Precisely phase-modulated VPO catalysts with enhanced inter-phase conjunction for acrylic acid production through the condensation of acetic acid and formaldehyde

Article abstract of DOI:10.1016/j.jcat.2019.04.032

A type of precisely phase-modulated vanadium phosphorus oxides (VPOs)was fabricated for the first time by quantitatively mixing the pure phase of γ-VOPO₄, δ-VOPO₄, and (VO)₂P₂O₇ in certain ratios through a mechanical milling approach. The resulting materials were used as catalysts for very efficient condensation of acetic acid and formaldehyde to acrylic acid as well as methyl acrylate. The modulated dual phase VPO catalyst outperforms the single-phase counterpart, and by optimizing the reaction conditions especially the contact time, we can further considerably enhance the catalyst performance. Over an optimized catalyst of δ-VOPO₄/γ-VOPO₄ (w/w, 3/1), the (AA + MA)yield of 84.2% with the equivalent (AA + MA)formation rate of 1.71 mmol·g^?1·h^?1 or the (AA + MA)yield of 41.8% with the equivalent (AA + MA)formation rate of 4.25 mmol·g^?1·h^?1 is achievable at 360 °C on the formaldehyde input basis. This sort of catalyst is rather durable within a period of 110 h, and fully recovered by simple air-treatment at the reaction temperature. Due to the enhanced conjunction at the interface of two phases, the surface property of phase-modulated catalyst such as the P-O bond length and surface acidity is notably modified/enhanced, accounting for the promising catalytic performance.

Full text of DOI:10.1016/j.jcat.2019.04.032

Journal of Catalysis 374 (2019) 171–182
Contents lists available at ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Precisely phase-modulated VPO catalysts with enhanced inter-phase
conjunction for acrylic acid production through the condensation of
acetic acid and formaldehyde
Jun Liu , Pengcheng Wang , Yina Feng , Zhijia Xu , Xinzhen Feng , Weijie Ji a, , Chak-Tong Au b
a
a
a
a
a,
⇑
⇑
a
Key Laboratory of Mesoscopic Chemistry, MOE, School of Chemistry and Chemical Engineering, Nanjing University, Nanjing 210023, China
Department of Chemistry, Hong Kong Baptist University, Kowloon Tong, Hong Kong
b
a r t i c l e i n f o
a b s t r a c t
Article history:
A type of precisely phase-modulated vanadium phosphorus oxides (VPOs) was fabricated for the first
Received 2 February 2019
Revised 15 April 2019
Accepted 18 April 2019
time by quantitatively mixing the pure phase of
4 4 2 2 7
c-VOPO , d-VOPO , and (VO) P O in certain ratios
through a mechanical milling approach. The resulting materials were used as catalysts for very efficient
condensation of acetic acid and formaldehyde to acrylic acid as well as methyl acrylate. The modulated
dual phase VPO catalyst outperforms the single-phase counterpart, and by optimizing the reaction con-
ditions especially the contact time, we can further considerably enhance the catalyst performance. Over
Keywords:
4 4
an optimized catalyst of d-VOPO /c-VOPO (w/w, 3/1), the (AA + MA) yield of 84.2% with the equivalent
Vanadium phosphorus oxide
Phase-modulation
Inter-phase conjunction
Acetic acid
Formaldehyde
Condensation
À1 À1
À1 À1
(
+
AA + MA) formation rate of 1.71 mmolÁg Áh or the (AA + MA) yield of 41.8% with the equivalent (AA
MA) formation rate of 4.25 mmolÁg Áh is achievable at 360 °C on the formaldehyde input basis. This
sort of catalyst is rather durable within a period of 110 h, and fully recovered by simple air-treatment at
the reaction temperature. Due to the enhanced conjunction at the interface of two phases, the surface
property of phase-modulated catalyst such as the P-O bond length and surface acidity is notably modi-
fied/enhanced, accounting for the promising catalytic performance.
Acrylic acid
Ó 2019 Elsevier Inc. All rights reserved.
1
. Introduction
As a versatile unsaturated organic compound, acrylic acid (AA)
formaldehyde becomes relatively low-priced, especially due to the
overproduction capacity of acetic acid. It is highly desirable to
develop new process using acetic acid as a starting material to
the downstream and value-added product such as acrylic acid. In
recent years, one-step condensation of acetic acid and formalde-
hyde to acrylic acid has received particular attention [14–17].
There are different kinds of catalysts applied for the condensa-
tion, such as basic or acidic oxides, and acid-base bi-functional
oxide catalysts [18–20]. It was reported that catalysts consisting
of acidic and/or basic oxides are active for the target reaction
[21–23]. For example, Ai found that the condensation of formalde-
hyde with acetic acid/methyl acetate [24–26] or acetone [27] to
acrylic acid (and its derivative) or methyl vinyl ketone can be effec-
is widely used in manufacture of paint additives, adhesive, textile,
and leather treating agents [1,2]. Furthermore, it is also extensively
utilized in manufacture of industrial monomers, methyl acrylate,
butyl acrylate, ethyl acrylate, 2-hydroxyethyl ester, and polymers
etc. [3,4]. Currently, most of acrylic acid is produced by the conven-
tional two-step oxidation of propylene via acrolein intermediate
[
5–10]. However, the feedstock of propylene is petroleum-based,
and the volatility of world oil market raises a strong motivation
to develop alternative route of acrylic acid production particularly
from renewable raw materials [11–13]. AA generation through the
condensation of acetic acid with formaldehyde is considered as a
promising approach. Acetic acid is mostly fabricated via the Mon-
santo process (catalytic carbonylation of methanol) whereas
formaldehyde is produced by the oxidation of methanol. Methanol
is now obtainable in huge scale from natural gas or coal especially
in North America and Asian regions. Therefore, both acetic acid and
2 5 2 5
tively catalyzed by the V O -P O type catalysts. Ai et al. [25]
reported a yield of 65% for acrylic acid and methyl acrylate (on
formaldehyde input basis) in condensation of acetic acid with
methanol over a V/Ti/P ternary oxide catalyst. Gogate et al. [28]
observed that a methyl methacrylate yield of 10.5% (based on
methyl propionate) would be achievable in the condensation of
methyl propionate with formaldehyde over a V-Si-P ternary cata-
lyst. The vanadium phosphorus oxide (VPO) catalyst is recognized
as one of the most complicated catalyst systems, and many efforts
⇑
Corresponding authors.
E-mail addresses: fengxinzhen1008@163.com (X. Feng), jiwj@nju.edu.cn (W. Ji).
https://doi.org/10.1016/j.jcat.2019.04.032
021-9517/Ó 2019 Elsevier Inc. All rights reserved.
0

1
72
J. Liu et al. / Journal of Catalysis 374 (2019) 171–182
have been made to advance our understanding of catalyst nature in
the partial oxidation of n-butane to maleic anhydride [29–42].
Very recently, Ivars-Barceló et al. reported the distinct functions
of bulk phase, near surface, and outermost atomic layer of VPO cat-
alysts in the oxidative dehydrogenation of ethane [43]. The VPO
type catalysts have been employed for conversion of acetic acid
phoric acid (85%) was added drop wise to reach a P/V atomic ratio
being 1.05/1.0. The suspension was refluxed for another 6 h, and
then the solid was filtered out, and washed with acetone. It was
further dried at 100 °C in air for 24 h to obtain catalyst precursor.
Before characterization and performance evaluation, all the pre-
cursors were activated according to the following procedures.
The activation of catalyst precursor was performed in a quartz
tube reactor at different flowing atmospheres and temperatures
to obtain distinct phase of VPO. Generally, the precursor powder
of 10 g was charged into the reactor which was set in the thermo-
static zone of a tubular furnace.
(
methyl acetate) and formaldehyde to acrylic acid (methyl acry-
late) via a condensation route. Yan et al. [44] reported a AA yield
of 21.9% with the selectivity of ca. 98% in the condensation of acetic
acid and formaldehyde over a VPO catalyst. The supported VPO
catalyst had also been investigated on the condensation of
formaldehyde with carboxylic acid and the corresponding esters
(a) The precursor was activated at 400 °C for 15 h under a flow-
[
19,45], as well as acetic acid/methyl acetate [46–48]. Li and
ing atmosphere (certain volume percentage of O
, 40 mL/min) [60]. A series of VPOs comprising essentially d-
VOPO phase were obtained.
(b) The precursor was activated at 680 °C for 12 h under a flow-
ing atmosphere (certain volume percentage of O in the balanced
, 60 mL/min) [60]. A series of VPOs consisting of essentially
VOPO phase were obtained.
2
in the balanced
coworkers recently reported the deactivation behavior on the
VPO and VPO-Zr catalysts in the condensation of methyl acetate
and formaldehyde [49].
N
2
4
In our previous studies [18,50–59], efforts were made on prepa-
ration chemistry and catalytic behaviors in partial oxidation of n-
butane to MA over various VPO catalysts. It is obvious that the
preparation history of VPO type catalyst can dramatically impact
on its physicochemical properties such as phase composition, mor-
phology, particle size, surface P/V ratio, and vanadium oxidation
state. Notably, we found out that the efficient condensation of
formaldehyde with acetic acid or methyl acetate can be accom-
plished over the controllably activated VPO catalysts using differ-
ent reactors under the specific atmospheres [18], and the d-
2
N
2
c-
4
(c) The precursor was activated at 400 °C for 15 h under a flow-
ing atmosphere (1.5 vol% n-butane/air, 90 mL/min) [18]. A VPO
sample comprising essentially (VO)
The phase-modulated VPOs containing quantified combination
of binary phases, i.e., (VO) /d-VOPO , (VO) -VOPO , or
d-VOPO -VOPO with a mass ratio of 1/3, 1/1, and 3/1 respec-
2 2 7
P O phase was obtained.
P
2 2
O
7
4
P
2 2
O
7
/c
4
4
/c
4
tively, were obtained through mechanical ball-milling for a certain
period. In the mechanical ball-milling process, 50 little agate balls
were added into a 50 mL agate jar and 25 mL cyclohexane was
served as milling medium. To clarify the inter-phase conjunction
over the phase-modulated VPOs subjected to mechanical ball-
milling, a series of reference VPOs with the same phase constitu-
VOPO
target reaction. In the current work, we first synthesized the VPO
samples of essentially pure phase, including d-VOPO -VOPO
and (VO) , in a controlled way; then the phase composition
4
phase was identified for the first time to be critical for the
4
,
c
4
,
2 2 7
P O
of a catalyst was precisely modulated by quantifying the ratio of
different pure phases. The inter-phase conjunction was further sys-
tematically tuned by mechanically ball milling various phase-
modulated samples. The derived materials were subjected to calci-
nations under the controlled atmospheres and evaluated for the
condensation of formaldehyde with acetic acid. The results demon-
strated that the phase constitution and the involved inter-phase
conjunction did show a significant impact on the catalytic behavior
of the received materials. The highest formation rate of acrylic
acid/methyl acrylate was achieved so far over the specifically
phase-modulated, mechanically ball-milled, and controllably acti-
vated catalyst. The current study disclosed a successful approach
to developing highly efficient VPO type catalyst and tuning unique
inter-phase conjunction significant for the desired reaction path-
tions were made through simple mixing, i.e.,
4
c-VOPO and d-
VOPO (w/w, 1/3) or (VO) and d-VOPO
4
2
P
2
O
7
4
(w/w, 1/1) were
added into a 50 mL round-bottomed flask and 25 mL cyclohexane
was added as the mixing medium. Then the mixture was stirred
for 12 h followed by the same drying and activation.
2.3. Characterization
X-ray powder diffraction (XRD) patterns were recorded on a
Philips X’Pert MPD Pro X-ray diffractometer with graphite-
monochromatized Cu Ka radiation (k = 0.1541 nm). Raman spectra
were recorded at RT on an InVia-Reflex Raman spectrometer (laser
source: 488 nm). Scanning electron microscopy (SEM) images were
3
1
way. Techniques including XRD, Raman, XPS, H
2
-TPR, NH
3
-/CO
2
-
taken on a SEM instrument (S-4800). The P magic angle spinning
solid state NMR spectra were recorded on a Bruker AVANCE AVIII
WB 400 spectrometer at a resonance frequency of 162.31 MHz
with a 4 mm MAS NMR rotor, a pulse width of 20 ms, and a delay
time of 5 s between scans were applied. The BET surface areas were
measured on a ASAP 3020 material physical structure determina-
3
1
TPD, and P magic angle spinning solid state NMR were employed
to clarify the detailed catalyst characteristics and to establish the
structure-performance correlation.
2
. Experimental
tor. The samples were degassed at 573 K for 6 h, after which N
adsorption was conducted at 77 K. Hydrogen temperature-
programmed reduction (H -TPR) was carried out using a U-
2
2
.1. Chemicals
2
shaped quartz reactor. The catalyst (50 mg) was first pretreated
in an Ar flow (40 mL/min) at 100 °C for 1 h. After the temperature
was cooled down to room temperature (RT), the catalyst was
heated from RT to 850 °C at a rate of 10 °C/min in a flow of 5%
The benzyl alcohol (ꢀ99.0%), PEG-6000, vanadium pentoxide
(
2
V O
5
), phosphoric acid (H
3
PO
4
, 85%), acetic acid (ꢀ99.0%) are ana-
lytical grade. There is a small fraction of methanol (ca. 6%) existing
in the commercially available formaldehyde solution (37%) serving
as a stabilizing agent.
H
2
/Ar (v/v, flow rate = 40 mL/min), and the dehydrated effluent
gas was analyzed using a thermal conductivity detector (TCD).
The CuO was used as a reference to calibrate the H consumption.
X-ray photoelectron spectroscopy (XPS) measurement was per-
formed on a PHI5000 Versa Probe instrument with Al K radiation.
2
2
.2. Catalyst preparation
a
The preparation of catalyst precursors was described previously
The binding energy (BE) was calibrated against the C1s signal
(284.6 eV) of contaminant carbon. Elemental surface composition
was estimated based on the peak areas normalized using Wagner
[
18,53]. Benzyl alcohol was employed as the preparation media.
was first refluxed in benzyl alcohol at 140 °C for 5 h, after that
certain amount of PEG 6000 was introduced. One hour later, phos-
2 5
V O
factors.
Ammonia
temperature-programmed
desorption

J. Liu et al. / Journal of Catalysis 374 (2019) 171–182
173
(
NH
catalyst (50 mg) was first pretreated in a He flow (40 mL/min) at
00 °C for 1 h. After the temperature was cooled down to 100 °C,
NH adsorption was performed at 100 °C for 1 h. Then, the catalyst
was purged with a He flow until a steady baseline was obtained.
Finally, NH -TPD was carried out in a He flow (40 mL/min) with
the sample being heated to 450 °C at a rate of 20 °C/min, and the
effluent gas was analyzed by TCD. The amount of desorbed NH
3
-TPD) was carried out using a U-shaped quartz reactor. The
Detailed examples how to determine the HCHO conversion, (AA
+ MA) yield/formation rate, selectivity to all the products as well as
carbon balance are demonstrated in the Supplementary Material.
In our previous study [18], the issue of mass transport limita-
tion had been studied carefully, the results indicated that the
impact of internal and external diffusion was proved to be insignif-
icant under the applied operating conditions. In the current study,
the operating conditions are nearly identical to those employed in
our previous study; therefore, the mass transport limitations are
essentially eliminated.
2
3
3
3
(
(
(
in
0.01 mol/L) was used to absorb the released NH
0.01 mol/L) was used as the titrant. CO -TPD was performed sim-
lmol/g) was determined by a titration, in which a HCl solution
3
. A NaOH solution
2
ilarly and the effluent gas was analyzed by TCD.
For the V element of different oxidation state, the relative sur-
face concentration can be estimated through deconvolution analy-
sis of a specific XPS peak. For the same batch of sample measured
employing identical conditions as well as the same parameters
3. Results and discussion
3.1. Xrd
4
+
5+
adopted for deconvolution analysis, comparison of the V /V
The XRD patterns of the VPO samples activated under various
conditions plus the reference phase PDFs are shown in Fig. S1.
The XRD patterns of the three single-phase VPO samples and three
phase-modulated VPO samples are displayed individually with the
PDF(s) of the related reference phase(s) (Fig. 1). In some cases, the
phase-modulated VPO samples prepared via simple mixing and
mechanical ball milling are displayed together for direct compar-
ison (Fig. 1e-f). According to the literature works [18,60,61], the
phase composition of catalyst can be significantly affected by the
applied activation conditions (atmosphere and temperature). As
shown in Fig. 1a, all the diffraction lines (at 2h = 19.5°, 22.0°,
ratios over different samples is reasonable. The deviation of such
analysis is generally < ±5% in relative amount.
2.4. Catalyst evaluation
Catalyst evaluation was carried out at atmospheric pressure. All
the catalyst powders were pressed, crushed, and sieved to 20–40
mesh for activity evaluation. Two reactors were used for catalyst
evaluation, one has an ID of 10 mm without a thermocouple jacket,
and the other has an ID of 12 mm with a thermocouple jacket
whose outside diameter is 3 mm. The reaction data derived from
the two reactors were proved to be reproducible. Catalyst of 3 g
was charged into the reactor, and the space above the catalyst
bed was filled with quartz chips to preheat the in-coming liquid.
Before feedstock introduction, the sample was heated up in a flow
4
24.2°, 28.5°, etc.) are typical of the d-VOPO phase (PDF #47-
0951), corresponding to the (0 0 2), (1 1 1), (0 1 2), and (0 2 0)
plane, respectively [60]. Similarly, the diffraction lines (at
2h = 18.1°, 22.7°, 23.2°, 25.4°, etc.) presented in Fig. 1b are typical
of the
4
c-VOPO phase (PDF #47-0950), corresponding to the
(0 3 1), (2 2 1), (0 4 0), and (2 3 0) plane, respectively [60]. The
of N
2
(30 mL/min) to a desired temperature at a rate of 10 °C/min
diffraction lines (at 2h = 23.0°, 28.4°, 29.9°, etc.) in Fig. 1c are typ-
and kept at this temperature for 2.5 h. When a mixed solution of
acetic acid and formaldehyde (2.5/1, n/n) was fed, a mixture of
ical of the (VO)
(2 0 0), (0 2 4), and (0 3 2) plane, respectively [18]. The observa-
tions indicate that nearly pure d-VOPO -VOPO , and (VO)
entities are obtainable when the precursors are activated under
the conditions of pure O (40 mL/min) at 400 °C, 75 vol% O /N
2 2 7
P O phase (PDF #50-0380), corresponding to the
N
2
and air was served as carrier gas. The liquid feed rate was
.33–6.65 mL/h (formaldehyde feed rate, 6.1–30.5 mmol/h). Under
the reaction conditions, each component (acetic acid, formalde-
hyde, N , and O ) exists in gaseous state and its feed concentration
is indicated in vol% (mol%) together with the total gas hourly space
4
,
c
4
2 2 7
P O
1
2
2
2
2
2
(60 mL/min) at 680 °C, and 1.5 vol% n-butane/air (90 mL/min) at
400 °C, respectively. In addition, as shown in Fig. S1a, when the
À1
À1
velocity (h ). Typically, the GHSV of 1496 h and feed composi-
tion of 15.3/6.1/17.3/2.2/71.5 (molar fraction) was employed. The
products were collected in a cold trap. After 2.5-h reaction, the col-
lected liquid phase was analyzed using a gas chromatograph
equipped with a flame ion detector (FID) and a HP-FFAP capillary
column (0.32 mm  25 m). Valeric acid and iso-butyl alcohol were
used as internal standards for component quantification. All the
catalysts were first evaluated by screening their performances in
terms of the (AA + MA) yield in the collected liquid sample based
on formaldehyde input. Further evaluations were made on a few
representative catalysts. In these circumstances, the off-gas was
on-line analyzed by a GC equipped with TCD and TDX-01 packed
column. It is worth noting that the formaldehyde component can-
not be measured by GC analysis, therefore, the formaldehyde con-
version cannot be directly determined by using the GC analysis
data. In some cases, the unreacted HCHO content was analyzed
by the iodometry method, to directly determine the HCHO conver-
sion. Yield value is usually dependent upon a few parameters such
as catalyst amount, concentration of reactants, gas hourly space
velocity (GHSV), and time on stream (TOS). In order to minimize
the influence of these parameters in comparison of catalyst activ-
ity, we generally reported the formation rate(s) for the desired pro-
duct(s), namely, the amount of product(s) (in mmoles) generated
on per unit mass of catalyst per hour.
precursor is activated under pure O
new diffraction peak (2h = 11.9°) appears and could be ascribed
to the (0 0 1) plane of VOPO O phase (PDF #36-1472). The
Á2H
oxygen rich atmosphere favors the combination of -VOPO and
O to generate VOPO O. On the other hand, as shown in
Á2H
Fig. S1b, when the precursor is activated under 75 vol% O /N
(40 mL/min) at 400 °C, a diffraction peak at 2h = 23.0° is observed,
corresponding to the (2 0 0) plane of the (VO) phase (PDF
2
(60 mL/min) at 680 °C, a
4
2
c
4
H
2
4
2
2
2
2 2 7
P O
#50-0380). In this case the oxygen concentration in the activation
atmosphere is not high enough to fully oxidize the V4+ species. As
demonstrated in Fig. S1c and d, when the precursor is activated
under pure N
phase (PDF #50-0380 and 53-1051). Clearly, when the precur-
sor is activated under an oxygen-free atmosphere, it is usually
transformed into the (VO) phase no matter when activated
2 2 2
, all the diffraction lines correspond to the (VO) P -
O
7
2 2 7
P O
at 400 °C or 680 °C. When the activation atmosphere becomes
oxygen-rich, the sample activated at 400 °C would comprise a high
fraction of d-VOPO
oxygen, the essentially pure d-VOPO
-VOPO entity is obtainable when the precursor is activated under
an atmosphere of 75 vol% O
When the pure d-VOPO
4
phase. If the precursor is activated in the pure
4
entity is obtained. The pure
c
4
2
/N
2
at a higher temperature of 680 °C.
, and (VO) entities are
4
,
c-VOPO
4
2 2 7
P O
mixed with each other in certain ratios and subjected to mechan-
ical ball milling or simple stirring, no additional diffraction lines

1
74
J. Liu et al. / Journal of Catalysis 374 (2019) 171–182
Fig. 1. XRD patterns of the pure phase and phase-modulated VPOs: (a) d-VOPO
4
(activated in O
2
, 400 °C), (b)
c
-VOPO
4
(activated in 75 vol% O
2 2 2 2 7
/N , 680 °C), (c) (VO) P O
(
activated in 1.5 vol% n-butane/air, 400 °C), (d) (VO) -VOPO (w/w, 1/3), (e) (VO)
2
P
2
O
7
/
c
4
2
P
2
O
7
/d-VOPO
4
(w/w, 1/1), (f)
c
-VOPO
4
/d-VOPO (w/w, 1/3).
4
are observed except for those of the constitutional
4
c-VOPO , d-
3.2. Raman
VOPO , and (VO) phases, suggesting that the applied mechan-
4
2 2 7
P O
ical ball milling or simple stirring does not alter the original phases
or generate the new species. Notably, however, there are still visi-
ble differences in the diffraction intensity of ball-milled and simply
stirred samples (Fig. 1e and f). For instance, in the ball-milled sam-
The Raman spectra of the six representative samples are
demonstrated in a similar way. There are no standard Raman spec-
tra of the single-phase VPO samples available, but as described in
the experimental section, these single-phase samples are carefully
synthesized according to the procedures of the literature works
[18,60], and their Raman spectra are also carefully checked with
respect to those reported in the same literature (Fig. 2a–c) [60].
ple the diffraction peak of the (0 0 4) plane (
ously weakened whereas the diffraction peak of the (1 1 1) plane
d-VOPO ) is considerably enhanced (Fig. 1f). The mechanical ball
4
c-VOPO ) is more obvi-
(
4
milling is more energetic than the simple stirring and can more
effectively change the particle size and morphology, resulting in
alteration of facet orientation of mixed crystallites and enhanced
contact between the facets of different phases.
Therefore, the character of single-phase composition of d-VOPO
4
,
c
-VOPO , and (VO) samples is further confirmed. On account
4
2 2 7
P O
of the recognitions, the Raman bands of the phase-modulated VPO
samples are assigned in detail (Fig. 2d-f), and the trend of phase

J. Liu et al. / Journal of Catalysis 374 (2019) 171–182
175
Fig. 2. Raman spectra of the pure phase and phase-modulated VPOs: (a) d-VOPO
4
(activated in O
2
, 400 °C), (b)
c
-VOPO
4
(activated in 75 vol% O
2 2 2 2 7
/N , 680 °C), (c) (VO) P O
(
activated in 1.5 vol% n-butane/air, 400 °C), (d) (VO) -VOPO (w/w, 1/3), (e) (VO)
2
P
2
O
7
/
c
4
2
P
2
O
7
/d-VOPO
4
(w/w, 1/1), (f)
c
-VOPO
4
/d-VOPO (w/w, 1/3).
4
contact/conjunction is visibly reflected in the corresponding
Raman spectra (band shape change-noticeably narrower and more
symmetric as well as band shift), especially when the samples sub-
jected to simple mixing and mechanical ball milling are directly
compared (Fig. 2e, f), demonstrating the generation of structurally
more uniform and intensified conjunction between the two phases.
(b) only exhibit one peak whereas the spectra (c), (d), (e), and (f)
can be deconvoluted into two separate components. According to
the previous studies [63,64,83], the BE of V2p3/2 around 516.6 eV
4
+
5+
and 517.6 eV can be assigned to that of V 2p3/2 and V 2p3/2
respectively. Therefore, it is clear that the V species in the VPO cat-
alyst activated at 680 °C under an atmosphere of 75 vol% O /N is
,
2
2
5
+
in the form of V , whereas the V species in the VPO catalyst acti-
3
.3. Xps
vated at 400 °C under an atmosphere of 1.5 vol% n-butane/air is in
4
+
the form of V . In the case of spectrum (c), there is a minor peak at
4+
The surface elemental composition of the constitutional
c
-
516.0 eV, suggesting the presence of small quantity of V in the
4
+
5+
VOPO d-VOPO (VO) and the corresponding phase-
4
,
4
,
2
2
P O
7
4
nearly pure d-VOPO counterpart. The V 2p3/2 and V 2p3/2 sig-
modulated samples was measured by XPS. The binding energies
are calibrated with respect to 284.6 eV to eliminate the impact of
testing circumstance [62]. As shown in Fig. 3, the spectra (a) and
nals can be clearly observed on spectra (d), (e) and (f).
Through the deconvolution analysis, the surface elemental com-
position can be estimated over the different samples and the

1
76
J. Liu et al. / Journal of Catalysis 374 (2019) 171–182
broadening and resonance shift occurs associating with the 31P
solid-state NMR signals of phosphorus atoms in the proximity of
paramagnetic vanadium species in VPO catalysts.
3
1
Fig. 5 shows the P solid-state MAS NMR spectra of the binary
phase modulated catalysts of (VO) /d-VOPO (w/w, 1/1) pre-
2 2
P O
7
4
pared through a mechanical milling or simple stirring. All the spec-
tra display the main peaks at À17.6 and À8.4 ppm, attributable to
4
the presence of d-VOPO [60,75]. Notably, there is a small peak at
around 27 ppm in spectrum (a); however, this signal is absent in
spectra (b) and (c). It is supposed to be the association between
4
+
5+
P and V /V ions [70,76]. In other words, the applied mechanical
4+
5+
ball-milling process of the current study enhances the V -V
4
+
5+
interaction and also the coupling between P and V /V ions as a
result of intensified inter-phase contact/conjunction. In terms of
4
+
5+
the catalyst activities (Section 4), the enhanced V -V interaction
caused by ball-milling process is favorable for the reaction.
3
1
Fig. 6 exhibits the P solid-state MAS NMR spectra of the binary
phases modulated catalysts of -VOPO /d-VOPO (w/w, 3/1) pre-
pared through a mechanical milling or simple stirring. According
to the literature [41,58], the -VOPO entity has two characteristic
entity has two
c
4
4
Fig. 3. Curve-fitting analysis of the V2p3/2 peaks of (a) (VO)
.5 vol% n-butane/air, 400 °C), (b) -VOPO (activated in 75 vol% O
d-VOPO (activated in O , 400 °C), (d) (VO) /d-VOPO (w/w, 1/1, ball-milled), (e)
/d-VOPO (w/w, 1/3, ball-milled), and (f) -VOPO /d-VOPO (w/w, 1/3,
2
P
2
O
7
(activated in
c
4
1
c
4
/N , 680 °C), (c)
2 2
4
2
2
P
2
O
7
4
peaks at À21.2 and À17.3 ppm, and the d-VOPO
4
c
-VOPO
4
4
c
4
4
characteristic peaks at À17.6 and À8.4 ppm, respectively. All spec-
tra in Fig. 5 show two main peaks at À13.6 and À21.5 ppm, differ-
stirred).
ent from that of individual
the result of strong coupling effect between
c
-VOPO
4
or d-VOPO
4
. It is thought to be
and d-VOPO
4
c-VOPO
4
results are summarized in Table 1. The measured surface P/V ratios
of all catalysts are found to be higher than the nominal value. This
is expected because the P element is usually enriched on the VPO
caused by the efficient milling process. In addition, there is a shoul-
der at À9.5 ppm in spectra (a) and (b), and it becomes more inten-
sive in the former spectrum corresponding to an extended milling
process. It is believed that this signal is closely related to the inter-
type catalyst [53,65,66]. Over the constitutional d-VOPO
4
, the
4
+
5+
V /V ratio is approximately 0.16, whereas on the constitutional
, neither V5 in the former sample nor V
+
4+
4 4
phase conjunction between c-VOPO and d-VOPO which is inten-
(
2 2 7 4
VO) P O and c-VOPO
sified by an extended milling. Note also that, the trend of signal
intensity at À9.5 ppm fits that of activity order (Section 4), indicat-
ing an inherent correlation between the activity and the inter-
phase contact/conjunction.
in the latter one is detectable.
3.4. Sem
Fig. 4 shows the typical SEM images of the constitutional
VOPO , d-VOPO , and (VO) components and the phase-
modulated samples subjected to mechanical milling. The
VOPO entity is ‘‘rose-like’’ in morphology with the assembled
plates of 1–2 m whereas the (VO) and d-VOPO counterparts
are irregular blocks in morphology with the particle dimension of
ca. 1.5–3.5 or 1.5–4.0 m, respectively. On the other hand, all the
phase-modulated VPO catalysts after mechanical milling exhibit
fine particle in morphology with the average particle size of 0.3–
0.5 lm. The ball-milling process cracks the primary particles, thus
c-
4
4
2
2
P O
7
3.6. H -Tpr
2
c
-
4
The H -TPR profiles over the typical catalysts were measured
2
l
2
P
2
O
7
4
and the results are shown in Fig. 7. With the calibration of the
reduction peak (H2 consumption), the reducibility of surface V5+
is estimated and the results are summarized in Table S1. Clearly,
l
5
+
4+
the reduction of surface V /V species is closely associated with
the specific phase structure as well as the inter-phase conjunction
especially over the phase-modulated samples.
reduces the particle size and also smoothens somewhat the parti-
Within the adopted temperature range, the reduction peaks
cle morphology.
centered around 500–600 °C in the TPR profiles can be attributed
5
+
to the reduction of V species, while the reduction peaks in the
3
1
4+
3
.5. P solid-state MAS NMR
700–850 °C range attributable to the reduction of V species. Both
c
-VOPO
the constitutional V species (Fig. 7e, f) whereas the (VO)
displays the major reduction peak corresponding to the constitu-
4 4
and d-VOPO show their reduction peaks attributable to
5
+
The solid-state NMR spectroscopy was used to study the VPO
type catalysts [67–75], because it is highly sensitive to the local
structures of V/P elements. According to the literature [75], line
2 2 7
P O
4
+
tional V species and almost no reduction peak appears in the
Table 1
XPS data of the various VPO catalysts.
Catalyst
Relative surface concentration (at%)
P/V ratio
V4+/V5+ ratio
C1s
V2p
P2p
O1s
a
b
c
d
e
f
23.0
21.8
13.6
15.1
26.7
33.2
8.5
7.3
8.3
8.3
6.8
7.0
11.5
11.2
12.9
13.8
11.3
11.0
57.0
59.8
65.1
62.9
55.3
48.8
1.35
1.53
1.55
1.66
1.66
1.57
/
/
0.16
1.30
0.11
0.16
(
2
a) (VO) P
2
O
7
(activated in 1.5 vol% n-butane/air at 400 °C), (b)
c
-VOPO
4
(activated in 75 vol% O
2
/N
2
at 680 °C), (c) d-VOPO
4
2 2 2 7 4
(activated in O at 400 °C), (d) (VO) P O /d-VOPO
(w/w, 1/1, ball-milled), (e)
c
-VOPO /d-VOPO (w/w, 1/3, ball-milled), (f) c-VOPO /d-VOPO (w/w, 1/3, stirred).
4
4
4 4

J. Liu et al. / Journal of Catalysis 374 (2019) 171–182
177
Fig. 4. SEM images of (a) (VO)
2
P
2
O
7
(activated in 1.5 vol% n-butane/air, 400 °C), (b)
c
-VOPO
4
(activated in 75 vol% O
2
/N
2
, 680 °C), (c) d-VOPO
(w/w, 1/3).
4
(activated in O
2
, 400 °C), (d)
ball-milled -VOPO /d-VOPO (w/w, 1/3), (e) ball-milled (VO)
c
4
4
2
P
2
O
7
/d-VOPO (w/w, 1/1), and (f) ball-milled (VO)
4
2
P
2
O
7
/
c
-VOPO
4
Fig. 5. 31P solid-state MAS NMR spectra of (VO)
O
(w/w, 1/1) milled or
Fig. 6. 31P solid-state MAS NMR spectra of
c
-VOPO
2
P
2
7
/d-VOPO
4
4 4
/d-VOPO (w/w, 1/3) milled or
simply mixed for different time: (a) 24 h ball-milled, (b) 4 h ball-milled, and (c) 12 h
simply mixed.
simply mixed for different time: (a) 24 h ball-milled, (b) 12 h ball-milled, (c) 4 h
ball-milled, and (d) 12 h simply mixed.

1
78
J. Liu et al. / Journal of Catalysis 374 (2019) 171–182
Fig. 7. H
-VOPO
-VOPO /d-VOPO
80 °C), (f) d-VOPO
2
-TPR profiles of (a) (VO)
(w/w, 1/3, ball-milled), (c)
(w/w, 1/3, ball-milled), (e)
(activated in O , 400 °C), (g) (VO)
2
P
2
O
7
/d-VOPO
-VOPO
-VOPO
4
(w/w, 1/1, ball-milled), (b) (VO)
/d-VOPO (w/w, 1/3, stirred), (d)
(activated in 75 vol% O /N
(activated in 1.5 vol% n-
2
-
Fig. 8. NH
/d-VOPO
milled), (d) -VOPO
% n-butane/air, 400 °C), (f)
VOPO (activated in O
3
-TPD profiles of (a)
c
-VOPO
4
/d-VOPO
4
(w/w, 1/3, ball-milled), (b) (VO)
-VOPO (w/w, 1/3, ball-
(activated in 1.5 vol
/N , 680 °C), (g) d-
2
-
P
c
6
2
O
7
/
c
4
c
4
4
P
2
O
7
4
(w/w, 1/1, ball-milled), (c) (VO)
2
P
2
O
7
/
c
4
4
4
c
4
2
2
,
c
4
/d-VOPO
4
(w/w, 1/3, stirred), (e) (VO)
2 2 7
P O
4
2
2
P
2
O
7
c
-VOPO (activated in 75 vol% O
4
2
2
butane/air, 400 °C).
4
2
, 400 °C).
5
+
5
00–600 °C range (Fig. 7 g). Besides, the reduction of V in d-
VOPO and -VOPO occurs at different temperature (560 °C vs.
78 °C). Over the phase-modulated VPO catalysts, the reduction
as a result of ball-milling treatment, the conjunction at the inter-
face of two phases is enhanced, and the surface property of
phase-modulated catalyst such as the P-O bond length and surface
acidity can be modified accordingly, which in turn significantly
improves the catalytic performance. On the other hand, in terms
of the results shown in Fig. S5 and Table S11, the quantity of basic
sites is much less than that of the acidic ones on the various cata-
lysts, and there is no clear correlation observed between the cata-
lyst basicity and the catalyst performance. Therefore, it is thought
that over the current VPO type catalysts, the reaction could be
mainly catalyzed by surface acidity other than basicity.
4
c
4
5
5
+
of V in the constitutional phases is still distinguishable, but the
behavior is influenced by the degree of inter-phase contact/con-
junction. The ball milling process not only remarkably reduces
the catalyst particle size but also effectively enhances the inter-
phase contact (SEM images, Fig. 4d-f) and causes the structurally
more uniform environment for the surface V5 species to be
reduced (Fig. 7c vs. d). The mechanical ball milling is more ener-
getic than the simple stirring and can more effectively change
the particle size and morphology, resulting in alteration of facet
orientation of mixed crystallites and enhanced contact between
the facets of different phases.
+
4
. Catalytic performance
4.1. Effects of atmosphere and temperature for activation
3
.7. NH
3
2
-/CO -Tpd
The VPO samples, activated at distinct atmospheres and tem-
The NH
3
-TPD profiles were collected over the representative
peratures, were examined in aldol condensation of formaldehyde
with acetic acid to AA (MA) at atmospheric pressure. The reaction
catalysts and the results are shown in Fig. 8 and summarized in
Table 2. Within the adopted temperature range, the desorption
peaks centered around 230–240 °C are attributed to the weak acid
sites, the desorption peaks centered around 340–350 °C attributa-
ble to the medium strong acid sites, and the desorption peaks cen-
tered around 430–440 °C are ascribed to the strong acid sites
46,77]. With the calibration of the corresponding NH desorption
3
peak area, the acid site distribution and the total acid site density
can be determined (Table 2).
Notably, the total acid site densities of the phase-modulated
and ball-milled catalysts are significantly greater than that of the
phase-modulated but simply-mixed counterpart as well as the
single-phase catalysts. It is known that the density of medium
strong acid sites on the VPO type catalyst is a crucial factor for cat-
alytic activity. The density of medium strong acid sites shows the
temperature was 360 °C, GHSV (gas hourly space velocity) was
À1
1496 h , with the feed composition of HAc/HCHO/H
2
O/O
2
/
2
N = 15.3/6.1/17.3/2.2/71.5 (molar fraction) [18]. From Table S2,
clearly, the atmosphere and temperature employed for activation
is highly influential on catalyst performance. Regardless the activa-
tion temperature at 400 °C or 680 °C, the (AA + MA) yield increases
with increasing oxygen concentration to 75 vol% in the activation
atmosphere and then decreases with further increasing oxygen
concentration. Hence, the optimal oxygen concentration for activa-
tion is 75%. On the other hand, the VPO activated at 680 °C is more
active than the one activated at 400 °C. According to the character-
[
ization results, the catalyst activated at 400 °C consists of d-VOPO
together with (VO) , and the content of d-VOPO increases
with increasing oxygen concentration up to 100% at which nearly
pure d-VOPO entity is obtained. Therefore, as for the d-VOPO
phase, the catalytic activity is proportional to its content. Interest-
ingly, the catalyst activated in 75% O /N is slightly better than that
activated in pure oxygen, demonstrating that a mixture of major d-
VOPO plus (VO) functions more favorable than nearly pure
d-VOPO . In other words, the interaction between the d-VOPO
and (VO) entities enhances catalytic performance, in good
agreement with our previous observation [18]. Similarly, the
4
2 2
P O
7
4
following order:
-VOPO (w/w, 1/3, ball-milled) >
(w/w, 1/1, ball-milled) > -VOPO /d-VOPO
red) > d-VOPO > (VO) , coincident with the trend in activity
of the same catalyst serial (see below). Interestingly, the phase-
modulated but simply-mixed -VOPO /d-VOPO (w/w, 1/3) shows
c
-VOPO
4
/d-VOPO
4
(w/w, 1/3, ball-milled) > (VO)
-VOPO > (VO) /d-
(w/w, 1/3, stir-
2
-
4
4
P
2
O
7
/c
4
c
4
2 2 7
P O
VOPO
4
c
4
4
2
2
4
2 2 7
P O
4
2 2 7
P O
c
4
4
4
4
remarkably lower surface acidity (especially the density of med-
ium strong/strong as well as overall acid sites). As discussed above,
2 2 7
P O

J. Liu et al. / Journal of Catalysis 374 (2019) 171–182
179
Table 2
The surface acidity of the VPO catalysts determined by means of NH
3
-TPD.
Catalyst
Acid site distribution (lmol/gcat
)
Total acid site density (lmol/gcat)
Weak
Medium
Strong
a
b
c
d
e
f
103.5
89.7
70.6
76.6
49.3
85.3
60.3
133.8
74.6
106.2
68.5
44.4
93.0
56.1
186.2
236.4
152.6
107.9
56.1
423.5
400.6
329.3
252.9
149.8
241.6
152.2
63.2
35.7
g
(
a)
c
-VOPO
4
/d-VOPO
4
(w/w, 1/3, ball-milled), (b) (VO)
2
P
2
O
7
/d-VOPO
4
(w/w, 1/1, ball-milled), (c) (VO)
2
P
2
O
7
/
c
-VOPO
4
(w/w, 1/3, ball-milled), (d) -VOPO
c
4 4
/d-VOPO (w/w, 1/3,
stirred), (e) (VO)
2
P
2
O
7
(activated in 1.5 vol% n-butane/air, 400 °C), (f)
c
-VOPO (activated in 75 vol% O
4
/N
2 2
, 680 °C), (g) d-VOPO
4
(activated in O , 400 °C).
2
catalytic activity is closely dependent on the content of
phase. It is worth noting that the catalyst activated in pure O
c
-VOPO
4
VPO catalysts generally outperform the unitary phase catalysts;
and furthermore, the phase constitution is also critical to deter-
2
at
6
6
80 °C is less active than the one activated in 75 vol% O
80 °C, and this is mainly due to the generation of VOPO
2
/N
Á2H
2
at
2
O
mine catalytic performance. The
tant role in the reaction even if its content is in small fraction.
Over a phase-modulated VPO catalyst of (VO) /d-VOPO (w/
w, 1/1), its activity is enhanced most significantly, suggesting the
maximum inter-phase conjunction between these two phases with
similar structural feature.
In all cases, the intermediate (3-hydroxyl propanoic acid) of
condensation reaction, has not been detected over the various cat-
alysts. Moreover, we conducted the following experiments: the 3-
hydroxyl propanoic acid was directly employed as a reactant and
4
c-VOPO phase plays an impor-
4
species which is thought to be rather inactive for the reaction. Nev-
ertheless, through optimizing the activation atmosphere and tem-
perature, the phase constitution of catalyst can be tuned
P
2 2
O
7
4
À1 À1
accordingly. The highest formation rate of 1.29 mmolÁg Áh for
(
AA + MA) is achieved among these catalysts, higher than the one
previously reported by us [18].
4.2. Effect of phase constitution
performed the reaction over the single-phase d-VOPO
and the phase-modulated and ball-milled -VOPO /d-VOPO
w, 1/3) as well as the simply mixed -VOPO /d-VOPO (w/w, 1/3)
4
,
c
-VOPO
4
,
c
4
4
(w/
In Section 4.1, the catalyst phase constitution can be tuned
c
4
4
somehow by varying activation atmosphere and temperature.
However, the variation in phase constitution is not controllable,
and the quantification of different phases is essentially unknown.
In this section, the catalytic performances of the phase-
modulated VPO catalysts have been systematically investigated.
Here, the catalyst phase constitution is intentionally modulated
under the same operating conditions. It was found that nearly
identical (AA + MA) yield was obtainable over various catalysts,
suggesting that the dehydration of 3-hydroxyl propanoic acid is
not a rate-determining step regarding the catalyst phase constitu-
tion, and the condensation step is more critically influenced by the
enhanced conjunction over the dual phase catalyst (still phase
dependent).
through pre-mixing of different VPO phases [(VO)
and -VOPO ] in certain ratio (w/w = 1/3, 1/1, and 3/1) followed
2 2 7 4
P O , d-VOPO ,
c
4
by mechanical milling to enhance inter-phase coupling interaction.
As shown in Table 3, the modulated phase constitution exhibit sig-
4.3. Effect of the way of phase mixing
nificant impact on catalyst performance. When (VO)
and milled with -VOPO in different ratios, the resulting materials
are superior to the individual components in activity. Clearly, the
activity increases with increasing -VOPO content, for instance,
even the -VOPO content is only 25 wt%, the (AA + MA) yield
reaches 61.7%. When the d-VOPO is increased to 50 wt%, the
AA + MA) yield is 63.2%, significantly higher than that on the indi-
vidual (VO) (36.8%) or d-VOPO (51.5%). When the d-VOPO is
mixed and milled with -VOPO in a ratio of 3:1 (w/w), the yield of
2 2 7
P O is mixed
c
4
Catalytic performances of the dual phase-modulated VPO cata-
2
2
7
4
4 4
lysts of (VO) P O /d-VOPO (w/w, 1/1) and c-VOPO /d-VOPO (w/
c
4
w, 1/3) through ball-milling or simple mixing are presented in
Table S3. Clearly, only the ball-milled catalysts exhibit remarkably
enhanced activities. Catalytic performance also increases with
extended grinding time. To study the key factor for catalytic per-
formance in the ball-milling process, the specific surface areas of
the catalysts were obtained from the BET measurement and
showed in Table S4. There is an order in the surface area of individ-
c
4
4
(
2
P
2
O
7
4
4
c
4
(
AA + MA) reaches 67.6%, corresponding to a formation rate of
À1 À1
1
.37 mmolÁg Áh at 360 °C. Clearly, the dual phase-modulated
2 2 7 4 4
ual phases, namely, (VO) P O > d-VOPO > c-VOPO . The applied
milling process considerably increases the surface area of the indi-
vidual phases. However, this order is not coincident with the activ-
Table 3
ity order [Table S5,
4 4 2 2 7
c-VOPO > d-VOPO > (VO) P O )]. Moreover,
Catalytic activities of the phase-modulated catalysts subjected to ball-milling.
the activity of the individual phase changes insignificantly before
and after milling (Table S5), suggesting that increment in surface
area of individual phase show little impact on catalytic activity.
Moreover, it is thought that the effective inter-phase contact and
coupling interaction can only be accomplished after the ball-
milling procedure and a period of 12–24 h milling is satisfactory
to reduce the particle size and to intensify the inter-phase
contact/conjunction.
Notably, all the phase-mixed (or phase-modulated) catalysts
outperform the corresponding single-phase catalyst especially
when subjected to mechanical ball milling. The ball milling process
not only remarkably reduces the catalyst particle size but also
effectively enhances the inter-phase contact (SEM images, Fig. 4-
Catalyst
Yield of
AA + MA) (%)
(MA + AA) formation
À1 À1
(
rate (mmolÁg Áh
)
(
(
(
(
(
(
VO)
VO)
VO)
VO)
VO)
VO)
2
2
2
2
2
2
P
2
P
2
P
2
P
2
P
2
P
2
O
O
O
O
O
O
7
7
7
7
7
7
/
/
/
c
c
c
-VOPO
-VOPO
-VOPO
4
4
4
4
4
4
(w/w, 1/3)
(w/w, 1/1)
(w/w, 3/1)
(w/w, 1/3).
(w/w, 1/1).
(w/w, 3/1).
(w/w, 3/1)
(w/w, 1/1)
(w/w, 1/3)
66.0
63.4
61.7
60.1
63.2
47.0
63.7
63.1
67.6
1.34
1.29
1.25
1.22
1.28
0.95
1.30
1.28
1.37
/d-VOPO
/d-VOPO
/d-VOPO
/d-VOPO
/d-VOPO
/d-VOPO
c
c
c
-VOPO
-VOPO
-VOPO
4
4
4
4
4
4
À1
GHSV = 1496 h , T = 360 °C, HAc/HCHO/H
fraction).
2
O/O
/N
2 2
= 15.3/6.1/17.3/2.2/71.5 (molar

1
80
J. Liu et al. / Journal of Catalysis 374 (2019) 171–182
d-f) and causes the structurally more uniform environment for the
surface V species (H -TPR profiles, Fig. 7c vs. d), leading to an
2
4.5. Correlation between characterization and evaluation results
5
+
enhanced conjunction at the interface of two phases (Figs. 1e, f
and 2e, f). It does not mean; however, the target reaction needs a
prerequisite, namely, the existence of specific conjunct interface
of two phases. It is worth mentioning that, the enhanced conjunc-
tion at the interface of two phases, does have an impact on the PAO
bond strength (Fig. 2f) [78], which in turn can influence on the sur-
face acidity of catalyst and ease of the H extraction from CAH bond
The data of HCHO conversion and selectivity to (AA + MA) as
well as all the by-products measured over the seven representative
catalysts are presented in Table S12 where the total carbon bal-
ances are also included for reference.
Table S13 summarizes the (AA + MA) yield, formation rate,
4
+
5+
selectivity, together with the V /V ratio as well as the density
of medium strong acid sites. The (AA + MA) yield is closely related
to the density of medium strong acid sites of catalyst which is
dependent on the phase constitution of catalyst. Therefore, the
phase constitution of catalyst largely impacts on the density of
medium strong acid sites on the current catalysts, and the latter
is crucial to determine the catalyst performance. As the activity
is normalized to per surface area of catalyst, one can observe that
the area-normalized (AA + MA) formation rate is comparable over
catalysts b, c, e, and f except for catalysts a and d, suggesting that
the surface area is also an influential factor on catalytic activity. It
is worth noting that the selectivity to (AA + MA) is enhanced over
the phase-modulated and ball-milled catalysts especially on
[
79]. According to the NH
face acid sites is considerably increased over the phase-modulated
and ball-milled samples, especially on -VOPO /d-VOPO (w/w,
/3). Be aware of the fact that the target reaction is largely acid-
3
-TPD results (Fig. 8), the density of sur-
c
4
4
1
catalyzed in nature (this point is interpreted in more detail below).
It is true the surface region of VPO catalyst could be rather amor-
phous under n-butane oxidation, it is uncertain yet the surface
state of current catalysts under the working conditions. Neverthe-
less, the two reactions are chemically distinct (oxidation vs. con-
densation), and the reaction atmosphere (1.5% n-butane/air vs.
acetic acid/formaldehyde/water/nitrogen/oxygen) and tempera-
ture (400–420 °C vs. 360 °C) are also quite different; therefore, it
is reasonably expected that for the two reactions the surface state
of catalyst under the working conditions is rather incomparable
and fairly dependent on the phase composition of catalyst (essen-
4 4
VOPO /d-VOPO (w/w, 1/3), demonstrating that the ball-milling
process not only efficiently increases catalyst surface area but also
enhances the formation of the desired products. A comparison
between catalysts e and f with originally identical phase constitu-
tion clearly indicates that the applied ball-milling enhances the
tially the (VO)
2
2
P O
7
for n-butane oxidation vs. the most effective
c-
VOPO /d-VOPO
4
4
(w/w, 1/3) for the current condensation). Even if
2
inter-phase conjunction (also verified by the Raman, H -TPR, and
the catalyst surface is possibly somehow structural disorder for
the current system, due to the enhanced conjunction at the inter-
face of two phases, the surface property of phase-modulated cata-
lyst such as the PAO bond length and surface acidity is proved to
be largely influenced by the original phase constitution/structure.
solid-state MAS NMR investigations), which in turn results in a sig-
nificant increment in the density of medium strong acid sites and a
more selective reaction. Furthermore, there seems a direct correla-
4
+
5+
tion between the V /V ratio and the catalytic behavior. Basically,
5
+
4+
the V looks more favorable than V for the reaction. The pres-
4
+
ence of V impacts slightly on selectivity but more noticeably on
(AA + MA) formation rate. This is also confirmed by the deactiva-
tion behavior of catalyst: the yield of (AA + MA) continuously
decreases while the corresponding selectivity changes insignifi-
cantly with time on stream, accompanying by the continuous
4.4. Effects of operating parameters
The effects of reaction temperature, gas hourly space velocity,
and feed condition on (AA + MA) yield and formation rate were
investigated over d-VOPO -VOPO (w/w, 3/1) and the results
are shown in Table S6–S8. From Table S6, at 340 °C, the (AA
5
+
4+
4
/c
4
reduction of surface V to V during the reaction course.
+
1
3
MA) yield and formation rate declines to 49.2% and
4
.6. The reaction scheme
À1 À1
.00 mmolÁg Áh , respectively. As the temperature is raised to
80 °C, the catalyst performance changes insignificantly compared
The involved reactions are demonstrated in Scheme 1 regarding
to the values obtained at 360 °C. Thus 360 °C is a suitable temper-
ature for the reaction.
The data presented in Table S7 indicate that the carrier flow rate
shows dramatic impact on activity. The carrier flow rate was con-
the fed substances in the reaction and the measured products dis-
tribution. The major reaction route is the condensation between
acetic acid and formaldehyde to AA which can further react with
methanol to MA. There are minor reaction routes from acetic acid
plus methanol to methyl acetate which can also react with
formaldehyde to MA. Acetic acid itself can subject to a side reaction
to produce acetone/CO
on our catalysts. The CO
reactants/products.
trolled by the N
O, and O ) were remained, as the carrier flow rate is increased
from 30 mL/min to 40 mL/min, the (AA + MA) yield increases from
2
flow rate and the other flow rates (HAc, HCHO,
H
2
2
2
, but this route is found to be very limited
is resulted from the deep oxidation of
6
1
5
4
7.6% to 84.2%, corresponding to a formation rate as high as
À1 À1
x
.71 mmolÁg Áh . If the carrier flow rate is further increased to
0 mL/min, the (AA + MA) yield changes little. So a flow rate of
0 mL/min for the carrier is appropriate for the reaction. It appears
that the contact time is critical to determine the selective forma-
tion of AA and MA. On one hand, prompt desorption of AA/MA from
catalyst surface avoids further disintegration of the desired prod-
ucts; and on the other hand, quick recover of surface sites speeds
up the reaction cycles [70].
5. Catalyst durability
This reaction through a continuous gas phase heterogeneous
process has been studied so far by a few research groups
[14,18,80,81], and in most cases, only the initial activities (2.5 h
reaction period or even shorter) were reported and compared.
According to the limited examples of durability tests reported by
others and us [18,44], there is continuous decline in activity with
time on stream regardless the deactivation is relatively slow or
fast. The phase composition of VPO catalyst could be variable upon
operating conditions such as reaction atmosphere and tempera-
The influence of feed rate on catalytic performance was also
studied (Table S8). As the feed rate (based on HCHO) varies from
À1
6
.1 to 30.5 mmolÁh , the formation rate of (AA + MA) increases
steadily, meanwhile the yield of (AA + MA) declines simultane-
À1
À1 À1
ously. At a feed rate of 18.3 mmolÁh , the (MA + AA) formation
rate reaches a level of 3.29 mmolÁg Áh , corresponding to a yield
À1
of (AA + MA) being 54%. At the highest feed rate (30.5 mmolÁh
)
4 4
ture. We selected the most efficient catalyst [d-VOPO /c-VOPO
(w/w, 3/1)] as the representative system to conduct durability test
for approximately 110 h, with the GHSV of 1823 h and feed com-
employed, the (MA + AA) formation rate and yield is
.25 mmolÁg ÁhÀ1 and 41.8%, respectively.
À1
À1
4

J. Liu et al. / Journal of Catalysis 374 (2019) 171–182
181
with time on stream, and such evolution of V oxidation state is
reversible upon air-treatment. Interestingly, accompanying with
the variation in V oxidation state, the density of weak/medium
strong acid sites decreases accordingly and is significantly recov-
ered upon regeneration (Fig. S4). These characterization findings
over the used catalysts of different period reasonably account for
the variation in catalytic performance with time on stream up to
1
20 h. It is worth noting that the catalytic performance is not sim-
ply determined by the reducibility of d-VOPO or -VOPO phase.
According to Fig. 7, although -VOPO is more difficult to reduce
than d-VOPO , the single
phase-modulated d-VOPO
4
c
4
c
4
4
c
c
-VOPO
-VOPO
4
is obviously inferior to the
(w/w, 3/1) in catalytic perfor-
4
/
4
mance, suggesting the phase-coupling is critical in determining
catalytic activity.
3
+
To verify whether there is V species existing in the fresh and
used catalysts, the representative catalyst of
4 4
c-VOPO /d-VOPO (w/
w, 1/3, ball-milled) was first subjected to a H
2
-TPR procedure (from
RT to 850 °C) and then analyzed by XPS. As shown in Fig. S6a, there
is one peak centered at 515.8 eV attributable to V 2p3/2 [82], con-
Scheme 1. A reaction scheme in terms of the fed substances in the reaction and the
measured products distribution. The thick arrow line corresponds to the major
reaction route whereas the thin arrow lines refer to the minor reaction routes.
3
+
3
+
firming that the V is dominated after the H
2
-TPR procedure. Over
4
+
the same catalyst used for 72 h (Fig. S6b), though V is dominant
on the catalyst surface, a small quantity of V can also be identi-
fied. In contrast, the fresh catalyst (Fig. S6c) shows the two de-
convoluted peaks at 516.6 and 517.6 eV, corresponding to V
2p3/2 and V 2p3/2, respectively [63,64]. Therefore, it is clarified
that the V species are essentially undetectable on the fresh cata-
lyst surface, but could be detected in a low content on the used cat-
alyst surface with a TOS of 72-h.
3
+
2 2 2
position being HAc/HCHO/H O/O /N = 15.3/6.1/17.3/2.2/96.0 (in
molar fraction, Fig. 9). The initial activity is the highest known to
date and the deactivation is comparatively slower during 110 h
and can be fully recovered by simple air-treatment at the reaction
temperature.
To verify the possible phase variation of catalyst during the
reaction period, the used catalyst subjected to different reaction
period (2.5 h, 27 h, and 72 h, respectively) and regeneration was
4+
5+
3+
6
. Concluding remarks
3
collected and analyzed by means of Raman, XPS, and NH -TPD.
The reaction performances are listed in Table S9 and the character-
ization results are presented in Figs. S2–S4 and Table S10. Notably,
the HCHO conversion gradually decreases with time on stream.
However, the selectivity for target products essentially retains
In this study, the precisely phase-modulated VPO catalysts were
fabricated and the inter-phase conjunction can be effectively
enhanced by the approach of mechanical grinding and controlled
activation. The resulting materials were applied for the gas phase
condensation of acetic acid and formaldehyde to acrylic acid
(
the inset of Fig. 9), suggesting that the deactivation is associated
to the decrease in conversion rather than selectivity. As demon-
strated in Figs. S2 and S3, over the used sample (72 h), the Raman
(
methyl acrylate). The present work is the first example to demon-
À1
5+
strate the significance of quantification of VPO phase constitution
in determining catalyst performance for the target reaction. Sys-
tematic characterizations reveal the important information on
the phase type of VPO component, the surface concentration and
oxidation state of V element, the surface V/P ratio, the redox
behavior, and the surface acidity of unitary phase and dual
phase-modulated catalysts. Through systematic tuning the phase
type and constitution as well as the activation conditions, we
obtained the most efficient VPO catalyst system so far for the tar-
band of
s 2 2 7
t (P-O) shifts to 932 cm [(VO) P O ] and the surface V
4
+
is mostly reduced to V . Over the regenerated catalyst, the Raman
À1
4+ 5+
band of
s
t (P-O) shifts back to 941 cm and the surface V /V
ratio is recovered to 0.60 (Table S10), indicating that in the surface
5+
region of catalyst, the V species in the original d-VOPO
4
/
c
-VOPO
phases is gradually reduced to V4 species [(VO)
2 2 7
P O -like entity]
4
+
4 4
get reaction. Over an optimized catalyst of d-VOPO /c-VOPO (w/
w, 3/1), the (AA + MA) yield of 84.2% with the equivalent (AA
À1 À1
+
MA) formation rate of 1.71 mmolÁg Áh or the (AA + MA) yield
of 41.8% with the equivalent (AA + MA) formation rate of
À1 À1
4
.25 mmolÁg Áh are achievable on the formaldehyde input basis
under the controlled conditions. This sort of catalyst can be repro-
ducibly fabricated, rather durable within a period of 110 h, and
fully recovered by simple air-treatment at the reaction
temperature.
Acknowledgement
The financial support of National Nature Science Foundation of
China (21673112) is greatly appreciated.
Appendix A. Supplementary material
Supplementary data to this article can be found online at
https://doi.org/10.1016/j.jcat.2019.04.032.
4 4
Fig. 9. Durability test over c-VOPO /d-VOPO (w/w, 1/3) at 360 °C for a period of
approximately 120 h.

1
82
J. Liu et al. / Journal of Catalysis 374 (2019) 171–182
[
[
[
44] D. Yang, C. Sararuk, K. Suzuki, Z.X. Li, C.S. Li, Chem. Eng. J. 300 (2016) 160–168.
45] M. Ai, Stud. Surf. Sci. Catal. 162 (2006) 457–464.
46] D. Yang, C. Sararuk, H. Wang, S.J. Zhang, Z.X. Li, C.S. Li, Ind. Eng. Chem. Res. 57
References
[
1] H. Danner, M. Neureiter, L. Madzingaidzo, M. Gartner, R. Braun, Appl. Biochem.
Biotechnol. 70 (1998) 895–903.
(
2018) 93–100.
47] H. Zhao, C.C. Zuo, D. Yang, C.S. Li, S.J. Zhang, Ind. Eng. Chem. Res. 55 (2016)
2693–12702.
48] D. Yang, D. Li, H.Y. Yao, G.L. Zhang, T.T. Jiao, Z.X. Li, C.S. Li, S.J. Zhang, Ind. Eng.
Chem. Res. 54 (2015) 6865–6873.
[
[
[
[
[
[
[
[
[
[
[
2] X.B. Xu, J.P. Lin, P.L. Cen, Chin. J. Chem. Eng. 14 (2006) 419–427.
3] P.T. Anastas, J.C. Warner, Green Chemistry: Theory and Practice, Oxford
University Press, New York, 1998, pp. 11–56.
1
[
[
[
[
[
[
4] C.C. Zuo, Y.P. Li, C.S. Li, S.S. Cao, H.Y. Yao, S.J. Zhang, AlChE J. 62 (2016) 228–240.
5] M.M. Lin, Appl. Catal., A 207 (2001) 1–16.
6] X.J. Yang, R.M. Feng, W.J. Ji, C.-T. Au, J. Catal. 253 (2008) 57–65.
7] N. Mizuno, M. Tateishi, M. Iwamoto, Appl. Catal., A 128 (1995) L165–L170.
8] U. Takashi, N. Hiroya, K. Yukio, W. Shin, US Patent 5380933, 1995.
9] M.M. Lin, Appl. Catal., A 250 (2003) 305–318.
49] D. Yang, G. Wang, H. Wu, X.P. Guo, S.J. Zhang, Z.X. Li, C.S. Li, Catal. Today 316
(
2018) 122–128.
50] X. Feng, Y. Yao, Q. Su, L. Zhao, W. Jiang, W.J. Ji, C.-T. Au, Appl. Catal., B 164
2015) 31–39.
51] X.S. Wang, W.Y. Nie, W.J. Ji, X.F. Guo, Q.J. Yan, Y. Chen, Chem. Lett. 30 (2001)
96–697.
52] W.J. Ji, L.J. Xu, X.S. Wang, Z. Hu, Q.J. Yan, Y. Chen, Catal. Today 74 (2002) 101–
10.
53] X.S. Wang, L.J. Xu, X. Chen, W.J. Ji, Q.J. Yan, Y. Chen, J. Mol. Catal. A: Chem. 206
2003) 261–268.
54] W.Y. Nie, Z.Y. Wang, W.J. Ji, Y. Chen, C.-T. Au, Appl. Catal., A 244 (2003) 265–
72.
55] C.Y. Xiao, X. Chen, Z.Y. Wang, W.J. Ji, Y. Chen, C.-T. Au, Catal. Today 93–95
2004) 223–228.
(
6
[
10] P. Botella, J.M. López Nieto, B. Solsona, A. Mifsud, F. Márquez, J. Catal. 209
2002) 445–455.
(
1
[
[
11] W. Fang, Q.J. Ge, J.F. Yu, H.Y. Xu, Ind. Eng. Chem. Res. 50 (2011) 1962–1967.
12] R.N. d’Alnoncourt, L.-I. Csepei, M. Hävecker, F. Girgsdies, M.E. Schuster, R.
Schlögl, A. Trunschke, J. Catal. 311 (2014) 369–385.
(
[
[
[
[
[
[
[
[
13] B.Y. Jo, S.S. Kum, S.H. Moon, Appl. Catal., A 378 (2010) 76–82.
14] M. Ai, J. Catal. 107 (1987) 201–208.
2
15] O.H. Bailey, R.A. Montag, J.S. Yoo, Appl. Catal., A 88 (1992) 163–177.
16] M. Ai, H. Fujihashi, S. Hosoi, A. Yoshida, Appl. Catal., A 252 (2003) 185–191.
17] J.R. Tai, R.J. Davis, Catal. Today 123 (2007) 42–49.
18] X.Z. Feng, B. Sun, Y. Yao, Q. Su, W.J. Ji, C.-T. Au, J. Catal. 314 (2014) 132–141.
19] M. Ai, Appl. Catal., A 288 (2005) 211–215.
(
[56] Z.Q. Zhou, H.Y. Xu, W.J. Ji, Y. Chen, Catal. Lett. 96 (2004) 221–226.
[57] X.K. Li, W.J. Ji, J. Zhao, Z.B. Zhang, C.-T. Au, J. Catal. 238 (2006) 232–241.
[58] R.M. Feng, X.J. Yang, W.J. Ji, Y. Chen, C.-T. Au, J. Catal. 246 (2007) 166–176.
[59] Q. Jiang, J. Zhao, X.K. Li, W.J. Ji, Z.B. Zhang, C.-T. Au, Appl. Catal., A 341 (2008)
20] J.B. Yan, C.L. Zhang, C.L. Ning, Y. Tang, Y. Zhang, L.L. Chen, S. Gao, Z.L. Wang, W.
X. Zhang, J. Ind. and Eng. Chem. 25 (2015) 344–351.
7
0–76.
60] F. Ben Abdelouahab, R. Olier, N. Guilhaume, F. Lefebvre, J.C. Volta, J. Catal. 134
1992) 151–167.
[
[
[
21] R.K. Zeidan, M.E. Davis, J. Catal. 247 (2007) 379–382.
22] B. Liu, S.J. Wu, X.F. Yu, J.Q. Guan, Q.B. Kan, J. Colloid Interface Sci. 362 (2011)
(
[
[
[
61] T.P. Moser, G.L. Schrader, J. Catal. 92 (1985) 216–231.
6
25–628.
23] B. Li, R.Y. Yan, L. Wang, Y.Y. Diao, Z.X. Li, S.J. Zhang, Catal. Lett. 143 (2013) 829–
38.
62] E. Bordes, Catal. Today 1 (1987) 499–526.
[
63] P.T. Nguyen, R.D. Hoffman, A.W. Sleight, Mater. Res. Bull. 30 (1995) 1055–
8
1063.
[
[
[
[
[
[
[
[
24] M. Ai, J. Catal. 112 (1988) 194–200.
25] M. Ai, Appl. Catal. 59 (1990) 227–235.
26] M. Ai, Stud. Surf. Sci. Catal. 72 (1992) 101–108.
27] M. Ai, J. Catal. 106 (1987) 273–279.
28] M.R. Gogate, J.J. Spivey, J.R. Zoeller, Catal. Today 36 (1997) 243–254.
29] B.K. Hodnett, Cat. Rev. Sci. Eng. 27 (1985) 373–424.
30] G. Centi, F. Trifiro, J.R. Ebner, V.M. Franchetti, Chem. Rev. 88 (1988) 55–80.
31] G.J. Hutchings, A. Desmartin-Chomel, R. Olier, J.C. Volta, Nature 368 (1994) 41–
[
64] F.D. Hardcastle, I.E. Wachs, J. Phys. Chem. 95 (1991) 5031–5041.
[
65] Y. Jia, D.Q. Du, J.C. Bai, J. Ding, Q. Zhong, X.L. Ding, Atmos. Pollut. Res. 6 (2015)
184–190.
[
[
[
[
66] M.J. Cheng, W.A. Goddard, J. Am. Chem. Soc. 135 (2013) 4600–4603.
67] B.M. Reddy, I. Ganesh, E.P. Reddy, J. Phys. Chem. B 101 (1997) 1769–1774.
68] V.I. Bukhtiyarov, Catal. Today 56 (2000) 403–413.
69] J. Kasperkiewicz, J.A. Kovacich, D. Lichtman, J. Electron. Spectrosc. Relat.
Phenom. 32 (1983) 123–132.
4
5.
[
70] R.A. Overbeek, A.R.C.J. Pekelharing, A.J. van Dillen, J.W. Geus, Appl. Catal., A 135
[32] R. Nilsson, T. Lindblad, A. Andersson, J. Catal. 148 (1994) 501–513.
[33] M.T. Sananes, G.J. Hutchings, J.C. Volta, J. Catal. 154 (1995) 253–260.
[34] C.J. Kiely, A. Burrows, G.J. Hutchings, K.E. Bere, J.C. Volta, A. Tuel, M. Abon,
Faraday Discuss. 105 (1996) 103–118.
(
1996) 231–248.
[
71] R.A. Overbeek, P.A. Warringa, M.J.D. Crombag, L.M. Visser, A.J. van Dillen, J.W.
Geus, Appl. Catal., A 135 (1996) 209–230.
[72] J. Li, M.E. Lashier, G.L. Schrader, B.C. Gerstein, Appl. Catal. 73 (1991) 83–95.
[73] M.T. Sananes, A. Tuel, J.C. Volta, J. Catal. 145 (1994) 251–255.
[74] M.T. Sananes, A. Tuel, Solid State Nucl. Magn. Reson. 6 (1996) 157–166.
[75] K.E. Birkeland, S.M. Babitz, G.K. Bethke, H.H. Kung, G.W. Coulston, S.R. Bare, J.
Phys. Chem. B 101 (1997) 6895–6902.
[
[
35] S. Albonetti, F. Cavani, F. TrifirÒ, Cat. Rev. Sci. Eng. 38 (1996) 413–438.
36] G.W. Coulston, S.R. Bare, H. Kung, K. Birkeland, G.K. Bethke, R. Harlow, N.
Herron, P.L. Lee, Science 275 (1997) 191–193.
[
[
37] J.M. Herrmann, P. Vernoux, K.E. Béré, M. Abon, J. Catal. 167 (1997) 106–117.
38] V.V. Guliants, S.A. Holmes, J.B. Benziger, P. Heaney, D. Yates, I.E. Wachs, J. Mol.
Catal. A: Chem. 172 (2001) 265–276.
[
76] M.T. Sananes-Schulz, A. Tuel, G.J. Hutchings, J.C. Volta, J. Catal. 166 (1997) 388–
392.
[
39] G.J. Hutchings, J.A. Lopez-Sanchez, J.K. Bartley, J.M. Webster, A. Burrows, C.J.
Kiely, A.F. Carley, C. Rhodes, M. Hävecker, A. Knop-Gericke, R.W. Mayer, R.
Schlögl, J.C. Volta, M. Poliakoff, J. Catal. 208 (2002) 197–210.
[
[
77] A. Tuel, M.T. Sananes-Schulz, J.C. Volta, Catal. Today 37 (1997) 59–68.
78] O.B. Lapina, D.F. Khabibulin, A.A. Shubin, V.M. Bondareva, J. Mol. Catal. A:
Chem. 162 (2000) 381–390.
[
40] N. Hiyoshi, N. Yamamoto, N. Ryumon, Y. Kamiya, T. Okuhara, J. Catal. 221
[
[
[
[
[
79] V. Robert, S. Petit, S.A. Borshch, B. Bigot, J. Phys. Chem. A 104 (2000) 4586–
(
2004) 225–233.
4591.
[
41] M. Conte, G. Budroni, J.K. Bartley, S.H. Taylor, A.F. Carley, A. Schmidt, D.M.
Murphy, F. Girgsdies, T. Ressler, R. Schlögl, G.J. Hutchings, Science 313 (2006)
80] J. Frey, C. Lieder, T. Schölkopf, T. Schleid, U. Nieken, E. Klemm, M. Hunger, J.
Catal. 272 (2010) 131–139.
81] N.H. Batis, H. Batis, A. Ghorbel, J.C. Vedrine, J.C. Volta, J. Catal. 128 (1991) 248–
63.
82] B. Horvath, J. Strutz, J. Geyer-Lippmann, E.G. Horvath, Z. Anorg, Allg. Chem. 483
1981) 181–192.
83] T.P. Moser, G.L. Schrader, J. Catal. 104 (1987) 99–108.
1
270–1273.
[
[
42] J.K. Bartley, N.F. Dummer, G.J. Hutchings, Vanadium phosphate catalysts metal
oxide catalysis, in: S.D. Jackson, J.S.J. Hargreaves (Eds.), WILEY-VCH Verlag,
Weinheim, 2009 (Chapter 12).
43] F. Ivars-Barceló, G.J. Hutchings, J.K. Bartley, S.H. Taylor, P. Sutter, P. Amorós, R.
Sanchis, B. Solsona, J. Catal. 354 (2017) 236–249.
2
(