Phosphonic acid promotion of supported Pd catalysts for low temperature vanillin hydrodeoxygenation in ethanol

Article abstract of DOI:10.1016/j.apcata.2018.05.008

Bifunctional catalysts with activity for both hydrogenation and dehydration have been frequently investigated for hydrodeoxygenation (HDO). Here, we report the application of organophosphonic acids (PAs) to Pd/Al₂O₃ catalysts for low-temperature vanillin HDO. Reaction studies indicated that PA-modification significantly improved the liquid-phase HDO activity; the yield to the desirable product, p-creosol (CR), increased from 2.5% to 87% at 50 °C. This improvement was attributed to the creation of metal/acid bifunctional sites upon PA modification. In addition, HDO activity positively correlated with the Br?nsted acidity of the PA modifier, which could be tuned by adjusting the PA tail functionality.

Full text of DOI:10.1016/j.apcata.2018.05.008

AppliedCatalysisA,General561(2018)1–6
Contents lists available at ScienceDirect
Applied Catalysis A, General
journal homepage: www.elsevier.com/locate/apcata
Phosphonic acid promotion of supported Pd catalysts for low temperature
vanillin hydrodeoxygenation in ethanol
Pengxiao Hao, Daniel K. Schwartz, J. Will Medlin⁎
Chem. Biol. Eng UCB 596, University of Colorado Boulder, CO, 80309, United States
A R T I C L E I N F O
A B S T R A C T
Keywords:
Bifunctional catalysts with activity for both hydrogenation and dehydration have been frequently investigated
for hydrodeoxygenation (HDO). Here, we report the application of organophosphonic acids (PAs) to Pd/Al2O3
catalysts for low-temperature vanillin HDO. Reaction studies indicated that PA-modification significantly im-
proved the liquid-phase HDO activity; the yield to the desirable product, p-creosol (CR), increased from 2.5% to
87% at 50 °C. This improvement was attributed to the creation of metal/acid bifunctional sites upon PA mod-
ification. In addition, HDO activity positively correlated with the Brønsted acidity of the PA modifier, which
could be tuned by adjusting the PA tail functionality.
Bifunctional catalyst
Surface modification
Brønsted acid site
Hydrodeoxygenation
Vanillin
1. Introduction
effective acid component on AC; HDO activity linearly increased with
the availability of carboxyl groups on the support.
Catalytic reactions typically involve multiple steps and pathways
that often require different active-site environments [1]. One approach
to achieve high rates for complex reactions is to introduce bifunctional
or multifunctional catalysts, by integrating various functions to a single
material to synergistically activate different pathways. For example,
hydrodeoxygenation (HDO) of biomass-derived oxygenates, an im-
portant step in biofuel upgrading, usually involves hydrogenation and
hydrogenolysis as two processes that call for different catalytic func-
tions [2,3].
In order to achieve high HDO activity, late-transition-metal/
Brønsted acid bifunctional catalysts have frequently been used [2].
During the reaction, the metal component catalyzes the hydrogenation
step due to its activity for hydrogen dissociation, while Brønsted acid
sites cooperatively facilitate hydrogenolysis by activating the CeO
bond [2,4,5]. A convenient approach to introduce this bifunctionality is
to employ supported metal catalysts, where the support material carries
Brønsted acidity to synergistically catalyze HDO with the metal coun-
terpart. Therefore, support materials that intrinsically contain Brønsted
acid sites have often been applied or synthesized, e.g., silica-alumina
[6,7] and hybrid metal-organic-framework (MOF)/partially reduced
graphene oxide (PRGO) [8]. Recently, Wang el. al. reported that
amorphous activated carbon (AC) synergistically catalyzed room-tem-
perature HDO of vanillyl alcohol with Pd [9]. Reaction studies sug-
gested that the high HDO activity was due to acid sites on the support.
X-ray photoelectron spectroscopy (XPS) and temperature-programmed
desorption analysis further revealed that carboxyl groups were the
In addition to employing pristine acidic support materials, acid sites
can also be introduced by surface modification with acidic functional-
ities [10–13]. This approach maintains the intrinsic structure of the
native support, while also providing the ability to design the near-
surface environment in a controllable manner [14–16]. For instance,
phosphate/phosphonate modifiers have been applied to metal oxide
supports to introduce Brønsted acidity for improving HDO activity for
various reaction substrates [10–12]. Nelson et. al. reported that by
modifying the reducible CeO2 support with phosphates, the ratio of
redox to acidic catalytic sites could be optimized to achieve a high
activity for guaiacol HDO [10]. More recently, phosphonate modifiers
were applied to Al2O3 supports for vapor-phase HDO reactions [11,12].
By changing the tail structure, Brønsted acidity could be tuned to
strongly affect HDO activity.
To the best of our knowledge, the use of phosphonic acid mono-
layers has not been investigated for application in low-temperature li-
quid-phase reactions, where interactions between the surface and the
solvent are expected to play a significant role [17]. Ultimately, we
expect that such conditions may allow one to realize the greatest ben-
efits from use of ligand-modified catalysts, since low-temperature op-
eration can help avoid issues with ligand stability, and the ligands can
be tuned to influence solvent-catalyst interactions. In this contribution,
the organophosphonate-modification approach has therefore been ap-
plied to a commercial Pd/Al2O3 catalyst for liquid-phase HDO reactions
for the first time. We have used vanillin as the probe reactant, a direct
product of biomass pyrolysis and a model compound for bio-oil
⁎
Corresponding author.
E-mail address: Will.Medlin@colorado.edu (J.W. Medlin).
https://doi.org/10.1016/j.apcata.2018.05.008
Received 27 February 2018; Received in revised form 25 April 2018; Accepted 13 May 2018
0926-860X/©2018ElsevierB.V.Allrightsreserved.

P. Hao et al.
AppliedCatalysisA,General561(2018)1–6
upgrading [3,18–20]. The reaction has been reported to strongly de-
pend on reaction temperature [3,18,20]. Below 150 °C, vanillin mainly
undergoes hydrogenation to vanillyl alcohol (VA), or deoxygenation of
the carbonyl group to the desirable product p-creosol (CR). In order to
improve CR selectivity, various Brønsted acid materials such as N-
doped carbon [3], zeolites [20], and metal–organic frameworks (MOFs)
[13], have been used to support Pd. In addition, other reaction systems,
such as water-oil Pickering emulsions, have been reported to improve
vanillin HDO due to enhanced product separation [18]. Despite the
various approaches used, CR production generally requires a tempera-
ture higher than 90 °C; lower temperatures usually lead to complete
hydrogenation to VA, while a high selectivity to CR has rarely been
reported [19]. In this manuscript, phosphonic acid modifiers (PAs)
were applied to Pd/Al2O3 catalysts for liquid-phase vanillin HDO. By
introducing Brønsted acid sites to the metal oxide support, high yields
of CR could be observed below 50 °C. Furthermore, various PA modi-
fiers were utilized to tune the Brønsted acidity, which significantly af-
fected the HDO activity.
form a monolayer on the support surface. The mixture was then cen-
trifuged to remove the THF supernatant, followed by annealing at
120 °C in ambient air for 6 h to obtain the chemisorbed phosphonate-
modified catalyst. The catalyst was then rinsed with THF for four times
to remove any physisorbed phosphonic acid, followed by drying in air
overnight.
The PA-modified catalysts were characterized using infrared spec-
troscopy. The experiments were performed using a Thermo Scientific
Nicolet 6700 for diffuse reflectance infrared Fourier transform spec-
troscopy (DRIFTS) in an open cell with N2 flux. For each sample, the
background was collected with 400 scans at 1 cm−1 resolution, and 200
scans were used to analyze the samples. In order to further investigate
the adsorption geometry of the phosphonic acids, second derivatives of
the spectra were obtained to analyze the peak shoulders. The second
derivatives were generated using Norris derivative method in the
OMNIC software, with a 3-point segment and a gap between segments
of 7. BET measurements and inductively coupled plasma mass spec-
trometry (ICP-MS) were performed to determine the surface area and
phosphorus content following modification, respectively. BET mea-
surement was conducted using
a Micromeritics ChemiSorb 2720
2. Experimental
system. The catalyst was pretreated in-situ in 30% N2 in He at 180 °C for
two hours and cooled to room temperature, followed by N2 saturation
in the same atmosphere for measurement.
2.1. Materials
In order to compare the density of acid sites on different catalysts,
ammonia pulse chemisorption was performed in a quartz tube reactor
with a mass spectrometer (same as the CO chemisorption system). The
catalyst was pretreated at 120 °C in H2 for 30 min, followed by 270 °C in
20 sccm He flow for 3 h, and cooled down to 100 °C. Ammonia che-
misorption was carried out in 100 sccm He flow at 100 °C by manual
injection. Error bars were calculated based on replicate measurements.
The 5 wt % Pd/Al2O3, palladium black, vanillin, vanillyl alcohol, 2-
methoxy-4-methylphenol (p-creosol), phosphoric acid, styrene, acetic
acid, ammonia solution (2.0 M in ethanol), and tetrahydrofuran (THF)
were purchased from Sigma-Aldrich. Methylphosphonic acid (Alfa
Aesar), (chloromethyl)phosphonic acid (Acros Organics), methanol
(VWR), and ethanol (Decon Laboratories) were purchased from sup-
pliers as specifically indicated. All chemicals were with > = 98%
purification grades and used as received without further treatments.
Zero-grade air and ultra-high purity H2 were obtained from Airgas.
2.3. Catalytic reactions, yield, and rate calculations
2.2. Catalyst preparation and characterization
Vanillin hydrogenation/hydrodeoxygenation (HDO) reactions were
performed in a 100 mL liquid-phase semi-batch reactor at 50 °C in
200 psi H2 for one hour, with a stirring rate of 1200 rpm. The reaction
solution included 32 mL ethanol (solvent), 267 mg (0.05 M) vanillin
(reactant), 20 mg supported Pd catalysts (PdTotal:Vanillin = 1:176 mol/
mol), and 1 mL methanol (internal standard for gas chromatography
analysis). For Pd black catalyst, 2 mg (PdTotal:Vanillin = 1:88 mol/mol)
was used to reach a similar conversion with the Pd/Al2O3 catalysts after
the reaction course of 1 h. Styrene hydrogenation reactions were per-
formed in the same reactor at room temperature, with a lower H2
The surface area and dispersion of Pd catalysts were characterized
by CO pulse adsorption in a custom quartz tube reactor (6.35 mm inner
diameter) with a Pfeiffer mass spectrometer. Prior to the measurement,
the catalyst was pretreated in situ by a 30-min reduction in a 10 sccm
H2/20 sccm He flow at 120 °C, followed by 30-min purging with He at
the same temperature. CO adsorption was performed at 50 °C. The ac-
tive surface area and dispersion of the catalysts given by the mea-
surement are shown in Table 1.
The Pd/Al2O3 catalyst was further modified with phosphonic acids,
which are well-known to form organic monolayers on Al2O3 and other
oxide surfaces [21,22]. The modification was performed by immersing
the catalyst into a 10 mM phosphonic acid solution in THF with vig-
orous stirring for 16 h at room temperature. The total amount of
phosphonic acid was controlled at 5–10 times of the amount required to
pressure
of
30 psi
and
a
lower
catalyst
loading
(Pdtotal:styrene = 1:628 mol/mol) to obtain a moderate reaction rate.
Seven 1 mL liquid samples were taken from the reactor during the 1 h
reaction period for vanillin hydrogenation/HDO. For styrene hydro-
genation, a total of six samples were taken at 30 s intervals in the first
2.5 min of the reaction to calculate initial rates. For recycling reactions,
Table 1
Characterization of Pd and the Al2O3 Support upon PA Modification.
Catalysts
Pd/Al2O3
MPA@ Pd/Al2O3
H3PO4@ Pd/Al2O3
ClMPA@ Pd/Al2O3
Pd black
STY Hydrogenation Initial Ratea (s−1 mol/molTotalPd
)
2.8
0.19
81
0.7
0.04
1
3.3
0.04
40
0.5
0.02
1
2.5
0.04
41
0.3
0.02
1
2.4
0.05
63
0.2
0.01
2
0.8
0.3
0.002
CO Uptake (molCO/molTotalPd
BET Surface Area (m2/g)
PA Densityb
)
0.025
50d
N.A.
8.5
0.7
7.48
0.03
4.3
0.4
N.A.
(#PA/nm2(Uncoated-Al2O3))c
a
Reaction conditions: 0.1 M styrene, 15 mg supported Pd catalyst (PdTotal:STY = 1:628 mol/mol), 32 mL ethanol (solvent), 1 mL methanol (internal standard),
298 K, 30 psi H2, and stirring at 1200 rpm. The initial rate was measured during the reaction course of first 2.5 min.
b
For PA density and acid site density measurements, the error bars were estimated by 2–3 repeated measurements of different batch of catalysts prepared. The
same batch of catalysts were used for repeated BET measurements.
c
P content was characterized by ICP-MS analysis. The total surface area was given by BET measurement of the native Pd/Al2O3 catalyst, used as the denominator
to calculate PA density.
d
Data source: reference [11].
2

P. Hao et al.
AppliedCatalysisA,General561(2018)1–6
the catalyst was rinsed twice with 20 mL methanol in a sonication
cleaner and subsequently centrifuged. The methanol supernatant was
then poured out and the catalyst was dried in a vacuum desiccator
before reuse. The activity of the catalyst was characterized by the rate
of creosol production during the first 5 min of the reaction. With vig-
orous stirring, the solid catalyst was expected to be uniformly dispersed
in the reaction mixture; the ratio of the catalyst to the reaction solution
was thus held constant within the reactor by simultaneously sampling
liquid and catalyst with continuous stirring, and subsequently filtering
out the catalyst using a syringe filter (0.22 μm, nylon). The liquid
samples were analyzed by an Agilent 7890 A gas chromatography with
a flame ionization detector, using an Agilent HP-5 capillary column.
The concentrations of reactants and products in the reaction solu-
tion were determined from peak areas obtained from GC analysis using
corresponding response factors, which were measured using a series of
standard solutions. The concentrations of the standard solutions were
comparable to those in reaction conditions. The product yield (the ratio
of product to initial reactant) and rate calculations were both based on
concentration. To avoid the ambiguity of turnover frequency (TOF)
measurements, the rates here were reported as moles of reactant con-
sumed per mole of total metal catalyst instead of per surface site. Error
bars were calculated based on replicate reactions.
significant increase in the pyridinium band attributed to Brønsted acid
sites. This Brønsted acidity was hypothesized to be carried through
−OH groups on the PA molecules. Extensive characterization by pyr-
idinium DRIFTS at elevated temperatures indicated that the Brønsted
acidity of the modifiers could be tuned by the electron-withdrawing
group on the PA molecule [11] or via the use of ligands capable of
intramolecular hydrogen-bonding interactions [12]. It should be noted,
however, that the adsorption of PA on the support may involve various
combinations of configurations for different PAs (see Fig. S1 and as-
sociated discussion). Depending on the Brønsted and Lewis acidity of
the surface, different PA adsorption geometries could be formed, only
some of which contained Brønsted acid sites [30] (Scheme 1). Based on
the prior work, we hypothesized that Brønsted acid sites formed from
PA modification could promote liquid-phase HDO activity.
3.2. Reaction study of PA-modified Pd/Al2O3 for vanillin hydrogenation/
HDO
The liquid phase vanillin hydrogenation/HDO (Scheme 2) was
performed in a semi-batch liquid phase reactor at 323 K under a con-
stant H2 pressure of 200 psi. Because vanillin HDO was hypothesized to
be enhanced by acid sites [8,31], ClMPA was investigated as a re-
presentative modifier among other PA coatings due to its higher acidity
(Table S1). Fig. 1 shows the characteristic reaction profile of Pd/Al2O3
before and after ClMPA modification. The native Pd catalyst exhibited a
high hydrogenation activity to produce vanillyl alcohol (VA), leading to
a yield of ∼97% after 60 min, while the hydrodeoxygenation (HDO)
selectivity remained low at all conversions. After ClMPA modification,
the selectivity profile was drastically changed, reaching a creosol (CR)
yield of 87% after 60 min. The reaction pathway from vanillin to CR
had been previously proposed to be either 1) a one-step direct hydro-
genolysis of the C]O bond, or 2) a multi-step hydrogenation-hydro-
genolysis process with VA as the intermediate product [2,3]. In this
study, the reaction profile indicated that the latter mechanism domi-
nated, as VA was generated in the early stage of the reaction and
continued to form CR after the complete consumption of vanillin.
Though direct vanillin hydrogenolysis to CR could not be ruled out,
either case suggested that the high CR yield was due to an improved
HDO activity, instead of suppressing the hydrogenation pathway. In
fact, the ClMPA coating increased the initial VA production rate based
on total amount of available Pd sites (Table 2), indicating that ClMPA
was likely to enhance C]O hydrogenation as well, as discussed in more
detail below. Therefore, we hypothesized that the ClMPA modifier
improved vanillin HDO activity by introducing Brønsted acid sites, and
moderately promoted hydrogenation to VA while strongly promoting
hydrogenolysis to CR.
To further verify this assumption, ClMPA-modified Pd/Al2O3 was
treated by NH3 to poison the acid sites prior to reaction testing. The
NH3 treatment was carried out by immersing the catalyst in a 2 M NH3/
ethanol solution overnight, followed by pouring out the ethanol su-
pernatant and heating to 60 °C in ambient environment for 4 h to re-
move excess NH3 and ethanol [9]. As shown in Table 2, after poisoning
the acid sites, the catalyst performed similarly to the uncoated Pd, i.e.
the yield to CR dropped by more than 70% after 60 min of reaction
(entry 6, 7) with a lowered initial rate similar to that of the native
catalyst. The same NH3 treatment was also conducted on the native
catalyst, and showed no notable difference from the untreated case
(entry 1, 3), indicating that the NH3 treatment did not affect the Pd
surface sites relevant to the reaction. These control experiments sug-
gested that the high HDO activity on the ClMPA-modified Pd/Al2O3
could be attributed to the Brønsted acid sites introduced by the modi-
fier. Moreover, when acetic acid was used as a homogeneous acid, a
similar improvement of HDO activity was observed for both supported
and unsupported Pd catalysts (Table 2, entry 1, 2, 9, 10), again con-
firming the key role of Brønsted acid sites in vanillin HDO.
3. Results
3.1. Characterization of PA-modified Pd/Al2O3 surface
Three PA modifiers – phosphoric acid (H3PO4), methylphosphonic
acid (MPA), and (chloromethyl)phosphonic acid (ClMPA) – were used
to modify Pd/Al2O3 catalysts. As shown in Table 1, their coverages
varied from 4 to 8 molecules/nm2 in the order H3PO4 ≈ MPA >
ClMPA. Compared to previously reported values (2.1 to 5.0 molecules/
nm2) [23–25], the density measured in this study was generally higher.
The higher coverage could be attributed to PA-multilayer formation by
physisorption, as well as organophosphonate adsorption on Pd metal
sites as suggested in recent studies [11,12].
To test if the PA modification affected Pd site availability, Pd surface
area was measured by CO chemisorption. As an additional measure of
site availability, we also evaluated the catalysts for activity in liquid-
phase styrene (STY) hydrogenation. We employed STY hydrogenation
as a probe reaction, hypothesizing that the activity for this reaction
would depend chiefly on the number of surface Pd sites due to the
structure-insensitivity of this reaction [26,27]. As shown in Table 1,
while a decrease of surface sites on modified catalysts was observed by
CO adsorption, this apparent blocking effect did not affect the activity
for STY hydrogenation within the experimental error. This result was
similar to previous findings, in which the blocking effect on a Pd by
surface modifiers did not strongly affect active sites for vapor-phase
hydrogenation reactions [11,28,29]. For instance, McCue et. al. ob-
served the promotion effect of triphenylphosphine modifiers for acet-
ylene hydrogenation on Pd/TiO2 catalysts [28], despite a notable de-
crease in the number of active sites as measured by CO pulse
chemisorption. The authors partly attributed this specific promoting
effect to the inaccessibility of hollow sites and a change of electronic
structure by triphenylphosphine adsorption, which weakened CO ad-
sorption to decrease the quantity of measured active sites without
substantially decreasing the availability of sites for acetylene hydro-
genation. Here in this study, we speculated that there was a similar
effect for PA modification that maintained the STY hydrogenation ac-
tivity while blocking a portion of active sites.
The successful PA modification was further confirmed by observa-
tion of PeO stretching modes in DRIFT spectra (Fig. S1). In general, the
spectra were similar to those from previous work, which showed that
PAs introduced Brønsted acidity to Al2O3 supports (Scheme 1) [11,12].
Those previous studies used infrared spectroscopy after pyridine ad-
sorption to show that PA modification was accompanied with a
We noted that the Brønsted acid sites were hypothesized to
3

P. Hao et al.
AppliedCatalysisA,General561(2018)1–6
Scheme 1. Possible binding geometries of PA
adsorbed on Pd/Al2O3 [30].
acid and metal sites exist on the same support, confirming that the PA
modification improved the HDO activity in a bifunctional way.
We hypothesized that the degree of improvement of HDO activity by
the bifunctional acid/metal catalyst was related to the Brønsted acidity
of the modifier. As shown in Fig. 2, when different PA modifiers were
applied to the Pd/Al2O3 catalyst, the yield to the HDO product CR ex-
hibited a positive correlation with PA acidity; the highest CR yield was
obtained on ClMPA-modified catalyst, which hypothetically had the
highest Brønsted acidity. However, it should be noted that comparisons
between different coatings are complicated by the significant differ-
ences in P densities; for example, the relatively high CR yield with the
H3PO4-modifier could be related to its higher acid surface coverage
compared to ClMPA (Table S2). However, the general observation that
more acidic PAs lead to more active catalysts was consistent with
previous work for vapor-phase HDO reactions [6,9,11]. This enhance-
ment in HDO activity was found to be insensitive to temperature across
the range in the current study (Fig. S3).
Scheme 2. Reaction pathway of vanillin hydrogenation.
Reaction conditions: 0.05 M vanillin, Pd catalyst (PdTotal:Vanillin = 1:176 mol/
mol), 32 mL ethanol (solvent), 1 mL methanol (internal standard), 323 K,
200 psi H2, and stirring at 1200 rpm.
3.3. Reusability of ClMPA-modified Pd/Al2O3
The reusability of ClMPA-modified Pd/Al2O3 was tested by mea-
suring the initial CR production rate during recycling reactions. After
each run, the catalyst was rinsed twice with 20 mL methanol in a so-
nication cleaner to remove the carbonaceous surface species generated
during the reaction. The mixture was then centrifuged and the me-
thanol supernatant was poured out, followed by drying in a vacuum
desiccator before reuse. As shown in Fig. 3, the native Pd catalyst ex-
hibited a constant low CR production rate during the recycling runs. For
ClMPA-modified Pd/Al2O3, its high HDO activity was generally re-
tained, exhibiting a 1–2 order of magnitude increase compared to the
native Pd/Al2O3 catalyst. Furthermore, FT-IR indicated that the ClMPA
modifier was stable on the surface after three consecutive runs; the
original vibration modes remained after recycling reactions (Fig. S4).
4. Discussion
4.1. Effect of Brønsted acid sites on vanillin hydrogenation/HDO
As discussed above, vanillin HDO appeared to proceed mainly
through a two-step hydrogenation-hydrogenolysis process. We hy-
pothesized that: 1) PA introduced Brønsted acid sites to dramatically
accelerate the hydrogenolysis step; 2) the acids sites were required to be
adjacent to metal sites for synergetic catalysis; 3) HDO activity de-
pended on the pKa of the PA modifier.
However, in addition to activating hydrogenolysis, the Brønsted
acid sites were also likely to enhance vanillin hydrogenation, as shown
in Table 2. This was most evident on the ClMPA-coated surface where
the initial rate doubled after the modification, while upon poisoning the
acid sites by NH3 treatment, the rate decreased to a similar value as the
native catalyst (entries 1, 6, and 7). Unlike the frequently reported
catalyzing effect of Brønsted acids for the hydrogenolysis or dehydra-
tion step in HDO, the role of Brønsted acids in C]O hydrogenation had
not been generally recognized. Recently, Hensley et. al. suggested that
compared to surface hydrogen atoms, Brønsted acid sites increased the
capability of phenol hydrogenation by shuttling protons to and from the
reactant [33]. As suggested by density functional theory (DFT) calcu-
lations, the proton-mediated mechanism significantly decreased the
activation energy of ketone hydrogenation. This promotion of the hy-
drogenation step further changed the dominant reaction pathway of
Fig. 1. Reaction profile of vanillin hydrogenation/hydrodeoxygenation on a)
native and b) ClMPA-modified Pd/Al2O3 catalysts. Reaction conditions: 0.05 M
vanillin, 20 mg Pd catalyst (PdTotal:Vanillin = 1:176 mol/mol), 32 mL ethanol
(solvent), 1 mL methanol (internal standard), 323 K, 200 psi H2, and stirring at
1200 rpm.
cooperatively catalyze HDO with the active metal counterpart. To test
this hypothesis, the native Pd/Al2O3 was physically mixed with ClMPA-
modified γ-Al2O3 that contained Brønsted acid sites (entry 8); this re-
action showed least improvement of HDO activity compared to the
native catalyst. Therefore, the enhanced HDO activity required that
4

P. Hao et al.
AppliedCatalysisA,General561(2018)1–6
Table 2
Influence of Brønsted acidity on vanillin hydrogenation/hydrodeoxygenation over different catalysts.a.
Entry
Catalyst
Time (h)
Conv. (%)
CR Yieldb (%)
Initial Ratec (s−1 molVaniillin/molTotalPd
)
1
2
3
4
5
6
7
8
Pd/Al2O3
1
1
1
1
1
1
1
1
1
1
> 99
> 99
97
17
18
> 99
> 99
> 99
72
2.0
75
2.7
2.4
4.6
87
14
9.8
13
0.9
14
1.8
1.2
0.8
11
9
1.8
13
22
0.29
0.35
0.3
0.004
0.002
0.55
0.22
0.28
0.16
0.18
0.03
0.05
0.1
0.002
0.002
0.08
0.03
0.07
0.09
0.05
Pd/Al2O3, acetic acid
Pd/Al2O3, NH3 treated
Al2O3
1
7
10
Al2O3, acetic acid
ClMPA@Pd/Al2O3
ClMPA@Pd/Al2O3, NH3 treated
Pd/Al2O3&ClMPA@Al2O3
Pd black&Al2O3
9
10
15
11
Pd black&Al2O3, acetic acid
89
76
a
Reaction conditions: 0.05 M vanillin, 20 mg 5 wt% supported Pd or 2 mg Pd black (PdTotal:Vanillin = 1:176 (1:88 for Pd black) mol/mol), 32 mL ethanol
(solvent), 1 mL methanol (internal standard), 323 K, 200 psi H2, and stirring at 1200 rpm.
b
Yield to creosol was reported after 1 h of the reaction.
Initial rates were based on vanillin consumption, obtained during the reaction course of the first minute.
c
4.2. Comparison of PA-promoted HDO in liquid-phase vs. vapor-phase
conditions
The same PA-modification strategy was previously applied to im-
prove HDO of aromatic oxygenates, such as benzyl alcohol, m-cresol,
and phenol, in vapor-phase conditions [11,12]. Similar to liquid-phase
reactions, the addition of PA modifiers increased the activity and se-
lectivity for vapor-phase HDO, the activity of which could be tuned by
the acid strength or the structure of PA tail. However, some differences
were also observed between the two reaction conditions. Zhang et. al.
found that the H3PO4 modifier outperformed organophosphonic acid
modifiers at very short times on stream [11], possibly because H3PO4
produced stronger acid sites at the catalyst-vapor interface, as indicated
by temperature-dependent pyridine DRIFTS measurements. However, a
drastic deactivation was observed for the H3PO4 modified catalyst
under vapor-phase HDO conditions, due to acid-promoted coke for-
mation. In liquid-phase conditions, however, neither of these two fea-
Fig. 2. Yields of vanillin hydrogenation/hydrodeoxygenation on native and
modified Pd/Al2O3 catalysts, and Brønsted acidity of the modifiers [32]. Re-
tures was observed. In contract to vapor-phase reactions, ClMPA had
comparable initial activity with H3PO4 for liquid phase HDO. Because
acidity was proved to be critical for both liquid- and vapor-phase HDO,
we speculated that the relative acidity between ClMPA and H3PO4 may
be affected by the presence of solvent. As Gould and Xu recently re-
vealed, the choice of solvent could in fact affect the protonation ability
of acid sites on zeolites, leading to different liquid- and vapor-phase
acidity, and thus in this work, affecting the deoxygenation reaction and
possible acid-catalyzed coking [34].
action
conditions:
0.05 M
vanillin,
20 mg
Pd
catalyst
(PdTotal:Vanillin = 1:176 mol/mol), 32 mL ethanol (solvent), 1 mL methanol
(internal standard), 323 K, 200 psi H2, and stirring at 1200 rpm.
5. Conclusions
PA modification of Pd/Al2O3 significantly enhanced the rate and
selectivity of low temperature vanillin HDO. Reaction studies indicated
that vanillin HDO underwent a hydrogenation-hydrogenolysis pathway,
creating VA as the intermediate product. While vanillin hydrogenation
to VA dominated on the unmodified catalyst, PA-modification sig-
nificantly activated the HDO pathway. This activation was due to
Brønsted acid/metal synergistic catalysis that required adjacent metal
and acid sites. In addition, by changing the electron-withdrawing group
on the PA modifier, the Brønsted acidity could be varied to tune HDO
activity. While more studies are required to fully understand the reac-
tion mechanism, the current results demonstrated that metal/acid bi-
functionality was a key factor to improve HDO activity. Compared to
conventional methods that incorporate Brønsted acidity into catalytic
systems, PA-modification offered a robust strategy to modify metal
oxide-supported catalysts, adding the degree of freedom to precisely
control the near-surface environment.
Fig. 3. Reusability of the phosphonate-coated Pd/Al2O3. Note that these data
are also reported on a linear scale in the supporting information (Fig. S5).
phenol deoxygenation compared to the case without available Brønsted
acid sites.
In this study, the enhanced vanillin HDO by Brønsted acids was
hypothetically due to an activated hydrogenolysis step, or similarly to
the acid-promoted phenol deoxygenation, a possible change of reaction
pathway.
5

P. Hao et al.
AppliedCatalysisA,General561(2018)1–6
Acknowledgements
[12] P.D. Coan, L.D. Ellis, M.B. Griffin, J.W. Medlin, J. Phys. Chem. C 122 (2018)
6637–6647.
[13] F. Zhang, Y. Jin, Y. Fu, Y. Zhong, W. Zhu, A.A. Ibrahim, M.S. El-Shall, J. Mater.
Chem. A 3 (2015) 17008–17015.
[14] C.A. Schoenbaum, D.K. Schwartz, J.W. Medlin, Acc. Chem. Res. 47 (2014)
1438–1445.
[15] C.-H. Lien, J.W. Medlin, J. Catal. 339 (2016) 38–46.
[16] C.H. Lien, J.W. Medlin, J. Phys. Chem. C 118 (2014) 23783–23789.
[17] K.R. Kahsar, D.K. Schwartz, J.W. Medlin, Appl. Catal. a-Gen. 445 (2012) 102–106.
[18] S. Crossley, J. Faria, M. Shen, D.E. Resasco, Science 327 (2010) 68–72.
[19] R. Nie, H. Yang, H. Zhang, X. Yu, X. Lu, D. Zhou, Q. Xia, Green Chem. 19 (2017)
3126–3134.
This work was supported by the Department of Energy, Office of
Science, Basic Energy Sciences Program, Chemical Sciences,
Geosciences, and Biosciences Division [Grant No. DE-SC0005239]. The
authors would also like to acknowledge Patrick Coan for assistance with
DRIFTS and for useful discussions.
Appendix A. Supplementary data
[20] J. Kayalvizhi, A. Pandurangan, Mol. Catal. 436 (2017) 67–77.
[21] W. Gao, L. Dickinson, C. Grozinger, F.G. Morin, L. Reven, Langmuir 12 (1996)
6429–6435.
[22] S. Pawsey, K. Yach, L. Reven, Langmuir 18 (2002) 5205–5212.
[23] R. Luschtinetz, A.F. Oliveira, J. Frenzel, J.O. Joswig, G. Seifert, H.A. Duarte, Surf.
Sci. 602 (2008) 1347–1359.
Supplementary material related to this article can be found, in the
online version, at doi:https://doi.org/10.1016/j.apcata.2018.05.008.
References
[24] T. Hauffman, O. Blajiev, J. Snauwaert, C. van Haesendonck, A. Hubin, H. Terryn,
Langmuir 24 (2008) 13450–13456.
[1] R.A. van Santen, M. Neurock, Molecular Heterogeneous Catalysis : A Conceptual
and Computational Approach, Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim,
Germany, 2006, p. 75.
[2] A.M. Robinson, J.E. Hensley, J.W. Medlin, ACS Catal. 6 (2016) 5026–5043.
[3] X. Xu, Y. Li, Y. Gong, P. Zhang, H. Li, Y. Wang, J. Am. Chem. Soc. 134 (2012)
16987–16990.
[4] P.M. De Souza, R.C. Rabelo-Neto, L.E.P. Borges, G. Jacobs, B.H. Davis, T. Sooknoi,
D.E. Resasco, F.B. Noronha, ACS Catal. 5 (2015) 1318–1329.
[5] H. Li, Z. Fang, R.L. Smith, S. Yang, Prog. Energy Combust. Sci. 55 (2016) 98–194.
[6] C.R. Lee, J.S. Yoon, Y.W. Suh, J.W. Choi, J.M. Ha, D.J. Suh, Y.K. Park, Catal.
Commun. 17 (2012) 54–58.
[7] C. Zhao, S. Kasakov, J. He, J.A. Lercher, J. Catal. 296 (2012) 12–23.
[8] A.A. Ibrahim, A. Lin, F. Zhang, K.M. AbouZeid, M.S. El-Shall, ChemCatChem 9
(2017) 469–480.
[25] D.T. Fakult, Structural Characterization of Self-Assembled Monolayers and Their
Application in Organic Thin-Film Memory Transistors Strukturcharakterisierung
von Selbstorganisierenden Monolagen Und Ihre Anwendung in Organischen D ¨,
Erlangen, Friedrich-Alexander-Universität Erlangen-Nürnberg, 2015.
[26] J.M. Badano, C. Betti, I. Rintoul, J. Vich-Berlanga, E. Cagnola, G. Torres, C. Vera,
J. Yori, M. Quiroga, Appl. Catal. A Gen. 390 (2010) 166–174.
[27] N. Tiengchad, O. Mekasuwandumrong, C. Na-Chiangmai, P. Weerachawanasak,
J. Panpranot, Catal. Commun. 12 (2011) 910–916.
[28] A.J. McCue, F.-M. McKenna, J.A. Anderson, Catal. Sci. Technol. 5 (2015)
2449–2459.
[29] W.-J. Kim, S.H. Moon, Catal. Today 185 (2011) 2–16.
[30] P.J. Hotchkiss, S.C. Jones, S.A. Paniagua, A. Sharma, B. Kippelen, N.R. Armstrong,
S.R. Marder, Acc. Chem. Res. 45 (2012) 337–346.
[31] F. Zhang, S. Zheng, Q. Xiao, Y. Zhong, W. Zhu, A. Lin, M. Samy El-Shall, Green
Chem. 18 (2016) 2900–2908.
[32] L.D. Freedman, G.O. Doak, Chem. Rev. 57 (1957) 479–523.
[33] A.J.R. Hensley, Y. Wang, D. Mei, J.-S. McEwen, ACS Catal. 8 (2018) 2200–2208.
[34] N.S. Gould, B. Xu, J. Catal. 358 (2017) 80–88.
[9] Q. Wang, N. Gupta, G. Wen, S.B.A. Hamid, D.S. Su, J. Energy Chem. 26 (2017)
8–16.
[10] N.C. Nelson, Z. Wang, P. Naik, J.S. Manzano, M. Pruski, I.I. Slowing, J. Mater.
Chem. A 5 (2017) 4455–4466.
[11] J. Zhang, L.D. Ellis, B. Wang, M.J. Dzara, C. Sievers, S. Pylypenko, E. Nikolla,
J.W. Medlin, Nat. Catal. 1 (2018) 148–155.
6