ACS Catalysis
Letter
results mainly from its internal pore structure (internal surface
area), whereas the surface area of the nonporous SA particles
originates exclusively from the outer accessible surface.
Quantitative 1H MAS NMR investigations with the
adsorption of ammonia was used to determine the surface
Brønsted acid sites on dehydrated SAs. The adsorption of
acetone-13C has been used to distinguish Brønsted and Lewis
acid sites by 13C MAS NMR, and the relative intensity of Lewis
sites to Brønsted sites was applied to determine the number of
Lewis sites. Comparison of the amount of Brønsted and Lewis
acid sites of the SAs (Table 1, column 5 and 6) indicates that all
SAs are mainly populated with Brønsted acid sites. Hardly any
Lewis acid sites could be detected by 13C MAS NMR
spectroscopy21 at an Al content lower than 30%. In contrast,
De-Al-HY contains a 2-times-higher amount of Lewis acid sites
compared with Brønsted acid sites and exposes a higher
number of both Brønsted and Lewis acid sites than the SA
catalysts.
higher than SA/70). PG conversion with this reference catalyst
was 90% after 6 h of reaction, compared with 100% on SA/70.
A comparison of the average turnover frequencies (TOFs)
confirms the high activity of the SAs for PG conversion (Table
1). The average TOFs of the flame-derived SAs after 6 h were
between 10.2 and 10.5 h−1, which is ∼20 times higher than that
determined for De-Al-HY (0.57 h−1). Note that De-Al-HY has
much stronger acid sites (both strength and population
density) and a 2−3 times higher surface area (671 m2/g,
Table. 1, column 2) compared with SA/10 and SA/30 (377 and
248 m2/g, respectively).21,22 Striking differences between the
SAs and De-Al-HY are the strength of the acid sites and their
location (accessibility). The acid sites of De-Al-HY are located
inside the nanopores, whereas the SAs are virtually nonporous,
and constraints imposed by intraparticle diffusion phenomena
can be ruled out. On the large pore (∼1.1 nm) zeolite De-Al-
HY, PG (∼0.67 nm) can enter into the cages; however, the rate
of this size-confined diffusion of PG inside micropores is
expected to be considerably lower than that of the interparticle
free diffusion of PG to the surface acid sites on SAs. The slow
reactant and product diffusion inside the pores of De-Al-HY
obviously slows down the overall reaction rate, and
consequently, the higher population density of acid sites on
this catalyst does not lead to enhanced PG conversion. This
scenario is further supported when comparing the reaction
kinetics observed with De-Al-HY and SA/70 shown in Figure 2.
At the beginning of the reaction, De-Al-HY with 16.5 times
more acid sites and 3.5 times higher surface area showed only
slightly higher conversion than SA/70. With the progress of the
reaction, the conversion on SA/70 became higher compared
with that on De-Al-HY. It appears that the large-size products
formed inside the zeolite pores and the decrease in the reactant
concentration slowed the global reaction rate on De-Al-HY as a
result of intraparticle diffusional limitations.
The catalytic reactions were carried out using a PG (0.4 M)
ethanol solution at 363 K over 0.05 g of catalyst in a stirred
batch reactor. The results obtained with SAs possessing
different surface area and population density of Brønsted and
Lewis acid sites are summarized in Table 1. Conversions of PG
and selectivities to ethyl mandelate (EM) as a function of
reaction time with SA/10, 30, 50, and 70 are shown in Figure 2.
As reported earlier,11 dealuminated zeolite USY in the Lewis
acidity domain (73% extra-framework Al as Lewis acid sites,
27% framework Al contribute Brønsted acid sites) showed high
selectivity (97%) and high yield (95%) to ethyl mandelate.
Therefore, Lewis acid sites were proposed to contribute mainly
to the formation of ethyl mandelate, whereas Brønsted acid
sites preferentially generated the corresponding acetals. Lewis
acid sites are also dominant in De-Al-HY, which afforded ethyl
mandelate with 90% selectivity and 81% yield in this work.
Interestingly, SA/70 with mainly Brønsted acid sites showed
higher selectivity and yield in ethyl mandelate production
(97%). Even with SA/10 exposing only Brønsted acidity, PG
was converted with a selectivity of 93% to ethyl mandelate.
Interestingly, flame-derived pure alumina with dominant Lewis
acid sites also afforded 94% selectivity and 87% yield to ethyl
mandelate. Thus, ethyl mandelate is the preferred product on
both Brønsted acidic and Lewis acidic flame-derived SAs, and
increasing the strength of acid sites21 has only little influence on
the product selectivity.
Results of the catalytic test carried out with different alcohols
as reactants over the best-performing catalyst SA/70 are
summarized in Figure 3 and Table 2. Figure 3 shows how the
conversion and the selectivity to the corresponding alkyl
mandelates (AM) developed with reaction time. Conversion of
PG varied significantly among the alkyl alcohols used, being
lowest for MeOH and higher with longer alkyl chains. The
selectivity to alkyl mandelates was high for all alcohols, except
for n-butyl alcohol (87%), and much less dependent on the
alkyl chain length of the alcohol. The reference catalyst De-Al-
Figure 2. Catalytic conversion of PG in ethanol (―) and selectivity
□
○
△
to ethyl mandelate (---) over SA/70 ( ), SA/50 ( ), SA/30 ( ), SA/
▽
10 ( ), and De-Al-HY (☆) as a function of 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 nonacidic SA/0 was inactive for the target reaction. After
introduction of aluminum into the silica matrix, by adding Al to
the precursor feed solution of the flame reactor, acidic sites
were formed on the surface of the SAs21 (Table 1). With
increasing aluminum content, the population density of acidic
sites increased, resulting in strikingly higher conversion and
slightly higher selectivity to the target compound ethyl
mandelate (Figure 2). The conversion of PG on different SAs
after 6 h was 60% for SA/10, 71% for SA/30, 85% for SA/50,
and 100% for SA/70. The increase in PG conversion correlates
well with the increasing population density of acid sites on the
surface of the SAs.
For comparison, the PG reaction was also performed with
dealuminated zeolite H-Y (De-Al-HY) containing a large
number of Lewis acid sites (1.75 mmol/g, 219 times higher
than SA/70) and Brønsted acid sites (0.865 mmol/g, 6 times
1575
dx.doi.org/10.1021/cs400271e | ACS Catal. 2013, 3, 1573−1577