Strongly enhanced acidity and activity of amorphous silica–alumina by formation of pentacoordinated AlV species

Article abstract of DOI:10.1016/j.jcat.2019.02.007

Tailoring high-performance aluminosilicates plays a key role in the efficient and clean production of high-value chemicals. Recent work reveals that pentacoordinated Al (Al^V) species can significantly enhance the Br?nsted acidity of amorphous silica–alumina (ASA), compared with that typically dominated by tetracoordinated Al species. However, the controlled synthesis of Al^V-rich ASAs is challenging. Employing xylene as the solvent in a flame-spray pyrolysis process, we synthesized Al^V-rich ASAs successfully. The high combustion enthalpy of xylene (36.9 kJ/ml) results in a high flame temperature, promoting the formation and distribution of metastable Al^V species in the silica network forming Br?nsted acid sites. This provides a promising route for the controlled synthesis of Al^V-rich ASAs with higher Br?nsted acidity. As an example, Al^V-rich ASAs are shown to exhibit superior catalytic performance in phenylglyoxal conversion to ethyl mandelate in ethanol compared with that achieved with other acid catalysts, attaining an ethyl mandelate yield of 99.8%.

Full text of DOI:10.1016/j.jcat.2019.02.007

Journal of Catalysis 372 (2019) 1–7
Contents lists available at ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Strongly enhanced acidity and activity of amorphous silica–alumina
by formation of pentacoordinated AlV species
e,
Zichun Wang a,b, Yijiao Jiang b, Fangzhu Jin a, Catherine Stampfl c, Michael Hunger d, , Alfons Baiker
⇑
⇑
,
Jun Huang a,
⇑
a Laboratory for Catalysis Engineering, School of Chemical and Biomolecular Engineering, Sydney Nano Institute, the University of Sydney, New South Wales 2006, Australia
b Department of Engineering, Macquarie University, Sydney, New South Wales 2109, Australia
c School of Physics, Sydney Nano Institute, the University of Sydney, Sydney, New South Wales 2006, Australia
d Institute of Chemical Technology, University of Stuttgart, D-70550 Stuttgart, Germany
e Institute for Chemical and Bioengineering, Department of Chemistry and Applied Bioscience, ETH Zürich, Hönggerberg, HCI, CH-8093 Zürich, Switzerland
a r t i c l e i n f o
a b s t r a c t
Article history:
Tailoring high-performance aluminosilicates plays a key role in the efficient and clean production of
high-value chemicals. Recent work reveals that pentacoordinated Al (AlV) species can significantly
enhance the Brønsted acidity of amorphous silica–alumina (ASA), compared with that typically domi-
nated by tetracoordinated Al species. However, the controlled synthesis of AlV-rich ASAs is challenging.
Employing xylene as the solvent in a flame-spray pyrolysis process, we synthesized AlV-rich ASAs suc-
cessfully. The high combustion enthalpy of xylene (36.9 kJ/ml) results in a high flame temperature, pro-
moting the formation and distribution of metastable AlV species in the silica network forming Brønsted
acid sites. This provides a promising route for the controlled synthesis of AlV-rich ASAs with higher
Brønsted acidity. As an example, AlV-rich ASAs are shown to exhibit superior catalytic performance in
phenylglyoxal conversion to ethyl mandelate in ethanol compared with that achieved with other acid
catalysts, attaining an ethyl mandelate yield of 99.8%.
Received 23 August 2018
Revised 12 December 2018
Accepted 3 February 2019
Keywords:
Amorphous silica–alumina
Flame-spray pyrolysis
Pentacoordinated aluminum
Solid-state NMR
Acidity
Phenylglyoxal
Ethyl mandelate
Ó 2019 Elsevier Inc. All rights reserved.
1. Introduction
proposed to further enhance the acid strength of neighboring mod-
erate BAS (e.g., BAS-AlIV) to zeolitic strength or even higher [13].
Amorphous silica–aluminas (ASAs) are among the most widely
used solid acid catalysts and supports for the production of petro-
chemicals, fine chemicals, and renewable energy [1–3]. The cat-
alytically active Brønsted acid sites (BAS) originate from
aluminum centers, commonly accepted as tetracoordinated alu-
minum (AlIV) species, flexibly coordinated to neighboring SiOH
groups in ASAs [4–6]. However, a very low Si/Al ratio (<7/3) leads
to the formation of alumina phases [7,8] and to a decrease of the
Brønsted acidity in ASAs [9,10].
This enables AlV-rich ASA to exhibit excellent catalytic perfor-
mance; as an example, a turnover frequency more than an order
of magnitude higher than that of dealuminated zeolite H-Y was
achieved in the catalytic conversion of phenylglyoxal (PG) [14].
Moreover, AlV species can act as Lewis acid sites (LAS) and anchor
active metal species on the ASA surface [15,16]. Therefore, enrich-
ment of ASA with AlV species not only can enhance their Brønsted
acidity, but also can improve their catalytic performance as multi-
functional catalysts or excellent support materials for reactions.
AlV species are intermediates or metastable species between
AlIV and AlVI species [15,17–19]. Various techniques, including
classical wet-chemical synthesis [20,21], chemical liquid deposi-
tion [6], and atomic layer deposition [22], generally produce
AlIV-rich ASAs, rather than AlV–rich ASAs. Flame spray pyrolysis
(FSP), which we employed in this work, is a single-step method
that can generate multicomposition nanoparticles in milliseconds,
combining synthesis and calcination on a large scale [23–25].
Taking advantage of super-high temperatures (up to 2000 K) and
fast cooling rates, FSP is able to produce materials dominated by
metastable phases and polymorphs [24], and thus, FSP techniques
Recently, 27Al-{1H} dipolar-mediated heteronuclear multiple
quantum correlation (D-HMQC) two-dimensional (2D) NMR stud-
ies have evidenced that pentacoordinated aluminum (AlV) species
in spatial proximity to SiOH groups contribute to a new type of
BAS (BAS-AlV) [11]. Coexisting AlV and AlIV species can elevate
the BAS population density in ASA, reaching seven times higher
densities than in conventional AlIV-rich ASAs [12]. AlV species are
⇑
Corresponding authors.
E-mail addresses: michael.hunger@itc.uni-stuttgart.de (M. Hunger), baiker@
chem.ethz.ch (A. Baiker), jun.huang@sydney.edu.au (J. Huang).
https://doi.org/10.1016/j.jcat.2019.02.007
0021-9517/Ó 2019 Elsevier Inc. All rights reserved.

2
Z. Wang et al. / Journal of Catalysis 372 (2019) 1–7
are suitable for preparing AlV-enriched ASAs [13,26]. The AlV con-
tent in ASAs can be tuned from 0 to 41.6% to promote BAS forma-
tion with increasing Al loading up to 70% [13]. At the same Al
loading, a higher precursor flow rate results in a higher AlV content,
but decreases the number of BAS, which is attributed to the larger
particle size of the ASAs [26].
materials were virtually nonporous. Prior to isotherm measure-
ment, each sample was degassed at 423 K for 12 h under vacuum.
The BET surface areas of the ASAs were determined from the N2
adsorption isotherms measured at 77 K using the Brunauer–Emm
ett–Teller (BET) method.
Powder X-ray diffraction (XRD) studies were performed on a
Catalytic conversion of PG in water and alcohols over solid acids
enables single-step production of mandelic acid and mandelates
[14,27–29], which are valuable intermediates for producing phar-
maceuticals and fine chemicals [30,31]. Nonporous ASA catalysts
facilitate a turnover frequency an order of magnitude higher than
that of microporous dealuminated HY zeolites [14]. The catalytic
performance of ASAs can be improved significantly by increasing
their Brønsted acidity [14], which provides a much higher ethyl
mandelate (EM) yield (98.6%) than that achieved with other solid
acids reported earlier [27–29].
Siemens D5000 instrument with CuKa radiation in the range
10°–70° with scanning steps of 0.02°.
2.3. Solid-state MAS NMR spectroscopy
For the 27Al and 29Si MAS NMR investigations, all samples were
fully hydrated by overnight exposure to the saturated vapor of a Ca
(NO3)2 solution at ambient temperature in a desiccator. 27Al and
29Si MAS NMR investigations were carried out on a Bruker Avance
III 400 WB spectrometer at resonance frequencies of 104.3 and
79.5 MHz, respectively. 27Al MAS NMR spectra were recorded at
a sample spinning rate of 8 kHz using 4-mm MAS rotors after
To our knowledge, the deliberate generation of AlV species to
promote Brønsted acidity in ASAs has so far not been reported,
but would be promising for improving the catalytic performance
of ASAs for various reactions. With this in mind, we used FSP to
prepare ASAs with significantly improved AlV density, resulting
in higher Brønsted acidity of the surface sites and superior catalytic
performance. A series of AlV-rich ASAs with different Si/Al ratios
were synthesized using xylene as the solvent, which has a higher
combustion enthalpy (36.9 kJ/ml), resulting in a higher flame tem-
perature than that of the methanol/acetic acid (Me/AA) mixture
single-pulse
MAS NMR measurements were performed with a sample spinning
rate of 4 kHz using a 7-mm MAS rotor after single-pulse /2 exci-
p
/6 excitation with a repetition time of 0.5 s. 29Si
p
tation, high-power proton decoupling, and a recycle delay of 20 s.
Before the 1H and 13C MAS NMR experiments, the samples filled
into glass tubes were dehydrated for 12 h at 723 K and at a pres-
sure of less than 10ꢁ2 bar. These dehydrated samples were sealed
in glass tubes or directly loaded with ammonia or acetone-2-13
C
(volume ratio 1/1) with
a
combustion heat of ’25.4 kJ/ml
(99.5% 13C-enriched, Sigma-Aldrich) on a vacuum line. Subse-
quently, the loaded samples were evacuated at 373 K for 1 h (for
ammonia) or at room temperature for 2 h (for acetone) to remove
weakly physisorbed molecules. Then the samples were transferred
into the MAS NMR rotors under dry nitrogen gas inside a glove box.
1H and 13C MAS NMR investigations were performed on the same
spectrometer at resonance frequencies of 400.1 and 100.6 MHz,
respectively, with a sample spinning rate of 8 kHz using 4-mm
MAS rotors. 1H MAS NMR spectra were recorded after single-
employed in previously reported ASA syntheses [13]. It will be
shown that the use of xylene as a solvent in the liquid precursor
flame feed results in significantly enhanced AlV concentration at
the same Al loading and thus in a larger number of BAS, as revealed
by quantitative 1H magic-angle spinning (MAS) NMR spectroscopy.
Finally, the catalytic performance of AlV-rich ASAs prepared via the
above-mentioned route was evaluated for the reaction of PG in
ethanol to demonstrate their remarkable catalytic activity com-
pared to other solid acids.
pulse
p/2 excitation with a repetition time of 20 s. Quantitative
1H MAS NMR measurements were carried out using a zeolite H,
Na-Y (35% ion-exchanged) as an external intensity standard. 13C
cross-polarization (CP) MAS NMR spectra were recorded with a
contact time of 4 ms and a repetition time of 4 s.
2. Experimental methods
2.1. Preparation of ASAs by FSP
2.4. Catalytic conversion of phenylglyoxal in ethanol
Aluminum acetylacetonate (Al(acac)3, purity ꢀ 99.9%), tetra-
ethyl orthosilicate (TEOS) (purity ꢀ 99.9%), and xylene
(purity ꢀ 98.5%) were all purchased from Sigma-Aldrich and used
for producing ASA catalysts. The precursor solutions used for the
preparation of ASAs by FSP were obtained by dissolving appropri-
ate amounts of the precursor materials in xylene (0.5 M concentra-
tion by metal), followed by filtration using a glass filter. Then the
FSP precursor solutions were used in the FSP process as described
previously [13]. In brief, the precursor solutions were pumped
through a capillary at a rate of 5 ml/min and nebulized at 5 L/
min O2. The resulting spray was ignited by an annular supporting
methane/oxygen flame (1.5/0.9 L/min). Particles were collected
on a cooled Whatman GF6 filter (diameter 257 mm). A Busch SV
1040C vacuum pump aided in particle recovery. The synthesized
silica–alumina powders are designated as SA/Xx, where X is 5,
10, 30, or 70, representing the atom% of Al in the precursor, and
x indicates the use of xylene as a solvent to distinguish it from
the methanol/acetic acid (denoted as SA/X) used in earlier work
[13].
The conversion of PG (Sigma-Aldrich, >97%) to EM was utilized
to study the catalytic performance of the SA/Xx materials. An
amount of 0.05 g of the catalyst was employed and activated over-
night under an N2 flow of 50 ml/min at 723 K. After cooling down
in flowing N2 gas, the catalyst was transferred into a glass reactor
with a volume of 15 ml. Subsequently, 1.25 ml of the ethanol solu-
tion containing 0.4 M PG and 0.05 M octane (as GC internal stan-
dard) was added and mixed with the activated catalyst under
magnetic stirring. The reaction was carried out in a tightly closed
glass reactor immersed in an oil bath at 363 K for 6 h. The reaction
products were determined using a Shimadzu QP2010 Ultra GC–MS
equipped with a Rtx-5MS column (30 m ꢂ 0.25 mm ꢂ 0.25
and quantified by a Shimadzu GC2010 equipped with an flame ion-
ization detector and a Rtx-5 column (30 m ꢂ 0.32 mm ꢂ 0.25 m).
lm)
l
The selectivity for the specific products i (Si) was calculated as Si
(%) = 100 ꢂ (i)/[(PG)0 ꢁ (PG)], where (i) is the molar concentration
of the product i and (PG)0 and (PG) correspond to the molar con-
centrations of PG before and after the reaction, respectively. The
used catalyst was washed with ethanol three times and dried over-
night at 373 K before being calcined at 773 K for 3 h for complete
removal of organic residues. The reaction mixture after 6 h reaction
was analyzed by atomic emission spectroscopy with an inductively
coupled plasma (ICP-AES, Perkin-Elmer, Plasma 400).
2.2. Structural characterization
N2 adsorption–desorption isotherms measured at 77 K on an
Autosorb IQ-C system indicated that the synthesized ASA powder

Z. Wang et al. / Journal of Catalysis 372 (2019) 1–7
3
metastable species (e.g., AlV) through combustion at high temper-
ature, and the resulting well-mixed product molecules and clusters
underwent nucleation and grew to nanoparticles. The high cooling
rate facilitated the freezing of metastable species in the final prod-
uct. Compared with Me/AA, xylene has a higher combustion
enthalpy, providing a higher flame temperature, and thus a more
severe temperature drop, which enhanced the freezing of meta-
stable AlV species.
3. Results and discussion
3.1. Catalyst morphology
The amorphous nature of the SA/Xx catalysts was confirmed
by XRD patterns (Fig. 1). All samples showed a broad peak at
2h = 20°–30° attributed to amorphous silica [32], and no crystalline
silica or alumina phases could be detected. This finding indicates
that aluminum species are well dispersed in the amorphous silicon
oxide network.
The surface areas and primary particle sizes of the SA/Xx cata-
lysts are summarized in Table 1. The specific surface areas of these
materials were in the range of 241–391 m2/g, significantly higher
than those of ASAs previously prepared using Me/AA mixtures as
solvent (SA/X materials, 200–377 m2/g) [13]. As observed for the
SA/X materials [13], the surface areas of SA/Xx catalysts gradually
decreased with increasing Al content. The average particle sizes of
SA materials were estimated from BET measurement (Table 1) and
confirmed by SEM images (Fig. S1 in the Supporting Information).
Interestingly, using xylene instead of Me/AA in the FSP precursor
solution did not strongly affect the particle size of SA materials
(7–17 vs. 7–20 nm).
Recently, the formation of BAS based on AlV species has been
evidenced by 27Al-{1H} D-HMQC 2D NMR experiments [11]. There-
fore, the effect of increasing the AlV content on the surface
Brønsted acidity of the ASAs was studied. 1H MAS NMR spec-
troscopy is a powerful tool for the characterization of surface
hydroxyl protons [5]. As shown in Fig. 3a–e, top, the signals of non-
acidic and acidic surface OH groups are strongly overlapping, and
only a strong peak at d1H = 1.8 ppm assigned to SiOH groups can
be observed in the 1H MAS NMR spectra of dehydrated ASA cata-
lysts [5,13]. Adsorption of a strong base, such as ammonia, is a use-
ful technique for identifying BAS on ASA [5,13]. Upon ammonia
loading, a signal arises at d1H = ca. 6.7 ppm (Fig. 3a–e, bottom),
which indicates the formation of ammonium ions via ammonia
protonation at BAS.
The number of BAS can be estimated by integration of the
ammonium signal (d1H = ca. 6.7 ppm) in Fig. 3 [5], leading to the
results summarized in Table 1. Since AlVI species mainly afford sur-
face Lewis acidity [35–38], the enhancement of BAS (both density
and fraction) with increasing Al content (from 5 to 70%) obviously
depends on the significant increase in AlV concentration (from 0 to
55.5%) rather than on the strong decrease in AlIV concentration
(from 97 to 16.3%). Moreover, the larger amount of AlV species con-
tributes to the higher BAS density (6.8–19.8 ꢂ 10ꢁ2 mmol/g)
obtained for SA/Xx using xylene as solvent compared with ASAs
prepared by other methods [12,39], and with FSP-derived SAs pre-
pared with Me/AA as solvent (9.8–15.1 ꢂ 10ꢁ2 mmol/g) [13].
In ASAs, the formation of BAS is accepted as being caused by Al
atoms in the vicinity of SiOH groups. The correlation between Al
atoms and neighboring SiOH groups was probed by a 1H/27Al TRAP-
DOR (transfer of population in double resonance) MAS NMR exper-
iment (Fig. 4) [4,40,41]. In a TRAPDOR experiment, the dipolar
dephasing of 1H–27Al pairs can be suppressed by continuous 27Al
irradiation, which influences the intensity of 1H spin–echo MAS
NMR signals caused by OH groups in the vicinity of Al species.
Thus, the difference spectrum of SA/70x recorded without and with
27Al irradiation (Fig. 4 bottom) evidenced the interaction between
SiOH groups (d1H = 1.8 ppm) and neighboring Al atoms [4,5]. More-
over, a strong low-field shift of the main peak in the 29Si MAS NMR
spectra of SA/Xx catalysts with increasing Al content (Fig. 5)
demonstrates a continuous incorporation of Al into the silica net-
work [13,42]. It correlates well with the increasing number of
BAS, which indicates that increasing aluminum incorporation con-
tributes to the formation of these surface sites on ASA.
3.2. Surface acidity and local structure of SA/Xx materials
The effect of using xylene as a solvent in the FSP synthesis of
AlV-rich ASA was investigated by 27Al MAS NMR spectroscopy
(Fig. 2). In SA/5x, the strong signal at d27Al = 51 ppm and the weak
signal at d27Al = 0 ppm represent the commonly observed AlIV and
AlVI species in ASAs. With the Al content increased to 10 mol.%
(SA/10x), a signal at d27Al = ca. 28 ppm assigned to AlV species
was observed [11,13,33], which was strongly enhanced when the
Al content was increased to 70 mol% (SA/70x).
The molar ratios of Al species are summarized in Table 1. Obvi-
ously, using a solvent with higher combustion enthalpy, such as
xylene, can enhance the formation of AlV species in ASAs. The AlV
species was observed in SA/10x (17.6%), in this research but not
in SA/10 in our previous study [13]. At 30 to 70 mol.% Al loading,
the fraction of AlV species synthesized by FSP using xylene as sol-
vent was 1.3–1.6 times higher than that of corresponding catalysts
produced with Me/AA as solvent.
During the FSP synthesis [23,24,34], the metal precursors and
the solvent were atomized and sprayed to form the precursor
vapor, which in turn was converted to product vapor containing
Since ASA catalysts are often utilized under anhydrous condi-
tions, the study of the correlation between the number of BAS on
the dehydrated materials and the AlV species as a function of Al
loading is of great importance (Fig. 6). The concentrations of AlV
species were determined by decomposing the 27Al MAS NMR spec-
tra of corresponding dehydrated SA/Xx catalysts as shown in the
Supporting Information (Fig. S1 and Table S2). Compared to SAs
prepared with Me/AA as solvent, the increase of the BAS density
from using xylene as solvent correlates well with the increase of
AlV species (44% vs. 39%, 52% vs. 48%, 57% vs. 51%, and 61% vs.
55% for Al loadings of 10%, 30%, 50%, and 70%, respectively) while
the AlIV fraction decreased (54% vs. 59%, 43% vs. 51%, 38% vs.
42%, and 32% vs. 38% for Al loadings of 10%, 30%, 50%, and 70%,
respectively). Clearly, using xylene as a solvent yields a larger pop-
ulation of AlV species, and thus promotes the connection between
Fig. 1. XRD patterns of FSP-derived SA/Xx catalysts with different Al contents.

4
Z. Wang et al. / Journal of Catalysis 372 (2019) 1–7
Table 1
Properties of FSP SA/Xx catalysts with Al content X of 5–70 atom%: BET surface areas (ABET), particle sizes (DBET), total densities of surface OH groups (TOH), densities of Brønsted
acid sites (NBAS), and fractions of Al species with different oxygen coordination.
Sample
ABET (m2/g)
DBET (nm)a
TOH (mmol/g)b
NBAS (mmol/g)b
AlIV (atom%)c
AlV (atom%)c
AlVI (atom%)c
AlV (atom%)d
SA/5x
391
369
289
267
241
7
8
9
12
17
1.4
1
1.3
1.2
1.6
6.8 ꢂ 10ꢁ2
7.4 ꢂ 10ꢁ2
14.8 ꢂ 10ꢁ2
15.8 ꢂ 10ꢁ2
19.8 ꢂ 10ꢁ2
97
56.7
23
20
16.3
—
17.6
43
55
55.5
3
25.7
34
25
28.2
—
0
20.3
40.5
41.6
SA/10x
SA/30x
SA/50x
SA/70x
a
The particle diameter (DBET) of SA/Xx materials derived from BET surfaces according to DBET (nm) = 6000/(BET surface area in m2/g) ꢂ (density in g/cm3).
b
Total number of OH groups (TOH) and population of Brønsted acidic OH groups (NBAS). Determined from quantitative 1H MAS NMR using NH3 as a probe molecule as
shown in Section 3.2.
c
The atomic concentrations of Al species were determined from Fig. S2.
d
AlV concentration obtained from Ref. [13] using Me/AA as a solvent.
Fig. 2. 27Al MAS NMR spectra of FSP-derived SA/5x (a), SA/10x (b), SA/30x (c), SA/
50x (d), and SA/70x (e).
AlV species and SiOH groups in their vicinity [11] to generate more
BAS. Notably, both SA/30x and SA/50, as well as SA/50x and SA/70,
which contain similar AlV and AlIV fractions but higher BAS densi-
ties, are obtained for SA prepared using xylene at a much lower Al
content. More significantly, SA/30x has a BAS density similar to
that of SA/70 having a factor of 2.33 higher Al content at lower
AlV concentration. This finding indicates that using a solvent with
higher combustion enthalpy (e.g., xylene) not only increases the
AlV concentration, but can also promote the distribution of Al spe-
cies over the silica network and thus the formation of BAS. There-
fore, this approach provides a potential route for the synthesis of
ASA materials with higher acidity at lower Al content.
The effect of AlV species on the acid strength of ASA was scaled
by the 13C NMR chemical shift of the carbonyl groups of acetone-
2-13C (CH313COCH3) as a probe molecule [5]. A greater adsorbate-
induced low-field shift of the 13C CP/MAS NMR signals indicates
a higher Brønsted acid strength of surface OH groups and vice
versa. As shown in Fig. 7, the 13C CP/MAS NMR spectra of SA/Xx
samples are all dominated by a signal at 214–216 ppm, assigned
to acetone-2-13C adsorbed onto BAS with moderate strength. The
broad hump at ca. 225 ppm in Fig. 7b and 7c indicates a wide dis-
tribution of acidic OH groups with higher strength on SA/Xx, while
the weak signal at ca. 240 ppm in SA/70x (Fig. 7c) was caused by
acetone adsorbed onto Lewis acidic sites [13]. Other d13C signals
are due to aldol reaction of the acetone-2-13C on SA/x catalysts
and nonenriched carbon atoms in methyl groups, respectively
[43–45].
Fig. 3. 1H MAS NMR spectra of dehydrated FSP-derived SA/5x (a), SA/10x (b), SA/
30x (c), SA/50x (d), and SA/70x (e), recorded before (top) and after (bottom) loading
with NH3 and subsequent evacuation at 373 K for 1 h.
Obviously, the surface OH groups of the SA/Xx catalysts exhibit
a Brønsted acid strength from moderate to comparable to or even
stronger than that of zeolites H-Y (d13C = 220 ppm) [45] and HZSM-
5 (d13C = 223 ppm) [44,45], which is similar to that of FSP-derived
ASA synthesized using Me/AA as a solvent [13]. Particularly, the
spectrum of SA/10x shows a hump at d13C ca. 225 ppm, which
was not observed in the spectrum of SA/10 using Me/AA as solvent
and which was dominated by a signal at d13C = 213 ppm only [13].
In terms of the slightly lower BAS density, this indicates that the

Z. Wang et al. / Journal of Catalysis 372 (2019) 1–7
5
Fig. 4. 1H/27Al TRAPDOR spectra of dehydrated SA/70x.
Fig. 7. 13C CP/MAS NMR spectra of SA/5x (a), SA/10x (b), and SA/70x (c) recorded
after loading with acetone-2-13C and evacuation at room temperature for 1 h.
3.3. Catalytic conversion of phenylglyoxal to ethyl mandelate
The catalytic performance of the SA/Xx materials was evaluated
using the conversion of PG in ethanol to EM at 363 K. The catalytic
results are summarized in Table 2. The conversion of PG and the
selectivity to EM as a function of time are shown in Fig. 8 (also
see Fig. S3 with error bars). Pure silica is inactive for this reaction
[14]. Already, SA/5x, having mainly Brønsted acidity, showed a
PG conversion of 61.6% and a selectivity to EM of 98.7% after a reac-
tion time of 6 h. The enhancement of the Brønsted acidity of SA/Xx
materials with increasing Al content up to 70% significantly
improved the PG conversion from 61.6% to 100% under the same
conditions, and the selectivity of EM remained higher than 98.6%
for all samples. With SA/70x, an EM yield of 99.8% was achieved,
which to our knowledge is the highest reported so far with solid
acid catalysts [29,46–50].
Fig. 5. 29Si MAS NMR spectra of FSP-derived SA/5x (a), SA/10x (b), SA/30x (c), and
SA/70x (d).
Compared with AlV–rich FSP-derived ASA materials synthesized
using Me/AA as the solvent [13], a significant increase of the EM
yield (89.8% vs. 56.5%, 95.2% vs. 67%, and 99.8% vs. 97% for
SA/10x, SA/30x, and SA/70x, respectively) was acquired on SA/Xx
using xylene as the solvent. The major difference between these
two batches is the remarkable increase in the AlV concentration,
which improves the Brønsted acidity. The effect of AlV species on
the catalytic activity of the BAS was evaluated by comparing the
rate constants k of PG conversion (Table 2, columns 5 and 6).
SA/10x showed
a
2.1 times higher
slightly smaller number of BAS
k
value (0.0054 sꢁ1 vs.
0.0026 sꢁ1
)
with
a
(7.4 ꢂ 10ꢁ2 mmol/g vs. 9.8 ꢂ 10ꢁ2 mmol/g) than SA/10. This cat-
alytic property is attributed to the enhanced BAS strength
(d13C = 214–225 ppm for SA/10x vs. 213 ppm for SA/10), which
facilitates the activation of the C@O bond in PG as the initiating step
of the reaction. The high BAS densities of the AlV–rich SA/30x and
SA/70x catalysts resulted in 2.8 and 1.3 times higher k values than
for the FSP-derived ASA using Me/AA as solvent (0.0097 sꢁ1 vs.
0.0035 sꢁ1 and 0.0141 sꢁ1 vs. 0.0106 sꢁ1). Moreover, SA/30x, which
has a number of BAS similar to that for SA/70, afforded a similar
PG conversion and k value, but at half the Al content, demonstrating
the preparation of high-performance ASA materials with much lower
Al content. Therefore, the efficiency of the catalytic conversion of PG
to EM can be significantly improved by enhancing the Brønsted
acidity of ASA via increasing the AlV concentration. The novel route
for preparing FSP-derived ASA using xylene as a solvent represents
a promising way to enhance the Brønsted acidity of ASAs and to
improve their catalytic performance.
Fig. 6. Correlation between the concentration of Brønsted acidic OH groups (BAS,
solid line) and AlV species (dashed line) as a function of the Al content in dehydrated
FSP-derived SA/Xx prepared with xylene ( ) and MeAA (j) as solvents. The density
of AlV species was obtained by simulation of the corresponding 27Al NMR spectra of
dehydrated SA/Xx samples using the parameters obtained in MQMAS NMR
investigations (Fig. S1). The simulation parameters are summarized in Table S1.
higher AlV concentration of SA/10x causes a higher Brønsted acid
strength. With increasing Al content up to 70%, both the density
and strength of BAS are promoted by the increase of AlV species
in SA/Xx compared with other ASA catalysts. Therefore, the
FSP-derived ASA synthesized using xylene as solvent is expected
to exhibit higher catalytic performance for the target reaction than
ASA catalysts prepared via other routes.

6
Z. Wang et al. / Journal of Catalysis 372 (2019) 1–7
Table 2
Catalytic properties of FSP-derived SA/x catalysts with different Al content.a
b
c
d
f
Sample
CPG (%)
YEM (%)
SEM (%)
ke (sꢁ1
)
k0 (sꢁ1
)
TOFg (hꢁ1
)
TOF0g, (hꢁ1
)
SA/5x
61.6
90.1
95.4
98.9
100
61
98.7
98.8
98.6
99.8
99.8
0.0023
0.0054
0.0097
0.0126
0.0141
—
24.4
35.2
20.5
35.3
28.4
—
SA/10x
SA/30x
SA/50x
SA/70x
89.8
95.2
98.7
99.8
0.0026
0.0035
0.0051
0.0106
27.5
19.8
31.6
25.3
a
b
c
d
e
f
Conditions: 1.25 ml of ethanol solution containing 0.4 M PG and 0.05 g catalyst at 363 K for 6 h with stirring.
CPG = conversion of PG.
YEM = yield of ethyl mandelate in mol.%.
SEM = selectivity to ethyl mandelate at 50% conversion.
k = reaction rate from fitting kinetic data.
k0 obtained from the kinetic data of FSP SA catalysts using the mixture of methanol and acetic acid (1:1 by volume) as solvent.
TOF and TOF0 were calculated on the basis of 8–10% PG conversion and the number of BAS on SA/Xx and SA/X, respectively.
g
significant activity loss (Fig. S4). Furthermore, no Al leaching could
be detected in the reaction mixture, and the 27Al MAS NMR spectra
of SA/70x indicated no significant change before and after five
recycling runs (Fig. S5), confirming the high stability of SA/70x in
the liquid-phase reaction under the current conditions.
4. Conclusions
AlV-rich ASA catalysts with different Al content have been suc-
cessfully synthesized by FSP using a solvent with high combustion
enthalpy, xylene, for the precursor solution fed to the flame. The
local structure and acidity of as-prepared ASAs have been charac-
terized using suitable probe molecules and various solid-state
NMR methods. The studies show that the concentration of AlV spe-
cies on the ASA surface can be tuned by applying solvents with dif-
ferent combustion enthalpies, resulting in different flame
temperatures. The higher combustion enthalpy of xylene used as
solvent causes a higher flame temperature and a sharper tempera-
ture drop during cooling than with previously used solvents, which
enhances the freezing of metastable AlV species in the final prod-
uct. High AlV concentration promotes the formation of BAS in the
ASA. The AlV concentration in as-prepared ASAs is much higher
than those in ASAs prepared by other methods, where AlIV species
is predominant [12,39]. Moreover, as-prepared ASAs exhibit com-
parable BAS density at significantly lower Al content than previ-
ously prepared ASAs.
The FSP-derived ASAs using xylene as solvent for the precursor
solution showed outstanding catalytic performance in the conver-
sion of PG in ethanol to EM, affording 99.8% yield, which to our
knowledge is the highest yield of EM reported so far with a solid
acid catalyst. The present work shows that the concentration of
AlV species and the Brønsted acidity of FSP-synthesized ASAs can
be controlled by the choice of the solvent used in the FSP process,
providing together with the optimization of the Al content a
promising strategy for tuning the acidity of ASAs.
Fig. 8. Catalytic conversion of PG in ethanol (closed symbols) and selectivity to
ethyl mandelate (open symbols) over SA/5x (j), SA/10x (d), SA/30x (▲), SA/50x
(r), and SA/70x ( ) as a function of reaction time. Conditions: 1.25 ml of ethanol
solution containing 0.4 M PG, 0.05 g catalyst, at 363 K for 6 h with stirring.
The catalytic performance of individual active sites on the sur-
face was evaluated by turnover frequency (TOF) measurements.
The TOFs of SA/Xx at 8–10% PG conversion, summarized in Table 2,
are in the range of 20.5–38.3 hꢁ1, slightly higher than those
obtained with SA/X (19.8–31.6 hꢁ1, TOF0 in Table 2) at the same
Al content. Thus, increasing the AlV concentration is able to boost
the catalytic performance of individual sites (TOF) mainly at low
Al content, similarly to the enhancement of the strength of BAS
observed with SA/10x compared with SA/10 due to greater acid
strength, as detected by 13C NMR experiments (Fig. 7).
Flame-derived SA catalysts exhibit much higher catalytic per-
formance than microporous and mesoporous silica–alumina cata-
lysts reported in the literature [14,28,29]. For example, SA/70x
affords a PG conversion of 95% and an EM yield of 94% after 4 h
reaction, while with dealuminated HY (de-Al-HY) zeolite and [Al]
MCM-41 (Si/Al = 15) PG conversions of only 82% and 84% and EM
yields of 74% and 80% were achieved. Remarkably, the TOFs
obtained in this work are 3–6 times higher than those achieved
with de-Al-HY zeolite (6.2 hꢁ1 at 9% PG conversion), the best cata-
lyst for this reaction reported so far in the literature [14,29]. Since
de-Al-HY zeolite has much higher density and strength of acid
sites, the enhanced catalytic performance of flame-made non-
porous catalysts may be ascribed partly to better diffusional mass
transfer than with the diffusional constraints inherent in the
microporous zeolite.
Acknowledgments
This work was supported by Australian Research Council
Discovery Projects (DP150103842 and DP180104010), the SOAR
Fellowship, and the Sydney Nano Grand Challenge from the
University of Sydney from the University of Sydney.
Appendix A. Supplementary material
Finally, the recycling behavior of the best-performing catalyst
(SA/70x) was tested in five recycling runs, which showed no
Supplementary data to this article can be found online at
https://doi.org/10.1016/j.jcat.2019.02.007.

Z. Wang et al. / Journal of Catalysis 372 (2019) 1–7
7
[26] K.D. Kim, S. Pokhrel, Z. Wang, H.J. Ling, C.F. Zhou, Z.W. Liu, M. Hunger, L.
Madler, J. Huang, ACS Catal. 6 (2016) 2372–2381.
References
[27] Z. Wang, Y. Jiang, M. Hunger, A. Baiker, J. Huang, ChemCatChem 6 (2014)
2970–2975.
[28] Z. Wang, Y. Jiang, R. Rachwalik, Z. Liu, J. Shi, M. Hunger, J. Huang,
ChemCatChem 5 (2013) 3889–3896.
[29] P.P. Pescarmona, K.P.F. Janssen, C. Delaet, C. Stroobants, K. Houthoofd, A.
Philippaerts, C. De Jonghe, J.S. Paul, P.A. Jacobs, B.F. Sels, Green Chem. 12
(2010) 1083–1089.
[30] S.H. Li, A.M. Zheng, Y.C. Su, H.J. Fang, W.L. Shen, Z.W. Yu, L. Chen, F. Deng, Phys.
Chem. Chem. Phys. 12 (2010) 3895–3903.
[31] A. Corma, S. Iborra, A. Velty, Chem. Rev. 107 (2007) 2411–2502.
[32] C. Gerardin, S. Sundaresan, J. Benziger, A. Navrotsky, Chem. Mater. 6 (1994)
160–170.
[33] Z. Wang, S. Pokhrel, M. Chen, M. Hunger, L. Mädler, J. Huang, J. Catal. 302
(2013) 10–19.
[34] W.Y. Teoh, R. Amal, L. Mädler, Nanoscale 2 (2010) 1324–1347.
[35] K.J.D. MacKenzie, J. Temuujin, M.E. Smith, P. Angerer, Y. Kameshima,
Thermochim. Acta 359 (2000) 87–94.
[36] G.H. Kuehl, H.K.C. Timken, Micropor. Mesopor. Mater. 35–6 (2000) 521–532.
[37] D.C. Koningsberger, J.T. Miller, The development of strong acidity by non-
framework aluminum in H-USY determined by Al XAFS spectroscopy, Elsevier
Science, Amsterdam, 1996, pp. 841–850.
[38] W.E.E. Stone, G.M.S. Elshafei, J. Sanz, S.A. Selim, J. Phys. Chem. 97 (1993)
10127–10132.
[39] F. Leydier, C. Chizallet, A. Chaumonnot, M. Digne, E. Soyer, A.A. Quoineaud, D.
Costa, P. Raybaud, J. Catal. 284 (2011) 215–229.
[40] M. Xu, W. Wang, M. Seiler, A. Buchholz, M. Hunger, J. Phys. Chem. B 106 (2002)
3202–3208.
[41] C.R. Wilke, P. Chang, AICHE J. 1 (1955) 264–270.
[42] J. Huang, Y. Jiang, N. van Vegten, M. Hunger, A. Baiker, J. Catal. 281 (2011) 352–
360.
[43] A.I. Biaglow, R.J. Gorte, D. White, J. Catal. 150 (1994) 221–224.
[44] S. Li, A. Zheng, Y. Su, H. Zhang, L. Chen, J. Yang, C. Ye, F. Deng, J. Am. Chem. Soc.
129 (2007) 11161–11171.
[1] A. Corma, Chem. Rev. 95 (1995) 559–614.
[2] G. Busca, Chem. Rev. 107 (2007) 5366–5410.
[3] G.W. Huber, S. Iborra, A. Corma, Chem. Rev. 106 (2006) 4044–4098.
[4] Q. Luo, F. Deng, Z.Y. Yuan, J. Yang, M.J. Zhang, Y. Yue, C.H. Ye, J. Phys. Chem. B
107 (2003) 2435–2442.
[5] Y. Jiang, J. Huang, W. Dai, M. Hunger, Solid State Nucl. Magn. Reson. 39 (2011)
116–141.
[6] M. Valla, A.J. Rossini, M. Caillot, C. Chizallet, P. Raybaud, M. Digne, A.
Chaumonnot, A. Lesage, L. Emsley, J.A. van Bokhoven, C. Coperet, J. Am.
Chem. Soc. 137 (2015) 10710–10719.
[7] A. Matsumoto, H. Chen, K. Tsutsumi, M. Grun, K. Unger, Micropor. Mesopor.
Mater. 32 (1999) 55–62.
[8] G.A. Eimer, L.B. Pierella, G.A. Monti, O.A. Anunziata, Catal. Lett. 78 (2002) 65–
75.
[9] K. Tanabe, Solid acids and bases and their catalytic properties, Academic Press,
New York, 1970.
[10] M. Hunger, D. Freude, H. Pfeifer, H. Bremer, M. Jank, K.P. Wendlandt, Chem.
Phys. Lett. 100 (1983) 29–33.
[11] Z. Wang, Y. Jiang, O. Lafon, J. Trebosc, K.D. Kim, C. Stampfl, A. Baiker, J.P.
Amoureux, J. Huang, Nat. Commun. 7 (2016) 13820.
[12] B. Xu, C. Sievers, J.A. Lercher, J.A.R. van Veen, P. Giltay, R. Prins, J.A. van
Bokhoven, J. Phys. Chem. C 111 (2007) 12075–12079.
[13] J. Huang, N. van Vegten, Y. Jiang, M. Hunger, A. Baiker, Angew. Chem. Int. Ed. 49
(2010) 7776–7781.
[14] Z. Wang, Y. Jiang, A. Baiker, J. Huang, ACS Catal. 3 (2013) 1573–1577.
[15] R. Wischert, P. Laurent, C. Coperet, F. Delbecq, P. Sautet, J. Am. Chem. Soc. 134
(2012) 14430–14449.
[16] H. Deng, Y.B. Yu, F.D. Liu, J.Z. Ma, Y. Zhang, H. He, ACS Catal. 4 (2014) 2776–
2784.
[17] M.F. Williams, B. Fonfe, C. Sievers, A. Abraham, J.A. van Bokhoven, A. Jentys, J.A.
R. van Veen, J.A. Lercher, J. Catal. 251 (2007) 485–496.
[18] B.M. Dewitte, P.J. Grobet, J.B. Uytterhoeven, J. Phys. Chem. 99 (1995) 6961–
6965.
[45] J.F. Haw, J.B. Nicholas, T. Xu, L.W. Beck, D.B. Ferguson, Acc. Chem. Res. 29
(1996) 259–267.
[19] P.J. Chupas, K.W. Chapman, G.J. Halder, J. Am. Chem. Soc. 133 (2011) 8522–
8524.
[46] S.S. Hall, A.M. Doweyko, F. Jordan, J. Am. Chem. Soc. 98 (1976) 7460–7461.
[47] S.S. Hall, A.M. Doweyko, F. Jordan, J. Am. Chem. Soc. 100 (1978) 5934–5939.
[48] A.E. Russell, S.P. Miller, J.P. Morken, J. Org. Chem. 65 (2000) 8381–8383.
[49] K. Maruyama, Y. Murakami, K. Yoda, T. Mashino, A. Nishinaga, J. Chem. Soc.-
Chem. Commun. (1992) 1617–1618.
´
[20] S. Nishikawa, E. Matijevic, J. Colloid Interface Sci. 165 (1994) 141–147.
[21] J.H. de Boer, Discuss. Faraday Soc. 52 (1971) 109–112.
[22] A.R. Mouat, C. George, T. Kobayashi, M. Pruski, R.P. van Duyne, T.J. Marks, P.C.
Stair, Angew. Chem. Int. Ed. 54 (2015) 13346–13351.
[23] R. Strobel, A. Baiker, S.E. Pratsinis, Adv. Powder Technol. 17 (2006) 457–480.
[24] R. Koirala, S.E. Pratsinis, A. Baiker, Chem. Soc. Rev. 45 (2016) 3053–3068.
[25] R. Strobel, S.E. Pratsinis, J. Mater. Chem. 17 (2007) 4743–4756.
[50] K. Ishihara, T. Yano, M. Fushimi, J. Fluorine Chem. 129 (2008) 994–997.