Supercritical water treatment for cello-oligosaccharide production from microcrystalline cellulose

Article abstract of DOI:10.1016/j.carres.2014.10.012

Microcrystalline cellulose was treated in supercritical water at 380 °C and at a pressure of 250 bar for 0.2, 0.4, and 0.6 s. The yield of the ambient-water-insoluble precipitate and its average molar mass decreased with an extended treatment time. The highest yield of 42 wt % for DP2-9 cello-oligosaccharides was achieved after the 0.4 s treatment. The reaction products included also 11 wt % ambient-water-insoluble precipitate with a DP_w of 16, and 6.1 wt % monomeric sugars, and 37 wt % unidentified degradation products. Oligo- and monosaccharide-derived dehydration and retro-aldol fragmentation products were analyzed via a combination of HPAEC-PAD-MS, ESI-MS/MS, and GC-MS techniques. The total amount of degradation products increased with treatment time, and fragmented (glucosyl_n-erythrose, glucosyl_n-glycolaldehyde), and dehydrated (glucosyl_n-levoglucosan) were identified as the main oligomeric degradation products from the cello-oligosaccharides.

Full text of DOI:10.1016/j.carres.2014.10.012

Carbohydrate Research 401 (2015) 16–23
Contents lists available at ScienceDirect
Carbohydrate Research
journal homepage: www.elsevier.com/locate/carres
Supercritical water treatment for cello-oligosaccharide production
from microcrystalline cellulose
Lasse K. Tolonen a, Minna Juvonen b, Klaus Niemelä c, Atte Mikkelson c, Maija Tenkanen b, Herbert Sixta a,
⇑
a Aalto University, Department of Forest Products Technology, PO Box 16300, FI-00076 Aalto, Finland
b Department of Food and Environmental Sciences, PO Box 27, FI-00014 University of Helsinki, Finland
c VTT, PO Box 1000, FI-02044 VTT, Finland
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 28 August 2014
Received in revised form 15 October 2014
Accepted 17 October 2014
Available online 25 October 2014
Microcrystalline cellulose was treated in supercritical water at 380 °C and at a pressure of 250 bar for 0.2,
0.4, and 0.6 s. The yield of the ambient-water-insoluble precipitate and its average molar mass decreased
with an extended treatment time. The highest yield of 42 wt % for DP2-9 cello-oligosaccharides was
achieved after the 0.4 s treatment. The reaction products included also 11 wt % ambient-water-insoluble
precipitate with a DPw of 16, and 6.1 wt % monomeric sugars, and 37 wt % unidentified degradation prod-
ucts. Oligo- and monosaccharide-derived dehydration and retro-aldol fragmentation products were ana-
lyzed via a combination of HPAEC-PAD–MS, ESI-MS/MS, and GC–MS techniques. The total amount of
degradation products increased with treatment time, and fragmented (glucosyln-erythrose, glucosyln-
glycolaldehyde), and dehydrated (glucosyln-levoglucosan) were identified as the main oligomeric degra-
dation products from the cello-oligosaccharides.
Keywords:
Biorefinery
Supercritical water
Cellulose
Oligosaccharide
Prebiotics
Ó 2014 Elsevier Ltd. All rights reserved.
1. Introduction
required to hydrolyze the b-glucosidic bond in the cello-oligosac-
charides.3 Fermentation activity in the human intestine was
Cellulose, being globally and abundantly available, is probably
the most important biopolymer that is currently used in a wide
variety of applications such as paper products, textile fibers and
for the production of biofuels and platform chemicals.1 Cellulose
is also the natural source for the manufacture of cello-oligosaccha-
rides. In general, oligosaccharides are water-soluble saccharide
polymers, typically consisting of two to ten monosaccharide units,
although the definition varies. An example of the possible applica-
tions for oligosaccharides is their use as prebiotics that are added
in increasing quantities to foods with the focus on beverages and
milk products.2–4 By definition, prebiotics are food ingredients that
are not digested by humans and have zero metabolizable energy
value. Instead they provide a source of carbon for the intestinal
microflora stimulating the beneficial bacteria in the colon.2 Not
all oligosaccharides are, however, recognized or used as prebiotics
and this applies also to cello-oligosaccharides.2 Yet there is a rea-
son to believe that cello-oligosaccharides are potential prebiotics
because the human digestion system lacks the enzyme that is
reported for cellobiose indicating that it acts as energy source for
certain bacteria.5
An obvious reason for the lack of extensive research on cello-
oligosaccharides is their limited availability and high price, which
is caused by cellulose’s recalcitrant nature compared with other
polysaccharides like starch. Cellulose is a linear polysaccharide
consisting of b(1 ? 4) linked
D-anhydroglucopyranose units
(AGU) in which every second AGU is rotated 180° in the plane,
adjacent units forming a cellobiose. Cellulose exists as a polymer
with the degree of polymerization (DP) up to 10,000 AGUs.1 In
the naturally existing cellulose I allomorph, the cellulose chains
are aligned parallel, forming sheets which are stacked on top of
each other, thus forming ordered crystalline domains interrupted
by less ordered domains.1 In the crystalline domains, a rigid intra
and interchain hydrogen bond network is formed in the cellulose
sheets whereas these sheets are held together by Van der
Waals—forces which were in fact reported to be the pivotal factor
for cellulose recalcitrance.6 Also hydrophobic interaction, which is
caused by the affinity of water molecules to each other, is impor-
tant regarding the insolubility of cellulose in aqueous systems.7
As a result, in order to produce water-soluble cello-oligosaccha-
rides, the cellulose must be depolymerized in a controlled manner
which will render it water-soluble. The challenge is to hydrolyze
cellulose as oligomers instead of obtaining mere monomers, which
⇑
Corresponding author. Tel.: +358 503841746.
E-mail addresses: lasse.tolonen@aalto.fi (L.K. Tolonen), minna.juvonen@helsinki.
fi
(M. Juvonen), klaus.niemela@vtt.fi (K. Niemelä), atte.mikkelson@vtt.fi
(A. Mikkelson), maija.tenkanen@helsinki.fi (M. Tenkanen), herbert.sixta@aalto.fi
(H. Sixta).
http://dx.doi.org/10.1016/j.carres.2014.10.012
0008-6215/Ó 2014 Elsevier Ltd. All rights reserved.

L.K. Tolonen et al. / Carbohydrate Research 401 (2015) 16–23
17
is typically not possible via acid-catalyzed hydrolysis at near-
ambient temperatures. Therefore, a separate dissolution step is
typically needed to create a homogeneous environment for the
hydrolysis of cellulose to oligosaccharides.
The overall sampling and analysis procedure is illustrated in
Scheme 1. In order to avoid the earlier reported problem with
the re-deposition of precipitated cellulose, the time required for
the removal of the undissolved residue was reduced to approxi-
mately 15 s.13 The solution from the reactor’s outlet was taken into
a 10 mL syringe and immediately pressed through a syringe filter
(Acrodisk, PN4523T). In total 50 mL of reaction product solution
was filtered in this way. The syringe filters were then dried at
105 °C, and the amount of undissolved cellulose residue was deter-
mined gravimetrically. The filtered solution was stored in a cold
room to allow the precipitation of cellulose take place. The formed
precipitate was separated by centrifugation for analytical size
exclusion chromatography.
In the early 1990s it was discovered that crystalline cellulose
can be converted to water-soluble species via a noncatalytic sub-
critical or supercritical water treatment.8 This process has the
advantage that the hydrolyzed cellulose is dissolved not as a
monomer but rather as oligomers or short polymers whose DP
depends on the treatment temperature.9 No actual swelling and
dissolution of the cellulose crystallites take place in subcritical
water at least below 300 °C,10 while at temperatures higher than
300 °C and at a pressure of 25 MPa, a crystalline-to-amorphous
transformation was reported.11 This transformation results in a
rapid destruction of the crystallites in near critical and supercriti-
cal water.12–14 When the temperature reaches the critical point
(374 °C and 22.1 MPa) at a pressure of 25 MPa, the internal energy
of the system increases almost stepwise and the physical proper-
ties of water are drastically changed. These factors were attributed
to a stepwise acceleration of cellulose dissolution.12,14 Overall, the
rate of cellulose dissolution increases faster than the correspond-
ing rate of degradation of sugar, which enables the recovery of
the dissolved compounds in high yields.12
Although supercritical water treatment was shown to dissolve
and hydrolyze cellulose in one stage without a dedicated cellulose
solvent, several degradation reactions concomitantly occur during
the supercritical water treatment. These formed reaction products
may have an effect on the use of the produced oligosaccharides in
applications where a high purity is required. In subcritical water
the proton concentration is higher than in ambient water catalyz-
ing hydrolytic depolymerization and dehydration; in near- and
supercritical water dehydration and fractionation via retro-aldol
reactions become more prominent favoring reaction products such
as levoglucosan, 5-hydroxymethylfurfural, erythrose, methylgly-
oxal, glycolaldehyde, and dihydroxyacetone.12,15,16 These degrada-
tion reactions are not restricted to monosaccharides but take place
also at the reducing end groups of polysaccharides.17
Besides the samples collected with the syringe filters, several
liters of reaction solution were collected and stored in a cold room
for more than 48 h. During that time nearly all of the dissolved cel-
lulose chains precipitated. The solid fraction containing the undis-
solved residue and the formed precipitate was removed using
filters (Whatman Polycap heavy duty 5.0/10.0 lm). The filters were
dried overnight at 105 °C and the total amount of undissolved res-
idue and precipitate was determined gravimetrically. The amount
of cellulose precipitate was obtained by subtracting the amount
of undissolved residue from the total amount of residue and
precipitate.
A part of the 0.4 s treated, filtrated sample was fractionated by
preparative-scale size exclusion chromatography (Prep-SEC). First
the sample was concentrated to 1:20 of its volume by a rotary
evaporator (55 °C and 80 mbar) to compensate for the dilution in
In this study we investigated the dissolution and depolymeriza-
tion of microcrystalline cellulose to water-soluble cello-oligosac-
charides and the formation of degradation products thereof. The
yields and molar mass distributions were determined from ambi-
ent-water-soluble oligosaccharides and cellulose precipitate which
was insoluble in ambient water. The formed monomeric and
oligomeric degradation products were investigated by HPAEC-
PAD–MS, ESI-MS/MS, and GC–MS techniques.
2. Experimental
Commercial microcrystalline cellulose (MCC) powder was pur-
chased from Merck and used as raw material. The mass average
molar mass of the MCC was 32.9 kg molꢀ1 and polydispersity index
2.03.10 The carbohydrate composition of the MCC after total hydro-
lysis was 97.8% glucose, 1.0% xylose, and 1.2% mannose, analyzed
by HPAEC-PAD with a CarboPac PA20 column after the hydrolysis
to monosugars as described earlier.18 Water–MCC suspension
was prepared by mixing the MCC powder thoroughly with deion-
ized water. Nitrogen purging was applied in order to remove oxy-
gen from the suspension. The prepared cellulose suspension was
treated with a bench-scale tubular flow reactor system described
earlier.13 Three experiments were conducted at the temperature
of 380 °C employing treatment times of 0.20 s, 0.40 s, and 0.60 s.
The pressure was held at 25.0 MPa in all experiments. The concen-
tration of the cellulose suspension was 0.50 wt % in the feeding
tank and 0.20 wt % in the beginning of the supercritical water
treatment after dilution by supercritical heating water.
Scheme 1. Experimental procedure. Conducted analyses indicated by italics.

18
L.K. Tolonen et al. / Carbohydrate Research 401 (2015) 16–23
the subsequent Prep-SEC in which it was fractioned to 93 fractions,
10.3 mL each, using Bio-Gel P-2 (Bio-Rad) column (height and
diameter 94 cm and 5 cm, respectively) with a particle size of
0.45–0.55 V. Mass spectra were processed using DataAnalysis soft-
ware (Bruker Daltonik GmbH).
The DP3 and DP4 fractions were also analyzed by Dionex ICS
ICS-3000 system coupled to a MSQ Plus mass spectrometer
(Thermo Scientific Dionex) (HPAEC-PAD–MS). The samples were
analyzed without pretreatment after fractionation by prepara-
tive-SEC. The analytes were separated using a CarboPac PA200
45–90 lm. The Prep-SEC fractionation was done at room tempera-
ture, and degassed MilliQ-water was used as eluent. A flow rate of
0.9 mL minꢀ1 was applied.
The molar mass of the polymers in the precipitate was deter-
mined by analytical size exclusion chromatography (SEC). A sol-
vent exchange sequence [water–acetone–N,N-dimethylacedamide
(DMAc)] was carried out in 2 mL Eppendorf-tubes using centrifu-
gation for the separation of the precipitate and solvent. The sam-
ples were dissolved in 90 g Lꢀ1 anhydrous lithium chloride (LiCl)/
DMAc at room temperature under occasional shaking, diluted
according to their estimated concentration to 1.0 g Lꢀ1, and filtered
column (3 mm ꢁ 250 mm) together with
guard column (3 mm ꢁ 50 mm, DionexCorp, USA) at 30 °C using
flow rate 0.3 mL minꢀ1. After injection of a 100
L sample (fil-
tered with a 0.45 m filter), 100 mM NaOH was run through
a CarboPac PA200
l
l
the column for 5 min, after which gradient of 100 mM to
300 mM NaOAc/100 mM NaOH in 33 min was applied. Then col-
umn was washed with 100 mM NaOH/300 mM NaOAc and
300 mM NaOH. After the analytical column, a T-piece diverted
one part of the flow directly to pulsed amperometric detection
(PAD) while the other part of the eluent was desalted through
a Thermo Scientific Dionex ASRS suppressor. After the suppres-
sor, the eluent was mixed with a solution of lithium chloride
with 0.2 lm syringe filters. The SEC-analysis was performed with a
Dionex Ultimate 3000 chromatography system that comprised one
guard column (PLgel Mixed-A, 7.5 ꢁ 50 mm, Agilent Technologies)
and four analytical columns (PL-gel Mixed-A, 7.5 ꢁ 300 mm) in
series and an RI-detector (Shodex RI-101). The analysis was carried
out at room temperature using 9 g Lꢀ1 LiCl/DMAc solution as the
eluent (0.75 mL minꢀ1). Double injections were conducted for each
(500 l
mol Lꢀ1) via second T-piece in order to detect carbohy-
drates by mass spectrometry as lithium adducts. The positive
electrospray ionization-single quadrupole mass spectrometer
was operated at the following conditions: probe temperature
400 °C, nitrogen pressure, cone voltage 90 V, needle voltage
3 kV and scan range of 100–1050 m/z was applied. The delay
times of both detectors were synchronized automatically by
the Chromeleon software package.
Unfractionated water-soluble products were analyzed by GC–
MS as their trimethylsilyl (TMS) derivatives, for the determination
of the monomeric degradation products additional to glucose. The
applied procedure and equipment are described in19 together with
references to relevant mass spectral data. Additional references
were now consulted for the final identification of glycolaldehyde
in cyclic forms and erythrose.20,21
sample (100 lL). Narrow pullulan standards (343 Da–708 kDa,
Polymer Standard Service GmbH, and 1600 kDa, Fluka GmbH) were
used to calibrate the system. The molar mass distributions were
calculated by using a MATLAB script written at Aalto University.
High-performance anion exchange chromatography system
(Dionex ICS-3000) with pulsed amperometric detection (HPAEC-
PAD) was employed for the analysis of water-soluble oligo- and
monomers. The system comprised one Carbopac PA100 guard col-
umn and an analytical column of similar type. Flow rate of
0.600 mL minꢀ1 and temperature of 22 °C were applied. The eluent
concentration was 100 mM for NaOH(aq) throughout all the runs.
The linear NaOAc gradients for the analysis of the oligosugar frac-
tions was 10, 10, 200, 250, 50, and 50 mM at 0, 3, 30, 31, 35, 36, and
45 min, respectively. The SEC-fractions were analyzed using a
modified linear eluent profile in which the NaOAc profile was 15,
15, 200, 250, 50, and 50 mM at 0, 10, 30, 31, 35, 36, and 45 min,
respectively. Cellotriose (ꢂ95%), cellotetraose (ꢂ95%), cellopenta-
3. Results and discussion
3.1. Undissolved residue and cellulose precipitate
ose (ꢂ95%), and cellohexaose ꢂ90% from Megazyme and
D(+)-glu-
cose, p.a. and (+)-cellobiose (Fluka 22150-10G, >99%) were used
D
Supercritical water treatment at 380 °C resulted in the rapid
decomposition of the microcrystalline cellulose. In the supercriti-
cal water treatment and the subsequent separation steps, micro-
crystalline cellulose was fractioned into undissolved residue,
cellulose chains that dissolved in supercritical water but precipi-
tated in ambient water, denoted hereafter as precipitate, and
water-soluble compounds. The shares of these fractions depended
on the treatment conditions (Table 1). After the 0.2 s treatment, a
small amount of solid matter was recovered. This fraction was con-
sidered to be undissolved crystalline cellulose residue although its
low quantity did not allow for analyzing whether the residue had
retained cellulose Ib allomorph. After the 0.4 s and 0.6 s treatments
no solid residue was found, indicating a complete dissolution of the
cellulose crystallites.
The formation of white precipitate was observed a few minutes
after the solution was recovered from the reactor system and con-
tinued for hours. This precipitating fraction has been earlier iden-
tified as cellulose II allomorph, showing that it originates from
completely dissolved cellulose chains.12,13,17,22 The yield and molar
mass of the cellulose precipitate depended on the treatment time:
a longer treatment time resulted in a lower yield of precipitate
(Tables 1 and 2). This is explained by the depolymerization of the
dissolved chain, which was observed as the shifted molar mass
distributions, reducing the degree of polymerization below the
solubility limit of cellulose therefore rendering the chains water-
soluble (Table 2).
to calibrate the analysis. Flory–Schulz type distributions
xꢀ1
ðcðxÞ ¼ bxð1 ꢀ aÞ Þ where c is concentration as a function of x
which is degree of polymerization, and a and b are constants) were
fitted to the observed oligosaccharide concentrations using the
MATLAB’s curve fit toolbox.
The negative ion mass spectrometry experiments of isolated
DP3 and DP4 fractions were performed with Agilent XCT Plus
model quadrupole ion trap mass spectrometer, equipped with an
electrospray ionization source (Agilent Technologies) (ESI-MS/
MS). Five microliter aliquots of the fractions were diluted with
50% methanol (v/v) to a final volume of 200
cellotriose and cellotetraose, were diluted with 50% methanol
(25
g mLꢀ1). Ammonium chloride was added (50 g mLꢀ1) for
[M+Cl]ꢀ-adduct ion formation. Sample and standard solutions
were infused into the ESI source at a flow rate of 5
L minꢀ1 via
lL. Standard samples,
l
l
l
a syringe pump. The electron spray capillary voltage was set to
3200 V and end plate off set to ꢀ500 V. Other parameters were
set automatically by target ion mass (m/z 539 for DP3 fraction
and m/z 701 for DP4 fraction). Nitrogen gas was used as both neb-
ulizing gas and drying gas. The drying gas temperature was set to
325 °C. The drying gas was flow 4 L minꢀ1 and nebulizer pressure
was 1.03 bar (15 psi). The range of m/z 100–1500 was scanned
and 5 scans were averaged for a spectrum. Each spectrum was pro-
duced by accumulating data for 1 min. In MS/MS analysis the frag-
mentation amplitude for collision-induced dissociation (CID) was

L.K. Tolonen et al. / Carbohydrate Research 401 (2015) 16–23
19
Table 1
calibration lines of the existing DP2-6 standards and these known
response factors were then extrapolated to the DP7-9. In order to
avoid problems with integration in the presence of several side
peaks, the calibration and analysis was done against the peak
heights.
The main fractions and identified water-soluble compounds from supercritical water
treatment as percentages on initial microcrystalline cellulose with different treat-
ment times
0.2 s
0.4 s
0.6 s
Main fractions
The concentrations of the identified water-soluble oligosaccha-
rides were clearly affected by the treatment time: the total quan-
tity of water-soluble oligosaccharides increased from 29% after
0.2 s to 42% after 0.4 s (Table 1 and Fig. 1) but decreased when
the treatment was extended to 0.6 s. The increase in total oligosac-
charide concentration between the 0.2 s and 0.4 s treatments is
undoubtedly due to the depolymerization of longer chains, which
was seen as the decreasing molar mass and amount of the precip-
itate. The concentrations of DP1-6 cello-oligosaccharides were
used to fit Flory–Schulz distributions in Figure 1. Good agreement
between the measured data and the distributions supports the ran-
dom depolymerization mechanism. In addition, the Flory–Schulz
distributions match well with the DP7-9 concentrations thereby
indicating that the extended calibration was valid. The parallel
analysis using a PA20 column revealed only low concentrations
of xylose, mannose, and fructose (Table 1), and verified the glucose
concentrations from the oligomer analysis (Fig. 1).
Undissolved residue
Precipitate
8
0
11
89
1100
0
<1
100
1024
35
57
922
Water-soluble comp.a
Total concentration
Identified constituents
DP2-9 COSb
29
42
30
Glucoseb
2.4
0.3
0.2
0.1
2.1
23
5.0
0.5
0.3
0.3
3.5
37
7.7
0.5
0.4
0.5
3.5
57
Xyloseb
Mannoseb
Fructoseb
Degradation productsc
Unident. constituentsd
a
From TOC as anhydroglucan + gravimetric analyses of undissolved residue and
precipitate.
b
HPAEC-PAD with a PA20 column.
GC–MS (see Table 3).
Calculated from the mass balance.
c
d
Table 2
3.3. Dehydration and retro-aldol fragmentation products
Molar mass averages of cellulose precipitate (g molꢀ1) with different treatment times.
0.2 s
0.4 s
0.6 s
A large number of smaller unidentified peaks were detected in
the HPAEC-PAD analysis. In order to get more information on the
other oligomeric products, the sample treated for 0.4 s was frac-
tionated with Prep-SEC and the fractions were analyzed by
HPAEC-PAD. The contour map in Figure 2 revealed a series of the
main oligomer peaks that eluted according to their DP. The main
cello-oligosaccharide peaks of each fraction exhibited a bimodal
shape (insert in Fig. 2) which was reported earlier.22 In addition
to the bimodal main peak, two substantial side peaks were
detected for each cello-oligosaccharide together with several
unidentified peaks in the monosugar range. No oligosaccharides
above DP8 were found in the analyzed fractions, and it is probable
that they precipitated during the storage time and the concentra-
tion step prior to the Prep-SEC fractionation. The slight bending
of peaks seen in Figure 2 was probably due to the drift of the pH
because we used high injection concentrations to ensure the detec-
tion of all side peaks.
Number average
Mass average
2072
3238
6
47
1.6
1851
2643
6
36
1.4
1560
1917
5
21
1.2
a
DP5%
DP95%
b
Polydispersity index
a
5% of precipitate has a lower DP. Rounded up to next integer.
95% of precipitate has a lower DP. Rounded down to next integer.
b
The analytical-SEC revealed that after the 0.2 s treatment
90 wt % of the cellulose precipitate comprised fractions between
DP9 and DP47 (Table 2), and the molar mass distribution was
shifted toward a low molar mass with a prolonged treatment time.
In general, the average molar masses of cellulose precipitate deter-
mined in this study were lower than those reported by Sasaki
et al.12 or Ehara and Saka17 who reported viscosity average values
at around DP50-30 and a distribution from DP13 to DP100, respec-
tively. Caution is required in particular with the low and high
molar mass tails in the distribution because those may be affected
by the band-broadening effect.23
To identify the different oligomeric degradation products from
cellulose, DP3 and DP4 fractions were analyzed with the ESI-MS
on the negative mode as chloride adducts (Fig. 3). The base peaks
The number-weighted average molar mass decreased nearly
linearly against the reciprocal treatment time, suggesting that
depolymerization occurs randomly in supercritical water. In this
case the resulting polydispersity index for high polymers would
be two.24 However, the PDI values are clearly lower than 2
(Table 2), probably due to a fractionation upon precipitation, which
automatically results in a truncated molar mass distribution of
precipitate.
3.2. Water soluble cello-oligosaccharides
The fraction that remained soluble in ambient water was quan-
tified based on the carbon balance (Table 1). The cello-oligosaccha-
rides together with monomeric glucose, which are the primary
depolymerization products of cellulose, were detected and quanti-
fied by HPAEC-PAD analysis. The reliable quantification of all
detected oligosaccharide peaks was challenging because commer-
cial calibration standards were not available beyond DP6. Hence
we used an approach described by Yu and Wu,22 where the
response factors (nC mgꢀ1) were obtained from the slopes of
Figure 1. Analyzed cello-oligosaccharide concentrations. White markers: oligomer
analysis by a PA100 column. Black markers: Monomer analysis by a PA20 column.
Dashed lines represent Flory–Schulz type of molar mass distributions fitted on the
DP1-6 oligosaccharide concentrations.

20
L.K. Tolonen et al. / Carbohydrate Research 401 (2015) 16–23
Figure 2. Contour map of Prep-SEC–HPAEC-PAD analysis for the fractions treated for 0.4 s at 380 °C. The insert shows the chromatogram of the fraction 53, x-axis time (min),
y-axis signal (nC). The minimum contour level 10 nC, baseline signal subtracted from the signal. The and b peaks are discussed in the text.
a
spectrum (Fig. 4A). Precursor ion of m/z 521 was identified as a
dehydration product of cellotriose [Glc3ꢀ18 Da+Cl]ꢀ. In the MS/
MS spectrum of this compound, cross-ring cleavage fragmentation
products were not observed. Generally, the cross-ring fragments
are formed from reducing oligosaccharides if a chloride adduct is
attached and the absence of them indicates that the dehydrated
compound does not have an anomeric carbonyl group available
for retro-aldol reaction.25 Thus the MS results suggest a 1,6-anhy-
drous structure without the anomeric carbonyl group, that is, cel-
lobiose-levoglucosan, for the ion m/z 521 (Fig. 4B). The formation
of the 1,6-anhydroglucose (levoglucosan) has been reported in
the supercritical water treatment of cellulose,26 and cellobiose27,28
by hydrolysis and dehydration reactions, and was also detected in
this study (see below). 1,6-anhydro-b-D-cellooligosaccharides (b-D-
glucopyranosyln-levoglucosans) from cellulose were earlier found
by Ehara and Saka.17
Two other minor compounds, m/z 479 and m/z 451 ions, were
identified as cross-ring fragmented products from the reducing
end of cellotriose. Based on their MS/MS spectra, the m/z 479 ion
(Fig. 4C) was identified as cellobiose–erythrose by loss of 60 Da
glycolaldehyde (C2H4O2). The m/z 451 (Fig. 4D) ion was identified
as hemiacetal form of cellobiose-glycolaldehyde and methanol
present in the MS solvent [Glc3ꢀ120 Da+MeOH+Cl]ꢀ. Small
amount of cellobiose–glycolaldehyde hydrate (hemiacetal with
water) m/z 437/599 [Glc3/4ꢀ120 Da+H2O+Cl]ꢀ and cellobiose–gly-
colaldehyde m/z 419/581 [Glc3/4ꢀ120 Da+Cl]ꢀ were also identi-
fied in the MS spectra (Fig. 3). Similar findings of hydrates of
degradation products have recently been reported applying
HPAEC-PAD–MS.28
Figure 3. ESI-MS spectra of fraction 53 (A) and fraction 45 (B) as chloride anion
adducts. The base peaks were identified as cellotriose (m/z 539) in the DP3 fraction
and as cellotetraose (m/z 701) in the DP4 fraction. In fraction 53 (DP3) minor peaks
were identified as 1,6-anhydro-b-
D
-cellotriose (m/z 521, [Glc3ꢀ18 Da+Cl]ꢀ), cello-
biose–erythrose (m/z 479, [Glc3ꢀ60 Da+Cl]ꢀ), and cellobiose–glycolaldehyde (m/z
451, [Glc3ꢀ120 Da+MeOH+Cl]ꢀ). In fraction 45 (DP4) minor peaks were cellotet-
raose (m/z 759 [Glc4+NaCl+Cl]ꢀ), 1,6-anhydro-b-
D-cellotetraose (m/z 683,
[Glc4ꢀ18 Da+Cl]ꢀ), cellotriose–erythrose (m/z 641, [Glc4ꢀ60 Da+Cl]ꢀ), and cello-
triose–glycolaldehyde
(m/z
613,
[Glc4ꢀ120 Da+MeOH+Cl]ꢀ,
m/z
599
[Glc4ꢀ120 Da+H2O+Cl]ꢀ and m/z 581 [Glc4ꢀ120 Da+Cl]ꢀ). Small amounts of
DPn+1 compounds were also detected outside of currently presented spectrum
range.
The degradation products identified in the ESI-MS/MS analysis
were linked to their corresponding peaks in Figure 2 by employing
the HPAEC-PAD analysis coupled with mass spectrometry (HPAEC-
PAD–MS). Thus the HPAEC-PAD–MS analysis revealed the follow-
ing elution order of the cello-oligosaccharides and their degrada-
tion products: Glcn ꢀ18 Da, Glcn ꢀ120 Da, Glcn, and Glcn ꢀ60 Da,
where Glcn is the corresponding cello-oligosaccharide. In particu-
in the spectra were m/z 539 and m/z 701 in the DP3 and DP4 frac-
tions, respectively. As earlier with HPAEC-PAD analysis, some
minor peaks were also observed in the MS spectra.
The DP3 fraction was further analyzed by ESI-MS/MS to get
more structural information from the compounds in this fraction
and on the oligomeric products in general. The base peak of the
DP3 fraction (m/z 539) was confirmed to be a cellotriose (Glc3)
by comparing the MS/MS spectrum with the cellotriose standard
lar, the a- and b-peaks in Figure 2 were found to be glucosyln-levo-
glucosan (ꢀ18 Da) and glucosyln-erythrose (ꢀ60 Da), respectively.
The lack of reducing end group in the dehydrated products

L.K. Tolonen et al. / Carbohydrate Research 401 (2015) 16–23
21
(glucosyln-levoglucosan) is indeed expected to result in limited
retention in the HPAEC column. As Glcn and Glcn ꢀ120 Da elute
very closely, their order may vary depending on the column
(CarboPacPA100, CarboPacPA200) and applied elution conditions.
Low molar mass constituents from each of the non-fractionated
samples were analyzed via the GC–MS, which resulted in the iden-
tification and quantification of several C4–C6 carbohydrates and
their lower fragments, aliphatic hydroxy or oxocarboxylic acids,
and miscellaneous compounds (Table 3). In addition, several other
compounds could be only partially characterized or they remained
totally unknown.
In the GC–MS chromatograms, up to ten peaks were found elut-
ing between the glucopyranose and cellobiose peaks. Their mass
spectra exhibited an intense m/z 361 peak, characteristic of trim-
ethylsilylated hexose disaccharides and related compounds.29,30
It is therefore reasonable to assume that they represent partially
fragmented cellobiose, such as glucosyl-erythrose and glucosyl-
glycolaldehyde, which supports our results from HPAEC-PAD–MS
and ESI-MS/MS measurements and is in agreement with the liter-
ature.27 Also the detected fructose, levoglucosan, erythrose, dihy-
droxyacetone, and glycolaldehyde are well-established products
from the sub- and supercritical water treatment of cellulose,17,31
cellobiose,16 and glucose.32
The main identified monomeric carbohydrates included fruc-
tose, levoglucosan, mannose, xylose, and erythrose, but small
amounts of xylulose were also present. The fragments of lower
molar mass included glycolaldehyde, found in two dimeric forms,20
and monomeric dihydroxyacetone. The presence of mannose and
xylose, and its isomer xylulose, can readily be explained as trace
impurities in the applied cellulose material, although some man-
nose may also be formed by isomerization of glucose-derived
fructose.
Several carbohydrates or carbohydrate-derived compounds,
usually found in small amounts, could be only partially character-
ized. Their mass spectra as the TMS derivatives usually contained
several ions characteristic of carbohydrate derivatives (e.g., those
at m/z 204, 205, 217 and 218), but in many cases intense atypical
ion peaks were also found (e.g., m/z 155 or 348) hampering final
identifications. It is possible that at least some of them represent
mixed dimers of glycolaldehyde, methylglyoxal, dihydroxyacetone,
or glyceraldehyde.33
Table 3
Minor degradation products formed from cello-oligosaccharides (mg mLꢀ1
)
Compound
0.2 s
0.4 s
0.6 s
Levoglucosan
Fructose
3
3
8
7
5
8
Xylose
2
3
4
Xylulose
Erythrose
<1
2
<1
4
<1
3
Dihydroxyacetone
Glycolaldehyde
Other sugar-type compd
Glycolic acid
<1
1
2
<1
2
3
<1
2
5
4
5
4
Lactic acid
Pyruvic acid
Glyceric acid
2,4-Dihydroxybutanoic
Erythronic acid
Gluconic acid
5-Hydroxymethylfurfural
Reductic acid (tentative)
Catechol
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
2
1
<1
<1
<1
<1
<1
<1
1
<1
<1
<1
<1
<1
2
Figure 4. ESI-MS/MS spectra of fraction 53 (DP3) and proposed fragmentation
pathways of (A) cellotriose (m/z 539), (B) 1,6-anhydro-b- -cellotriose (cellobiose–
1
1
D
<1
<1
3
<1
<1
3
levoglucosan) (m/z 521), (C) cellobiose–erythrose (m/z 479), and (D) cellobiose–
glycolaldehyde (m/z 451) as chloride adduct ion. The actual position of chloride
anion is unknown.
Pyrogallol
Miscellaneous

22
L.K. Tolonen et al. / Carbohydrate Research 401 (2015) 16–23
Of the non-volatile aliphatic carboxylic acids, only glycolic acid
and at the pressure of 250 bar using treatment times of 0.2–0.6 s,
with the highest yield of 42% for DP2-9 cello-oligosaccharides after
0.4 s of treatment. The residual reaction products also consisted of
11% precipitate with a DPw of 16, and 6.1% monomeric sugars.
Attention was paid to the structural analysis of oligomeric and
monomeric degradation products. It was shown that cello-oligo-
saccharides contain fragmented (glucosyln-erythrose, glucosyln-
glycolaldehyde) and dehydrated (glucosyln-levoglucosan) sugar
end-groups. These derivatives may limit the use of such oligosac-
charide product, for example, in human foods.
occurred in moderate amounts (Table 3). It has frequently been
found, together with lactic and pyruvic acids, after the sub- and
supercritical water treatments of cellulose or cellulosic materials,
for example, by Yoshida et al.34 The identified glyceric, 2,4-dihydr-
oxybutanoic, erythronic, and gluconic acids represent known
oxidation or fragmentation products of different carbohydrates of
which the 2,4-dihydroxybutanoic acid is typically formed during
alkaline treatments.
3.4. Summarizing remarks
Acknowledgments
The current results show that crystalline cellulose is dissolved
and depolymerized in supercritical water treatment forming dis-
solved polymers, which are insoluble in ambient water, ambient-
water-soluble cello-oligosaccharides, monosaccharides, and degra-
dation products thereof. In the present study, the maximum yield
of undegraded cello-oligosaccharides was 42%, which compares
well with the yields of 32% for oligosaccharides and 47% for total
water-soluble sugars reported by Ehara and Saka with a single-
stage treatment for 0.2 s at 400 °C.31 Zhao et al. have also reported
a 39% yield of DP2-5 oligosaccharides using a batch reactor at
380 °C for 16 s.26
In supercritical water the depolymerization of dissolved cellu-
lose chains occurs according to a random cleavage of the glycosidic
bonds owing to the detected Flory–Schulz type of distribution of
the formed cello-oligosaccharides. Depolymerization via hydroly-
sis is not, however, the only degradation mechanism in supercrit-
ical water treatment. Also thermal chain cleavage without water
addition results in the formation of a cello-oligosaccharide with a
levoglucosan end group, and dehydration and fractionation reac-
tions take place in the reducing end groups as in the present and
earlier studies show.35 In general, the quantity of degraded com-
pounds increased when treatment time is extended. This is due
to the dehydration and fractionation of the end groups but also
due to the formation of new end groups by the hydrolytic depoly-
merization, which become susceptible to aforementioned reac-
tions. On the contrary, when the treatment time is very short,
the amount of degraded AGUs can be limited to less than 4%; at
the same time a significant amount of the dissolved cellulose
typically remains insoluble in ambient water.12,36,37
The authors would like to acknowledge Sanna Koutaniemi for
her help with the preparative size exclusion chromatography, Tarja
Tamminen for discussions and interpretation of the data, and Sep-
po Jääskeläinen and his crew for their efforts with the reactor sys-
tem. This work was funded by the Finnish Funding Agency for
Technology and Innovations (TEKES), and Finnish Bioeconomy
Cluster Ltd as a part of the Future Biorefinery Program.
Supplementary data
Supplementary data (molar mass distributions of cellulose pre-
cipitate, a description of the extended calibration method used for
DP7-9 cello-oligosaccharides, and the HPAEC-PAD and HPAEC-
PAD–MS chromatograms of the discussed analyses) associated
with this article can be found, in the online version, at http://
dx.doi.org/10.1016/j.carres.2014.10.012.
References
1. Klemm, D.; Heublein, B.; Fink, H.; Bohn, A. Angew. Chem., Int. Ed. 2005, 44,
3358–3393.
2. Roberfroid, M. J. Nutr. 2007, 137, 830S–837S.
3. Mussatto, S. I.; Mancilha, I. M. Carbohydr. Polym. 2007, 68, 587–597.
4. Cardenas-Toro, F. P.; Alcazar-Alay, S. C.; Forster-Carneiro, T.; Meireles, M. A. A.
Food Publ. Health 2014, 4, 123–139.
5. Nakamura, S.; Oku, T.; Ichinose, M. Nutrition 2004, 20, 979–983.
6. Gross, A. S.; Chu, J. J. Phys. Chem. B 2010, 114, 13333–13341.
7. Bergenstråhle, M.; Wohlert, J.; Himmel, M. E.; Brady, J. W. Carbohydr. Res. 2010,
345, 2060–2066.
8. Adschiri, T.; Hirose, S.; Malaluan, R.; Arai, K. J. Chem. Eng. Japan 1993, 26, 676–
680.
9. Yu, Y.; Wu, H. Energy Fuels 2010, 24, 1963–1971.
To obtain an optimal yield of water-soluble cello-oligosaccha-
rides it is essential to optimize the treatment conditions in a way
which minimizes the formation of the water-insoluble precipitate,
the monomeric sugars and their degradation products. The frag-
mentation reactions become more favored over hydrolysis reac-
tions in supercritical water compared with subcritical water.15,31
Therefore it may be preferable to limit supercritical water treat-
ment only for the dissolution of the cellulose crystallites, and carry
out the final hydrolysis to water-soluble oligomers in subcritical
water. This has been demonstrated by Ehara and Saka with prom-
ising results in terms of a low concentration of unwanted degrada-
tion products.31 On the other hand, the formation of furfurals and
phenols through dehydration is known to be pronounced under
ionic conditions in subcritical water.15 Therefore, a short treatment
in near critical or supercritical water appears desirable in order to
reduce the overall treatment time and avoid the formation of furf-
urals and phenols which are known to polymerize easily resulting
in humin or soot formation.
10. Tolonen, L. K.; Zuckerstatter, G.; Penttilä, P. A.; Milacher, W.; Habicht, W.;
Serimaa, R.; Kruse, A.; Sixta, H. Biomacromolecules 2011, 12, 2544–2551.
11. Deguchi, S.; Tsujii, K.; Horikoshi, K. Green Chem. 2008, 10, 191–196.
12. Sasaki, M.; Adschiri, T.; Arai, K. AIChE J. 2004, 50, 192–202.
13. Tolonen, L. K.; Penttilä, P. A.; Kruse, A.; Serimaa, R.; Sixta, H. Cellulose 2013, 20,
2731–2744.
14. Cantero, D. A.; Bermejo, M. D.; Cocero, M. J. J. Supercrit. Fluids 2012, 75, 48–57.
15. Kruse, A.; Gawlik, A. Ind. Eng. Chem. Res. 2003, 42, 267–279.
16. Sasaki, M.; Furukawa, M.; Minami, K.; Adschiri, T.; Arai, K. Ind. Eng. Chem. Res.
2002, 41, 6642–6649.
17. Ehara, K.; Saka, S. Cellulose 2002, 9, 301–311.
18. Testova, L.; Chong, S.; Tenkanen, M.; Sixta, H. Holzforschung 2011, 65, 535–542.
19. Borrega, M.; Niemelä, K.; Sixta, H. Holzforschung 2013, 67, 863–870.
20. Novina, R. Chromatogrophia 1984, 18, 96–98.
21. Havlicek, J.; Peterson, G.; Samuelson, O. Acta Chem. Scand. 1972, 26, 2205–2215.
22. Yu, Y.; Wu, H. Ind. Eng. Chem. Res 2009, 48, 10682–10690.
23. Baumgarten, J. L.; Busnel, J. P.; Meira, G. R. J. Liq. Chromatogr. Related Technol.
2002, 25, 1967–2001.
24. Flory, P. J. J. Am. Chem. Soc. 1936, 58, 1877–1885.
25. Vinueza, N. R.; Gallardo, V. A.; Klimek, J. F.; Carpita, N. C.; Kenttämaa, H. I. Fuel
2012, 105, 235–246.
26. Zhao, Y.; Lu, W.; Wang, H. Chem. Eng. J. 2009, 150, 411–417.
27. Kabyemela, B. M.; Takigawa, M.; Adschiri, T.; Malaluan, R. M.; Arai, K. Ind. Eng.
Chem. Res. 1998, 37, 357–361.
28. Yu, Y.; Shafie, Z. M.; Wu, H. Ind. Eng. Chem. Res. 2013, 52, 17006–17014.
29. Niemelä, K. Carbohydr. Res. 1989, 194, 37–47.
4. Conclusions
30. Kochetkov, N. K.; Chizhov, O. S.; Molodtsov, N. V. Tetrahedron 1968, 24, 5587–
5593.
31. Ehara, K.; Saka, S. J. Wood Sci. 2005, 51, 148–153.
32. Kabyemela, B. M.; Adschiri, T.; Malaluan, R. M.; Arai, K. Ind. Eng. Chem. Res.
1999, 38, 2888–2895.
Supercritical water treatment appears to be a promising process
for the manufacture of water-soluble cello-oligosaccharides. The
present study demonstrated this at the temperature of 380 °C

L.K. Tolonen et al. / Carbohydrate Research 401 (2015) 16–23
23
33. Gardiner, D. Carbohydr. Res. 1966, 2, 234–239.
36. Sasaki, M.; Iwasaki, K.; Hamaya, T.; Adschiri, T.; Arai, K. Kobunshi ronbunshu
34. Yoshida, K.; Miyafuji, H.; Saka, S. J. Wood Sci. 2009, 55, 203–208.
35. Matsumura, Y.; Sasaki, M.; Okuda, K.; Takami, S.; Ohara, S.; Umetsu, M.;
Adschiri, T. Combust. Sci. Technol. 2006, 178, 509–536.
2001, 58, 527–532.
37. Cantero, D. A.; Bermejo, M. D.; Cocero, M. J. Bioresour. Technol. 2012, 135, 697–
703.