Deoxygenation of octanoic acid catalyzed by hollow spherical Ni/ZrO2

Article abstract of DOI:10.1016/j.apcata.2016.10.014

A series of Ni located on hollow spherical ZrO₂ catalysts was prepared, and their catalytic performances for octanoic acid deoxygenation were investigated. The ZrO₂ hollow spheres saturated by the water phase served as a nanoreactor that captured octanoic acid to react with Ni located inside a hollow spherical ZrO₂. The complete conversion of octanoic acid deoxygenation could be achieved by migration of reactants and products in the water-oil biphasic phase. Oxygen vacancies on ZrO₂ that resulted from calcination temperature improved the catalytic activity of deoxygenation. Propionic acid was taken as a model compound instead of octanoic acid in view of their same functional group (carboxyl), to study the mechanisms of hydrodeoxygenation, decarbonylation, and decarboxylation through density functional theory of calculations. The main reaction route was decarboxylation which was consistent with the experimental results.

Full text of DOI:10.1016/j.apcata.2016.10.014

Accepted Manuscript
Title: Deoxygenation of Octanoic Acid Catalyzed by Hollow
Spherical Ni/ZrO2
Author: Hao Chen Yulong Wu Suitao Qi Yu Chen Mingde
Yang
PII:
S0926-860X(16)30509-9
DOI:
Reference:
http://dx.doi.org/doi:10.1016/j.apcata.2016.10.014
APCATA 16030
To appear in:
Applied Catalysis A: General
Received date:
Revised date:
Accepted date:
30-6-2016
7-10-2016
17-10-2016
Please cite this article as: Hao Chen, Yulong Wu, Suitao Qi, Yu Chen, Mingde Yang,
Deoxygenation of Octanoic Acid Catalyzed by Hollow Spherical Ni/ZrO2, Applied
Catalysis A, General http://dx.doi.org/10.1016/j.apcata.2016.10.014
This is a PDF file of an unedited manuscript that has been accepted for publication.
As a service to our customers we are providing this early version of the manuscript.
The manuscript will undergo copyediting, typesetting, and review of the resulting proof
before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that
apply to the journal pertain.

Deoxygenation of Octanoic Acid Catalyzed by Hollow Spherical Ni/ZrO2
Hao Chen,[a] Yulong Wu,*[b] Suitao Qi,*[a] Yu Chen [b] and Mingde Yang[b]
[a] Department of Chemical Engineering, School of Chemical Engineering and Technology, Xi’an Jiaotong University,
Xi’an 710049 (P.R. China).
[b] Institute of Nuclear and New Energy Technology, Tsinghua University, Beijing 100084 (PR China)
E-mail: wylong@tsinghua.edu.cn(Yulong Wu); suitaoqi@mail.xjtu.edu.cn(Suitao Qi)
Graphical abstract
Highlights
1: Deoxygenation of octanoic acid in oil-water system.
2: A noval catalyst with Ni located inside of hollow spherical ZrO2 was prepared.
3: The hollow spherical ZrO2 performs the role of nanoreactor.
4: Complete conversion of octanoic acid deoxygenation could be achieved.
5: The mechanism of deoxygenation for carboxyl function group was studied by DFT.
1

Abstract: A series of Ni located on hollow spherical ZrO2 catalysts was prepared, and their catalytic
performances for octanoic acid deoxygenation were investigated. The ZrO2 hollow spheres saturated by
the water phase served as a nanoreactor that captured octanoic acid to react with Ni located inside a
hollow spherical ZrO2. The complete conversion of octanoic acid deoxygenation could be achieved by
migration of reactants and products in the water-oil biphasic phase. Oxygen vacancies on ZrO2 that
resulted from calcination temperature improved the catalytic activity of deoxygenation. Propionic acid
was taken as a model compound instead of octanoic acid in view of their same functional group
(carboxyl), to study the mechanisms of hydrodeoxygenation, decarbonylation, and decarboxylation
through density functional theory of calculations. The main reaction route was decarboxylation which
was consistent with the experimental results.
Keywords: Hollow spherical Ni/ZrO2 catalyst, Deoxygenation of biooil, Octanoic acid, Density
functional theory, Reaction mechanism.
1. Introduction
Fossil fuel depletion has triggered concerns about environmental problems and national energy
security. Thus, interest in seeking renewable and environmental-friendly alternative fuels has increased.
Fossil resources generally take over millions of years to form and are non-renewable; hence, replacing
fossil resources with biomass is important to produce various platform molecules and liquid fuels[1-4].
Bio-oils produced from biomass resources are a promising source of renewable energy[5-7]. However,
the direct use of bio-oil as liquid fuel is currently unfeasible because of its high oxygen content,
instability, low heating value, serious chemical instability, and corrosiveness. Upgrading bio-oil to
liquid fuels with low oxygen contents and high heating values provides significant environmental,
economic, and strategic advantages for future prospects[8-10].
Catalytic deoxygenation is the most important and promising method for crude bio-oil upgrading[11].
Crude bio-oil contains a large amount of water (10–30 wt%), and bio-oil upgrading produces
2

hydrophilic or hydrophobic components[12, 13]. Thus, this reaction system is a compound liquid with
two immiscible phases (water and oil). Diffusion of reactants, intermediates, and products between the
water and oil phases reduces the activity and selectivity of catalysts employed in deoxygenation. To
resolve this problem, a recent study has applied homogeneous and emulsion catalysts in bio-oil
deoxygenation. Homogeneous catalysts display excellent activity[14] but are difficult to separate from
final product mixtures. Emulsion-stabilizing solids with additional catalytic functional groups facilitate
a broad range of reactions. Reactions in Pickering emulsion catalysis are performed in a biphasic
mixture of two immiscible solvents, and emulsifiers are added to enhance the interfacial surface area
and thus promote the mass transfer of molecules between different phases[15-18]. However, catalysts
stabilized by the effect of tension in Pickering emulsions are unstable at the interface under vigorous
stirring[19]. In biphasic phase-involved deoxygenation reactions, the dispersion of catalysts in the
medium must be enhanced to ensure efficient contact between the reactants and the active metal sites.
Stabilizing catalysts on a permeable shell that encloses water in the oil phase allows the reactants and
products to migrate through the permeable shell from the water to oil phase and the catalysts to
deoxygenate crude bio-oil.
In our previous work[20], octanoic acid can be used as transportation fuel to replace gasoline[21-23].
Octanoic acid was chosen as the model to investigate deoxygenation catalyzed by Ni/ZrO2. Ni/ZrO2
exhibited good catalytic performance for the deoxygenation of octanoic acid in the oil phase. In the
present work, Ni inside a hollow spherical ZrO2 (ZrO2-HS) was prepared as a potential candidate for the
catalytic deoxygenation of octanoic acids to obtain high-grade biofuels. As Scheme 1 shows, octanoic
acid is slightly soluble in the hot water phase from the oil phase through the permeable ZrO2 shell. Then,
the octanoic acid is catalytically deoxygenated by contacting with the active site on the Ni surface. After
deoxygenation, the products that are insoluble in the water phase but easy to dissolve in the oil phase,
such as octane or heptane, transfer to the oil phase from the water phase. Deoxygenation is promoted in
the nanoreactor of a hollow spherical shell; the reactant is captured by the hollow spherical shell first
and then reacts on the Ni particles inside the ZrO2 shell. Although the solubility of octanoic acid and
3

intermediates (octanol) is low the in water phase, the motivation of deoxygenation could continuously
proceed through product-phase migration from the water to oil phase. To our knowledge, this study is
the first to deoxygenate octanoic acid in a water–oil biphase. The effect of oxygen vacancies results
from calcination temperature, and the reaction routes of deoxygenation are studied. Deoxygenation
includes three routes: hydrodeoxygenation (HDO), decarbonylation (DCN), and decarboxylation (DCX).
To understand the effect of oxygen vacancies on the ZrO2 surface and to investigate the main reaction
route of deoxygenation, density functional theory (DFT) is used to provide detailed information on the
properties of the catalyst[24-28]. Long-chain carboxylic acids cannot easily be studied computationally;
thus, the mechanisms of the three routes are investigated using propanoic acid instead of octanoic acid
on the Ni surface because the former is a significant model reactant for various long-chain carboxylic
acids[24].
2. Experimental and Computational Methods
2.1. Catalyst preparation
Ni located inside hollow spherical ZrO2 (Ni/ZrO2-HS) was prepared as the following step. The
synthetic methods are shown in scheme 2.
Firstly, SiO2 template was prepared by Stöber Method[29]. In a typical synthesis, 10 mL of ammonia
(28 wt%), 20 mL of deionized water and 70 mL of ethanol were mixed and transferred to a three-necked
flask. Then, 10 mL of tetraethoxysilicate (TEOS) were dissolved in 90 mL of ethanol, then; the solution
was added into the former mixed solution and vigorously stirred at 40oC for 2.5h. The obtained white
suspension was separated by centrifugation and washed four times with ethanol, and then dried in air
and the spherical SiO2 template was obtained.
Secondly, Ni(NO3)2 with desired Ni loadings was loaded on spherical SiO2 by incipient wetness
impregnation. Then the precursor was dried at 110oC for 1h and calcined at 450oC for 3h.
Thirdly, 1.50 g of the as-prepared Ni/SiO2 nanospheres were ultrasonically dispersed in 100 mL of
propyl alcohol, and then 7g zirconium propoxide were dissolved in 100 mL of propyl alcohol and
subsequently added to the former mixed solution, vigorously stirred at room temperature for 3 h. The
4

solid were separated by centrifugation and washed four times with deionized water, and then dried in an
oven at 80oC for 6 h.
Finally, the above solid precursor was dispersed in 35 mL of 2.00M NaOH solution, and then the
obtained suspension was transferred to 50 mL Teflon-sealed autoclave and maintained at 80oC for 4 h.
The resulting products were washed four times with deionized water and dried at 60oC under vacuum
overnight. At last, the sample was calcined and reduced by H2 at 450oC for 2h respectively to obtain the
Ni/ZrO2-HS. In sum, the synthetic route: (Ni loading→ZrO2 coating→SiO2 removal).
2.2. Catalyst characterization
The surface areas were calculated by the Brunauer-Emmett-Teller (BET) method. Transmission
electron microscope (TEM) was carried out on a FEI tecnai F20 instrument. The FT-IR absorption
spectra were obtained with a Bruker model Tensor 27 equipment in the range of 4000-400cm-1, using
KBr powders. Fourier transform infrared spectroscopy of adsorbed pyridine on the catalysts were
recorded on a NEXUS 670 FTIR spectrometer. Electron Paramagnetic Resonance (EPR) spectra were
run on a X-band CW-EPR Bruker EMX spectrometer equipped with a cylindrical cavity operating at
100 kHz field modulation. Bulk elemental composition was determined by inductively coupled plasma
(ICP) atomic emission spectrometry. Powder X-ray diffraction (XRD) patterns were recorded on a
Bruker D8 Advance X-ray diffractometer using Cu Kα radiation (λ= 0.1541 nm, 36 kV, 20 mA,
scanning step = 2omin−1). Data were recorded in the 2 range 5~80o. The average crystallite size was
calculated using the following well-known Scherrer formula[30]:
k
S 
 cos
where S is the average size(nm) of Ni particles; λ is the X-ray wavelength in nanometer (nm) (0.154 nm
for Cu-Kα); β is the peak width of the diffraction peak profile at the half maximum height (FWHM)
resulting from small crystallite size in radians and k is a constant related to crystallite shape, normally
taken as 0.9, with an average = 1 for polymers.
2.3. Catalytic activity tests
5

Octanoic acid deoxygenation was carried out in a 150 mL autoclave in batch mode. 0.5 g catalysts
presoaked by water in vacuum until hollow spheres were saturated with water. Then the catalysts were
highly dispersed into 50 mL decane solution by ultrasound with 1 g octanoic acid and then the
compound was moved into the autoclave. The reactor was purged three times with H2 and then
pressurized with H2 to 3 MPa with constant stir rate (300 rpm). Then the reactor was heated to the
desired reaction temperature, and maintained for 12h. After the autoclave was cooled down to room
temperature, the reaction product were analyzed on gas chromatograph (Agilent GC6820) using a
GsBP-Inowax (30 m × 0.32 mm × 0.25 m) capillary column with FID. The equation of conversion of
octanoic acid and yield of each product are as follows:
M
C  (1 R )100%
MI
M
Y  P 100%
MI
where C is the conversion of octanoic acid; MI is the mole of the starting octanoic acid; MR is the moles
of remaining octanoic acid; Y is the yield of each product; MP is the moles of each product.
2.4. Computational methods
The DFT calculations were conducted using the program package DMol3 in the Materials Studio
software (version 6.1). Generalized gradient approximation (GGA) using exchange-correlation
functional developed by PW91 was employed and the localized double-numerical basis setting with
polarization functions (DNP) were used. The core electrons of the metal atoms were treated using the
effective core potentials (ECP). The convergence criteria for configuration optimization was set to the
tolerance for SCF, energy, maximum force, and a maximum displacement of 1.0e−5 Ha, 2.0e−5 Ha,
0.004 Ha/Å and 0.005 Å, respectively. Throughout calculations, a Fermi smearing of 0.005 Ha was used
to improve computational performance and all of the computations were performed with
spin-polarization.
The periodic surfaces in the study were modeled using four-layer slabs inside a 3 × 3 unit cell that is
separated from its neighboring images by a 12 Å width vacuum in the direction perpendicular to the
6

surface. The Brillouin-zone integrations were performed using a 3 × 3 × 1 k-point grid. The geometries
of the adsorbates and the top two layers of the metal surface were allowed to relax while the bottom two
layers were frozen during the geometry optimizations[31].
The mechanisms for deoxygenation on Ni surfaces were explored with the transition states
determined by using the complete LST/QST method, where LST and QST stand for linear synchronous
transit and quadratic synchronous transit, respectively. The transition state structures were determined
accordingly and proved to have only single imaginary frequency that corresponds to the reaction
pathway from the reactant to the product[32, 33].
The adsorption energy is calculated using the following formula:
Eads  EA/slab  Eslab  EA(g)
where EA/slab is the total energy of the slab with an adsorbate bound to it; Eslab is the energy of the clean
metal slab; EA(g) is the energy of the adsorbate in gas-phase.
The energies of activation barrier(Ea) and energies of reaction(Erxn) are calculated using the following
formula:
Ea  ETS  EIS
Erxn  EFS  EIS
where ETS is the energy of the transition state; EIS is the energy of the initial state; EFS is the energy of
the final state.
3. Results and Discussion
3.1.Catalyst characterization
Five catalysts were prepared, and their characterization parameters are listed in Table 1. 5Ni/Z450,
10Ni/Z450, and 15Ni/Z450 refer to the different Ni contents inside the Z450 support (hollow spherical
ZrO2 calcined at 450 °C and marked as Z450). 10Ni/Z750 is the Ni content inside the Z450 support
(hollow spherical ZrO2 calcined at 750 °C and marked as Z750). 10Ni′/Z450 is the Ni content outside
Z450. 10Ni′/Z450 was prepared similarly to 10Ni/Z450, but the sequence of Ni impregnation differed in
these two catalysts (see SI). The BET surface area of ZrO2 was found decreasing upon increasing
7

calcination temperature from 450 to 750 °C and the surface area values being 324.9 and 133.9 m2g-1 for
Z450 and Z750 catalysts respectively. The surface areas in the present work were greater than that in
Peng’ s work(98 m2g-1)[34] and our previous work(5.046 m2g-1)[20], indicating that a hollow spherical
structure could greatly promote the surface area of ZrO2. The Ni content of each catalyst was confirmed
by ICP analysis.
The different ZrO2 phases affected by calcination temperature were compared by calcining two
groups of hollow spherical ZrO2 at 450 °C (5Ni/Z450, 10Ni/Z450, 15Ni/Z450, and 10Ni′/Z450) and
750 °C (10Ni/Z750). The five catalysts were characterized via XRD (Fig. 1). XRD patterns show that
10Ni/Z750 has diffraction peaks at 2θ=30.6°, 35.3°, 50.9°, 60.5°, 73.5°, and 74.3°, which are
characteristic peaks of tetragonal ZrO2. In particular, 73.5° and 74.3° are the respective characteristic
peaks of tetragonal (004) and (400), which have been extensively used to distinguish the tetragonal
from the cubic structure in ZrO2[35, 36]. The XRD patterns of 5Ni/Z450, 10Ni/Z450, 15Ni/Z450, and
10Ni′/Z450 show no characteristic peaks of ZrO2, indicating the amorphous structure of ZrO2. Thus, the
crystal is affected by different calcination temperatures. The diffraction peaks of Ni/Z450 at 2θ=44.6°,
51.9°, and 76.8° are characteristic peaks of Ni (111), (200), and (220), respectively. Other crystalline
phases are not detected in Ni/Z450, indicating the complete reduction of Ni nanoparticles. In addition,
Ni (111) is the main crystal growth in the Ni particles. Narrow and intensive diffraction peaks of Ni in
15Ni/Z450 indicate a well-crystallized structure. The size of Ni particles was calculated by Scherrer
equation, and the results are listed in Table 1. The size of the Ni particles in 5Ni/Z450, 10Ni/Z450,
10Ni′/Z450 and 10Ni/Z750 is about 10 nm, but the size of the Ni particles in 15Ni/Z450 is about 30 nm.
The size and dispersion of the Ni particles could also be directly observed in the TEM images. The Ni
particles of the catalysts disperse well on the supports, except for 15Ni/Z450, in which the Ni particles
clearly aggregate.
The nitrogen adsorption–desorption isotherms of the five catalysts were representative type IV curves
with the sharp capillary condensation steps for both catalysts (Fig. S1a)[37]. The relative pressures(P/P0)
of type H2 hysteresis loop are higher than 0.9, which indicates that the pores on the ZrO2 shell form by
8

particle packing[38]. As shown in Fig. S1b, the four catalysts supported by Z450 possess narrow
pore-size distributions centered at about 4.5 nm, whereas the catalyst supported by Z750 displays a
pore-size distribution centered at about 2.0 nm. The crystal transformation of Z750 results in the
decrease of its pore size, and its pore size at 40–60 nm results from the rupture of the spherical shell at a
high calcination temperature.
Z450 and Z750 differ in morphology, pore size, and specific surface area. Fig. 2 shows further
comparison of 10Ni/Z450 and 10Ni/Z750. Figs. 2(a)–2(c) show TEM images of 10Ni/Z450. The hollow
spherical structure of Z450 is shown in Fig. 2a. As shown in Fig. 2b, the shell thickness and diameter of
Z450 are about 25 and 400 nm, respectively. Ni particles about 12 nm are dispersed inlay in the ZrO2
shell. Fig. 2c, the amplification of the area in the white dashed circle of Fig. 2b, shows that Ni particles
are embedded inside the ZrO2 shell. The interplanar spacing of the Ni particles is 0.202 nm in the
HRTEM image, which shows that the crystal face is Ni(111)[39, 40]. Compared with 10Ni/Z450, a
certain amount of spherical shell of 10Ni/Z750 is ruptured (Fig. 2d), which is consistent with the
pore-size distributions. Fig. 2f, the amplification of the area in the white dashed circle of Fig. 2e, shows
Ni particles located inside the ZrO2 shell. In addition, ZrO2 of Z750 in Fig. 2f is more compact than that
of Z450 in Fig. 2c, which implies that Z750 possesses higher crystallinity and smaller pore size than
Z450.
As shown in Fig. 3, the EPR spectra of Z450 and Z750 show signals that indicate the presence of
intrinsic defects in the prepared material. The symmetric signal, resonating at g=2.0024, represents the
free electron—an electron trapped in an oxygen vacancy[41]. Compared with Z750, Z450 signal grows
in intensity, indicating more oxygen vacancies are present in Z450 than in Z750. The Py-FTIR studies
of the 10Ni/Z750 and 10Ni/Z450 catalysts are shown in Fig. S2. The bands at 1445 and 1536 cm-1 are
associated with Lewis and Brönsted acid sites, respectively[42]. Compared with 10Ni/Z750, 10Ni/Z450
displays a stronger relative intensity of Lewis acid sites, indicating that more oxygen vacancies are
present in 10Ni/Z450 than in 10Ni/Z750[43]. DFT calculation was applied to demonstrate that oxide
9

vacancies could promote R–COO adsorbance on the ZrO2 surface. This phenomenon is discussed
thoroughly below.
3.2.Octanoic acid deoxygenation over the Ni/ZrO2-HS catalyst
3.2.1. Octanoic acid deoxygenation at different temperatures
Temperature is an important parameter in deoxygenation reaction. Considering the small amount of
water (about 6%), we only detected the products in the oil phase. As shown in Table 2, the conversion
of octanoic acid catalytic deoxygenation over 10Ni/Z450 at different reaction temperatures shows that
the conversion is considerably low at 180 °C but dramatically increases with temperature. The
conversion reaches 98.3%, and the heptane yield reaches 87.4% at 240 °C. The yield of octane increases
accompanies with the decrease of octanol from 180°C to 260°C. It indicates octanol prefer to dehydrate
to octane at higher temperature, which attributed to that HDO prevailed at high temperatures. Thus,
energy barriers for the processes leading to HDO should be higher than the energy barriers of the steps
that lead to DCX and DCN over the Ni/Z450 catalyst. The effect of different Ni loadings was studied,
and results show that the performance of 10Ni/Z450 is the best among the tested catalysts. The
conversion of octanoic acid catalyzed by 5Ni/Z450 also reaches 98%, but the heptane and octane yields
only reach 79.2% and 7.1%, respectively. However, the conversion of octanoic acid over 15Ni/Z450
reaches 95.5%, which is lower than that over 10Ni/Z450. 10Ni/Z450 was selected for a follow-up study.
As previously mentioned, different calcination temperatures result in different crystal systems. To
investigate the effect of calcination temperature on octanoic acid deoxygenation, 10Ni/Z750 catalysts
were also prepared for octanoic acid deoxygenation. Compared with 10Ni/Z450, octanoic acid
conversion catalyzed by 10Ni/Z750 only reaches 63.1%. In addition, octanol yield reaches 8.5%, which
is higher than that in the 10Ni/Z450 system. O-vacancies decrease against reactant adsorption on the
catalyst, which is proven by DFT calculation. Moreover, the cracked hollow spherical shells do not
encapsulate reactants effectively; thus, the reactants hardly contact active sites. Finally, 10 wt% Ni
loaded on commercial nanoZrO2 (10Ni/c-ZrO2) was also tested. Octanoic acid conversion only reaches
53.2%. Heptane yield reaches 23.3%, which is lower than that of 10Ni/Z450 and 10Ni/Z750. This result
10

indicates that the reactions are commendably promoted by reactant capture via the hollow spherical
shell.
3.2.2. Catalytic activity influenced by the location of Ni particles and the effect of water
In octanoic acid deoxygenation, 10Ni′/Z450 was applied to understand further the catalytic activity
influenced by Ni particles located on the ZrO2 spherical shell. The TEM images in Fig. S3 show the
difference of Ni location between the two preparation methods. 10Ni′/Z450 in Fig. S3a shows the Ni
located outside the ZrO2 spherical shell, whereas 10Ni/Z450 in Fig. S3b shows the Ni located inside the
shell. The two catalysts were applied for octanoic acid deoxygenation. The changes in the conversion
and yield of products as a function of reaction time were traced (depicted in Fig. 4). In Fig. 4a, the
complete conversion of octanoic acid is observed after reaction at 240 °C for 12 h over 10Ni/Z450.
Heptane yield increases continuously to 89.1% at 12 h, whereas octane yield grows slowly to 5.9% after
12 h. Octanol yield gradually increases to 13.9% at 6 h and then gradually decreases, accompanied by
an increase in octane yield. In Fig. 4b, the conversion of octanoic acid only reaches about 80% after 6 h
reaction over 10Ni′/Z450 and hardly increases further. Octanol yield is higher than the one catalyzed by
10Ni/Z450. In addition, octanol yield reaches about 14% but does not decrease subsequently, which
differs from that over 10Ni/Zr450. Compared with the Ni located outside the ZrO2 shell, the one inside
is easier to contact the reactant captured by the hollow spherical shell. However, octanoic acid should
overcome the mass transfer resistance across the ZrO2 shell to contact active sites on 10Ni/Z450. Thus,
the reaction rate over 10Ni/Z450 is slower than that over 10Ni′/Z450. To substantiate the effect of water
in this reaction system, 10Ni/Z450 was not water-soaked before the test. As observed in Fig. 4c, the
conversion of octanoic acid and the yield of main products under 10Ni/Z450 catalysis without water
soaking are lower than those under 10Ni/Z450 catalysis with water soaking before the test. In Fig. 4d,
octanoic acid deoxygenation is catalyzed by 10Ni′/Z450 without water soaking. The mass transfer
resistance of the ZrO2 shell is eliminated because of the Ni located outside the hollow spherical ZrO2 in
the experiments of Figs. 4b and 4d. The octanoic acid conversion and heptane yield in the experiment of
Fig. 4d are clearly lower than that in the experiment of Fig. 4b, which indicates that water intervenes
11

with the reaction mechanism in the experiment of Fig. 4b. To summarize, the reasons for deoxygenation
improvement by water participation are as follows. Water participation is conducive for reactant
transfer through the ZrO2 shell. In addition, the aqueous-phase reformation of octanoic acid may
facilitate the in situ formation of H2[44, 45]. Moreover, CO produced by DCN reaction responds with
water and then produces H2 through water–gas shift reaction. This finding suggests the possibility of a
self-sustaining process, where CO is not only a reaction product but also a key intermediate for H2
formation required by DCN[46].
3.2.3. Effect of H2 pressure on octanoic acid deoxygenation
To study the mechanism underlying the deoxygenation reaction of octanoic acid, we investigated the
influence of H2 pressures on the catalytic activity of 10Ni/Z450 (Fig. 5). For clarity, octanoic acid
conversion increases with the augmentation of H2 pressure. For H2 pressure at 0 MPa, the octanoic acid
conversion and heptane yield respectively reach 48.4% and 43.6%, whereas the octane yield is 0%. This
result illustrates that the main reaction for octanoic acid deoxygenation is DCX without H2. With the
increase in H2 pressure, the octanoic acid conversion and heptane yield significantly increase.
Meanwhile, the octane yield increases slowly, which means that a part of octanoic acid follows the
DCN reaction. When the pressure of hydrogen is further elevated to 3 MPa, a high yield of heptane is
achieved with the complete conversion of octanoic acid. When the H2 pressure exceeds 3 MPa, heptane
yield decreases while octane yield remarkably increases, which indicates a certain amount of octanoic
acid following HDO reaction. Hydrogen is required to convert octanoic acid into octanal and further
quickly into octanol; thus, H2 participation shifts the reaction following DCN and HDO, whereas
decreasing the amount of H2 has the opposite effect, which leads to DCX. To substantiate the selectivity
of DCN and HDO reactions influenced by H2 pressures, we studied octanal, the intermediate of octanoic
acid deoxygenation, as the initial reactant for deoxygenation. As shown in Fig. S4a, octanal conversion
under each H2 pressure is similar. The heptane yield (Fig. S4b) decreases while the octane yield (Fig.
S4d) increases with the increase in H2 pressure, which indicates that H2 promotes the selectivity of
HDO. In Fig. S4c, the octanol yield initially increases and then decreases, which demonstrates that
12

octanal is converted into octanol and further into octane under H2. Furthermore, octanol as the initial
reactant is converted into an independent experiment on 10Ni/Z450 under conditions identical to those
in octanoic acid conversion. In Fig. S5, the octane yield increases almost linearly with octanol
conversion, which is formed via a tandem dehydration–hydrogenation route.
Combining these experiments allows us to formulate the overall reaction pathway for octanoic acid
conversion (Scheme 3). The process involves the initial adsorption of octanoic acid to generate the
carboxylate group on the surface of the ZrO2 support before diffusing to the Ni surface. The white
arrow demonstrates the DCX pathway that octanoic acid directly decarboxylates to generate heptane. In
the DCN pathway depicted by yellow arrows, octanoic acid initially hydrogenates and dehydrates to
generate octanal and then decarbonylates to generate heptane. In addition, octanal, which follows the
HDO pathway in red arrows, initially hydrogenates to octanol and then dehydrates to generate octane.
The mechanism underlying the Ni/ZrO2-catalyzed deoxygenation of the carboxylate group is further
investigated by DFT.
3.2.4. Stability tests of the 10Ni/Z450 catalyst
As shown in in Fig. 6, the stability of the 10Ni/Z450 catalyst was evaluated, and 15 successive times
were performed under the same reaction conditions. During the first six cycles, the octanoic acid
conversion and the heptane yield remain constant which is about 98% and 87%, respectively. The Ni
nanoparticles located inside the ZrO2 shell could protect the catalytic active component[47-49].
However, the octanoic acid conversion and heptane yield sharply decrease after nine cycles. Compared
with the fresh catalysts (Fig. S6a), the spherical shell partly ruptures, and the shell thickness gradually
thins after 10 cycles (Fig. S6b). After 15 cycles, the structure of the hollow sphere is completely
destroyed (Fig. S6c). This catalyst inactivation is irreversible, and the stability of the frame structure of
catalysts will be studied in future works.
3.3.DFT calculations
As shown in Scheme 3, a possible mechanism is supposed for octanoic acid on 10Ni/Z450. In this
section, the adsorption of the carboxyl function group influenced by oxygen vacancies on the ZrO2
13

surface and the deoxygenation mechanism of the carboxyl function group on the Ni surface were
simulated by DFT. The main reaction route was computationally studied to verify with the experiment
results. Considering efficiency and economy, propionic acid instead of octanoic acid was studied for
deoxygenation. The deoxygenation mechanism of propionic acid could analogically substitute that of
octanoic acid because these two acids have the same functional group.
3.3.1. Adsorption of carboxyl influenced by oxygen vacancies on supports
To explore the influence between COO and oxygen vacancies, we investigated propionic acid
adsorbed on the ZrO2 surface with O vacancies. For this purpose, 27 surfaces of monoclinic
ZrO2(m-ZrO2), cubic ZrO2(c-ZrO2), and tetragonal ZrO2(t-ZrO2) shown in Fig. S7 were modeled
without O vacancies (ZrO2-0), with one O vacancy (ZrO2-1), and with two O vacancies (ZrO2-2). The
adsorption energies of C2H5COO on these surfaces are listed in Table 3, and the geometric structures
are shown in Fig. S8. According to adsorption energies, C2H5COO adsorbs on t-ZrO2(100), t-ZrO2(110),
t-ZrO2(111), m-ZrO2(110), and c-ZrO2(110) surfaces without oxygen vacancies. The adsorption
energies of C2H5COO on the ZrO2 surface with oxygen vacancies are lower than that those on the ZrO2
surface without oxygen vacancies. A growing number of oxygen vacancies on the ZrO2 surface results
in lower energy of C2H5COO adsorption. This result indicates that oxygen vacancies are conducive for
C2H5COO to adsorb on ZrO2 and that C2H5COO further contacts with the active sites of Ni on ZrO2.
3.3.2. Mechanism of deoxygenation on Ni surface
According to XRD results, Ni(111) is the main style of crystal growth on the Ni particles, and it is the
most stable surface among Ni(100), Ni(110), and Ni(111)[50]; thus, the following study was discussed
on Ni(111). The interplanar distance of Ni(111), calculated by DFT with 0.204 nm, is close to the
experiment results of TEM with 0.202 nm, which indicates that the calculation method in this work is
reasonable for DFT study. Before addressing the deoxygenation routes for propionic acid, we first
investigated the adsorption of key reaction species involved in the deoxygenation reaction on the
Ni(111) surface. The adsorption geometries of the reactants, products, and possible intermediates and
14

the adsorption energies of each species are shown in Fig. 7. The parameters, distance between adsorbate
and Ni surface, are listed in Table S1.
The entire process consists of 10 elementary steps, which are defined as DCX, DCN, HDO, and
hydrogenation. Each elementary reaction step investigated for deoxygenation mechanism of propanoate
will be discussed thoroughly. All energies of activation barrier (Ea) and reaction are summarized in
Table 4. The related transition states are illustrated in Fig. 8, and the calculated potential energy profiles
of deoxygenation for propanoate on the Ni catalyst models are shown in Fig. 9.
For DCX, CO2 abstraction is a one-step reaction from CH3CH2COO. The geometrical structure of
CH3CH2COO in Fig. 7(1) is the initial state for the elementary reaction. The α-C−β-C cracks to form
CO2, which directly adsorbs on the Ni surface. Transition states (TS1) have a C−COO distance of 2.144
Å in Fig. 8(1), and this reaction requires the system to overcome the activation energy of 1.36 eV. The
elementary reaction is an endothermic process with a reaction energy of 0.59 eV.
DCN is subdivided into two elementary reactions. In the first step, the geometrical structure of
CH3CH2COO in Fig. 7(2) is the initial state for the elementary reaction. In TS2 (Fig. 8(2)), one O-atom
cracks and is close to the fcc hollow of the Ni surface. The α-C closes to the top site of Ni with a C-Ni
bond length of 3.141 Å, which has a C−O distance of 2.011 Å and an Ea of 1.12 eV. This elementary
reaction is an endothermic process with a reaction energy of 0.29 eV. In the second step, the bond of
α-C−β-C cracks and the C atom of CO becomes close to the fcc hollow site. TS3 (Fig. 8(3)) has a
α-C−β-C distance of 1.788 Å, which is stretched from 1.473 Å. The elementary reaction is an
endothermic process with a reaction energy of 0.82 eV, and it is the rate-determining step of DCN with
an activation energy of 1.47 eV.
HDO is subdivided into four elementary reactions. In the first step, H addition to the α-C of
propanoate forms C2H5-CHOO with 2.03 eV Ea, which is a high-activation energy barrier and the
rate-determining step of HDO. This elementary reaction is an endothermic process with a reaction
energy of 0.90 eV. In TS4 for this step (Fig. 8(4)), the H atom from the top site of Ni is close to the α-C
of C2H5-CHOO, and the C-H distance is 1.695 Å. This structure is unstable so it soon transfers to the
15

second step wherein one O-atom deoxygenates from C2H5-CHOO to form O and C2H5-CHO. Thus, in
the second step, TS5 (Fig. 8(5)) has a considerably low activation barrier with 0.28 eV, and it is an
exothermic process with a reaction energy of −0.13 eV. The O atom breaks from α-C and adsorbs on the
fcc site of the Ni surface, in which the α-C−O distance of TS5 is 1.938 Å. The third step is when one H
atom closes to C2H5-CHO and forms C2H5-CH2O. TS6 (Fig. 8(6)) has a α-C−H distance of 1.580 Å and
0.57 eV Ea. It is also an exothermic process with a reaction energy of −0.38 eV. The fourth step is when
another O atom cracks and forms C2H5-CH2 and O. TS7 (Fig. 8(7)) has a C−O distance of 1.794 Å, and
this process with 1.04 eV Ea is endothermic with a reaction energy of 0.48 eV.
For hydrogenation, C2H5, C2H5-CH2, and C2H5-CH2O are hydrogenated to form C2H6, C2H5-CH3,
and C2H5-CH2OH, respectively. In TS8 (Fig. 8(8)), the H and C atoms of C2H5 bind to the same Ni
atom with a C−H bond length of 1.633Å and an Ea of 0.71 eV. In TS9 (Fig. 8(9)), the process is similar
to C2H5-CH2 hydrogenation, whose C−H distance is 1.589 Å and Ea is 0.85 eV. For C2H5-CH2O
hydrogenation, TS10 (Fig. 8(10)) has an O−H distance of 1.47 6Å and an Ea of 0.39 eV. The three
processes are all endothermic with reaction energies of 0.37, 0.39, and 0.14 eV, respectively.
As aforementioned, the selectivity of octanoic acid conversion in the presence of hydrogen over the
Ni catalysts is determined by the dominant reaction pathway under reaction conditions. Experimental
results show that DCX and DCN are the main reaction, and HDO is the side reaction. However, these
results do not distinguish which is the dominant reaction. According to the calculated results, the
activation barrier of the rate-determining step in DCX with 1.36 eV is lower than that in DCN with 1.47
eV and HDO with 2.03 eV in the three reactions, which indicates that DCX on the Ni(111) surface is
kinetically more favorable than DCN and HDO.
4. Conclusion
In this work, a novel approach was disclosed for the transformation of octanoic acid into alkanes over
the Ni catalysts supported by hollow spherical ZrO2. Ni catalysts stabilized inside a hollow spherical
ZrO2 wraps the water phase catalytically, deoxygenating octanoic acid to alkanes. By reactant capture
and product removal, octanoic acid deoxygenation continuously occurs until it nearly completes the
16

conversion in the water phase despite the low solubility of octanoic acid. In addition, the oxygen
vacancies on ZrO2 are an important factor in promoting octanoic acid deoxygenation. In DFT
calculation, propionic acid instead of octanoic acid was applied to study the effect of oxygen vacancies
and deoxygenation mechanism. DFT calculation verifies that the carboxyl group is favorable to adsorb
on ZrO2 with oxygen vacancies, which is conducive for further deoxygenation on the Ni surface. DFT
results also show that DCX is the dominant route resulting in the main product as heptane instead of
octane. Our results highlight the applications of solid catalysts localized at the interface between two
liquid phases, which will facilitate a broad range of reactions.
Acknowledgements
This work was supported by the National Natural Science Foundation of China (No. 21576155, No.
21376140 and No.21176142), Program for Chang jiang Scholars and Innovative Research Team in
University （ IRT13026 ） and Program for New Century Excellent Talents in University (No.
NCET-12-0308). The authors thank Computer Network Information Center in Chinese Academy of
Science for access to the software of Materials Studio.
References
[1] J. Payormhorm, K. Kangvansaichol, P. Reubroycharoen, P. Kuchonthara, N. Hinchiranan,
Bioresource Technol. 139 (2013) 128-135.
[2] M.Z. Jacobson, Energ. Environ. Sci. 2 (2009) 148-173.
[3] P. Langan, S. Gnanakaran, K.D. Rector, N. Pawley, D.T. Fox, D.W. Cho, K.E. Hammel, Energ.
Environ. Sci. 4 (2011) 3820-3833.
[4] M.S. Mettler, D.G. Vlachos, P.J. Dauenhauer, Energ. Environ. Sci. 5 (2012) 7797-7809.
[5] Y. Tang, W. Yu, L. Mo, H. Lou, X. Zheng, Energ. Fuel 22 (2008) 3484-3488.
[6] J. Yang, M. Xu, X. Zhang, Q. Hu, M. Sommerfeld, Y. Chen, Bioresource Technol. 102 (2011)
159-165.
[7] Z. Zhang, M. Tang, J. Chen, Appl. Surf. Sci. 360 (2016) 353-364.
[8] L. Petrus, M.A. Noordermeer, Green Chem. 8 (2006) 861-867.
[9] E. Butler, G. Devlin, D. Meier, K. McDonnell, Renew. Sust. Energ. Rev. 15 (2011) 4171-4186.
[10]T.P. Vispute, H. Zhang, A. Sanna, R. Xiao, G.W. Huber, Science 330 (2010) 1222-1227.
[11]D.M. Alonso, J.Q. Bond, J.A. Dumesic, Green Chem. 12 (2010) 1493-1513.
[12]J. He, C. Zhao, J.A. Lercher, J. Am. Chem. Soc. 134 (2012) 20768-20775.
[13]S. Czernik, A. Bridgwater, Energ. Fuel 18 (2004) 590-598.
[14]F. Mahfud, S. Bussemaker, B. Kooi, G. Ten Brink, H. Heeres, J. Mol. Catal. A 277 (2007) 127-136.
[15]X. Yang, Y. Liang, X. Zhao, Y. Song, L. Hu, X. Wang, Z. Wang, J. Qiu, RSC Adv. 4 (2014)
31932-31936.
[16]Z. Zhu, H. Tan, J. Wang, S. Yu, K. Zhou, Green Chem. 16 (2014) 2636-2643.
[17]S. Crossley, J. Faria, M. Shen, D.E. Resasco, Science 327 (2010) 68-72.
17

[18]D. Ni, L. Wang, Y. Sun, Z. Guan, S. Yang, K. Zhou, Angew. Chem. Int. Edit. 49 (2010)
4223-4227.
[19]I. Masalova, E. Kharatyan, A.Y. Malkin, J. Disper. Sci. Technol. 34 (2013) 1074-1078.
[20]Y. Duan, Y. Wu, Q. Zhang, R. Ding, Y. Chen, J. Liu, M. Yang, J. Mol. Catal A 398 (2015) 72-78.
[21]S. Harnos, G. Onyestyák, J. Valyon, Appl. Catal. A 439 (2012) 31-40.
[22]K. Sun, T.C. Schulz, S.T. Thompson, H.H. Lamb, Catal. Today 269 (2016) 93-102.
[23]G. Onyestyák, S. Harnos, M. Štolcová, A. Kaszonyi, D. Kalló, Catal. Commun. 40 (2013) 32-36.
[24]J. Lu, S. Behtash, A. Heyden, J. Phys. Chem. C 116 (2012) 14328-14341.
[25]J. Lu, M. Faheem, S. Behtash, A. Heyden, J. Catal. 324 (2015) 14-24.
[26]S. Behtash, J. Lu, A. Heyden, Catal. Sci. Technol. 4 (2014) 3981-3992.
[27]Q. Meng, Y. Shen, J. Xu, X. Ma, J. Gong, Surf. Sci. 606 (2012) 1608-1617.
[28]M. Salciccioli, W. Yu, M.A. Barteau, J.G. Chen, D.G. Vlachos, J. Am. Chem. Soc. 133 (2011)
7996-8004.
[29]M. Nakamura, K. Ishimura, Langmuir 24 (2008) 5099-5108.
[30]D. Akbar, Appl. Surf. Sci. 362 (2016) 63-69.
[31]B. Liu, M.T. Lusk, J.F. Ely, A.C. van Duin, W.A. Goddard III, Mol. Simulat. 34 (2008) 967-972.
[32]B. Wang, L. Song, R. Zhang, Appl. Surf. Sci. 258 (2012) 3714-3722.
[33]R. Zhang, L. Song, Y. Wang, Appl. Surf. Sci. 258 (2012) 7154-7160.
[34]B. Peng, C. Zhao, S. Kasakov, S. Foraita, J.A. Lercher, Chem. Eur. J. 19 (2013) 4732-4741.
[35]A. Benedetti, G. Fagherazzi, F. Pinna, J. Am. Ceram. Soc. 72 (1989) 467-469.
[36]R. Srinivasan, R.J. Deangelis, G. Ice, B.H. Davis, J. Mater. Res. 6 (1991) 1287-1292.
[37]M. Selvaraj, K. Shanthi, R. Maheswari, A. Ramanathan, Energ. Fuel 28 (2014) 2598-2607.
[38]K.M.S. Khalil, M.I. Zaki, Powder Technol. 120 (2001) 256-263.
[39]B. Liu, L. Liu, X. Liu, M. Liu, Y. Xiao, Mater. Sci. Technol. 28 (2012) 1345-1348.
[40]F.J. Li, S. Zhang, J.-W. Lee, J. Guo, T.J. White, B. Li, D. Zhao, J. Cryst. Growth 404 (2014) 26-33.
[41]C. Gionco, M.C. Paganini, E. Giamello, R. Burgess, C. Di Valentin, G. Pacchioni, Chem. Mater. 25
(2013) 2243-2253.
[42]C. Cativiela, F. Figueras, J. Fraile, J. Garcia, J. Mayoral, L. de Me, E. Pires, Appl. Catal. A 101
(1993) 253-267.
[43]T. Prasomsri, M. Shetty, K. Murugappan, Y. Román-Leshkov, Energ. Environmen. Sci. 7 (2014)
2660-2669.
[44]D.J.M. de Vlieger, L. Lefferts, K. Seshan, Green Chem. 16 (2014) 864-874.
[45]T. Nozawa, Y. Mizukoshi, A. Yoshida, S. Naito, Appl. Catal. B 146 (2014) 221-226.
[46]C. Miao, O. Marin-Flores, S.D. Davidson, T.T. Li, T. Dong, D.F. Gao, Y. Wang, M. Garcia-Perez,
S.L. Chen, Fuel 166 (2016) 302-308.
[47]I. Lee, Q. Zhang, J. Ge, Y. Yin, F. Zaera, Nano Res. 4 (2011) 115-123.
[48]Q. Zhang, D.Q. Lima, I. Lee, F. Zaera, M. Chi, Y. Yin, Angew. Chem. 123 (2011) 7226-7230.
[49]Q. Zhang, I. Lee, J.B. Joo, F. Zaera, Y. Yin, Accounts Chem. Res. 46 (2012) 1816-1824.
[50]H. Chen, J. Yue, Y. Li, C. Yi, B. Yang, S. Qi, Computat. Theor. Chem. 1057 (2015) 1-6.
18

19

10Ni/Z750
15Ni/Z450
10Ni/Z450
5Ni/Z450
(004)t
(400)t
10Ni'/Z450

72
73
74
75







10
20
30
40
50
60
70
80
2
Fig.1. X-ray diffraction patterns of the catalysts: (●) Tetragonal ZrO2, (□) Ni.
20

Fig. 2. TEM images: (a,b,c) 10Ni/Z450; (d,e,f)10Ni/Z750;
21

10Ni/Z750
10Ni/Z450
g=2.0024
300
310
320
330
340
B(mT)
350
360
370
380
Fig. 3. EPR spectra of Z450 and Z750.
22

(a)
100
80
60
40
20
0
Coversion
Octane
Heptane
Octanol
Others
0
0
0
2
4
4
4
6
8
8
8
10
10
10
12
12
12
Time(h)
(b)
100
80
60
40
20
0
Conversion
Octane
Heptane
Octanol
Others
2
6
Reaction time(h)
(c)
Conversion
Octane
Heptane
Octanol
Others
100
80
60
40
20
0
2
6
Reaction time(h)
23

(d)
100
80
60
40
20
0
Conversion
Octane
Heptane
Octanol
Others
0
2
4
6
8
10
12
Reaction time(h)
Fig. 4. Octanoic acid conversion and yield of main products over 10Ni/Z450 soaked by water(a), 10Ni′/Z450 soaked
by water (b), 10Ni/Z450 without soaked by water(c) and 10Ni′/Z450 without soaked by water (d) as a function of
reaction time (Reaction condition: Temperature=240oC; 3MPa H2 pressure; 0.5g catalyst; 1g octanoic acid).
24

100
80
60
40
20
0
Conversion
Octane
Heptane
Accelerating hydrogenation
of octanal and dehydration
of octanol
Accelerating
decarbonylation
of octanal
0
1
2
3
4
5
6
H2 pressure (MPa)
Fig. 5. Octanoic acid conversion and yield of main products over 10Ni/Z450 catalysts as a function of H2 pressure
(Reaction condition: Temperature=240oC; 0.5g catalyst; 1g octanoic acid; Reaction time=12h).
25

100
80
60
40
20
0
Coversion
Heptane
0
2
4
6
8
10
12
14
16
Cycles
Fig. 6. Cycle tests of 10Ni/Z450 catalyst (Reaction condition: Temperature=240oC; 3MPa H2 pressure; 0.5g catalyst;
1g octanoic acid; One cyclic time= 12h)
26

Fig. 7. Side (upper panel) and top view (lower panel) of the adsorption structure of various intermediates of
deoxygenation for propanoic acid on Ni(111).
27

Fig. 8. Transition states of deoxygenation on the Ni(111) surface.
28

(a)
1.0
0.5
TS8
TS1
0.0
TS3
TS2
-0.5
-1.0
-1.5
-2.0
-2.5
-3.0
Decarbonylation
Decarboxylation
Hydrogenation
1.5
1.0
(b)
TS4
TS9
0.5
TS7
TS6
TS5
0.0
TS10
-0.5
-1.0
-1.5
-2.0
-2.5
Hydrodeoxygenation
Hydrogenation
Fig. 9. Potential energy profiles of DCX, DCN, HDO and hydrogenation on Ni(111) surface.
29