Carbohydrate Polymers 83 (2011) 17661774

Contents lists available at ScienceDirect

Carbohydrate Polymers
journal homepage: www.elsevier.com/locate/carbpol

Aerocellulose from celluloseionic liquid solutions: Preparation, properties and

comparison with celluloseNaOH and celluloseNMMO routes
Romain Sescousse, Roxane Gavillon 1 , Tatiana Budtova
Mines ParisTech, Centre de Mise en Forme des Matriaux - CEMEF2 , UMR CNRS/Ecole des Mines de Paris 7635, BP 207, 06904 Sophia-Antipolis, France

a r t i c l e

i n f o

Article history:
Received 31 August 2010
Received in revised form 8 October 2010
Accepted 19 October 2010
Available online 27 October 2010
Keywords:
Cellulose
Imidazolium-based ionic liquids
Regeneration kinetics
Diffusion
CelluloseNaOHwater solution
Mechanical properties
Porosity
Morphology
Aerogel

a b s t r a c t
Ultra-light and highly porous cellulose material, aerocellulose, is prepared via cellulose dissolution
in imidazolium-based ionic liquids, 1-ethyl-3-methylimidazolium acetate (EMIMAc) and 1-butyl-3methylimidazolium chloride (BMIMCl), followed by regeneration and drying in supercritical CO2
conditions. Regeneration kinetics of cellulose in water is studied. The diffusion coefcients of EMIMAc
and BMIMCl were obtained using a Fickian approach; they are analysed as a function of cellulose concentration and compared with the previously obtained values for NaOH and N-methyl-morpholine N-oxide
(NMMO). Density, morphology and porosity of aerocellulose from celluloseionic liquid solutions are
investigated and compared with the corresponding values from NaOH and NMMO routes. The mechanical properties of aerocelluloses under compression from all three routes are studied and correlated with
the moduli obtained from mercury porosimetry. For the ionic liquid and NaOH routes the Youngs modulus scales aerocellulose density with the exponent close to three, a value typical for silica aerogels.

1. Introduction
The use of cellulose, an inexhaustible natural polymer, is reconsidered in our days due to the discoveries of new non-toxic cellulose
solvents and the possibilities of making materials with various
functionalities. Ultra-light and highly porous cellulose is a new
very promising material offering a wide range of potential applications, from bio-medical and cosmetics (delivery systems, scaffolds)
to insulation and electro-chemical (when pyrolysed).
Two main ways of making ultra-light cellulose are known. One
is inspired by the preparation of aerogels. Cellulose is dissolved in
a direct solvent (such as N-methyl-morpholine N-oxide (NMMO)
monohydrate (Innerlohinger, Weber, & Kraft, 2006; Liebner,
Potthast, Rosenau, Haimer, & Wendland, 2008), 8% NaOHwater
(Gavillon & Budtova, 2008; Sescousse & Budtova, 2009), LiCl/DMAc
(Duchemin, Staiger, Ticker, & Newman, 2010), calcium thiocyanate (Jin, Nishiyama, Wada, & Kuga, 2004) or ionic liquid

Corresponding author. Tel.: +33 4 93 95 74 70; fax: +33 4 92 38 97 52.

E-mail address: [email protected] (T. Budtova).
1
Present address: LOreal, 188-200, rue Paul Hochart, 94550 Chevilly-Larue,
France.
2
Member of the European Polysaccharide Network of Excellence (EPNOE),
www.epnoe.eu.
0144-8617/$ see front matter 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.carbpol.2010.10.043

2010 Elsevier Ltd. All rights reserved.

(Aaltonen & Jauhiainen, 2009; Deng, Zhou, Du, Kasteren, & Wang,
2009; Tsioptsias, Stefopoulos, Kokkinomalis, Papapdoupoulou, &
Panayiotou, 2008), regenerated (or coagulated) in a non-solvent
(water, alcohols) and then dried in a special way that prevents
pores collapse, i.e. either via freeze-drying (Deng et al., 2009;
Duchemin et al., 2010; Jin et al., 2004), or under CO2 supercritical conditions (Aaltonen & Jauhiainen, 2009; Gavillon & Budtova,
2008; Innerlohinger et al., 2006; Liebner et al., 2008; Sescousse &
Budtova, 2009; Tsioptsias et al., 2008). In all cases cited no chemical cross-linking is used to stabilize the cellulose network;
it is formed during cellulose regeneration either from a solution
(Aaltonen & Jauhiainen, 2009; Deng et al., 2009; Duchemin et al.,
2010; Innerlohinger et al., 2006; Jin et al., 2004; Liebner et al.,
2008) or from a physical gel (Gavillon & Budtova, 2008; Sescousse &
Budtova, 2009). These cellulose II aerogel-like materials, referred to
as aerocellulose (Duchemin et al., 2010; Gavillon & Budtova, 2008;
Innerlohinger et al., 2006; Sescousse & Budtova, 2009), have wide
pore size distribution, from tens of nanometers to several microns,
and a high specic surface area of several hundreds of m2 /g. The
density of aerocellulose depends on the initial cellulose concentration in solution and on the control of drying. Foaming agents can
be added to increase aerocellulose porosity (Gavillon & Budtova,
2008).
The second way of making ultra-light porous cellulose is to
use cellulose nanobers, which can be either bacterial cellulose

R. Sescousse et al. / Carbohydrate Polymers 83 (2011) 17661774

(Liebner et al., 2010; Maeda, Nakajima, Hagwara, Sawaguchi, &

Yano, 2006) or microbrillated cellulose prepared via native cellulose mechanical disintegration and enzymatic treatment (Paakko
et al., 2008). In both cases the starting material is a network
of cellulose I nanobers lled with water. Water can then be
extracted using the same approaches as described for aerocellulose, i.e. freeze-drying or drying in supercritical CO2 . The resulting
material is thus also a network of cellulose I nanobers. It looks
similar to aerocellulose but with higher porosities and lower densities because of lower initial cellulose concentrations and absence
of contraction during drying due to higher skeleton crystallinity
and molecular weight. However, the preparation of porous cellulose from the nanober route either does not allow the variation
of cellulose concentrations (case of bacterial cellulose), or requires
high energy consumption (in the case of microbrillated cellulose).
In general, aerocellulose can be prepared from any cellulose solution. Many cellulose solvents are known, but most
are either toxic, or with some drawback effects or, like (79%)
NaOHwaterurea (or thiourea) mixtures, do not allow the dissolution of cellulose of high molecular weights and concentrations
(Egal, Budtova, & Navard, 2007). Room-temperature imidazoliumbased ionic liquids (IL), such as 1-ethyl-3-methylimidazolium
acetate
(EMIMAc),
1-butyl-3-methylimidazolium
chloride
(BMIMCl) and 1-allyl-3-methylimidazolium chloride, are now
accepted to be direct cellulose solvents (Swatloski, Spear, Holbrey,
& Rogers, 2002; Zhang, Wu, Zhang, & He, 2005). Processing of
cellulose from ionic liquids, which occurs without a derivatisation
step, is very attractive: the dissolution procedure is simple and
ionic liquids are nonvolatile and have a high thermal stability making them so called green solvents. Cellulose of high molecular
weight and at rather high concentrations, around 1520%, can be
dissolved in the above mentioned ionic liquids. Various cellulose
objects such as bers (Cai, Zhang, Guo, Shao, & Hu, 2010; Quan,
Kang, & Chin, 2010; Wendler, Kosan, Kreig, & Meister, 2009), lms
(Turner, Spear, Holbrey, & Rogers, 2004), composites (Zhao et al.,
2009) and beads (Lin, Zhan, Liu, Fu, & Lucia, 2009) were prepared
from celluloseionic liquid solutions.
In this work we present an extensive study of the preparation and properties of new aerocellulose prepared from
celluloseEMIMAc and celluloseBMIMCl solutions, via dissolution/regeneration route and drying under supercritical CO2 . Carbon
dioxide is one of the most commonly used supercritical uids
in polymer chemistry and technology due to its low cost, noninammability and low critical point temperature and pressure.
Literature reports the preparation conditions and nal morphology/porosity of aerocellulose prepared from either BMIMCl
(Aaltonen & Jauhiainen, 2009; Deng et al., 2009) or AMIMCl
(Tsioptsias et al., 2008). We start with a study of cellulose regeneration kinetics from ionic liquid solutions. To the best of our
knowledge there is no publication describing this unavoidable step
in cellulose processing from ionic liquids, whatever is the nal
material. Regeneration kinetics from ionic liquids is characterized
by EMIMAc and BMIMCl diffusion into regenerating bath (in our
case, water); the results obtained are compared with the ones
known for cellulose regeneration from celluloseNMMO monohydrate and cellulose NaOHwater systems (Gavillon & Budtova,
2007). Then we discuss aerocellulose density, porosity and morphology from celluloseEMIMAc and again compare with the
same from NMMO and NaOH routes. Finally, a systematic study
of the mechanical properties of aerocelluloses made from EMIMAc, NMMO monohydrate and 8% NaOHwater were performed.
The approaches developed for aerogels and foams were applied
to analyse aerocellulose mechanical properties. Young modulus,
compressive yield strength and absorption energy were obtained
and studied as a function of aerocellulose density and preparation
route.

1767

2. Experimental
2.1. Materials
Avicel PH-101 microcrystalline cellulose (cellulose in the following) with degree of polymerization DP = 180, was purchased
from SigmaAldrich and used for most of the studies (ionic liquid and NaOH routes). Other celluloses, Solucell from Lenzing AG,
Austria, of DP = 950 (Solucell in the following) and cotton also of
DP = 950 were used for NMMO route.
The ionic liquids EMIMAc and BMIMCl were used as received
from Fluka. NaOH in pellets (97% purity) and acetone (98% purity)
were purchased from VWR. Alkyl polyglycoside surfactant Simulsol
SL8 was from Seppic Inc., Faireld, USA. Distilled water was used
for preparing celluloseNaOHwater solutions and for regenerating bath. CO2 for drying in supercritical conditions was supplied by
Air Liquide with a purity of 99.9%.
2.2. Methods
2.2.1. Preparation of samples: cellulose solutions, samples for
regeneration studies and making dry aerocellulose
Cellulose was dried at 50 C in vacuum prior to use. Cellulose8%
NaOHwater solutions were prepared as described elsewhere (Egal
et al., 2007; Gavillon & Budtova, 2008; Sescousse & Budtova, 2009).
Briey, an aqueous solution of 12 wt% NaOH was cooled down to
6 C. Cellulose was swollen in distilled water and kept at 5 C. 12%
NaOHwater and swollen-in-water-cellulose were mixed at 6 C
with a stirring rate of 1000 rpm for 2 h in certain proportions to
obtain various cellulose concentrations (wt%) in 8% NaOHwater.
Solutions were stored at 5 C to avoid aging.
CelluloseIL solutions were prepared by mixing and stirring cellulose and solvent in a sealed reaction vessel at 80 C for about 48 h
to ensure complete dissolution. Cellulose solutions were stored at
room temperature and protected against moisture absorption.
In order to study cellulose regeneration kinetics from
celluloseIL solutions and to use the Fickian approach which
was proved to work well for celluloseNaOHwater and
celluloseNMMO monohydrate solutions (Gavillon & Budtova,
2007; Sescousse & Budtova, 2009), the sample must have well
dened dimensions. An innite plane approximation was selected
because of the ease of sample preparation (Gavillon & Budtova,
2007; Sescousse & Budtova, 2009). Within this approximation sample thickness must not be larger than 1/10 of its diameter in case
the sample is a disk. A special cylindrical mould of 3 cm diameter
and thickness 2l = 3 mm was constructed. The walls of the mould
were made of a stainless steel grid with the holes of 50 m 50 m.
The solution was pored into the mould, closed with the cover
made of the same grid and placed into regenerating bath. Because
celluloseEMIMAc and celluloseBMIMCl solutions are of a rather
high viscosity (see, for example, Gericke, Schlufter, Liebert, Heinze,
& Budtova, 2009; Sescousse, Le, Ries, & Budtova, 2010), they were
not leaking through the holes. This set-up allowed free solvent/nonsolvent diffusion through the mould walls and kept sample size
constant during regeneration.
To prepare dry aerocellulose samples from cellulose8%
NaOHwater and celluloseIL, the procedure described in
Gavillon and Budtova (2008), Sescousse and Budtova (2009)
was used, i.e. drying in supercritical CO2 conditions. For the
celluloseNaOHwater route, solution gelation was performed
prior regeneration: the solutions were kept at 50 C for 2 h (Roy,
Budtova, & Navard, 2003). For celluloseIL route, the same mould,
as described above, was used. Solutions and gels were then placed
in water regenerating bath which was regularly changed until all
solvent is washed out (control of pH for NaOH route and refractive
index for ionic liquids, see next section). The samples were then

1768

R. Sescousse et al. / Carbohydrate Polymers 83 (2011) 17661774

lulose was regenerating, small amounts of bath liquid were taken

at different times. Because the concentrations of IL in regenerating
bath were very low, it was not possible to use directly the calibration dependence. First, water from the probes was evaporated
in a controlled way in order to increase IL concentration. Then the
refractive index was measured and IL concentration in the bath was
calculated with the help of the calibration dependence and knowing the weights of the probes before and after evaporation. These
data were used to describe the kinetics of IL release from the sample (see example for cellulose regeneration from celluloseEMIMAc
solution in Fig. 1). Each experiment was repeated 35 times and
the mean value of the diffusion coefcient was calculated as will
be described in Section 3. The error in concentration determination
is within 5% and for the resulting diffusion coefcient it was about
15%.

C EMIMAc, %
0%

5%
10%

2
n

y = 0.0018x + 1.3336

1.4

1.35
y = 0.0016x + 1.3336

CIL, %
1.3
0

10

20

30

40

0
0

100

200

300

time, min
Fig. 1. Example of EMIMAc concentration increase in the regenerating bath in time
during cellulose regeneration from celluloseEMIMAc solutions at 22 C. Cellulose
initial concentrations shown are 0, 5 and 10%. Dashed lines are given to guide the
eye. Inset: Calibration curves for EMIMAc (dark points) and BMIMCl (open points)
aqueous solutions: refractive index vs IL concentration at 22 C.

washed in acetone to remove water which is not compatible with

CO2 . Swollen-in-acetone samples were dried under supercritical
CO2 as described earlier (1 L autoclave, 80 bar, 35 C) (Gavillon &
Budtova, 2008; Sescousse & Budtova, 2009). After depressurisation
(4 bar per hour at 37 C), aerocellulose samples were extracted and
analysed.
In some cases of NaOH route, the surfactant Simulsol was used
as a foaming agent to increase aerocellulose porosity, as described
in Gavillon and Budtova (2008). Simulsol was added at various concentrations (0.1%, 0.5%, and 1% in weight) to the ready cellulose8%
NaOHwater solution. The mixture was stirred at 1000 rpm for
5 min at +5 C, leading to the formation of air bubbles. The foamed
solution was then immediately gelled at 50 C for 2 h in order to
freeze the bubbles. Simulsol dissolved in water during the regeneration step.
Aerocellulose prepared from Solucell and cotton dissolved in
NMMO monohydrate was used to perform compression experiments to determine the mechanical properties and compare with
NaOH and IL routes. The samples were prepared in the R&D laboratory of Lenzing AG, Austria. SolucellNMMO monohydrate and
cottonNMMO monohydrate solutions were poured into cylindrical moulds, cooled down to room temperature and regenerated in
water at ambient temperature. Water was exchanged with acetone
of analytical grade. The samples were dried in CO2 under supercritical conditions in Natex, Ternitz, Austria.
2.2.2. Cellulose regeneration kinetics from ionic liquid solutions
The kinetics of cellulose regeneration from solutions in ionic
liquids was studied as follows. The mould with celluloseIL solution of a given weight and volume was placed in the regenerating
bath (water) of a xed temperature. The proportion between sample/bath weights was about 1/30 with the bath volume of 100 mL.
Magnetic stirrer with a gentle stirring rate (300 rpm) was used
for concentration homogenization.
The concentration of ionic liquid in water was measured using
an Abbe refractometer. First, the calibration dependence was
obtained by measuring the refractive index n of ILwater solutions
as a function of known IL concentration CIL (see inset of Fig. 1).
Calibrations were carried out at each bath temperature. When cel-

2.2.3. Mechanical measurements

Compression mechanical measurements were performed on dry
aerocelluloses from IL, NaOH and NMMO routes. The samples were
cylinders with the height being 1.5 times larger than the diameter.
The universal testing machine, Dartec Testwell was used for
the uniaxial compression tests. The compressive machine was
equipped with a 50 kN load cell. The cylindrical sample was inserted
between two steel circular compression plates. A deformation rate
of 0.0027 s1 was applied on the top surface of each specimen.
All measurements were made under ambient conditions at about
2022 C and 5060% relative humidity.
2.2.4. Porosity measurements
Porosity of aerocellulose was measured in SAFT, Bordeaux,
France, using mercury intrusion with Micromeritics Autopore IV
porosimeter and nitrogen absorption with Belsorp Mini II. The
details on data treatment will be given in Section 3.
The density of all aerocelluloses was determined by measuring
the weight and dimensions of the samples.
2.2.5. Scanning electron microscopy
Scanning electron microscopy (SEM) observations of aerocellulose morphology were performed using a PHILIPS XL30 ESEM at the
acceleration voltage of 1015 kV. Thin layers of gold were deposited
by sputtering onto the surface of the cross sections.
3. Results and discussion
3.1. Cellulose regeneration kinetics from celluloseIL solutions
As demonstrated in Sescousse and Budtova (2009), Gavillon
and Budtova (2007), Biganska and Navard (2005), regeneration of
cellulose from celluloseNaOHwater and celluloseNMMO monohydrate solutions is a diffusion-controlled process and can be
described by a Fickian approach. The same is to be expected for the
description of cellulose regeneration from celluloseIL solutions:
indeed, cumulated
amount of solvent M(t) released in time t as a

function of t is approximated with a straight line up to 80% of

M(t), which indicates a diffusion-controlled process.
The method used for the monitoring the amount of IL released
into the bath in time allows determination of a mean value of diffusion coefcient which does not depend on the position inside the
cellulose sample and is supposed to be constant. Several simplications can be made for the calculation of IL diffusion coefcient;
they are presented in Table 1. A relative mass of substance released
in time M(t)/M (M being maximal amount of substance in the bath
which corresponds to the mass of IL in cellulose sample) was plotted against t/l2 (l being half thickness of the sample disk) for each
bath temperature and various cellulose concentration (see examples in Fig. 2). Diffusion coefcients of EMIMAc and BMIMCl were

R. Sescousse et al. / Carbohydrate Polymers 83 (2011) 17661774

Table 1
Approximations used to calculate the diffusion coefcient of a substance released
from an innite plane.
Early time: 0
Half-time:

M(t)
M

M(t)
M

= 1/2

Late-time: 0.4

1.2

0.4

M(t)
M

 Dt 1/2

M(t)
M

=4

D=

(t/l2 )

M(t)
M

=1

l2

0.049
atM(t)/M=0.5

8
2

exp

 2 Dt 
l2

M(t)/M

0.8

0.6

0.4

0.2

0
0

2000

4000

6000

t/l

Fig. 2. Cumulated mass evolution of EMIMAc during regeneration of cellulose from

5% celluloseEMIMAc solution at 22 C (dark points) and at 54 C (open points).
Solid and dashed lines correspond to late time and early time approximations,
respectively.

then calculated for each approximation and the average value was
taken.
The diffusion coefcients of EMIMAc and BMIMCl at 22 C are
shown in Fig. 3 as a function of cellulose concentration. It was not
possible to obtain DBMIMCl at Ccell = 0% because of extremely high
BMIMCl hygroscopicity and thus difculties in sample preparation. The measured DEMIMAc in water at Ccell = 0 was compared with
DEMIM using data available in literature for the self-diffusion of
EMIM ions at 27 C (Liu, Sale, Holmes, Simmons, & Singh, 2010)
0.001
0

10

15

Ccell, %

EMIMAc
BMIMCl
EMIMAc (Liu et al, 2010)
EMIMAc (Seki et al, 2010)

0.0001

D, mm2/s
0.00001
Fig. 3. . Diffusion coefcients of EMIMAc and BMIMCl as a function of cellulose
concentration during cellulose regeneration in water at 22 C. The line is given to
guide the eye. DEMIM at Ccell = 0 calculated using data from Liu et al. (2010) and Seki
et al. (2010) are also shown.

1769

and 30 C (Seki et al., 2010). First, the radius R of the ion was calculated using Einstein formula DEMIM (T ) = kB T/6(T )R where
kB is Boltzmann constant, T is temperature and  is the viscosity of the medium (EMIMAc) which was taken from Gericke et al.
(2009) at the corresponding temperatures, 27 C and 30 C. Then
the radius obtained was used to calculate the diffusion coefcient
of EMIM in water taking water viscosity of 103 Pa s. For such a
rough estimation, and also considering some difference in temperature (experimental vs used in literature), the agreement between
measured and estimated EMIMAc diffusivity in water is very good,
3 104 mm2 /s (experimental value obtained in this work at 22 C)
vs 4 104 at 27 C calculated with self-diffusion data from Liu et al.
(2010) and 7 104 at 30 C from Seki et al. (2010).
The values of DEMIMAc and DBMIMCl at the same cellulose concentration practically coincide within experimental error which was
expected as far as both solvent are very similar. At a given cellulose
concentration, DEMIMAc and DBMIMCl are four to ve times lower than
DNaOH (Gavillon & Budtova, 2007; Sescousse & Budtova, 2009) and
about twice lower than DNMMO (Gavillon & Budtova, 2007) because
of the larger size of the diffusing entity (ionic liquid). The presence of low amount of cellulose, 1%, strongly decreases IL diffusion
coefcient as compared with Ccell = 0. Further increase of cellulose
concentration leads to a very small decrease in DEMIMAc and DBMIMCl
(Fig. 3). Such a weak dependence of solvent diffusion coefcient
on polymer concentration is rather unusual. The same trend was
observed for DNMMO during cellulose regeneration while DNaOH
was found to be concentration-dependent (Gavillon & Budtova,
2007). It was shown that free volume and hydrodynamic models
seem to describe well the dependence of NaOH diffusion coefcient on cellulose concentration during regeneration, contrary
to the case of celluloseNMMO monohydrate system (Gavillon &
Budtova, 2007). There are several reasons for this different concentration behaviour, DNMMO and DIL vs DNaOH . First, celluloseNMMO
monohydrate or IL are real solutions and celluloseNaOH is a gel
which is formed due to the preferential cellulosecellulose and not
cellulosesolvent interactions at room temperature. The gelation
of celluloseNaOHwater solution with the increase of time and
temperature is accompanied by a micro-phase separation. At the
beginning of regeneration cellulose in 8% NaOHwater is already
partly coagulated, it is a sort of a network where NaOH can freely
diffuse, while NMMO or ionic liquid are diffusing in a polymer
solution which is undergoing a phase separation when placed in a
non-solvent. Second, the viscosity and composition of the medium
where the solvent entity is diffusing varies a lot from the beginning
to the end of cellulose regeneration for NMMO and ionic liquid
system while it does not vary that much for NaOH system. For
example, at the beginning of regeneration, in a 3% cellulose sample
there is 97% of ionic liquid, 82% of NMMO and only 8% of NaOH.
At the end of the experiment the concentration of IL, NMMO and
NaOH is theoretically zero. There is thus a strong gradient of IL and
NMMO concentration in the regenerating cellulose. Thus the average diffusion approach used here cannot be applied for interpreting
the dependence of ionic liquid and NMMO diffusion coefcients on
cellulose concentration.
The diffusion coefcients of EMIMAc and BMIMCl at various
bath temperatures were measured for 5% cellulose solutions. As
expected, the higher bath temperature, the quicker the cellulose
regeneration (see example in Fig. 2). The activation energy was calculated from the temperature dependence of EMIMAc and BMIMCl
diffusion coefcients. It was found to be 1520 kJ/mol which is
comparable with the values of the activation energies obtained for
NaOH and NMMO diffusion coefcients at this cellulose concentration (Gavillon & Budtova, 2007). Overall, the kinetics of cellulose
regeneration from celluloseEMIMAc and celluloseBMIMCl solutions is very similar. Knowing the diffusion coefcient at a given
temperature, it is then easy to estimate the time needed to regen-

1770

R. Sescousse et al. / Carbohydrate Polymers 83 (2011) 17661774

Fig. 4. . SEM images of aerocellulose from 3% (a and b) and 15% (c) celluloseEMIMAc solutions, and from gelled 5% cellulose8% NaOHwater (d, published with permission
from Gavillon & Budtova, 2008, Copyright 2008 American Chemical Society), solid 3% SolucellNMMO (e, published with permission from Gavillon & Budtova, 2008, Copyright
2008 American Chemical Society) and molten 3% SolucellNMMO (f, published with permission from Gavillon & Budtova, 2008, Copyright 2008 American Chemical Society).

erate cellulose of a given shape. For example, it will take about 50 h

to regenerate a 1 cm thick 3% cellulose disk.
3.2. Aerocellulose morphology, density and pore size distribution
A representative morphology of aerocellulose prepared from
celluloseEMIMAc solutions of low (3%) and high (15%) cellulose concentrations are shown in Fig. 4. The system is with the
open pores of the size from few hundreds of nanometers to a
few microns. Cellulose is in a form of globules attached to each
other; their size decreases with the increase of cellulose concentration (see Fig. 4 a and c). The globules themselves are composed
of brous cellulose (Fig. 4b). Similar open-pores morphology
with the same pore sizes was obtained for the aerocelluloses from

celluloseNaOHwater and celluloseNNMO solutions (Gavillon &

Budtova, 2008; Sescousse & Budtova, 2009).
An interesting trend can be seen in the ne morphology of
different aerocelluloses made from celluloseionic liquid solutions (Fig. 4ac), from gelled celluloseNaOH and celluloseNMMO
solutions (Fig. 4df, images taken from Gavillon & Budtova,
2008). The globular structure is observed for celluloseionic
liquid solutions (Fig. 4ac) and for aerocellulose made from
molten celluloseNMMO monohydrate solution (Fig. 4f) while a
net of cellulose strands is seen for aerocellulose from gelled
celluloseNaOHwater and solid celluloseNMMO monohydrate
solutions (Fig. 4d and e, respectively). The reason of this difference was suggested in Gavillon and Budtova (2008) where
aerocelluloses from molten celluloseNMMO monohydrate, solid

R. Sescousse et al. / Carbohydrate Polymers 83 (2011) 17661774

1771

, g cm-3

0.3

0.2

0.1

Ccell, %
0
0

10

15

20

Fig. 5. Aerocellulose density as a function of the initial cellulose concentration

in solutions of celluloseEMIMAc (open points), cellulose8% NaOHwater (dark
points) and SolucellNMMO (crosses). The line is linear approximation y = 0.0168x.

celluloseNMMO monohydrate and gelled celluloseNaOHwater

solutions were compared. In the two latter cases the sample
before regeneration is already phase separated into free solvent
(crystals of NMMO monohydrate or NaOHwater hydrates) and
cellulose + bound solvent. Regeneration occurs in two steps: water
is rst diluting regions with free solvent, and then removing the rest of solvent bound to cellulose. In the liquid (or
molten) celluloseNMMO monohydrate solution cellulose is spatially homogeneously distributed and phase separation occurs in
one step, via spinodal decomposition (Biganska & Navard, 2009)
creating regular small spheres (Fig. 4f). The morphology obtained
for the aerocellulose from celluloseEMIMAc is very similar to the
one from molten celluloseNMMO monohydrate: indeed, these
systems are in the same solution state, contrary to gelled or
solid (crystalline) solutions; phase separation during regeneration
occurs in one step.
The density  of aerocellulose as a function of the initial cellulose concentration is shown in Fig. 5 for EMIMAc, NaOH and
NMMO routes. Data shown for cellulose8% NaOHwater system
do not include aerocellulose prepared with a foaming agent as far
as it was not possible to calculate the initial cellulose concentration in the samples with air bubbles. While the data for ionic liquid
and NaOH routes fall on the same linear dependence, aerocellulose from NMMO route has a slightly higher density. The most
probable reason is different drying procedure parameters and thus
different samples densication: aerocelluloses from ionic liquid
and NaOH were dried in Centre Energtique et Procds, Mines
ParisTech, France (the procedure is described in Section 2.2: mild
conditions and slow depressurization) while aerocellulose from
NMMO monohydrate solutions was dried in Natex, Austria (higher
CO2 temperature and quicker depressurization (Gavillon, 2007)).
One of the commonly used ways of determination of pore size
distribution is mercury porosimetry. However, aerocellulose samples were completely compressed under applied pressure and no
mercury entered the pores (Fig. 6). This phenomenon had already
been described for some aerogels (for example, based on cellulose
acetate (Fischer, Rigacci, Pirard, Berthon-Fabry, & Achard, 2006;
polyurethane (Pirard et al., 2003), silica (Majling, Komarneni, &
Fajnor, 1995; Scherer, Smith, Qiu, & Anderson, 1995) and silicazirconia (Pirard, Blacher, Brouers, & Pirard, 1995)). A buckling
theory for hyper-porous materials was developed by Pirard et al.
(1995) allowing building pore size distribution according to the
equation correlating the size of the largest pores L remaining after

Fig. 6. Cumulative volume as a function of applied pressure for the aerocellulose

from celluloseEMIMAc of various initial cellulose concentrations.

compression at pressure P:
L=

k
P 0.25

(1)

where k is a constant depending on pore wall thickness and Young

modulus and is to be determined in separate experiments from
porous volume evolution during isostatic compression (Pirard et al.,
2003).
The constant k had never been determined for aerocelluloses
and this task was not in the scope of our work. To give a general idea
of the pore size distribution, arbitrary units for pore diameter were
used supposing, in the rst approximation, that k is constant for various initial cellulose concentrations that remains to be proved. The
result is shown in Fig. 7 for three initial cellulose concentrations. As
expected, higher cellulose concentration, smaller are pores: the size
distribution is shifted towards lower values. The results obtained
with mercury porosimetry experiments will be used for the calculation of sample Young modulus; it will be compared with the
mechanical properties of aerocellulose obtained from direct compression measurements in the next section.
The distribution of pore sizes in the range of mesopores
was obtained using nitrogen adsorbtiondesorption method.
The result of BET analysis for two representative aerocellu30

dV/dD
3%
10%

20

15%

10

D, arb. units
0
0.01

0.1

10

Fig. 7. Qualitative pore size distribution in aerocellulose from celluloseEMIMAc

solutions of 3, 10 and 15% obtained by applying Eq. (1) to mercury porosimetry
data.

1772

R. Sescousse et al. / Carbohydrate Polymers 83 (2011) 17661774

0.1

E, MPa

60

dV/dR, cm /(g nm)

y = 531.88x1.6552

cellulose-EMIMAc
0.08

cellulose-8%NaOH

5%

(Solucell and cotton)-NMMO

40
0.06
3.376

10%

y = 5713.3x

0.04

20

y = 3666.4x2.9777

0.02

,, g/cm3

R, nm
0

0
0

10

20

30

Fig. 8. Pore size distribution in the mesopores range for aerocellulose obtained from
5 and 10% celluloseEMIMAc solutions. The lines are given to guide the eye.

loses from celluloseEMIMAc is shown in Fig. 8. The radius of

the pores in the nanometer range is around 10 nm; it slightly
decreases with the increase of the initial cellulose concentration. The surface area varies from 230 m2 /g for the aerocellulose
from 3% celluloseEMIMAc solution to 130 m2 /g for 15% solution. Similar results were reported for aerocellulose made from
celluloseNMMO route (Liebner et al., 2009).
3.3. Mechanical properties of aerocelluloses
Stressstrain curves obtained from the uniaxial compression
of aerocellulose prepared from celluloseEMIMAc solutions are
presented in Fig. 9. During compression no sample buckling was
noticed, the cross-section area did not change and thus, in the
rst approximation, Poisson ratio can be taken equal to zero, as
it was reported for some other natural foams (cork, for example).
Stressstrain dependence are characterised by three regimes: (a)
linear elastic region from which Young modulus E can be determined; (b) a stress plateau, corresponding to progressive cell walls
elastic buckling followed by their collapse when plastic yield is
reached and (c) densication region, when opposing cell walls
touch each other, broken fragments pack together and further
deformation start to compress the wall material itself. After the
end of compression experiment aerocellulose does not recover its
shape, the density increases by several times. Aerocellulose thus

Fig. 9. Stressstrain data for aerocellulose from celluloseEMIMAc solutions of various initial cellulose concentrations.

0.0

0.1

0.2

0.3

Fig. 10. Young modulus vs density for aerocelluloses made from low-DP cellulose
(Avicel) dissolved in 8% NaOHwater and in EMIMAc and high-DP cellulose (Solucell
and cotton) dissolved in NMMO. Lines are power law approximations.

behaves as an elasticplastic foam. The linear elastic behaviour

ends when the material starts to yield and reaches its compressive yield strength,  pl . The second region of the compression curve
is characterised by a low slope as compared with the other two
regions. This plateau enables porous materials to absorb large
amount of energy, W (J/m3 ), without experiencing a large increase
in stress. The absorption energy during the plastic deformation is
dened as the area under the stressstrain curve taken up to 40%
strain. The third region is characterised by a steep increase in the
stressstain curve corresponding to material densication. Each
material characteristic, E,  pl and W, will be compared for aerocelluloses obtained from different routes (ionic liquids, NaOHwater
and NMMO monohydrate); they will be analysed as a function
of density. In general, the increase of cellulose initial concentration (which leads to aerocellulose density increase) leads to the
decrease of pore size and increase of cell wall thickness. The resistance to cell wall bending and collapse is thus increased resulting in
a higher modulus and plateau stress, and the cell walls touch each
other sooner reducing the strain at which densication begins, as
demonstrated in Fig. 9.
For porous materials, foams and aerogels, a scaling relationship
between material mechanical characteristics and bulk density was
suggested, such as for example, E n . The open-cell regular foam
model predicts n 2 (Gibson & Ashby, 1997); for aerogels the exponent was shown to be higher, n 2.54 (see, for example n = 2.7 for
resorcinolformaldehyde aerogels (Pekala, Alviso, & LeMay, 1990),
n = 3.2 and n = 3.7 for silica aerogels (Cross, Goswin, Gerlach, &
Fricke, 1989; Woignier, Phalippou, & Vache, 1989), respectively).
The reason of this difference can be found in the formation of aerogel network itself during a solgel transition: gel structure has a
lot of defects such as dangling ends and loops which do not participate to material response to mechanical stresses. Aerocelluloses
are neither aerogels that are obtained via chemical gelation nor
foams that are made by introducing gas bubbles into a continuous
medium. Aerocellulose structure is formed during cellulose regeneration (coagulation) from a solution into a non-solvent; many
structural defects should also be created during this process and
thus the scaling exponent is expected to be similar to the one found
for aerogels.
Young modulus of various aerocelluloses as a function of density is shown in Fig. 10. Here the results obtained on aerocellulose
from foamed gelled cellulose8% NaOHwater route are included.
The values of E for the samples from the same low molecular
weight microcrystalline cellulose, obtained via dissolution in 8%

R. Sescousse et al. / Carbohydrate Polymers 83 (2011) 17661774

1.2

(a)

,, Pa

1000

E, MPa
Hg porosimetry

cellulose-EMIMAc

1.0

1773

100

cellulose-8%NaOH

cellulose-EMIMAc
cellulose-8%NaOH

0.8

0.6

10

y = 28.043x 2.2204

0.4
1
0.2

, g/cm3

, g/cm3

0.0
0

1200

0.05

0.1

0.15

0.2

0.25

0.1

Fig. 12. Young modulus as a function of density for aerocelluloses made from
celluloseEMIMAc and cellulose8% NaOHwater solutions (dark points, they are
the same data as in Fig. 10) together with E calculated from mercury porosimetry
data (open circles) for the aerocellulose prepared from celluloseEMIMAc solutions.
Line is the power law approximation with the exponent 3.2.

(b)

W, kJ/m 3

0.1
0.01

cellulose-EMIMAc
cellulose-8%NaOH
800

2.5128

y = 36477x

400

, g/cm3
0
0

0.05

0.1

0.15

0.2

0.25

Fig. 11. Compressive yield strength (a) and the absorption energy (b) vs density for
aerocelluloses from celluloseEMIMAc and cellulose8% NaOHwater routes. Lines
are power law approximations.

NaOH or in EMIMAc, are very similar, and show the same trend
vs density (Fig. 10). Their scaling exponents are n = 33.4 conrming the hypothesis that aerocelluloses are aerogel-like materials.
Higher moduli values for aerocellulose from NMMO route were
obtained; the most probable reason is larger cellulose molecular
weight. Similar Young modulus values, of 110 MPa, were reported
for aerocellulose from celluloseNMMO route (Liebner et al., 2009)
for samples made from various celluloses of the initial 3% solutions
and nal densities around 0.06 g cm3 . Surprisingly, we found a
very low scaling exponent for our aerocellulose from NMMO route,
n = 1.7 (Fig. 10). More data on the mechanical properties of this type
of aerocellulose are needed to conrm this result.
The compressive yield strength and the absorption energy as a
function of aerocellulose density are presented in Fig. 11a and b,
respectively. The results for both preparations routes, EMIMAc and
8% NaOH, fall on the same scaling trends:  2.2 and W 2.5 . The
exponents obtained are again very similar to the ones observed for
aerogels (Pekala et al., 1990).
As mentioned in the previous section, the results obtained from
mercury porosimetry can be used to deduce the mechanical properties of the porous material (Cross et al., 1989; Majling et al.,
1995; Pirard & Pirard, 1997; Pirard et al., 1995; Scherer et al., 1995).
The bulk modulus K is known to be a constant correlating applied
pressure P and volumetric strain V/V: P = K(V/V). V/V is a prod-

uct of the cumulative volume and sample (bulk) density . Thus

the bulk modulus can be calculated from the slope of the early
linear stage of compression dependence, pressure vs cumulative
volume (inverted Fig. 6) as K = /. Young modulus was then calculated from the known rule correlating E, K and Poisson ratio ,
the latter was taken equal to zero: E = 3K(1 2) = 3K. The results
are summarized in Fig. 12 where E determined from porosimetry experiments on aerocellulose made from celluloseEMIMAc
is plotted together with the Young modulus obtained from compression data for the same aerocellulose. Data for the compressed
aerocellulose from cellulose8% NaOHwater route are also shown.
The power law obtained for E vs density from porosimetry, with
the scaling exponent n = 3.2, shows an excellent correspondence
with the scaling obtained from direct compressions measurements: for aerocellulose from celluloseEMIMAC n = 3.4 and from
cellulose8% NaOHwater n = 3. A very similar exponent, n = 3.3,
was obtained for aerogels based on a cross-linked cellulose acetate
(Fischer et al., 2006).
4. Conclusions
Aerocellulose from celluloseionic liquid solutions was prepared via cellulose dissolution, regeneration and drying in
supercritical CO2 conditions. The properties of aerocellulose were
compared with the ones of samples made with the same procedure but via cellulose dissolution in NMMO monohydrate and in
8% NaOHwater.
The kinetics of cellulose regeneration in water from
celluloseEMIMAc and celluloseBMIMCl was investigated.
Diffusion coefcients of EMIMAc and BMIMCl were obtained and
found to weakly depend on cellulose concentration within the
range of cellulose concentration studied.
The porosity of aerocellulose was studied using mercury intrusion and nitrogen adsorption methods. Samples were compressed
under mercury pressure and thus only qualitative pore size distribution was deduced. Nitrogen adsorption experiments showed the
presence of mesopores of the average size of 1020 nm and specic
surface area of about 150200 m2 /g. The density of aerocelluloses
was from 0.06 to 0.20 g/cm3 . The bead-like morphology of aerocellulose from celluloseionic liquid solutions was similar to the
one of samples obtained from molten celluloseNMMO monohy-

1774

R. Sescousse et al. / Carbohydrate Polymers 83 (2011) 17661774

drate solutions. The difference with the net-like morphology of

aerocellulose from solid celluloseNMMO monohydrate or gelled
celluloseNaOHwater solutions was explained by different paths
on a phase diagram that the system is taking during coagulation
process.
The mechanical properties of aerocelluloses from EMIMAc,
NMMO and 8% NaOH were measured and compared. For the
samples made from cellulose of the same molecular weight and
prepared in the same drying conditions Youngs modulus was only
controlled by density. It was found that the Youngs modulus is
power-law density dependent with the scaling exponent of about
3, similar to aerogels. This aerogel-like and not foam-like behaviour
was explained by the presence of many defects in the overall
network structure of aerocellulose formed during cellulose regeneration. Finally, a good match was found between the Young moduli
obtained from mechanical compression and mercury experiments.
Acknowledgments
This work was supported by two ANR (France) projects, Carbocell ANR-06-MAPR-0004 and Nanocell ANR-09-HABISOL-010,
and the EC 6th framework, Aerocell project NMP3-CT2003505888. Authors are grateful to P. Ilbizian (Centre Energtique et
Procds, Mines ParisTech, France) for performing supercritical CO2
drying and to B. Simon (Saft, France) for the measurements of aerocellulose porosity.
References
Aaltonen, O., & Jauhiainen, O. (2009). The preparation of lignocellulosic aerogels
from ionic liquid solutions. Carbohydrate Polymers, 75, 125129.
Biganska, O., & Navard, P. (2005). Kinetics of precipitation of cellulose from
celluloseNMMOwater solutions. Biomacromolecules, 6, 19481953.
Biganska, O., & Navard, P. (2009). Morphology of cellulose objects regenerated
from celluloseN-methylmorpholine N-oxidewater solutions. Cellulose, 16,
179188.
Cai, T., Zhang, H., Guo, Q., Shao, H., & Hu, X. (2010). Structure and properties of cellulose bers from ionic liquids. Journal of Applied Polymer Science, 115, 10471053.
Cross, J., Goswin, R., Gerlach, R., & Fricke, J. (1989). Mechanical properties of SiO2 aerogels. Revue de Physique Applique, Colloque C4, Supplment au n 4 avril, 24.
C4-185C4-190
Deng, M., Zhou, Q., Du, A., Kasteren, J. M. N., & Wang, Y. (2009). Preparation of
nanoporous cellulose foams from celluloseionic liquid solutions. Materials Letters, 63, 18511854.
Duchemin, B. J. C., Staiger, M. P., Ticker, N., & Newman, R. H. (2010). Aerocellulose based on all-cellulose composites. Journal of Applied Polymer Science, 115,
216221.
Egal, M., Budtova, T., & Navard, P. (2007). Structure of aqueous solutions of microcrystalline cellulose/sodium hydroxide below 0 C and the limit of cellulose
dissolution. Biomacromolecules, 8, 22822287.
Fischer, F., Rigacci, A., Pirard, R., Berthon-Fabry, S., & Achard, P. (2006). Cellulosebased aerogels. Polymer, 47, 76367645.
Gavillon, R. (2007). Preparation and characterisation of ultra-porous cellulose material.
PhD thesis, CEMEF/Ecole des Mines.
Gavillon, R., & Budtova, T. (2007). Kinetics of cellulose regeneration from
celluloseNaOHwater gels and comparison with celluloseNMMOwater solutions. Biomacromolecules, 8, 424432.
Gavillon, R., & Budtova, T. (2008). Aerocellulose: New highly porous cellulose prepared from celluloseNaOH aqueous solutions. Biomacromolecules, 9, 269277.
Gericke, M., Schlufter, K., Liebert, T., Heinze, T., & Budtova, T. (2009). Rheological
properties of cellulose/ionic liquid solutions: From dilute to concentrated states.
Biomacromolecules, 10, 11881194.
Gibson, L. J., & Ashby, M. F. (1997). Cellular solids. Structure and properties (2nd ed.).
Cambridge University Press.

Innerlohinger, J., Weber, H. K., & Kraft, G. (2006). Aerocellulose: Aerogels and
aerogel-like materials made from cellulose. Macromolecular Symposia, 244,
126130.
Jin, H., Nishiyama, T., Wada, M., & Kuga, S. (2004). Nanobrillar cellulose aerogels.
Colloids and Surfaces A: Physicochemical Engineering Aspects, 240, 6367.
Liebner, F., Haimer, E., Potthast, A., Loidl, D., Tschegg, S., Neouze, M.-A., et al. (2009).
Cellulosic aerogels as ultra-lightweight materials. Part 2. Synthesis and properties. Holzforschung, 63, 311.
Liebner, F., Haimer, E., Wendland, M., Neouze, M. A., Schlufter, K., Miethe, P., et al.
(2010). Aerogels from unaltered bacterial cellulose: Application of scCO(2)
drying for the preparation of shaped, ultra-lightweight cellulosic aerogels.
Macromolecular Bioscience, 10, 349352.
Liebner, F., Potthast, A., Rosenau, T., Haimer, E., & Wendland, M. (2008). Cellulose aerogels: Highly porous, ultra-lightweight materials. Holzforschung, 62,
129135.
Lin, C.-X., Zhan, H.-Y., Liu, M.-H., Fu, S.-Y., & Lucia, L. A. (2009). Novel preparation
and characterisation of cellulose microparticles functionalised in ionic liquids.
Langmuir, 25, 1011610120.
Liu, H., Sale, K. L., Holmes, B. M., Simmons, B. A., & Singh, S. (2010). Understanding the
interactions of cellulose with ionic liquids: A molecular dynamics study. Journal
of Physical Chemistry B, 114, 42934301.
Maeda, H., Nakajima, M., Hagwara, T., Sawaguchi, T., & Yano, S. (2006). Preparation
and properties of bacterial cellulose aerogel. Kobunshi Ronbunshu, 63, 135137.
Majling, J., Komarneni, S., & Fajnor, V. S. (1995). Mercury porosimetry as a means to
measure mechanical properties of aerogels. Journal of Porous Materials, 1, 9195.
Paakko, M., Vapaavuori, J., Silvennoinen, R., Koosen, H., Ankerfors, M., Lindstrom,
T., et al. (2008). Long and entangled native celluloseI nanobers allow exile
aerogels and hierarchically porous templates for functionalities. Soft Matter, 4,
24922499.
Pekala, R. W., Alviso, C. T., & LeMay, J. D. (1990). Organic aerogels: Microstructural
dependence of mechanical properties in compression. Journal of Non-crystalline
Solids, 125, 6775.
Pirard, R., Blacher, S., Brouers, F., & Pirard, J. P. (1995). Interpretation of mercury
porosimetry applied to aerogels. Journal of Materials Research, 10, 21142119.
Pirard, R., & Pirard, J.-P. (1997). Aerogel compression theoretical analysis. Journal of
Non-Crystalline Solids, 212, 262267.
Pirard, R., Rigacci, A., Marechal, J. C., Quenard, D., Chevalier, B., Achard, P., et al. (2003).
Characterisation of hyperporous polyurethane-based gels by non-intrusive mercury porosimetry. Polymer, 44, 48814887.
Quan, S.-L., Kang, S.-G., & Chin, I.-J. (2010). Characterisation of cellulose bers electrospun using ionic liquid. Cellulose, 17, 223230.
Roy, C., Budtova, T., & Navard, P. (2003). Rheological properties and gelation of
aqueous celluloseNaOH solutions. Biomacromolecules, 4, 259264.
Scherer, G. W., Smith, D. M., Qiu, X., & Anderson, J. M. (1995). Compression of aerogels. Journal of Non-Crystalline Solids, 186, 316320.
Seki, S., Kobayashi, T., Kobayashi, Y., Takei, K., Miyashiro, H., Kikuko, H., et al. (2010).
Effects of cation and anion on physical properties of room-temperature ionic
liquids. Journal of Molecular Liquids, 152, 913.
Sescousse, R., & Budtova, T. (2009). Inuence of processing parameters on regeneration kinetics and morphology of porous cellulose from celluloseNaOHwater
solutions. Cellulose, 16, 417426.
Sescousse, R., Le, K. A., Ries, M. E., & Budtova, T. (2010). Viscosity of
celluloseimidazolium-based ionic liquid solutions. Journal of Physical Chemistry
B, 114, 72227228.
Swatloski, R. P., Spear, S. K., Holbrey, J. D., & Rogers, R. D. (2002). Dissolution of cellulose with ionic liquids. Journal of the American Chemical Society, 124, 49744975.
Tsioptsias, C., Stefopoulos, A., Kokkinomalis, I., Papapdoupoulou, L., & Panayiotou, C.
(2008). Development pf micro- and nano-porous composite materials by processing cellulose with ionic liquids and supercritical CO2 . Green Chemistry, 10,
965971.
Turner, M. B., Spear, S. K., Holbrey, J. D., & Rogers, R. D. (2004). Production of
bioactive cellulose lms reconstituted from ionic liquids. Biomacromolecules, 5,
13791384.
Wendler, F., Kosan, B., Kreig, M., & Meister, F. (2009). Possibilities for the physical
modication of cellulose shapes using ionic liquds. Macromolecular Symposia,
280, 112122.
Woignier, T., Phalippou, J., & Vache, R. (1989). Parameters affecting elastic properties
of silica aerogels. Journal of Materials Research, 4, 688692.
Zhang, H., Wu, J., Zhang, J., & He, J. (2005). 1-Allyl-3-methylimidazolium chloride
room temperature ionic liquid: A new and powerful nonderivatizing solvent for
cellulose. Macromolecules, 38, 82728277.
Zhao, Q., Yam, R. C. M., Zhang, B., Yang, Y., Cheng, X., & Li, R. K. Y. (2009). Novel
all-cellulose ecocomposites prepared in ionic liquids. Cellulose, 16, 217226.