Shape-selective Valorization of Biomass-derived Glycolaldehyde using Tin-containing Zeolites

Article abstract of DOI:10.1002/cssc.201600757

A highly selective self-condensation of glycolaldehyde to different C₄ molecules has been achieved using Lewis acidic stannosilicate catalysts in water at moderate temperatures (40–100 °C). The medium-sized zeolite pores (10-membered ring framework) in Sn-MFI facilitate the formation of tetrose sugars while hindering consecutive aldol reactions leading to hexose sugars. High yields of tetrose sugars (74 %) with minor amounts of vinyl glycolic acid (VGA), an α-hydroxyacid, are obtained using Sn-MFI with selectivities towards C₄ products reaching 97 %. Tin catalysts having large pores or no pore structure (Sn-Beta, Sn-MCM-41, Sn-SBA-15, tin chloride) led to lower selectivities for C₄ sugars due to formation of hexose sugars. In the case of Sn-Beta, VGA is the main product (30 %), illustrating differences in selectivity of the Sn sites in the different frameworks. Under optimized conditions, GA can undergo further conversion, leading to yields of up to 44 % of VGA using Sn-MFI in water. The use of Sn-MFI offers multiple possibilities for valorization of biomass-derived GA in water under mild conditions selectively producing C₄ molecules.

Full text of DOI:10.1002/cssc.201600757

DOI: 10.1002/cssc.201600757
Full Papers
Very Important Paper
Shape-selective Valorization of Biomass-derived
Glycolaldehyde using Tin-containing Zeolites
Søren Tolborg,[a, b] Sebastian Meier,[a] Shunmugavel Saravanamurugan,[a] Peter Fristrup,[a]
Esben Taarning,[b] and Irantzu Sꢀdaba*[b]
A highly selective self-condensation of glycolaldehyde to differ-
ent C4 molecules has been achieved using Lewis acidic stanno-
silicate catalysts in water at moderate temperatures (40–
1008C). The medium-sized zeolite pores (10-membered ring
framework) in Sn-MFI facilitate the formation of tetrose sugars
while hindering consecutive aldol reactions leading to hexose
sugars. High yields of tetrose sugars (74%) with minor
amounts of vinyl glycolic acid (VGA), an a-hydroxyacid, are ob-
tained using Sn-MFI with selectivities towards C4 products
reaching 97%. Tin catalysts having large pores or no pore
structure (Sn-Beta, Sn-MCM-41, Sn-SBA-15, tin chloride) led to
lower selectivities for C4 sugars due to formation of hexose
sugars. In the case of Sn-Beta, VGA is the main product (30%),
illustrating differences in selectivity of the Sn sites in the differ-
ent frameworks. Under optimized conditions, GA can undergo
further conversion, leading to yields of up to 44% of VGA
using Sn-MFI in water. The use of Sn-MFI offers multiple possi-
bilities for valorization of biomass-derived GA in water under
mild conditions selectively producing C4 molecules.
Introduction
Glycolaldehyde (GA), a two-carbon monosaccharide, has the
potential to become a valuable biomass-derived platform
chemical. It is currently obtained in high yields in supercritical
water or as a main component of bio-oil, and other processes
to efficiently transform sugars to GA in high yields are current-
ly being developed.[1]
availability could provide a new platform for biomass-derived
products. Especially products that are otherwise difficult to
access from cheaper pentose and hexose sugars are attractive.
The sweetener erythritol, for example, is currently obtained by
fermentation of glucose and sucrose, but could be obtained
through hydrogenation of the tetrose sugars.[5] Under hydro-
thermal alkaline conditions the monomer lactic acid has also
been obtained, by way of aldol condensation and subsequent
conversion of GA.[6] Similarly, several interesting products have
been obtained by employing Lewis-acid catalysts, including
the a-hydroxyacids lactic acid/methyl lactate (LA/ML), methyl
vinyl glycolate (MVG), and methyl 4-methoxy-2-hydroxy-3-bu-
tenoate (MMHB).[7] In addition to these, the formation of the
pharmaceutical precursor a-hydroxy-g-butyrolactone (HBL)
from smaller sugars (C1 to C3) has been reported.[3,7b,8]
GA can be catalytically converted into many interesting
chemicals. For example, ethylene diamine or ethanolamine,
with applications in polymer production and pharmaceuticals,
can be produced by direct amination.[2] Glyoxal, glycolic acid,
or glyoxylic acid can be obtained by oxidation, whereas ethyl-
ene glycol can be produced by reduction (hydrogenation).
Alternatively, the aldehyde functionality can be employed in
CÀC bond formations, such as aldol condensations.[3] The self-
condensation of GA, leading to the formation of tetrose
sugars, permits the production of uncommon and expensive
sugars. The tetroses currently have limited potential uses, for
example within pharmaceutical treatments,[4] but increased
The condensation of GA with small oxygenates such as
formaldehyde (FA) has long been considered a plausible path-
way in the abiotic synthesis of sugars.[9] In the catalytic self-
condensation of GA, the challenge is to control the selectivity
towards tetrose sugars while avoiding the formation of hexo-
ses (Scheme 1). Various strategies have been utilized to control
the aldol condensation of smaller sugars towards the desired
condensation products. The most common catalysts used in
this reaction include bases,[10] amino acids,[11] and peptides.[12]
Inorganic hydroxides such as calcium hydroxide have been ex-
tensively studied, alongside various minerals such as borates,[13]
silicates,[14] and phosphates.[9b] A combined yield of tetrose de-
rivatives of 86% was obtained using GA and borates at moder-
ate temperatures (658C).[13b]
[a] S. Tolborg, Dr. S. Meier, Dr. S. Saravanamurugan,+ Dr. P. Fristrup
Department of Chemistry
Technical University of Denmark
Kemitorvet, 2800-Kgs. Lyngby (Denmark)
[b] S. Tolborg, Dr. E. Taarning, Dr. I. Sꢀdaba
New Business R&D
Haldor Topsøe A/S
Haldor Topsøes Allꢁ 1, 2800-Kgs. Lyngby (Denmark)
E-mail: irsz@topsoe.dk
[+] Current address:
Center of Innovative and Applied Bioprocessing (CIAB)
Mohali, 160071-Punjab (India)
One of the major advantages of zeolite catalysis is the ability
to control product selectivity by choosing materials with suita-
ble pore sizes.[15] Zeolites have previously been used with
Supporting Information and the ORCID identification numbers for the
authors of this article can be found under http://dx.doi.org/10.1002/
cssc.201600757.
ChemSusChem 2016, 9, 1 – 9
1
ꢁ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Full Papers
We show here that the medium-pore zeolite Sn-MFI catalyz-
es the formation of tetrose sugars and vinyl glycolic acid with
high selectivity (up to 98%) resulting in a combined yield of C4
products of up to 80% (at 90% conversion), while avoiding
the formation of hexoses (<2%). Owing to the mild reaction
conditions (808C), the catalyst shows good reusability over
several cycles.
Results and Discussion
Catalyst preparation and characterization
Crystalline (MFI and Beta) zeolites and ordered amorphous
(MCM-41 and SBA-15) silicates were chosen for their differen-
ces in channel diameter, ranging from 5.5 ꢂ (MFI) and 6.5 ꢂ
(Beta) to ca. 30 ꢂ (MCM-41) and ca. 40 ꢂ (SBA-15).[20] All tin-
containing catalysts (MFI, Beta, MCM-41, and SBA-15) were pre-
pared by direct hydrothermal synthesis, following procedures
described in the literature.[18a–c,21] Sn-MFI was prepared follow-
ing two different approaches: i) utilizing hydroxide ions as min-
eralizing agents, yielding the catalyst designated Sn-MFI (100,
OHÀ); or ii ) using fluoride ions, yielding the catalyst designated
Sn-MFI (400, FÀ).[18a] The number in parentheses is the nominal
Si/Sn ratio.
The structure and physical properties of the micro- and mes-
oporous materials were confirmed by X-ray diffraction and N2-
adsorption/desorption measurements (see Supporting Informa-
tion, Figures S1–S3, and Table 1). The prepared Beta and MFI
zeotypes all showed high crystallinity of their respective zeolite
phases (see Figure S1). As expected, the morphology and size
of the prepared Sn-MFI catalyst crystals were highly influenced
by the preparation method. Scanning electron microscopy
(SEM) revealed that the hydroxide route yielded small (0.5–
1.5 mm), spherical crystals, whereas the fluoride route resulted
in much larger, prism-shaped crystals (see Figure 1a–b), ex-
plaining the differences in textural properties seen in Table 1,
entries 1–2. The expected capped square bipyramidal morphol-
ogy was found for the unseeded Beta crystals (Figure 1c).[21]
No discernible morphology was observed for Sn-MCM-41 and
Sn-SBA-15 (not shown).[18c]
Scheme 1. Schematic representation of the conversion of hexose sugars to
the C2 sugar glycolaldehyde (reported elsewhere) followed by aldol conden-
sation to the tetrose sugars and subsequent aldol condensation to hexoses
(a) or isomerization and b-elimination to vinyl glycolic acid (VGA, b).
some success to control the condensation of formaldehyde to
selectively form the C3 sugar 1,3-dihydroxyacetone.[16] Recently,
solid Lewis-acid catalysts have been shown to achieve high se-
lectivity in aldol and cross-aldol reactions for the formation of
a range of chemicals.[3,7a,17] Different micro- and mesoporous
metallosilicates are available through either direct hydrother-
mal synthesis or post-synthetic procedures, including the tin-
containing zeotypes Sn-Beta, Sn-MFI, Sn-USY, and Sn-MWW, as
well as the ordered amorphous Sn-MCM-41 and Sn-SBA-15.[18]
The choice of framework has proven important for carbohy-
drate conversion. For the isomerization of sugars, the large
channels (12-membered ring, with an average diameter of 6.5–
7 ꢂ) in Sn-Beta could easily facilitate the conversion of all mon-
osaccharides (C3 to C6). In contrast, the medium-size channels
(10-membered ring, with an average diameter of 5.5–6 ꢂ) of
the MFI framework were found suitable for the conversion of
triose sugars, while the reactions of pentose and especially
hexose sugars was much slower, owing to the difference in
size of the substrates.[19] Recently, such shape selectivity was
identified in the conversion of tetrose sugars, where different
catalyst structures lead to significantly different product distri-
butions.[20] With this in mind, the objective of the present
study was the selective formation of tetrose sugars from GA in
water under mild reaction conditions, investigating the influ-
ence of catalyst framework on the product selectivity.
For the amorphous materials, low-angle X-ray diffraction
(08–58 2q) showed the characteristic (100), (110), and (200) re-
flections appearing as a consequence of diffraction from the
hexagonally ordered structure (Supporting Information, Fig-
Table 1. Structure and physical properties of the catalysts.[a]
Entry Sample
Si/Sn ratio[a]
Surface area Pore volume
(elemental analysis) [m2 gÀ1
]
[mLgÀ1
]
[b]
[c]
SBET
Smicro Vtotal Vmicro
1
2
3
4
5
Sn-MFI (100, OHÀ) 82
404
273 0.25 0.14
337 0.20 0.13
395 0.28 0.21
Sn-MFI (400, FÀ)
Sn-Beta (150)
383
186
377
492
857
918
Sn-MCM-41 (150) 148
Sn-SBA-15 (200) 271
–
1.09
–
111 1.09 0.06
[a] Determined by ICP analysis. [b] Brunauer–Emmett–Teller (BET) surface
area. [c] Micropore volume calculated using the t-plot method.
&
ChemSusChem 2016, 9, 1 – 9
www.chemsuschem.org
2
ꢁ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Papers
Figure 1. Electron microscope images of a) Sn-MFI (100, OHÀ), b) Sn-MFI (400, FÀ) and c) Sn-Beta (150, FÀ).
ure S2).[18c,22] The measured angles correspond to d100 spacings
of ca. 40 and ca. 100 ꢂ for Sn-MCM-41 and Sn-SBA-15, respec-
tively. The large pore volume and surface area anticipated for
these two types of materials were confirmed by N2 adsorption/
desorption measurements showing total surface areas of ca.
900 m2 gÀ1 and a pore volumes of ca. 1.2 mLgÀ1 for both mate-
rials (Table 1).
Elemental analysis was performed to evaluate the metal con-
tent in the material (see Table 1). For both tin-containing MFI
zeolites, a slightly higher Si/Sn ratio was obtained than expect-
ed based on the nominal values, pointing at some loss of sili-
con during synthesis. Incorporation of tin for Sn-MCM-41 (Si/
Sn=148) was found to be close to the nominal values (150)
(Table 1, entry 4). For the rest of the catalysts (Table 1, entries 3
and 5), the Si/M ratios were slightly higher than the nominal
ratio, meaning that a loss of metal occurred during synthesis.
Condensation of glycolaldehyde with different catalysts
As expected in consecutive reactions, products are not stable
since they can continue reacting. Therefore, selectivity was
found to be highly dependent on the degree of conversion. As
a consequence, catalysts are henceforth compared as a func-
tion of conversion rather than by showing the conventional ki-
netics (yields over time can be found in the Supporting Infor-
mation, Table S1 and Figure S4). This depiction allows the com-
parison of different catalysts under equivalent conditions, re-
gardless of the intrinsic activity of the catalysts due to metal
loading.
Figure 2. Yield of tetrose sugars obtained with different catalysts from glyco-
laldehyde as a function of conversion. The grey dotted line represents a se-
lectivity of 100%. Reaction conditions: 808C, 5 wt% glycolaldehyde (dimer)
in water, 0.075 g of catalyst, 2.5 g of water, reaction times 1–48 h (corre-
sponding data can be found in Table S1).
narily high selectivity (up to 96%) for tetrose sugars. Sn-MFI
(100, OHÀ and 400, FÀ) and Amberlyst A21 are clearly the pre-
ferred catalysts for aldol condensation of GA, leading to high
tetrose yields of up to 74%. In the three cases the tetrose
sugar yield drops drastically above 80% conversion. This trend
was independent of the temperature (40–1008C, Supporting
Information Table S4 and Figure S5). It is, however, evident
from Figure 2 that Sn-MFI (100, OHÀ) maintains selectivity for
tetrose sugar formation to higher conversion values than all
the other selective catalysts tested herein (dashed lines). As
a result, Sn-MFI (100, OHÀ) reaches a yield of 74% of tetrose
sugar after only 30 mins (Figure S5) at 76% conversion before
a drop in tetrose sugar yield is observed. Clearly, with an in-
crease in conversion the probability for an aldol reaction be-
tween two GA molecules diminishes and instead alternative re-
action pathways are preferred (discussed later).
Figure 2 shows the combined yield of tetrose sugars as
a function of conversion for several Lewis-acid and -base cata-
lysts. Apart from the stannosilicates, materials without a defined
pore structure (zirconia, Amberlyst A21) and homogeneous
catalysts such as SnCl4·5H2O were used to probe the effect of
the pore structure. A weak base resin, Amberlyst A21, contain-
ing dimethyl amino functional groups, was used as a reference
for base-catalyzed aldol condensation. For some of the cata-
lysts (Sn-MCM-41, Sn-SBA-15, and ZrO2), a high selectivity (near
100%) towards the desired aldol products is only obtained as
long as conversion is kept low (<20%), quickly decreasing
with increased conversion. The tested catalysts fall into two
different groups in terms of product yield (Figure 2): for one
group (marked in blue) the tetroses are merely intermediates,
whereas the other group (marked in green) provides extraordi-
When comparing the stability of the Sn-MFI catalyst and the
basic resin, significant differences were observed during reuse
ChemSusChem 2016, 9, 1 – 9
www.chemsuschem.org
3
ꢁ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Full Papers
accumulated within the pore system. Thermogravimetric meas-
urements of a spent Sn-MFI catalyst after the washing step re-
vealed a negligible mass loss of 2.9 wt% (Supporting Informa-
tion, Figure S6). In the case of Amberlyst A21, a regeneration
procedure based on the conditions provided by the supplier
(DOW) was attempted. The regeneration involved exposing
the catalyst to different concentrations of NH4OH for 1 h, fol-
lowed by thorough washing to remove excess base adsorbed
to the resin. Nevertheless, the regeneration procedures used
did not improve the catalytic activity in the consecutive cycles
(Supporting Information, Figure S7).
Regarding the formation of other byproducts in the reaction,
Table 2 summarizes the yield of all products detected for the
different catalysts at comparable conversion levels (70Æ6%).
Sn-MFI (100, OHÀ) showed the highest tetrose sugar selectivity
(96%) and carbon balance (99%) followed by Sn-MFI (400, FÀ)
and Amberlyst A21 (entries 1, 2, and 7, respectively). The main
by-products identified in the reaction are formed by further
conversion of the formed tetrose sugars by (i) consecutive con-
densations with GA leading to the formation of hexoses, or (ii)
dehydration and 1,2-hydride shifts to form vinyl glycolic acid
(VGA) and HBL.[7a,23] Stannosilicate materials (Sn-MFI, Sn-Beta,
Sn-MCM-41, and Sn-SBA-15) have been reported to catalyze
the isomerization of sugars, including tetrose sugars.[7a,18c] As
a result, the distribution of tetrose isomers changes over time
towards erythrulose (ERU; Supporting Information, Figure S8a–
d).[24] For Amberlyst A21, isomerization is hardly observed and
the primary tetrose isomer observed is threose (THR) (Fig-
ure S8e). The use of homogeneous tin chloride (Table 2,
entry 6) resulted in some formation of tetrose sugars (16%) at
74% conversion. Several volatile unidentified products with
high boiling points were observed on CG, resulting in a lower
mass balance (45%). Tin chloride could potentially precipitate
under the reaction conditions to yield tin hydroxide species
with a different reactivity compared with the other tested ma-
terials, as observed by Dusselier et al.[7b] In line with previous
reports, no formation of tetrose sugars was observed using
SnO2-Beta zeolites (Sn-Beta prepared using SnO2 as the tin
source). From this it can be concluded that tin incorporated in
the zeolite must be the active sites for the aldol condensation
as neither tin nor the silicious framework show any activity in
the reaction (Supporting Information, Table S3).[21,25]
Figure 3. Yields of tetrose sugars obtained from glycolaldehyde when reus-
ing Sn-MFI (100, OHÀ) or Amberlyst A21. The catalysts were washed repeat-
edly with water and dried over night at 808C between runs. Reaction condi-
tions: 808C, 5 wt% glycolaldehyde in water, substrate-to-catalyst ratio of 1.7,
2 h, 600 rpm stirring. Data and carbon balances can be found in Table S2.
experiments of Amberlyst A21 and Sn-MFI (100, OHÀ) after
washing of the used catalyst with water (Figure 3). The tetrose
yield was found to be almost constant in the case of Sn-MFI
(100, OHÀ), with only a small decrease from 69 to 65% over
several consecutive cycles, demonstrating good catalyst stabili-
ty. For Amberlyst A21, a deactivation of the catalyst was ob-
served, with a decrease of tetrose yield from 64 to 25% over
five cycles. Both catalysts generally require regeneration in-be-
tween uses; for zeolites, calcination at temperatures over
5008C is often performed to remove carbonaceous species
trapped within the pore system and Amberlyst A21 can be
treated in a flow of aqueous base to regenerate basicity. It is
interesting to note that in Figure 3, only thorough washing
and drying (808C) overnight was done to regenerate the cata-
lysts. For Sn-MFI, washing of the zeotype catalyst was found to
be fully sufficient in removing almost all carbonaceous species
Table 2. Comparison of product distribution using different catalysts at comparable conversion levels (70Æ6%).[a]
Entry
Catalyst
t [h]
Conversion [%] Product distribution[b] [%]
THR ERY ERU VGA+HBL hexose sugars
Selectivity (total C4) [%] Carbon balance [%]
1
2
3
4
5
6
7
Sn-MFI (100, OHÀ)
Sn-MFI (400, FÀ)
Sn-MCM-41 (150)
Sn-SBA-15 (200)
Sn-Beta (150)
0.5 76
6.5 74
19
19
11
12
6
15
16
5
8
1
39
30
16
15
16
3
1
2
3
–
16
-
1
1
7
6
5
3
4
96
88
49
55
49
21
83
99
95
76
78
75
45
92
24
66
65
69
74
70
48
3
SnCl4·5H2O[c]
24
2
9
35
4
14
Amberlyst A21
9
–
[a] Reaction conditions: 0.125 mg glycolaldehyde dimer, 0.075 g catalyst, 2.5 g demineralized water, 808C, 600 rpm, 0.25–48 h. [b] Yield (carbon percent) of
threose (THR), erythrose (ERY), erythrulose (ERU), vinyl glycolic acid (VGA), and hexose sugars from glycolaldehyde (GA). [c] 0.025 g of catalyst used.
&
ChemSusChem 2016, 9, 1 – 9
www.chemsuschem.org
4
ꢁ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Papers
Byproduct formation
hexoses are too large to enter the pores.[19b] Additionally, the
formation of large products that might accumulate in the
pores and cavities of the MFI framework can be ruled out, be-
cause negligible amounts of carbon were left after washing in
water (Figure S6). Therefore, we are led to speculate that the
smaller porous system of the MFI zeolite hinders condensation
of substrates larger than glycolaldehyde and that the high se-
lectivity is a consequence of shape selectivity.[26]
From data in Table 2 shown at 65–75% conversion, it is clear
that all the catalysts tested form small amounts of hexose
sugars by additional aldol condensation reactions between te-
trose sugars and GA. When comparing the formation of
hexose sugars as a function of conversion (Figure 4), it can be
seen that all catalysts except Sn-MFI form hexoses already at
low conversion, increasing up to a yield of 8–10% at almost
full conversion. Despite achieving high selectivity towards the
tetrose sugars, Amberlyst A21 (Table 2, entry 7) likewise formed
large quantities of hexoses, especially when approaching full
conversion (up to 10%, Figure 4). Amberlyst A21 showed no
formation of VGA. Sn-MCM-41 and Sn-SBA-15 (Table 2, en-
tries 3–4) behaved similarly, resulting in modest selectivities to-
wards the tetrose sugars with a similar formation of hexose
sugars (up to 6%, Figure 4) and additional formation of low
yields of VGA (up to 4%, Table S1). On the other hand, Sn-MFI
zeolites both maintained a low formation of larger sugars until
a conversion of 75%, after which the yield of hexoses in-
creased to 4%.
To validate the formation of hexoses, we chose to analyze
the reaction mixtures by multidimensional NMR because of its
ability to distinguish similar analytes and isomers, especially
hexoses.[27] Figure 5 shows the anomeric region of a highly re-
1
solved H–13C NMR region for reactions using Sn-MFI (400, FÀ)
(Figure 5a) and Amberlyst A21 (Figure 5b) run to ca. 85% con-
version in D2O. As expected, both catalysts form large quanti-
ties of aldotetroses. ERY and THR are detected in this region as
a- and b-furanose forms in addition to hydrated open-chain
forms (Figure 5). All eight aldohexoses (allose, altrose, glucose,
mannose, gulose, idose, and talose) are detected for Amberlyst
A21 (Figure 5b). In contrast, only minor amounts of mannose,
glucose and gulose are detectable for the Sn-MFI catalyzed re-
The similar product distribution of both Sn-MFI materials
(OHÀ and FÀ), despite having very different crystal sizes
(Figure 1), suggests that the high selectivity towards the te-
trose sugars is not limited by the physical properties of the
materials and is rather a result of the confined active tin sites
within the narrow channels of the zeolite framework.[20] When
comparing Sn-MFI and Sn-Beta, it is clear that larger products
(hexose sugars) are obtained using the 12-membered ring
framework of Sn-Beta (Table 2 entry 5, Figure 4). It has been
shown previously that Sn-MFI shows very poor activity when
hexose sugars are used as substrate, which might indicate that
Figure 5. 1H-13C HSQC spectra of hemiacetal groups in the reaction mixture
produced by the catalytic conversion of GA dimer (808C, D2O) using a) Sn-
MFI (100, OHÀ) and b) Amberlyst A21. Samples are measured at ~85% con-
version. Compounds were identified by reference compounds and by homo-
and heteronuclear NMR assignment spectra recorded on the mixture. Infor-
mation on abbreviations used is available as Supporting Information.
Figure 4. Yield of C6 sugars obtained with different catalysts from pure gly-
colaldehyde as a function of conversion. Reaction conditions: 808C, 5 wt%
glycolaldehyde (dimer) in water, 0.075 g of catalyst, 2.5 g of water, reaction
times 0.25–48 h (corresponding data in Table S1).
ChemSusChem 2016, 9, 1 – 9
www.chemsuschem.org
5
ꢁ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Full Papers
action (Figure 5a). These findings confirm that all stereoiso-
mers of hexose sugars can be formed from consecutive aldol
reactions at high conversion, whilst the confinement in narrow
pores of Sn-MFI restricts hexose formation.
In addition to the formation of larger sugars, the formed
tetroses can undergo dehydration and 1,2-hydride shift to
yield a-hydroxyacid vinyl glycolic acid (VGA) and a-hydroxy-g-
butyrolactone (HBL). This group of chemicals has been ob-
tained from a range of different sugars (C3–C6) with promising
application in the formulation of new materials.[7a,23,28] For ex-
ample, using either condensation or metathesis, VGA has been
shown to form functional polyesters and polyamides.[7b,29] In
a previous study by the group of Sels, several alkyl ester ana-
logues of VGA were formed using SnCl4·5H2O as catalyst in al-
cohol as solvent. Under these conditions, methyl-4-methoxy-2-
hydroxy-3-butanoate (MMHB) was the main product, with
a maximum yield of 55%.[7b] This molecule does not contain
the vinyl group characteristic for VGA. From the data present-
ed in Table 2, it is clear that not all the catalysts tested form
HBL and VGA in significant amounts. The most active catalysts
for HBL and VGA formation are Sn-MFI and Sn-Beta. The results
also suggest that the selectivity towards the different co-prod-
ucts is not only affected by the type of catalyst, but also by
the concentration of GA in the medium. At the beginning of
the reaction, the reaction medium is rich in GA and the possi-
bility for two molecules of GA reacting with each other is high.
With diminishing GA substrate and tetrose sugar formation, re-
actions can progress by formation of hexose sugars or further
conversion of tetroses to VGA becomes more likely. This effect
was also observed when the starting concentration of GA was
changed (Supporting Information, Table S5 and Figure S9). At
low GA concentrations, VGA is formed in higher relative yields
(10% vs 3%, for initial concentrations of GA of 2% and 8%, re-
spectively, after 2 h, using the same amount of catalyst). Bear-
ing this difference in selectivity in mind, we tried to optimize
the reaction conditions for the production of VGA.
Figure 6. Conversion of glycolaldehyde over time using a) Sn-MFI (100, OHÀ)
and b) Sn-Beta (150). Reaction conditions: 808C, 5 wt% glycolaldehyde in
water, 0.075 g of catalyst, 2.5 g of water, 0–48 h. Data can be found in
Table S1.
ucts of 74% were obtained with 97% selectivity after 0.5 h at
808C in water. The formation of hexose sugars by consecutive
aldol condensation reactions is minimized when the medium-
pore-size zeolite MFI is used, compared to catalysts containing
larger pores. We further observe that tin sites in zeolites MFI
and Beta catalyze the additional transformation of tetrose
sugars into vinyl glycolic acid (VGA). The reaction conditions
can be tuned not only to obtain tetrose sugars in high yields
and at high selectivity, but also good yields (up to 44%) of
VGA can be achieved under mild conditions.
Sn-Beta and Sn-MFI were found to form significant amounts
of VGA in water, leading to yields of 39% using Sn-MFI (100,
OHÀ) and 30% using Sn-Beta (150) after 48 h (Figure 6 and
Supporting Information, Figure S10). When the amount of Sn-
MFI catalyst was increased, the dehydration reaction was fa-
vored and the yield of VGA reached 44% (Supporting Informa-
tion, Table S6) at 1008C after 4.5 h. This is the highest VGA
yield reported hitherto in water. In the case of Sn-Beta, similar
conditions lead to the formation of up to 33% HBL and 27%
VGA after 48 h at 808C to give a combined yield of ~60%
(Figure 6).
Experimental Section
Catalyst preparation
MFI zeotype materials used in this study were prepared following
two different synthesis routes initially described by Mal et al.[18a]
For the preparation of Sn-MFI (100, OHÀ), SnCl4·5H2O (98%, Aldrich)
was dissolved in 5 g of demineralized water and added to 15.6 g
of tetraethyl orthosilicate (TEOS, 98%, Aldrich) and stirred for 30
mins. 13.4 g of tetrapropylammonium hydroxide (TPAOH, 40%, Ap-
pliChem) in 13.4 g of demineralized water was then added to the
mixture and stirred for 1 h. Following this, an additional 60 g of
demineralized water was added and the solution was stirred for
another period of ca. 20 h, whereafter the solution was added to
Conclusions
The valorization of biomass-derived glycolaldehyde (GA) using
tin-containing zeotypes in water has potential for establishing
a platform for tetrose-derived products. By exploiting the
shape-selectivity of the zeolite frameworks with different pore
sizes, C4 products can be obtained with high selectivity. The
main products of the reaction of GA in water using medium
pore Sn-MFI as catalyst are C4 sugars. Total yields of C4 prod-
&
ChemSusChem 2016, 9, 1 – 9
www.chemsuschem.org
6
ꢁ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Papers
a Teflon-lined autoclave and synthesized at 1608C for 2 days under
static conditions. For the preparation of Sn-MFI (400, FÀ), 5.35 g of
ammonium fluoride (NH4F, Sigma–Aldrich) was dissolved in 25 g of
demineralized water. To this solution, tin(IV)chloride pentahydrate
(SnCl4·5H2O, 98%, Aldrich) was dissolved in demineralized water
and was added under rapid stirring. Following this, 9.8 g of tetra-
propylammonium bromide (TPABr, Sigma–Aldrich) dissolved in
56 g of demineralized water was added slowly and 8.6 g of fumed
silica was then added to mixture under ample stirring for 3 h. The
resulting gel was transferred to a Teflon-lined autoclave and
heated to 2008C for 6 days.
Sn-Beta (Si/Sn=150) was synthesized by following the alkali-free
synthesis route described in Ref. [21]. In a typical synthesis proce-
dure, 30.6 g of TEOS (98%, Aldrich) was added to 33.1 g of tetrae-
thylammonium hydroxide (TEAOH, Sigma–Aldrich, 35% in water)
under careful stirring. After 1 h, SnCl4·5H2O (98%, Aldrich) dis-
solved in 2 mL of demineralized water was added drop wise and
stirred for 5 h. Finally, 3.1 g hydrofluoric acid (HF, Fluka, 47–51%)
diluted with 1.6 g of demineralized water was added to the gel.
The sample was then homogenized and transferred to a Teflon-
container placed in a stainless steel autoclave and placed at 1408C
for 14 days.
The ordered mesoporous stannosilicate, Sn-MCM-41 (Si/Sn=150),
was prepared according to the route described by Li et al.[22] In
a typical synthesis, 26.4 g of tetraethylammonium silicate (TMAS,
Aldrich, 15–20 wt% in water, ꢀ99.99%) was slowly added to a solu-
tion containing 13.0 g of hexadecyltrimethylammonium bromide
(CTABr, Sigma, ꢀ99.0%) dissolved in 38.0 g of water. This mixture
was then allowed to stir for 1 h. At this point, SnCl4·5H2O (98%, Al-
drich) and hydrochloric acid (HCl, Sigma–Aldrich, min. 37%) in
2.1 g of water was added dropwise to the solution and stirred for
1.5 h. To this solution, 12.2 g of TEOS (98%, Aldrich) was added
and stirred for an additional 3 h and transferred to a Teflon-lined
container placed in a stainless steel autoclave and heated to 1408C
for 15 h.
sorb3 software. Thermogravimetric analysis (TG/DSC) was per-
formed on a TGA/DSC-1 (Mettler-Toledo).
Catalytic activity
Catalyst (0.075 g), glycolaldehyde dimer (SAFC, 0.125 g), and deion-
ized water (2.5 g) were added in a 15 mL vial (ACE pressure tube)
and heated at 808C under vigorous stirring (600 rpm) for between
10 min and 48 h. After finishing, the reaction was quenched in cold
water and filtered samples were retrieved for analysis on an Agi-
lent 1200-series HPLC with a BIORAD Aminex HPX-87H column and
equipped with refractive index (RI) and diode array (DA) detectors
operating at 658C using an eluent of 0.004m H2SO4 solution in
water at 0.6 mLminÀ1
. Unconverted glycolaldehyde and the
formed tetrose sugars (erythrose, threose, and erythrulose) and
larger sugars (hexoses) were quantified using the RI detector. Stan-
dard solutions of the tetrose sugars were obtained from Omicron
Biochemicals, Inc. The formation of larger hexose sugars was quan-
tified using the response factor of the aldohexose; glucose. Vinyl
glycolic acid (VGA) was quantified using the DA detector at a wave-
length of 210 nm with a standard (95% purity) obtained from En-
amine. a-Hydroxy-g-butyrulactone (HBL) was tentatively quantified
on HPLC using the RI detector and a standard (technical grade) ob-
tained from Aldrich. High-field NMR spectroscopy was conducted
on a Bruker Avance II 800 MHz spectrometer equipped with a TCI
Z-gradient CryoProbe and an 18.7 T magnet (Oxford Magnet Tech-
1
nology, Oxford, UK). Sufficiently 13C-resolved H–13C HSQC spectra
were acquired by sampling 1024 complex data points in the direct
(1H) dimension and 1024 complex data points in the indirect (13C)
dimension, during acquisition times of 143 and 93 milliseconds, re-
spectively. All spectra were recorded at 378C. Identification of com-
pounds in the reaction mixture was performed by the use of pure
reference standards for glycolaldehyde, tetroses, and hexoses.
Sn-SBA-15 (Si/Sn=200) was prepared following the synthesis route
described by Ramaswamy et al.[30] Initially, 8.0 g of Pluronic P-123
(PEG-PPG-PEG polymer, Mw =5800 gmolÀ1) was dissolved in 60 g of
demineralized water, followed by the addition of 1.0 g of hydro-
chloric acid (HCl, 37 wt%) in 140 g of demineralized water. The so-
lution was then stirred for 2 h. To the synthesis mixture, 18.0 g of
TEOS (98%, Aldrich) was added followed by SnCl4·5H2O (98%, Al-
drich) dissolved in 2.0 g of demineralized water. The mixture was
then stirred for 24 h at 408C and transferred to a Teflon-lined auto-
clave and heated to 1008C for 24 h.
Acknowledgements
The work of S.T. is funded by the Bio-Value platform (http://bio-
value.dk) under the SPIR initiative by The Danish Council for Stra-
tegic Research and The Danish Council for Technology and Inno-
vation, case number 0603-00522B. S.M. gratefully acknowledges
funding by Grant 2013_01_0709 of the Carlsberg Foundation.
800 MHz NMR spectra were recorded on the spectrometer of the
Danish National Instrument Center for NMR Spectroscopy of Bio-
logical Macromolecules at the Technical University of Denmark.
All prepared catalysts were recovered by filtration, washed with
ample water, and dried overnight at 808C. The materials were final-
ized by calcination, heating the sample to 5508C at 28CminÀ1 in
static air and maintaining this temperature for 6 h.
Keywords: aldol condensation · biomass conversion · tetrose
sugars · tin · zeolites
Characterization techniques
Powder X-ray diffraction (XRD) patterns of the calcined samples
were measured on an X’Pert diffractometer (Philips) using CuKa ra-
diation. The elemental composition of the prepared materials was
measured using inductively coupled plasma–atomic emission spec-
troscopy (ICP–OES) on a PerkinElmer model Optima 3000 (Varian
Vista). Surface area and pore volume measurements were per-
formed using multipoint N2-adsorption/desorption on an Autosorb
automatic surface area and pore size analyzer (Quantachrome In-
struments). The total surface area of the samples was obtained
using the Brunauer–Emmett–Teller (BET) method and the micro-
pore volume was calculated by the t-plot method using the Auto-
[1] a) M. Sasaki, K. Goto, K. Tajima, T. Adschiri, K. Arai, Green Chem. 2002, 4,
285–287; b) D. Mohan, C. U. Pittman, P. H. Steele, Energy Fuels 2006, 20,
848–889; c) R. Vinu, L. J. Broadbelt, Energy Environ. Sci. 2012, 5, 9808–
9826.
[2] W. Mꢃgerlein, J. P. Melder, J. Pastre, J. Eberhardt, T. Krug, M. Kreitsch-
mann, US20120271068A1, 2012.
[3] S. Van de Vyver, Y. Romꢀn-Leshkov, Angew. Chem. Int. Ed. 2015, 54,
12554–12561; Angew. Chem. 2015, 127, 12736–12744.
[4] R. J. Jariwalla, US 20040092549A1, 2001.
[5] R. Ooms, M. Dusselier, J. A. Geboers, B. Op de Beeck, R. Verhaeven, E.
Gobechiya, J. A. Martens, A. Redl, B. F. Sels, Green Chem. 2014, 16, 695–
707.
ChemSusChem 2016, 9, 1 – 9
www.chemsuschem.org
7
ꢁ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Full Papers
[6] a) H. Kishida, F. Jin, X. Yan, T. Moriya, H. Enomoto, Carbohydr. Res. 2006,
341, 2619–2623; b) X. Yan, F. Jin, K. Tohji, A. Kishita, H. Enomoto, AIChE
J. 2010, 56, 2727–2733.
[7] a) M. S. Holm, Y. J. Pagꢀn-Torres, S. Saravanamurugan, A. Riisager, J. A.
Dumesic, E. Taarning, Green Chem. 2012, 14, 702–706; b) M. Dusselier,
P. Van Wouwe, S. De Smet, R. De Clercq, L. Verbelen, P. Van Puyvelde,
F. E. Du Prez, B. F. Sels, ACS Catal. 2013, 3, 1786–1800.
[8] a) S. Yamaguchi, T. Matsuo, K. Motokura, Y. Sakamoto, A. Miyaji, T. Baba,
ChemSusChem 2015, 8, 853–860; b) S. Yamaguchi, T. Matsuo, K. Moto-
kura, A. Miyaji, T. Baba, ChemCatChem 2016, 8, 1386–1391.
[9] a) A. N. Simonov, L. G. Matvienko, O. P. Pestunova, V. N. Parmon, N. A.
Komandrova, V. A. Denisenko, V. E. Vas’kovskii, Kinet. Catal. 2007, 48,
550–555; b) A. N. Simonov, O. P. Pestunova, L. G. Matvienko, V. N. Snyt-
nikov, O. A. Snytnikova, Y. P. Tsentalovich, V. N. Parmon, Adv. Space Res.
2007, 40, 1634–1640; c) I. Delidovich, A. Simonov, O. Pestunova, V.
Parmon, Kinet. Catal. 2009, 50, 297–303.
11736–11739; Angew. Chem. 2012, 124, 11906–11909; e) Q. Guo, F. Fan,
E. A. Pidko, W. N. P. van der Graaff, Z. Feng, C. Li, E. J. M. Hensen, Chem-
SusChem 2013, 6, 1352–1356; f) X. Yang, L. Wu, Z. Wang, J. Bian, T. Lu,
L. Zhou, C. Chen, J. Xu, Catal. Sci. Technol. 2016, 6, 1757–1763.
[19] a) J. Jae, G. A. Tompsett, A. J. Foster, K. D. Hammond, S. M. Auerbach,
R. F. Lobo, G. W. Huber, J. Catal. 2011, 279, 257–268; b) C. M. Lew, N. Ra-
jabbeigi, M. Tsapatsis, Microporous Mesoporous Mater. 2012, 153, 55–58.
[20] R. De Clercq, M. Dusselier, C. Christiaens, J. Dijkmans, R. I. Iacobescu, Y.
Pontikes, B. F. Sels, ACS Catal. 2015, 5, 5803–5811.
[21] S. Tolborg, A. Katerinopoulou, D. D. Falcone, I. Sꢀdaba, C. M. Osmund-
sen, R. J. Davis, E. Taarning, P. Fristrup, M. S. Holm, J. Mater. Chem. A
2014, 2, 20252–20262.
[22] L. Li, C. Stroobants, K. Lin, P. A. Jacobs, B. F. Sels, P. P. Pescarmona, Green
Chem. 2011, 13, 1175–1181.
[23] M. Dusselier, P. Van Wouwe, F. de Clippel, J. Dijkmans, D. W. Gammon,
B. F. Sels, ChemCatChem 2013, 5, 569–575.
[10] T. Matsumoto, S. Inoue, J. Chem. Soc. Perkin Trans. 1 1982, 1975–1979.
[11] a) A. Cꢄrdova, I. Ibrahem, J. Casas, H. Sundꢅn, M. Engqvist, E. Reyes,
Chem. Eur. J. 2005, 11, 4772–4784; b) J. Kofoed, J.-L. Reymond, T.
Darbre, Org. Biomol. Chem. 2005, 3, 1850–1855.
[24] S. Saravanamurugan, A. Riisager, Catal. Sci. Technol. 2014, 4, 3186–
3190.
[25] M. S. Holm, S. Saravanamurugan, E. Taarning, Science 2010, 328, 602–
605.
[12] a) S. Pizzarello, A. L. Weber, Science 2004, 303, 1151; b) A. L. Weber, S.
Pizzarello, Proc. Natl. Acad. Sci. USA 2006, 103, 12713–12717.
[13] a) A. Ricardo, M. A. Carrigan, A. N. Olcott, S. A. Benner, Science 2004,
303, 196; b) H.-J. Kim, A. Ricardo, H. I. Illangkoon, M. J. Kim, M. A. Carri-
gan, F. Frye, S. A. Benner, J. Am. Chem. Soc. 2011, 133, 9457–9468.
[14] J. B. Lambert, S. A. Gurusamy-Thangavelu, K. Ma, Science 2010, 327,
984–986.
[26] P. B. Weisz, Pure Appl. Chem. 1980, 52, 2091–2103.
[27] a) M. Bøjstrup, B. O. Petersen, S. R. Beeren, O. Hindsgaul, S. Meier, Anal.
Chem. 2013, 85, 8802–8808; b) B. O. Petersen, O. Hindsgaul, S. Meier,
Analyst 2014, 139, 401–406.
[28] S. Tolborg, S. Meier, I. Sadaba, S. G. Elliot, S. K. Kristensen, S. Saravana-
murugan, A. Riisager, P. Fristrup, T. Skrydstrup, E. Taarning, Green Chem.
2016, 18, 3360–3369.
[15] S. M. Csicsery, Zeolites 1984, 4, 202–213.
[29] a) M. Dusselier, P. Van Wouwe, A. Dewaele, P. A. Jacobs, B. F. Sels, Science
2015, 349, 78–80; b) A. Dewaele, L. Meerten, L. Verbelen, S. Eyley, W.
Thielemans, P. Van Puyvelde, M. Dusselier, B. Sels, ACS Sustain. Chem.
Eng. 2016, in press, DOI: 10.1021/acssuschemeng.6b00807.
[30] V. Ramaswamy, P. Shah, K. Lazar, A. V. Ramaswamy, Catal. Surv. Asia
2008, 12, 283–309.
[16] a) S. Trigerman, E. Biron, A. H. Weiss, React. Kinet. Catal. Lett. 1977, 6,
269–274; b) H. Tajima, K. Tabata, T. Niitsu, H. Inoue, J. Chem. Eng. Jpn.
2002, 35, 564–568.
[17] a) J. D. Lewis, S. Van de Vyver, Y. Romꢀn-Leshkov, Angew. Chem. Int. Ed.
2015, 54, 9835–9838; Angew. Chem. 2015, 127, 9973–9976; b) Y. Wang,
J. D. Lewis, Y. Romꢀn-Leshkov, ACS Catal. 2016, 6, 2739–2744.
[18] a) N. K. Mal, V. Ramaswamy, P. R. Rajamohanan, A. V. Ramaswamy, Micro-
porous Mater. 1997, 12, 331–340; b) A. Corma, L. T. Nemeth, M. Renz, S.
Valencia, Nature 2001, 412, 423–425; c) C. M. Osmundsen, M. S. Holm,
S. Dahl, E. Taarning, Proc. R. Soc. London Ser. A 2012, 468, 2000–2016;
d) C. Hammond, S. Conrad, I. Hermans, Angew. Chem. Int. Ed. 2012, 51,
Received: June 7, 2016
Revised: July 11, 2016
Published online on && &&, 0000
&
ChemSusChem 2016, 9, 1 – 9
www.chemsuschem.org
8
ꢁ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

FULL PAPERS
Zeo moduz? A highly selective self-con-
densation of glycolaldehyde to different
C4 molecules is achieved with Lewis-
acidic stannosilicate catalysts in water at
moderate temperatures. Sn-MFI (green)
catalyzes selectively the condensation
of glycolaldehyde (black) to C4 sugars.
Catalysts with larger pores, such as Sn-
MCM-41 (blue) and Sn-Beta (red) favor
the formation of hexoses. Under opti-
mized reaction conditions, other inter-
esting C4-derived products are formed.
S. Tolborg, S. Meier, S. Saravanamurugan,
P. Fristrup, E. Taarning, I. Sꢀdaba*
&& – &&
Shape-selective Valorization of
Biomass-derived Glycolaldehyde using
Tin-containing Zeolites
ChemSusChem 2016, 9, 1 – 9
www.chemsuschem.org
9
ꢁ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ