Quantitative NMR Approach to Optimize the Formation of Chemical Building Blocks from Abundant Carbohydrates

Article abstract of DOI:10.1002/cssc.201700587

The future role of biomass-derived chemicals relies on the formation of diverse functional monomers in high yields from carbohydrates. Recently, it has become clear that a series of α-hydroxy acids, esters, and lactones can be formed from carbohydrates in alcohol and water solvents using tin-containing catalysts such as Sn-Beta. These compounds are potential building blocks for polyesters bearing additional olefin and alcohol functionalities. An NMR approach was used to identify, quantify, and optimize the formation of these building blocks in the Sn-Beta-catalyzed transformation of abundant carbohydrates. Record yields of the target molecules can be achieved by obstructing competing reactions through solvent selection.

Full text of DOI:10.1002/cssc.201700587

Accepted Article
Title: Quantitative NMR Approach to Optimize the Formation of
Chemical Building Blocks from Abundant Carbohydrates
Authors: Sebastian Meier, Samuel Elliot, Søren Tolborg, Irantzu
Sadaba Zubiri, and Esben Taarning
This manuscript has been accepted after peer review and appears as an
Accepted Article online prior to editing, proofing, and formal publication
of the final Version of Record (VoR). This work is currently citable by
using the Digital Object Identifier (DOI) given below. The VoR will be
published online in Early View as soon as possible and may be different
to this Accepted Article as a result of editing. Readers should obtain
the VoR from the journal website shown below when it is published
to ensure accuracy of information. The authors are responsible for the
content of this Accepted Article.
To be cited as: ChemSusChem 10.1002/cssc.201700587
Link to VoR: http://dx.doi.org/10.1002/cssc.201700587
A Journal of
www.chemsuschem.org

10.1002/cssc.201700587
ChemSusChem
COMMUNICATION
Quantitative NMR Approach to Optimize the Formation of
Chemical Building Blocks from Abundant Carbohydrates
Samuel G. Elliot,[a] Søren Tolborg,[b] Irantzu Sádaba,[b] Esben Taarning,[b] and Sebastian Meier*[a]
The future role of biomass-derived chemicals relies on the formation
of diverse functional monomers in high yields from carbohydrates.
Recently, it has become clear that a series of α-hydroxy acids,
esters and lactones can be formed from carbohydrates in alcohols
and water using tin-containing catalysts such as Sn-Beta. These
compounds are potential building blocks for polyesters with
additional olefin and alcohol functionalities. We employ an NMR
approach to identify, quantify and optimize the formation these
building blocks in the chemocatalytic transformation of abundant
carbohydrates by Sn-Beta. Record yields of the target molecules can
be achieved by obstructing competing reactions through solvent
choice.
The need to establish more sustainable ways to obtain the
chemicals needed by society for the production of food,
materials, fuels and energy is widely recognized. The current
chemical industry is based on the availability of a small number
of petroleum-derived building blocks. Biomass can be utilized to
provide access to both existing and new types of building blocks
and research into this area has steeply increased during the last
Scheme 1. Overview of major pathways in the catalytic conversion of xylose
by Sn-Beta and the major products identified in the reaction mixtures at high
decade. It is important for these chemicals to be accessible at
low cost and at the same time to have useful properties, in order
to be utilized commercially. Therefore, direct conversion of
sugars to a target chemical by heterogeneous catalysis, offers
the best chance of lowering the process costs.[1-4] A plethora of
reaction products have been identified in the acid-catalyzed
conversion of C5 and C6 carbohydrates.[5] Amidst them acyclic
α-hydroxy esters and acids have recently emerged as an
attractive group of bio-monomers (Scheme 1).
temperature (>100 °C). The products are shown for reactions in alcohol (R =
alkyl) or water (R = H) and are grouped (from left to right) as 3,4-dideoxy
esters, 3-deoxy esters, retro-aldol products and furanics.
Table 1. Conversion of C4, C5 and C6 carbohydrates by Sn-Beta to
homologous 3,4-dideoxy esters with declining yield.
Carbon
Backbone
Compound
#
Substrate
Structure
Yield
For the use in biopolymer applications, the production,
polymerization and upgrading of the C3 building block lactate (7)
and the C4 building block vinyl glycolate (8) have previously
been demonstrated.[3,6-11] Recently, the production and
polymerization of the C5 building block 2,5-dihydroxy-3-
pentenoic acid methyl ester (Me-1) as well as the formation of
the C6 building block trans-2,5,6-trihydroxy-3-hexenoic acid
methyl ester (Me-12) have been described (Table 1).[12-14]
C4
C5
C6
Erythrulose
Me-8
Me-1
56%[7]
Xylose
33%[13]
16%[14]
Glucose
Me-12
The conversion of C5 and C6 carbohydrates to the acyclic
unsaturated α-hydroxy compounds 1 and 12 has been achieved
using Sn-Beta, in methanol[13-14] or water[12] at temperatures
above 100 °C in the absence of further additives. A Sn-Beta
zeolite synthesized under hydrothermal conditions with
hydrofluoric acid as the mineralizing agent has so far proven
most efficient for this reaction.[12-14] C5 and C6 carbohydrates
yield around 30% for C5 carbohydrate substrate (Me-1) and
around 15% for C6 carbohydrate substrate (Me-12), respectively,
while the yields for the formation of the C4 analogue vinyl
glycolate methyl ester (Me-8) can reach above 50% (Table
1).[7,13-14] Addition of alkali salts has been shown to reduce those
yields and to strongly enhance the formation of smaller
compounds such as methyl lactate, instead.[8]
[a]
[b]
S. G. Elliot, Dr. S. Meier
Department of Chemistry
Technical University of Denmark
Kemitorvet, Bygning 207, 2800 Kgs. Lyngby, Denmark
E-mail: semei@kemi.dtu.dk
Dr. S. Tolborg, Dr. I.Sádaba, Dr. E. Taarning,
Haldor Topsøe A/S, Haldor Topsøes Alle 1, 2800 Kgs. Lyngby,
Denmark.
Supporting information for this article is given via a link at the end of
the document.
This article is protected by copyright. All rights reserved.

10.1002/cssc.201700587
ChemSusChem
COMMUNICATION
The detection and optimization of carbohydrate conversion
to new chemicals requires consideration of the optimum
analytical approaches employed. Thus, methods that rely on the
use of reference standards for identification and quantification,
including liquid and gas chromatography, may be challenging to
apply in work streams towards new chemicals. In cases where
commercial standards are unavailable, either production of
standards or calculations to estimate response factors are
necessary.[10] In contrast, methodologies that combine detailed
structural information and the possibility of obtaining an accurate
quantitative signal may be more valuable in the push towards
bio-based economies.
for compound identification. Band-selective 13C excitation[19] and
optimized decoupling sequences[20-21] were used to suppress
non-informative signals and artefacts and to obtain higher quality
1H-13C spectra. NMR spectroscopy has notorious shortcomings
in the detection of heteroatoms beyond protons and carbons, but
this problem was negligible for reactions involving carbohydrate
fragmentation or dehydration, as oxygen positions could be
inferred from 13C chemical shifts. At the same time, the detection
of discrete signals for individual atomic positions by NMR
spectroscopy allows the distinction of isomers. Such distinction
of isomers is crucial, as several potential products in
carbohydrate dehydration cannot be distinguished based on
their mass alone.
Major products that were identified in reaction mixtures
produced from xylose at 160 ºC using Sn-beta zeolite are shown
in Scheme 1. These compounds were identified through de novo
structure determination and 1H/13C chemical shift assignments in
unpurified reaction mixtures in six different protic solvents (water,
methanol, ethanol, n-propanol, iso-propanol and n-butanol). This
approach was aided by the use of high-field NMR
instrumentation (18.7 Tesla magnets) equipped with
cryogenically cooled detection electronics to reduce electronic
noise approximately 3-fold. The identified reaction products
derive from pathways including C-C bond breakage in retro-aldol
reactions to yield C2, C3 and C4 fragments, which may
In the current study, we combine in-situ NMR spectroscopy
for the identification of reaction products and an accurate
quantitative NMR (qNMR) methodology to assess solvent effects
in the Sn-Beta catalyzed conversion of abundant carbohydrates.
The methodology operates on
a timescale comparable to
commonly used chromatography methods but avoids
instrumental response factors altogether. Identification of new
chemicals in situ makes their purification and characterization
obsolete. Water and a series of short-chain alcohols are used as
solvents that provide sufficient substrate solubility for the
carbohydrates. The change of solvents from methanol and
water to longer-chain alcohols is motivated by the varying
physicochemical solvent properties and by the stoichiometric
participation of the solvent as a nucleophile in the reaction. Thus, subsequently undergo dehydration to various α-hydroxy esters
the choice of solvent will affect the formation of alkyl-glycoside
and acetal type intermediates during the reaction, modulate the
microenvironment and Lewis acid properties of tin active sites in
the stannosilicates and alter molecular dynamics and energetics
in the reaction path.[9-10] Hence, solvent variation is a means of
optimizing operation conditions towards increased profitability of
bioprocesses.
For the analysis of solvent effects, we use optimized
spectra on complex reaction mixtures without prior purification or
sample pre-treatment in protonated organic solvents. The
approach does not rely on the availability of purified reference
compounds. Changes in product structure due to reaction with
the solvent do not critically complicate product identification and
quantification for the 11 main reaction products.
(7-8). Alternatively, direct dehydration of the C5 compound to α-
hydroxy, 3-deoxy or 3,4-dideoxy esters (1-6) or triple
dehydration to furanics (9-11) can occur. The most interesting of
these products may be the trans-2,5-dihydroxy-3-pentenoic acid
alkyl ester (1). We recently showed that this prospective
chemical building block could be co-polymerized enzymatically
in a selective 1,5 polymerization reaction with ethyl 6-hydroxy-
hexanoate, yielding polymers that could be specifically
functionalized at the secondary alcohol group or at the olefinic
bond of the monomer.[13] Thus, 2,5-dihydroxy-3-pentenoic acid
alkyl esters could form the platform for a vast variety of
functional materials derived from C5 carbohydrates.
Standard reaction conditions were defined based on
optimizations of trans-2,5-dihydroxy-3-pentenoic acid methyl
ester (Me-1) formation in methanol.[13] In these studies, a Sn-
Beta catalyst was found to be most effective for the formation of
Me-1, and therefore such a catalyst was also employed in this
study of solvent effect.
Results and Discussion
One-dimensional 13C NMR spectra of reaction mixtures
carbonyl regions obtained in various solvents are shown in
Figure 1, assigned compounds corresponding to the spectral
signals are shown in Scheme 1. In these reaction mixtures,
careful inspection allowed the identification of the minor cis-2,5-
dihydroxy-3-pentenoic acid alkyl ester (2) in addition to the
dominant trans-form (1). The 1H-13C spectral region showing
both molecules and their chemical shift assignments in methanol
Detection of products in reaction mixtures without relying
on reference compounds
NMR spectroscopy is widely used for chemical structure
elucidation and mixture analysis. The use of NMR spectroscopy
for component identification and quantification is particularly well
developed for biological samples, mostly biofluids, extracts and
foods, but has also gained popularity within biomass conversion,
for instance in the study of carbohydrate isomerization
reactions[15-17] or of lignin structure and depolymerization
reactions.[18] Here, we use a suite of homo- and heteronuclear
assignment spectra for the detection and identification of
carbohydrate degradation products in situ. Specifically, DQF-
COSY, TOCSY, 1H-13C HMBC, conventional and edited 1H-13C
2
is displayed in Figure 2. Long-range correlations through JHC
couplings across the double bond are indicated by white lines
and serve to identify the connected signals as parts of the same
NMR spin system. The cis- and trans-configurations were
3
identified by their characteristic JHH scalar couplings across the
1
HSQC as well as H-13C HSQC-TOCSY spectra were employed
olefinic bond and by the characteristic signal shifts to lower
frequencies for carbons adjacent to the cis-double bond. This
This article is protected by copyright. All rights reserved.

10.1002/cssc.201700587
ChemSusChem
COMMUNICATION
cis-isomer was identified as the minor isomer in all solvents.
Together with these products, 6 additional 3-deoxy ester/acid
compounds were identified in the carbonyl region shown in
Figure 1. These signals cluster within different spectra regions in
accordance with their functionality: olefinic esters (1-3, 8) at 172-
174 ppm, alkyl esters (4, 5, 7) at 174-176 ppm and lactones (6)
above 177 ppm. Further inspection of the 13C NMR spectra of
Figure 1 indicates that rather dramatic compositional changes
result from changing the solvent from water to alcohol.
Furthermore, it was possible to distinguish the diastereomers of
compounds 4-6 signified as by an apostrophe.
Quantification of bio-based chemicals in reaction mixtures
Reliable quantifications of the compound series shown in
Scheme 1 were subsequently pursued by quantitative NMR
(qNMR) spectroscopy. As qNMR can operate at the highest
level of quantitative measurement[22-23] the method is free of
empirical factors in the uncertainty analysis of the experiment
and signal areas are directly proportional to the concentration of
atoms contributing to the signal.
Cryogenically cooled detection electronics and high-field
instrumentation enabled absolute quantifications of reaction
mixtures by quantitative 1D 13C NMR spectroscopy at natural 13
C
isotope abundance (1.11%). Experiments that only use inverse
gated decoupling were used to avoid enhancement of the 13C
signals by the nuclear Overhauser effect.[24] Crude reaction
mixtures, subjected to quantitative 13C NMR spectroscopy,
contained dimethyl sulfoxide as an internal standard for absolute
quantification. Dilute dimethyl sulfoxide showed good recovery
after reactions without detectable degradation, while being
miscible with the solvents tested in this study and did not result
1
in H or 13C NMR signal overlap with analyte signals. Therefore,
dimethyl sulfoxide was chosen and deemed preferable to
compounds like mesitylene, dioxane, glycerol or other alditols
tested for the spectroscopic characterization of reaction mixtures
formed in catalytic carbohydrate conversion.
Figure 1. 13C NMR spectra after converting 360 mg D-xylose with 180 mg Sn-
Beta in 5 mL solvent. The carboxyl spectral region is shown and signals are
assigned to the molecules shown in Scheme 1; an apostrophe denotes the
diastereomer of a compound. Chemical shifts are relative to deuterated
methanol (added to 10% v/v) set to 47.85 ppm for all solvents.
Quantitative 1D 13C NMR rather than the more commonly
1
used H NMR spectroscopy was employed for quantification, as
it provides excellent signal resolution and sharp non-split signals.
Protonated carbon positions were generally used for
quantification due to their shorter T1-NMR relaxation times,
yielding quantitative signal for inter-scan relaxation delays on the
sub-minute timescale. As an alternative, improved 13C NMR
spectra could be obtained upon addition of 1 mM GdCl3 as a
A
relaxation agent to the NMR sample, shortening the carbonyl 13
C
T1 relaxation time at room temperature and 18.7 T magnetic field
to the ~1 second time scale. The reduced 13C T1 relaxation time
permits the accurate measurement of 13C NMR signal areas also
for quaternary carbons without the need for long inter-scan
recycle delays. The signal-to-noise ratio obtained within one
hour of experiment time per sample translated to an estimated
error of determination of 0.2-1.0% for product yields. Such small
experimental uncertainty was validated by performing the
analyses for reaction mixtures in all six different solvents in
replicate, yielding near-identical 13C NMR spectra for repetitions
in each solvent (Figure S1).
B
5.50
O
5.83
4.10
127.28
5.80
173.43
O
61.30 126.81
173.32
133.68
OMe
3.74
HO
OMe
1
2
OH
51.45
6.01
3.75
HO
Solvent effect on Sn-Beta catalyzed reaction
4.98
OH
51.45
132.36
4.34/4.20
57.78
67.35
4.71
70.85
Quantifications of the eight major compounds in non-purified
reaction mixtures in different solvents are displayed in Figure 3.
Various trends become evident upon variation of the solvent.
Upon increasing the alkyl chain of the alcohol solvent, the
amount of retro-aldol products (7-8) decreases, consistent with
previously reported observations.[7] Simultaneously, an
increased yield of 3,4-dideoxy esters (1-3) and related 3-deoxy
esters (4-6) is observed. Notably, all longer chain alcohols
3
3
J
H3H4=16 Hz
JH3H4=11 Hz
Figure 2. (A) 1H-13C HSQC spectral region displaying the olefinic signals of
compound 1 (trans-isomer) and 2 (cis-isomer) in a reaction mixture. JCH
correlations across the double bond are indicated by thin white lines. (B)
Chemical shift assignments of compound 1 (trans-isomer) and compound 2
(cis-isomer).
2
This article is protected by copyright. All rights reserved.

10.1002/cssc.201700587
ChemSusChem
COMMUNICATION
tested gave higher yields of 3,4-dideoxy esters than methanol
did, even at conditions that were specifically optimized for
reactions in methanol. The formation of 3,4-dideoxy esters was
highest in ethanol with yields at around 42% (38% trans (1) and
4% cis (2-3) form) and thus 1.2-fold higher than in methanol.
Consistently increased yields were also found for the lactone (6),
yielding 15% in ethanol as compared to 8% in methanol. Overall,
the 3-deoxy and 3,4-dideoxy esters (1-6) account for 66% of the
carbon balance in ethanol and in iso-propanol. Quantifications of
these related compounds are summarized in Table 2. Furfural
derivatization to alkyl acetals predominates for methanol and is
and acetalization. Due to this catalytic activity of Brønsted acids,
their continuous formation will impact the catalytic system and
alter the product distribution, thus rationalizing the principally
different product composition in water and alcohols. Especially
where the 3,4-dideoxy compounds are targeted, water is overall
less useful as the solvent compared to the alcohols.
We finally note that reaction selectivity has been
partially attributed to steric hindrance at the Sn active site in
zeolite pores.[17] In the current solvent study, the use of less
nucleophilic and more bulky solvent increases the selectivity
towards the formation of C5 lactones (compound 6). This
less pronounced in larger alcohols and especially in iso-propanol. change in selectivity is consistent with a favored intramolecular
reaction rather than a reaction with the solvent for bulkier
alcohols in Sn-Beta zeolite pores.
Temperature effect in ethanol
As improved yields of 3,4-dideoxy esters were achieved in other
alcohols than in methanol, even under conditions that had been
optimized for methanol, we evaluated the prospect for further
improvements of the reaction using ethanol as the solvent. In the
range of 140-160 °C, the product selectivity was reasonably
constant with similar yields of 3,4-dideoxy esters (42%, see
Table S2). Overall, we find the formation of the 3-deoxy and 3,4-
dideoxy dehydration products (1-6) to be only weakly sensitive
to temperature variation between 120 and 180 ºC. Nevertheless,
exploring solvent and temperature effects resulted in
a
combined yield above 40% for 3,4-dideoxy esters, an almost
10% increase relative to previously achieved yields in methanol
(33%, Table 1).
Temperature-dependent differences are detected in the
formation of formation of the cyclic γ-lactone compound 6
relative to the open-chain 3-deoxy compounds 4 and 5. Lower
temperatures favor the formation of compound 6 relative to the
open-chain compounds 4 and 5. This observation is consistent
with an increasing fraction of acyclic forms due to endothermic γ-
lactone solvolysis to compound 5.[25] Tabulated yields of the
temperature study are collected in the SI (Table S2).
Figure 3. Product distribution of reactions in alcohols or water. The reactions
were conducted with 180 mg Sn-Beta, 360 mg D-xylose, 5 mL solvent and 50
mg dimethyl sulfoxide as internal standard, and run for 2 hours at 160 °C with
600 rpm stirring. An experimental uncertainty within ±1% was achieved in all
cases. Legend numbers reference to the compounds in Scheme 1.
In all solvents, 3,4-dideoxy esters are primarily formed as the
trans compound (1), with yields that are also higher than both
the cis 3,4-dideoxy form (2-3) and the potentially related furfural
forms (9-11) combined.
Table 2. Yield of 3,4-dideoxy compounds (1-3) and 3-deoxy compounds (4-6)
from xylose in various solvents.
n-
n-
iso-
Compound Water Methanol Ethanol
Propanol Butanol Propanol
1-3
4-6
4%
33%
14%
42%
24%
40%
14%
35%
21%
38%
28%
14%
17%
47%
66%
54%
56%
66%
Sum
Reactions conducted in water exhibited a significantly
altered composition than reactions conducted in alcohols, also
exhibiting a greater tendency towards humin formation and
catalyst discoloration. Thus, 3,4-dideoxy compounds are formed
at lower levels than retro-aldol products, furanics and 3-deoxy
compounds in water. A principal difference in the use of water
and alcohol arises in the formation of free Brønsted acids as
products in reactions conducted in water. Brønsted acids are
known to catalyze a multitude of reactions including dehydration
Figure 4. Product distribution of reactions between 120 °C and 180 °C using
ethanol as the solvent. The reactions were conducted using 180 mg Sn-Beta,
360 mg D-xylose, 5 mL ethanol and 50 mg dimethyl sulfoxide as internal
standard, and run for 2 hours at the selected temperature. An uncertainty
within ±1% was achieved in all cases. Legend numbers reference to the
compounds in Scheme 1.
This article is protected by copyright. All rights reserved.

10.1002/cssc.201700587
ChemSusChem
COMMUNICATION
Time-resolved reaction analysis
mechanistic effects of Brønsted acid formation. The formation of
alkyl lactates is largely suppressed in longer alcohols, consistent
with previous studies at conditions that were optimized for
lactate formation. Here, maximum alkyl lactate yields had been
reported to drop from 65% in methanol to approximately 35% in
ethanol and 25% in isopropanol.[7] Overall, the beneficial effect
of ethanol relative to methanol in the formation 1-3 can be
ascribed to the less nucleophilic and larger ethanol disfavoring
side reactions.
Additional insight into solvent effects on the acyclic reaction of
xylose to potential polyester building blocks was sought by
tracking the reaction progress in water, methanol and ethanol
over time. To this end, reactions were conducted under
microwave heating for varying time and were subsequently
analyzed by quantitative 13C NMR spectroscopy (full data
provided in Table S3). The experiments show continued
formation of products even after full xylose conversion. This
observation is attributed to the accumulation of glycoside
intermediates,[14] some of which subsequently can be converted
to the full range of products.
The formation of compound 1, the main product, in
methanol and ethanol, was tracked in an initial-rate experiment
at 160 ºC under quasi-steady state conditions as shown in
Figure 5. Due to the initial surplus of substrate, pseudo-first
order kinetics were obtained, resulting in a linear accumulation
of product during the short initial time period shown in Figure 5.
Although compound 1 formed with higher selectivity in ethanol
(39%) than in methanol (31%) and water (approximately 5%) in
the kinetic experiments, the initial rate of formation is lower in
ethanol than in methanol. This finding on the formation of
compound 1 is paralleled by a higher rate of conversion of
20
Methanol, v=0.85 mM/min
Ethanol, v=0.57 mM/min
Water, v=0.20 mM/min
10
0
0
20
40
Time/s
60
80
100
Figure 5. Initial rate experiments for the formation of product 1 from 360 mg
D-xylose 180 mg by Sn-Beta at 160ºC in 5 mL methanol, water and ethanol.
The initial rate is lower in ethanol than in methanol, albeit a higher final
concentration of 1 is reached in ethanol.
xylose in methanol than in ethanol and
conversion in water (Table S3).
a lower rate of
Kinetic profiles underline the complexity of the reaction
pathway and hint at a combination of solvent effects. Solvent
molecules compete with the substrate for binding to tin active
sites. A weaker binding to these sites is expected for alcohols
relative to water, leading to a better availability of active sites for
the substrate in alcohols than in water.[26-27] This rationalization
may explain the lower rate of formation of compound 1 and
concomitantly lower rate of xylose conversion observed in water.
In the Sn-Beta catalyzed conversion of xylose in alcohol, the
alcohol enters the catalytic cycle as a nucleophilic reactant. The
lower rate of forming compound 1 in ethanol than in methanol is
thus consistent with previous reports for C4 carbohydrate
products suggesting that lower nucleophilicity and steric
limitations may alter product selectivity towards the 3,4-dideoxy
compound.[9]
In methanol and ethanol the formation of 1-6 is continuous
during the reaction course (Figure S2). In contrast, the yield of
compound 1 in water shows a maximum and subsequent
decrease during the reaction, while maintaining a consistent
collective increase of 3-deoxy and 3,4-dideoxy compounds (1-6).
This finding indicates that some rehydration of compound 1 can
occur in water, especially in the presence of accumulating
Brønsted acidity. Formation of compound H-7 occurs more
rapidly in water than formation of compounds Me-7 and Et-7 in
alcohols (Figure S3). Accumulation of compound 7 in water is
followed by an increase in formation of compound 7, indicative
of an alteration of the mechanism due to an increase in
Brønsted acidity. Final levels of furanics decrease slightly from
water to methanol and ethanol (Figure 3). Lower carbon balance
in water and the discoloring of the reaction mixture also indicate
the formation of undetetcted polymeric species (humins). The
distinct selectivity of Sn-Beta-catalyzed xylose conversion in
water relative to alcohols can be ascribed to lower reactivity at
the Sn active sites due to competing solvent absorption and to
Improved yield of chemical building blocks from glucose
The abundance of glucose in biomass is even greater than that
of xylose and glucose is the most abundant carbohydrate in
nature. Glucose is available from starch and cellulose, which are
both homopolymers composed of glucopyranose sub-units.
Hexoses are able to form compound 12 (trans-2,5,6-trihydroxy-
3-hexenoic acid methyl ester, Figure 6A), the C6 analogue to
compound 1, although in lower yields. In order to substantiate
that the solvent effect discussed herein is applicable to the
formation of chemical building blocks from C6 carbohydrates, as
well, reactions using glucose as the substrate were performed.
The beneficial effect of ethanol relative to methanol in the
formation of compound 12 is also valid for glucose, as can be
seen from the NMR spectra of the olefinic region (Figure 6B).
The solvent change from methanol to ethanol increases the
formation of compound 12 approximately 1.2-fold (from 13.8% to
16.0% under the standard reaction conditions used herein). The
formation of the hexono-lactone equivalent of compound 6 also
increases, from 16.6% to 19.4%, while the formation of alkyl
lactate 7 drops from 31.5% to 12.6%. All of these effects closely
follow trends observed for the conversion of xylose and are
consistent with previous studies using homogeneous and
heterogeneous tin catalysts on the less abundant acyclic C4
carbohydrate erythrulose.[9-10] Thus, a shift towards improved
selectivity for the formation of commercially interesting 3,4-
deoxy esters at the expense of retro-aldol cleavage can be
achieved for a series of carbohydrates using Sn-Beta catalyst in
longer alcohols than methanol.
This article is protected by copyright. All rights reserved.

10.1002/cssc.201700587
ChemSusChem
COMMUNICATION
O
OH
A
Experimental Section
12
Reactions were conducted with a Biotage Initiator+ microwave reactor in
5 mL glass reaction vials. Reactions were typically conducted with 180
mg Sn-Beta (Si/Sn = 200, hydrothermally synthesized), 360 mg D-xylose,
5 mL solvent and 50 mg dimethyl sulfoxide as internal standard.
RO
OH
OH
B
Ethanol
area 1.000
DMSO
12
12
Sn-Beta was produced according to the synthesis procedure described in
the ESI, based on a modification of the procedure described by Valencia
et al,[30-31] yielding the Sn-Beta catalyst with a Si/Sn ratio of 200. The
product was confirmed by ICP (Si/Sn = 200), XRD (*BEA, Figure S4) and
N2-adsorption (see ESI for details).
0.254 0.004
0.253 0.004
Methanol
12
0.206 0.002
12
0.206 0.002
For discerning solvent effects on product composition and pathway
usage, each sample was heated to 160 °C for 2 hours. 500 µL aliquots
were taken from each sample, 50 µL of methanol-d4 (Sigma-Aldrich) was
added and the mixture transferred to a 5 mm NMR tube for immediate
analysis. In kinetic experiments, samples in water, methanol or ethanol
were prepared in the composition detailed above and heated to 160 °C
for a variable duration (1 s, 10 s, 1 min, 10 min, 1 h, 2 h). Samples were
rapidly cooled with compressed air and 500 µL aliquots were taken from
each sample, mixed with 50 µL of methanol-d4 and transferred to a 5 mm
NMR tube.
area 1.000
38
DMSO
130
125
40
36
13
Chemical Shift ( C, ppm)
Figure 6. (A) Structure of the C6 analogue of compound 1, trans-2,5,6-
trihydroxy-3-hexenoic acid methyl ester (12). (B) Spectral comparison of the
areas of the olefinic carbon shifts of compound 12 in ethanol and methanol
reaction solvents. Duplicate samples are shown, including errors of
determination of signal areas relative to DMSO as an internal quantification
standard.
All NMR spectra were acquired on an 800 MHz Bruker Avance III NMR
spectrometer equipped with a TCI CryoProbe and a SampleJet sample
changer. 1D 13C NMR spectra were acquired by sampling 64k complex
data points during and acquisition time of 1.36 seconds, and using an
inter-scan recycle delay of 45 seconds. A pulse sequence with 1H
irradiation only during the signal acquisition was used to minimize
distortions of signal integrals by the nuclear Overhauser effect.
Protonated 13C carbons were used for quantification and several carbon
sites per molecule were used to improve the statistics of the signal area,
using integrations in Bruker Topspin 3.5 pl5 (see e.g. Figure 8). Standard
DQF-COSY, TOCSY, 1H-13C HMBC, standard and edited 1H-13C HSQC
as well as 1H-13C HSQC-TOCSY spectra were employed for compound
identification in samples after evaporating the solvent overnight in a fume
hood, prior to re-dissolution in deuterated solvents. Figure S6
demonstrates the accuracy of this method by comparison of a reaction
sample with readily available standard compounds. Standard 1H-13C and
edited HSQC as well as 1H-13C HSQC-TOCSY were used in protonated
solvents to validate the assignments. All spectra were processed with
ample zero filling in all spectral dimensions using Bruker Topspin 3.5 pl5.
Conclusion
In conclusion, we employ quantitative 13C NMR for the detection
and accurate quantification of previously unknown products in
the conversion of abundant C5 and C6 carbohydrates by Sn-
Beta, without using reference standard compounds.
Identification and quantification are feasible in different solvents,
thus permitting the elucidation of solvent effects on reaction
selectivity. The exchanging of methanol for longer alcohols
results in increased yields of C5 3,4-dideoxy esters, from 33% in
methanol to 42% in ethanol. This increase in yield can be
ascribed to the less nucleophilic and larger ethanol disfavoring
competing reactions. The improved yield of C5 3,4-dideoxy
esters in ethanol relative to methanol and water correlates with a
lower rate of formation, although the pathway towards 3,4-
dideoxy acids or esters is assumed to occur under kinetic
control.[9] The formation especially of retro-aldol products is
diminished in ethanol and longer alcohols. Nonetheless, the
reaction pathways towards the formation of 3-deoxy and 3,4-
dideoxy compounds are broadly conserved in various protic
solvents.
Simple alcohols have been considered as environmentally
preferable solvents and ethanol has been described as
preferable to methanol in terms of environmental, health and
safety regulations.[28-29] Hence, the formation of 2,5-dihydroxy-3-
pentenoic acid esters with 42% yield and of related 3-deoxy-
esters at combined yields near 66% from xylose in ethanol is
encouraging for the development of environmentally benign
processes in a future bio-based chemical industry.
Acknowledgements
This work was funded by the Innovation Fund Denmark (case
number 5150-00023B) and by Grant 2013_01_0709 of the
Carlsberg Foundation. 800 MHz NMR spectra were recorded on
the spectrometer of the DTU NMR center supported by the
Villum foundation.
Keywords: carbohydrates • green solvents • NMR • Sn-Beta •
solvent effect
References
[1]
[2]
Y. Román‐Leshkov, J. N. Chheda, J. A. Dumesic, Science 2006, 312,
1933.
J. B. Binder, R. T. Raines, J. Am. Chem. Soc. 2009, 131, 1979-1985.
This article is protected by copyright. All rights reserved.

10.1002/cssc.201700587
ChemSusChem
COMMUNICATION
[3]
[4]
[5]
[6]
P. Gallezot, Chem. Soc. Rev. 2012, 41, 1538-1558.
[19] C. Gaillet, C. Lequart, P. Debeire, J.-M. Nuzillard, J. Magn. Reson.
1999, 139, 454-459.
C. Li, Z. Zhang, Z. K. Zhao, Tetrahedron Lett. 2009, 50, 5403-5405.
A. Corma, S. Iborra, A. Velty, Chem. Rev. 2007, 107, 2411-2502.
E. Taarning, S. Saravanamurugan, M. S. Holm, J. Xiong, R. M. West, C.
H. Christensen, ChemSusChem 2009, 2, 625-627.
[20] Ē. Kupce, R. Freeman, G. Wider, K. Wüthrich, J. Magn. Reson., Ser. A
1996, 122, 81-84.
[21] Z. Zhou, R. Kümmerle, X. Qiu, D. Redwine, R. Cong, A. Taha, D.
Baugh, B. Winniford, J. Magn. Reson. 2007, 187, 225-233.
[22] M. Weber, C. Hellriegel, A. Rueck, J. Wuethrich, P. Jenks, J. Pharm.
Biomed. Anal. 2014, 93, 102-110.
[7]
[8]
[9]
M. S. Holm, S. Saravanamurugan, E. Taarning, Science 2010, 328,
602-605.
S. Tolborg, I. Sádaba, C. M. Osmundsen, P. Fristrup, M. S. Holm, E.
Taarning, ChemSusChem 2015, 8, 613-617.
[23] H. Jancke, F. Malz, W. Haesselbarth, Accred. Qual. Assur. 2005, 10,
421-429.
R. De Clercq, M. Dusselier, C. Christiaens, J. Dijkmans, R. I. Iacobescu,
Y. Pontikes, B. F. Sels, ACS Catal. 2015, 5, 5803-5811.
[24] R. Freeman, H. D. W. Hill, R. Kaptein, J. Magn. Reson. 1972, 7, 327-
329.
[10] M. Dusselier, P. Van Wouwe, F. de Clippel, J. Dijkmans, D. W.
Gammon, B. F. Sels, ChemCatChem 2013, 5, 569-575.
[11] A. Sølvhøj, E. Taarning, R. Madsen, Green Chem. 2016, 18, 5448-
5455.
[25] M. T. Pérez-Prior, J. A. Manso, M. del Pilar García-Santos, E. Calle, J.
Casado, J. Org. Chem. 2005, 70, 420-426.
[26] W. N. P. van der Graaff, C. H. L. Tempelman, G. Li, B. Mezari, N.
Kosinov, E. A. Pidko, E. J. M. Hensen, ChemSusChem 2016, 9, 3145-
3149.
[12] H. S. Chen, A. Wang, H. Sorek, J. D. Lewis, Y. Román‐Leshkov, A.
Bell, ChemistrySelect 2016, 1, 4167-4172.
[13] S. G. Elliot, C. Andersen, S. Tolborg, S. Meier, I. Sadaba, A. E.
Daugaard, E. Taarning, RCS Adv. 2017, 7, 985-996.
[27] J. R. Christianson, S. Caratzoulas, D. G. Vlachos, ACS Catal. 2015, 5,
5256-5263.
[14] S. Tolborg, S. Meier, I. Sádaba, S. G. Elliot, S. K. Kristensen, S.
Saravanamurugan, A. Riisager, P. Fristrup, T. Skrydstrup, E. Taarning,
Green Chem. 2016, 18, 3360-3369.
[28] C. Capello, U. Fischer, K. Hungerbuhler, Green Chem. 2007, 9, 927-
934.
[29] M. Tobiszewski, J. Namieśnik, F. Pena-Pereira, Green Chem. 2017,
10.1039/c6gc03424a.
[15] S. Saravanamurugan, A. Riisager, E. Taarning, S. Meier, Chem.
Commun. 2016, 52, 12773-12776.
[30] S. Valencia, A. Corma, UOP LLC, United States, 1999.
[31] S. Tolborg, A. Katerinopoulou, D. D. Falcone, I. Sádaba, C. M.
Osmundsen, R. J. Davis, E. Taarning, P. Fristrup, M. S. Holm, J. Mater.
Chem. A 2014, 2, 20252-20262.
[16] S. Saravanamurugan, A. Riisager, E. Taarning, S. Meier,
ChemCatChem 2016, 10.1002/cctc.201600783.
[17] S. Tolborg, S. Meier, S. Saravanamurugan, P. Fristrup, E. Taarning, I.
Sádaba, ChemSusChem 2016, 10.1002/cssc.201600757.
[18] E. A. Capanema, M. Y. Balakshin, J. F. Kadla, J. Agric. Food Chem.
2004, 52, 1850-1860.
This article is protected by copyright. All rights reserved.

10.1002/cssc.201700587
ChemSusChem
COMMUNICATION
Entry for the Table of Contents
COMMUNICATION
A quantitative NMR approach is used
to detect, quantify and optimize the
formation sugar-derived building
blocks without depending on
Samuel G. Elliot, Søren Tolborg, Irantzu
Sádaba, Esben Taarning, and Sebastian
Meier*
Page No. – Page No.
reference compounds. Record yields
of target molecules are achieved by
obstructing competing reactions
through solvent choice.
Quantitative NMR Approach to
Optimize the Formation of Chemical
Building Blocks from Abundant
Carbohydrates
This article is protected by copyright. All rights reserved.