Dissolution of wood in α-keto acid and aldehydic carboxylic acids and fractionation at room temperature

Article abstract of DOI:10.1039/c4gc00504j

Wood pulverised using a ball mill was dissolved in an α-keto acid, pyruvic acid, and two aldehydic carboxylic acids, namely glyoxylic acid and formic acid, at room temperature. Viscosity and optical transmittance analyses revealed that the wood particles dissolved in glyoxylic acid and formic acid in several hours, while they dissolved in pyruvic acid in several days. Size exclusion chromatography demonstrated that substances with number average molecular mass exceeding 10⁵ were present in every solution. Incubation of these carboxylic acids with methanol produced the corresponding ester and/or (hemi)acetal derivatives, suggesting that the dissolution mechanism involves the substitution of hydroxyl groups in biomass. After dissolution, all solutions demonstrated non-Newtonian fluid behaviour with a non-Newtonian index n ranging from 0.41 to 0.72. When the dissolution time increased, the non-Newtonian index of the formic acid solution increased towards the value for a Newtonian fluid (n = 1). This rise was accompanied by partial depolymerisation. Evaporation of the wood solution gave a smooth transparent film. Using the water-immiscible solvent 2-methyltetrahydrofuran (2-MeTHF), the formic acid solution was separated into cellulosic porous solids, lignin-rich 2-MeTHF-soluble fraction and other components at room temperature. The dissolution of woody biomass at room temperature creates a new biorefinery process for the conversion of wood components in various solution reactions. In particular, in this study, pyruvic acid, which plays a central role in the metabolism of living organisms, is used as a new solvent for wood, thus bridging biochemistry and wood biorefinery. This journal is the Partner Organisations 2014.

Full text of DOI:10.1039/c4gc00504j

Green Chemistry
View Article Online
View Journal | View Issue
PAPER
Dissolution of wood in α-keto acid and
aldehydic carboxylic acids and fractionation
at room temperature†
Cite this: Green Chem., 2014, 16,
3569
Yuri Nishiwaki-Akine and Takashi Watanabe*
Wood pulverised using a ball mill was dissolved in an α-keto acid, pyruvic acid, and two aldehydic car-
boxylic acids, namely glyoxylic acid and formic acid, at room temperature. Viscosity and optical transmit-
tance analyses revealed that the wood particles dissolved in glyoxylic acid and formic acid in several
hours, while they dissolved in pyruvic acid in several days. Size exclusion chromatography demonstrated
that substances with number average molecular mass exceeding 105 were present in every solution. Incu-
bation of these carboxylic acids with methanol produced the corresponding ester and/or (hemi)acetal
derivatives, suggesting that the dissolution mechanism involves the substitution of hydroxyl groups in
biomass. After dissolution, all solutions demonstrated non-Newtonian fluid behaviour with a non-Newto-
nian index n ranging from 0.41 to 0.72. When the dissolution time increased, the non-Newtonian index of
the formic acid solution increased towards the value for a Newtonian fluid (n = 1). This rise was
accompanied by partial depolymerisation. Evaporation of the wood solution gave a smooth transparent
film. Using the water-immiscible solvent 2-methyltetrahydrofuran (2-MeTHF), the formic acid solution
was separated into cellulosic porous solids, lignin-rich 2-MeTHF-soluble fraction and other components
at room temperature. The dissolution of woody biomass at room temperature creates a new biorefinery
process for the conversion of wood components in various solution reactions. In particular, in this study,
pyruvic acid, which plays a central role in the metabolism of living organisms, is used as a new solvent for
wood, thus bridging biochemistry and wood biorefinery.
Received 22nd March 2014,
Accepted 14th May 2014
DOI: 10.1039/c4gc00504j
www.rsc.org/greenchem
lated numerous studies addressing the solubilisation of ball-
Introduction
milled wood in various solvent systems such as DMSO–LiCl,2
Widely available in large quantities and non-competing with DMSO–pyridine,3 urea and NaOH–boric acid.4 The effective-
food supply, woody biomass is expected to serve as a platform ness of ionic liquids has also been examined,5 but the solvent
for green chemicals and effectively mitigate greenhouse gas cost limits their industrial implementation.
emission effects. The separation and conversion of its cell wall
Simple, biomass-derived and highly biodegradable chemi-
components have emerged as important technologies for the cals that dissolve and separate woody biomass under mild con-
sustainable production of renewable fuels and chemicals. ditions are attractive for the creation of a new biorefinery
Major components of woody biomass comprise cellulose, process. This study aims at harnessing these solvent properties
hemicelluloses and lignin, which is a heterogeneous aromatic to develop a new wood solubilisation and fractionation tech-
polymer. Cell wall polysaccharides are coated with lignin; thus, nology at room temperature. Here, we report the dissolution of
their dissolution and separation is difficult and hinders their ball-milled wood in highly biodegradable acids such as
utilisation. In 2003, ball-milled wood was dissolved in pyruvic, glyoxylic and formic acids at room temperature; more-
dimethyl sulphoxide (DMSO) mixtures with N-methylimidazole over, the resulting solution in formic acid was fractionated by
or tetrabutylammonium fluoride, and the components were precipitation. These solvents are biologically important com-
analysed by NMR spectroscopy.1 This pioneering report stimu- pounds produced by various enzymatic reactions as well as
organic synthesis. The simplest α-keto acid, pyruvic acid, is a
biologically and chemically synthesised compound that is
Research Institute for Sustainable Humanosphere, Kyoto University, Uji, Kyoto 611-0011,
important for the metabolism of glucose in living organisms.
Japan. E-mail: twatanab@rish.kyoto-u.ac.jp; Fax: +81 774 38 3681;
The simplest carboxylic acid, formic acid, has been produced
Tel: +81 774 38 3640
by free radical reaction of oxalic acid and thermochemical
treatment of plant biomass. Glyoxylic acid, which is an inter-
†Electronic supplementary information (ESI) available. See DOI: 10.1039/
c4gc00504j
This journal is © The Royal Society of Chemistry 2014
Green Chem., 2014, 16, 3569–3579 | 3569

View Article Online
Paper
Green Chemistry
mediate of the glyoxylate cycle, has also been synthesised by Reaction between acids and methanol
oxidation of glyoxal. In aqueous solutions, glyoxylic acid has
an equilibrium with its hydrated form.
Pyruvic acid (0.44 g, 5.0 mmol) and methanol (0.24 g,
7.5 mmol) were mixed at room temperature for 30 min. Next, a
small portion of the mixture (5–10 mg) was dissolved in CDCl3
(0.5 mL) and transferred into a 5 mm NMR tube for analysis.
Similarly, mixtures of glyoxylic acid monohydrate (0.23 g,
2.5 mmol) and methanol (0.12 g, 3.75 mmol) as well as formic
acid (0.46 g, 10 mmol) and methanol (0.48 g, 15 mmol) were
prepared and analysed by NMR spectroscopy.
Formic acid has been used to delignify corn cob at 60 °C6
and birch wood at boiling temperature.7 Lignin has been
extracted from Miscanthus8 and sugarcane bagasse9 with
refluxing formic acid solutions containing hydrochloric or sul-
phuric acids. The hydrolysis of pulp was also conducted in
formic acid between 180 and 220 °C.10 Glyoxylic acid has dis-
solved pulp above 50 °C;11 however, no report has addressed
the dissolution of wood by these carboxylic acids without addi-
tives at room temperature. Pyruvic acid is a new solvent for
woody biomass that connects biological metabolisms and
wood biorefinery. This study provides a new process that dis-
solves whole wood at room temperature and separates porous
cellulosic solids from the solution by precipitation using water
and 2-methyltetrahydrofuran (2-MeTHF), a water-immiscible
solvent derived from furfural produced from biomass. Precipi-
tation with 2-MeTHF facilitates the separation between solid,
water and solvent, thus enabling solvent recovery with
minimum energy input. Therefore, this new solubilisation and
fractionation process is environmentally benign.
1
The H NMR spectra were recorded on a JNM Lambda-400
spectrometer (JEOL Ltd, Tokyo, Japan) at 400 MHz using tetra-
methylsilane as an internal reference. The products (1–5) were
quantified by 1H NMR spectra. Yields were calculated using
integrals of the 1H NMR peaks corresponding to the remaining
unreacted acid and the newly formed products. In reactions
involving glyoxylic acid, yields were calculated from the
amount of unreacted methanol because the amount of
unreacted acid could not be directly determined by NMR
spectroscopy.
Gas chromatography mass spectrometry (GC–MS) analysis
was performed using a Shimadzu GCMS-QP5050A system (Shi-
madzu, Co., Ltd, Kyoto, Japan) equipped with a DB-5MS
column (30 m × 0.25 mm i.d., 0.25 μm film thickness, Agilent
Technologies, USA). The GC oven temperature was maintained
at 50 °C for 3 min, then increased to 300 °C at a heating rate
of 8 °C min−1 and maintained at 300 °C for 20 min. Helium
was used as a carrier gas at a flow rate of 1.6 mL min−1. The
mass spectrometer was operated in the electron impact ioniza-
tion mode at 70 eV for a mass/charge (m/z) scan range of
40–600. The split ratio was 1 : 10, and the interface tempera-
ture was set at 300 °C.
Experimental
Materials and general methods
Chemicals were purchased from Wako Pure Chemical Industry
Co., Ltd (Osaka, Japan), Nacalai Tesque Inc. (Kyoto, Japan) and
Tokyo Chemical Industry Co., Ltd (Tokyo, Japan).
Eucalyptus globulus wood chips were purchased from Oji
Paper Co., Ltd (Tokyo, Japan). The wood chips were ground to
20 mesh using a Wiley mill, extracted with ethanol–toluene
(1 : 2, v/v) for 10 h and dried at 105 °C overnight. The dried
wood particles were ground using a ball mill for 48 h under a
nitrogen atmosphere.
The products were identified by comparing the NMR and
mass peaks with data reported in the Wiley and SDBS libraries
(http://sdbs.db.aist.go.jp (National Institute of Advanced
Industrial Science and Technology), 2013/10).
The water content of the ball-milled wood was determined
to be 7.85% 0.05% after weight reduction by drying at 105 °C Quantification of soluble and non-filterable materials
for 19 h. Yields were calculated based on the initial amount of
Milled wood and formic acid were mixed and stirred for 1, 18
milled wood excluding the water content. The average and
or 24 h. The mixture was then filtered under reduced pressure
standard deviation of three experimental data were determined
through a glass filter (ADVANTEC, GA-100, Φ 5.5 cm), which
for non-filterable material quantification, viscosity, and optical
was pre-dried for 1 h at 105 °C. The collected non-filterable
transmittance measurements.
materials were dried overnight with a vacuum freeze dryer and
weighed. The filtrate was concentrated to dryness using a
Dissolution or dispersion of milled wood in carboxylic acids
rotary evaporator and vacuum freeze dryer, and weighed.
by immersion
Dissolution in liquid carboxylic acids was achieved by adding
milled wood (0.2 g) to carboxylic acid (4 g) with stirring at room
Viscosity and fluid property measurements
temperature. For solid carboxylic acids, the acids were diluted The viscosity of the solubilised samples was determined using
with distilled water to prepare saturated aqueous solutions and a TVE-25H rotating viscometer (Toki Sangyo Co., Ltd, Tokyo,
milled wood (0.2 g) was dissolved in the resulting aqueous car- Japan) equipped with a standard cone plate rotor (1°34′ × R24)
boxylic acid solution (4 g). In each case, a 4.8% (w/w) solution in the mode H (η < 1296 mPa s) or mode U (η ≥ 2592 mPa s).
was obtained after complete dissolution. The volume of the Measurements were performed at a rotational speed of 5 rpm
mixture was scaled up for viscosity measurements. For UV- for 1 min. The viscometer temperature was maintained at
visible spectroscopic measurements, the 4.8% solutions were 30 °C by water circulation using a MC-01 micro chiller
diluted 100 or 200 times with the acid used for dissolution.
(Advanced Thermal Sciences, CA, USA).
3570 | Green Chem., 2014, 16, 3569–3579
This journal is © The Royal Society of Chemistry 2014

View Article Online
Green Chemistry
Paper
For fluid property investigations, viscosities were measured
at different rotational speeds (0.5, 1, 2.5, 5, 10 and 20 rpm) in
the mode U under the same conditions, unless the viscosity
became out of range at high rotational speeds.
The shear rate γ (s−1) was calculated by multiplying the
˙
value of rotational speed (rpm) by the shear rate coefficient
3.83 (for a standard rotor). Shearing stress σ (Pa) was obtained
by multiplying the machine reading (%) by the shearing stress
coefficient 1.986 (mode U). Thus, the viscosity η (Pa s) was
then determined using eqn (1):
η ¼ σ=˙γ
ð1Þ
According to the Ostwald–de Waele relationship,12,13 the
shear rate γ can be related to σ or η, as shown in eqn (2) and
˙
Fig. 1 Fractionation of
diluted with water and 2-MeTHF.
a
ball-milled wood solution in formic acid
(3):
σ ¼ K˙γ n;
η ¼ K˙γnꢀ1
ð2Þ
ð3Þ
;
Fractionation
Wood components in a ball-milled wood solution in formic
acid were separated using a water-immiscible solvent, 2-MeTHF.
Water and 2-MeTHF were added to the milled wood solution
and fractionation was performed as shown in Fig. 1. The water
and 2-MeTHF solutions phase-separated and precipitates were
formed. The upper 2-MeTHF layer was isolated and the lower
aqueous suspension was filtered to separate the water layer
from the precipitate. The precipitate was freeze-dried to obtain
fraction 1 while water and the 2-MeTHF layer were evaporated
and freeze-dried to obtain fractions 2 and 3, respectively.
Phase separation profiles of ball-milled wood solutions in
glyoxylic and pyruvic acids were also examined by addition of
water and 2-MeTHF at various weight compositions.
where K is the consistency index and n is the non-Newtonian
(or flow behaviour) index calculated by the power
approximation.
Optical transmittance measurements
About 4 g of a liquid was placed in a test tube (inner diameter:
16 mm) and the transmittance T (%) was measured at 660 nm
using a Miniphoto 518R optical density (OD) monitor (TAITEC,
Saitama, Japan). The transmittance of an opaque liquid
showing a negative reading was assumed to be 0.
Determination of molecular weight distributions by size
exclusion chromatography (SEC)
A mixture of milled wood and acid was diluted with dimethyl-
formamide (DMF, 80 mg mL−1) and the solution was filtered
through a 0.45 μm polytetrafluoroethylene filter. The average
number molecular mass (Mn) distributions were determined
using a Prominence SEC system (Shimadzu Corporation,
Kyoto, Japan) equipped with a photodiode array detector
SPD-M20A set at 280 nm, a refractive index detector RID-10A
and two Shodex KD-803 columns (8.0 × 300 mm, Showa Denko
KK, Tokyo, Japan) linked to a guard column. A 10 μL sample
was injected and analysed at 40 °C (column temperature)
using DMF as an eluent at a flow rate of 0.8 mL min−1. The
columns were calibrated using polyethylene glycol and poly-
ethylene oxide standards presenting molecular masses ranging
from 586 to 107 000 g mol−1. For the calibration, the time
difference between UV and RI detectors (0.2 min) was cor-
rected using a polystyrene standard.
Pyrolysis GC–MS of ball-milled wood fractions obtained from
the formic acid solution
Compounds in each fraction were analysed by pyrolysis gas
chromatography mass spectrometry (Py–GC–MS). The pyrolysis
was performed using a pyrolyser PY-2020D (Frontier Labora-
tories, Fukushima, Japan) connected to a Shimadzu GC–MS
QP5050A system (Shimadzu, Co., Ltd, Kyoto, Japan) equipped
with a DB-5MS column (30 m × 0.25 mm i.d., film thickness:
0.25 μm; Agilent Technologies, USA). The sample (ca. 0.1 mg)
was placed in a small cup, which was placed in a pyrolysis
chamber. The pyrolysis was performed at 500 °C for 12 s and
the interface temperature was 250 °C. Subsequently, the pro-
ducts were introduced into the GC–MS system with a split ratio
of 1 : 97. The GC–MS analysis was performed using helium
(1.5 mL min−1) as a carrier gas. The GC oven temperature was
maintained at 45 °C for 4 min, increased to 300 °C at a
heating rate of 4 °C min−1 and maintained at 300 °C for
UV-visible light absorption measurements
UV-visible absorption spectra of the milled wood solution were 20 min. The mass spectrometer was operated in the electron
recorded between 190 and 800 nm on a U-3310 spectropho- impact ionization mode at 70 eV for the m/z scan range of
tometer (HITACHI, Ltd, Tokyo, Japan) at a scan speed of 40–600. Products were identified by comparing their mass
300 nm min−1 in wavelength scan mode using a quartz cell peaks with spectral data from the Wiley and NIST libraries,
(light path length: 10.0 mm).
and other reports.14–17
This journal is © The Royal Society of Chemistry 2014
Green Chem., 2014, 16, 3569–3579 | 3571

View Article Online
Paper
Green Chemistry
Compositional analysis of the fractions
The holocellulose content was determined according to the
Wise chlorite method.18 Klason lignin was determined from
the weight ratio of the solid residue after a two-step acid hydro-
lysis using 72% and 1.9% sulphuric acid according to the
TAPPI T222 om-98 procedure. The acid-soluble lignin in the fil-
trate was determined at 204 nm using a UV-vis spectropho-
tometer U-3310 (Hitachi Ltd, Tokyo, Japan) after 10-fold
dilution using 3% H2SO4 according to Tappi UM 250. An
absorption coefficient of 106 L g−1 cm−1 was used.19
Monosugars obtained by neutralization following a two-step
acid hydrolysis were characterised using a HPLC equipped
with a post-column and a fluorometric detector20 using the
Prominence HPLC system (Shimadzu, Co., Ltd, Kyoto, Japan)
equipped with two Aminex HPX-87P columns (Bio-Rad, CA,
USA) connected to two guard columns. For fluorescence detec-
tion, the samples were labelled at 150 °C using an aqueous
solution of L-arginine (1%) and boric acid (3%). The sample
solution (5 μL) was injected and analysed at 58 °C using water
Fig. 2 Photographs showing ball-milled wood mixed with pyruvic acid,
aqueous glyoxylic acid and formic acid at a concentration of 0.5%. (a) Clear
solution obtained using pyruvic, aqueous glyoxylic and formic acids 1 week
after mixing. (b) Suspension obtained 1 day after mixing with pyruvic acid.
Table 1 Solubility of the ball-milled wood in various carboxylic acids
after standing for 7 daysa
Acid
Optical
as an eluent at a flow rate of 0.3 mL min−1
.
concentration transmittance
21
pKa
(wt%)
9.0
T (%), 660 nm Solubility
Scanning electron microscopy
Oxalic acid–
water
1.25,
3.81
2.39
3.18
0.05 0.03
Insoluble
Scanning electron microscopy (SEM) images of the original
ball-milled E. globulus and the glass filter before and after fil-
tration of the formic acid solution were taken using a 3D Real
Surface View Microscope VE-9800 (Keyence Corporation,
Osaka, Japan). The sample was mounted on a sample stub
using a double-sided carbon tape and gold-coated using a
magnetron sputter coater MSP-1S (Vacuum Device Inc.,
Ibaraki, Japan). The SEM image of the precipitate (Fr. 1)
obtained by 2-MeTHF–water fractionation of the wood solution
in formic acid was acquired.
Pyruvic acid
Glyoxylic acid–
water
Formic acid
DL-Lactic acid
Acetic acid
100
64
7.99 0.5
7.28 1.9
Soluble
Soluble
3.75
3.86
4.76
100
100
100
1.65 0.2
0.03 0.03
0.006 0.00
Soluble
Insoluble
Insoluble
a The solubility was evaluated using optical transmittance.
Wiley-milled wood (20 mesh) under similar conditions,
suggesting that ball-milling was necessary to disrupt wood cell
walls and realise an efficient dissolution.
A strong acid (low pKa) was expected to dissolve wood more
efficiently than a weak acid because trifluoroacetic acid (pKa =
0.5221) is known to dissolve cellulose.22 However, glyoxylic acid
(pKa = 3.18) dissolved the milled wood more rapidly than
pyruvic acid (pKa = 2.39), revealing that acidity is not the only
factor that determines the solubilisation efficiency.
Pyruvic, glyoxylic and formic acids contain an acyl and/or a
formyl group, suggesting that these groups may contribute to
their high solubilisation efficiency via acetal and/or hemiacetal
formation. Paraformaldehyde–DMSO23,24 and trichloroacetal-
dehyde (chloral)–DMF–pyridine25–27 are known cellulose sol-
vents and their solubilisation efficiency was attributed to the
formation of hemiacetal bonds.26
To verify this hypothesis, methanol was reacted with the
three solvents at room temperature and the products were ana-
lysed by NMR and GC–MS (Table 2). The reaction of glyoxylic
acid with methanol formed 2-hydroxy-2-methoxyacetic acid (3)
as a major product, indicating that hemiacetalisation readily
occurred only by mixing glyoxylic acid and alcohol. After several
days, the amount of compound 3 decreased while that of
methyl 2-hydroxy-2-methoxy acetate (6) increased. In contrast,
the formic acid reaction with methanol primarily afforded
Results and discussion
Dissolution of ball-milled wood in carboxylic acids
The solubility of the ball-milled wood in various carboxylic
acids was investigated by immersion in the acid solutions with
stirring at room temperature. The ball-milled wood showed
high solubility in 64% aqueous glyoxylic acid solution
(glyoxylic acid monohydrate–water = 80 : 20) as well as pyruvic
and formic acids (Fig. 2). However, DL-lactic acid, acetic acid
and aqueous oxalic acid did not dissolve the milled wood.
The dissolution of the milled wood was evaluated by measur-
ing the transmittance of the mixture after 7 days (Table 1). The
milled wood dissolved readily in pure pyruvic and formic acids
but only partially in aqueous carboxylic acids containing more
than 20% water, except for glyoxylic acid, which solubilised the
milled wood at high water content (36%).
The ball-milled wood was completely dissolved in pyruvic
acid at room temperature after 4–6 days. A shorter mixing time
(1 day) was not sufficient to obtain complete dissolution
(Fig. 2b). In contrast, formic acid and 64% aqueous glyoxylic
acid dissolved the milled wood in 4–6 h and 30–60 min,
respectively. However, the three acids did not dissolve the
3572 | Green Chem., 2014, 16, 3569–3579
This journal is © The Royal Society of Chemistry 2014

View Article Online
Green Chemistry
Paper
able materials became measurable. The quantification sup-
ported that the milled wood was mostly dissolved in formic
acid. In every measurement, the sum of filtrate and non-filter-
able materials exceeded 100% of the initial ball-milled wood.
This clearly indicates that formic acid was bound to the wood
by chemical modification such as esterification. The amounts
of filtrate and non-filterable materials in pyruvic or glyoxylic
acid could not be determined because the high viscosity hin-
dered filtration. The formic acid solution was less viscous than
pyruvic or glyoxylic acid solutions. However, the formic acid
solution was adhesive to the glass filter and the solution
adhered to the filter even after fully washing with the solvent,
as is evident from the SEM images before and after filtration
(Fig. 3(a) and (b)). This is the reason why the quantity of non-
filterable fraction at 18 h and 24 h was not 0% (Table 3). None
of the granular materials observed in the original ball-milled
wood (Fig. 3c) were detected by SEM after filtration, confirm-
ing the dissolution of ball-milled wood in formic acid.
Table 2 Yields of products in reactions of the carboxylic acids with
methanol at room temperature
Acetal or
hemiacetal
Ester, acetal or
hemiacetal
Acid
Ester
Pyruvic acid
N.D.a
1
2%
2
6%
Glyoxylic acid
N.D.a
3
37%
4
6%
Formic acid
N.D.a
N.D.a
5
14%
a N.D., not detected.
Each 4.8% milled wood solution in carboxylic acid was
more viscous than their corresponding solvent. Therefore, the
dissolution process was monitored using a rotating viscometer
(Fig. 4). When the milled wood was dissolved in 64% aqueous
glyoxylic acid (17 mPa s), the viscosity increased within 1 h
and reached a value of 4170 mPa s after 1 day before decreas-
ing. For pyruvic acid (17 mPa s), the viscosity increased to
1854 mPa s after stirring for 3 days and then decreased. For
formic acid (4 mPa s), the viscosity increased to 480 mPa s
after 6 h of stirring before decreasing. The dissolution of the
milled wood in pyruvic and formic acids occurred more slowly
than in glyoxylic acid, which is in agreement with the changes
in optical transmittance. The viscosity of solutions of cellulose
derivatives depends on the solvent and several solute para-
meters such as the molecule size and substituent distribution
in chemical structures. Among these factors, differences in the
type of substituent and solvent as well as the degree of poly-
merisation significantly influenced the viscosity maximum.
The decrease in viscosity after reaching the maximum is
methyl formate (5) rather than the hemiacetal. In the case of
pyruvic acid, both methyl pyruvate (1) and methyl 2,2-dimeth-
oxypropanoate (2) were detected in the reaction mixture, con-
sistent with the occurrence of acetalisation and esterification.
Consequently, the high solubilisation efficiency of pyruvic and
glyoxylic acids may stem from the acetalisation and/or hemiace-
talisation of formyl and acyl groups as well as the esterification
of the carboxyl groups.
Although conventional high-temperature liquefaction
afforded black wood solutions, the milled wood dissolution
using pyruvic and glyoxylic acids as solubilising agents pro-
duced clear reddish brown solutions. This reduced blackening
of these lignin solutions results from the mild, room tempera-
ture conditions. Moreover, the conversion of hydroxyl groups
into the corresponding acetals and/or hemiacetals may reduce
their reactivity leading to hyperdegradation.
Analysis of the dissolution process
To clarify the dissolution process, the time course of the amounts explained primarily by a decrease in molecular weight upon
of filtrate and non-filterable materials was analysed. When formic
acid was used as a solvent, the dissolution required a short time,
hydrolysis after dissolution.
The optical transmittance of the 4.8% wood solution at
as estimated from the increased viscosity 3–6 h after dissolution 660 nm was measured using an optical density monitor to
(Table 3, Fig. 4). However, the soluble material could not be quan-
tified by filtration after 3–6 h because of high viscosity.
assess the transparency of each milled wood solution (Fig. 5).
The optical transmittance profile changed in agreement with
After 18 h, the viscosity of the solution decreased due to the viscosity increase. Immediately after adding milled wood
hydrolysis (Fig. 4), and the amounts of filtrate and non-filter-
to carboxylic acid, the optical transmittance of each acid solu-
tion was around zero because of the insolubility. The transmit-
tance increased slightly for the glyoxylic acid solution after 1 h,
but remained close to zero for formic acid and pyruvic acid
solutions. After 1 day, the formic acid solution became trans-
parent with increased transmittance; however, the transmit-
tance of the pyruvic acid solution increased more slowly
because of slower dissolution. The transmittance of the formic
acid solution decreased after 2 days, accompanied by a colour
change to black.
Table 3 Effects of solubilisation time of ball-milled wood in formic
acid on filtration
Non-filterable
fraction (%)
Time (h)a
Filtrate (%)
1
18
24
92
16
14
5
6
5
24
116 33
113
3
5
a Mixtures stirred for 3–6 h were not measurable because of high
viscosity.
With regard to colour, the transmittance of the short wave-
lengths below 550 nm was lower than that of longer wave-
This journal is © The Royal Society of Chemistry 2014
Green Chem., 2014, 16, 3569–3579 | 3573

View Article Online
Paper
Green Chemistry
Fig. 3 SEM images of (a) the filter before filtration, (b) the filter after filtration of the ball-milled formic acid solution and (c) original ball-milled
E. globulus.
Fig. 4 Time course of viscosity changes in dissolution of ball-milled
wood in pyruvic, glyoxylic and formic acids.
Fig. 6 Transmittance of ball-milled wood solutions measured by spec-
trophotometry in the wavelength scan mode: (a) formic acid, glyoxylic
acid and pyruvic acid solutions (14 days) and (b) dilute solutions in
formic acid (2 days).
These results show that the viscosity and transmittance
provide a good assessment of solubility and molecular weight
changes after mixing.
Fluid properties of acid solutions
Fig. 5 Time course of optical transmittance at 660 nm for the dissol-
ution of ball-milled wood in pyruvic, glyoxylic and formic acids.
The viscosity was measured at different viscometer rotational
speeds to assess the fluid properties of 4.8% wood solutions.
Although this analysis is typically conducted using rhe-
lengths in the visible region between 400 and 750 nm (Fig. 6a), ometers, viscometers can be used to determine fluid types in
resulting in reddish brown solutions. The spectrophotometer solutions. A fluid is classified as Newtonian if its viscosity
and the optical density monitor gave almost the same trans- remains constant for different rotational speeds or non-Newto-
mittance at 660 nm. Dilute wood solutions exhibited a higher nian if its viscosity varies with rotational speed.
transmittance (Fig. 6b) and a higher transparency by visual
observation (Fig. 2, 0.5% solution).
All acid solutions exhibited shear thinning, characteristic of
non-Newtonian fluids (Fig. 7). Pyruvic acid (3 days), glyoxylic
3574 | Green Chem., 2014, 16, 3569–3579
This journal is © The Royal Society of Chemistry 2014

View Article Online
Green Chemistry
Paper
Fig. 7 Viscosity of 4.8% ball-milled wood solutions in pyruvic, glyoxylic
and formic acids at various rotational speeds.
Fig. 9 SEC of ball-milled wood solutions. (a) Pyruvic acid (6 days), (b)
glyoxylic acid (24 h) and (c) formic acid (24 h). The elution was moni-
tored at 280 nm using a UV detector. The dashed lines represent the
SEC of the solvents.
Fig. 8 Shear stress of 4.8% ball-milled wood solutions in pyruvic,
glyoxylic and formic acids at various shear rates converted from Fig. 7.
acid (1 day) and formic acid solutions (6 h) presented non-
Newtonian indices (n) of 0.57, 0.72 and 0.41, respectively,
indicative of their pseudo-plasticity (non-Newtonian index
n < 1) (Fig. 8).
Polymer solutions behave as non-Newtonian fluids in many
cases. Cellulose solutions have shown a non-Newtonian behav-
iour,24,28,29 suggesting that the non-Newtonian nature of a
liquid is evidence for the dissolution of polymers in wood. In a
previous study, the n values ranged between 0.6 and 0.95 for
wood pulp (degrees of polymerization (DPs) 532, 837) or cotton
Fig. 10 SEC of ball-milled wood dissolved in formic acid after 4 h, 24 h
and 7 days.
linter (DP 495) solutions in paraformaldehyde–DMSO and posed of lignin and polysaccharides or their aggregates. In the
decreased (non-Newtonian) with concentration and DP.24 The low molecular weight region (104–103), the formic acid solu-
tion showed a peak at 103, which originates from lignin–carbo-
formic acid solution demonstrated a non-Newtonian behaviour
(n = 0.41) after 6 h. However, after 6 days, its non-Newtonian hydrate complexes (LCCs) or polysaccharide-free lignin. For
index n increased to 0.96, indicating that the solution changed pyruvic acid or glyoxylic acid solution, low molecular weight
into a Newtonian fluid. This change is consistent with the
compounds could not be evaluated because of an overlap with
depolymerisation of wood polysaccharides induced by hydro- solvent absorbance peaks. Reported molecular weights
lysis with formic acid over a long time period.
amounted to 107–104, 105–104 and 104–102 for ball-milled Red
Maple,4 LCC from spruce30 and acetylated milled wood lignin
Molecular weight distributions of dissolved wood com-
ponents were evaluated by SEC (Fig. 9). High molecular weight from Norway spruce,31 respectively. The obtained molecular
regions (Mn ≥ 105) presented a peak at 5.4 × 105, 9.2 × 105 and weight distributions indicate that the dissolved wood retained
4.1 × 106 for pyruvic acid, glyoxylic acid and formic acid solu-
high molecular mass fractions exceeding Mn values of 105–103,
tions, respectively. The formic acid solution showed another in accordance with the previous studies.
peak at 5.0 × 105. Moreover, all the solutions showed sub-
stances with molecular weights exceeding 107 in the void
The time course of molecular weight distributions was ana-
lysed at 4 h, 24 h and 7 days by SEC for the formic acid solu-
volume region, corresponding to the elution of matrices com- tion (Fig. 10). SEC profiles after 4 and 24 h were similar, except
This journal is © The Royal Society of Chemistry 2014
Green Chem., 2014, 16, 3569–3579 | 3575

View Article Online
Paper
Green Chemistry
Fig. 12 Photograph of the film formed after evaporation of the ball-
milled wood solution in formic acid.
Fig. 11 UV absorption of the dilute milled wood solution in formic acid.
an environmentally benign separation of wood components
without intensive modification of the native structures.
for the high molecular weight intensities (Mn > 105) after 24 h.
This indicates that the 24 h immersion caused a higher
amount of lignin and LCCs to dissolve in the acid. After 7
days, high molecular mass components (Mn > 107) disap-
peared. Moreover, the absorbance intensity decreased between
5 × 105 and 5 × 106, concomitant with the formation of a new
peak at 1.3 × 105. This suggests that the hydrolysis of cell wall
polysaccharides induced by formic acid broke the lignin and
polysaccharide matrices, separating the non-covalently linked
components. The depolymerisation of LCCs may also explain
this shift.
UV-visible absorption spectra of milled wood solutions were
measured. Because pyruvic acid or glyoxylic acid showed
strong absorption below 400 nm, the absorption spectra were
recorded for formic acid solutions, which absorb UV light
below 260 nm (Fig. 11). Spectra obtained at two different con-
centrations of 0.048% and 0.024% showed a maximum at
280 nm. Although furfural derivatives present an absorption
peak at 280 nm, this peak is ascribed to lignin because carbo-
hydrates are stable at ambient temperature. The reported
absorption coefficient of lignin at 280 nm amounted to
20 L g−1 cm−1 (acetyl bromide, soft wood and hard wood),32
19 L g−1 cm−1 (3% H2SO4, spruce SEW)33 and 17.5 L g−1 cm−1
(lignosulphonate).34 Using an absorption coefficient of
19 L g−1 cm−1 and a specific gravity of 1.22 for formic acid,
the lignin content of milled wood was calculated to be ca.
15%–16%. These results confirm that the milled wood was dis-
solved in the acid.
The separation process relies on the addition of a water-
immiscible solvent that is stable against the acid used, justify-
ing the choice of 2-MeTHF, a furan derivative available from
biomass. The effect of 2-MeTHF and water addition to the
wood solution was analysed in detail at various relative compo-
sitions (Fig. 13). The balled-milled wood solution remained
homogeneous upon addition of water and 2-MeTHF at weight
ratios below 10% (I). However, at weight ratios exceeding
50 : 50 (w/w), precipitates formed in most cases upon addition
of water and 2-MeTHF. Although 2-MeTHF and water are
immiscible, mixing the solvents with acids resulted in homo-
geneous solutions at certain compositions (Fig. 13 (II)). When
acid, water and 2-MeTHF were mixed at a weight ratio of
around 20 : 40 : 40, water and 2-MeTHF layers were separated
and precipitates formed (III) (Fig. 14).
Fig. 13 Phase diagrams of a ternary mixtures of ball-milled wood solu-
tion in acid, 2-MeTHF and water.
Transparent film formation
A transparent brown film was obtained by evaporation of the
formic acid solution without fractionation (Fig. 12). This trans-
parency confirms that the ball-milled wood was actually dis-
solved in formic acid.
Precipitation in ether solvents and fractionation
The separation of wood components from wood solution is
important for their conversion and utilization. Because the dis-
solution of whole wood was performed at room temperature,
Fig. 14 Photograph showing three phase-separated layers (III) formed
by addition of 2-MeTHF and water to the ball-milled wood solution in
the fractionation of the solutions by precipitation will lead to formic acid.
3576 | Green Chem., 2014, 16, 3569–3579
This journal is © The Royal Society of Chemistry 2014

View Article Online
Green Chemistry
Paper
Table 4 Yields (%) of fractions isolated from a ball-milled wood solution in formic acid using 2-MeTHF and water
Fr. 1
Fr. 2
Fr. 3
Time of separation
after dissolution
Original
milled wood
No.
Precipitate
Aqueous layer
2-MeTHF layer
1
2
1 day
8 months
100
100
73
64
19
37
18
33
In the case of mixture III, the 2-MeTHF layer was isolated (Fr. 1) and a small portion of cellulose-derived substances was
and the aqueous suspension was filtered to provide a whitish, detected in the 2-MeTHF layer (Fr. 3). We also fractionated the
light brown solid. Water-miscible dioxane and DMSO have solution after 3 and 7 days. The components were almost the
been used as lignin solvents but have caused coloured sub- same as those of the 1 day sample. However, when the wood
stances to adhere to solids during fractionation. A water- solution was fractionated 8 months after dissolution (no. 2), a
immiscible solvent can dissolve these substances in the small amount of lignin-derived substances was present in the
organic phase to separate them from the solid. Although a precipitate and the 2-MeTHF layer contained an increased
poor solvent for these coloured substances, 2-MeTHF dissolved amount of cellulose-derived substances. This demonstrates
them when mixed with an acid.
that a long immersion decomposed lignin- and cellulose-
The fractionation of the ball-milled wood solution in derived substances and solubilised them by acid-enabled
formic acid provided three portions, Fr. 1 (precipitation), Fr. 2 chemical modification. The water layer (Fr. 2) contained hemi-
(aqueous layer) and Fr. 3 (2-MeTHF layer), using solvent cellulose-derived furfural (compound no. 2), cellulose-derived
system III (formic acid, water and 2-MeTHF, 20 : 40 : 40, w/w/w) levoglucosan and lignin-derived substances. Low molecular
(Table 4). The fractionation was conducted 1 day (no. 1) and weight compounds were assumed to dissolve in water.
8 months (no. 2) after dissolution. The total recovered weight
The compositional analysis revealed that the precipitate
exceeded 100% because a portion of formic acid was bound to obtained after 1 day contained a higher holocellulose amount
the wood components. A long immersion period increased the and a lower lignin amount than the original wood (Table 5),
total weight because the acid was incorporated into the wood consistent with Py–GC–MS results. Neutral sugar analysis indi-
components.
cated that this precipitate contained a higher glucose concen-
tration and a lower xylose concentration than the original
wood. Therefore, the precipitate obtained from the wood solu-
tion using 2-MeTHF and water mainly consists of cellulose as
well as hemicelluloses and lignin at lower concentrations than
in the original wood.
The cellulosic precipitate (Fr. 1) obtained from the formic
acid solution by 2-MeTHF–water fractionation were examined by
SEM. The precipitate (Fr. 1) comprised a porous material con-
taining small pores with diameters of up to ca. 10 μm (Fig. 16).
Precipitates from cellulose–ionic liquid solution have been
reported to have porous structures.35 These morphological fea-
tures confirm that the ball-milled wood was actually dissolved
in formic acid and then precipitated by addition of 2-MeTHF
Analysis of wood fractions
Fractions isolated by addition of 2-MeTHF and water to the
ball-milled wood solution in formic acid were characterised by
Py–GC–MS (Fig. 15) and compositional analysis. Area ratios of
compounds detected by Py–GC–MS are shown in S1. Levoglu-
cosan (compound no. 30), a product from cellulose pyrolysis,
was the main component of the precipitate (Fr. 1). The 2-
MeTHF layer (Fr. 3) primarily contained lignin-derived sub-
stances (compound no. 21 etc.). When the fractionation was
performed 1 day after dissolution (no. 1), compounds derived
from lignin and hemicelluloses were found in the precipitate
Table 5 Compositions (wt%) of the ball-milled wood and solid fraction
obtained by 2-MeTHF–water fractionation of the formic acid solution at
room temperature
Ball-milled
E. globulus
No. 1–Fr. 1a
Holocellulose
Klason lignin
Acid soluble lignin
Carbohydrates
63.1
20.3
2.3
0.0
13.9
0.0
77.7
14.0
2.8
0.0
7.9
0.0
0.0
58.7
Ara
Xyl
Man
Gal
Glc
0.0
45.6
Fig. 15 Py–GC–MS total ion chromatograms of each portion obtained
by 2-MeTHF–water fractionation of a ball-milled wood solution in
formic acid after 1 day (no. 1).
a Precipitate isolated from the ball-milled wood solution in formic acid
using 2-MeTHF and water (Fig. 1).
This journal is © The Royal Society of Chemistry 2014
Green Chem., 2014, 16, 3569–3579 | 3577

View Article Online
Paper
Green Chemistry
fatty acid, formic acid, has been used in various chemical
industries for its economic advantages. Moreover, this solvent
has a low boiling point that facilitates its recovery from reactive
media containing wood and its degradation products. Here, a
method was also developed to fractionate the wood solution
using the water-immiscible solvent 2-MeTHF. Because the dis-
solution and separation were achieved at room temperature
with simple compounds derivable from biomass, our results
are expected to lead to self-sufficient biorefinery systems that
transform wood components to chemicals, fuels and func-
tional polymers.
Fig. 16 SEM image of the precipitate (Fr. 1) obtained by 2-MeTHF–
water fractionation of the wood solution in formic acid.
Acknowledgements
This work was supported by a grant from the Ministry of Edu-
cation, Culture, Sports, Science and Technology of Japan. We
would like to thank Ms R. Mizuno for technical support for
the carbohydrate analysis.
and water. A transparent film exhibiting a high lignin content
was obtained by evaporation of the wood solution and a porous
solid was obtained by reprecipitation during the fractionation.
This demonstrates that the wood solution can produce different
solid forms by various post-treatment processes.
The dissolution of whole wood in an ionic liquid and other
solvents such as DMSO has been studied extensively over this
decade. However, its dissolution and fractionation using
simple solvents without any additives at room temperature
have been difficult to achieve. Our method offers a process
that solubilizes and separates whole wood using simple α-keto
acids or aldehydic carboxylic acids, in combination with pre-
cipitation using 2-MeTHF and water at room temperature. The
water-immiscible solvent used for separation can be recovered
from the wood by simple partitioning. This α-keto acid and
aldehydic carboxylic acid system is expected to find wide appli-
cations in solution reactions of wood components and liquid
fuels. It may also be implemented in the production of
polymer supports for adsorption and gradual release of
organic compounds by chemical and enzymatic modification.
Notes and references
1 F. Lu and J. Ralph, Plant J., 2003, 35, 535–544.
2 Z. Wang, T. Yokoyama, H.-M. Chang and Y. Matsumoto, J.
Agric. Food Chem., 2009, 57, 6167–6170.
3 H. Kim and J. Ralph, Org. Biomol. Chem., 2010, 8, 576–591.
4 M. Lawoko and A. R. P. van Heiningen, J. Wood Chem.
Technol., 2011, 31, 183–203.
5 P. Maki-Arvela, I. Anugwom, P. Virtanen, R. Sjoholm and
J. P. Mikkola, Ind. Crops Prod., 2010, 32, 175–201.
6 M. Zhang, W. Qi, R. Liu, R. Su, S. Wu and Z. He, Biomass
Bioenergy, 2010, 34, 525–532.
7 S. Zhou, L. Liu, B. Wang, F. Xu and R. Sun, Process Biochem,
2012, 47, 1799–1806.
8 J. J. Villaverde, J. Li, M. Ek, P. Ligero and A. de Vega,
J. Agric. Food Chem., 2009, 57, 6262–6270.
9 X. Zhao and D. Liu, Bioresour. Technol., 2012, 117, 25–32.
10 L. Kupiainen, J. Ahola and J. Tanskanen, Bioresour.
Technol., 2012, 116, 29–35.
Conclusions
The dissolution of whole wood has attracted major research 11 J. Hiltunen, WO Pat, 2012140328-A1, 2012.
efforts because of its impact on wood component separation 12 W. Ostwald, Kolloid Z., 1925, 36, 99–117.
and the biological and chemical modification of these com- 13 A. De Waele, J. Oil Colour Chem. Assoc., 1923, 6, 33–69.
ponents in solution. The wood solution can be used also as a 14 F. J. F. Lopes, F. O. Silverio, D. C. F. Baffa, M. E. Loureiro
renewable liquid fuel if the solvent is cheap enough for energy
recovery by burning. In this study, ball-milled wood dissolved
and M. H. P. Barbosa, J. Wood Chem. Technol., 2011, 31,
309–323.
in pyruvic, glyoxylic and formic acids at room temperature. 15 O. Faix, D. Meier and I. Fortmann, Holz Roh- Werkst., 1990,
The simplest α-keto acid, pyruvic acid, is an important inter- 48, 351–354.
mediate in several metabolic pathways such as glycolysis, glu- 16 E. R. E. van der Hage, M. M. Mulder and J. J. Boon, J. Anal.
coneogenesis, TCA cycle and amino acid and fatty acid Appl. Pyrolysis, 1993, 25, 149–183.
biosyntheses. The use of this safe and highly biodegradable 17 J. Ralph and R. D. Hatfield, J. Agric. Food Chem., 1991, 39,
α-keto acid as a wood solvent opens the door to new appli- 1426–1437.
cations related to lignocellulosic biorefinery processes. The 18 L. E. Wise, M. Murphy and A. A. D’Addieco, Pap. Trade J.,
simplest α-formyl acid, glyoxylic acid, is the key metabolite in 1946, 122, 35–43.
the glyoxylate path in microorganisms and plants. Biological 19 B. Swan, Sven. Papperstidn., 1965, 68, 791–795.
production of this acid can be achieved by blocking the trans- 20 H. Mikami and Y. Ishida, Bunseki Kagaku, 1983, 32, E207–
formation of glyoxylate into malate in this path. The simplest
E210.
3578 | Green Chem., 2014, 16, 3569–3579
This journal is © The Royal Society of Chemistry 2014

View Article Online
Green Chemistry
Paper
21 CRC Handbook of Chemistry and Physics, CRC Press, 93rd 28 N. Tamai, D. Tatsumi and T. Matsumoto, Biomacromole-
edn, 2012. cules, 2004, 5, 422–432.
22 M. Hasegawa, A. Isogai, F. Onabe and M. Usuda, J. Appl. 29 Y. Lv, J. Wu, J. Zhang, Y. Niu, C.-Y. Liu, J. He and J. Zhang,
Polym. Sci., 1992, 45, 1857–1863. Polymer, 2012, 53, 2524–2531.
23 D. C. Johnson, M. D. Nicholson and F. C. Haigh, Appl. 30 X. Du, G. Gellerstedt and J. Li, Plant J., 2013, 74, 328–338.
Polym. Symp., 1976, 28, 931–943.
24 C. He and Q. Wang, Polym. Adv. Technol., 1999, 10, 487–
492.
25 K. H. Meyer, M. Studer and A. J. A. van der Wyk, Monatsh.
Chem., 1950, 81, 151–154.
31 J. Asikkala, T. Tamminen and D. S. Argyropoulos, J. Agric.
Food Chem., 2012, 60, 8968–8973.
32 K. Iiyama and A. F. A. Wallis, Wood Sci. Technol., 1988, 22,
271–280.
33 M. Iakovlev and A. van Heiningen, J. Wood Chem. Technol.,
2011, 31, 233–249.
34 E. Sjostrom and P. Haglund, Tappi, 1964, 47, 286–291.
26 K. Kamide, K. Okajima, T. Matsui and S. Manabe, Polym. J.,
1980, 12, 521–534.
27 A. Isogai, A. Ishizu and J. Nakano, Cellul. Chem. Technol., 35 M. Hirano, Y. Suzuki and A. Kanda, JP Pat, 2012057137A,
1983, 17, 123–131.
2012.
This journal is © The Royal Society of Chemistry 2014
Green Chem., 2014, 16, 3569–3579 | 3579