1
52
KATSMAN et al.
high SA concentration, a high CO pressure, etc.), the
Calculation
selective Kn formation can reach 10%. The selectivity
of reversible anhydride Ah formation does not exceed
1%. Stearic acid can reversibly bind up to 16% HDe
into the ester ES, but heptadecene is released again as
the conversion increases.
0
0
0
0
0
0
.7
.6
.5
.4
.3
.2
y = 0.9833x
2
R = 0.9806
Therefore, the high selectivity of the nickel sulfide
catalyst in the higher olefin formation is not deter-
mined by the stearic acid conversions discussed above.
Heptadecene dimerization (oligomerization)—reac-
tion (8)—turns out to be the main route decreasing the
selectivity. According to the kinetic model, this process
decelerates with an increasing amount of catalyst and
hydrogen pressure (Table 4). It can be assumed that Ni
in the catalyst retains its capacity for reversible hydride
0
.1
0
0.1
0.2
0.3
0.4
0.5
0.6
Experiment
0.7
Fig. 3. Fit between the values calculated via the kinetic
model and the experimentally measured concentrations of
the reactants.
formation in an H atmosphere [12, 30]. According to
2
Bullock [31], metal hydrides are capable of terminating
radical reaction chains, apparently including oligomer-
ization. The consumption of metal hydrides is com-
pensated for by their continuous formation in the
hydrogen atmosphere [2]. This can cause a slowdown
of HDe oligomerization and the corresponding
increase in the olefin selectivity. Note that metal
hydrides are capable of terminating almost all similar
reactions involving not only radicals but also carbe-
nium cations, carbanions, and metal complexes [31].
This can be called “negative” side reaction catalysis
that enhances the target process selectivity. Certainly,
the yield of higher olefins, particularly α-olefins, can
be increased by means of their continuous removal
from the reaction zone, e.g., by distillation [32].
accelerated. Because of the consecutive character of
the reaction, the selectivity may exceed 90% at low
stearic acid conversions.
In the presence of the palladium catalyst, up to a half
of the entire HD forms via stearic acid decarboxylation
[
12]. A similar situation is observed for the nickel sulfide
catalyst: 5 to 50% of the resulting HD is due to decar-
boxylation, while the rest of HD is due to HDe hydro-
genation. Here, HD is not the main reaction product
(
HD selectivity is 3 to 17%). The reason lies in the low
activity of this catalyst in both HDe hydrogenation and
stearic acid decarboxylation.
Let us discuss the main features of the kinetics of the
stearic acid deoxygenation in the presence of the cata-
lyst under study in comparison with a Pd/Al O catalyst
The amount of the ketone Kn (product of the non-
catalytic reaction) forming at a high catalyst concen-
tration and a low stearic acid concentration is negligi-
ble. When the catalytic reaction slows down (at a low
catalyst concentration, a reduced hydrogen pressure, a
2
3
[
12]. As in the case of the palladium catalyst [12], the SA
transformation starts from reactant adsorption on the
active site (Z). Then, as in the case of other of catalysts
[
2], R–COOH bond breaking in the adsorbed SA mol-
ecule occurs in the key reaction step.
Table 4. Equilibrium constants of the adsorption steps
The bound –COOH group undergoes transforma-
tion on hydrogen- or water-containing active sites (see
Kinetic Model of the Reaction). As in the presence of
other catalysts [2], decarbonylation is the dominant
process: the total ratio between the contributions from
this route and decarboxylation is ~97 : 3. In other
words, deoxygenation over the nickel sulfide catalyst
proceeds by two parallel steps (1) and (3), whereof
step (1) is accelerated by hydrogen and step (3) by
water vapor. In step (1), as in the case of Pd, the entire
SA undergoes decarbonylation. But, as distinct from
what is observed for the Pd catalyst, only a small part
of SA is decarboxylated by step (3) in the presence of
Determination
Eq.
logK*
error for
log(K /K )**
i
VI'
I'
II'
2.99
7.47
4.79
0
6.99
5.66
5.85
4.61
0.21
0.23
0.23
>1
0.22
–
0.18
0.23
III'
IV'
V'
VI'
VII'
VIII'
*
The unit of measure of the equilibrium constants in Eqs. (I'), the nickel sulfide catalyst. For this reason, step (3) is
–3 –4
(
(
II'), and (VI') is (mol/L) ; in Eq. (III'), (mol/L) ; in Eqs.
written as two equations, Eqs. (3′) and (3′′), whereof
step (3′) describes the decarbonylation and step (3′′)
describes decarboxylation. An additional difference
lies in the fact that two water molecules rather than
one as is in the case of Pd are present in the corre-
–2
IV'), (V'), (VII'), and (VIII'), (mol/L) . The adsorption equi-
librium constants are assumed to be temperature-independent.
(
This assumption has no effect on the adequacy of kinetic data
description.)
* Equilibrium constant ratios are uniquely determined.
*
KINETICS AND CATALYSIS
Vol. 58
No. 2
2017

KINETICS AND MECHANISM OF THE PRODUCTION
153
sponding adsorption complex (see Kinetic Model of the C dienes opens the way to preparing these valuable
1
7
Reaction).
HDe hydrogenation via reaction (2) proceeds via
products from renewable raw materials.
two parallel kinetic pathways, namely, adsorption com-
plex containing HDe and hydrogen (2-1) and the inter-
action of HDe from the reaction medium with a hydro-
gen-containing active site (2-2) according to the Eley–
Rideal mechanism [33]. According to the kinetic mod-
eling results, pathway (2-1) includes the formation of
the adsorption complex Z(HDe)(SA)(H O)(H ). An
Kinetic Model of the Reaction
The reactant–Z adsorption complexes selected by
modeling form via the following reactions:
Z + SA + 2H = Z(SA)(H ) ,
(I')
2
2 2
Z + SA + 2H O = Z(SA)(H O) ,
(II')
2
2
2
2
2
analysis of the kinetic model showed that SA totally
excels the other reactants in terms of the adsorption
coverage of active sites.
Z + HDe + SA + H O+ H2
2
(III')
= Z(HDe)(SA)(H O)(H ),
2
2
Therefore, the probability of HDe adsorption by a
free active site is low, and HDe binds mainly to the
active site that is already occupied by SA. As was noted
above, the possibility of multimolecular binding by
active sites in nanocatalysts was demonstrated in a
series of works, e.g., [12, 29]. On the other hand, the
conclusion that higher fatty acid molecules bind much
more strongly to the active sites of sulfided catalysts
than the molecules of compounds of other classes was
drawn in Ref. [13] from catalytic data.
The SA conversion mechanism includes thermal
steps along with catalytic steps. Reversible noncata-
lytic steps (4) and (5), shifted to the left, yield low con-
centrations of anhydride Ah and ester Es, respectively.
Step (6) represents thermal Ah decarboxylation to
ketone Kn.
Z + H O+ H = Z(H O)(H ),
(IV')
(V')
2
2
2
2
Z + 2H O = Z(H O) ,
2
2
2
Z+ 2CO+ H = Z(CO) (H ),
(VI')
2
2
2
Z+ CO+ H O = Z(СO)(H O), (VII')
2
2
Z + 2H = Z(H ) .
(VIII')
2
2 2
Table 4 lists the equilibrium constants (Ki) of these
reactions.
These data make it possible to estimate the reactant
coverages for the active sites. The fraction of free active
sites during the reaction does not exceed 0.01%. The
fraction of Z(SA)(H O) is up to 80%, and that of
2
2
Z(H O) is up to 20%. The values of the rest of the
coverages are substantially lower.
2
2
The greatest difference in catalytic properties
between the nickel sulfide and palladium catalysts is
observed in the deoxygenation of oleic acid, both pure
and mixed with SA (1 : 1). Over the Pd catalyst, it
undergoes hydrogenation to SA and the deoxygen-
ation of the resulting SA begins only after OA is con-
sumed [8, 12, 34]. In the presence of the nickel sulfide
catalyst, OA undergoes deoxygenation concurrently
with hydrogenation. Here, heptadecadiene with ter-
minal and internal double bonds should form primar-
ily. Using the GC–MS method, one ion with m/e 236
was detected, which is due to α,ω-heptadecadiene. Its
occurrence can be explained by the isomerization of
the primary product.
The rate equations of the steps proceeding via dif-
ferent routes and the corresponding rate constants are
presented in Table 5.
The rate equations for the main routes of SA deox-
ygenation expressed in terms of reactant concentra-
tions appear as
2
H
w = k K C C C /Den decarbonylation,
1
1
I
SA
Ni
2
w2 = k2(1)K C C C C C /Den
III HDe SA H O
H
Ni
2
2
2
H
+
k2(2)CHDeKVIIIC C /Den hydrogenation,
Ni
2
2
w = k K C C C /Den decarboxylation,
3
3
8
II SA H O Ni
2
w
= k
C
HDe/(CH2
C
Ni)0.5 oligomerization,
Therefore, oleic acid almost does not inhibit stearic
acid deoxygenation in the presence of the nickel sulfide
catalyst as opposed to palladium catalysts [8, 12, 34].
8
where k is the step j rate constant, K is the adsorption
j
i
2
Apparently, the active sites of the nickel sulfide catalyst step i equilibrium constant, and Den = 1 + K C C
+
+
I' SA
H
2
are not blocked by oleic acid molecules, as distinct from
those of the palladium catalysts. The outcomes of the
2
K C C
+ K C CHDeC CH2 + K C CH2
H O
III' SA IV’ H O
II' SA H 2O
2
2
2
2
2
H
deoxygenation of pure stearic and oleic acids and their K C
+ K C C + K C C
+ KVIII/C . The
total active site concentration is conventionally assumed
V’ H 2O
VI' CO H2
VII' CO H O
2
2
mixture differ little.
to be equal to C .
Apparently, the diene hydrogenation rate in the
presence of the nickel sulfide catalyst is quite high and
a significant part of the dienes is saturated with hydro- kinetic experiments at 325°C were used. Table 5 lists
gen to heptadecenes. Nevertheless, our finding that the significant rate constants and the errors of their
the nickel sulfide catalyst can convert oleic acid pres- calculation determined by Fisher matrix inversion [22,
ent in oils and fats in significant quantities into higher 35]. The activation energies of the corresponding steps
Ni
To determine the activation parameters, the data of
KINETICS AND CATALYSIS Vol. 58 No. 2 2017

1
54
KATSMAN et al.
Table 5. Rate equations and step parameters
logk
E ,
kcal/mol
Route
a
Rate equation
logK
(
step)
350°C
325°C
1
k [Z(SA)(H ) ]
k2–1[Z(HDe)(SA)(H O)(H )]
1.78 ± 0.77
1.11 ± 0.09
–0.88 ± 0.10
–0.68 ± 0.18 –0.88 ± 0.13
–2.65 ± 0.22
–4.95
1.16 ± 0.11
0.85 ± 0.09
–
–
–
–
–
–3.49
42.4
3.9
–
13.7
–
1
2 2
2
2
-1
-2
2 2
k2–2[HDe][Z(H2)2]
k [Z(SA)(H O) ]
3
4
3
2
2
2
–
k ([SA] – [Ah][H O]/K ])
4
2
4
5
6
7
8
k ([SA][ HDe] – [Es]/K ])
–
–1.14 ± 0.16
–
–2.93
–
1.30
–
–
52.5
–
5
5
k6[Ah]
–0.37
k [Z(CO)(H O)](1 –[H ][CO ]/(K [CO][H O])) –0.72 ± 0.10
7
2
2
2
7
2
k [HDe](C ) 0.5[H2]–0.5
–
43.8
–3.26 ± 0.07 –3.91 ± 0.26
8
Ni
The concentrations are expressed in mol/L (liquid and gaseous compounds in the corresponding phases); time, in min.
k and K are determined as a product with an error of ±0.56.
6
4
k and K are determined as a quotient (the reverse reaction rate constant) with an error of ±0.32.
5
5
CNi is the catalyst loading in g-mol Ni/(L liquid phase).
“
–” means that the table does not provide insignificant values at 325°C (see main text). Accordingly, there are no activation energy val-
ues as well. Equilibrium constants for irreversible steps are not presented either.
calculated via the Arrhenius equation are also provided. constant K was chosen to be K . Table 4 lists the loga-
VI
j
Table 5 does not present the rate constants of the steps rithmic errors for these NPFs. K and k are deter-
4
6
whose contribution is insignificant. At 325°C, these
include Es formation, steam CO reforming, and HDe
hydrogenation via the Eley–Rideal mechanism.
mined as a product with a logarithmic error of ±0.56.
The reason lies in the relatively rapid reversible forma-
tion of Ah via step (4) prior to its decomposition to Kn
via step (6). The constants k and K are determined as
The ketone Kn and olefin dimer formation steps are
characterized by the maximum activation energy values.
Therefore, reducing the temperature somewhat
enhances the olefin selectivity of the reaction. However,
in the case of temperature reduction, the low activation
energy of step (3) will lead to some increase in the con-
5
5
a quotient (logarithmic error of ±0.32). This quotient is
equal to the rate constant of Es dissociation, which is
the reverse to the reaction of Es formation from HDe
and SA. This constant is unambiguously determined.
The kinetic model of the reaction made it possible to
tribution from the decarboxylation step, which has only estimate the relative contributions from the routes of
a slight effect on the result because step (3) proceeds stearic acid conversion. The primary transformations of
according to the decarbonylation stoichiometry to the SA proceed via routes (1), (3), and (6) and result almost
extent of 98%. The kinetic model at 325°C adequately quantitatively in the heptadecene formation (above
describes the experimental data.
97%). The other products are heptadecanes and revers-
ibly forming anhydride Ah. Subsequently, olefin dimer
accumulation starts. As the reaction proceeds, the con-
tribution from this route increases to ~50% at an SA
conversion of ~80%. Therefore, route (8) is the main
pathway for olefin loss. A growth in the contribution
from heptadecanes from 3 to 15% as a result of heptade-
cene hydrogenation and stearic acid decarboxylation is
also observed. Depending on the reaction conditions,
the contribution from Kn formation can increase from
We carried out a numerical analysis of the identifi-
ability of the model parameters obtained by solving the
reverse kinetic problem (Tables 4, 5). Our experimental
design (Table 1) provided a high degree of informativity
with respect to the number and quality of estimates of
kinetic and adsorption parameters. A series of rate and
equilibrium constants were unambiguously estimated.
The error values for such rate constants are specified in
Table 5. The other parameters were determined in the
form of nonlinear parametric functions (NPFs), so-
called complexes, for which errors were calculated by
dimensionless Fisher matrix inversion as well [22, 35].
Below, the formulas for these NPFs are provided and
the causes of their appearance are specified.
1
to 10% with a decrease in the catalyst concentration
and hydrogen pressure and an increase in CO pressure.
Up to 15% of the CO turns into CO via route (7)
2
through steam reforming.
CONCLUSIONS
In this work, we have demonstrated new opportuni-
The ratios of the adsorption step rate constants
rather than constants themselves were unambiguously
determined. The reason lies in the high coverage of the ties for designing catalysts and a process for the synthesis
catalyst active sites. The corresponding NPFs are writ- of higher olefins and higher dienes from higher fatty
ten as K /K . For the certainty of values, the equilibrium acids obtained via the processing of oils and fats, tradi-
i
j
KINETICS AND CATALYSIS
Vol. 58
No. 2
2017

KINETICS AND MECHANISM OF THE PRODUCTION
155
tional type of renewable raw material, using stearic and
oleic acids as examples. It has been found that the nickel
sulfide catalyst obtained by treating alumina-supported
nickel sulfate with hydrogen is active and selective in
higher fatty acid decarbonylation. In the presence of this
catalyst, the higher olefins selectivity can exceed 90%.
Unsaturated higher fatty acids, e.g., oleic acid, turn into
higher dienes, which were not earlier prepared using
thid method. For the first time, the kinetics has been
studied, a kinetic model has been constructed, and a
mechanism of higher fatty acid decarbonylation over
9. Dragu, A., Kinayyigit, S., García-Suárez, E.J., Flo-
rea, M., Stepan, E., Velea, S., Tanase, L., Collière, V.,
Philippot, K., Granger, P., and Parvulescu, V.I., Appl.
Catal., A, 2015, vol. 504, p. 81.
0. Ahmadi, M., Nambo, A., Jasinski, J.B., Ratnasamy, P.,
and Carreon, M.A., Catal. Sci. Technol., 2015, vol. 5,
no. 1, p. 380.
1
1
1. Peng, B., Zhao, C., Kasakov, S., Foraita, S., and Ler-
cher, J.A., Chem. Eur. J., 2013, vol. 19, no. 15, p. 4732.
1
2. Berenblyum, A.S., Podoplelova, T.A., Katsman, E.A.,
Shamsiev, R.S., and Danyushevsky, V.Ya., Kinet.
Catal., 2012, vol. 53, no. 5, p. 595.
the nickel sulfide catalyst has been proposed, which, in 13. Brillouet, S., Baltag, E., Brunet, S., and Richard, F.,
particular, makes it possible to control this reaction. It
has been shown that the oligomerization of the target 14. Kubička, D. and Kaluža, L., Appl. Catal., A, 2010,
product is the main route that decreases the decarbon-
vol. 372, no. 2, p. 199.
Appl. Catal., B, 2014, vols. 148–149, p. 201.
ylation selectivity. Kinetic evidence has been obtained 15. Senol, O.I., Ryymin, E.-M., Viljava, T.-R., and
Krause, A.O.I., J. Mol. Catal. A: Chem., 2007,
vol. 268, p. 1.
for the possibility of increasing the olefin yield by means
of the catalytic inhibition of oligomerization with nickel
hydrides forming on the catalyst.
1
6. Anand, M., Farooqui, S.A., Rakesh, K., Rakesh, J.,
Rohit, K., Sibi, M.G., Singh, H., and Sinba, A.K.,
Appl. Catal., A, 2016, vol. 516, p. 144.
ACKNOWLEDGMENTS
17. Brimont, M.R., Dupont, C., Daudin, A., Geantet, C.,
and Raybaud, P., J. Catal., 2012, vol. 286, p. 153.
This work was supported by the Russian Founda-
1
8. Gorskii, V.G., Planirovanie kineticheskikh eksperimen-
tov (Design of Kinetic Experiments), Moscow: Nauka,
984.
9. Laikov, D.N., Chem. Phys. Lett., 1997, vol. 281, p. 151.
0. Laikov, D.N., Chem. Phys. Lett., 2005, vol. 416, p. 116.
1. Kirkpatrick, W.J., Adv. Catal., 1951, vol. 3, p. 329.
tion for Basic Research (grant nos. 14-03-00105 and
1
5-03-02906) and by the Russian Federal State Pro-
1
gram of the Ministry of Education and Science of the
Russian Federation (grant no. 114120870179, project
no. 564).
1
2
2
We are grateful to A.E. Gekhman for assistance in
making GC–MS analyses, to A.A. Kozlov and
V.M. Karpov for measuring specific surface areas, and
to P.V. Mel’nikov for determining the nickel content of
the catalyst.
2
2. Berenblyum, A.S., Wadhaf, H.A., Katsman, E.A., and
Flid, V.R., Kinet. Catal., 2011, vol. 52, no. 2, p. 296.
3. Habashi, F., Mikhail, S.A., and Van, K.V., Can. J.
Chem., 1976, vol. 54, p. 3646.
2
24. Nanomaterials in Catalysis, Serp, Ph. and Philippot, K.,
Eds., Weinheim, Germany: Wiley–VCH, 2013.
2
2
2
2
5. Samsonov, G.V. and Drozdova, S.V., Sul’fidy (Sul-
REFERENCES
fides), Moscow: Metallurgizdat, 1972.
1
. Lane, J., What Does $60 Oil Mean for the Biofuel Indus-
try, 2015. http://www.renewableenergyworld.com/
rea/news/article/2015/01/what-does-60-oil-mean-for-
the-biofuel-industry.
. Berenblyum, A.S., Danyushevsky, V.Ya., Kuznetsov, P.S.,
Katsman, E.A., and Shamsiev, R.S., Pet. Chem., 2016,
vol. 56, no. 5, p. 663.
6. Kumar, N., Raman, N., and Sundaresan, A., J. Solid
State Chem., 2013, vol. 208, p. 103.
7. Sowa, H., Ahsbahs, H., and Schmitz, W., Phys. Chem.
Miner., 2004, vol. 31, no. 5, p. 321.
2
8. Mortensen, P.M., Gardini, D., Carvalho, H.W.P.,
Damsgaard, C.D., Grunwaldt, J.-D., Jensen, P.A.,
Wagner, J.B., and Jensen, A.D., Catal. Sci. Technol.,
3
. Berenblyum, A.S., Danyushevsky, V.Ya., Katsman, E.A.,
Podoplelova, T.A., and Flid, V.R., Pet. Chem., 2010,
vol. 50, no. 4, p. 305.
2014, vol. 4, p. 3672.
2
9. Berenblyum, A.S., Al-Wadhaf, H.A., and Katsman, E.A.,
Pet. Chem., 2015, vol. 55, no. 2, p. 118.
0. Antonov, V.E., J. Alloys Compd., 2002, vols. 330–332,
p. 110.
4
5
6
. Berenblyum, A.S., Podoplelova, T.A., Shamsiev, R.S.,
3
Katsman, E.A., and Danyushevsky, V.Ya., Pet. Chem.,
2011, vol. 51, no. 5, p. 336.
31. Bullock, R.M., Comments Inorg. Chem., 1991, vol. 12,
. Berenblyum, A.S., Podoplelova, T.A., Shamsiev, R.S.,
Katsman, E.A., Danyushevsky, V.Ya., and Flid, V.R.,
Catal. Ind., 2012, no. 3, p. 209.
. Berenblyum, A.S., Shamsiev, R.S., Podoplelova, T.A.,
and Danyushevsky, V.Ya., Russ. J. Phys. Chem. A, 2012,
vol. 86, no. 8, p. 1199.
. Berenblyum, A.S., Danyushevsky, V.Ya., Katsman, E.A.,
Shamsiev, R.S., and Flid, V.R., Pet. Chem., 2013, vol. 53,
no. 6, p. 362.
no. 1, p. 1.
3
3
3
3
2. Chiappero, M., Do, P.T.M., Crossley, S., Lobban, L.L.,
and Resasco, D.E., Fuel, 2011, vol. 90, p. 1155.
3. Krylov, O.V., Geterogennyi kataliz (Heterogeneous
Catalysis), Moscow: Akademkniga, 2004.
4. Mäki-Arvela, P., Kubickova, I., Snåre, M., Eränen, K.,
and Murzin, D.Yu., Energy Fuels, 2007, vol. 21, p. 30.
5. Gorskii, V.G., Katsman, E.A., Klebanova, F.D., and
Grigor’ev, A.A., Teor. Eksp. Khim., 1987, vol. 23, p. 191.
7
8
. Simakova, I., Rozmysłowicz, B., Simakova, O., Mäki-
Arvela, P., Simakov, A., and Murzin, D.Yu., Top.
Catal., 2011, vol. 54, p. 460.
Translated by E. Boltukhina
KINETICS AND CATALYSIS Vol. 58 No. 2 2017