Bioresource Technology 100 (2009) 5162–5169

Contents lists available at ScienceDirect

Bioresource Technology
journal homepage: www.elsevier.com/locate/biortech

Tannin-based rigid foams: A survey of chemical and physical properties

G. Tondi a, W. Zhao b, A. Pizzi a, G. Du b, V. Fierro c, A. Celzard d,*
a
ENSTIB-LERMAB, Nancy-University, 27 rue du Merle Blanc, BP1041, 88051 Epinal cedex 9, France
b
Department of Wood Science and Technology, South West Forestry University, Kunming, Yunnan, People’s Republic of China
c
Institut Jean Lamour, UMR CNRS 7198, CNRS – Nancy-Université – UPV-Metz, Département Chimie et Physique des Solides et des Surfaces,
Faculté des Sciences and Techniques, BP 239, 54506 Vandoeuvre-lès-Nancy cedex, France
d
Institut Jean Lamour, UMR CNRS 7198, CNRS – Nancy-Université – UPV-Metz, Département Chimie et Physique des Solides et des Surfaces, ENSTIB,
27 rue du Merle Blanc, BP 1041, 88051 Épinal cedex 9, France

a r t i c l e i n f o a b s t r a c t

Article history: Tannin-based rigid foams, prepared from 95% natural material, are suggested for replacing synthetic phe-
Received 23 March 2009 nol–formaldehyde foams in various applications. For that purpose, a few physical properties were mea-
Received in revised form 18 May 2009 sured and reported here: resistance to fire and chemicals, absorption of various liquids, permeability,
Accepted 20 May 2009
thermal conductivity and mechanical (compressive and tensile) strength. Modifying the composition
Available online 2 July 2009
through the use of boric and/or phosphoric acid allowed substantial increase of fire resistance. The mate-
rials were also found to present good resistance to strong acid and bases, and to solvents. High affinity for
Keywords:
water, but limited one for organic liquids, was also evidenced. Finally, slightly anisotropic mechanical
Tannins foams
Phenolic foams
properties were measured. The materials present a brittle behaviour, whether tested in compression or
Physical properties traction; nevertheless, their strengths, as well as their thermal conductivities, are fully comparable with
Chemical resistance those of their phenolic counterparts. We show that such materials of vegetable origins can compete with
synthetic ones for most of traditional applications.
Ó 2009 Elsevier Ltd. All rights reserved.

1. Introduction phenolic foams. Especially, foamed phenolic resoles are customar-

Structural thermoset foams are extensively used in transporta-
tion, packaging, cushioning, insulation, but also in building mate-
rials, automobiles, aircrafts and marine structures, as well as in
electronic applications and flame retardants. For all these applica-
tions, flame, smoke and toxicity performances are critical. From
that point of view, the performances of phenolic foams are much
higher than those of polyurethane, polyvinyl chloride or polysty-
rene foams. Due to their good insulating properties, low density,
outstanding fire characteristics, low smoke emission, no dripping
of molten plastic when exposed to flame and low cost, phenolic
foams bring a suitable solution as insulation and sandwich core
materials. They are also highly resistant to chemicals and solvents.
Finally, due to their brittle character, phenolic foams may dissi-
pate energy irreversibly by damage and rupture of the skeleton,
and find applications in the field of crash protection and
packaging.
Foams based on natural products might replace industrial phe-
nol–formaldehyde (i.e., phenolic resoles) foams in most of applica-
tions, as far as they present comparable properties. It is the
purpose of the present work to show that thermoset tannin-de-
rived foams are relevant materials for traditional applications of
* Corresponding author. Tel.: +33 329 29 61 14; fax: +33 329 29 61 38.
E-mail address: Alain.Celzard@enstib.uhp-nancy.fr (A. Celzard).
ily used as open-celled flower foams, closed celled insulation foam
for civil engineering, foam for fire protection elements, filler for
mining and tunnelling, etc. Such light materials might be replaced
by vegetable tannins-based ones, given that tannins indeed com-
bine high reactivity, ‘‘green” origin, and low cost (typically 650€
per metric ton). Mimosa (Acacia mearnsii formerly mollissima, De
Wildt) bark tannin-based rigid foams have been known since
1994 (Meikleham and Pizzi, 1994). These tannin–furanic foams
typically comprise 95% of natural products. Polyflavonoid tannins,
representing more than 55% of the foam, are vegetable extracts de-
rived from bark. Around 20% of the final weight of the foam is con-
stituted by furfuryl alcohol which is obtained throughout catalytic
reduction of furfural, a natural derivative obtained by hydrolysis of
the sugars from several agricultural crops (Aguilar et al., 2002).
Tannin-based foams are prepared from a viscous liquid mixture
of tannin–formaldehyde resin and furfuryl alcohol. The latter is
used as a heat-generating agent through both its auto-polymerisa-
tion and for its reaction with the tannin under acid conditions; it
also has an advantageous strengthening effect. Expansion of this li-
quid mixture to low-density foam is obtained by simultaneous
cross-linking of the tannin–furfuryl alcohol mixture and evapora-
tion of a low boiling-point solvent used as blowing agent (here
diethyl ether). In the same time, polymerisation occurs because
tannins quickly react with aldehydes, especially through the nucle-
ophilic sites of their A ring.

0960-8524/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.biortech.2009.05.055
 G. Tondi et al. / Bioresource Technology 100 (2009) 5162–5169
Characterisation of tannin-based rigid foams was just initiated
in a recent work (Tondi and Pizzi, 2008). The present study reports
further advances and more detailed information. Besides, signifi-
cant improvements are reported here, through simple composition
modifications of reference materials. Physical properties of the lat-
ter, such as liquid absorption and permeability, thermal conductiv-
ity and compressive and tensile resistance, are also measured and
discussed. As far as the available data from the open literature al-
low it, systematic comparison of these properties with those of
synthetic phenolic foams is made. It will be demonstrated that tan-
nin-based foams present very similar performances, and can thus
be considered as ecological, cheap, alternative materials. Finally,
the measured properties are modelled on the basis of the foams’
structure, and useful relationships are suggested for the first time,
for predicting the behaviour of such materials as a function of their
apparent density.
2. Experimental
2.1. Preparation of foams
Standard foams (i.e., without additives) were prepared as fol-
low: furfuryl alcohol (5.2 g), formaldehyde 37% water solution
(3.7 g) and water (3.0 g) were mixed with 15 g of tannin extract
under mechanical stirring. When the mixture became homoge-
neous, diethylether (foaming agent: variable amount between 1
and 2 g) and then para-toluene-4-sulphonic acid (pTSA) 65% water
solution (6.0 g) were added and mixed for 10 s. Diethyl ether was
chosen among other blowing agents because of its very low boiling
point (35 °C) and good miscibility with the other components of
the foam. The mixture was then poured in a 6  6 cm plastic lab-
made PET box. The resin foamed within 2 min of mixing, leading
to black foam of density ranging from 0.04 to 0.12 g cm3. The tem-
perature rose to about 40 °C. After 10 min the skin of the foam was
cut out and the remaining structure was left to age for at least 24 h.
During this period, evaporation of residual blowing agent trapped
into the foam occurred. Due to the rising of the gas vertically dur-
ing foaming, the foams are orthotropic materials. They indeed
present slightly elongated cells in the direction of cell growth,
whereas the pores appear to be isotropic when observed in the
orthogonal plane. Therefore, two measurement directions were de-
fined: the so-called growing (or z-) direction, corresponding to the
(vertical) direction of foaming, and the flat (xy-) direction, corre-
sponding to the orthogonal (horizontal) plane.
Several modified foams have been prepared in order to increase
their fire resistance: (a) foam modified with boron, prepared by
addition of a solution of boric acid (H3BO3: 50.0 g L1) instead of
water; (b) foam modified with phosphorus, prepared by addition
of 1.5 g of ortho-phosphoric acid (H3PO4: 85%) and 4.5 g of pTSA
65% water solution, instead of just 6.0 g of pTSA; Foam modified
by both phosphorus and boron, prepared with the two, simulta-
neous, aforementioned modifications. For all these materials, the
amount of diethylether was always kept constant at 1.5 g.
2.2. Characterisation of foams
2.2.1. Porosity and surface area
Given their composition (mainly natural polymers cross-linked
in aqueous solution), measuring their surface area through the tra-
ditional BET method based on nitrogen adsorption at 77 K is not
possible. The materials indeed need to be outgassed by heat-treat-
ment under vacuum, and continuously release volatile matters in
these conditions. Thus, obtaining a sufficiently low working pres-
sure in the adsorption apparatus is never possible. Therefore, the
surface area can only be estimated, and compared with that of
5163
carbon foams (obtained by pyrolysis in inert gas) having the same
origin and the same pore structure.
Unpublished works from our group revealed, on the basis of
SEM observations and tomography studies, that the pore structure
may be modelled by a close packing of connected spherical pores. If
the latter all have a radius r, then the specific surface area (m2 g1)
reads:
3C
S¼ ð1Þ
dr
pffiffiffi
C is the compacity and is equal to p 2=6 ¼ 74%, whether the close-
packing is FCC or HCC type, and d is the bulk density, defined as the
mass of material divided by the total volume it occupies. The pore-
size distribution being shouldered at low pore diameters, about 10%
of the total porosity is based on spherical pores having a radius typ-
ically r/10, and located between the bigger cells of radius r. The
additional surface area brought by this narrower porosity is then:
3
Sadd ¼ ð2Þ
dr
so the total surface area is Stot = S + Sadd, hence:
3ð1 þ C Þ
Stot ¼ ð3Þ
dr
The porosity was determined from the bulk density of the foam and
from the skeletal density, ds, (i.e., the density of the polymer from
which the foam is made), which was measured by helium pycnom-
etry in a lab-made apparatus; ds was found to be 1.59 g cm3,
slightly higher than that of pure cured phenol–formaldehyde resin:
1.47. The porosity, U, reads:
d
U¼1 ¼ 1  dr ð4Þ
ds
where dr is the so-called relative density (dimensionless). The sur-
face area could be calculated from average pore sizes, derived from
the observation of a high number of cells by scanning electron
microscope (SEM Hitachi S 4800).
2.2.2. Resistance to fire and chemicals
Measuring such properties is of interest, given that these foams
can be suitable materials for industrial thermal insulation in severe
conditions. Resistance to fire was tested as follows. Foam samples
of similar densities (around 0.058 g cm3) and dimensions
(2  2  2 cm) were exposed facially for 20 s to the oxidising flame
of a Bunsen burner, and especially to the top of the blue cone of the
flame (at least 1200 °C and 10 kW m2). The flame was considered
as extinguished when the generated glow disappeared and the
foam reacquired its original black colour. During the 10 min test,
the weight loss of each sample was measured as a function of time.
Tannin-based rigid foams might be used as filters for corrosive
chemicals. For that reason, some tests of chemical resistance to
acid and alkali media were achieved. Resistance to strong bases
and acids was tested as follows. 30  30  15 mm foam samples
of density 0.06 g cm3 were soaked for 24 h into either 33%
H2SO4 or 33% NaOH aqueous solutions. Samples were then dried
by towel-paper and next during 12 h in air at 60 °C. The compres-
sive strengths were finally measured once the samples were totally
dry (see Section 2.2.6).
2.2.3. Absorption of liquids
Absorption of water on one hand, and various organic fluids on
the other hand, was investigated. In order to measure the maxi-
mum amount of water that can be absorbed by foams of different
densities, cubic 2  2  2 cm samples were placed in an impregna-
tion chamber filled with water. The pressure of the air above water
was decreased to 10 mbar, which value was maintained for
5164 G. Tondi et al. / Bioresource Technology 100 (2009) 5162–5169

60 min; then, the materials were slowly left to return to ambient
pressure. This process has been repeated three times.
Absorption of other liquids was also measured; for the sake of
comparison, all the tests were carried out on identical foam sam-
ples having a density of 0.08 g cm3. Thus, 25 cubic 2  2  2 cm
samples were soaked in the following liquids: water, 1-butanol,
xylene, petroleum ether and sunflower oil. The weight increase
of the foams was thus measured for various soaking times.
2.2.4. Permeability to water
Cubic samples of side 17 mm were glued with hot-melt glue
into a cylindrical copper tube of 26 mm diameter and 30 mm
length. The glue neither intruded the porosity nor covered the
opposite faces of the foam through which water was forced to flow.
Before starting the experiment, two to three pressure–vacuum cy-
cles were repeated in order to saturate the porosity with water.
The permeability, k, of these samples was then determined by
measuring the amount of water passing throughout during times
ranging from 3 to 10 min, and applying Darcy’s law:
L g 
k¼Q
A DP
Q is the flow rate, L and A are sample thickness and cross-sectional
area respectively, g is the dynamic viscosity of water (103 Pa s),
and DP is the pressure drop over the sample.
3. Results and discussion
3.1. Porosity and surface area
SEM images of foam samples having densities of 0.05, 0.085 and
0.12 g cm3 are not shown in the manuscript, but may be seen in
the Electronic Annex 1 in the online version of this article. Their
corresponding porosity and average pore sizes are listed in Table
1. Given the anisotropy of the materials, two values of pore size,
i.e., along xy- and z-directions, are introduced. The values are close
to what is usually reported for phenolic foams: from 100 to 200 lm
(Kim and Lee, 2008; Shen and Nutt, 2003; Auad et al., 2007). From
these data, the surface area could be estimated using Eq. (3), in
which the radius of the pore was taken as the half of the pore size.
The highest and lowest values of surface area given in Table 1 thus
correspond to lowest (xy) and highest (z) values of the pore size,
respectively. The surface areas are typically within the range 0.5–
1 m2 g1, in good agreement with what could be derived from mi-
cro-tomography studies on one hand (Tondi et al., 2009a), and
from nitrogen adsorption carried out with carbon foams derived
ð5Þ from the present materials on the other hand, the latter having
close densities and similar pore texture (Tondi et al., 2009b). More-
over, surface areas within the range 0.5–1.75 were also measured
for phenolic foams whose densities are similar to the ones reported
here (Denslow and Rickle, 1989).

2.2.5. Thermal conductivity 3.2. Resistance to fire and chemicals

The thermal conductivity was measured using a lab-made
guarded heat flow (hot plate) method, already described elsewhere As expected, all tannin-based rigid foams are flame resistant
(Py et al., 2001, and references therein). For that purpose, cubic and self extinguish. Such fire resistance, which is achieved without
samples of side close to 2.5 cm were used. The temperature gradi- additives, indeed originates from the chemical structure of the
ent within the material, installed between one hotter and one cured resin, based on many aromatic rings of high binding energy.
colder plate, was measured by a number of thermocouples whose Just like for phenolic resole foam, the structure carbonises in fire,
probes were placed at different depths. The whole structure was and the carbon layer protects the inner parts of the material; be-
protected by polystyrene foam blocks to avoid excessive side heat sides, very few volatile matters are released. A typical (non-modi-
loss and, once the system reached steady-state equilibrium, the fied) standard tannin-based foam, undergoing 20 s of direct
thermal conductivity was derived through a computer programme exposure to the oxidising flame of a Bunsen burner (top of the blue
taking account all the necessary parameters. The samples were cone of the flame), thus needs around 60 s after flame withdrawal
tested in two directions, namely the direction of foam growth before self-extinguishing in air. Such already excellent fire resis-
and the direction normal to this latter. Several samples having dif- tance could be even improved through the formulation changes
ferent densities were analysed. described in Section 2.1. The results are reported in Table 2. Self-
extinguishing time of modified foams is always much lower than
2.2.6. Mechanical resistance
The compression strength of foam samples of dimensions
3  3  1.5 cm was measured with an Instron 4206 universal test-
Table 1
ing machine equipped with a 1 kN head, and using a load rate of Pore texture characteristics of various tannin-based rigid foams.
2.0 mm min1. The complete strain–stress characteristics were re-
corded for a broad range of bulk densities, but only the mechanical Bulk density Porosity xy-Pore size z-Pore size Estimated surface area
(g cm3) (%) (lm) (lm) (m2 g1)
resistance – or compressive strength, also called fracture stress –
will be discussed here. The strength is defined as the height of 0.050 96.9 215 428 0.49 < Stot < 0.97
0.085 94.7 180 308 0.40 < Stot < 0.68
the long serrated plateau. Since the plateau is sometimes not com-
0.120 92.5 81 118 0.74 < Stot < 1.07
pletely horizontal, the compressive strength was always taken at
20% strain. Samples were tested along the two aforementioned
orthogonal axes: growing z-direction and flat xy-direction.
Table 2
For measuring the resistance to traction, opposite ends of foam Average extinguishing time and final aspect of tannin-based foams after exposure to
samples of dimensions 50  50  15 mm were tightly fixed to two an oxidising flame.
5  6 cm flat steel plates using hot-melt glue. The glued samples
Foam Bulk Extinguishing Aspect after 10 min of
were tested in traction, again along two orthogonal directions, density time (s) after flame exposure
with an Instron 4467 universal testing machine using a load rate (g cm3) exposition to
of 2.0 mm min1. The tensile strength was deduced from the flame for:
strain–stress characteristics as the point of highest stress before 20 s 60 s
the material suddenly cracks. Such experiments were carried out
Standard 0.0562 49.5 382 Ashes
on a lower range of densities, given the difficulty to prepare H3BO3 – modified 0.0531 21.1 278 Ashes
sufficiently homogeneous long samples of density higher than H3PO4 – modified 0.0572 0 0 Several cracks
0.06 and 0.075 g cm3 along growing z- and flat xy-directions, H3BO3 + H3PO4 – 0.0593 0 0 A low number of cracks
modified
respectively.
 G. Tondi et al. / Bioresource Technology 100 (2009) 5162–5169
that of the reference material. Ignition time of phenolic foam ex-
posed to 50 kW m2 is 6 s (Auad et al., 2007); in the case of tan-
nin-based foams, no ignition at all was observed, so ignition time
can be considered to be infinite in the present experimental
conditions.
The aqueous acid solution used in the composition of the vege-
table foam catalyses the cross-linking reactions, allowing the resin
to cure near room temperature. Replacing pure water by small
amounts of an aqueous solution of boric acid led to upgraded
foams which were able to self-extinguish twice as fast. However,
the resultant foam was not a completely inert material, and pro-
longing flame exposure to 10 min degraded the material until
ashes. Phosphoric acid modification was more effective. Complete
replacement of para-toluene sulphonic acid (pTSA) by H3PO4 was
not possible because tannins in the foam are sensitive to complex-
ation reactions and do not allow obtaining homogeneous materi-
als. However, replacement of only 25% pTSA noticeably improved
fire resistance properties. H3PO4 – modified foams, in fact, do not
glow red at all; no red spot appears on the foam surface. In other
words, the ignition time again tends to infinity. Such high thermal
stability and insulation to fire are properties of interest for applica-
tion to aircraft industry; sandwich part with phenolic foam cores
were indeed shown to be relevant for replacing sandwiches with
honeycomb core structures in cabins of aircraft (Haberstroh and
Schlumm, 2006). Moreover, foams are also better from the point
of view of acoustic and thermal insulation.
Higher flame exposure times produced cracking of the foams
because of the sudden temperature-induced devolatilisation of
water and other light chemicals produced by thermal decomposi-
tion of the material. However, addition of boric acid to phosphoric
acid significantly lowered the cracking of the materials (see Table
2).
In order to quantify the fire resistance of the samples, their
weight loss was measured as a function of time when directly ex-
posed to a flame. The results are presented in Fig. 1, in which the
protective effect of ortho-phosphoric acid is clearly demonstrated.
Such an increased resistance to oxidation is similar to that already
encountered in lignin–phosphoric acid mixtures. H3PO4 was in-
deed shown to increase the carbon yield of pyrolysed organic
materials in two ways: (i) increase of dehydration and promotion
of a structural rearrangement of the solid, which also reduced
the emission of volatile compounds and (ii) prevention of combus-
tion in the air through the formation of a layer of phosphorus
hemipentoxide (see Fierro et al., 2005, 2006 for details). Higher
char yields are desirable because char forms an isolation layer
improving the fire resistance by limitation of heat transfer to the
inside of the material during the ignition process. Moreover, the
100
80
Weight loss (%)
60
40
Standard foam
Boron-modified
20
Phosphorus-modified
Boron-Phosphorus-modified
0 sities within the range 0.057–0.068 g cm3 (Riahi et al., 2007). De-
0 2 4 6 8 10
Time (min)
Fig. 1. Weight loss of tannin-based rigid foams (average bulk density 0.058 g cm3)
as a function of time when directly exposed to an oxidising flame.
5165
glass-like carbon produced by pyrolysis of tannin-based foams
was already shown to be highly resistant to high temperatures
(Tondi et al., 2009b). Modifying the formulation of the foams with
phosphoric acid then visibly protected the foam, so more heat was
necessary to oxidise the hardened polymer network.
Soaking standard tannin-based rigid foams into a 33% sulphuric
acid water solution changed the colour of the solution to pinkish,
probably because pTSA adsorbed on foam surface was leached off
by H2SO4. After drying, the aspect of the foam structure was not
modified at all. Whereas still appearing to be quite strong, mechan-
ical measurements (carried out on foam of density 0.06 g cm3)
evidenced a decrease of about 70% of the compression strength.
Nevertheless, the foam thus appears to be very resistant since it
survived 24 h of soaking followed by 12 h of drying at 60 °C, the ex-
cess acid remaining in the porosity being highly corrosive in such
conditions.
When soaked into a 33% NaOH water solution, the foam swelled
and became elastic and flexible after 24 h of immersion. However,
shaping the original rigid foam seems impossible using this meth-
od, since homogeneous drying of the sample was very difficult, a
great amount of solution remaining inside the foam.
3.3. Absorption behaviour
Investigation of water absorption by tannin-based rigid foams
was initiated in a previous work (Tondi and Pizzi, 2008). Given that
the foam cells are open, absorption occurs by simple soaking but is
slow (see below). By contrast, vacuum–pressure cycles led to fast
and complete filling of the porosity by water. The results are pre-
sented in Fig. 2a as a function of foam’s bulk density. All the data
points could be fitted by the simple following power law:
x
Water update / d ð6Þ
where x is an exponent very close to 1, suggesting that the ab-
sorbed amount is nearly inversely proportional to the bulk density
of the foam. A straightforward calculation based on the data of
Fig. 2a easily shows that the volume of absorbed water is very close
to that of the available pore space of the dry material (deduced from
Eq. (4)). Consequently, the observed phenomenon is indeed absorp-
tion, and not adsorption that might have been accounted for by Eq.
(3) (which suggests that the inner surface is also inversely propor-
tional to bulk density). The amount of water absorbed was defined
as:
mfinal  minitial mwater dwater
water uptake ¼ ¼ ¼ U ð7Þ
minitial msimple d
where m are masses and dwater is the density of water (1 g cm3). Gi-
ven the range of bulk densities investigated here, the porosity does
not change much (from 91.2% to 96.8%) and it may be roughly as-
sumed that U is a constant. The inverse proportionality between
water absorption and density is thus recovered through Eq. (7).
The fact that the observed exponent is slightly different from –1
might be explained by the low change of porosity with bulk density,
but most probably by the presence of a small amount of closed
pores.
The observed trend, i.e., decreasing amount of absorbed water
at higher density, was also reported for other phenolic (resole)
foams; however the water uptake was much slower, since corre-
sponding to almost closed cells foams: 16–4% for densities within
the range 0.06–0.35 g cm3 (Shutov, 1984), or 19.3–13.5% for den-
spite their high affinity for water, long-term weathering
experiments already proved that tannin-based foams are water-
proof and highly resistant to moisture (Tondi et al., 2008), just like
phenolic foams used in marine equipments.
5166 G. Tondi et al. / Bioresource Technology 100 (2009) 5162–5169

2000 chemical resistance of tannin-based foams to these solvents.

2000
Transport in the porosity is now examined.

water uptake (%)

1500
Water uptake (wt. %)

3.4. Permeability to fluids

1500 1000
The permeability of a number of foams measured along z- and
500
xy-directions is presented in Fig. 3a as a function of density. As ex-
6 10 14
3
18
-1
22 pected, low-density foams, i.e., having the biggest pores and the
1/d (cm g )
widest pore windows (see Table 1 and Tondi et al. (2009a)), are
1000
much more permeable then the ones having higher density. Again,
a power law such as:
n
k/d ð8Þ
500 was found to fit the data very correctly, whatever the measurement
0.04 0.06 0.08 0.1 0.12 0.14 0.16
-3 direction. A seen in Fig. 3a, the exponent n is close to 2. At very
Bulk density (g cm ) low density, k should tend to infinity as the porosity becomes 100%.
(a) However, given the very small range of porosity explored, and
given the high porosities involved, for which the permeability
1000 changes very strongly, it is difficult to discuss what should be the
theoretical permeability behaviour. It is similar to the one ob-
800 served by Xu et al. (2008), whose permeability data plotted as a
Liquid uptake (wt. %)

function of 1  U in the same range of porosity (i.e., as a function

Water
of d, just like in our case) may also be very well fitted by Eq. (8).
600 1-butanol
Xylene
Petroleum ether
400 Sunflower oil -10
1.2×10
200 -10
-10
slope = -2.03
1.0×10
Permeability (m2)

-10.4

log k
0 8.0×10
-11
0 4 8 12 16 20
-10.8 slope =
Time (days) -11
6.0×10 -2.40
(b) -11.2
-1.4 -1.2 -1 -0.8
-11
Fig. 2. (a) Maximal water absorption of tannin-based rigid foams having different 4.0×10 log d
densities. The solid curve is calculated from the power law (Eq. (6)) in which the
exponent is 0.965. The inset shows the validity of Eq. (6); (b) absorption of tannin- -11
based foams soaked into various liquids, as a function of time. 2.0×10 xy
z
0
0.04 0.06 0.08 0.1 0.12 0.14
Absorption of other liquids was compared with that of water.
The results gathered in Fig. 2b evidence that water is the fluid most
Bulk density (g cm-3)
absorbed by tannin-based foams. Absorption of other liquids is not (a)
only much lower than that of water, but quickly saturates, whereas
water uptake is almost linear until complete pore filling is -9
10
achieved. These behaviours may be essentially explained on the
basis of polarity, viscosity and molecular size of the liquids in-
Calculated permeability (m )
2

volved. Especially, extremely small, mobile and polar molecules

such as H2O strongly interact with the high number of hydroxyl -10
groups present everywhere on the foam surface that is based on 10
flavonoid units. Water thus quickly fills the most external cells,
and then progressively spreads all over the surface so it can fill
the inner porosity. Water being present in the composition of the xy
-11
foam itself, it may be exchanged by foreign H2O molecules, which 10
z
consequently are able to permeate the pore walls. Comparing
Fig. 2a and b, it can be deduced that approximately 30% of the foam
is filled by water in 1 h, 40% in 24 h and around 70% in 15 days.
Less polar molecules intrude the porosity much slower than 10
-12
water due to the lower interaction with the surface. Molecules 10
-12
10
-11
10
-10
10
-9

being also bigger, they probably can not permeate throughout pore
walls, so absorption finally stops in the absence of pressure. A Measured permeability (m2)
higher viscosity may also contribute to a lower uptake, just like (b)
in the case of oil. Xylene is likely to present affinities with the fla-
Fig. 3. (a) Permeability of tannin-based rigid foams as a function of density,
vonoid rings through p–p interaction, possibly justifying the rather measured along xy- and z-directions. The lines are calculated from power laws (Eq.
high absorbed amount at initial times. For all the solvents tested, (8)) in which the exponents are close to 2. The inset shows the validity of Eq. (8);
finding a stabilised sample weight after 10 days evidences the (b) calculated permeabilities according to Eq. (11) as a function of measured ones.
 G. Tondi et al. / Bioresource Technology 100 (2009) 5162–5169 5167

However, our value of n is higher than theirs. This result is partic-
ularly difficult to analyse, since the permeability of foams depends
on both porosity (at constant cell size) and pore diameter (at a
fixed porosity) (Paek et al., 2000; Biasetto et al., 2007; Bhattach-
arya et al., 2002). Since cell size and porosity are interdependent
in the case of the present foam (see Table 1), the exponent of Eq.
(8) cannot be predicted.
Permeability along the xy-direction is higher than along z-direc-
tion. Interconnection between cells is indeed more likely when the
cells have higher contact areas. Considering the anisotropy of the
material, side-by-side connectivity is expected to be higher than
in the direction of bubble growth.
The data given in Fig. 3a may be tentatively recovered basing on
the theory of Kozeny and Carman (see Celzard et al., 2005, and ref-
erences therein). In this context, the permeability reads:
U3
k¼ ð9Þ
fS2sp
where f is the so-called Kozeny constant (f = 2 for cylindrical tubes:
not relevant here, f = 5 in many porous materials), and Ssp (m2 m3)
is the internal accessible surface area per unit volume of solid mate-
rial. Given that:
3ð1 þ CÞ
Ssp ¼ d Stot ¼ ð10Þ
r Values of mechanical strength measured along the two xy- and
then the permeability may be rewritten as:
U3 r 2
k¼ ð11Þ
f½3ð1 þ CÞ2
Eq. (11) thus has the same form as that formerly developed by Brace
(1977): k = AD2U3, in which A is an empirical constant and D is the
average diameter of the cells.
The permeabilities were calculated from Eq. (11), using the val-
that the cells are fully open; therefore, even better insulating prop-
erties are expected with closed cells tannin foams.
3.6. Mechanical properties
When submitted to a compression, all the foams present a brit-
tle failure mode, characterised by three distinct regions of the
strain–stress curve: linear elastic (up to 10% strain, on average),
collapse and densification. The long serrated plateau, see Fig. 4a,
typically ranging from 20% to 80% strain, originates from the co-
existence of collapsed and uncollapsed zones, typical of brittle
foam undergoing successive cell wall fractures. Beyond the pla-
teau, densification takes place and the stress rises sharply as com-
plete densification begins. These features are also typical of
phenol–formaldehyde foams (Shen and Nutt, 2003; Okzul and
Mark, 1994; Rangari et al., 2007; Gontier et al., 2001). The tensile
strain–stress characteristics, see Fig. 4b, are rather different. Even
if the same elastic region may be observed, the fracture is typical
of a brittle solid that suddenly cracks for a low critical strain. Only
compressive and tensile strengths are discussed here as a function
of foam density. It was not the purpose of the present paper to con-
sider the changes of elastic modulus or plateau length; these points
will be developed in a future work purely focusing on mechanical
properties.
z-directions under compression on one hand, and under traction on
the other hand, are given in Fig. 5a and b, respectively. More ex-
0.8
0.7
0.6
xy
Stress (MPa)

ues of U obtained with Eq. (4), while f was assumed to be 5; the 0.5 z
results are compared with the measurements in Fig. 3b. Given
0.4
the limitation of Kozeny–Carman’s theory, which best
describes the behaviour of a bundle of cylindrical, parallel, tortuous 0.3
tubes, the agreement between calculation and measurements may
0.2
be considered as very satisfactory.
0.1
3.5. Thermal properties 0
0 20 40 60 80 100
Measurements of thermal conductivity confirm the high insu- Strain (%)
lating properties of tannin-based foams. Average values as low as (a)
0.024 and 0.030 W m1 K1 were indeed measured for materials
having densities of 0.08 and 0.12 g cm3, respectively. No clear dif- 0.08
ferences were observed between z- and xy-directions in this short xy
range of densities. Such values compare very well with those of 0.07
phenolic insulation foams: 0.029 and 0.057 W m1 K1 for a lab- 0.06 z
made (micro-wave assisted) foam and a commercial one, both hav-
Stress (MPa)

ing a bulk density of 0.12 g cm3 (Kim and Lee, 2008). Since the 0.05
skeletal density of the phenol–formaldehyde resin (1.47 g cm3) 0.04
is different from that of the tannin-based polymer, these values
of thermal conductivity correspond to materials having a relative 0.03
density dr = 0.082 (see definition in Eq. (4)). The corresponding
0.02
bulk density of tannin-based foam would then be
dr  1.59 = 0.13 g cm3. The thermal performances of our material 0.01
thus appear to be close to those of the commercial phenolic foam
0
of similar relative density. 0 0.2 0.4 0.6 0.8 1
The thermal conductivities measured in this work also compare
Strain (%)
well with those of other phenolic (0.04–0.08 W m1 K1 for densi-
ties 0.04–0.2 g cm3 (Shutov, 1984), 0.057 W m1 K1 at (b)
0.12 g cm3 (de Carvalho et al., 2003)) and lignophenolic rigid Fig. 4. Typical strain–stress characteristics of (a) a rigid tannin-based foam (density
foams (0.072 W m1 K1 at 0.45 g cm3 (de Carvalho et al., 0.053 g cm3) undergoing compression, and (b) a rigid tannin-based foam (density
2003)). Permeability and liquid absorption experiments evidenced 0.047 g cm3) undergoing traction.
5168 G. Tondi et al. / Bioresource Technology 100 (2009) 5162–5169

actly, the logarithm of strength was plotted as a function of the log Table 3
of density, in order to check that strength, r, depends on bulk den- Compressive strengths (MPa) of tannin-based and phenolic foams, as a function of
their relative density.
sity according to a power law. The following equation was indeed
evidenced by extensive analytical, numerical and experimental Relative Tannin-based foamsa Phenolic foams References for
works (see for example Gibson and Ashby, 1997): density phenolic foams
xy- z- xy- z-
Direction Direction Direction Direction
r / da ð12Þ
0.016 0.03 0.03 0.06 Okzul and Mark
The exponent a is associated to the structure and deformation (1994)
mechanics of the cellular material. According to Sanders and Gibson 0.037 0.12 0.12 0.085 Riahi et al. (2007)
0.04 0.14 0.14 0.1 Gontier et al. (2001)
(2003), the theoretical values of a are 1.5, 2 and 1.36 if the foams are 0.054 0.22 0.23 0.65 0.76 Shen and Nutt (2003)
based on open-cells, closed-cells and hollow spheres, respectively. 0.054 0.22 0.23 0.4 Mendelsohn et al.
Fig. 5a and b both shows similar behaviour: mechanical resis- (1979b)
tance increases with density according to Eq. (12). The correspond- 0.068 0.32 0.33 0.88 Mendelsohn et al.
(1979a)
ing exponents are close to 1.5 in most cases (1.55 and 1.65 for
0.08 0.41 0.43 0.4 Gontier et al. (2001)
compression along xy- and z-directions, respectively, and 1.64 0.082 0.42 0.45 0.8 de Carvalho et al.
and 1.54 for traction along xy- and z-directions, respectively). (2003)
The value of the compression exponents is near that of open-celled 0.102 0.59 0.65 0.8 Shutov (1984)
foam, as expected from SEM observations and permeability and 0.136 0.93 1.04 1.85 2.17 Auad et al. (2007)
0.156 1.15 1.31 1 Shutov (1984)
absorption experiments. That of the z-direction is slightly higher 0.161 1.21 1.38 5.4 Rangari et al. (2007)
than the one of the xy-plane, as also noticed by Auad et al. 0.306 3.26 3.97 1 de Carvalho et al.
(2007) who measured the mechanical properties of phenolic foams (2003)
along the same axes. The exponents for the tensile strength are a
Deduced (extrapolated) from Fig. 5a.
more difficult to appraise, given the lack of bibliographic refer-
ences and the scattering of the results.
Anisotropy can be considered as negligible as far as compressive
behaviour is concerned, whereas the tensile strength is higher along the growing z-direction. This result is readily explained by
the preferential orientation of the struts, leading to a higher
strength along the z-direction. Nevertheless, the resistances be-
-0.2 come identical at the lowest densities. In other words, anisotropy
increases with density, and especially as far as traction tests are
considered. This finding is also typical of the so-called graphite
log compressive strength

-0.4 foams, being fully isotropic at low density, and more and more
anisotropic when the porosity decreases (Celzard et al., 2003). It
-0.6 is interesting to note that the aspect ratio of the cells paradoxically
decreases with density, as evidenced in Table 1, whereas anisot-
ropy increases. However, the mechanical strength is controlled
-0.8 by the weakest struts, whose thickness tends to be very low at
xy
the lowest densities, thus explaining the quasi isotropic behaviour.
z Despite more rounded cells, foams having higher densities are
-1
characterised by more dissimilar strut thicknesses, depending on
the measurement directions. This assumption is supported by the
-1.2 observed anisotropy of electrical conductivity, measured on carbon
-1.3 -1.1 -0.9 foams obtained by pyrolysis of rigid tannin-based foams (Zhao
log density et al., 2009), which also increases with density.
(a) Because skeletal densities of phenol–formaldehyde and tannin-
based resins are different, mechanical properties should be only
-0.8 compared for identical relative densities (calculated by Eq. (4)).
Values of compressive strengths of tannin-based and phenolic
foams are gathered in Table 3. It can be seen that, in most cases,
-0.9 the reported fracture stresses are close to each other, especially
log tensile strength

at low density. Natural foams derived from tannins thus definitely

-1 compare very correctly with their synthetic phenol-based
counterparts.

-1.1 xy 4. Conclusion
z
-1.2 This work clearly demonstrated that tannin-based foams can
compete with synthetic, commercial, phenolic foams because they:

-1.3  Are cheap, fast and easy to prepare.

-1.4 -1.2 -1
 Self extinguish and present high fire resistance, which can even
log density
be significantly improved through the modification of their
(b) composition.
Fig. 5. Logarithm of strength (MPa) of tannin-based foams as function of log of
 Are highly resistant to chemicals (strong acids or bases,
density (g cm3): (a) compressive strength; (b) tensile strength. The straight lines solvents).
are calculated from Eq. (12).  Are very resistant to moisture and weathering.
 G. Tondi et al. / Bioresource Technology 100 (2009) 5162–5169
 Present extremely low thermal conductivity.
 Are shock-mitigating materials, having compression strengths
close to those of phenolic foams, especially at low density for
which the insulating properties are the highest.
Acknowledgements
The authors gratefully acknowledge the financial support of the
CPER 2007-2013 ‘‘Structuration du Pôle de Compétitivité Fibres
Grand’Est” (Competitiveness Fibre Cluster), through local (Conseil
Général des Vosges), regional (Région Lorraine), national (DRRT
and FNADT) and European (FEDER) funds. Two of us (WZ and
GD) also thank the Natural Science Foundation of Yunnan Province
(China).
Appendix A. Supplementary data
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.biortech.2009.05.055.
References
Aguilar, R., Ramırez, J.A., Garrote, G., Vazque, M., 2002. Kinetic study of the acid
hydrolysis of sugar cane bagasse. J. Food Eng. 55, 309–318.
Auad, M.L., Zhao, L., Shen, H., Nutt, S.R., Sorathia, U., 2007. Flammability properties
and mechanical performance of epoxy modified phenolic foams. J. Appl. Polym.
Sci. 104, 1399–1407.
Bhattacharya, A., Calmidi, V.V., Mahajan, R.L., 2002. Thermophysical properties of
high porosity metal foams. Int. J. Heat Mass Transfer 45, 1017–1031.
Biasetto, L., Colombo, P., Innocentini, M.D.M., Mullens, S., 2007. Gas permeability of
microcellular ceramic foams. Ind. Eng. Chem. Res. 46, 3366–3372.
Brace, W., 1977. Permeability from resistivity and pore shape. J. Geophys. Res. 82,
3343–3349.
Celzard, A., Marêché, J.F., Furdin, G., 2003. Describing the properties of compressed
expanded graphite through power laws. J. Phys. Condens. Matter 15, 7213–
7226.
Celzard, A., Marêché, J.F., Furdin, G., 2005. Modelling of exfoliated graphite. Prog.
Mater. Sci. 50, 93–179.
De Carvalho, G., Pimenta, J.A., dos Santos, W.N., Frollini, E., 2003. Phenolic and
lignophenolic closed cells foams: thermal conductivity and other properties.
Polym. Plast. Tech. Eng. 42, 605–626.
Denslow, K.R., Rickle, G.K., 1989. Surfactant effects in phenolic foam resins. J. Cell.
Plast. 25, 31–42.
5169
Fierro, V., Torné-Fernández, V., Montané, D., Celzard, A., 2005. Study of the
decomposition of kraft lignin impregnated with orthophosphoric acid.
Thermochim. Acta 433, 153–159.
Fierro, V., Torné-Fernández, V., Celzard, A., 2006. Kraft lignin as a precursor for
microporous activated carbons prepared by impregnation with ortho-
phosphoric acid: synthesis and textural characterisation. Micropor. Mesopor.
Mater. 92, 243–250.
Gibson, L.J., Ashby, M.F., 1997. Cellular Solids: Structure and Properties, second ed.
Cambridge University Press, Cambridge.
Gontier, C., Bouchou, A., Vinot, C., 2001. A mechanical model for the computation of
phenolic foams in compression. Int. J. Mech. Sci. 43, 2371–2384.
Haberstroh, E., Schlumm, M., 2006. Properties and manufacturing of sandwich parts
with phenolic foam cores. J. Polym. Eng. 26, 213–226.
Kim, B.G., Lee, D.G., 2008. Development of microwave foaming method for phenolic
insulation foams. J. Mater. Proc. Technol. 201, 716–719.
Meikleham, N.E., Pizzi, A., 1994. Acid- and alkali-catalyzed tannin-based rigid
foams. J. Appl. Polym. Sci. 53, 1547–1556.
Mendelsohn, M.A., Meier, J.F., Rudd, G.E., 1979a. Effects of density and structure on
load bearing and permeability characteristics of a family of shock mitigating
phenolic foams. J. Appl. Polym. Sci. 23, 333–339.
Mendelsohn, M.A., Meier, J.F., Rudd, G.E., 1979b. Effects of composition and
processing conditions on the properties of a series of shock mitigating
phenolic foams. J. Appl. Polym. Sci. 23, 341–351.
Okzul, M.H., Mark, J.E., 1994. The effect of preloading on the mechanical properties
of polymeric foams. Pol. Eng. Sci. 34, 794–798.
Paek, J.W., Kang, B.H., Kim, S.Y., Hyun, J.M., 2000. Effective thermal conductivity and
permeability of aluminium foam materials. Int. J. Thermophys. 21, 453–464.
Py, X., Olives, R., Mauran, S., 2001. Paraffin/porous–graphite-matrix composite as a
high and constant power thermal storage material. J. Heat Mass Transfer 44,
2727–2737.
Rangari, V.K., Hassan, T.A., Zhou, Y., Mahfuz, H., Jeelani, S., Prorok, B.C., 2007. Cloisite
clay-infused phenolic foam nanocomposites. J. Appl. Polym. Sci. 103, 308–314.
Riahi, F., Doufroune, R., Bouremel, C., 2007. Substitution of pentane by a mineral
clay as a blowing agent for phenolic foam. Int. J. Polym. Mater. 56, 929–938.
Sanders, W.S., Gibson, L.J., 2003. Mechanics of hollow sphere foams. Mater. Sci. Eng.
A 347, 70–85.
Shen, H., Nutt, S., 2003. Mechanical characterization of short fibre reinforced
phenolic foam. Compos. A 34, 899–906.
Shutov, F.A., 1984. Phenolic foams in the USSR. Cellular Polym. 3, 95–104.
Tondi, G., Pizzi, A., 2008. Tannin-based rigid foams: Characterization and
modification. Ind. Crops Prod. 29, 356–363.
Tondi, G., Pizzi, A., Olives, R., 2008. Natural tannin-based rigid foams as insulation
material in wood construction. Maderas Ciencia y Tecnologia 10 (3), 219–227.
Tondi, G., Blacher, S., Léonard, A., Pizzi, A., Fierro, V., Leban, J.M., Celzard, A., 2009a.
X-ray microtomography studies of tannin-derived organic and carbon foams.
Micros. Microanal, accepted for publication.
Tondi, G., Fierro, V., Pizzi, A., Celzard, A., 2009b. Tannin-based carbon foams. Carbon
47, 1480–1492.
Xu, W., Zhang, H., Yang, Z., Zhang, J., 2008. Numerical investigation on the flow
characteristics and permeability of three-dimensional reticulated foam
materials. Chem. Eng. J. 140, 562–569.
Zhao, W., Celzard, A., Tondi, G., Pizzi, A., Fierro, V., 2009. Unpublished results.