Supported iron nanoparticles for the hydrodeoxygenation of microalgal oil to green diesel

Article abstract of DOI:10.1016/j.jcat.2014.04.009

Iron nanoparticles supported on mesoporous silica nanoparticles (Fe-MSN) catalyze the hydrotreatment of fatty acids with high selectivity for hydrodeoxygenation over decarbonylation and hydrocracking. The catalysis is likely to involve a reverse Mars-Van Krevelen mechanism, in which the surface of iron is partially oxidized by the carboxylic groups of the substrate during the reaction. The strength of the metal-oxygen bonds that are formed affects the residence time of the reactants facilitating the successive conversion of carboxyl first into carbonyl and then into alcohol intermediates, thus dictating the selectivity of the process. The selectivity is also affected by the pretreatment of Fe-MSN, the more reduced the catalyst the higher the yield of hydrodeoxygenation product. Fe-MSN catalyzes the conversion of crude microalgal oil into diesel-range hydrocarbons.

Full text of DOI:10.1016/j.jcat.2014.04.009

Journal of Catalysis 314 (2014) 142–148
Contents lists available at ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Supported iron nanoparticles for the hydrodeoxygenation of microalgal
oil to green diesel
Kapil Kandel a,b, James W. Anderegg a, Nicholas C. Nelson a,b, Umesh Chaudhary a,b, Igor I. Slowing a,b,
⇑
a U.S. Department of Energy, Ames Laboratory, Ames, IA 50011-3020, USA
b Department of Chemistry, Iowa State University, Ames, IA 50011-3111, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
Iron nanoparticles supported on mesoporous silica nanoparticles (Fe-MSN) catalyze the hydrotreatment
of fatty acids with high selectivity for hydrodeoxygenation over decarbonylation and hydrocracking. The
catalysis is likely to involve a reverse Mars–Van Krevelen mechanism, in which the surface of iron is
partially oxidized by the carboxylic groups of the substrate during the reaction. The strength of the
metal–oxygen bonds that are formed affects the residence time of the reactants facilitating the successive
conversion of carboxyl first into carbonyl and then into alcohol intermediates, thus dictating the selectiv-
ity of the process. The selectivity is also affected by the pretreatment of Fe-MSN, the more reduced the
catalyst the higher the yield of hydrodeoxygenation product. Fe-MSN catalyzes the conversion of crude
microalgal oil into diesel-range hydrocarbons.
Received 11 November 2013
Revised 4 April 2014
Accepted 9 April 2014
Keywords:
Mesoporous silica nanoparticles
Iron nanoparticles
Fatty acids
Hydrodeoxygenation
Microalgae oil
Ó 2014 Elsevier Inc. All rights reserved.
Green diesel
1. Introduction
more economical sulfur-free catalyst to upgrade renewable oils,
Lercher has demonstrated the hydrodeoxygenation of microalgae
The increasing energy demand and concerns over the gradual
depletion of fossil fuels have attracted significant amount of
research to the exploration of alternative energy sources [1,2]. In
this context, microalgae are considered as one of the most promis-
ing renewable energy resources owing to their short harvest cycle,
small cultivation area, high lipid content (up to 80% of their dry
weight), and minimum greenhouse gas emission [3,4].
The major components of microalgal oil are free fatty acids
(FFAs) and triglycerides. These can be converted to fatty acid
methyl esters (FAMEs) by catalytic reaction with methanol and
used as biodiesel. However, due to the degree of unsaturation
and high oxygen content of the FAMEs, issues such as poor storage
stability, marginal cold flow, and engine compatibility limit their
widespread use [5,6]. An alternate technology to produce biofuels
from microalgal oil is through hydrotreating with Ni, Co, and Mo
sulfides or noble metal catalysts such as Pd and Pt supported on
metal oxides [7–14]. While the high price of the noble metals
can be avoided by using the sulfided catalysts, slow desulfurization
reduces their activity and contaminates the fuel [7,9,15]. Further-
more, these catalysts have shown poor selectivity, favoring crack-
ing and decarbonylation over hydrodeoxygenation to produce
broad hydrocarbon distributions [16]. In an effort to establish a
oil to alkanes by cascade reactions on bifunctional catalysts based
on Ni and an acidic zeolite [14,17]. Following work by the same
group illustrated the selectivity toward decarbonylation route by
supporting Ni catalyst on ZrO2 which directed the conversion
through two parallel pathways [18].
The success of Ni in the conversion of renewable feedstocks into
green diesel [14,19,28] stimulates the exploration of other inexpen-
sive transition metals as catalysts for the process. Considering its
rich redox-chemistry, high natural abundance and low price, iron
emerges as an appealing candidate for this kind of conversion. While
many researchers have been studying iron catalysts in the Fischer–
Tropsch synthesis for several decades [20,21], the activity and selec-
tivity of these species for the hydrodeoxygenation of fatty acids has
not been much explored. To contribute to the efforts for economical
and efficient catalysts for upgrading renewable feedstocks to green
diesel, we report the synthesis of iron nanoparticles supported on
mesoporous silica nanomaterials (Fe-MSN), and their application
in the hydrotreatment of fatty acids and crude microalgal oil.
2. Experimental
2.1. Materials
⇑
Corresponding author at: 2756 Gilman Hall, Iowa State University, Ames, IA
50011-3111, USA. Fax: +1 515 294 4709.
Pluronic P104 (>99.8%) was generously provided by BASF.
E-mail address: islowing@iastate.edu (I.I. Slowing).
Tetramethyl orthosilicate (TMOS, 98%), oleic acid (P99.0%) and
http://dx.doi.org/10.1016/j.jcat.2014.04.009
0021-9517/Ó 2014 Elsevier Inc. All rights reserved.

K. Kandel et al. / Journal of Catalysis 314 (2014) 142–148
143
Sylon (N,O-bis(trimethylsilyl)trifluoroacetamide, 99.3% and tri-
methylchlorosilane, 99.3%, 99:1) were purchased from Sigma
Aldrich. Iron (III) Nitrate [Fe(NO3)3Á9H2O] (100%), hydrochloric acid
(37.3%, ACS certified) and hexanes (certified, mixture of isomers,
boiling range 1.0 °C) were purchased from Fisher Scientific. All
reagents were used as received without further purification.
accomplished by shifting the spectra so that silicon 2p peak was
at 103.3 eV.
2.4. Catalytic activity measurements
All catalytic reactions were performed in a 100-mL batch reac-
tor (Parr Instruments). In a typical experiment, the catalyst (10 mg)
and oleic acid solution in hexanes (1 mM, 10 mL) were added in
the reactor. The reactor was purged three times with H2 and was
then pressurized with H2 to 30 bar at ambient temperature. For
kinetics study, the reaction was carried out at 290 °C for 1, 2, 3,
4, 5, and 6 h with constant stir rate (500 rpm). The reaction was
allowed to cool to room temperature and the catalyst was sepa-
rated. The reaction product was mixed with 1 mL Sylon (N,O-
bis(trimethylsilyl)trifluoroacetamide and trimethylchlorosilane,
99:1) and heated to 70 °C for 2 h for further derivatization. The
final mixture was analyzed in an Agilent GC–MS (7890A, 5975C)
with a HP – 5MS column. Runs started at 100 °C for 5 min, then
ramped to 200 °C at a rate of 20 °C minÀ1 held for 25 min, and then
ramped to 280 °C at 20 °C minÀ1 holding for 5 min at this temper-
ature. Methyl nonadecanoate was used as an internal standard.
Conversion was defined as mole% and calculated as moles of con-
verted oleic acid per mole of starting oleic acid times 100%. Yields
were defined as mole% and were calculated as moles of each prod-
uct per mole of starting oleic acid times 100%.
2.2. Catalyst preparation
MSN was prepared using a nonionic block co-polymer Pluronic
P104 surfactant [22]. In a typical synthesis, P104 (7.0 g) was dis-
solved in aqueous HCl (273.0 g, 1.6 M). After stirring for 1 h at
56 °C, tetramethylorthosilicate (TMOS, 10.64 g) was added and
stirred for additional 24 h. The resulting mixture was further
hydrothermally treated for 24 h at 150 °C in a high-pressure reac-
tor. Upon cooling to room temperature, the white solid was col-
lected by filtration, washed with copious amounts of methanol,
and dried in air. To remove the surfactant P104, the MSN material
was heated at a ramp rate of 1.5 °C minÀ1 and maintained at 550 °C
for 6 h. MSN was then mixed with water and stirred at room tem-
perature in order to rehydrate and regenerate the silanol groups,
followed by filtration and drying. For impregnation, Fe(NO3)3Á9H2O
(0.40 mmol, 0.16 g) was completely dissolved in water (0.48 mL).
To this solution, the rehydrated MSN (0.4 g) was added and mixed.
The solid mixture was calcined in air at a heating rate of
10 °C minÀ1 to 300 °C and maintained at that temperature for 3 h
followed by reduction at 400 °C for 6 h in a constant flow of H2
(1.67 mL sÀ1).
Similar experiment was conducted on crude microalgal oil
obtained from Solix Biofuels, Inc. by adding Fe-MSN catalyst
(10 mg) to a solution of microalgal oil (10 mg in 10 mL hexanes)
and heating to 290 °C under 30 bar H2 for 6 h.
2.3. Characterization
3. Results and discussion
Surface analysis of the catalyst was performed by nitrogen sorp-
tion isotherms at À196 °C in a Micromeritics Tristar analyzer. The
surface areas were calculated by the Brunauer–Emmett–Teller
(BET) method, and the pore size distribution was calculated by
the Barrett–Joyner–Halenda (BJH) method. Pretreatment of sam-
ples for surface area measurement was done by flowing N2 for
6 h at 100 °C. Powder X-ray diffraction patterns were obtained
with a Rigaku Ultima IV diffractometer using Cu target at 40 kV
and 44 mA, and samples were analyzed in the 0.8–90 2h° at a scan
rate of 1 2h° minÀ1. Cu Kb was removed using a monochromator.
Crystallite size was estimated from modeling the diffraction at
44.6 2h° with OriginPro software and incorporating the FWHM into
the Scherrer equation (d = Kk/bcosh, where d is the estimated crys-
3.1. Synthesis of Fe-MSN
The textural properties of MSN support and Fe-MSN catalyst are
summarized in Table 1. ICP measurement indicated that 6.0 wt% Fe
was immobilized on the MSN. Formation of the Fe nanoparticles
led to approximately 10% decrease in the surface area and pore vol-
ume of the support; however, its nitrogen sorption isotherm
remained type IV confirming retention of the mesoporous charac-
ter (Fig. 1a) [23]. TEM and STEM imaging suggested that the Fe
nanoparticles were located mainly inside the pores of MSN
(Fig. 1b). Low-angle XRD analysis confirmed that the structure of
the support was not affected by the formation of Fe nanoparticles,
as it preserved the p6mm pattern characteristic of SBA-15 type
materials (Fig. 1c) [24]. Wide-angle XRD showed a pattern of peaks
corresponding to the body-centered cubic phase of crystalline iron
nanoparticles (JCPDS card No. 89-7194, Fig. 1d) [25,26]. The wide
reflections indicated small crystallite size of the iron nanoparticles.
Estimation using Scherrer equation indicated that their size
(9.9 nm) was similar to the width of the mesopores (10.9 nm), sug-
gesting nanoparticle growth was restricted by pore width [27].
This observation was supported by estimation of the average size
of the Fe nanoparticles from TEM images (10.7 nm), which was
only slightly smaller than the pore width (Fig. S1).
tallite size, K is the shape factor, k is the wavelength of the Cu Ka, b
is the line broadening at half the maximum intensity in radians,
and h is the Bragg angle) [27]. For transmission electron micros-
copy measurements, an aliquot of the powder was sonicated in
methanol for 15 min. A single drop of this suspension was placed
on a lacey carbon-coated copper TEM grid and dried in air. The
TEM examination was completed on a Tecnai G2 F20 electron
microscope operated at 200 kV. Average particle size was calcu-
lated using ImageJ software based on five representative TEM
images (100 particles). Fourier transform infrared (FT-IR) spectra
were recorded on Nicolet Nexus 470. Samples were diluted with
KBr (about 5 wt%) and made into pellets for analysis in transmis-
sion mode. To measure the Fe loading, samples (2.0 mg) were
digested for 20 h in aqueous HF and HCl solution (0.18% and 5%
respectively) and analyzed in a Perkin Elmer Optima 2100 DV
ICP-OES. Temperature-programmed reduction was performed in
a Micromeritics AutoChem II using a flow of H2 in Argon (10.13%,
Table 1
Textural properties of the support and catalyst.
Material
Surface area
Pore volume
Pore diameter
(nm)
50 mL minÀ1
) ramping from 40 °C to 500 °C at a rate of
(m2 gÀ1
)
(cm3 gÀ1
)
10 °C minÀ1. XPS analysis was done with a PHI 5500 multi-tech-
nique system using a standard Al X-ray source. Since the samples
were mounted on two-sided scotch tape, charge correction was
MSN
Fe-MSN
331
295
0.97
0.88
11.1
10.9

144
K. Kandel et al. / Journal of Catalysis 314 (2014) 142–148
Fig. 1. (a) N2 sorption isotherms of MSN (red) and Fe-MSN (blue), (b) TEM (left) and HAADF-STEM (right) images of Fe-MSN, (c) XRD patterns of MSN (red) and Fe-MSN (blue),
insets show 10Â magnified 110 and 200 reflections, and (d) wide-angle XRD of Fe-MSN. (For interpretation of the references to color in this figure legend, the reader is
referred to the web version of this article.)
3.2. Oleic acid hydrotreatment with Fe-MSN
of n-octadecane increased linearly with the decrease in 1-octade-
canol until 5 h, supporting the notion that the alcohol is an
intermediate in the hydrodeoxygenation pathway (Fig. S2).
The formation of n-octadecane from 1-octadecanol could take
place as a two-step process, first involving dehydration of 1-octa-
decanol to give 1-octadecene, which would then be hydrogenated
to the saturated product (Scheme 1, pathway a). n-Heptadecane
could form either by direct decarboxylation of stearic acid or by
decarbonylation of octadecanal, which could also be an intermedi-
ate in the formation of 1-octadecanol (Scheme 1, pathway b). The
quick conversion of oleic acid to stearic acid suggests that the
hydrogenation of double bonds must be very fast under the reac-
tion conditions. Recent reports show that suspensions of iron
nanoparticles catalyze alkene hydrogenation at room temperature
and pressures as low as 1 bar H2 [29]. Therefore, if octadecene and
heptadecene form as intermediates, they should be transformed at
a high rate into octadecane and heptadecane, respectively. Indeed,
performing the Fe-MSN-catalyzed hydrogenation of oleic acid
under milder conditions allowed the detection of both alkenes
(about 2% yields at 10 bar H2 and 270 °C).
The kinetics of hydrotreatment of oleic acid (1 mM, 10 mL) with
Fe-MSN (10 mg) at 290 °C and 30 bar H2 pressure is shown in
Fig. 2. No oleic acid was detected after 0.5 h and the major hydro-
carbon product obtained after 6 h was n-octadecane, indicating
hydrodeoxygenation was the major reaction route. The yield of
n-octadecane (C18) increased continuously to 83% at 6 h, while that
of n-heptadecane (C17) grew slowly to 12% after 6 h (Fig. 2a). In
sharp contrast to the hydrogenation of oleic acid using nickel sup-
ported on MSN under the same reaction conditions (72% hydro-
cracking, 25% C17 and 3% C18) [28], the hydrocracking was almost
eliminated with Fe-MSN catalyst as it was only observed after
6 h reaction (3% yield).
Stearic acid and 1-octadecanol were observed with highest
yields at early reaction times, peaking before 0.5 and 2 h respec-
tively and then decreasing, which suggested that both species are
reaction intermediates (Fig. 2b). Since stearic acid disappeared
earlier than octadecanol, it is likely that the reaction proceeded
initially by a fast hydrogenation of C@C, followed by reduction of
the COOH group to alcohol, which eventually underwent
hydrodeoxygenation to give the major reaction product. The yield
The Fe-catalyzed reduction of acetic acid to acetaldehyde takes
place at 1 bar H2 in the range of 250–350 °C [30–32]. This suggests
Fig. 2. Kinetics of oleic acid hydrotreatment catalyzed by Fe-MSN: (a) hydrocarbon products, and (b) reaction intermediates.

K. Kandel et al. / Journal of Catalysis 314 (2014) 142–148
145
Scheme 1. Possible mechanisms for the conversion of oleic acid into n-octadecane
(blue, a) and n-heptadecane (red, b) from the intermediates 1-octadecanol and
octadecanal, respectively. R = n-pentadecyl. (For interpretation of the references to
color in this figure legend, the reader is referred to the web version of this article.)
that Fe-MSN should be able to catalyze the hydrogenation of stea-
ric acid to octadecanal under the conditions employed in this work.
Yet, the equilibrium constant for the hydrogenation of octadecanal
to 1-octadecanol at 260 °C is approximately 57, which explains
why the aldehyde is scarce under our reaction conditions [18].
Octadecanal could only be observed when performing the reaction
at a lower temperature (2.4% yield after 6 h at 30 bar H2 and
230 °C). Because aldehyde was not detected at 250 °C or higher
reaction temperatures, the aldehyde reduction rate to the alcohol
was concluded to be fast over Fe-MSN under the reaction condition
employed. Thus, if under our reaction conditions the rate of
interconversion between the aldehyde and alcohol is fast, the ratio
of n-heptadecane to n-octadecane products in the overall reaction
must be controlled by the relative rates of aldehyde decarbonyla-
tion and alcohol dehydration. Fe-MSN may favor the latter reac-
tion, for iron-based catalysts promote the dehydration of ethanol
at temperatures above 200 °C [33].
The production of liquid hydrocarbons was very low at 230 °C
but increased dramatically with temperature to almost 100% at
290 °C (Fig. 3a). Analysis of hydrocarbon distribution (Fig. 3b and
c) revealed that the increase of yield in this range was mainly
due to n-octadecane (from under 1% at 230 °C to 83% at 290 °C).
The yield of n-hepadecane also increased with temperature, but
to a smaller extent (from under 1% at 230 °C to 16% at 310 °C).
The products of hydrocracking were not observed at 230 °C but
were detected only at temperatures higher than 250 °C, with an
increase from 1% at 250 °C to 7% at 310 °C. It must be noted that
while the yields of both decarbonylation and hydrocracking prod-
ucts kept increasing from 290 °C to 310 °C, the yield of n-octade-
cane dropped from 83% at 290 °C to 72% at 310 °C. This suggests
that temperatures higher than 290 °C tend to favor decarbonyla-
tion and hydrocracking at the expense of hydrodeoxygenation.
These observations suggest that the apparent activation energies
for the processes leading to decarbonylation and hydrocracking
should be higher than the apparent activation energies of the steps
that lead to hydrodeoxygenation when Fe-MSN is used as a
catalyst.
Fig. 3. Effect of temperature on (a) the yield and (b) distribution of hydrocarbons in
the Fe-MSN-catalyzed hydrotreatment of oleic acid (6 h yields, 30 bar H2). (c) Same
plot as (b) showing details in the scale 0–20%. Blue circles correspond to n-
octadecane, red squares to n-heptadecane and black rhombi to hydrocracking
products. (For interpretation of the references to color in this figure legend, the
reader is referred to the web version of this article.)
[2,18]. This dependence of selectivity on hydrogen pressure is
likely the result of the participation of H2 in the equilibrium
between octadecanal and 1-octadecanol, the branching step in
the process. As hydrogen is required to convert the octadecanal
into 1-octadecanol, increasing the amount of the gas shifts the
equilibrium toward the alcohol favoring the route to n-octadecane,
while decreasing the amount of the gas has the opposite effect
leading to the n-heptadecane pathway.
3.3. Mechanism of oleic acid hydrotreatment with Fe-MSN
As mentioned above, iron oxides have been reported as cata-
lysts for the selective reduction of carboxylic acids to aldehydes
[34]. Pestman and co-workers demonstrated that the reduction
of acetic acid to acetaldehyde over iron oxides required the forma-
tion of a Fe(0) phase on the surface of the oxide [31]. They pro-
posed that hydrogen was first bound at the metallic sites and
then spilled over to reduce the acid, which was bound at defect
sites in the oxide [32,35,36]. Kinetic studies by Rachmady and Van-
nice supported this mechanism as they demonstrated that the
reaction fits a Langmuir–Hinshelwood model for two reactants
binding at two different types of site [30,37]. XRD analysis of our
catalyst showed only metallic iron and gave no evidence of the
oxide (Fig. 1d). However, given its redox potential, Fe can be read-
ily oxidized by mere exposure to air, forming sub-nanometer lay-
ers of oxide in less than 1 min [38,39]. Indeed, XPS analysis
revealed the presence of iron oxide in the catalyst (Fig. S3). Thus,
the absence of iron oxide peaks in the XRD was likely due to low
concentration and/or lack of crystallinity. Interestingly, the spent
catalyst showed a less-defined XRD pattern, in which the Fe(0)
As expected, the conversion of oleic acid was proportional to
the pressure of hydrogen applied (Fig. 4a). However, at 270 °C,
the total hydrocarbon yield showed little sensitivity to an increase
in pressure from 10 to 20 bar, going only from 33% to 36% and
requiring higher pressures to approach to full conversion. Consis-
tent with previous reports, low hydrogen pressures increased the
selectivity for decarbonylation and hydrocracking, whereas high
hydrogen pressures favored heavily hydrodeoxygenation (Fig. 4b)

146
K. Kandel et al. / Journal of Catalysis 314 (2014) 142–148
Fig. 4. Effect of H2 pressure on (a) the yield and (b) distribution of hydrocarbons in the Fe-MSN-catalyzed hydrotreatment of oleic acid at 270 °C (6 h yields). In (b) blue circles
correspond to n-octadecane, red squares to n-heptadecane and black rhombi to hydrocracking products. (For interpretation of the references to color in this figure legend, the
reader is referred to the web version of this article.)
Scheme 2. Proposed mechanism of the hydrogenation of carboxylic groups on the surface of the partially oxidized Fe nanoparticles of Fe-MSN: (a) reduction to aldehyde, and
(b) further reduction to alcohol. The oxygen atoms eliminated from the FFA may either escape as water or become part of a growing iron oxide lattice.
peak almost disappeared and a weak reflection at 35°, possibly
belonging to the oxide, emerged (Fig. S4). These observations sug-
gested that the process involved an active transformation of the
surface of the catalyst as it was continuously reduced by the H2
and oxidized by the carboxylic acid, through a reverse Mars–Van
Krevelen mechanism. In contrast to the work by Pestman et al.,
our original catalyst was Fe(0) rather than iron oxides, and the
reaction was performed at much higher H2 pressure. These differ-
ences had important consequences, as starting with Fe(0) provided
more surface for H2 to bind, which along with the higher H2 con-
centration prevented the reaction from stopping at the carbonyl
stage (Scheme 2). Consistently with the proposed reaction path-
ways, if the reduction stopped at the carbonyl stage, the main
product of hydrogenation under our reaction conditions should
have been n-heptadecane (decarbonylation). Conversely, if the
reduction continued to form the alcohol, the main product would
have been n-octadecane (hydrodeoxygenation), which indeed
was the case with Fe-MSN.
temperature-programmed reduction (TPR) profile of the parent
material (iron nitrate-impregnated MSN after calcination in air)
indicated two reduction steps with Tm at 240 °C and 345 °C
(Fig. S5), suggesting the transition between different oxidized
states of iron-MSN and the conversion into the final Fe(0)-MSN,
respectively. XRD analyses of the original calcined iron oxide-
MSN and of samples reduced at 240 °C and 350 °C gave no reflec-
tions, suggesting lack of crystallinity of the iron species. However,
XPS analyses of the same samples confirmed the presence of oxi-
dized iron in them (Fig. S6). Consistent with the hypothetical
mechanism, the C18:C17 selectivities of the hydrotreatment of oleic
acid using the non-reduced material and the samples reduced at
240 °C, 350 °C, 400 °C and 500 °C were proportional to the temper-
ature of reduction of iron in the catalysts (Fig. 5). The non-reduced
material gave the lowest yield of n-octadecane, suggesting that a
significant fraction of oleic acid stopped at the carbonyl stage lead-
ing the way to the decarbonylation product n-heptadecane. To the
contrary, the reduced materials led to higher selectivities for n-
octadecane, which results from the formation of 1-octadecanol.
This behavior is also consistent with the observed changes in
hydrocarbon selectivity with time (decrease for n-octadecane and
increase for n-heptadecane), as the Fe-MSN catalyst gradually
To evaluate if the degree of oxidation of iron controlled whether
the reaction stopped at the carbonyl or the alcohol stage, we com-
pared the selectivities of hydrotreatment of oleic acid using iron-
MSN materials reduced with H2 at increasing temperatures. The

K. Kandel et al. / Journal of Catalysis 314 (2014) 142–148
147
that the fatty acid composition of the extract consisted mainly of
unsaturated C16, C18 and C20 fatty acids (72.1 mol%) and saturated
C14 and C16 fatty acids (27 mol%, Table 2). Hydrotreatment of the
crude oil for 6 h using Fe-MSN as a catalyst gave 67% conversion,
the products being 16% alcohols, 33% unsaturated hydrocarbons
and 18% saturated hydrocarbons (Figs. 6 and S6). All of the remain-
ing 33% FFAs were saturated, even if most of the acids in the origi-
nal extract were unsaturated (72.1%). This behavior is consistent
with our observation that the hydrogenation of double bonds
was the first step in the hydrotreatment of oleic acid. The high ratio
of C18:C17 products (6.4:1) obtained during the hydrotreatment of
microalgal extract is very similar to the ratio we obtained during
the hydrotreatment of oleic acid (6.9:1), which is also consistent
with our observation that Fe-MSN favors hydrodeoxygenation over
decarbonylation.
Only a fraction of the fatty acids in microalgal oil exist as free
acids, while another part occur as di- or triglycerides. Thus, the
reduction of the fatty acid glycerides to hydrocarbons requires an
initial hydrogenolysis step. This additional step implies a higher
hydrogen consumption, which consequently should affect product
yield and distribution [40]. The presence of octadecene after
hydrotreatment of the crude microalgae oil is consistent with the
observation of unsaturated hydrocarbons at low H2 pressure
(Fig. 4), as the availability of H2 is diminished by its consumption
in the hydrodeoxygenation and hydrogenolyses. Further studies
to elucidate the detailed kinetics and mechanism on hydrotreat-
ment of triglycerides with Fe-MSN and pathways to decrease the
hydrogen consumption are currently under investigation in our
laboratory.
Fig. 5. Product selectivity as a function of the oxidation state of the catalyst.
Table 2
Fatty acid composition of microalgal oil.a
b
C12
C14
C16:2
1.13
C16:1
C16
C18:3
3.94
C18:2
C18
C20:5
0.32
4.59
21.88
22.34
14.79
0.67
30.32
a
Crude microalgal oil from Solix Biofuels, Inc.
First subindexes correspond to number of carbon atoms, subindexes following
colons correspond to the number of unsaturations in the chains.
b
changed from Fe(0) to a mixture of the metallic phase and an oxide
(Figs. 2a and S1).
It is likely that the tendency of Fe to favor hydrodeoxygenation
over decarbonylation in the hydrotreatment of oleic acid is directly
related to the strength of the Fe–O bonds that form during the
reaction. The strength of metal–oxygen bonds can be estimated
from the heats of formation of the highest oxides of the metal
[35]. For instance, Fe forms stronger metal–oxygen bonds than
Ni, which favors decarbonylation over hydrodeoxygenation when
supported on silica [18,28]. Thus, the capacity to form stronger
bonds allows for longer retention times of the oxygenated interme-
diates, which prevents the reaction from stopping at the carbonyl
stage and lead to higher hydrodeoxygenation yields.
4. Conclusions
Supported iron nanoparticles are efficient and sulfur-free alter-
native catalysts for converting microalgal oil into green diesel. Fe-
MSN catalyzes the hydrogenation of oleic acid to stearic acid,
which is further reduced to aldehyde and alcohol intermediates.
The H2 pressure of the reaction controls the equilibrium between
aldehyde and alcohol intermediates directing it to two main path-
ways: either decarbonylation of the aldehyde or dehydration of the
alcohol. Both of these processes give unsaturated hydrocarbons,
which get further hydrogenated to give liquid alkanes. The process
that involves dehydration of the alcohol has a lower temperature
barrier than that of decarbonylation and hydrocracking when Fe-
MSN is used as a catalyst and is the major pathway at 290 °C and
30 bar H2. This pathway gives full hydrodeoxygenation of oleic acid
with the highest carbon economy. Fe-MSN catalyzes this reaction
by dynamically changing its surface composition through a reverse
3.4. Conversion of microalgae oil into green diesel
A crude microalgal oil extract (Solix Biofuel, Inc., 10 mg in
10 mL hexanes) was directly hydrotreated with 10 mg of Fe-MSN
at 290 °C under 30 bar H2 in batch mode. GC–MS analysis showed
Fig. 6. Distribution of liquid products of Fe-MSN-catalyzed hydrotreatment of crude microalgal oil extract (Solix biofuels, Inc.; 6 h 290 °C and 30 bar H2). First subindexes
correspond to number of carbon atoms, subindexes following colons correspond to the number of unsaturations in the chains, OH designates alcohol.

148
K. Kandel et al. / Journal of Catalysis 314 (2014) 142–148
ˇ
ˇ
[5] P. Šimácek, D. Kubicka, G. Šebor, M. Pospíšil, Fuel 88 (2009) 456.
Mars–Van Krevelen mechanism in which carboxylic acids oxidize
the metallic surface and react with spilled-over hydrogen. It is
likely that the high selectivity of Fe-MSN for hydrodeoxygenation
results from the strength of the Fe–O bond, which allows for resi-
dence times that are long enough to complete the reduction of sub-
strate and intermediates. Tuning the degree of oxidation of Fe in
the catalyst allows controlling the ratio of hydrodeoxygenation to
decarbonylation of the hydrocarbon products. Using Fe-MSN as a
catalyst in the hydrotreatment of crude microalgal extract leads
to diesel-range hydrocarbons. In addition to the hydrodeoxygen-
ation of the FFAs, Fe-MSN catalyzed the hydrogenolysis of glyce-
rides. Further investigation of the mechanism and conditions of
Fe-MSN-catalyzed hydrogenolysis of glycerides are ongoing in
our laboratory. We envision that additional understanding of the
various conversions involved in the hydrotreatment of complex
oils will lead to the creation of more inexpensive and efficient cat-
alysts for converting biorenewable feedstocks into liquid hydrocar-
bon fuels.
ˇ
ˇ
[6] P. Priecel, D. Kubicka, L. Capek, Z. Bastl, P. Ryšánek, Appl. Catal. A 397 (2011)
127.
[7] G.W. Huber, P. O’Connor, A. Corma, Appl. Catal. A 329 (2007) 120.
ˇ
ˇ
ˇ
[8] D. Kubicka, L. Kaluza, Appl. Catal. A 372 (2010) 199.
ˇ
[9] D. Kubicka, J. Horácek, Appl. Catal. A 394 (2011) 9.
ˇ
[10] M. Snåre, I. Kubicková, P. Mäki-Arvela, K. Eränen, J. Wärnå, D.Y. Murzin, Chem.
Eng. J. 134 (2007) 29.
[11] S. Lestari, P.i. Mäki-Arvela, H. Bernas, O. Simakova, R. Sjöholm, J. Beltramini,
G.Q.M. Lu, J. Myllyoja, I. Simakova, D.Y. Murzin, Energy Fuels 23 (2009) 3842.
[12] B. Rozmysłowicz, P. Mäki-Arvela, S. Lestari, O. Simakova, K. Eränen, I.
Simakova, D. Murzin, T. Salmi, Top. Catal. 53 (2010) 1274.
[13] P. Do, M. Chiappero, L. Lobban, D. Resasco, Catal. Lett. 130 (2009) 9.
[14] C. Zhao, T. Bruck, J.A. Lercher, Green Chem. 15 (2013) 1720.
[15] T.R. Viljava, R.S. Komulainen, A.O.I. Krause, Catal. Today 60 (2000) 83.
[16] Y. Yang, C. Ochoa-Hernández, V.A. de la Peña O’Shea, J.M. Coronado, D.P.
Serrano, ACS Catal. 2 (2012) 592.
[17] B. Peng, Y. Yao, C. Zhao, J.A. Lercher, Angew. Chem. Int. Ed. 51 (2012) 2072.
[18] B. Peng, X. Yuan, C. Zhao, J.A. Lercher, J. Am. Chem. Soc. 134 (2012) 9400.
[19] W. Song, C. Zhao, J.A. Lercher, Chem. Eur. J. 19 (2013) 9833.
[20] G.P. Van Der Laan, A.A.C.M. Beenackers, Catal. Rev. 41 (1999) 255.
[21] D.L. Huber, Small 1 (2005) 482.
[22] T.-W. Kim, I.I. Slowing, P.-W. Chung, V.S.-Y. Lin, ACS Nano 5 (2010) 360.
[23] Y. Wan, D.Y. Zhao, Chem. Rev. 107 (2007) 2821.
[24] D. Zhao, J. Feng, Q. Huo, N. Melosh, G.H. Fredrickson, B.F. Chmelka, G.D. Stucky,
Science 279 (1998) 548.
[25] S. Peng, C. Wang, J. Xie, S. Sun, J. Am. Chem. Soc. 128 (2006) 10676.
[26] L.M. Lacroix, N. Frey Huls, D. Ho, X. Sun, K. Cheng, S. Sun, Nano Lett. 11 (2011)
1641.
[27] H. Borchert, E.V. Shevchenko, A. Robert, I. Mekis, A. Kornowski, G. Grübel, H.
Weller, Langmuir 21 (2005) 1931.
[28] K. Kandel, C. Frederickson, E.A. Smith, Y.-J. Lee, I.I. Slowing, ACS Catal. 3 (2013)
2750.
Acknowledgments
This research was supported at the Ames Laboratory by the U.S.
Department of Energy, Office of Basic Energy Sciences. Ames Labo-
ratory is operated for the U.S. Department of Energy by Iowa State
University under Contract No. DE-AC02-07CH11358.
[29] P.-H. Phua, L. Lefort, J.A.F. Boogers, M. Tristany, J.G. de Vries, Chem. Commun.
(2009) 3747.
[30] W. Rachmady, M.A. Vannice, J. Catal. 208 (2002) 158.
[31] E.J. Grootendorst, R. Pestman, R.M. Koster, V. Ponec, J. Catal. 148 (1994) 261.
[32] R. Pestman, R.M. Koster, E. Boellaard, A.M. van der Kraan, V. Ponec, J. Catal. 174
(1998) 142.
Appendix A. Supplementary material
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.jcat.2014.04.009.
[33] T. Zaki, J. Colloid Interface Sci. 284 (2005) 606.
[34] J.C. Kuriacose, S.S. Jewur, J. Catal. 50 (1977) 330.
[35] R. Pestman, R.M. Koster, J.A.Z. Pieterse, V. Ponec, J. Catal. 168 (1997) 255.
[36] C. Doornkamp, V. Ponec, J. Mol. Catal. A 162 (2000) 19.
[37] W. Rachmady, M.A. Vannice, J. Catal. 208 (2002) 170.
[38] A. Cabot, V.F. Puntes, E. Shevchenko, Y. Yin, L. Balcells, M.A. Marcus, S.M.
Hughes, A.P. Alivisatos, J. Am. Chem. Soc. 129 (2007) 10358.
[39] Y.-P. Sun, X.-Q. Li, J. Cao, W.-X. Zhang, H.P. Wang, Adv. Colloid Interface Sci. 120
(2006) 47.
References
[1] R. Xing, A.V. Subrahmanyam, H. Olcay, W. Qi, G.P. van Walsum, H. Pendse, G.W.
Huber, Green Chem. 12 (2010) 1933.
[2] S. Lestari, P. Mäki-Arvela, J. Beltramini, G.Q.M. Lu, D.Y. Murzin, ChemSusChem
2 (2009) 1109.
[3] E. Molina Grima, E.H. Belarbi, F.G. Acién Fernández, A. Robles Medina, Y. Chisti,
Biotechnol. Adv. 20 (2003) 491.
[40] B. Donnis, R. Egeberg, P. Blom, K. Knudsen, Top. Catal. 52 (2009) 229.
[4] Y.C. Sharma, B. Singh, J. Korstad, Green Chem. 13 (2011) 2993.