Active sites in Sn-beta for glucose isomerization to fructose and epimerization to mannose

Article abstract of DOI:10.1021/cs500466j

Framework Lewis acidic tin sites in hydrophobic, pure-silica molecular sieves with the zeolite beta topology (Sn-Beta) have been reported previously to predominantly catalyze glucose-fructose isomerization via 1,2 intramolecular hydride shift in water and glucose-mannose epimerization via 1,2 intramolecular carbon shift in methanol. Here, we show that alkali-free Sn-Beta predominantly isomerizes glucose to fructose via 1,2 intramolecular hydride shift in both water and methanol. Increasing extents of postsynthetic Na⁺ exchange onto Sn-Beta, however, progressively shifts the reaction pathway toward glucose-mannose epimerization via 1,2 intramolecular carbon shift. Na ⁺ remains exchanged onto silanol groups proximal to Sn centers during reaction in methanol solvent, leading to nearly exclusive selectivity toward epimerization. In contrast, decationation occurs with increasing reaction time in aqueous solvent and gradually shifts the reaction selectivity to isomerization at the expense of epimerization. Decationation and the concomitant selectivity changes are mitigated by the addition of NaCl to the aqueous reaction solution. Preadsorption of ammonia onto Sn-Beta leads to near complete suppression of infrared and ¹¹⁹Sn nuclear magnetic resonance spectroscopic signatures attributed to open Sn sites and of glucose-fructose isomerization pathways in water and methanol. These data provide evidence that Lewis acidic open Sn sites with either proximal silanol groups or Na-exchanged silanol groups are respectively the active sites for glucose-fructose isomerization and glucose-mannose epimerization.

Full text of DOI:10.1021/cs500466j

Research Article
pubs.acs.org/acscatalysis
Active Sites in Sn-Beta for Glucose Isomerization to Fructose and
Epimerization to Mannose
‡
‡
†
Ricardo Bermejo-Deval, Marat Orazov, Rajamani Gounder, Son-Jong Hwang, and Mark E. Davis*
Chemical Engineering, California Institute of Technology, Pasadena, California 91125, United States
*
S Supporting Information
ABSTRACT: Framework Lewis acidic tin sites in hydrophobic, pure-silica molecular
sieves with the zeolite beta topology (Sn-Beta) have been reported previously to
predominantly catalyze glucose−fructose isomerization via 1,2 intramolecular hydride
shift in water and glucose−mannose epimerization via 1,2 intramolecular carbon shift in
methanol. Here, we show that alkali-free Sn-Beta predominantly isomerizes glucose to
fructose via 1,2 intramolecular hydride shift in both water and methanol. Increasing
+
extents of postsynthetic Na exchange onto Sn-Beta, however, progressively shifts the
reaction pathway toward glucose−mannose epimerization via 1,2 intramolecular carbon
+
shift. Na remains exchanged onto silanol groups proximal to Sn centers during reaction
in methanol solvent, leading to nearly exclusive selectivity toward epimerization. In
contrast, decationation occurs with increasing reaction time in aqueous solvent and
gradually shifts the reaction selectivity to isomerization at the expense of epimerization.
Decationation and the concomitant selectivity changes are mitigated by the addition of
NaCl to the aqueous reaction solution. Preadsorption of ammonia onto Sn-Beta leads to
1
19
near complete suppression of infrared and
Sn nuclear magnetic resonance
spectroscopic signatures attributed to open Sn sites and of glucose−fructose isomerization
pathways in water and methanol. These data provide evidence that Lewis acidic open Sn
sites with either proximal silanol groups or Na-exchanged silanol groups are respectively the active sites for glucose−fructose
isomerization and glucose−mannose epimerization.
KEYWORDS: tin-beta, open site, closed site, silanol, Bilik reaction, epimerization, isomerization, fructose, glucose, mannose,
sodium exchange
1
. INTRODUCTION
inhibition or deactivation that otherwise occurs in the presence
of liquid water.
We have previously shown that tetravalent Lewis acidic metal
centers (Sn4 and Ti ) isolated within hydrophobic, pure-silica
+
4+
When Sn-Beta was used to react glucose in methanol solvent,
we obtained the surprising result that mannose was produced
selectively via a Lewis acid-mediated 1,2 intramolecular carbon
molecular sieves with the zeolite beta framework topology (Sn-
Beta and Ti-Beta, respectively) catalyze the isomerization of
glucose to fructose in aqueous media.
mediate the ring-opening of glucose and coordinate with
glucose O1 and O2 atoms prior to isomerization via an
2
1
−4
shift mechanism known as the Bilik reaction. Although
Framework Sn sites
9
−11
homogeneous molybdate anions
complexes
and nickel(II) diamine
1
2−14
have been reported to catalyze the epimeriza-
tion of glucose to mannose by the Bilik reaction, Sn-Beta is the
intramolecular hydride shift from the C2 to C1 position (1,21
first example of a heterogeneous catalyst that can mediate this
intramolecular hydride shift) in the ring-opened glucose chain.
2
reaction. Only framework Sn sites in Sn-Beta were able to
This glucose isomerization reaction pathway is analogous to
that observed in metalloenzymes such as D-xylose isomerase XI
that contain two divalent Lewis acid metal centers (e.g., Mg
or Mn ) confined within a hydrophobic pocket.
framework SnO clusters located within hydrophobic micro-
form mannose via the Bilik reaction in methanol, as intrazeolitic
2+
SnO clusters isomerized glucose to fructose by base-catalyzed
x
2
2+
5−7
Extra-
LdB−AvE rearrangement as also occurred in water. SnOx
clusters deposited on external zeolite crystal surfaces and on
amorphous silica also isomerized glucose to fructose in liquid
x
pores of pure-silica zeolite beta, but not at external crystallite
surfaces or on amorphous supports, are also able to isomerize
glucose to fructose in aqueous solutions. Unlike the framework
Sn centers, however, these extra-framework intrazeolitic SnOx
clusters act as solid bases that catalyze glucose isomerization via
LdB−AvE (Lobry de Bruyn−Alberda van Ekenstein) rearrange-
2
methanol by the enolate mechanism, in contrast with their
inability to do so in liquid water. Thus, only zeolites that
contained framework Sn sites showed differences in the
Received: April 8, 2014
Revised: May 29, 2014
2
,8
ments involving enolate intermediates, and the hydrophobic
surrounding voids appear to protect SnO surface sites from
x
©
XXXX American Chemical Society
2288
dx.doi.org/10.1021/cs500466j | ACS Catal. 2014, 4, 2288−2297

ACS Catalysis
Research Article
17
2
predominant mechanism by which glucose is reacted in
aqueous and methanol solvent.
et al., then our previous experimental results suggest that
active sites in Sn-Beta in methanol may be altered in a manner
that precludes the involvement of the neighboring silanol in the
reaction pathway.
Framework Sn centers in Sn-Beta were proposed by Corma
1
5
et al. to be present in both “closed” and “open” forms that
respectively correspond to a nonhydrolyzed Sn site (Sn-(OSi)4)
Here, we provide new experimental results that further
examine the structure of active Sn sites in Sn-Beta and their
influence on glucose isomerization and epimerization reactivity.
The mechanistic role of the silanol group adjacent to the open
Sn site is examined by exchanging its proton with a sodium
cation. We provide evidence that Na-exchanged Sn-Beta
catalyzes the epimerization of glucose to mannose via 1,2
intramolecular carbon shift with high selectivity in methanol
and in concentrated aqueous NaCl solutions. In water, the
selectivity to isomerization to fructose via 1,2 intramolecular
and a partially hydrolyzed Sn site ((HO)-Sn-(OSi) ) (Scheme
3
1a and b, respectively). The open site was proposed to be more
Scheme 1. Schematic Representation of the Dehydrated
States of (a) Closed and (b) Open Sites in Sn-Beta, (c) the
Na-Exchanged Open Site, and (d) the NH -Dosed Open
3
a
Site
+
hydride shift increases with reaction time because Na ions are
removed during reaction and neighboring silanol groups are
restored. These data clearly show that the open Sn site is the
active site for both glucose isomerization and epimerization
reactions, with isomerization prevailing when adjacent silanol
groups are in their proton form and epimerization prevailing
+
when adjacent silanol groups are exchanged with Na .
2
. EXPERIMENTAL METHODS
.1. Synthesis of Sn-Beta, 119Sn-Beta, Na-Sn-Beta, and
2
119
Si-Beta. Sn-Beta and Sn-Beta were synthesized according to
previously reported procedures. A total of 15.25 g of
a
“X” denotes framework O−Si units.
1
1
5
reactive in the Baeyer−Villiger oxidation of cyclic ketones,
and we have also similarly proposed the open site to be more
reactive in the isomerization of glucose to fructose. Previously,
we attempted to increase the proportion of open Sn sites by
substitution of SnCl4 precursors with Sn(CH )Cl during
crystallization and subsequent Na exchange prior to
1
1
19
3
3
+
calcination to prevent condensation of proximal Sn−OH and
1
Si−OH groups. The reactivity of these modified Sn-Beta
samples were indistinguishable from Sn-Beta crystallized using
SnCl precursors, indicating that open and closed Sn sites were
interconvertible during calcination and reaction conditions. As
a result, while we were able to distinguish between open and
closed Sn sites in Sn-Beta in Sn NMR spectra, our previous
4
1
1
19
experimental data could not conclusively determine which site
(5 wt % of SiO in gel) to this gel and mixed. The final gel was
2
1
(
or both) was the active site for glucose isomerization.
Quantum chemical studies suggest that glucose−fructose
transferred to a Teflon-lined stainless steel autoclave and heated
at 413 K in a static oven for 40 days. The recovered solids were
centrifuged, washed extensively with water, and dried at 373 K
overnight. The dried solids were calcined in flowing air (1.67
isomerization pathways are catalyzed with lower barriers on
open than on closed sites. Khouw and Davis exchanged Na+
1
16
3
−1
−1
onto silanol groups adjacent to open Ti sites ((HO)−Ti−
cm s , Air Liquide, breathing grade) at 853 K (0.0167 K s )
(
OSi) ) in TS-1 and completely inhibited catalytic alkane
for 10 h to remove the organic content located in the crystalline
3
1
6
119
oxidation reactions with hydrogen peroxide, providing
conclusive evidence that open Ti sites were the active sites
for alkane oxidation. Thus, it seems plausible that the silanol
group adjacent to the open Sn site in Sn-Beta could influence
the rates and selectivities of glucose isomerization catalysis. Rai
material.
Sn-Beta was calcined twice under the same
conditions.
Na-Sn-Beta was synthesized using the same procedure as Sn-
Beta, but with the addition of NaNO (Sigma-Aldrich, ≥99.0%)
3
to the synthesis gel. The final molar composition of the gel was
1 SiO /x NaNO /0.0077 SnCl /0.55 TEAOH/0.54 HF/7.52
1
7
et al. used density functional theory to calculate a lower
energy glucose−fructose isomerization pathway when glucose
binds to an open Sn site in a monodentate mode that involves
the adjacent silanol group, relative to that when glucose binds
to an open Sn site in a bidentate mode that does not involve
the neighboring silanol group. Conversely, the bidentate
binding mode results in a lower energy pathway for glucose−
mannose epimerization than does the monodentate binding
2
3
4
H O, where “x” was 0.010, 0.017, and 0.033 (Na-Sn-Beta-100,
2
-60, and -30, respectively). The gel was transferred to a Teflon-
lined stainless steel autoclave and heated at 413 K in a static
oven for 25 days. The recovered solids were washed, dried, and
calcined using the same procedure as for Sn-Beta. Synthesis gels
with a Si/Na ratio lower than 30 yielded a heterogeneous
material with small black particles dispersed among the zeolite.
These black particles were separated from the zeolite by hand
and were found to be amorphous, having a Si/Sn and Na/Sn
ratio of 15 and 2.28, respectively.
1
7
mode. If epimerization is a lower energy pathway than
isomerization when silanol groups adjacent to Sn sites are not
involved in the mechanism, a detail that is not addressed by Rai
2
289
dx.doi.org/10.1021/cs500466j | ACS Catal. 2014, 4, 2288−2297

ACS Catalysis
Research Article
Si-Beta was prepared by adding 10.01 g of tetraethylammo-
prior to recording dry sample weight. For Sn-Beta-NH , the
3
nium fluoride dihydrate (Sigma-Aldrich, 97% (w/w) purity) to
temperature during the degassing procedure never exceeded
−1
10 g of water and 4.947 g of tetraethylorthosilicate (Sigma-
473 K (0.167 K s ). Relative pressures (P/P ) were measured
0
between 10− and 1 at 87 K with precise volumetric Ar doses.
Deuterated acetonitrile dosing and desorption experiments
were performed according to the procedure described else-
7
Aldrich, 98% (w/w)). This mixture was stirred overnight at
room temperature in a closed vessel to ensure complete
hydrolysis of the tetraethylorthosilicate. The targeted H O/
2
18
SiO ratio was reached by complete evaporation of the ethanol
where. A Nicolet Nexus 470 Fourier transform infrared
2
and partial evaporation of the water. The final molar
(FTIR) spectrometer with a Hg−Cd−Te (MCT) detector was
−1
−1
composition of the gel was SiO /0.55 TEAF/7.25 H O. The
used to record spectra in 4000−650 cm range with a 2 cm
−2
2
2
gel was transferred to a Teflon-lined stainless steel autoclave
and heated at 413 K in a rotation oven (60 rpm) for 7 days.
The solids were recovered by filtration, washed extensively with
water, and dried at 373 K overnight. The dried solids were
resolution. Self-supporting wafers (10−20 mg cm ) were
pressed and sealed in a heatable quartz vacuum cell with
3
removable KBr windows. The cell was purged with air (1 cm
−1
s , Air Liquide, breathing grade) while heating to 373 K
(0.0167 K s ), where it was held for 12 h, followed by
3
−1
−1
calcined in flowing air (1.67 cm s , Air Liquide, breathing
−1
grade) at 853 K (0.0167 K s ) for 10 h to remove the organic
content located in the crystalline material.
evacuation at 373 K for >2 h (<0.01 Pa dynamic vacuum; oil
diffusion pump), and cooling to 308 K under a dynamic
+
+
2
.2. Na and H Ion Exchange of Zeolite Samples. Each
vacuum. CD CN (Sigma-Aldrich, 99.8% D atoms) was purified
3
ion exchange step was carried out for 24 h at ambient
temperature, using 45 mL of exchange or wash solution per 300
mg of starting solids. For the procedures involving multiple ion-
exchange steps, the ion-exchange solution was replaced every
by three freeze (77 K), pump, thaw cycles, then dosed to the
sample at 308 K until the Lewis acid sites were saturated. At
this point, the first FTIR spectrum in the desorption series was
recorded. The cell was evacuated down to 13.3 Pa, and the
second spectrum was recorded. Then, the cell was evacuated
under a dynamic vacuum while heating to 433 K (0.0167 K
24 h without intermediate water washing. One, two, and three
successive sodium ion exchanges (Sn-Beta-1Ex, Sn-Beta-2Ex,
and Sn-Beta-3Ex, respectively) were performed by stirring
−1
s ). Concurrently, a series of FTIR spectra were recorded (2
min for each spectrum) at 5 min intervals. The resulting spectra
were baseline-corrected, and the most illustrative spectra were
chosen for presentation. The spectra are not normalized by the
number of Sn sites. Spectral artifacts known as “interference
fringes” were removed using a computational method based on
calcined Sn-Beta in a solution of 1 M NaNO (Sigma-Aldrich,
3
≥
99.0%) and 10−4 M NaOH (Alfa Aesar 97%) in distilled
water. The final material was recovered by centrifugation, and
washed three times with 1 M NaNO in distilled water. Acid-
3
washed Sn-Beta (Sn-Beta-AW) was made by stirring the triply
19
sodium-exchanged Sn-Beta (Sn-Beta-3Ex) in 1 M H SO4
digital filtering techniques and Fourier analysis.
2
(
Macron Fine Chemicals, >51%) for 1 h at ambient
Solid-state magic angle spinning nuclear magnetic resonance
(MAS NMR) measurements were performed using a Bruker
Avance 500 MHz spectrometer equipped with a 11.7 T magnet
and a Bruker 4 mm broad band dual channel MAS probe. The
temperature, followed by separation by filtration and washing
with 1 L of distilled water in 100 mL batches. Finally, the
material was dried in room temperature air and calcined in
3
−1
1
flowing air (1.67 cm s , Air Liquide, breathing grade) at 853
operating frequencies were 500.2 and 186.5 MHz for H and
−1
119
K (0.0167 K s ). We note that the dehydration of sodium-
exchanged materials resulted in changes in their catalytic
properties; therefore, to ensure comparable saturation of the
samples with water, 24 h prior to reaction testing, all samples
were placed in a chamber whose humidity was controlled by a
saturated NaCl solution.
Sn nuclei, respectively. Approximately 60−80 mg of powder
was packed into 4 mm ZrO rotors and spun at 14 kHz for
2
MAS or cross-polarization (CP) MAS experiments under
1
19
1
ambient conditions. The
Sn{ H} CP condition was
optimized at a radiofrequency pulse power of 62.5 kHz ± νr,
where ν is spinning frequency, and spectra were recorded using
r
2
.3. Ammonia Adsorption onto Sn-Beta. Ammonia gas
a 2 ms contact time. The recycle delay times were 20 and 2 s
for 1 Sn MAS and CPMAS experiments, respectively. Signal
averaging over 8000 scans was performed for the CPMAS
19
dosing experiments were performed on Sn-Beta samples after
drying in a Schlenk flask at 473 K for 2 h under a vacuum. The
dried Sn-Beta was cooled under a dynamic vacuum to ambient
temperature, and the flask was backfilled with 101 kPa of
anhydrous ammonia gas (Matheson Tri-Gas, 99.99%). After 24
h, the excess ammonia was evacuated, and the sample was
spectrum of 1 Sn-Beta dehydrated after NH dosing, while
19
3
averaging over 30 000 scans was performed for the CPMAS
119
spectrum of Sn-Beta dehydrated after three Na-exchanges.
Liquid 13C NMR spectra were recorded using a Varian
INOVA 500 MHz spectrometer equipped with an auto-x pfg
broad band probe. Carbon chemical shifts are reported relative
exposed to the atmosphere (Sn-Beta-NH ). The ammonia-
saturated material was regenerated by calcination (Sn-Beta-
3
3
−1
13
NH -Cal) in flowing air (1.67 cm s , Air Liquide, breathing
to the residual solvent signal. C NMR spectra were acquired
3
−1
grade) for 6 h at 853 K (0.0167 K s ).
.4. Characterization Methods. Scanning electron
microscopy (SEM) with Energy Dispersive X-ray Spectroscopy
EDS) measurements were recorded on a LEO 1550 VP FE
with 2000 scans.
2
2.5. Reaction Procedures. Reactions with D-glucose
(Sigma-Aldrich, ≥99%) were conducted in 10 mL thick-walled
glass reactors (VWR) that were heated in a temperature-
controlled oil bath. Reactions were prepared with a 1:100 Sn/
glucose molar ratio using 5.0 g of a 1% (w/w) glucose solution
with approximately 20 mg of catalyst. For reactions performed
to investigate the effects of addition of NaCl to aqueous glucose
reactant solution, 0.2 g of NaCl were added per 1.0 g of 1% (w/
w) glucose solution. Reactors were placed in the oil bath at 353
K, and approximately 50 mg aliquots were extracted at 10, 20,
and 30 min. These reaction aliquots were mixed with 50 mg of
(
SEM at an electron high tension (EHT) of 15 kV. The
crystalline structures of zeolite samples were determined from
powder X-ray diffraction (XRD) patterns collected using a
Rigaku Miniflex II diffractometer and Cu Kα radiation.
Ar adsorption isotherms at 87 K were obtained using a
Quantachrome Autosorb iQ automated gas sorption analyzer.
−1
Zeolite samples were degassed at 353 K (0.167 K s ) for 1 h,
93 K (0.167 K s ) for 3 h, and 623 K (0.167 K s ) for 8 h
−1
−1
3
2
290
dx.doi.org/10.1021/cs500466j | ACS Catal. 2014, 4, 2288−2297

ACS Catalysis
Research Article
a 1% (w/w) D-mannitol (Sigma-Aldrich, ≥98%) solution as an
internal standard for quantification, diluted with 0.3 mL of
the Sn atoms in the synthesis gels form alkali stannates that are
part of the amorphous phase impurity, thereby lowering the Sn
and Na content of the crystalline Na-Sn-Beta that is formed.
The total micropore volumes of the samples were
determined from Ar adsorption isotherms (87 K, Supporting
Information Figures S.5−S.14) and are listed in Table 1. The
H O, and filtered with a 0.2 μm PTFE syringe filter.
2
Glucose conversions and product yields were calculated by
XGluc(t) = n
Gluc(t = 0) − nGluc(t)
× 100 [%]
nGluc(t = 0)
(1)
(2)
Table 1. Site and Structural Characterization of Samples
Used in This Study
ni(t)
(t = 0)
Yi(t) = nGluc
× 100 [%]
b
Si/
Na/ Ar micropore volume
a
3 −1
a
IR bands (cm−1)
c
catalyst
Sn-Beta
Sn
Sn
(cm g )
where XGluc(t) is the glucose conversion at time t, in percent;
Yi(t) is the fructose or mannose yield at time t, in percent;
nGluc(t = 0) is the initial moles of glucose in the reactor; nGluc(t)
95
0.00
3.80
4.38
4.85
0.27
0.12
0.26
0.94
0.00
0.00
0.19
0.16
0.15
0.16
0.19
0.18
0.19
0.15
0.17
0.19
2315, 2307, 2276,
and 2266
Sn-Beta-1Ex
Sn-Beta-2Ex
Sn-Beta-3Ex
Sn-Beta-AW
115
159
140
104
113
127
91
2310, 2280, and
2274
is the moles of glucose in the reactor at time t; and n(t) is the
i
moles of fructose or mannose in the reactor at time t.
Recyclability experiments were performed with Sn-Beta-3Ex
reacted with glucose in water and methanol under the
previously stated reaction conditions (353 K for 30 min in a
2310, 2280, and
2
274
2310, 2280, and
274
2
2315, 2307, 2276,
and 2266
1
% (w/w) with 1:100 Sn/glucose molar ratio) and washed
d
once with the solvent used in the reaction. The solids were
centrifuged and dried with ambient temperature air.
Na-Sn-Beta-
100
n.d.
d
Reaction aliquots were analyzed by high performance liquid
chromatography (HPLC) using an Agilent 1200 system
Na-Sn-Beta-
n.d.
6
0
Na-Sn-Beta-
2310, 2280, and
2274
(
Agilent) equipped with diode array and evaporative light
3
0
scattering detectors. Glucose, fructose, mannose, and mannitol
fractions were separated with a Hi-Plex Ca column (6.5 × 300
mm, 8 μm particle size, Agilent) held at 358 K, using ultrapure
Sn-Beta-
NH3
105
117
2306 and 2270
Sn-Beta-
2315, 2307, 2276,
and 2266
−1
NH -Cal
water as the mobile phase at a flow rate of 0.6 mL s .
3
1
3
a
Reactions with labeled C glucose at the C1 position
Determined by Energy Dispersive X-ray Spectroscopy (EDS). The
1
3
highest measured Si/Sn standard deviation for three scans of different
parts of the same material was ±30, while the highest Na/Sn standard
deviation was of ±1.25. These maximal standard deviations may be
used to estimate the uncertainty of measurement for all samples.
(Cambridge Isotope Laboratories, 1- C D-glucose, 98−99%)
and deuterium (D) in the C2 position of glucose (Cambridge
Isotope Laboratories, D-glucose-D2, >98%) were conducted
under the same conditions as those with D-glucose. The
reaction was ended by quenching after 30 min. The reaction
solution was filtered and rotavaporated to separate the solvent
from the reactant−product mixture. These recovered solids
b
c
Determined from the Ar adsorption isotherm (87 K). After CD CN
3
d
adsorption. n.d., not determined.
were dissolved in deuterium oxide and analyzed using 13
NMR.
C
micropore volumes of all Na exchanged materials decreased,
perhaps because of excess NaNO that remains on the solid
3
after Na exchange. The final wash in the exchange procedure
3
. RESULTS AND DISCUSSION
.1. Characterization of Microporous Materials. The
was performed with 1 M NaNO because washing with distilled
3
+
3
water results in partial Na removal. The FTIR spectra (Figure
powder X-ray patterns of Sn-Beta, Sn-Beta-1Ex, Sn-Beta-2Ex,
Sn-Beta-3Ex, Sn-Beta-AW, Sn-Beta-NH , Sn-Beta-NH -Cal and
S.15) of the Na exchanged materials indeed show a broad
−1
shoulder in the 1300 to 1500 cm range as would be expected
3
3
for the NO3− ion. Sn-Beta-AW has the same micropore
21
Na-Sn-Beta (Si/Na = 100, 60 and 30) (Supporting Information
Figures S.1 and S.2) show that each of the samples is highly
crystalline and has the zeolite beta framework topology. No
diffraction lines were observed at 2θ values of 26.7° and 34.0°
3
−1
volume (0.19 cm g ) as the parent Sn-Beta material, showing
that the measured decrease in microporosity for the Na-
exchanged materials is not due to a loss of crystallinity, but
that are characteristic of bulk SnO . SEM images (Supporting
likely due to the excess NaNO . Na-Sn-Beta-60 and Na-Sn-
2
3
Information Figure S.3) indicate that the crystallite size of Sn-
Beta is between 5 and 8 μm, and does not change significantly
after exchange with NaNO /NaOH or treatment with NH .
Beta-100 exhibit a similar micropore volume to Sn-Beta (Table
1), but Na-Sn-Beta-30 has a lower micropore volume of 0.14
cm g . This significant decrease in micropore volume is likely
3
−1
3
3
Na-Sn-Beta-30 (SEM images in Supporting Information Figure
S.4) and other materials with gel Si/Na ratios >30 (results not
presented here because of the high impurity content) contain
an impurity that consists of dark, amorphous (by powder XRD)
particles that are not observed in Na-Sn-Beta-60 and Na-Sn-
Beta-100. Thus, synthesis gels with high amounts of Na formed
contaminating amorphous solids with high Na and Sn contents
due to the amorphous particle impurities. The ammonia-dosed
3
−1
Sn-Beta showed a lower micropore volume of 0.17 cm g ,
which increased to that of the parent Sn-Beta sample after
3
−1
calcination (0.19 cm g ).
Table 1 lists the Sn and Na contents for all of the samples in
this study. The Na/Sn ratio increased with the number of
consecutive sodium ion exchanges, with the highest ratio being
20
(
Si/Sn = 15, Na/Sn = 2.28). Bellusi et al. proposed that the
4.38 after three consecutive exchanges with NaNO /NaOH.
3
insertion of titanium into the silicate framework (TS-1) is
inhibited when alkali metal ions are present in the synthesis gel
due to the formation of alkali titanates. Here, it is possible that
Na/Sn ratios above unity likely reflect the presence of excess
NaNO deposited on the sample and some Na exchange
3
occurring at silanol groups other than the ones adjacent to
2
291
dx.doi.org/10.1021/cs500466j | ACS Catal. 2014, 4, 2288−2297

ACS Catalysis
Research Article
open Sn centers. Acid treatment removed most of the sodium
from the zeolite, as the Na/Sn ratio in Sn-Beta-AW decreased
to 0.27. The Na/Sn ratio in the solids synthesized in the
presence of sodium (Na-Sn-Beta-100, -60, and -30) as the
sodium concentration increased in the synthesis gels.
After NaNO /NaOH treatments (Figures 1b, S.16, and
3
S.17), IR bands associated with CD CN bound to the open and
3
closed sites disappear or diminish in intensity, while a single
broad IR band with low intensity appears at ∼2310−2312
−1
cm . We speculate that this broad band may reflect multiple
3
.2. Structural Characterization of the Sn Sites in Sn-
contributions from residual nonexchanged Sn sites, by
24
Beta. The nature of Lewis acidic Sn sites in Sn-Beta and
postsynthetically treated Sn-Beta samples was probed by
monitoring changes in IR bands for CN stretching vibrations
extension of previous reports by Corma et al. showing that
a similar broad band at 2310 cm− in Sn-MCM-41 may be
deconvoluted into multiple bands that correspond to different
Sn environments. Interestingly, a more prominent IR band
appears at 2280 cm− in Na-exchanged Sn-Beta materials
(Figures 1b, S.16, and S.17), which we tentatively associate with
the Lewis acid site responsible for the reactivity of these
1
−
1
22
of adsorbed deuterated acetonitrile (2260−2340 cm )
1
during temperature-programmed desorption experiments (Fig-
ures 1 and S.16−S.19). The IR spectra for Sn-Beta exposed to
−1
samples. The lower frequency of this new band (2280 cm )
−1
compared to the open Sn site in Sn-Beta (2315 cm ) may
22
suggest a weaker interaction of CD CN with Lewis acid sites
3
in Na-exchanged Sn-Beta. The CD CN that gives rise to the
3
280 cm− band, however, desorbs more slowly than CD CN
1
2
3
−1
bound to the closed site (2307 cm ) and at comparable rates
to CD CN bound to the open site (Figure 1a and b). These
3
findings suggest that, in addition to the direct electron donation
of CD CN to the Lewis acidic Sn center, secondary interactions
3
of CD CN with the site or its surrounding environment may
3
influence the binding strength and thus the ν(CN) of
−1
CD CN. Strongly bound CD CN at 2280 cm (Figure S.18)
3
3
was not present on Na-exchanged Si-Beta, confirming that this
IR band is not a result of CD CN adsorbed to Na-exchanged
3
terminal silanol groups and requires the presence of a
framework Sn site. The synthetic Na-Sn-Beta-30 sample
showed a similar desorption profile to that of Na-exchanged
+
Sn-Beta (Figure S.19), suggesting that Na ions introduced
during or after synthesis have similar effects on the ability of
Lewis acid sites in Sn-Beta to bind CD CN.
3
CD CN adsorption onto Sn-Beta-NH gives rise to IR bands
3
3
−1
at 2306 cm , associated with the closed Sn sites (as confirmed
119
by Sn NMR, vide infra) and a previously unobserved band at
1
2
270 cm− (Figure 1c) for sites that desorb CD CN at a rate
3
similar to that of the closed sites. The IR band of CD CN
3
bound to the open site at 2315 cm− was not observed (Figure
1
1
c). These data indicate that NH remains bound only to open
3
Sn sites in Sn-Beta-NH after exposure to ambient air and
3
treatment in a vacuum at 373 K prior to CD CN exposure
3
(
Scheme 2), consistent with proposals that open Sn sites are
15
stronger Lewis acid sites than closed sites. We propose that
Scheme 2. Schematic Representation of Open (top row) and
Closed (bottom row) Sites in Sn-Beta after Different
a
Treatment Procedures
Figure 1. Baseline corrected IR spectra with decreasing CD CN
3
coverage on (a) Sn-Beta, (b) Sn-Beta-3Ex, and (c) Sn-Beta-NH3.
CD CN show bands at 2315, 2307, 2276, and 2266 cm−
1
3
−1
(
Figure 1a). The CD CN IR bands at 2276 and 2266 cm
3
have been assigned to CD CN coordinated to silanol groups
3
and physisorbed CD CN, respectively, while the bands at 2315
3
and 2307 cm− have been assigned to CD CN coordinated to
1
3
2
2,23
Lewis acid sites.
al., who assigned the 2316 cm band to CD CN bound at
the open Sn site and the 2308 cm band to CD CN bound at
These results are consistent with Corma et
a
(a) Hydrated open and closed sites after (b) dehydration and
1
5
−1
3
saturation with NH , followed by subsequent (c) exposure to ambient
3
−1
3
atmosphere and heated evacuation (373 or 393 K for IR and NMR
a weaker Lewis acid site proposed to be the closed Sn site.
studies, respectively). “X” denotes framework O−Si units.
2
292
dx.doi.org/10.1021/cs500466j | ACS Catal. 2014, 4, 2288−2297

ACS Catalysis
Research Article
open Sn sites with bound NH in Sn-Beta-NH (Scheme 1d)
−400 to −480 ppm shown in Figure S.20 of the Supporting
3
3
1
19
are more electron-rich, and in turn bind CD CN more weakly,
Information. Dehydration of the triply-Na-exchanged Sn-
3
+
than open Sn sites without NH ligands in Sn-Beta (Scheme
Beta shows that the introduction of Na to the sample causes a
3
1
b). The open Sn site with preadsorbed NH (Scheme 1d)
shift of the open Sn site resonance at −423 ppm (Figure 2d) to
3
−1
seems to be a likely candidate for the IR band at 2270 cm
−419 ppm (Figure 2e) but does not shift the closed site
1
19
(Figure 1c), which arises from weakly bound CD CN that
resonance at −443 ppm (Figure 2d and e). The Sn-Beta
3
−1
119
disappears more rapidly than the IR band at 2315 cm for
CD CN bound at open Sn sites (Figure 1a). The presence of
the 2306 cm IR band in Sn-Beta-NH suggests that any NH3
initially bound to the closed site (Scheme 2b) desorbs after
exposure to ambient air or vacuum treatment at 373 K (Scheme
NMR spectrum of the dehydrated triply-Na-exchanged Sn-
3
Beta also contains a small shoulder at −435 ppm that was
−1
3
confirmed not to be a spinning sideband of another resonance
1
119
(
Figure 2e). The H− Sn CPMAS NMR spectrum of
119
dehydrated triply-Na-exchanged Sn-Beta shows a resonance
2
c) and forms a closed Sn site similar to that found in the
at −419 ppm, indicating that these Sn centers have a proton
untreated Sn-Beta.
119
source nearby that cross-polarizes the Sn atom (Figure S.21).
The 1 Sn NMR spectra of 119Sn-Beta after calcination and
exposure to ambient conditions, which allows the Sn centers to
become hydrated, shows a main resonance centered at −688
ppm (Figure 2a) for octahedrally coordinated framework
19
This observation suggests that Na exchanges only one of the
two available protons present in the silanol and stannanol
groups in the dehydrated open Sn site (Scheme 1b). The
silanol proton is the more likely position for Na exchange
(
Scheme 1c) in light of the proposed mechanisms for glucose
isomerization and epimerization on Sn-Beta that require
17
bonding to glucose through the stannanol group. We
1
119
experienced difficulties in optimizing H− Sn CPMAS
conditions, partly due to poor rf pulse coverage over 300
ppm during contact period at high spinning speeds (14 kHz in
1
119
this case), that led us to acquire H− Sn CPMAS spectra
(
e.g., Figure S.21b) by averaging over 30 000 transients. The
resonances detected in the tetrahedral range of the dehydrated
Na-exchanged 1 Sn-Beta do not allow us to characterize the
origin of the small −435 ppm shoulder (Figure 2e).
19
Adsorption of ammonia onto 1 Sn-Beta gives rise to two
19
groups of broad resonances centered at −669 and −708 ppm
(
Figure 2c). Dehydration of this sample (evacuation at 393 K, 2
h) gives rise to a resonance for the closed tetrahedral site
−443 ppm, Figure 2f) but not for the open tetrahedral site
(
119
found in Sn-Beta (−423 ppm, Figure 2d). New resonances
1
19
are detected in the dehydrated spectrum of Na- Sn-Beta in
the −500 to −600 ppm range, suggesting the presence of a
different Sn coordination environment, which may originate
from the open site of Sn-Beta depicted in Scheme 1d. The
1
119
119
H− Sn CPMAS NMR spectrum of the
Sn-Beta
dehydrated after ammonia adsorption confirms that there is
no proton source in the neighborhood of the closed site (−443
ppm), or of any tetrahedrally coordinated Sn sites, after these
treatments (Figure S.22).
Figure 2. 119Sn MAS solid state NMR spectra of 119Sn-Beta after
different treatments: (a) calcination, (b) three Na-exchanges after
calcination, (c) NH adsorption after calcination, (d) dehydration after
3
These 119Sn NMR results are consistent with the
interpretations of the IR spectra of Sn-Beta-NH3 after
calcination, (e) dehydration after three Na-exchanges, and (f)
dehydration after NH adsorption.
3
CD CN adsorption, and lead us to propose the Sn structures
and coordinations in Scheme 2. Our findings suggest that the
open Sn site is a stronger Lewis acid site than the closed Sn site
3
1
,2,25
Sn.
Upon treatment in vacuum at 393 K to remove the
coordinating water, the Sn resonances shift to −423 and −443
ppm (Figure 2d) that are characteristic of tetrahedrally
coordinated framework Sn. We have shown previously through
(
Scheme 2a), and that it retains adsorbed NH (Scheme 2b)
3
after vacuum treatment at 373 or 393 K (Scheme 2c). Open Sn
1
119
sites that coordinate one NH ligand would appear as penta-
H− Sn CPMAS NMR that the open and closed sites
3
coordinated Sn sites in 1 Sn NMR spectra, which we speculate
could give rise to the resonances detected in −500 to −600
ppm range (Figure 2f). Penta-coordinated open Sn sites with
19
correspond to the resonances centered at −423 and −443 ppm,
respectively, because only the −423 ppm resonance was
detected when cross-polarization occurred from nearby
2
protons.
one NH
3
ligand would also bind CD
3
CN more weakly than
, and may give rise to
CN band observed in IR spectra (Figure
1c). These NMR data also suggest that NH bound to the
Three Na-exchanges performed on 119Sn-Beta decrease the
intensity of the sharp −688 ppm resonance (Figure 2a) and
form a broad shoulder that begins at −650 ppm and merges
open Sn sites without coordinated NH
3
the 2270 cm− CD
1
3
3
1
19
into the broader features of the Sn-Beta NMR spectrum
closed Sn sites desorbs upon dehydration (Scheme 2c), such
Figure 2b). 1 Sn NMR spectra for samples after dehydration
19
(
that the behavior of the closed Sn sites in Sn-Beta-NH is
3
treatments (vacuum at 398 K, 2h) are shown in Figure 2d−f,
with magnified spectra spanning the chemical shift range from
similar to their behavior in Sn-Beta samples that have not been
treated with NH3.
2
293
dx.doi.org/10.1021/cs500466j | ACS Catal. 2014, 4, 2288−2297

ACS Catalysis
Research Article
3
.3. Mannose Formation with Na Containing Sn-Beta.
from 10.3% to 0.0% with increasing Na/Sn ratio (Table 2).
Similarly, increasing the sodium content in the synthesis gel of
Sn-Beta led to samples that produced higher mannose yields
and lower fructose yields (0.0% fructose for Na-Sn-Beta-30,
Table 2). The large black particles of the amorphous phase
impurity formed from synthesis gels with Si/Na ratios less than
30 were isolated from the crystalline solids and did not react
with glucose in water, but were able to catalyze glucose−
fructose isomerization in methanol.
Na-containing Sn-Beta catalysts showed a higher selectivity
toward mannose for reactions in methanol than in water, and
fructose/mannose ratios significantly increased with increasing
reaction time for reactions in water (Tables 2 and S.1). These
results suggest that sodium decationation could be occurring in
aqueous media at a rate that would cause the selectivity to
change over the time frame of the experiment. Thus, we
investigated the effects of adding sodium salt to the aqueous
reaction solution in order to maintain the sodium content in
the solid more effectively during reaction (Tables 3 and S.2).
Glucose conversion and fructose and mannose yields after
reaction with different Sn-Beta samples in water and methanol
1:100 Sn/glucose ratio, 353 K) for 30 min are given in Table 2
(
Table 2. Glucose Conversion (X) and Fructose and Mannose
Yields (Y) in H O and CH OH Solvents
a
2
3
catalyst
Sn-Beta
solvent
XGluc. (%)
YFruc. (%)
YMann. (%)
H O
6.4
23.2
6.0
5.0
10.3
2.1
3.2
1.8
2.1
2.3
0.0
3.9
6.1
5.1
8.4
4.0
8.0
1.1
0.0
1.9
0.0
3.2
7.2
0.4
3.9
1.8
5.0
2.5
6.7
3.3
7.9
0.0
2.8
1.1
3.3
2.7
3.0
3.5
4.6
2.4
1.9
0.0
2.6
2
CH OH
3
Sn-Beta-1Ex
Sn-Beta-2Ex
Sn-Beta-3Ex
Sn-Beta-AW
Na-Sn-Beta-100
Na-Sn-Beta-60
Na-Sn-Beta-30
Sn-Beta-NH3
H O
2
CH OH
12.6
6.1
3
H O
2
CH OH
12.2
6.8
3
H O
2
CH OH
12.4
5.4
3
H O
2
CH OH
16.9
6.8
3
H O
2
CH OH
19.4
7.3
3
Table 3. Glucose Conversion (X) and Fructose and Mannose
H O
2
a
Yields (Y) with 0.2 g NaCl/g H O
2
CH OH
17.2
5.8
3
H O
2
catalyst
solvent
XGluc. (%)
Y
Fruc. (%)
Y
Mann. (%)
b
CH OH
6.8
Sn-Beta
H O-NaCl
9.8
10.9
10.7
11.5
4.5
2.6
2.5
0.0
4.1
5.2
6.0
7.5
3
2
H O
2
3.8
Sn-Beta-1Ex
Sn-Beta-2Ex
Sn-Beta-3Ex
H O-NaCl
2
CH OH
3.0
H O-NaCl
2
3
Sn-Beta-NH -Cal
H O
2
5.0
H O-NaCl
2
3
CH OH
17.6
a
3
Reaction conditions: 1% (w/w) glucose solutions, 1:100 metal/
b
a
Reaction conditions: 1% (w/w) glucose solutions, 1:100 metal/
glucose ratio, 353 K, 30 min.
glucose ratio, 353 K, 30 min. After reaction the catalyst had a Si/Sn
and a Na/Sn ratio of 115 ± 30 and 2.65 ± 1.25, respectively,
determined by Energy Dispersive X-ray Spectroscopy (EDS).
(data at 10 and 20 min given in Table S.1). Fructose is the
predominant product formed when Sn-Beta reacts with 1%
When glucose was reacted with Sn-Beta in aqueous NaCl
solutions, mannose and fructose were produced in nearly equal
yields (4.1% and 4.5%, respectively; Tables 3 and S.2), and the
solid had a Na/Sn ratio of 2.65 after reaction (Tables 3 and
aqueous glucose solutions (Table 2), with carbon balances
3
(84%) that were similar to those we have reported previously.
Incomplete closure of carbon balances likely reflects the
formation of side products, such as carboxylic acids via retro-
+
S.2) indicating that Na was exchanging into the solid during
26
aldol reactions of sugars. We have previously shown, using
reaction. Sn-Beta pre-exchanged with Na+ (Sn-Beta-3Ex)
13
solid state C MAS NMR, that these side products are present
in zeolites after adsorption of sugars on Sn-Beta in water at
maintained constant mannose selectivity during the course of
the reaction when NaCl was added to the aqueous reaction
solutions. No fructose or mannose formation was observed
without Sn-Beta in aqueous NaCl solution, showing that NaCl
does not catalyze isomerization reactions of glucose. These
results indicate that the presence of a sodium cation, whether
added synthetically or exchanged onto the material prior to or
during the reaction, shifts the reaction selectivity of Sn-Beta
from isomerization to fructose to epimerization to mannose. In
1
ambient temperature, demonstrating that the formation of side
products occurs at early reaction times and that these products
remain adsorbed on the zeolite after reaction. In order avoid
complications in data analysis associated with sugar degradation
side reactions that become more prevalent at higher
conversions, we focus here on low glucose conversions
(<30%) and the initial fructose and mannose products formed.
In water, Na-exchanged Sn-Beta samples led to similar
+
water solvent, the Na ion in the active site is replaced by a
glucose conversions (6.0−6.8%, Table 2) as Sn-Beta (6.4%,
Table 2) at equivalent reaction conditions. However, the
mannose yield increased systematically from 0.4% to 3.3%, and
the fructose yields decreased from 5.0% to ∼2% with increasing
Na content (Table 2). Similar results were observed with
increasing Na content for Na-Sn-Beta samples synthesized
directly (Table 2), suggesting that these selectivity differences
proton, causing a reversal towards the active site structure that
predominantly forms fructose, while in methanol the Na ion is
+
retained in the active site for longer times, resulting in
formation of mannose as the major product. The addition of
excess sodium salt to aqueous reaction mixtures increases the
extent to which Na+ exchanges onto Sn-Beta, in turn
maintaining the selectivity toward mannose during the course
of reaction.
+
do not depend on the method used to introduce Na cations
into Sn-Beta.
The Sn-Beta sample that was dosed with NH showed lower
3
Equivalent reaction conditions in methanol led to higher
glucose conversions on Sn-Beta (23.2%) and Na-containing Sn-
Beta samples (12.2−12.6%) than in water (Tables 2 and S.1).
As in the case of water, mannose yields increased systematically
from 3.9% to 7.9%, and fructose yields decreased systematically
glucose conversions in both water and methanol solvents (3.0−
3.8%, Table 2) than Sn-Beta and the Na-containing Sn-Beta
samples. Higher glucose conversions were observed with Sn-
Beta-NH in water and resulted in a dark yellow postreaction
3
solution, which may indicate the presence of humins formed
2
294
dx.doi.org/10.1021/cs500466j | ACS Catal. 2014, 4, 2288−2297

ACS Catalysis
Research Article
from NH OH that may have formed in situ from the desorption
mannose. All 13C NMR spectra in Figure 3 show the presence
of C in the C1 position (resonances at δ = 95.8 and 92.0
4
13
of NH . Calcination of the ammonia-dosed sample led to a
3
nearly full recovery of the reactivity in methanol and water
(
17.6% and 5.0% glucose conversion, respectively, Table 2).
The suppression of isomerization reactivity on Sn-Beta-NH3
Table 2) occurs together with the disappearance of the open
(
−
1
site CD CN IR band at 2315 cm (Figure 1c) and with the
3
1
19
disappearance of the open site Sn NMR resonance at −423
ppm (Figure 2f) in the dehydrated NH -dosed Sn-Beta. These
3
data corroborate our proposal that the open site is the active
site for the isomerization of glucose to fructose in the absence
of sodium and is the active site for the epimerization of glucose
to form mannose in the presence of sodium.
3
.4. Sodium Removal from Sn-Beta. The Sn-Beta-3Ex
+
sample was acid washed to remove Na from the sample (Sn-
Beta-AW) and probe whether the effects of sodium on the
reactivity of Sn-Beta were reversible. Sn-Beta-AW had much
less sodium (Na/Sn = 0.27) than Sn-Beta-3Ex (Na/Sn = 4.85).
The glucose conversion and the fructose and mannose yields
observed with Sn-Beta-AW were very similar to that of the
parent Sn-Beta (Table 2). The decrease in mannose yield and
concurrent increase in fructose yield after the acid treatment
demonstrates that the effects of sodium addition are reversible
and are not a result of a permanent poisoning of the site active
for glucose−fructose isomerization.
The effect of the reaction solvent on the recyclability of the
catalyst was probed by reacting Sn-Beta-3Ex with glucose in
water and methanol under the previously stated reaction
conditions (353 K for 30 min in a 1% (w/w) glucose solution)
and washing once with the solvent used in the reaction. This
cycle was repeated twice, and the reaction results after each
cycle are shown in Table 4. The Na/Sn ratio of the material
Figure 3. 13C NMR spectra for reactant and products with Sn-Beta in
a 1% (w/w) 13C-C1-glucose solution at 353 K for 30 min with the
following solvent mixtures (a) H O, (b) NaCl-H O, and (c) CH OH.
The abbreviations “gluc,” “fruc,” and “mann” stand for glucose,
fructose, and mannose, respectively.
2
2
3
ppm) of the α and β pyranose forms of the starting labeled
glucose, respectively. The fructose formed from reactions with
Sn-Beta in all three solvents showed 13C in the C1 position
Table 4. Glucose Conversion (X) and Fructose and Mannose
(
resonances at δ = 63.8 and 62.6 ppm) for β-pyranose and β-
a
Yields (Y) with Sn-Beta-3Ex in CH OH and H O
3
2
furanose forms of fructose, as expected from isomerization
1
13
b
b
mediated by 1,2 intramolecular hydride shift. The C label
was only observed in the C1 position (resonances at δ = 93.9
and 93.5 ppm) of α and β pyranose forms of mannose with
water and methanol solvents in Sn-Beta, indicating that
mannose was not formed by a 1,2 intramolecular carbon
shift, but likely via 1,2 intramolecular hydride shift of fructose
Cycle Si/Sn
Na/Sn
Solvent
XGluc.(%) YFruc.(%) YMann.(%)
1
2
3
1
2
3
115
136
123
115
132
119
4.38
0.93
0.26
4.38
1.26
0.82
H O
8.5
8.6
1.7
4.6
6.4
0.0
1.5
3.7
4.5
3.9
1.2
6.5
6.0
6.2
2
H O
2
H O
2
9.0
CH OH
9.4
3
CH OH
10.2
13.7
13
3
products into mannose. In contrast, the C label appeared in
CH OH
3
the C2 position (resonances at δ = 70.5 and 71.1 ppm) of the α
and β pyranose forms of mannose with Sn-Beta in aqueous
NaCl solutions, indicating that mannose was formed by the 1,2
a
After the first cycle the catalysts were washed with the solvent used in
the reaction and reused under the same reaction and solvent
conditions as the previous cycle. Reaction conditions: 1% (w/w)
glucose solutions, 1:100 metal/glucose ratio, 353 K, 30 min.
9
intramolecular carbon shift mechanism of the Bilik reaction.
b
The isotopic labeling experiments performed with Sn-Beta
were also conducted with Sn-Beta-3Ex in water, aqueous NaCl
Determined by Energy Dispersive X-ray Spectroscopy (EDS).
Uncertainty in Si/Sn is ±30. Uncertainty in Na/Sn is ±1.25.
solutions (0.2 g NaCl/g H O), and methanol as solvents, and
2
13
the resulting C NMR spectra are shown in Figure 4. In water,
the fructose products retained the 13C label in the C1 position
(resonances at δ = 63.8 and 62.6 ppm), with a lower intensity
decreased in each cycle, with a greater extent of sodium loss in
the case of aqueous media. A decrease in sodium content in the
zeolite after each cycle also led to a decrease in the mannose
yield and corresponding increase in the fructose yield (Table
1
3
relative to Sn-Beta, and the mannose product showed the
C
label only in the C2 position (resonances at δ = 70.5 and 71.1
ppm). These results, together with reaction data for earlier
reaction times in Table S.2, suggest that Sn-Beta-3Ex initially
forms mannose through the 1,2 intramolecular carbon shift in
water, but the loss of sodium from the active site results in the
formation of fructose without carbon rearrangement. When
methanol or concentrated aqueous NaCl solutions were used as
4
), consistent with the proposal that open Sn sites with Na-
exchanged silanol groups are active sites for the epimerization
reaction.
3
.5. Glucose Isomerization and Epimerization Mech-
1
3
anisms. One percent (w/w) glucose labeled with C at the C1
position ( C-C1-glucose) was reacted at 353 K for 30 min with
13
13
Sn-Beta in water, aqueous NaCl solutions (0.2 g NaCl/g H O),
solvents, mannose with C in the C2 position was observed as
the main product. These results confirm that the switch in
reaction mechanism from isomerization to epimerization of
2
and methanol as solvents to determine the mechanisms of
glucose isomerization to fructose and epimerization to
2
295
dx.doi.org/10.1021/cs500466j | ACS Catal. 2014, 4, 2288−2297

ACS Catalysis
Research Article
of mannose, respectively. This NMR evidence indicates that
with the sodium cation in the active site of Sn-Beta, the carbon
in the C2 position of glucose moves along with its deuterium to
the C1 position by the 1,2 intramolecular carbon shift to form
2
mannose, as we have observed previously. It is clear that the
presence of alkali metal cations in Sn-Beta can determine
whether epimerization can occur. In our initial report on
glucose epimerization into mannose via 1,2 intramolecular
2
carbon shift with Sn-Beta in methanol, we did not purposefully
add alkali metal cations to the synthesis gel for crystallizing Sn-
Beta. We have analyzed the samples used in that study and
found them to contain potassium. Although the exact origin of
the potassium remains unknown to us at this time, its presence
in those samples is likely the reason why we observed
epimerization of glucose in methanol solvent.
4
. CONCLUSION
Partially hydrolyzed Sn sites in zeolite beta (denoted as open
Sn sites) with proximal silanol groups are shown to be the
active sites for the isomerization of glucose into fructose via a
Lewis-acid mediated 1,2 hydride shift mechanism. The
Figure 4. 13C NMR spectra for reactant and products with Sn-Beta-
13
3
Ex in a 1% (w/w) C-C1-glucose solution at 353 K for 30 min with
+
the following solvent mixtures: (a) H O, (b) NaCl-H O, and (c)
exchange of Na onto the adjacent silanol group of the open
2
2
CH OH. The abbreviations “gluc,” “fruc,” and “mann” stand for
Sn sites results in sites that are active for the epimerization o+f
glucose into mannose via a 1,2 intramolecular carbon shift. Na
cations can be exchanged onto silanol groups in active Sn sites
of Sn-Beta either by postsynthetic ion exchange or by addition
3
glucose, fructose, and mannose, respectively.
sodium-exchanged materials is not directly dependent on the
solvent but rather on the presence of sodium in the active site.
Glucose epimerization into mannose can proceed via
reversible enolization upon abstraction of α-carbonyl protons
of low amounts of NaNO during synthesis. Acid washing of
3
Na-exchanged Sn-Beta resulted in nearly full recovery of the
initial reactivity of the parent alkali-free Sn-Beta, thereby
+
showing that any alterations to the active sites by Na are
(
LdB−AvE rearrangement), or via an intramolecular carbon
reversible. Na+ cations remain exchanged onto Sn-Beta in
methanol solvent, but decationation occurs gradually with
27
shift between C1 and C2 positions. In order to confirm that
the Sn-Beta containing Na was not epimerizing glucose to
+
increasing reaction time in aqueous solvent. The addition o+f
mannose by abstraction of the α-carbonyl proton, glucose with
deuterium at the C2 position (glucose-D2) was used as a
reactant. The mannose formed with Sn-Beta-3Ex after 30 min
at 353 K with 1% (w/w) glucose solution in methanol did not
show resonances at δ = 93.9 and 93.5 ppm (Figure 5) that
correspond to the C1 positions of the α and β pyranose forms
NaCl to aqueous reaction solutions appeared to preserve Na
cations exchanged onto silanol groups, as it led to an
enhancement in the selectivity toward epimerization of glucose
into mannose. These findings, in combination with recyclability+
studies performed in water and methanol, indicate that Na
cations are labile under reaction conditions and that the nature
of the solvent influences their lability.
ASSOCIATED CONTENT
Supporting Information
■
*
S
Catalyst characterization (X-ray diffractograms, SEM images,
and IR and solid state NMR spectra) and additional reaction
data are provided. This material is available free of charge via
the Internet at http://pubs.acs.org/.
AUTHOR INFORMATION
Corresponding Author
E-mail: mdavis@cheme.caltech.edu.
■
*
Present Address
†
School of Chemical Engineering, Purdue University, 480
Stadium Mall Drive, West Lafayette, IN 47907
Author Contributions
‡
R.B.D. and M.O. contributed equally to this work.
Notes
The authors declare no competing financial interest.
Figure 5. 13C NMR spectra for (a) reactant and products with Sn-
Beta-3Ex in a 1% (w/w) C-C1-glucose solutions at 353 K for 30 min
ACKNOWLEDGMENTS
This work was financially supported as part of the Catalysis
Center for Energy Innovation, an Energy Frontier Research
13
■
in CH OH and (b) mannose. The abbreviation “mann” stands for
3
mannose.
2
296
dx.doi.org/10.1021/cs500466j | ACS Catal. 2014, 4, 2288−2297

ACS Catalysis
Research Article
Center funded by the U.S. Department of Energy, Office of
Science, Office of Basic Energy Sciences under Award Number
DE-SC0001004. M.O. wishes to acknowledge funding from the
National Science Foundation Graduate Research Fellowship
Program under Grant No. DGE-1144469. Any opinions,
findings, and conclusions or recommendations expressed in
this material are those of the author(s) and do not necessarily
reflect the views of the National Science Foundation.
REFERENCES
■
(
1) Bermejo-Deval, R.; Assary, R. S.; Nikolla, E.; Moliner, M.;
Roman
́
-Leshkov, Y.; Hwang, S.-J.; Palsdottir, A.; Silverman, D.; Lobo,
R. F.; Curtiss, L. A.; Davis, M. E. Proc. Natl. Acad. Sci. U. S. A. 2012,
09, 9727−9732.
2) Bermejo-Deval, R.; Gounder, R.; Davis, M. E. ACS Catal. 2012, 2,
705−2713.
3) Moliner, M.; Roman
Sci. U. S. A. 2010, 107, 6164−6168.
4) Nikolla, E.; Roman-Leshkov, Y.; Moliner, M.; Davis, M. E. ACS
Catal. 2011, 408−410.
5) Collyer, C. A.; Henrick, K.; Blow, D. M. J. Mol. Biol. 1990, 212,
11−235.
6) Bhosale, S. H.; Rao, M. B.; Deshpande, V. V. Microbiol. Rev. 1996,
0, 280−300.
7) Kovalevsky, A. Y.; Hanson, L.; Fisher, S. Z.; Mustyakimov, M.;
1
(
2
(
́
-Leshkov, Y.; Davis, M. E. Proc. Natl. Acad.
(
́
(
2
(
6
(
Mason, S. A.; Forsyth, V. T.; Blakeley, M. P.; Keen, D. A.; Wagner, T.;
Carrell, H. L.; Katz, A. K.; Glusker, J. P.; Langan, P. Structure 2010, 18,
6
(
88−699.
8) Roy, S.; Bakhmutsky, K.; Mahmoud, E.; Lobo, R. F.; Gorte, R. J.
ACS Catal. 2013, 3, 573−580.
9) Hayes, M. L.; Pennings, N. J.; Serianni, A. S.; Barker, R. J. Am.
(
Chem. Soc. 1982, 104, 6764−6769.
(
(
10) Petrus, L. Chem. Zvesti (1947-1984) 1975, 29, 690−693.
11) Bílik, V.; Petrus, L.; Zemek, J. Chem. Zvesti (1947-1984) 1978,
3
2, 242−251.
12) Tanase, T.; Shimizu, F.; Kuse, M.; Yano, S.; Yoshikawa, S.;
Hidai, M. J. Chem. Soc., Chem. Commun. 1987, 2, 659.
13) Tanase, T.; Shimizu, F.; Yano, S.; Yoshikawa, S. J. Chem. Soc.,
Chem. Commun. 1986, 1001−1003.
14) Tanase, T.; Shimizu, F.; Kuse, M.; Yano, S.; Hidai, M.;
Yoshikawa, S. Inorg. Chem. 1988, 27, 4085−4094.
15) Boronat, M.; Concepcion, P.; Corma, A.; Renz, M.; Valencia, S.
(
(
(
(
J. Catal. 2005, 234, 111−118.
(
(
16) Khouw, C.; Davis, M. E. J. Catal. 1995, 151, 77−86.
17) Rai, N.; Caratzoulas, S.; Vlachos, D. G. ACS Catal. 2013, 3,
2
(
294−2298.
18) Gounder, R.; Davis, M. E. J. Catal. 2013, 308, 176−188.
19) Faggin, M. F.; Hines, M. A. Rev. Sci. Instrum. 2004, 75.
20) Bellussi, G.; Fattore, V. In Zeolite Chemistry and Catalysis
(
(
Proceedings of an International Symposium; Jacobs, P. A., Jaeger, N. I.,
Kubelkova, L., Wichterlova, B., Eds.; Elsevier: Amsterdam, 1991; Vol.
9, pp 79−92.
21) Miller, F. A.; Wilkins, C. H. Anal. Chem. 1952, 24, 1253−1294.
22) Pelmenschikov, A. G.; van Santen, R. A.; Janchen, J.; Meijer, E. J.
Phys. Chem. 1993, 97, 11071−11074.
23) Wichterlova, B.; Tvaruzkova,
Microporous Mesoporous Mater. 1998, 24, 223−233.
24) Boronat, M.; Concepcion, P.; Corma, A.; Navarro, M. T.; Renz,
M.; Valencia, S. Phys. Chem. Chem. Phys. 2009, 11, 2876−2884.
25) Corma, A.; Nemeth, L. T.; Renz, M.; Valencia, S. Nature 2001,
12, 423−425.
26) Holm, M. S.; Saravanamurugan, S.; Taarning, E. Science 2010,
28, 602−605.
27) Osanai, S. Top. Curr. Chem. 2001, 215, 43−76.
́
6
(
(
̈
(
́
̊
̌
́
Z.; Sobalík, Z.; Sarv, P.
(
́
(
4
(
3
(
2
297
dx.doi.org/10.1021/cs500466j | ACS Catal. 2014, 4, 2288−2297