Upload
holger
View
214
Download
0
Embed Size (px)
Citation preview
Fire retardancy of sol–gel derived titania wood-inorganiccomposites
Muhammad Shabir Mahr • Thomas Hubert •
Martin Sabel • Bernhard Schartel • Horst Bahr •
Holger Militz
Received: 23 March 2012 / Accepted: 28 May 2012 / Published online: 20 June 2012
� Springer Science+Business Media, LLC 2012
Abstract Sol–gel technology was applied in tailoring
novel wood-made-inorganic composites with improved
thermal and fire properties. In practice, composites materials
were prepared by impregnating pine sapwood wood with
nano-scaled precursor solutions derived from titanium(IV)
isopropoxide followed by a thermal curing process. Thermal
and fire properties were evaluated by thermal analysis and
cone calorimetry, whereas flammability was specified by
oxygen index (LOI) and UL 94 test. Peak heat release rates
were moderately reduced indicating fire retardance potential
in terms of flame spread attributed to the appropriate pro-
tection layer action of the titania-based depositions. LOI
(oxygen index) values of these composites were increased up
to 38 vol.% in comparison to 23 vol.% for untreated wood.
The flame retardancy performance depends on the fire sce-
nario and is strongly influenced by wood loading and crack-
free deposition of the titania layers inside the composite.
Introduction
Wood is one of the most important natural materials due to its
versatile properties and usage in all most in every sphere of
life. Despite of its excellent engineering properties, it has
some serious disadvantages such as low dimensional stability
in moist atmosphere, low resistance against UV-radiation,
heat and fire as well as low biological durability (low resis-
tance against insects, fungi, and other organisms) [1–4]. Due
to these draw backs an additional treatment is required for its
protection. In general, conventional wood preservatives
effectively protect timber from natural diseases, however, due
to their toxic effects some are under strict scrutiny since
decades. Therefore now, to address environmental concerns,
one is seeking environmentally friendly alternatives to wood
preservatives. Due to less toxic impact, cost effective pro-
cessing and easy handling makes sol–gel derived materials as
one of the potential candidate for wood modification and
preservation at industrial scale.
Sol–gel is a well adapted method for producing glass and
ceramic materials at low temperatures [5, 6]. Usually with this
process, metal oxides are formed by the hydrolysis and sub-
sequent condensation reactions initiating from precursors in
wet state (in the presence of acid or base catalyst) [7, 8] and
their desirable features such as chemical composition, struc-
ture can be tuned by varying synthesis parameters as well as
precursors used [9]. Saka et al. [10] applied this approach to
enhance the properties of wood by infiltrating sol–gel-based
solutions into wood to form a new class of novel composite
materials namely ‘‘wood-inorganic composites.’’ In later
trials, they formed a large variety of sol–gel derived wood-
inorganic composites by depositing different sol–gel precur-
sor solutions into untreated wood and studied extensively
their resistive role against moisture, fire and micro-organisms.
Results revealed better fire retardancy, dimensional, and
UV-stabilization [2, 11–17] as well as possessed high decay
resistance against white, brown, and soft rot fungi [18].
Similar studies were reported by other researchers. Donath
et al. [19] used sols derived from different silanes namely,
tetraethoxysilane (TEOS), methyl triethoxysilane (MTES)
and propyl trimethoxysilane (PTEO) by sol–gel process and
M. Shabir Mahr � T. Hubert (&) � M. Sabel � B. Schartel �H. Bahr
BAM Federal Institute for Materials Research and Testing,
Unter den Eichen 44-46, 12203 Berlin, Germany
e-mail: [email protected]
M. Shabir Mahr � H. Militz
Wood Biology and Wood Products, Burckhardt Institute,
Georg-August-University Gottingen, Busgenweg 4,
37077 Gottingen, Germany
123
J Mater Sci (2012) 47:6849–6861
DOI 10.1007/s10853-012-6628-3
found that wood properties such as wall bulking, anti-swelling
efficiency, moisture uptake, and bio-durability were
improved significantly by treating wood with these sol–gel
silane-based systems. In another study, they used sols derived
from amino-functional silanes [20] and reported a significant
enhancement in fungal decay resistance of the treated wood
with these systems. Bucker et al. [21] developed nano-sized
precursors derived from tetra ethoxysilane (TEOS) by sol–gel
process to get inorganic-wood composites with improved
properties. They stated that these materials showed better
dimensional stability as well as improved resistance against
water, fire, and bio-organisms. Most recently Hubert et al.
[22] used titanium(IV) n-butoxide and titanium(IV) iso-
propoxide as precursors to prepare TiO2 wood composites
by employing sol–gel route. Results indicated that these
wooden products exhibited better dimensional stability and
fire retardancy than their non treated counterparts.
Most of the researchers investigated the fire behavior of
sol–gel derived wood-inorganic composites using thermal
analysis (TGA, DTA) [2, 3, 14, 15, 22]. No doubt, thermal
analysis is very efficient, milligram based and fast method to
qualitatively characterize the thermal behavior of materials
and thus potent to deliver some hints for the fire behavior of a
material such as char yield and pyrolysis temperatures.
However, it should be noted that it is actually not sufficient to
estimate the fire properties of macroscopic specimens or
components in hardly any fire scenario or fire test. Thus,
more comprehensive studies about sol–gel-based wood-
inorganic composites that cover thermal and fire behavior
under different fire scenarios is still missing. The aim of this
contribution was to fill this gap by investigating the thermal
and fire behavior of newly prepared sol–gel derived TiO2
wood-inorganic composites under different fire conditions
and fire scenarios. In addition, it was also the goal of this
study to find out the influence of parameters like composi-
tion, size of the precursor species, and the depositions of
titania sol–gel precursor solutions, as well as dry TiO2 gel
present in resultant products after final curing on the fire
performance under different fire environments. Herein,
thermal properties were evaluated by employing thermo-
gravimetric analysis (TGA) and differential thermal analysis
(DTA) while fire behavior under forced flaming conditions
was characterized using the cone calorimeter. Flammability
in terms of reaction to small flame was investigated using the
limiting oxygen index (LOI) and UL 94 classification.
Experimental
Wood specimens
Wood specimens (pine sapwood, Pinus sylvestris L.) were
prepared for LOI (80 9 10 9 4 mm), UL 94 test
(125 9 13 9 1.6 mm), and cone calorimeter investiga-
tions (100 9 100 9 10 mm). All the samples were oven
dried (18 h, 103 �C) to get their reference weights and
volumes prior to impregnation. The masses of the wood
specimens were measured by means of an electric balance
(TR5003 Mettler Toledo).
Preparation of precursor solutions
Six different precursor solutions (T1 to T6) of different molar
ratios were prepared from titanium(IV) isopropoxide (TIP,
Alfa Aesar, 97 %) by adding in 2-propanol (ISP, Sigma-
Aldrich, 99.5 %) under vigorous stirring (see Table 1). In
each case a catalyst (65 % HNO3) was added to adjust the pH
value 2 of the solutions. TIP is highly reactive with water
therefore; it can readily hydrolyze during the synthesis using
water from the atmosphere. The catalyst contains some water
that can be used to start hydrolyzing the TIP. As a consequence
of TIP–water interaction, particles suspended in the solvent
will be formed. Thus, immediately after synthesis, all the
solutions were tested for evaluating the mean particle sizes of
the titanium containing particles formed and then were stored
at 17 �C in a storage tank until impregnation finished. Usually,
titanium alkoxide precursor solutions are stable for longer
periods (more than 3 months) and can be used for impreg-
nating wood in this period [22]. However, in this study, these
precursors were impregnated within a week after their prep-
aration without further stirring before impregnation.
Wood impregnation and curing treatments
Oven dried samples were impregnated under vacuum
conditions. All the samples were evacuated at 0.1–0.3 kPa
for 1 h before impregnate solutions. Impregnation took
place at 0.1–0.2 kPa subsequently specimens were soaked
at 5–10 kPa for 2 h at 20–23 �C. For each precursor
solution, all the test specimens were treated simulta-
neously. For some selective solutions, samples were doubly
impregnated in the same way. In order to insure complete
hydrolysis and well fixation of the precursor solutions into
Table 1 Characteristic parameters of sol–gel precursor solutions
Precursor
solution
TIP: ISP/
molar
ratio
Equivalent TiO2
solid content/
mass%
Species size
in solution/
nm
T1 1:120 1 0.9
T2 1:15 6.5 1.3
T3 1:10 9 1.6
T4 1:6 12 1.8
T5 1:3 16 4.3
T6 1:1 21 5.7
6850 J Mater Sci (2012) 47:6849–6861
123
the wooden structures, all samples were treated to a special
curing program which included different heating steps like
specimen storage in a desiccators at 20–23 �C and a rela-
tive humidity of 95 % (achieved by supersaturated KNO3
solution) for 1–7 days; open air drying for 1–7 days and a
final curing at 103 �C for 18 h. Described impregnation
and curing procedures are valid for relative small scaled
samples, whose sizes are demanded by standards test pro-
cedures. Until now, samples of (100 9 100 9 10) mm3
dimensions have been treated successfully. Upscaling to
commercial dimensions needs special equipment as well as
larger times for solution infiltration and subsequent soaking
and drying. Even the use of pressure treatment cannot be
ruled out for effective impregnation of sol–gel materials in
bigger wood slices.
Methods
Thermo-oxidative stability of the composites was assessed
by employing TGA/DTA using SETARAM TGA 92-16.
All the measurements were performed in the temperature
range of 23–800 �C at a constant heating rate of 2 K min-1
under air. Prior to all fire tests, the test specimens were
preconditioned at 23 �C with a relative humidity of 50 %
for 88 h. Cone calorimeter (Fire Testing Technology, East
Grinstead, UK) measurements were carried out according
to ISO 5660 [23]. Cone calorimeter works on oxygen
consumption principle to determine the heat release rate
(HRR) during combustion of a material [24]. A set of fire
risks and characteristics like total heat release (THR), time
to ignition, effective heat of combustion, smoke density,
mass loss rate, and CO-yield are measured as well. All
measurements were performed in the horizontal position at
a constant irradiation of 50 kW m-2 in triplicate and the
obtained values were averaged.
The limiting oxygen index (LOI) was determined
according to ISO 4589 [25] (apparatus: Stanton Redcroft,
East Grinstead, UK). LOI is the minimum amount of
oxygen in oxygen–nitrogen mixture required to sustain
candle like combustion in the specified set-up. The higher
the LOI value is, the more flame retardant the material is.
UL 94 horizontal and vertical classifications were deter-
mined using a Plastics HVUL Horizontal Vertical UL
Flame chamber (Fire Testing Technology, East Grinstead,
UK) in compliance with the standard IEC 60695-11-10
[26]. V-2, V-1, and V-0 UL 94 classification are defined by
self-extinguishing occurring within a specified time and
HB classification at least when horizontally burning speed
is below a defined limit.
Morphological information about the sol–gel derived
wood composites was obtained by employing environmental
scanning electron microscopy (ESEM) using a Philips
XL-30 linked to an EDAX energy dispersive X-ray analyz-
ing system (EDX). Typical specimens with cross sectional
cuts in 5 mm thickness were prepared with a Reichert sliding
microtome and sputtered with carbon before recording
ESEM images. EDX-mapping analysis was carried out to
observe the distribution of the impregnated sol–gel-based
inorganic particles present in the cell walls. The particle size
of the solution precursors was measured as hydrodynamic
particle diameter (considering them spherical objects) by
means of dynamic light scattering (DLS) using a particle
sizer (HPPS, Malvern, UK).
Results and discussion
Parameters influencing solid uptakes (WPG)
and composite morphology
All precursor solutions (T1 to T6) were color less liquids at
the time of synthesis but later on turned to light yellowish
upon storage. Characteristic details of titania precursor
solutions parameters like concentration (expressed in molar
ratios and equivalent solid titania content) and hydrody-
namic sizes of the precursor species are shown in Table 1.
Average diameters of titania precursor species present in
precursor solutions having uncertainty of about 0.1 nm
were in the range of 1–6 nm depend heavily on the con-
centration of titanium alkoxide. Precursor solution T1 with
molar ratio of TIP/ISP of 1–120 and an equivalent titania
solid content of 1 mass% contained species of 0.9 nm in
diameters while size was increased to 5.7 nm in T6 solu-
tion (equivalent solid titania 21 mass%). From the results,
it was noted that species size increased with the titania
solid content (titanium alkoxide concentration) and fol-
lowed the order on the basis of titania solid content i.e.,
T1 [ T2 [ T3 and so on (see Table 1).
All TiO2 wood composites were slightly lighter in color
compared to untreated controls due to TiO2 gel solidifica-
tion on the surfaces. However, these color variations were
very small and the original color of wood was not much
affected by the sol–gel treatment. As a result of vacuum
impregnation by the precursor solutions a mass gain of
material was observed due to their penetration into the
cellular wooden structures. Upon final curing infiltrated
material was deposited into wood blocks and termed as
weight percent gains (WPG) in literature [2–4, 19, 22].
Data concerning WPG are placed in Table 2 for all the
composites prepared applying precursor solutions T1 to T6
choosing specimens of various dimensions for performing
different fire tests (LOI, UL 94, and Cone calorimeter). It is
clear from the data that solid uptakes or WPG for all the
samples irrespective to their sizes and dimensions were
increased by increasing molar ratio of TIP/ISP or indirectly
J Mater Sci (2012) 47:6849–6861 6851
123
with titania solid content. The WPG values for composite
(TC1) derived from T1 precursor solution with the smallest
species size (0.9 nm) were about 3 mass% showing ‘WPG/
solid content’’ ratio equal to 3. For all other composites
‘‘WPG/solid content’’ values were not increased in such
extent as the penetration probability of the precursor
solutions tend to decrease due to containing larger size
titania precursor species. Noticeably, precursor solution T5
being more concentrated (solid content of 15 mass%
greater than T1) showed an increase of around 80 % in
WPG in the second deposition cycle for all treated samples
of different dimensions and geometries. T1 apparently
indicated a minor increase in WPG during second
impregnation too, even though the deviation is strictly
speaking covered by the uncertainty. Significant blockage
of cell wall micro voids as a consequence of first impreg-
nation is proposed as the reason for this phenomenon. This
implies higher solid content (titanium alkoxide concentra-
tion) in the precursor does not always lead to higher WPG
values. It may depend on the characteristic properties of the
precursor solution like species size, viscosity, and degree
of hydrolysis etc., as well as influenced by density,
dimensions, and geometry etc., of the wood impregnated.
ESEM investigations of radial and cross sectional parts
of the titania wood-inorganic composites were performed
to extract information about their microstructure. Typical
results are displayed in Fig. 1 a, b, c, d. Sol–gel precursor
solutions with nano-sized titania species (1–6 nm) were
effectively impregnated and fixated in the whole wood
matrix. Being larger cavities most of the material was
absorbed and deposited in the cell lumen in the form of
TiO2 gels upon hydrolysis and subsequent condensation of
the precursor solutions. These titania gels were placed there
in the form of nano-scaled films and coating the internal
surfaces of cell lumen. These depositions were not crack
free; however, very few cracks were found in composite
TC2 formed using T2 precursor with titania solid content
of 6.5 mass% (Fig. 1b). This less cracking may correlate to
low alkoxide concentration of the solution in such a way
that the thickness of the developed films was in the nano
range and thus can more sustain thermal treatment [27–29].
Composite TC5, prepared by relatively more concentrated
precursor solution T5 (titania solid content of 9.5 mass%
greater than T2) contained more cracks (see Fig. 1d). In
this case, the curing thick TiO2 gel films results in cracking
or delaminating due to differences in thermal expansions of
wood matrix and TiO2 deposited films.
Some samples were also subjected to EDX-mapping to
investigate the presence and distribution of titania gels in the
cell walls (see Fig. 2b, d). Figure 2d shows that few TiO2
gels were found in cell wall cavities of T5 (solid content
16 mass%) prepared titania wood composite (TC5);
although the TiO2 gels distribution was not uniform in the
entire cell wall areas. Relatively more gel were apparent in
the cell walls of TC2 composites (with similar random dis-
tribution) prepared with less concentrated solution T2 with
solid content 6.5 mass% as demonstrated in Fig. 2b. In
general, this kind of filling is caused by the penetration of
titanium containing species which are in the dimensions of
the wood cell wall voids (*2–8 nm). The infiltrated titanium
alkoxide solution of low solid content will be subsequently
converted to solid TiO2 upon curing [15, 22]. This further
implies that TiO2 content in the cell walls is dependent on the
concentration of the precursors. Due to limitations of larger
particle size and smaller dimensions of cell wall voids,
higher concentrated solutions were expected to be less
soaked into the cell walls. Furthermore, rapid hydrolysis and
condensation of precursor present in the lumen led to gela-
tion there that consequently interrupted the further diffusion
of liquid into the cell walls [15]. These limitations were well
over come by diluting the precursors. Larger cell wall
sorption can be attributed to small particulate size (1.3 nm)
of T2 solution as well as smaller deposition in cell lumen
because of the low titanium alkoxide concentration (small
solid content) of the solution rendered less hindrance in its
diffusion in cell walls. In contrary, previously reported
results for sol–gel derived TiO2 wood-inorganic composites
prepared from TIP where TiO2 gels were not detected in cell
wall regions [15]. This could be due to the effect of pre-
conditioning before impregnation. Herein, oven dried (0 %
moisture content) samples were used for impregnation while
in the reported work moisture conditioned samples with
20–25 % moisture content were subjected to impregnation
[15]. It is assumed that this huge amount of water in the
lumen speeded up the hydrolysis of TIP precursors followed
by fast condensation to form TiO2 gels in larger amounts into
wood lumen prevented further absorption of the liquid into
cell wall areas. Thus, as a result, the amount of TiO2 gels
present in the cell walls was too low to be detectable with
SEM-EDX.
Table 2 WPG values for titania wood-inorganic composites pre-
pared by single and double impregnation of titania precursor solutions
T1 to T6 and residual mass after TG experiment
Composites Impregnation
cycles
WPG/
mass%
TG residue/
mass%
UN 0 0 \1
TC1 1 3 ± 1 \2
TC1-2 2 4 ± 6 \2
TC2 1 10 ± 1 5
TC3 1 12 ± 2 7
TC4 1 21 ± 1 13
TC5 1 27 ± 3 13
TC5-2 2 49 ± 6 28
TC6 1 39 ± 6 18
6852 J Mater Sci (2012) 47:6849–6861
123
Thermo-oxidative decomposition
To study thermal decomposition, titania wood-inorganic
composites as well as for comparison untreated wood samples
were thermally analyzed with TGA and DTA under air.
Representative TG curves are displayed in Fig. 3a showing
mass loss as a function of temperature. From the TG curve of
untreated wood, three mass loss regimes (1, 2, and 3) were
identified in different temperature ranges. Below 220 �C
(region 1), a slight mass loss was observed. This was related to
moisture content liberated during measurement as well as
partial decomposition of least stable wood component hemi-
celluloses that began to degrade from 200 �C [30, 31]. A
considerable second and third mass loss (region 2 and 3) were
occurred in the ranges of 220–330 �C and 330–460 �C,
respectively. Mass loss in the range of 220–330 �C was
related to oxidative decomposition of wood components [32]
while mass loss in region 3 was attributed to char oxidation
[30]. After 460 �C, no further mass lost implied that untreated
wood was completely thermo-oxidized with very small resi-
due (\1 mass%). For titania wood-inorganic composites,
similar TG curves were obtained, consisting all three mass loss
regions. In region 1 relatively more mass loss corresponding to
release volatile organic residues (un-hydrolyzed alkoxide)
along with water dehydration. In the range of 220–330 �C, a
moss loss of around 40 mass% apparent for both untreated
wood and composites without showing pronounced mass loss
difference in this region. However, in region 3 composites
were decomposed with different final mass loss caused by
remaining titania as stable residues (see Fig. 3a). In this
region, all composites curves were hardly shifted to higher
temperatures indicating any considerable delay in the oxida-
tion of char. For all the titania wood-inorganic composites
yielded residual masses depended on WPG of the tested
composites. In comparison to WPG shown in Table 2, these
values are rather small displaying a disagreement in WPG and
the residual amounts. This implies that WPG contains con-
siderable amount of organic residuals (un-hydrolyzed alkox-
ide as a result of incomplete hydrolysis and condensation).
These organic residuals were thermally instable and released
during measurement leaving only stable titania residues at the
end of TG measurement.
DTA thermograms of untreated wood and all composites
are placed in Fig. 3b. Two distinct exothermic peaks appeared
in temperature ranges of 290–350 �C and 410–460 �C for
untreated wood. First exothermic narrow peak with larger
intensity was mainly ascribed to oxidative decomposition of
wood cell wall components (hemicelluloses and cellulose and
Fig. 1 ESEM micrographs of sol–gel derived titania wood composites: a and b from TC2 while c and d were developed from TC5 composites.
Images a and c are taken from cross sectional while b and d from radial cuts
J Mater Sci (2012) 47:6849–6861 6853
123
lignin) while second was attributed to oxidation of char as
similarly described above for respective TG curve. Appar-
ently, for all cases of composites, the intensity of the first peak
considerably reduced. The intensity of second peak was also
reduced gradually with increasing WPG and was slightly
shifted toward higher temperatures revealing a delay in oxi-
dation of char. These results are similar to previously reported
results for titania wood-inorganic composites [15, 22].
Fire properties
Fire behavior under forced flaming conditions
The investigation of the fire behavior in the cone calorimeter
results in two important characteristics to assess the fire risks.
The HRR related to flame spread and fire growth and the total
heat release (THR) which corresponds to fire load at the end
of the test [24, 33, 34]. The HRR profiles for measured
samples are illustrated in Fig. 4a, b and main cone calo-
rimeter results are summarized in Table 3. In addition, res-
idues of all tested samples collected after end of test are
visualized in Fig. 5. Always, the HRR representative curves
are a two peak profile typical for thick char forming materials
like wood. Out of two peaks, first one is caused by the
increasing amount of volatile fuel oxidizing within the
growing flame due to the rapid increase in mass release rate
before a char layer is formed limiting the pyrolysis. While
second peak at the end of burning corresponds to char layer
cracking and further decomposition [33, 34]. In comparison
to untreated wood, the HRR of titania wood-inorganic
composites was reduced in particular the second peak HRR
(PHRR) toward the end of burning accompanied by pro-
longed burning times (see Fig. 4a, b). These findings showed
similarities to DTA curves presented in Fig. 3b where sim-
ilar effects were observed in term of intensity reduction and
peak shifting in comparison to untreated wood. Of course,
the experiments were performed under different environ-
ments that led to different fire scenarios as well as pyrolysis.
Supported by residual amount obtained for untreated wood:
17.3 mass% in the cone calorimeter whereas\1 mass% in
TG (see Tables 2, 3), pyrolysis was mainly anaerobic in the
cone calorimeter [35] whereas thermo-oxidative decompo-
sition took place in TG under air. Therefore, the shifts and
reduction observed in DTA were not directly related to the
shifts observed for the second PHRR in the cone calorimeter.
Nevertheless, the similarities give a remarkable hint to a
general protection mechanism active in the residues of
homogeneous sol–gel derived titania composites. It is
noticed that HRR considerably decreased (but not tends to
zero) after the second PHRR. This fire response is assigned to
an after glowing of char through thermo-oxidative decom-
position. The reduction in heat release rate is irrelevant for
Fig. 2 Localization of TiO2 gels in the wood matrix of TiO2 wood-inorganic composites as characterizes by ESEM micrographs and EDX
mapping. a and b were taken from TC2 while c and d were developed from TC5 composites
6854 J Mater Sci (2012) 47:6849–6861
123
the first maximum in HRR. Beyond first PHRR, HRR
decreases because of the protection of carbonaceous char. It
is concluded that the protection performance was improved
by the applied titania depositions layers. It was reported
before that degree of reduction, shape, width, and shift
observed in the second PHRR depends on the protection
properties of residues [36–39]. Shift magnitude as well as
decrease in the second peak intensity of HRR was varied for
all the tested samples indicating a change in the residual
characteristics during combustion in the cone calorimeter.
Figure 4c revealed that the composites TC1 (WPG
3 mass%), TC1-2 (WPG 4 mass%) and TC2 (WPG
10 mass%) showed similar order of reduction (19, 26, and
30 %, respectively compared to untreated wood see Table 3)
with increased residual amounts in their second maximum of
HRR indicating similar properties of their residues. TC5 with
WPG 27 mass% and with 28.8 % residual content exhibited
a smaller reduction in the second PHRR indicating a reduced
protection action of the residues. This poorer performance
may be attributed to extensive cracking occurred in the char
formed as well as associated to cracks and imperfections
related to its composite micro structure as confirmed by
ESEM micrographs (Fig. 1d). An improvement in the pro-
tection action was observed when titania gels introduced in
the second impregnation healed most of the cracks formed
during first impregnation consequently reduced the second
PHRR to 25 % for TC5-2. On the basis of these results, it was
concluded that not the residual content (indirectly WPG) as
such controls the second PHRR of the tested composites.
Apparently, protection properties of the titania depositions
are not only dependent on the amount deposited into the
wood (WPG) but on the crack-free effective fixation of
material into the wooden structure that leads to a closed
barrier layer with greater capability to hinder the heat and
mass transportation in the condensed phase.
In contrast to untreated wood, the times to second PHRR
(2.TPHRR) were appreciably altered depending on the WPG
values for tested composites (see Fig. 4c; Table 3). Time
related to second maximum of HRR for TC2 composite (WPG
10 mass%) was delayed about 20 % while that of TC5-2
(49 mass% WPG) about 45 %. Burning times (time of
flameout) for all tested composites were greater than untreated
wood increasing with WPG. What is more, in contrast to the
second PHRR the time to second PHRR increases linearly
with the increasing fire residue. Thus, it is proposed that the
HRR was mainly controlled by the protection properties of
the residue, whereas the time to PHRR goes along also with
the heat capacity of the residue and thus its amount.
In contrary to HRR, total heat release evolved (THE =
THR at the end of the test) of all the investigated samples was
very slightly decreased except for TC5-2 composite. THE for
TC5-2 (WPG 49 mass%) prepared with double impregna-
tion was found to be 19 % greater in magnitude than THE
(57 MJ m-2) for untreated wood. This unexpected THE
value for TC5-2 composite supports the existence of organic
residuals as a large fraction of WPG in the composites as
already predicted by thermal analysis. The effective heat of
combustion deduced from THE/TML remained invariant for
all tested composites (Table 3) ruling out significant flame
retardancy mechanisms in the gas phase such as flame
inhibition or fuel dilution. The CO production for titania-
based composites TC1 and TC2 was minutely increased
while surprisingly, decreased for rest of tested composites
indicating slight changes in the volatile pyrolysis products.
Unlike to untreated wood, in particular total smoke release
was reduced considerably for tested titania composites. A
remarkable reduction in the range of 50–62 % in total smoke
release was achieved for all composites compared to
untreated wood indicating a worth seeing improvement in
reducing fire hazards by titania-based sol–gel treatment to
wood.
Figure 5 shows the residues of untreated wood and the
tested composites after the test in a cone calorimeter.
-1.0
-0.5
0.0
rela
tive
mas
s lo
ss
temperature / °C
TC6TC4TC3UN
(a)
1
2
3
0 200 400 600 800
200 400 600 800
End
o.E
xo.
(b)
TC6
TC5
TC4
TC3
TC2
TC1
untreated (UN)
temperature / °C
Fig. 3 Thermal behavior of sol–gel titania wood composites probed
by TGA (a) and DTA (b) at a heating rate of 2 K min-1
J Mater Sci (2012) 47:6849–6861 6855
123
Untreated wood was combusted to light brownish–black
residues. Cracking of the residue toward the end of burning
and subsequent after glowing finally decomposed the
material to black ash containing larger pieces disintegrate
to each other. Combustion of composites derived from
dilute precursor solutions (WPG up to 10 mass%) yields
white residual masses with more compact and integrated
structures. White appearance of residues attributed to tita-
nia gels network homogeneously distributed to whole wood
matrix. Larger cracks were found in the residual masses of
TC5 and TC5-2 (WPG 27 and 49 mass%, respectively). It
is assumed that thicker TiO2 coatings in the interior of
wood matrix tend to be more frail and hence to crack when
heated. As a result, more cracked and rigid residual mass
were obtained specially for TC5 composite accompanied
by a low reduction in its second maximum of HRR.
A comprehensive assessment of fire risks for untreated
wood and titania-based wood-inorganic composites for all
tested samples in terms of THE, PHRR, and time to igni-
tion (TTI) at an external heat flux of 50 kW m-2 is dis-
played in Fig. 4d. Method is also useful to identify the
system with minimum fire hazards [33]. The ordinate (total
0
50
100
150
200
250 UN TC1 TC2 TC5
HR
R /
kWm
-2
time / s
(a) (b) UN TC1-2 TC5-2
time / s
52
56
60
64
68
0 200 400 600 0 200 400 600
6 8 10 12 1416 20 24 28 32 36160
180
200
220
240
260
280
residue / mass%
2.P
HR
R /
kWm
-2
280
320
360
400
440UN
TC1
TC1-2
TC2
TC5
TC5-2
WPG
2.PHRR/TTI / kWm-2s-1
TH
E /
MJm
-2
UN
TC1
TC1-2
TC2TC5
TC5-22.
TP
HR
R /
s(c) (d)
Fig. 4 Assessing fire risks from
cone calorimeter data obtained
for titania-based wood-
inorganic composites at
50 kW m-2. Heat release rate
(HRR) as a function of time for
a single impregnated, b double
impregnated specimen,
c residue versus second maxima
of HRR and THRR, and d total
heat evolved against peak of
heat release rate per time to
ignition
6856 J Mater Sci (2012) 47:6849–6861
123
heat evolved) serves as a measure for the propensity to
cause long-lasting fire. The abscissa, peak of heat release
rate per time to ignition marking the propensity to cause a
quickly growing fire in the cone calorimeter experiment.
Based on this result, TC1-2 causes minimum fire hazards
among other tested composites because a strong decrease
not only in 2.PHRR/TTI as well as in THE was achieved in
a cone experiment for this system. Least emission of
smoke for TC1-2 strongly supported this finding. On the
other hand, 2.PHRR/TTI for TC5-2 is minimal but due to
unusual increase in THE relative to untreated wood proves
it the worst system in terms of fire hazards.
On the basis of results described above, it was extracted
that sol–gel derived titania wood-inorganic composites
showed slight fire retardancy in terms of reduced flame
spread as well as suppressed fire growth. Heat release rates
and second maximum of HRR for these systems were
decreased and corresponding burning times and time to
second maximum of HRR were increased moderately due
to change in their residual properties under force-flame fire
conditions. No further improvement was realized in terms
of fire load as total heat evolved was unchanged very
similar to untreated wood.
Flammability
Composites materials show better performance in terms of
flame retardancy, what is demonstrated in the following.
LOI value of untreated pine sapwood and their titania
derived composites are summarized in Table 4 along with
UL 94 results. Measured LOI value for untreated pine
wood is 23 vol.%. This value is smaller than the reported
value (29.4 vol.%) for pine wood [40], demonstrating that
this method delivers only relative (LOI) values depending
on origin and dimensions of the specimens. A remarkable
increase in LOI values ranging from 26 to 35 vol.% was
observed for singly impregnated composites. Increase in
LOI was even more promising in case of doubly impreg-
nated composites. Doubly impregnated composite TC1-2
only with 4 mass% WPG achieved LOI value of 29 vol.%
compared to 23 vol.% for pure wood exhibited 27 % rel-
ative improvement in oxygen index. Composites with
higher loadings such as TC5 (WPG 27 mass%) and TC5-2
(WPG 49 mass%) obtained the highest oxygen index val-
ues (33 and 38 vol.%, respectively) from the tested samples
(see Table 4). These results show that titania wood-inor-
ganic composites are excellent flame retardants even with
very low loadings of titania precursor materials.
In order to find the origin of flame resistance as well as
clarify the influence of parameters on flame retardancy,
LOI is plotted against and WPG and fire residue, respec-
tively, as well as second maximum of HRR (2.PHRR)
measured by cone calorimeter for the tested compositesTa
ble
3C
on
eca
lori
met
erd
ata
for
tita
nia
-bas
edw
oo
d-i
no
rgan
icco
mp
osi
tes
atan
irra
dia
tio
no
f5
0k
Wm
-2
Sp
ecim
enT
TI
(s)
TF
O
(s)
Mas
s
(g)
TM
L
(g)
Res
idu
e
(mas
s%)
TH
E
MJ/
m2
TH
E/T
ML
MJ/
m2g
1.P
HR
R
kW
/m2
1.T
PH
RR
(s)
2.P
HR
R2
.TP
HR
R
kW
/m2
TS
R
m2/m
2T
CO
P
(g)
CO
Y
kg
/kg
UN
20
40
94
9.1
40
.61
7.3
57
.01
.42
09
29
26
53
07
29
80
.21
10
.00
52
TC
12
34
71
50
.34
0.2
20
.05
2.9
1.4
22
83
32
14
35
31
42
0.2
25
0.0
06
1
TC
1-2
23
49
45
0.2
38
.12
4.1
51
.01
.32
06
32
19
63
59
11
20
.20
20
.00
52
TC
22
35
15
53
.84
0.4
24
.95
5.6
1.4
19
43
51
86
36
71
36
0.2
20
0.0
05
4
TC
53
35
03
61
.74
3.9
28
.85
6.6
1.3
17
84
12
26
39
31
47
0.1
80
0.0
04
0
TC
5-2
32
57
16
9.4
45
.93
3.9
67
.61
.52
06
60
19
94
46
14
60
.17
70
.00
38
TT
Iti
me
toig
nit
ion
,T
FO
tim
eo
ffl
ameo
ut
(bu
rnin
gti
me)
,T
ML
tota
lm
ass
loss
,T
HE
tota
lh
eat
evo
lved
,P
HR
Rp
eak
of
hea
tre
leas
e(fi
rst
and
seco
nd
),T
PH
RR
tim
eo
fP
HR
R(fi
rst
and
seco
nd
),
TS
Rto
tal
smo
ke
rele
ase,
TC
OP
tota
lC
Op
rod
uct
ion
,an
dC
OY
CO
-yie
ld
J Mater Sci (2012) 47:6849–6861 6857
123
(Fig. 6). Whereas the LOI roughly depends linearly from
the WPG, the linear relationship between LOI and residues
produced in cone calorimeter is rather perfect indicating
that LOI is strongly controlled by the char yield [41]. As
described before, a change in residual properties signifi-
cantly affects the second maxima of HRR in a cone
experiment. It is shown that the correspondence between
the LOI and 2.PHRR is lacking (Fig. 6b) similar to Fig. 4c.
Thus, it was concluded that the treatment of wood with
more concentrated titania precursor solutions as well as
higher titania loadings in the derived composites
remarkably enhance the flame retardancy of these materials
with respect to reaction to a small flame such as LOI.
It is reported before that those composite materials may
pass the UL 94 test with V-0 only when they have achieved
LOI values of larger than 40 vol.% [42]. Beside remarkably
high LOI values of the titania-based composites, their self-
extinguishing performance is still insufficient as depicted
by UL 94 test (see Table 4). Even though all the tested
samples have achieved horizontal burning HB in accor-
dance to UL 94 classification, the composites showed also
in UL 94 clearly reduced fire hazards. Most composites
Untreated wood (UN) TC5
TC5-2 TC2
TC1-2TC1
Fig. 5 Photographs of residue obtained in the cone calorimeter for untreated wood and sol–gel derived titania-based wood-inorganic composites
6858 J Mater Sci (2012) 47:6849–6861
123
specimens show self-extinguishing in HB before reaching
any testing marks in contrast to pure wood. Furthermore,
the flame spread velocity for the few burning specimen was
much slower than for untreated wood. These result well
corresponds to the LOI results and may correspond with
the cone results for the second maximum of HRR implying
that titania-based wood-inorganic composites are slow
burning or flame retardant materials. On the other hand, the
composites did not achieve rating vertical burning in
accordance to UL 94 showing that these systems were still
flammable materials in the more challenging vertical con-
figuration with ignition from the bottom. The difference in
LOI/horizontal burning UL 94 and vertical burning UL 94
testing indicates a strong influence of upward and down-
ward flame configurations on the flammability of the
investigated materials as it is often observed for charring
approaches. It was reported that heat transfer and burning
rates under downward flaming configuration (LOI and HB)
were greatly altered from the upward flaming configuration
(vertical burning UL 94) [43]. Moreover, in both test
geometries, the residual impact was changed that consid-
erably influenced the flame retarding effect [44].
Nevertheless, titania-based sol–gel wood composites are
valuable environmentally friendly materials that show
significant improvements in their most of properties
including fire and flame retardance at low titania loadings
(WPG). In practice, 3–4 mass% critical loading of titania
gels by single impregnation of titanium alkoxide precursors
to wood is sufficient to produce fire retardant wood-
inorganic composites with fewer fire risks such as low CO
and smoke release. An approach to use higher loadings by
infiltrating more concentrated titanium alkoxide solution is
senseless and hardly improves the fire performance of the
derived composites. In comparison with typical boron-
based fire retardants (boric acid, boric acid ? borax mix-
tures etc.) under similar loadings (3–5 mass%) similar fire/
flame retardancy enhancements in terms of second peak
HRR (2.PHRR) reduction as well as LOI increase in the
small scale flammability conditions were realized [40, 45,
46]. However, at higher loadings later performed better
flame retardance (based on their LOI values) than the under
investigated TiO2 wood composites. However, due to
better leach resistance, TiO2-based wooden materials may
have edge over boron-based ones especially under moist
conditions. Wood treated with nitrogen- and phosphorous-
based fire retardants showed almost three times more flame
resistance (increase in LOI) than TiO2 wood composites at
the similar respective loadings of nitrogen and phosphorous
in the tested specimens [47–49]. This comparison reveals
that TiO2 wood composites possess fire resisting character
but there current fire performance needs to improve further
to beat or become competitive with conventional fire
retardants.
In previous reports, on the basis of DTA/DTG results it
was stated that fire properties of sol–gel derived wood
composites were improved [2, 15]. However, the evidence
of the results is limited. Cone calorimeter, LOI and UL 94
results reported herein even provided first deep insight to
clarify the complex description of fire behavior of titania
impregnated wood composites.
21
24
27
30
33
36
39(a)
single impregnated double impregnated
LOI /
vol
.%
WPG / mass%0 6 12 18 24 30 36 42 48
21 24 27 30 33 36 3916
20
24
28
32
36
LOI / vol.%
resi
dues
/ m
ass%
160
180
200
220
240
260
2.P
HR
R /
MJm
-2
(b)UN
TC1
TC2
TC1-2
TC5
TC5-2
Fig. 6 LOI plotted against a weight percentage gains for the
composites and b residue as well as second maximum of HRR
(2.PHRR) obtained from the combustion of titania wood-inorganic
composites in a cone calorimeter at an irradiation of 50 kW m-2
Table 4 Flammability (reaction to small flame) of titania wood-
inorganic composites
Composites Impregnation
cycles
LOI ± 1/
vol.%
UL 94
classification
UN 0 23 HB
TC1 1 26 HB
TC1-2 2 29 HB
TC2 1 28 HB
TC3 1 29 –
TC4 1 30 –
TC5 1 33 HB
TC5-2 2 38 HB
TC6 1 35 –
J Mater Sci (2012) 47:6849–6861 6859
123
Since the main focus of this study was to elucidate the
fire/flame resistive potential through well accepted fire
experiments. However, it is realized that a number of wood
properties alter by applying sol–gel approach. For instance,
beside fire and flame retarding improvements, sol–gel
derived TiO2 wood-inorganic composites displayed excel-
lent anti-fungal characteristics [50]. These materials very
minutely degraded against brown rot fungi C. puteana and
P. placenta compared to untreated pine sapwood that
severely lost its mass due to fungal deterioration. Anti-
fungal efficacy of TiO2 wood composites was attributed to
better shielding by the gel thin layer precipitated on the cell
wood walls as well as the consequence of biocidal action of
the organic residues that acted like fungicides. In addition,
these composites absorbed significantly less moisture in the
moisture sorption test in comparison to untreated wood
exhibiting more dimension stability [22, 50]. Furthermore,
it is expected that these composites have better mechanical
properties like modulus of elasticity, modus of rapture, and
hardness etc., thereby the localization of TiO2 gels into the
cell wall cavities that provide additional strength to the
matrix. Similar mechanical improvements are reported in
literature [51]. Also, composites assume to be leaching
resistant because of the insolubility of TiO2 precipitates in
water as well as due to their better fixation into the matrix.
Conclusions
ESEM/EDX investigations showed that precursor solutions
can infiltrate the whole wood matrix and is fixated mainly
in the lumen forming thin films. It was visualized that these
depositions were not crack free and increased drastically in
composites prepared with precursor solutions of high alk-
oxide concentrations.
Thermo-oxidative decomposition of tested composite
materials was characterized employing thermal analysis. It
was obvious from TGA results that the amount of residues
formed during thermo-oxidative decomposition is lower
than the corresponding WPG values. This indicated that a
fraction of WPG was composed of organic residues, mainly
un-hydrolyzed alkoxide and solvents that released during
the measurement and left only stable inorganic residues.
For the development of titania wood composites with
optimum fire resistance capabilities, it is essential to avoid
possible content of organic residues by improving the
treatment process. DTA thermograms of composites
showed that the intensities were decreased markedly as
well as a shift occurred in the peaks with respect to tem-
perature qualitatively identified change in oxidative
decomposition reaction of these materials.
Cone calorimeter results revealed that heat release rates
after the initial PHRR and in particular, the heat release
rate of the second PHRR in a developing fire were mod-
erately reduced as well as the burning time was increased
significantly indicating some improvements in flame
retardance due to protection properties of the fire residue.
In addition, fire hazards such as CO and smoke production
were remarkably reduced. Beside these improvements, fire
behavior in terms of ignition, initial HRR increase, fire
growth index and fire load remained unchanged for the
tested composites.
An impressive improvement in LOI values up to 64 %
(relative to LOI for untreated wood) was noticed showing
better flame retardant properties of these materials when
reacting to small flame. All the UL 94 tested systems have
achieved HB (horizontal burning) but the composites show
dramatically decreased burning rates in the horizontal
configuration.
It is concluded that fire behavior and the flammability of
these systems readily depended on the fire scenarios. The
observed improvement in fire performance of the sol–gel
derived titania wood composites are due to stabilization of
the deposited thin layers of titania gels as well as fire
response of the depositing material. On the basis of
acquired understanding, it is expected that a considerable
improvement in the fire characteristics of these materials
can be achieved by minimizing the organic fraction of
WPG (by better processing) as well as stabilizing the
deposition layers of titania (crack free) by using more
dilute titanium alkoxide precursors.
Acknowledgements The authors thank Dr. Ina Stephan and Heidi
Lorenz of BAM Federal Institute for Materials Research and Testing
for supplying wood samples and assisting TG measurements.
References
1. Mahltig B, Swaboda C, Roessler A, Bottcher H (2008) J Mater
Chem 18:3180
2. Saka S, Ueno T (1997) Wood Sci Technol 31:457
3. Miyafuji H, Saka S (2001) J Wood Sci 47:483
4. Saka S, Miyafuji H, Tanno F (2001) J Sol-Gel Sci Technol
20:213
5. Klein LC (1988) Sol-gel technology for thin films, fibers,
preforms, electronics, and specialty shapes. Noyes Publishers,
Park Ridge
6. Dunn B, Zink IJ (1997) Chem Mater 9:2280
7. Ingo GM, Riccucci C, Bultrini G, Dire S, Chiozzini G (2001)
J Therm Anal Calorim 66:37
8. Tseng TK, Lin YS, Chen YS, Chu HA (2010) Int J Mol Sci
11:2336
9. Simonsen ME, Søgaard EG (2010) J Sol-Gel Sci Technol 53:485
10. Saka S, Sasaki M, Tanahashi M (1992) Mokuzai Gakaishi
38:1043
11. Koji O, Saka S (1993) Mokuzai Gakaishi 39:301
12. Saka S, Yakake Y (1993) Mokuzai Gakaishi 39:308
13. Koji O, Saka S (1994) Mokuzai Gakaishi 40:1100
14. Miyafuji H, Saka S (1996) Mokuzai Gakaishi 42:74
15. Miyafuji H, Saka S (1997) Wood Sci Technol 31:449
6860 J Mater Sci (2012) 47:6849–6861
123
16. Miyafuji H, Saka S, Yamamoto A (1998) Holzforschung 54:410
17. Miyafuji H, Saka S (1999) Mater Sci Res Int 5:270
18. Tanno F, Saka S, Takabe K (1997) Mater Sci Res Int 3:137
19. Donath S, Militz H, Mai C (2004) Wood Sci Technol 38:555
20. Donath S, Militz H, Mai C (2006) Holzforschung 60:210
21. Bucker M, Bocker W, Reinsch S, Unger B (2003) In: proceedings
of the 1st European conference on wood modification, Ghent,
2003, p 255. ISBN 9080656526
22. Hubert T, Unger B, Bucker M (2010) J Sol-Gel Sci Technol
53:384
23. ISO 5660–1 (2002) Reaction- to-fire tests—heat release, smoke
production and mass loss rate—part 1: heat release (cone calo-
rimeter method). ISO, Geneva
24. Babrauskas V (1984) Fire Mater 8:81
25. ISO 4589–2 (1996) Plastics-determination of burning behaviour
of oxygen index—Part 2: ambient temperature test. ISO, Geneva
26. IEC 60695–11-10 (2003) Amendment 1—Fire hazard testing-
part 11–10: test flames—50 W horizontal and vertical flame test
methods. ICE, Geneva
27. Bartl HM, Boettcher WS, Frindell LK, Stucky DG (2005) Acc
Chem Res 38:263
28. Fuertes CM, Colodrero S, Lozano G, Gonzalez-Elipe A, Grosso
D, Boissiere C, Sanchez C, Soler-Illia G, Miguez H (2008) J Phys
Chem C 112:3157
29. Krins N, Faustini M, Louis B, Grosso D (2010) Chem Mater
22:6218
30. Wood Wegner TH (1989) In: Mayer JA (ed) Encyclopedia of
polymer science and engineering. Wiley, New York
31. Shan DK, Gu S, Bridgwater AV (2010) J Anal Appl Pyrolysis
87:199
32. Kaur B, Gur IS, Bhatnagar HL (1986) J Appl Polym Sci 31:667
33. Schartel B, Hull TR (2007) Fire Mater 31:327
34. Parker WJ, Tran HC (1992) In: Babrauskas V, Grayson SJ (eds)
Heat release in fires, Chapt 4. Elsevier, New York, p 331
35. Lyon RE (2004) In: Harper CA (ed) Handbook of materials in fire
protection, Chapt 3. McGraw-Hill, New York, p 51
36. Braun U, Bahr H, Sturm H, Schartel B (2008) Polym Adv
Technol 19:680
37. Wu GM, Schartel B, Yu D, Kleemeier M, Hartwig A (2012)
J Fire Sci 30:69
38. Schartel B, Bartholmai M, Knoll U (2006) Polym Adv Technol
17:772
39. Bartholmai M, Schartel B (2004) Polym Adv Technol 15:355
40. Baysal E (2002) J Fire Sci 20:373
41. Van Krevelen DW (1975) Polymer 16:615
42. Braun U, Balabanovich AI, Schartel B, Knoll U, Artner J,
Ciesielski M, Doring M, Perez R, Sandler JKW, Altstadt V,
Hoffmann T, Pospiech D (2006) Polymer 47:8495
43. Weil ED, Hirschler MM, Patel NG, Said MM, Shakir S (1992)
Fire Mater 16:159
44. Braun U, Schartel B, Fichera AM, Jager C (2007) Polym Degrad
Stab 92:1528
45. Wang Q, Li J, Winandy JE (2004) Wood Sci Technol 38:375
46. LeVan SL, Tran HC (1990) The role of boron in flame-retardant
treatments. In: proceedings of the 1st international conference on
wood protection with diffusible preservatives, Forest Products
Research Society, Madison, 28–30 Nov 1990, p 39
47. Gao M, Pan DX, Sun CY (2003) J Fire Sci 21:189
48. Gao M, Zhu K, Sun YJ (2004) J Fire Sci 22:505
49. Gao M, Sun CY, Wang CX (2006) J Therm Anal Calorim 85:765
50. Shabir Mahr M, Hubert T, Stephan I, Militz H (2012) Int Biodeter
Biodegrad (accepted)
51. Wang X, Liu J, Chai Y (2012) BioResources 7(1):893
J Mater Sci (2012) 47:6849–6861 6861
123