A study of the mechanism of triglyceride hydrodeoxygenation over alumina-supported and phosphatized-alumina-supported Pd catalysts

Article abstract of DOI:10.1016/j.jcat.2021.08.052

The mechanism of catalytic hydrodeoxygenation (HDO) of fats, vegetable oils, and fatty acids was studied using alumina-supported Pd catalysts and tricaprylin and valeric acid as model reactants. The chemistry of fatty acid/catalyst interaction was studied by quasi-operando Diffuse Reflectance Infrared Fourier Transform Spectroscopy (DRIFTS). The Pd/γ–Al₂O₃ catalyst showed good activity in the hydrogenolysis reaction of the ester bonds to convert tricaprylin to caprylic acid, but they were of poor activity in the consecutive hydrodeoxygenation (HDO) of the acid to paraffin. The surface modification of the support alumina by phosphate groups significantly increased the HDO activity of the Pd catalyst and, consequently, the paraffin yield. The activity change was accounted partly for the partial replacement of the weak base Al–OH groups by weak acid P–OH groups but mainly for the partial elimination of Lewis acid (Al^⊕) – Lewis base (O^?) pair sites on the surface of the support. Both surface Al–OH and P–OH groups were shown to participate in the reaction with carboxylic acid and formed bidentate surface carboxylate species, which further reacted with hydrogen to give paraffin. Carboxylates of less basic surface sites were found to be more prone to HDO reaction than those of strong base sites. Monodentate carboxylates, formed on Al^⊕ O^? pair sites were of low reactivity. Phosphatizing eliminated most of the Lewis type acid-base pair sites, therefore, reactive bidentate carboxylates represented the most abundant surface intermediate (MASI) during the HDO reaction of triglyceride. The hydroxyl coverage of the carboxylated surface was shown to become somewhat higher under steady-state reaction conditions. The increased hydroxyl coverage implies that C–O bond hydrogenolysis of the surface carboxylate proceeds, regenerating OH groups and forming aldehyde that could be intermediate of paraffin formation.

Full text of DOI:10.1016/j.jcat.2021.08.052

Journal Pre-proofs
A study of the mechanism of triglyceride hydrodeoxygenation over alumina-
supported and phosphatized-alumina-supported Pd catalysts
Anna Vikár, Hanna E. Solt, Gyula Novodárszki, Magdolna R. Mihályi,
Róbert Barthos, Attila Domján, Jenő Hancsók, József Valyon, Ferenc Lónyi
PII:
S0021-9517(21)00362-6
https://doi.org/10.1016/j.jcat.2021.08.052
YJCAT 14318
DOI:
Reference:
To appear in:
Journal of Catalysis
Received Date:
Revised Date:
Accepted Date:
14 June 2021
24 August 2021
31 August 2021
Please cite this article as: A. Vikár, H.E. Solt, G. Novodárszki, M.R. Mihályi, R. Barthos, A. Domján, J.
Hancsók, J. Valyon, F. Lónyi, A study of the mechanism of triglyceride hydrodeoxygenation over alumina-
supported and phosphatized-alumina-supported Pd catalysts, Journal of Catalysis (2021), doi: https://doi.org/
10.1016/j.jcat.2021.08.052
This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover
page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version
will undergo additional copyediting, typesetting and review before it is published in its final form, but we are
providing this version to give early visibility of the article. 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.
© 2021 Published by Elsevier Inc.

A study of the mechanism of triglyceride hydrodeoxygenation over alumina-supported
and phosphatized-alumina-supported Pd catalysts
Anna Vikár a, Hanna E. Solt a, Gyula Novodárszki a, Magdolna R. Mihályi a, Róbert Barthos a,
Attila Domján b, Jenő Hancsók c, József Valyon a, Ferenc Lónyi a,*
a Institute of Materials and Environmental Chemistry, Research Centre for Natural Sciences,
Magyar tudósok körútja 2, 1117 Budapest, Hungary
b Centre for Structural Science, Research Centre for Natural Sciences, Magyar tudósok
körútja 2, 1117 Budapest, Hungary
c Research Centre for Biochemical, Environmental and Chemical Engineering, Department of
MOL Hydrocarbon and Coal Processing, University of Pannonia, Veszprém, 8200, Veszprém,
Hungary
* Corresponding author. E-mail address: lonyi.ferenc@ttk.hu (F. Lónyi)
Keywords: hydrodeoxygenation, Pd/-Al2O3, Pd/phosphatized -Al2O3, tricaprylin, valeric
acid
Abstract
The mechanism of catalytic hydrodeoxygenation (HDO) of fats, vegetable oils, and fatty acids
was studied using alumina-supported Pd catalysts and tricaprylin and valeric acid as model
reactants. The chemistry of fatty acid/catalyst interaction was studied by quasi-operando
Diffuse Reflectance Infrared Fourier Transform Spectroscopy (DRIFTS). The Pd/-Al2O3
catalyst showed good activity in the hydrogenolysis reaction of the ester bonds to convert
tricaprylin to caprylic acid, but they were of poor activity in the consecutive
hydrodeoxygenation (HDO) of the acid to paraffin. The surface modification of the support
alumina by phosphate groups significantly increased the HDO activity of the Pd catalyst and,
consequently, the paraffin yield. The activity change was accounted partly for the partial
replacement of the weak base Al–OH groups by weak acid P–OH groups but mainly for the
partial elimination of Lewis acid ( ) – Lewis base ( ) pair sites on the surface of the
Al
O
support. Both surface Al–OH and P–OH groups were shown to participate in the reaction with
1

carboxylic acid and formed bidentate surface carboxylate species, which further reacted with
hydrogen to give paraffin. Carboxylates of less basic surface sites were found to be more
prone to HDO reaction than those of strong base sites. Monodentate carboxylates, formed on
Al pair sites were of low reactivity. Phosphatizing eliminated most of the Lewis type acid-
O
base pair sites, therefore, reactive bidentate carboxylates represented the most abundant
surface intermediate (MASI) during the HDO reaction of triglyceride. The hydroxyl coverage
of the carboxylated surface was shown to become somewhat higher under steady-state
reaction conditions. The increased hydroxyl coverage implies that C–O bond hydrogenolysis
of the surface carboxylate proceeds, regenerating OH groups and forming aldehyde that could
be intermediate of paraffin formation.
2

1. Introduction
The extensive combustion of fossil carbon is most probably responsible for the growing
concentration of greenhouse gas CO2 in the atmosphere. The concerns about global warming
turned attention towards the production of biofuels by upgrading non-edible and waste
vegetable oils and animal fats [1–8]. The most widely used production method of diesel range
bio-oil, generally referred to as biodiesel, is catalytic transesterification of latter renewable
triglycerides by lower alcohols [1,6,7]. However, the biodiesel cannot fully replace
conventional diesel oil, because of its lower energy density, higher viscosity, moderate
oxidation stability, and limited compatibility with fossil fuel [4,6]. A better alternative of
triglyceride upgrading is deoxygenation via hydroprocessing that is providing a mixture of
hydrocarbons. The hydrocarbon mixture is second generation biofuel, often referred to as
biogasoil or green diesel. Biogasoil has comparable or even better fuel properties than
conventional diesel fuel [6,7,9].
Catalytic triglyceride HDO can be carried out at moderate temperature (200 – 400 C)
and hydrogen pressure (<50 bar). The applied catalysts are those, routinely used in the
petroleum industry for hydroprocessing, such as, sulfided cobalt and nickel molybdate
catalysts, supported noble (Pt, Pd, Ru) and non-noble metals (Ni, Cu, Co), metal phosphides,
metal oxides, etc. [3,4,6,10]. Application of monometallic transition metal catalysts are very
common for the deoxygenation of fatty acids and triglycerides [3,6,10]. Palladium based
catalysts are especially preferred because of the peculiar ability of the palladium metal to
activate hydrogen for reaction [10]. Nevertheless, there is a general agreement that the
catalyst support does not only provide high surface area to stabilize high metal dispersion, but
it also has significant contribution to the HDO activity and selectivity by its acid-base
property [4,6,10]. Supports, having strong Brønsted acidity, such as H-zeolites, are less
favored because they initiate cracking of long chain paraffins and condensation reactions,
producing coke precursors and coke that deactivates the catalyst [6,10]. Therefore, supports of
mild-to-moderate acidity, such as activated carbon, TiO2, ZrO2, SiO2, and Al2O3 were found
suitable for HDO catalysts [4,6,10]. The most often used support is -Al2O3 [4,8].
The hydroconversion of triglycerides to paraffins proceeds in the consecutive steps of
ester bond hydrogenolysis, giving carboxylic acid, and deoxygenation of the acid to paraffin.
The deoxygenation reaction is the rate determining step of paraffin formation [3,4,6,11].
Former studies showed that the deoxygenation reaction of the carboxylic acid intermediate
over Pd catalysts follows pathways resulting in the formation of CO (hydrodecarbonylation,
3

HDCO), CO2 (hydrodecarboxylation, HDCO2), and H2O (H2-reduction of oxygen, HDH2O).
Over supported metal catalysts the HDCO was found to be the major reaction route, whereas
the HDCO2 and HDH2O were reaction routes, which had only minor contribution to the
HDO reaction [7,8]. The mechanisms of these HDO pathways are not fully understood. It was
suggested that the HDCO reaction proceeded through direct hydrogenolysis of the carboxylic
acid to paraffin and formic acid (C–C bond scission), which reaction step was followed by
quick decomposition of HCOOH to CO and H2O [6,8,11]. Deoxygenation on the HDH2O
route was proposed to proceed by deoxygenation of carboxylic acid to aldehyde intermediate
[2,3,5,8]. Accordingly, formation of aldehyde from carboxylic acid involves
hydrogenation/dehydration reactions (hydrogenation of C=O bond to CH–OH followed by
H2O formation involving C–O bond scission). The surface-bound aldehyde intermediate is
then further hydrogenated to paraffins by releasing either H2O or CO, corresponding to routes
HDH2O and HDCO, respectively [3,5].
In the present study, alumina-supported Pd catalysts were prepared and studied to learn
more about the mechanism of triglyceride HDO reaction. The effect of support
phosphatization on the catalyst structure, acid-base properties, and activity was investigated.
The HDO activity was tested using tricaprylin and valeric acid model compounds.
Quasi-operando DRIFTS investigation provided insight in the chemistry of surface
intermediate formation during the catalytic reaction and permitted to come to important
conclusions, concerning some mechanistic details.
2. Experimental
2.1 Catalyst preparation
Alumina-supported palladium catalyst was perepared by impregnating 10 gramms of
γ-Al2O3 (Ketjen CK-300, Alfa Aesar) by 10 cm3 of aqueous Pd(NH3)4(NO3)2 (product of
Strem) solution. The concentration of the solution was adjusted to get catalyst of 0.5 wt.% Pd
content. The sample was dried at 110 °C for 16 h. To decompose the metal precursor salt the
sample was calcined. It was heated first at a heating rate of 2 °C min-1 to 150 °C, kept at this
temperature for 1 h, and then the temparature was raised at a heating rate of 4 °C min-1 to
350 °C. The catalyst was kept at this tempreture for additional 4 h. The obtained catalyst
sample was designated as Pd/Al2O3.
4

The phosphatized-alumina-supported Pd catalyst samples were prepared following the
same procedure as above, except that the alumina support was phosphatized first to different
extents. Supports with 1.0, 2.5, and 5.0 wt.% phosphorous content were prepared by
impregnating 10 g of γ-Al2O3 with 10 cm3 of a solution, containing calculated amount of
phosphoric acid. The impregnated samples were dried at 110 °C for 16 h then calcined in air
at 550 °C for 4 h to generate the phosphatized alumina supports. The thus obtained
phosphatized γ-Al2O3 catalyst supports are designated as Al2O3-1P, Al2O3-2.5P and Al2O3-5P,
respectively. These supports were used to prepare catalysts of 0.5 wt.% Pd content. The
corresponding supported Pd catalysts were designated as Pd/Al2O3-1P, Pd/Al2O3-2.5P, and
Pd/Al2O3-5P.
2.2 Catalyst characterization
2.2.1 Elemental analysis
The P and Pd content of the catalyst samples were determined by using Inductively
Coupled Plasma Optical Emission Spectroscopic (ICP-OES) method (Spectro Genesis ICP-
OES apparatus).
2.2.2 Specific surface area (SSA)
Nitrogen adsorption isotherms of the catalyst samples were determined at -196 °C using
an automatic, volumetric adsorption analyzer (The "Surfer", product of Thermo-Fisher
Scientific). The sample was dehydrated before the measurement at 250 °C under high vacuum
(10-6 mbar) for 4 h. The SSA of the catalyst samples was determined by the BET method,
whereas the pore-size distribution was calculated using the Barett-Joyner-Halenda (BJH)
method.
2.2.3 X-ray powder diffraction (XRPD)
The XRPD measurements were carried out using a Philips PW 1810/3710 X-ray powder
diffractometer in a Bragg-Brentano parafocusing arrangement applying monochromated Cu
Kα (λ = 1.5418 A) radiation.
2.2.4 Metal dispersion
The dispersion of Pd in the catalysts was determined by CO pulse chemisorption method.
About 100 mg of the sample was placed into a quartz microreactor (I.D.: 4 mm) and reduced
in situ in hydrogen flow at 450 °C for 1 h. It was flushed then by He flow at 450 °C for
30 min and cooled to room temperature in the He flow. In 3 min intervals carbon monoxide
5

pulses of 10 µL volume were injected sequentially into the He flow, passing through the
catalyst bed. The CO concentration of the reactor effluent was monitored using thermal
conductivity detector (TCD). The TCD signal was processed by computer. The introduction
of CO pulses was continued until the chemisorption sites were saturated. After calibration the
molar amount of chemisorbed CO was calculated from the areas of the TCD signals.
2.2.5 Temperature-programmed CO2 desorption (CO2-TPD)
The CO2-TPD measurement was used to characterize the basicity of the supports. About
150 mg of the sample was placed into a quartz microreactor (I.D.: 4 mm) and activated in O2-
flow at 550 °C for 1 h. The sample was then flushed with N2 for 15 min at 550 °C, evacuated
at the same temperature for 30 min and cooled to room temperature. Adsorption of CO2 was
carried out by contacting the sample with CO2 gas at 13.3 kPa for 15 min. The system was
flushed by He and the temperature of the reactor was ramped up at a rate of 10 °C min-1 to
700 °C. The CO2 concentration in the gas flow was monitored by TCD.
2.2.6 Fourier Transform Infrared Spectroscopy (FTIR)
A Nicolet 6700 FTIR (Thermo Scientific) instrument was used in transmission mode to
record the spectra of the surface species present on the neat supports and catalysts and
obtained from adsorption of compounds. Self-supporting wafer of the examined sample
having a “thickness” of 5–10 mg cm-2 was placed into the sample holder of a stainless steel
spectroscopic cell equipped with CaF2 windows and a furnace section for in situ activation of
the sample either in atmospheric gas flow or under high vacuum. Spectra were taken by
averaging 512 scans at a nominal resolution of 2 cm-1.
The acidity and basicity of the supports were studied by determining the spectra of
adsorbed pyridine (Py) or CO2, respectively. Prior to Py adsorption the sample was pretreated
at 450 °C under high vacuum (10-6 mbar) for 1 h then the temperature was lowered to 200 °C
and the sample was contacted with Py vapor at 5 mbar for 30 min. The sample was cooled
then to 100 °C. The Py vapor was removed from the cell by successive evacuation at
temperatures 100, 200, 300, 400, and 450 °C for 30 min at each temperature. After each
evacuation a spectrum was recorded at room temperature. The spectrum of the wafer,
recorded before Py adsorption, was subtracted from each spectrum to obtain the spectrum of
the adsorbed species only.
A procedure, similar to that described above, was followed to determine the spectra of the
species obtained from adsorption of CO2. Wafer of the activated sample was contacted with
CO2 at 15 mbar at room temperature for 30 min. The spectrum of the adsorbed species from
6

CO2 was measured after successive evacuation at room temperature, 100, 200, 300, and
400 °C under high vacuum for 30 minutes at each temperature.
The electronic state of Pd in the catalysts was characterized by analyzing the FTIR
spectrum of adsorbed species formed from CO. Prior to CO adsorption, the Pd-containing
catalyst wafer was reduced at 450 °C in H2 stream for 1 h. The catalyst was contacted with
CO gas at 5 mbar at room temperature for 20 min then the spectrum of the carbonyl species
formed was recorded after 20 min evacuation under high vacuum at room temperature. Each
absorbance spectra were scaled to 5 mg cm-2 wafer thickness to allow quantitative
comparisons.
2.2.7 Magic angle spinning solid-state nuclear magnetic resonance (MAS NMR)
Spectra were recorded using a Varian NMR System spectrometer operating at 600 MHz
1H frequency (242.74 MHz for 31P and 156.26 MHz for 27Al) with a Chemagnetics 3.2 mm
narrow-bore triple resonance T3 probe in double resonance mode. The 31P direct polarization
spectra were recorded (160 transients) at 20 °C with 12 kHz of spinning rate and 300 s
repetition delay. The 27Al – 1H cross polarization spectra were recorded (12000 transients)
with 0.5 ms contact time, 5 s of repetition delay at 20 °C with 12 kHz of spinning rate. For
both experiments SPINAL 1H decoupling was used. As chemical shift reference ammonium
dihydrogen phosphate (δiso = 0.81 ppm with respect to 85 wt.% H3PO4 solution) for the 31P
and sodium aluminate (δiso = 79.3 ppm) for the 27Al measurements was used.
2.2.8 Catalytic experiments
Hydroconversion of tricaprylin was investigated using a high pressure fixed-bed flow-
through microreactor system. The catalytic reactor (I.D.: 8 mm) was filled with 2.0 g of
catalyst using its 0.315-0.630 mm sieve fraction. Prior to the catalytic run, the catalyst was
reduced in situ in 50 cm3 min-1 flow of H2 at 450 °C for 2 hours at atmospheric pressure, then
the temperature was lowered to the desired reaction temperature (300 or 350 °C) and the
pressure was increased to 21 bar total pressure. The tricaprylin reactant was fed into the
reactor using a high-pressure syringe pump (ISCO) at a weight hourly space velocity
h ―1
, whereas the H2/tricaprylin molar ratio was 20. The
―1
(WHSV) of
4 gtricaprylin
g
catalyst
product mixture was cooled to room temperature and the liquid products were separated from
the gas products in a reflux condenser downstream of the reactor. The effluent gas leaving the
condenser through a back pressure regulator valve contained mainly H2, and CO (as main
product), CO2, and minor amount of CH4. The effluent was analysed on-line using a gas
chromatograph (GC) equipped with a TCD detector and a 60/80 Carbonex-1000 (L 15.0 ft ×
7

OD 1/8”) stainless steel column. The liquid product from the presurized collection vessel was
first transferred into a closed atmospheric vessel. During this transfer, the pressure of the
vessel reached the system pressure, which was then released via a transfer line connected to a
syringe, where the gas, expanded to atmospheric pressure, was collected. Note that this latter
expanded gas contained all of those products, which were in liquid phase under system
pressure but appeared as gases at atmospheric pressure (mainly propene, and some ethane in
H2). The liquid sample, drained now from atmospheric pressure, was analyzed by GC
equipped with flame ionization detector (FID) using CP-FFAP CB (L 25.0 m × ID 0.32 mm ×
df 0.3 μm) capillary column, whereas the expanded gas products were analysed by GC-TCD-
FID using ShinCarbon ST (L 2.0 m × ID 1/8 in. × OD 2.0 mm) column. Samples were taken
in every hour after the steady state was reached (after one hour time of stream). At high
tricaprylin conversions, when the reactant was converted almost completely into paraffins, the
liquid product split into two clear, colorless hydrocarbon and water phases. Since the lower
aqueous phase contained only a negligible amount of organic components, in such cases only
the composition of the upper organic phase was analyzed. In some cases, when the tricaprylin
conversion was not complete, and/or was not completely converted into paraffins, a single
phase liquid product was obtained, which was analysed for the organic components. In some
cases, when unknown components were also formed, GC-MS equiped with Rxi-5Sil MS
(L 30.0 m × ID 0.25 mm × df 0.25 μm) column was used for peak assignments.
2.2.9 Diffuse Reflectance Infrared Fourier Transform Spectroscopy (DRIFTS)
Quasi-operando DRIFT spectroscopic investigation of the adsorbed species formed from
carboxylic acid intermediate under catalytic conditions is crucial in order to reveal structure-
activity relationships of the catalyst in the HDO reaction of triglycerides or fatty acids. The
experiments were carried out using an FT-IR spectrometer (Thermo Nicolet iS10) equipped
with a Collector II diffuse reflectance mirror system and a flow-through DRIFT spectroscopic
reactor cell (I.D.: 5 mm, height of catalyst bed ~ 4 mm) filled with about 20 mg of powdered
catalyst sample. The design of the cell allows carrier gas or gas phase reactant mixture to flow
through the catalyst bed in the sample cup. The reactant is introduced into the cell by
switching the carrier gas flow to a gas saturator containing the reactant at room temperature.
Due to the experimental difficulties related to the very low vapor pressure of caprylic acid,
valeric acid (VA) was used as model compound in these experiments. First, the catalyst was
pre-treated in situ in a 50 cm3 min-1 H2 flow at 450 C for 2 h and then the spectrum of the
activated catalyst powder was collected (512 scans at a nominal resolution of 2 cm-1) at the
8

desired reaction temperatures (300 or 350 C). The reaction of VA was initiated by switching
the H2 or He flow (50 cm3 min-1) to the gas saturator. The thus obtained H2/VA or He/VA
mixture contained 258 ppm VA. The spectrum taken in the presence of the reacting gas
mixture at 300 or 350 C was corrected with the spectrum of the pure catalyst at the same
temperature. Since the contribution of the gas phase spectrum was found negligible under the
applied conditions, the thus obtained difference spectrum practically reflects the bands of
surface species formed (positive bands) or consumed (negative bands) in the VA
adsorption/reaction process.
The effluent gas leaving the reactor cell was continuously monitored by online mass
spectrometer (MS; VG ProLab, Thermo Scientific) following the characteristic masses of the
+
+
major reaction products: butane (m/z = 58, C4H10 ), pentane (m/z = 72, C5H12 ), CO (m/z = 28,
+
CO+), and CO2 (m/z = 44, CO2 ). All signals were corrected for the contribution of other
reaction products giving a fragment at the same m/z value.
3. Results
3.1 Catalyst composition and structure
The measured Pd and P contents of the catalysts, listed in Table 1, are in good agreement
with the values that follow from the applied conditions of catalyst preparation.
The nitrogen adsorption isotherms, shown in Fig. S1, are characteristic for mesoporous
oxides. They are classified as type IV isoterms, having H2 type hysteresis loop. The SSA of
the catalysts decreased as their phosphorous content was increased (Table 1). The SSA of the
Pd/Al2O3-5P catalyst is about 40 % lower than that of the Pd/Al2O3 catalyst. These results
suggest that phosphate groups can block some pores of the alumina support and thereby
decrease the SSA.
XRPD patterns of the phosphatized catalysts and that of the parent γ-Al2O3 support are
shown in Fig. 1. The diffractograms of the support and all the catalysts were similar, i.e., no
new crystalline phase could be detected (Fig. 1). The results suggest that the size of the Pd or
PdO particles on the support is well below the detection limit of the XRPD method (the
diameter was less than about 5 nm).
The Pd dispersion (DPd) was obtained as the ratio of the number of surface Pd atoms and
the total number of Pd atoms in the catalyst. The molar amount of chemisorbed CO was taken
to be equivalent with half of the molar amount of surface Pd atoms [12].
9

e
d
c
b
a
10
20
30
40
2 °
50
60
70
Fig. 1. XRD patterns of (a) parent -Al O , and (b) Pd/Al O , (c) Pd/Al O -1P,
γ
2
3
2
3
2
3
(d) Pd/Al2O3-2.5P, and (e) Pd/Al2O3-5P catalyst samples.
Table 1. Composition and properties of the catalysts.
CO, c
mol g-1
P, a
Pd, a
SSA, b Vp, b
m2 g-1 cm3 g-1
DPd, c dPd, c
Catalyst
wt.% wt.%
%
nm
1.3
1.6
1.5
1.5
Pd/Al2O3
0.0
0.9
0.45
0.46
0.46
0.47
212
183
167
132
0.52
0.46
0.43
0.36
18.5
15.1
16.0
16.4
86
70
74
74
Pd/Al2O3-1P
Pd/Al2O3-2.5P 2.6
Pd/Al2O3-5P 5.4
a From ICP-OES analysis. b Specific surface area (SSA) determined from the N2 adsorption isotherm
using the BET method; the total pore volume (Vp) determined using the Gurvich method from the
c
푝/푝 = 0.95
adsorption capacity at
. Determined by pulse CO chemisorption method; Pd dispersion
0
(DPd) was calculated assuming 1 to 2 CO to Pd adsorption stoichiometry [12]; the average Pd particle
size (dPd) was calculated by Eq. 1.
Assuming spherical particle shape and that the three low-index planes are in equal
proportions on the polycrystalline surface of the face-centered cubic crystals of the metal, the
mean Pd particle size (dPd) was calculated from the dispersion by the equation
푣
,
(1)
푑Pd = 6
푎 퐷Pd
where v is the volume occupied by a single Pd atom in the bulk of metal (1.47·10-2 nm3), and
is the average surface area occupied by one Pd atom (7.93·10-2 nm2) [13]. The Pd
푎
10

dispersions and the mean particle sizes, listed in Table 1, barely changed with the
phosphorous content of the catalysts.
3.2. Surface properties of the catalysts
3.2.1. Surface hydroxyls
The FTIR spectra measured for the phosphate-free and phosphatized γ-Al2O3 supports in
the range of stretching vibration of surface OH groups are shown in Fig. 2. The assignment of
the alumina OH bands, first given by Knözinger and Ratnasamy [14], was later refined by
Busca et al. [15–17]. The νOH bands above about 3700 cm-1 were assigned to terminal OH
groups. The band at 3792 cm-1, appearing as a weak shoulder in the spectrum of pure γ-Al2O3
(Fig. 2a) was attributed to νOH vibration of groups linked to tetrahedral aluminum ions. The
band at 3773 cm-1 belongs to OH groups next to a coordinately unsaturated tetrahedral
aluminum atom, i.e. next to a Lewis acid site. In a very recent paper [18] the assignment of
the band at about 3780–3770ꢀcm−1 was made more accurate. The band was shown to stem
from hydroxyl groups bound to octahedrally coordinated surface Al3+ ions (O5AlVI-OH sites)
that transforms to pentacoordinated Al3+ surface sites (“bare” O5AlV sites) upon thermal
dehydroxylation.
The relatively broad band, centered at 3730 cm-1 (Fig. 2a) was assigned to OH groups
bound to octahedrally coordinated Al3+ ions [15-17]. The presence of a coordinately
unsaturated octahedral aluminum atom adjacent to an OH group, results in OH band, shifted
to the 3740-3700 cm-1 frequency region [15–17]. However, we could not distinguish these
bands because of the band broadening and strong overlapping (Fig. 2a). The OH bands below
about 3700 cm-1 are attributed to different bridging OH species [15–17]. Thus, the broad band
centered approximately at 3670 cm-1 can be assigned to bridging OH species, whereas the
very broad feature around 3590 cm-1 can be attributed to triply bridging OH species or OH
groups in hydrogen bonding interactions [15–17].
The intensity drop of the OH bands (Fig. 2, b-d) clearly indicates OH consumption in a
reaction with phosphoric acid. The reaction has been described as sort of surface acid-base
neutralization reaction resulting in the formation of surface phosphate species and water [19-
21]. The formation of phosphate species is clearly indicated by the appearance of the band at
3677 cm-1, assigned to the νOH vibration of P–OH species (Fig. 2, b-d) and a strong νP=O
vibrational band of the phosphate groups in the range of 1150-1250 cm-1 frequency (not
shown). Interestingly, the characteristic νOH band of the terminal OH groups on tetrahedral
11

aluminum ions at 3792 cm-1 gained in intensity as the concentration of the surface phosphate
species increased (Fig. 2, a-d), suggesting that a surface reaction between phosphoric acid and
alumina surface is not a simple acid-base neutralization reaction.
d
Al2O3-5P
c
Al2O3-2.5P
b
Al2O3-1P
-Al2O3
a
3900
3800
3700
3600
3500
Wavenumbers, cm-1
Fig. 2. FTIR spectra of the νOH stretching region of (a) parent γ-Al2O3, and (b) Al2O3-1P,
(c) Al2O3-2.5P, and (d) Al2O3 5P phosphatized alumina supports. The samples were
pretreated in situ in the IR cell at 450 °C under high vacuum for 1 h.
3.2.2. Surface basicity
The surface basicity of the phosphate-free and phosphatized γ-Al2O3 supports was
characterized by the carbonate-like surface species obtained from CO2 adsorption. The IR
spectrum of these species was recorded and the spectral features were assigned on the basis of
the available literature [17,21,22] (Fig. 3). The CO2 uptake of the supports resulted mainly in
the formation of two types of bicarbonate species as indicated by the appearance of the
asymmetric νO-C=O vibrations at 1645 cm-1 for both B1 and B2 type bicarbonate species and
the symmetric vibrations at 1454 cm-1 of the B1 and 1482 cm-1 of the B2 type bicarbonate
species (Fig. 3A). The corresponding νOH bands of the bicarbonate species appear as positive
bands at 3610 (B2 type) and 3595 cm-1 (B1 type, as a shoulder) in the range of the O–H
vibration frequencies (Fig. 3B). The most intense negative νOH bands at 3773 and around
3700 cm-1 (Fig. 3B) suggest that mainly those OH groups were involved in the formation of
bicarbonates, which have a coordinately unsaturated tetrahedral or octahedral aluminum atom
in their neighborhood. Note that the involvement of these hydroxyl groups in the CO2
adsorption could be just due to their location near to Lewis acid sites [16]. The lack of
12

negative band at 3677 cm-1, where the νOH band of the phosphorus-bound hydroxyls appears,
implies that the P–OH groups do not participate in bicarbonate formation (Fig. 3B). In
harmony with the conclusion of other authors [21,23], this result permitted for us to conclude
that phosphatization does not generate basic centers for CO2 uptake, i.e., the P–OH group is
stronger acid than the CO2.
A
B
a
a
b
b
c
c
d
d
1800
1700
1600
1500
1400
3800
3700
3600
3500
Wavenumbers, cm-1
Wavenumbers, cm-1
Fig. 3. FTIR spectra of (A) surface species formed from CO2 on the parent -Al2O3 (a) and on
the Al2O3-1P (b), Al2O3-2.5P (c), and Al2O3-5P (d) phosphatized alumina supports and
(B) the corresponding difference spectra in the range of the O–H stretching vibrations.
Samples were pretreated in situ in the IR cell at 450 °C under high vacuum for 1 h then were
contacted with 15 mbar of CO2 at room temperature for 30 min. Spectra were taken after
30 min evacuation at room temperature.
The CO2 uptake on the supports results also in the formation of bidentate chelating
carbonate that gives weak IR band at 1670 cm-1 (shoulder) and monodentate carbonate bands
at 1542 and 1404 cm-1 (shoulders) frequencies (Fig. 3A) [17,22]. Formation of these
carbonate species probably takes place with the involvement of coordinately unsaturated
aluminum cation and oxide anion pairs, which oxide ions are Lewis base sites on the alumina
surface [16,17,21]. It is important to note that the strength of the characteristic absorption
bands of the carbonate species is inversely proportional to the phosphate loading (Fig. 3A, a-
d). These results suggest that formation of phosphate species is accompanied not only by
consumption of basic surface OH groups, but also by elimination of Lewis base sites.
The concentration and distribution of basic sites on the parent and phosphatized -Al2O3
supports were determined by CO2-TPD measurements. The TPD curves in Fig. 4 indicate the
13

presence of at least three overlapping component bands. According to Wang et al. [22], a peak
appearing around 80 C is due to the decomposition of bicarbonate species formed on weak
basic sites, whereas the component peaks observed around 160 and 250 C can be attributed
to decomposition of chelating bidentate carbonate species formed on medium strength basic
sites and monodentate carbonate species formed on strong basic sites, respectively. These
assignements are clearly supported by the observed thermal stability of different carbonate
species (Fig. S2). An additional high temperature peak, attributed to bridging bidentate
carbonate species formed on strong basic sites, may also appear at around 325 C
temperature. Therefore, the CO2-TPD curves shown in Fig. 4 were resolved by peak fitting
using four component peaks. The peak fitting process resulted essentially in three major
component peaks with a maximum around 85, 135, and 220 C, representing weak, medium
strength and strong basic sites (the fourth component peak had a negligible intensity on each
sample, therefore it was ignored). The total amount of different basic sites and their
distribution was calculated from the area under the corresponding curves using the result of a
previous calibration measurement (Table S1). The concentrations of the weak-to-moderate
strength basic sites and the strong basic sites are listed in Table 2. Results show that
introduction of phosphate groups significantly decreased the concentration of all types of
basic centers on the surface of the alumina support, which dropped by more than 90 % for the
support, having the highest phosphorous content.
120
100
a
80
60
b
40
c
20
0
d
25
100
200
300
400
500
Temperature, °C
Fig. 4. CO2-TDP curves measured for (a) parent γ-Al2O3, and (b) Al2O3-1P, (c) Al2O3-2.5P,
and (d) Al2O3-5P phosphatized alumina supports. Prior to the CO2-TPD run the sample was
activated in O2 stream at 550 °C. Adsorption was carried out at room temperature at 133 mbar
CO2 pressure. The TPD run was performed in He flow at a heating rate of 10 °C min-1 up to
700 °C.
14

Table 2. Basicity and acidity of the catalyst supports.
Basic sites,a mol g-1
Lewis acid sites,b mol g-1
Support
weak-to-medium
strong
34.7
26.3
8.2
weak-to-medium
strong
4.47
4.37
2.35
1.90
90.8
66.0
22.8
5.8
16.1
16.7
14.9
12.9
-Al2O3
Al2O3-1P
Al2O3-2.5P
Al2O3-5P
1.3
a Determined from the CO2-TPD measurements. Weak and medium strong basic sites were determined
from the component peaks at 85 and 135 C, whereas that of strong basic sites from the component
peak at 220 C. b Calculated from the Py adsorption examined by FT-IR spectroscopy. Concentration
of Lewis acid sites was calculated from the integrated absorbance of the 1450 cm-1 Py band after
degassing the wafer at 100 C (weak and medium strength Lewis acid sites) and 400 C (strong acid
Lewis sites), respectively. For the calculation, the integrated molar extinction coefficient of
2.22 cm µmol-1 was used [24].
3.2.3. Surface acidity
Adsorption of Py on the Lewis acid sites of alumina results in the formation of
coordinately bonded Py giving absorption bands in the range of 1630-1590 (8a band) and
1460-1430 cm-1 (19b band) [16,17,25]. The pair of bands observed at 1623 and 1456 cm-1 in
Fig. 5a can be assigned to Py adsorbed on tri-coordinated Al3+ cations (tetrahedral Al cations
with coordinative unsaturation), which represent the strongest acid Lewis sites of alumina.
The second pair of bands at 1615 and 1451 cm-1 can be also attributed to Py adsorbed on Al
ions with coordinative unsaturation, which are most probably in octahedral coordination
[16,17,25] and represent weaker acid Lewis sites than those of the coordinately unsaturated
tetrahedral aluminum cations.
The spectra of adsorbed CO2 support above identification of Py sorption sites. Bands at
+
2360 and 2345 cm-1, which belong to the so called
mode of linearly coordinated (end-on
Σu
adsorbed) CO2 [17,21,26], were found to develop in the presence of CO2 gas in the IR cell
(Fig. S3). These bands are most probably bands of CO2, bound to strong and weak,
coordinately unsaturated Lewis acid Al centers in tetrahedral and octahedral coordination,
respectively.
The intensity of Py bands is lower for the supports having higher phosphate concentration
(Fig. 5, a-d), suggesting that formation of surface phosphate affected the concentration of both
types of Lewis acid sites.
15

a
b
c
d
1650
1600
1550
1500
1450
Wavenumbers, cm-1
Fig. 5. FT-IR spectra of surface species formed from adsorption of Py on (a) parent γ-Al2O3,
and (b) Al2O3-1P, (c) Al2O3-2.5P, and (d) Al2O3-5P phosphatized alumina supports. Samples
were pretreated in situ in the IR cell at 450 °C under high vacuum for 1 h then were contacted
with 5 mbar of Py vapor at 200 °C for 30 min. Spectra were taken after evacuation at 100 °C
for 30 min (thick solid lines) and then after evacuation at 400 °C for 30 min (thin dotted
lines).
The bands of linearly adsorbed CO2 are also weaker for the phosphatized supports,
indicating that surface phosphate eliminates Lewis acid sites of alumina (Fig. S3, a-d).
Evacuation at 400 C resulted in the total desorption of Py from the weaker Lewis acid
sites (see the intensity drop of the bands at 1615 and 1451 cm-1), whereas the strongest Lewis
acid sites still withheld Py (see the bands at 1623 and 1456 cm-1) (Fig. 5, dashed lines).
Assuming that the total number of Lewis sites and the number of strong Lewis sites are
proportional to the band intensities observed after evacuation at 100 C and 400 C,
respectively, the corresponding concentrations of Lewis acid sites were determined by using
the extinction coefficient given in ref. [24]. The concentrations of the weak-to-moderate
strength Lewis acid sites and the strong Lewis acid sites are listed in Table 2.
3.2.4. 31P MAS NMR and 1H-27Al CP/MAS NMR results
The type of surface phosphate species was detected by 31P MAS NMR. The spectrum of
the Al2O3-1P sample, having the lowest P content, shows peaks at -10 and -22 ppm (Fig. 6A,
b), which peaks can be assigned to phosphorous in monomeric and polymeric phosphate
species, respectively [22,23,27,28]. At higher phosphorous contents, both peaks were
stronger, while the peak of the polymeric species gained even more intensity (Fig 6A, b-d). In
16

line with expectations [23], the higher surface phosphate concentration favors the formation
of polymeric species via condensation of P–OH groups.
A
B
Al2O3-5P
d
Al2O3-5P
d
Al2O3-2.5P
c
Al2O3-2.5P
Al2O3-1P
Al2O3-1P
b
c
a
-Al2O3
b
10
0
-10
-20
-30
-40
-50
100
80
60
40
20
0
-20
-40
31P chemical shift, ppm
27Al chemical shift, ppm
Fig. 6. 31P MAS NMR (A) and 1H-27Al CP/MAS NMR (B) spectra of (a) parent γ-Al2O3, and
(b) Al2O3-1P, (c) Al2O3-2.5P, and (d) Al2O3-5P phosphatized alumina supports.
The changing local environment of aluminum atoms on the surface of alumina upon
phosphate modification was characterized by 1H and 27Al CP/MAS NMR (Fig. 6B). The 1H
spectra reflect the local environment of those Al atoms which are near to protons at the
surface or near to surface-attached OH groups [23,29]. On the parent alumina support 27Al
resonance peaks can be observed at 14, 38, and 75 ppm (Fig. 6B, a), which can be attributed
to octahedral (AlVI), pentagonal (AlV), and tetrahedral (AlIV) surface aluminum atoms [27–
30]. Note that pentagonal aluminum atoms (AlV) are often observed in high surface area
transition aluminas in minor concentrations and their presence is associated with oxygen
defects adjacent to aluminum nucleus or substitution of lattice oxygen in octahedral symmetry
by hydroxyl groups [30,31]. When the alumina surface was modified with increasing amount
of phosphate, it was clearly visible that relatively broad component peaks developed at lower
chemical shifts near to these peaks at about 54, 26, and 5 ppm (Fig. 6B, b-d) indicating the
changes in the local environment of the corresponding surface Al atoms due to the formation
of Al-O-P bonds [27,32]. Results show that all types of surface Al atoms were involved in the
formation of Al-O-P bonds, suggesting the non-selective binding of the phosphate to the
alumina surface. This observation is in line with the spectral changes found in the νOH region
(Fig. 2) indicating the consumption of all the different types of OH groups upon phosphating
the alumina surface.
17

d
Pd/Al2O3-5P
c
Pd/Al2O3-2.5P
b
Pd/Al2O3-1P
Pd/Al2O3
a
2200
2100
2000
1900
1800
1700
Wavenumbers, cm-1
Fig. 7. FT-IR spectra of adsorbed species formed from carbon monoxide on (a) Pd/Al2O3,
(b) Pd/Al2O3-1P, (c) Pd/Al2O3-2.5P, and (d) Pd/Al2O3-5P catalyst samples. Samples were
pretreated in situ in the IR cell at 450 °C in H2-flow for 1 h then contacted with 5 mbar of CO
at room temperature for 30 min. Spectra were taken after removing gas phase CO from the
cell by evacuation at room temperature.
3.2.5. Characterization of the Pd surface by CO chemisorption
It is generally accepted that the carbonyl bands of CO adsorbed on highly dispersed Pd
catalysts appear in the spectral range below and above about 2000 cm-1, attributed to bridging
and linearly-bound CO, respectively [12,33,34]. The FTIR spectrum of the species formed
from CO adsorption on alumina-supported Pd catalysts are shown in Fig. 7. Carbonyl bands
are clearly discernible at about 1860, 1945, 2050, and 2085 cm-1. Following the band
assignments of Lear et al. [34,35], the broad band at about 1860 cm-1 can be assigned to µ3
hollow-bonded CO on Pd [111] planes or µ2 bridge-bonded CO on Pd [100] planes, whereas
the band near to 1945 cm-1 can be attributed to the µ2 bridge-bonded CO on Pd [100] facets
and CO, bridge bonded to particle edges. The linear CO peaks at around 2050 and 2085 cm-1
can be ascribed to CO bound to Pd [111]/[111] and Pd [111]/[100] particle edges, and CO
bound to particle corners, respectively [34,35]. These bands are all present both on the non-
phosphatized and phosphatized-alumina-supported Pd catalysts, although some deviation
from the published relative intensities, are apparent (Fig. 7). In particular, the linear features
relative to the bridge-bonded features became more pronounced at increasing phosphorous
content of the support, which suggest somewhat greater contribution of the Pd particle edges
18

and corners to the adsorption. Note that phosphating the alumina support hardly affected the
metal dispersion and the Pd particle size in the catalysts (Table 1).
3.3. Catalytic activity
Results of catalytic hydroconversion of tricaprylin (TC) are shown in Fig. 8. The organic
liquid product contained the unreacted TC (if any) and caprylic acid, propyl caprylate, 1-
octanol, octyl caprylate, heptane, octane, and some other minor products, mainly octanal, 8-
pentadecanon, 9-nonanone, and dicaprylates formed by the hydrogenolysis of only one ester
bond of TC. Caprylic acid and propyl caprylate is formed by the hydrogenolysis (HYS)
reaction of three or two ester bonds, respectively, whereas octyl caprylate could have been
formed by esterification of caprylic acid by octanol. Heptane and octane were produced via
deoxygenation of caprylic acid. Octanol is a possible intermediate of paraffin formation [11].
The effluent gas contained mainly propane (from HYS reaction) and CO. Minor amounts of
CO2, ethane, and methane could be also detected. The dominance of heptane over octane and
CO over CO2 in the liquid and gas phase product mixture, respectively, suggests that
hydrodecarbonylation (HDCO) is the main deoxygenation route, whereas
hydrodecarboxylation (HDCO2) and oxygen hydrogenation (HDH2O) represent minor
reaction routes [6,7].
Fig. 8. Catalytic hydroconversion of tricaprylin over Pd catalysts supported on parent
γ-Al2O3, and on the Al2O3-1P, Al2O3-2.5P, and Al2O3-5P phosphatized alumina supports.
Catalysts were pretreated in situ in H2-flow at 450 °C at atmospheric pressure. Catalytic tests
were carried out at 300 °C (left) or 350 °C (right) at a total pressure of 21 bars and at a
h ―1
―1
WHSV of
, whereas the H2/tricaprylin molar ratio was 20.
19
4 g
g
tricaprylin catalyst

At the reaction temperature of 300 C the conversion of TC was low (18.5 %) on the
Pd/Al2O3 catalyst, but reached virtually 100 % on the Pd/Al2O3-2.5P and Pd/Al2O3-5P
catalysts (Fig. 8, left side). When the reaction temperature was raised to 350 C, all the
phosphatized-alumina-supported Pd catalysts showed high activity in the HYS reaction
resulting in 100 % TC conversion (Fig. 8, right side). In line with earlier findings [11], these
results suggest that caprylic acid intermediate was formed by facile HYS reaction from TC,
which was followed by consecutive, rate-limiting deoxygenation (mainly HDCO) reaction of
the intermediate. Interestingly, the yield of the paraffin products (heptane and octane)
dramatically increases with the phosphorous content of the alumina support reaching nearly
100 % on the Pd/Al2O3-5P catalysts (Fig. 8, right side). These results clearly suggest that
phosphatization of alumina surface resulted in the change of catalyst structure so that the rate
of the hydrodeoxygenation (mainly HDCO) reaction was significantly enhanced.
3.4. In situ DRIFT spectroscopic investigation
The carboxylate species formed from adsorption of valeric acid, and their reactivity was
investigated under catalytic conditions by quasi-operando DRIFT spectroscopy. The results
obtained for the Pd/Al2O3 and Pd/Al2O3-5P catalysts are presented in Figs. 9 and 10,
respectively.
Molecularly adsorbed carboxylic acid could not be observed under the conditions of
experiments, i.e., the characteristic νC=O band of valeric acid expected to appear at
 1780 cm-1 could not be detected. The negative bands in the νOH region (Figs. 9-10, Section
A) and positive bands in the νO-C-O region (Figs. 9-10, Section B) clearly suggest that the
adsorption of the carboxylic acid resulted in the consumption of surface OH groups and in the
simultaneous formation of surface carboxylate species [36,37]. This surface reaction
(dissociative adsorption) is described as the deprotonation of the carboxylic acid by the
combination of acid hydrogen with a surface hydroxyl group to produce surface carboxylate
species and H2O [36–38]. Results shown in Figs. 9A and 10A indicate that practically all
types of OH groups on the support can be involved in the formation of carboxylate species,
including the P–OH groups of the phosphatized support (Fig. 2). Note that phosphatization
resulted in consumption of surface Al–OH groups and formation of new P–OH groups
(Fig. 2). Both the remaining Al–OH and the new P–OH groups are available for the
adsorption of carboxylic acid and their total number determines the surface concentration of
the carboxylate groups.
20

A
B
C
raise temp.
to 350 °C
a
b
switch to
H2/VA (300 °C)
CO
a
b
CO2
C4H10
C5H12
3900 3800 3700 3600 18001700160015001400
0
30 60 90 120 150
Time, min
Wavenumbers, cm-1
Wavenumbers, cm-1
Fig. 9. Difference DRIFT spectra in the range of the (A) O–H stretching and (B) C=O
stretching and C–H deformation vibrations on the Pd/Al2O3 catalyst in contact with (a)
He/valeric acid mixture (dotted lines) or H2/valeric acid mixture (solid lines) at 300 °C or (b)
at 350 °C. Part (C) shows the MS signals characteristic of CO, CO2, butane, and pentane as
HDO products in the effluent gas from the DRIFT reactor cell during the operando DRIFTS
experiment with H2/valeric acid mixture. Dashed lines indicate the point where spectra were
collected after reaching the steady state at the given reaction temperature.
A
B
C
raise temp.
to 350 °C
switch to
H2/VA (300 °C)
a
b
CO
a
b
C4H10
CO2
C5H12
3900 3800 3700 3600
18001700160015001400
0
30 60 90 120 150
Time, min
Wavenumbers, cm-1
Wavenumbers, cm-1
Fig. 10. Difference DRIFT spectra in the range of (A) O-H stretching and (B) C=O stretching
and C–H deformation vibrations on Pd/Al2O3-5P catalyst in contact with He/valeric acid
mixture (dotted lines) or H2/valeric acid mixture (solid lines) at (a) 300 °C and (b) 350 °C.
Part (C) shows the MS signals characteristic of CO, CO2, butane, and pentane as HDO
products in the effluent gas from the DRIFT reactor cell during the operando experiment with
H2/valeric acid mixture. Dashed lines indicate the point where spectra were collected after
reaching the steady state at the given reaction temperature.
21

The position of the asymmetric and symmetric νO-C-O stretching bands appearing over and
below about 1500 cm-1, respectively, in addition to the difference between their peak
positions (Δν=νas-νs) are indicative of the bonding structure of carboxylate species
[36,37,39,40]. The frequency of the νas vibration and the corresponding Δν value were shown
to increase in the following order: chelating bidentate < bridging bidentate ( free ionic) <
monodentate carboxyl species. The intense pair of bands observed at 1575 and 1470 cm-1
(Δν = 105 cm-1) for the Pd/Al2O3 sample (Fig. 9B) and a similar pair of bands at 1585 and
1470 cm-1 (Δν = 115 cm-1) for the Pd/Al2O3-5P sample (Fig. 10B) can be assigned to the
asymmetric and symmetric vibrations of chelating bidentate carboxylate species
[36,37,39,40]. A second type of carboxylate species gives an intense asymmetric νO-C-O
stretching band at 1650 cm-1 (Fig. 9B). The identification of its symmetric pair is, however,
difficult due to the appearance of overlapping C–H deformation vibrations in the frequency
range below 1500 cm-1, most probably due to the appearance of the as(CH3), s(CH2), and
s(CH3) vibrations of the -CH3 and -CH2- groups of the hydrocarbon chain (the s(CH3)
vibration is clearly discernible at around 1350 cm-1) [36,40,41]. However, we found a band at
about 1390 cm-1 that showed parallel of intensity with that of the 1650 cm-1 band, if reaction
conditions were varied. It was substantiated that a band at about 1390 cm-1 is the pair of the
1650 cm-1 band, stemming from symmetric νO-C-O stretching vibration (Figs. 9B, 10B, and S4,
S5). The relatively high frequency of the asymmetric νO-C-O vibration band (1650 cm-1) and
the large frequency separation from the corresponding symmetric vibration band (Δν = 1650 –
1390 = 260 cm-1) clearly suggest that this second type of carboxylate group can be identified
as monodentate carboxylate species bonding to surface aluminum atom [39,40]. Note that the
concentration of the monodentate carboxylate species is much lower over the Pd/Al2O3-5P
sample than over the Pd/Al2O3 sample (Fig. S4, Sections C and D). Phosphatization of
-alumina surface eliminated mainly those sites, where monodentate carboxylate species
could have been formed.
Upon contacting the catalysts with VA/He mixture at 300 C gradually carboxylate bands
developed (Fig. S4) until their surface concentration reached steady state (Figs. 9B and 10B,
spectra a). Virtually the same steady state concentrations were reached at 350 C (Figs. 9B
and 10B, spectra b). As expected, no reaction products were formed in the absence of H2.
When the reactant flow was changed to VA/H2, the intensity of the bands at
1575-1585 cm-1 and 1470 cm-1 decreased, suggesting that mainly the concentration of the
corresponding bidentate carboxylate species decreased, whereas the surface concentration of
22

the monodentate carboxylate species (bands at 1650 and 1390 cm-1) hardly changed. The
consumption of the bidentate carboxyl species was accompanied by the appearance of the
main deoxygenation reaction products, such as, butane, pentane, CO and CO2. Higher
reaction temperature resulted in a higher reaction rate and, therefore, in higher product
concentrations in the effluent gas (Figs. 9C and 10C). The dominance of butane and CO in the
product mixture suggests that the main route of VA deoxygenation is the HDCO reaction. The
results suggest that the surface of the Pd/Al2O3 catalyst is covered by more reactive bidentate
carboxylate species and less reactive monodentate carboxylate species. In contrast, the surface
coverage of the Pd/Al2O3-5P catalyst by the more active carboxylate is substantially higher,
and it is much lower by the less active species than the corresponding coverages of the
Pd/Al2O3 catalyst.
Product formation was accompanied by recovery of surface OH groups as indicated by
the intensity drop of some negative OH bands (Fig. 9A and 10A). Because the reaction of
carboxylic acid and OH groups is accompanied by release of H2O, the recovery of OH groups
had to involve C–O bond hydrogenolysis. Mainly lower-frequency OH groups recovered,
showing that the less basic OH groups were involved in the formation of reactive carboxylates
[16,17].
4. Discussion
The effect of phosphatization on the alumina surface seems to be twofold: it consumes
surface Al–OH groups (Fig. 2) and also reduces the concentration of Lewis acid sites (Table
2) and consequently the concentration of the Lewis acid (Al) – Lewis base ( ) pair sites.
O
Note that latter oxygen atoms can behave as Brønsted base (proton acceptor) or Lewis base
(electron pair donor) sites depending on the nature of the adsorbate [42].
The reaction of phosphoric acid with surface OH groups is well documented [20,21]. It is
considered to be an acid-base reaction, as shown in Scheme 1A. The reaction results in the
formation of surface phosphate groups and water. Phosphate species formed on adjacent OH
groups can condensate via P–OH groups to form polymeric phosphate species [23], which
appeared as dominating species at higher ( 2.5 wt.%) phosphate loadings (Fig. 6A). It is
important to note that phosphatization results not only in the appearance of P–OH groups but
also in the formation of non-reactive terminal AlIV-OH groups (Fig. 2). A similar
phenomenon was observed in a former study and was related to the reaction between a
23

bridging OH group (Al–O(H)–Al) and phosphoric acid giving a terminal OH group and a
[H2PO4]– ligand attached to a surface Al atom [20]; however, it was not clarified how the
charge neutrality was preserved in this process.
Scheme 1. Reaction of phosphoric acid with (A) surface OH groups, and (B) Lewis acid
(Al) – Lewis base ( ) pair sites.
O
The present study evidenced that formation of new terminal AlIV–OH groups was
accompanied by the elimination of strong acid Lewis sites (i.e., coordinately unsaturated
tetrahedral Al sites) and their charge balancing Lewis base (oxide ion) pair sites. We
rationalize these observations by the reaction shown in Scheme 1B. The strong Lewis acid
sites were suggested to become reversibly reconstructed by establishing a weak bond with a
nearby oxide ion and thus they could appear more as a distorted tetrahedral ion than as a tri-
coordinated one (see left side of Scheme 1B) [16]. However, the very weak coordination bond
in the strained species can be easily broken in the presence of a base or an acid [16,43]. Our
results substantiate that phosphoric acid reacts with these Lewis acid (Al) – Lewis base ( )
O
pair sites resulting in the formation of terminal AlIV-OH species and [H2PO4]– groups
completing the coordination sphere of coordinately unsaturated tetrahedral Al sites (right side
of Scheme 1B). Note that latter oxide ions behave as proton acceptor Brønsted base sites in
the process, whereas the [H2PO4]– groups cannot be distinguished from those formed with the
involvement of surface OH groups (vide supra). The process proposed here clearly indicates
how the charge neutrality can be maintained during the formation of the terminal OH groups,
and also accounts for the elimination of Lewis acid (Al) – Lewis base ( ) pair sites
O
24

(Scheme 1B). This is in line with the suggestion of DeCanio et al. [43] for the bonding of
fluoride ions to strong Lewis acid sites upon HF treatment of alumina support.
Scheme 2. Formation of carboxylate species via the reaction of carboxylic acid with (A)
surface OH groups, (B) Lewis acid (Al) – Lewis base ( ) pair sites, and (C) possible
O
formation of aldehyde intermediate from bidentate carboxylate species.
The quasi-operando DRIFTS investigation revealed the formation of two types of surface
carboxylate species under catalytic conditions (Fig. 9-10). Bidentate carboxylate species were
formed via acid-base reaction between fatty acid and a surface hydroxyl groups to produce
stable surface carboxylate and H2O as shown in Scheme 2A [36–38]. It is most likely that the
generated water desorbs from the surface at the reaction temperatures (300 and 350 C)
applied here during the DRIFTS investigations [37]. This process shows close resemblance to
that observed for the surface reaction with phosphoric acid (see above and in Scheme 1A).
The second type of carboxylate could be identified as monodentate carboxylate group [39,40].
It is rational to think that these latter species were formed in a similar process over Lewis acid
(Al) – Lewis base ( ) pair sites than that we proposed for the surface reaction with
O
phosphoric acid (cf. Scheme 1B and 2B). The result is a monodentate carboxylate group, in
which one of the oxygen forms an ester-like bond with a surface aluminum atom, while the
other oxygen forms an H-bond with the neighboring OH group (right side of Scheme 2B)
[39,44]. The reactive adsorption of carboxylic acid on strong Lewis acid (Al) – Lewis base (
25

) pair sites is clearly supported by the good correlation between the concentration of strong
O
Lewis acid sites (determined by Py adsorption) and the integrated absorbance values of the
asymmetric νO-C-O band at 1650 cm-1, assigned to monodentate carboxylate species
(Table S2).
In relation to the catalytic properties, the most important consequence of phosphatization
is that it reduces the concentration of Al– pair sites, where the less reactive monodentate
O
carboxylate species are formed. The DRIFTS investigations showed that both types of
carboxylates are strongly-bound surface intermediates, but the bidentate carboxylate species
are more susceptible to deoxygenation reaction leading to paraffin product than the
monodentate carboxylate intermediate (Figs. 9 to 10). The dominance of the bidentate
carboxyl species on the phosphatized alumina support explains the substantially higher
deoxygenation (mainly HDCO) activity of the phosphate catalyst as compared to the
Pd/Al2O3 catalyst (Fig. 8). Note that phosphatization did not affect noticeably either the
particle size, or the morphology of the Pd particles, therefore the enhanced deoxygenation
activity and paraffin selectivity can be attributed to changes in the surface properties of the
alumina support. Nevertheless, the reaction between the active surface intermediate and
hydrogen most probably should take place at the metal/support interface where activated
hydrogen is available for the reaction.
We found that preferentially weak base OH groups were recovered during the HDO
reaction (Figs. 9A and 10A). The recovery of surface OH groups must involve the scission of
a carboxylate C–O bond. The most likely product is aldehyde (Scheme 2C), which is
considered as an important intermediate of the HDO reaction of carboxylic acids [2,5,8]. Note
that the process depicted in Scheme 2C is the reverse process than that observed for surface
carboxyl formation from aldehyde on metal oxides [36]. Our results suggest that the bidentate
carboxylate species, formed in reaction with weak base hydroxyls, are more ready to react
with hydrogen than the carboxylates formed in reaction with strong base surface sites. In this
context, replacement of OH groups by P–OH groups favorably affects the HDO activity.
Paraffin formation proceeds mainly via hydrodecarbonylation (HDCO) of the aldehyde
intermediate, whereas oxygen reduction reaction (HDH2O) of aldehyde represents a minor
reaction route. The paraffin chain of fatty acids is shortened in HDCO but it is preserved in
HDH2O reaction [2,5,7]. This latter reaction was suggested to proceed via primary alcohol
intermediate and requires hydrogenation, dehydration, and/or hydrogenolysis steps [2,8]. One
can speculate that latter reaction goes on a less complex reaction pathway in which the
26

cleavage of both C–O bonds in the surface carboxylate species occurs, however, the
justification of such reaction route still requires further investigation.
5. Conclusions
This study provides insights into the structure – activity relationships, determining the
triglyceride HDO activity of alumina-supported Pd catalysts. The Pd/-A2O3 catalyst showed
relatively good activity in the ester bond hydrogenolysis of triglyceride, but poor activity in
the consecutive deoxygenation of the obtained carboxylic acid intermediate to paraffin
product.
Surface modification of the -alumina support with phosphate resulted in surprisingly
high activity enhancement of the catalyst in the later rate-determining step of paraffin
formation. The phosphatization of the alumina surface resulted in (i) partial elimination of
basic Al–OH groups and the concomitant formation of weak acid P–OH groups on the
alumina surface, (ii) a decrease in the number of Lewis acid (Al) – Lewis base ( ) pair sites,
O
whereas (iii) it did not affect noticeably either the particle size, or the morphology of the Pd
particles on the support.
Quasi-operando DRIFTS investigation under catalytic conditions revealed that both
surface Al–OH and P–OH groups serve as sites for fatty acids to form bidentate carboxylate
groups, which can further react with hydrogen to form paraffin products. If the OH groups,
involved in the carboxylate formation are less basic, the formed carboxylate groups are more
ready to react with hydrogen.
Carboxylic acids can also react with strong Lewis acid (Al) – Lewis base ( ) pair sites
O
giving monodentate carboxylate species, having low reactivity. Since phosphatization
significantly reduced the number of these pair sites, the concentration of the less reactive
monodentate carboxylate groups decreased substantially. The dominance of reactive bidentate
carboxylate groups on the phosphatized alumina surface explains the substantially higher
HDO, mainly hydrodecarbonylation (HDCO) activity of the phosphatized-alumina-supported
Pd catalyst than the HDO activity of the Pd/-Al2O3 catalyst.
The product formation from the bidentate carboxylate surface intermediate is
accompanied by OH group recovery, suggesting that the deoxygenation reaction must start
with the hydrogenolysis of a carboxylate C–O bond. This reaction leads to the formation of
aldehyde that must be the intermediate of deoxygenation to paraffin.
27

6. References
[1]
[2]
G.W. Huber, S. Iborra, A. Corma, Synthesis of transportation fuels from biomass:
Chemistry, catalysts, and engineering, Chem. Rev. 106 (2006) 4044–4098.
https://doi.org/10.1021/cr068360d.
L. Hermida, A.Z. Abdullah, A.R. Mohamed, Deoxygenation of fatty acid to produce
diesel-like hydrocarbons: A review of process conditions, reaction kinetics and
mechanism, Renew. Sustain. Energy Rev. 42 (2015) 1223–1233.
https://doi.org/10.1016/j.rser.2014.10.099.
[3]
[4]
[5]
[6]
K.A. Rogers, Y. Zheng, Selective deoxygenation of biomass-derived bio-oils within
hydrogen-modest environments: A review and new insights, ChemSusChem. 9 (2016)
1750–1772. https://doi.org/10.1002/cssc.201600144.
M. Ameen, M.T. Azizan, S. Yusup, A. Ramli, M. Yasir, Catalytic hydrodeoxygenation
of triglycerides: An approach to clean diesel fuel production, Renew. Sustain. Energy
Rev. 80 (2017) 1072–1088. https://doi.org/10.1016/j.rser.2017.05.268.
M. Anand, S.A. Farooqui, J. Singh, H. Singh, A.K. Sinha, Mechanistic in-operando
FT-IR studies for hydroprocessing of triglycerides, Catal. Today. 309 (2018) 11–17.
https://doi.org/10.1016/j.cattod.2017.12.021.
X. Li, X. Luo, Y. Jin, J. Li, H. Zhang, A. Zhang, J. Xie, Heterogeneous sulfur-free
hydrodeoxygenation catalysts for selectively upgrading the renewable bio-oils to
second generation biofuels, Renew. Sustain. Energy Rev. 82 (2018) 3762–3797.
https://doi.org/10.1016/j.rser.2017.10.091.
[7]
[8]
[9]
A. Sonthalia, N. Kumar, Hydroprocessed vegetable oil as a fuel for transportation
sector: A review, J. Energy Inst. 92 (2019) 1–17.
https://doi.org/10.1016/j.joei.2017.10.008.
S. Khan, A.N. Kay Lup, K.M. Qureshi, F. Abnisa, W.M.A. Wan Daud, M.F.A. Patah,
A review on deoxygenation of triglycerides for jet fuel range hydrocarbons, J. Anal.
Appl. Pyrolysis. 140 (2019) 1–24. https://doi.org/10.1016/j.jaap.2019.03.005.
P. Šimáček, D. Kubička, G. Šebor, M. Pospíšil, Hydroprocessed rapeseed oil as a
source of hydrocarbon-based biodiesel, Fuel 88 (2009) 456–460.
https://doi.org/10.1016/j.fuel.2008.10.022.
[10] A.N. Kay Lup, F. Abnisa, W.M.A. Wan Daud, M.K. Aroua, A review on reactivity and
stability of heterogeneous metal catalysts for deoxygenation of bio-oil model
compounds, J. Ind. Eng. Chem. 56 (2017) 1–34.
28

https://doi.org/10.1016/j.jiec.2017.06.049.
[11] L. Boda, G. Onyestyák, H. Solt, F. Lónyi, J. Valyon, A. Thernesz, Catalytic
hydroconversion of tricaprylin and caprylic acid as model reaction for biofuel
production from triglycerides, Appl. Catal. A 374 (2010) 158–169.
https://doi.org/10.1016/J.APCATA.2009.12.005.
[12] S.D. Jackson, N.J. Casey, Hydrogenation of propyne over palladium catalysts, J. Chem.
Soc. Faraday Trans. 91 (1995) 3269–3274. https://doi.org/10.1039/ft9959103269.
[13] G. Bergeret, P. Gallezot, Particle Size and Dispersion Measurements, in: G. Ertl, H.
Knözinger, F. Schüth, J. Weitkamp (Eds.), Handb. Heterog. Catal., Wiley-VCH Verlag
GmbH & Co. KGaA, Weinheim, Germany, 2008: pp. 738–765.
https://doi.org/10.1002/9783527610044.hetcat0038.
[14] H. Knözinger, P. Ratnasamy, Catalytic aluminas: Surface models and characterization
of surface sites, Catal. Rev. 17 (1978) 31–70.
https://doi.org/10.1080/03602457808080878.
[15] G. Busca, V. Lorenzelli, V.S. Escribano, R. Guidetti, FT-IR study of the surface
properties of the spinels NiAl2O4 and CoAl2O4 in relation to those of transitional
aluminas, J. Catal. 131 (1991) 167–177. https://doi.org/10.1016/0021-9517(91)90333-
Y.
[16] G. Busca, The surface of transitional aluminas: A critical review, Catal. Today. 226
(2014) 2–13. https://doi.org/10.1016/j.cattod.2013.08.003.
[17] C. Morterra, G. Magnacca, A case study: Surface chemistry and surface structure of
catalytic aluminas, as studied by vibrational spectroscopy of adsorbed species, Catal.
Today. 27 (1996) 497–532. https://doi.org/10.1016/0920-5861(95)00163-8.
[18] K. Khivantsev, N.R. Jaegers, J.-H. Kwak, J. Szanyi, L. Kovarik, Precise Identification
and Characterization of Catalytically Active Sites on the Surface of γ-Alumina, Angew.
Chemie Int. Ed. 60 (2021) 17522–17530. https://doi.org/10.1002/ANIE.202102106.
[19] J.A.R. Van Veen, P.A.J.M. Hendriks, R.R. Andréa, E.J.G.M. Romers, A.E. Wilson,
Chemistry of phosphomolybdate adsorption on alumina surfaces. 2. The
molybdate/phosphated alumina and phosphomolybdate/alumina systems, J. Phys.
Chem. 94 (1990) 5282–5285. https://doi.org/10.1021/j100376a022.
[20] J.M. Lewis, R.A. Kydd, Adsorption mechanism of phosphoric acid on γ-alumina, J.
Catal. 132 (1991) 465–471. https://doi.org/10.1016/0021-9517(91)90163-X.
[21] C. Morterra, G. Magnacca, P.P. Demaestri, Surface characterization of modified
aluminas: III. Surface-features of PO4-doped Al2O3, J. Catal. 152 (1995) 384–395.
29