Académique Documents
Professionnel Documents
Culture Documents
Abstract
The present study focuses on the mechanical properties and fire behaviour of Polyamide-
6/Kaolinite composites. Several kaolinites (exhibiting different shape factor) and different
surface treatments (showing more or less affinity with the matrix) reveal a high level of
disparity in their morphological properties. These morphologies characterised by their
dispersion (interparticular distance) and their distribution (median diameter) are directly
related to mechanical properties. Elastic properties are enhanced, as expected, by kaolinites
addition (with a most effective reinforcement for the kaolinites with the higher shape factor)
but quite unaffected to their surface treatment. The same trend is observed for the resilience.
The yield stress decreases as the median diameter increases: a good interfacial adhesion led to
a better particle desagglomeration and so a higher yield stress. All composites showed a
decrease in the peak heat release rate in comparison with the unfilled polyamide which seems
to be a direct consequence of both particles incorporation and their dispersion/distribution.
*Corresponding author
Tel.: 0033(0)466785631, Fax: 0033(0)466785680
E-mail address: Anne-Sophie.Caro@mines-ales.fr
URL: http://www.ema.fr
1. Introduction
Polymers are widely used since 50 years due to their good mechanical and physical properties
and easy processing. Nevertheless, they present a very low thermal resistance: heat and
oxygen combination leads to ignition and heat release. In this framework, a great challenge
concerns the improvement of polymer fire resistance.
Flame retardancy can be performed through different mechanisms:
Material surface protection: this process raises the question of adhesion/wear
between protection and sample
Material volume protection: flame retardant additives are directly introduced into
the polymer during the mixing stage. Most of the time, these additives lead to the
degradation of mechanical properties of final product and to an increased toxicity.
Mineral particles are among the main flame retardants. The action of these additives is mainly
physical. They can form a non-conductive coating and reduce heat transfer; additives
degradation reaction can also be endothermic and contributes to decrease the heat balance;
additives degradation can also produce inert gases which may dilute combustible gases. In
some cases, metal oxides could also act in a chemical way by absorbing active radicals or
promoting charring. Mineral particles (such as metal hydroxides (ATH, MDH)) offer a great
alternative in the framework of environmental constraint but the needed high level of
incorporation (until 65% by mass fraction) usually leads to poor mechanical properties. On
another hand, nano-mineral particles allow reaching high level of flame retardancy at low
rates (less than 5% by mass fraction). For example organically modified monmorillonite
(oMMT) are used to improve flame retardancy in various polymers: composites PP-g-
MA/oMMT [1], Polyimide(PI)/oMMT [2], PA6/oMMT [3], EVA/oMMT [4] where oMMT
particles are either intercalated or exfoliated, PMMA/Al2O3,TiO2 [5, 6]). Their successful
action in the heat balance can be attributed to
The creation of a diffusion barrier layer [4], [5], [7]
The modification of the polymer degradation: heterogeneous catalysis associated
to a char structure [4], [7]
The modification of the thermal diffusivity [5], [8]
The reduction of the macro molecular mobility (around the charges) [5]
The increase of polymer viscosity [9]
The trapping of radicals formed during the thermal degradation [5]
2
All these modes of action are highly dependent on the nanoparticle distribution
[5, 6].
Nevertheless nanoparticles are usually used in combination with classical flame retardants
because even if they lead to a more progressive heat release, the global heat balance remains
the same [10, 11, 12]. Moreover their processability limits their level of incorporation (always
under 10% by mass fraction).
Recent researches try to promote another kind of mineral particles: the kaolinites. The
papermaking industry is the main actor regarding its use and the advent of new technologies
promises a great disponibility of this raw material. Regarding its low shape factor, the main
interest is its high potential of incorporation rate in polymer matrix. The broadly potential use
of this material requires investigation on mechanical composite performances. Even if native
particles are submicronic, their level of dispersion in the matrix and their distribution are
usually responsible of poor mechanical properties. A surface treatment is therefore advisable.
The present study focuses on the use of kaolinites (30% by mass fraction) to improve both
mechanical and thermal behavior of PA6 material. At first the composites morphology is
characterized. Then mechanical (static/dynamic) and thermal (cone calorimeter) tests are
performed. Relation between morphology and thermal or mechanical properties is also
discussed. This article is a first step: the kaolinite is introduced alone, and so only a more
progressive heat release is expected. For a composite PA6/kaolinite a good processability is
still reached at 30wt%. A second step will be the addition of flame retardancy agent
(phosphonate additives for example) in the PA6/kaolinite composite.
2. Experimental
3
Fig. 1 Standard tests specimens (a) tensile (ISO 527-2 1A), thickness 4mm (b)
Charpy impact (c) cone calorimeter tests, thickness 4mm
Two kinds of lamellar kaolinite nanoparticles were selected for this study:
- a commercial one (K1): PARALUX with a specific surface area (BET) of 12.2 1.5
m2/g, a shape factor of 10 and an average primary particle size of 0.9 m,
- a non-commercial one (K2) with a specific surface area (BET) of 19.9 2 m2/g, a
shape factor of 5 and an average primary particle size of 0.2 m.
Both of them are supplied by Imerys and are mainly composed of alumina and silica
(Al2O3 wt % >38, SiO2 wt % >43).
Surface treatment was carried out as follows: kaolinites were dispersed in an
ethanol/water mixture (96/4). The solution was then placed under stirring and the pH was
changed by means of acetic acid to reach pH of 4.5. Coupling agents (treatment A: (3-
aminopropyl)triethoxysilane and treatment E: (3-glycidyloxypropyl)trimethoxysilane) were
added with a mass concentration of 10% of the kaolinite weight. This mixture was stirred
under reflux for 6h. At the end of the reaction, the solvent was evaporated and the kaolinite
was washed three times with ethanol to remove excess reagent.
Fillers are incorporated at 30% by mass fraction (equivalent to a volume incorporation rate
p=15%), which corresponds to a good compromise between processability and expected
flame retardancy. Six different formulations were thus realized and are presented in the Table
1.
4
Grafting -- -- A E -- A E
Table 1 : Materials under study
2.2 Characterization
Fig. 2 SEM micrographs of PA with 30%wt of K2, scale factors (a) 100 (b) 1000 (c) 20000
SEM micrographs (Fig. 3a) are converted in black and white images (Fig. 3b) and kaolinites
stacks sections are identified, filtered by area surface value (A) with respect to the scale of
observation (from 100 to 5000m (scale factor 100); from 15 to 100m (scale factor 1000)
and from 1 to 15 m (scale factor 20000)) and numbered. This procedure is repeated over all
the micrographs and the number of particles identified for each magnification is scaled by the
surface analysis area.
5
Fig. 3 Image analysis on SEM micrographs (a) initial image (b) selected areas
As selected objects have not a well-defined shape, it is useful for comparison to consider each
object as a disc occupying the same measured surface area. These discs are well described by
their diameters, denoted d. Collecting data of each magnitude analysis, a particle diameter
size distribution can be obtained from d=1m to 80m. For each nanocomposite a distribution
and a cumulated density in d are obtained (see Fig. 4 in the case of PAK1 specimen).
Fig.4 Diameter size distribution for the PAK1 sample: histogram of the normalized
frequencies, cumulative distributions (experiments (full line), identification (dotted line))
1 1 ln( d ) M
f (ln( d )) erf
(1)
2 2 SD 2
where M and SD are the mean and the standard deviation of the log-normal distribution and erf
is the Gauss error function; for each formulation two parameters (M and SD) are then extracted
6
from a well fitted result (see Fig. 4, full and dotted lines). A statistical analysis can therefore
be performed from these variables:
The equivalent diameter dm which is the median value of the distribution
dm eM (2)
2 2
(3)
S (e SD 2) e SD 1
where C is a constant value depending on the compactness stacking. For sake of simplicity
and for all the formulations C was assumed equal to 1 (cubic stacking).
Uniaxial tensile tests were carried out on a Zwick TH010 universal testing machine
according to the ISO 527 standard. The crosshead speed depends on investigated properties: 1
mm/min to describe linear behavior and 50mm/min for the non-linear behavior until breakage.
The software used is TestXpert and allows the recording of time, load (F) and elongation ().
The nominal stress is defined by the following expression: F / S0 where S0 is the initial
sample ISO 527-2 1A section (40mm). Youngs modulus (E) value is given by the slope,
7
calculated between 0 and 1 % elongation of the curve giving F vs. axial . We denote by Ec
the composites modulus, EPA the neat PA6 modulus and if necessary EK1 (equal to EK2) the
fillers modulus in the longitudinal direction. Out of this range of elongation, the whole
stress/deformation profile of these materials is given in Fig. 5. As commonly observed for
these kinds of material, the nominal stress reached a maximum (called the yield stress)
corresponding to the value yield at the deformation yield. We denote by c yield the composites
yield stress and PA yield the neat PA6 yield stress.
100
100
80 80
60 60
(MPa)
(MPa)
PA
40 40 PA-K2
PA PA-K2A
PA-K1 PA-K2E
20 PA-K1A 20
PA-K1E
0 0
0 5 10 15 20 25 0 5 10 15 20 25
(%)
(%)
Fig. 5 Stress vs. strain in uniaxial tensile tests for PA6 composites
where v (t ) is the instantaneous velocity of the impactor and v0 its impact velocity. Fig. 6
presents typical trend of the impact force vs. time for the PA6 composites; Sure enough the
impact behaviour is highly dependent on filler and/or surface treatment. All specimens were
totally broken after the impact and fillers provided additional toughening effects.
8
250
PA
PA-K1
200 PA-K1A
PA-K1E
150
Force (N)
100
50
0
5.0 5.5 6.0 6.5 7.0
Time (ms)
From these curves a useful global parameter, the energy, can be deduced:
u
(8)
J F (u)du
0
Name M SD del dm S A L
9
m m m m
PA - - - - - - -
PAK1 1.72 1.62 1.1 5.6 10.1 5.1 97.1
PAK1A 1.86 1.61 1.1 6.4 8.1 5.9 106.1
PAK1E 1.68 1.67 1.1 5.4 11.9 4.9 111.4
PAK2 1.90 1.60 0.5 6.7 7.7 13.4 107.7
PAK2A 1.77 1.56 0.5 5.9 7.7 11.8 83.7
PAK2E 1.76 1.64 0.5 5.8 10.3 11.6 109.0
Table 2: Morphological characteristics of the PA6 composites
The mechanical/thermal properties of all the PA6 composites have been summarized in Table
3.
10
PAK2E 4500 86.6 4.9 3.5 470 58 125 28.8
Table 3: Mechanical/thermal characteristics of the PA6 composites
E
MPa
K1 , K2 180 0.20
PA 3.4 0.35
Table 4: Elastic parameters for the Mori Tanaka modeling, p=15%
It appears that surface treatment doesnt impact so much material elasticity. In fact, as
reported in the literature, the Youngs modulus is evaluated at a low level of strain inducing a
little perturbation of the interface area [19]. Nevertheless surface treatment can modify both
matrix and filler dispersion [20]. Considering results of table 2, there is no obvious relation
between the median diameter and the Youngs modulus.
7000
7000
6000
6000
5000
5000
4000
Ec (MPa)
4000
Ec (MPa)
3000 3000
2000 2000
1000 1000
0 0
(a) PA PAK2 PAK2A PAK2E Mori Tanaka (b) PA PAK2 PAK2A PAK2EMori Tanaka
11
Usually, yield stress increases and yield strain decreases with the incorporation rate of
well dispersed rigid particles. The reason reported in the literature is the good adhesion
between fillers and matrix (which is responsible for the stress increasing) and the reduction of
the volume matrix which is enable to deform (which is responsible for the strain decreasing).
Some authors involve particle size and shape to explain the stress increasing [21-24]. The key
parameter for the yield stress seems to be the median diameter dm (Fig. 8a) for our study: c
yield depends linearly on the median diameter. Concerning the yield strain the same trend is
observed for each kind of kaolinites: c yield depends linearly on the level of aggregation (A)
(Fig. 8b).
100 40
(A) (B)
95 K1E 35 K2
K1
90 K2A 30 K2A
K2E
yield (MPa)
85 K2E 25
80 20
K1A
A
75 15
70 10 K1A
K2 K1E
65 5 K1
60 0
5.0 5.5 6.0 6.5 7.0 1 2 3 4 5 6
dm (m) yield (%)
Fig. 8 (a) Yield stress as a function of median diameter (b) Yield strain as a
function of agglomeration rate
12
4.5
4.0 K1
K1E
3.5 K2A
K2E
3.0
2.5 K1A
B
2.0
1.5
1.0 K2
0.5
0.0
5.0 5.5 6.0 6.5 7.0
dm (m)
It seems that the cohesion between matrix and aggregates decreases as the median
diameter increases. In fact the better is the interface adhesion the better is the dispersion.
Cohesion parameter is therefore a consequence of the surface treatment via its impact on
particle dispersion. For both kaolinite, E treatment leads to the lowest median diameter: Commented [AC1]: Belka il faut faire un laus l-dessus :
Marcos dit que les groupes poxy du traitement E peuvent avoir une
interaction plus importante avec les groupes amines du PA6. Pour le
traitement A cest diffrent. Tu confirmes ?
3.2.3 Resilience
Literature results reveal the undeniable relation between breakage and interfacial
quality in composites. In the case of no adhesion, rupture will be conditioned by the matrix
properties; if the interface is fragile, stress concentration around particles will be increased
and will lead to a fragile breakdown; On the contrary, a ductile interface will produce
particles cavitation [26]. Beyond these observations, it seems that rupture is first of all
conditioned by the biggest particles [27]. Some authors define a morphological criterion in
relation with the fragile/ductile composites breakdown [28, 29]. According to these authors, a
morphological analysis is performed from the lowest magnification (100) in order to qualify
only big agglomerates dispersion. These results are summarized in Table 4 with the associated
mode of failure.
Name M SD dm S L J Failure
m m m J mode
PA - - - - - 2.9 -
13
PAK1 3.16 1.99 23.5 12. 1723. 1.7 ductile
PAK1A 3.13 1.94 22.9 7.2 1344. 3.2 fragile
PAK1E 2.88 1.96 17.8 10.1 1149. 3.8 fragile
PAK2 3.48 1.99 32.7 7.1 2588. 1.2 ductile
PAK2A 2.66 1.81 14.3 8. 492. 4.6 fragile
PAK2E 3.00 1.95 20.1 12.5 1303. 3.5 fragile
Table 5: Morphological characteristics of the PA6 composites with the 100
magnification
A great correlation was found between resilience and interparticular distance through a sigmoid
function (Fig. 10). This graph confirms the existence of a critical interparticular value related to the
ductile-brittle material transition.
1.6
Ductile Fragile
1.2
J (J)
0.8
0.4
0.0
0 500 1000 1500 2000 2500 3000
L (m)
case of mineral composites the first maximum corresponds to the formation of a protective
Commented [I3]: Je nai pas observ la formation dune couche
char layer and the second to sample swelling. Nevertheless, in PA6 kaolinites composites la surface ni lexpansion des chantillons PA-Kaolinites. La
prsence dun plateau (ou deux pics selon la formulation) est
none of this phenomenon has been observed. The unfilled PA6 is fully pyrolized at the same difficile expliquer. Rodolphe disait que cela pourrait tre d
larrive de la flamme au fond de lchantillon et la chaleur,
time as its HRR curve reveals a single sharp peak [30]. The maximum peak reduction nayant plus une faon de svacuer est retransmisse au polymre
augmentant ainsi le HRR. Mais cela reste encore difficile a montrer.
(corresponding to the second one) is of 44% for PAK1, 54% for PAK1A, 65% for PAK1E Commented [AC4]: Rodolphe peux tu dire quelque chose l-
dessus ?
Merci
14
(42% for PAK2, 66% for PAK2A, 65% for PAK2E ). Regardless of their shape factor, the
kaolinites (and even more the treated ones) reduce the pHRR significantly. Moreover it seems
that pHRR evolution is well correlated with the median diameter (from 100 magnification)
(see Fig. 12). The better is the dispersion of the biggest agglomerates (the lower is the median
diameter) the lower is the pHRR. Some authors found qualitatively same trend on PP/MMT
composites even if there does not exist any theoretical background to describe the link
between the pHRR reduction and particle dispersion [32]. In [32] authors suggest that
nacomposites are better than microcomposites regarding pHRR reduction. Nevertheless, they
point out that if nanodispersion is achieved, the nanomorphology (exfoliation, intercalation
and presence of tactoids for clay particles) does not play any significant role. In conclusion, to
decrease the pHRR, a great challenge is the kaolinite global desagglomeration before and Commented [AC5]: Marcos je mets global pour ne pas rentrer
dans les dtails. a te va ?
during the melting process in order to reach a better dispersed system. Commented [I6]: On ne peut pas comparer directement la
dispersion de la kaolinite et la montmorillonite. Dans le cas de la
Regarding the THR values reported in Table 3, lower values are obtained introducing montmorilonite la dispersion est la intercalation/exfoliation que nous
ne regardons pas dans le cas de la kaolinite.
kaolinites. This is a direct consequence of inorganic fillers (so non-combustible)
incorporation. Finally, as indicated in table 3, EHC of all composites does not change
significantly indicating that the main mode-of-action of kaolinite occurs in the condensed
phase.
A notable difference between composites concerns the time to ignition (Table 3). All tested
specimens present a TTI never exceeding 2mn and follow this classification:
< 1 < 2 < 1 < 2 < 1 < 2
Mineral particles delay the ignition. This global augmentation of TTI can be attributed to an
improved thermal stability (TGA measurement could confirm this hypothesis) or to an Commented [AC7]: Marcos as-tu les ATG ?
increase of thermal diffusivity. The A treatment seems to be the best one considering this
parameter but no correlation with the morphology has been found. Commented [I8]: Cest difficile dexpliquer la diffrence
observ en ce qui concerne le TTI. Certains auteurs parlent de leffet
de la viscosit, dautres parlent de leffet de dgradation du
traitement des argiles ou dun effet catalytique. De toute faon, latg
des diffrentes compositions de montre pas une diffrence
significative (je nai pas lATG pour 30%, seulement pour 20%). Je
pense que le mieux cest de constater cette diffrence et donner des
explications de la littrature. Tu peux regarder larticle de lEVA
que nous avons soumis et utiliser la mme explication.
15
1500
PA
1250 PA-K1
PA-K1A
1000 PA-K1E
HRR (kW*m-2)
750
500
250
0
150 300 450 600
Time (s)
Fig. 11 : Heat Release Rate (HRR) curves of pristine PA6 and PA6 composites vs time
1000
800
pHRR (kW.m )
-2
600
400
200
0
15 20 25 30 35
dm (m)
16
Fig. 12 : Peak of Heat Release Rate (PHRR) vs median diameter (100 magnification)
4. Conclusion
17
the biggest agglomerates (scale magnification 100): the resilience decreases when the
interparticular distance increases and below a critical value (around 1500m) the composites
become fragile; the pHRR is reduced when a better dispersion of the biggest agglomerates is
reached (as dm is decreasing).
The next step is to combine kaolinites with a flame retardant additive; the expected
synergy between these two additives should lead to a pHRR and THR decrease.
18
References
[1] Blumstein A. Journal of Polymer and Science Part A 1965; 3:2665-2673
[2] Giannelis EP. Advanced Materials 1996; 8:29-35
[3] Gilman JW, Jackson CL, Morgan AB, Harris Jr R, Manias E, Giannelis EP,
Wuthenow M, Hilton D, Phillips SH 2000; 12:1866-1873
[4] Peeterbroeck S, Laoutid F, Taulemesse JM, Monteverde F, Lopez-Cuesta JM, Nagy
JB, Alexandre M, Dubois P. Advances Functional Materials 2007; 17:2787-2791
[5] Laachachi A, Leroy E, Cochez M, Ferriol M, Lopez-Cuesta JM. Polymer Degradation
and Stability 2005; 89:344-352
[6] Laachachi A, Ferriol M, Cochez M, Lopez-Cuesta JM, Ruch D. Polymer Degradation
and Stability 2009; 94:1373-1378
[7] Fina A, Abbenhuis HCL, Tabuani D, Camino G. Polymer Degradation and Stability
2006; 91(10):2275-2281
[8] Rothon RN, Hornsby PR. Polymer Degradation and Stability 1996; 54(2-3):383-385
[9] Razak MZA, Arsad A, Rahmat AR, Hassan A. Journal of Polymer Engineering 2012;
32:177-183
[10] Riva A, Zanetti M, Braglia M, Camino G, Falqui L. Polymer Degradation and
Stability 2002; 77(2):299-304
[11] Laoutid F, Ferry L, Leroy E, Lopez-Cuesta JM. Polymer Degradation and Stability
2006; 91(9):2140-2145
[12] Bourbigot S, Le Bras M, Dabrowski F, Gilman JW, Kashiwagi T. Fire and Materials
2000; 24:201-208
[13] Michler GH. Acta Polymerica 1993; 44:113-124.
[14] Hamming LM, Qiao R, Messersmith B, Brinson LC. Composites Science
andTechnology 2009; 69(11-12):1880-1886.
[15] Schartel B. In: Morgan AB, Wilkie CA, editors. Flame retardant polymer
nanocomposites. New York: John Wiley & sons; 2007. P. 107-130.
[16] Tandon G P, Weng G J. Polymer Composites1984; 5(4):327-333
[17] Shen Y L, Finot M, Needleman A, Suresh S. Acta Metallica 1994; 42(1):77-97
[18] Chen B, Evans JRG. Scripta Materialia 2006; 54(9):1581-1585
[19] Levita G, Marchetti A, Lazzeri A. Polymer Composites 1989; 10(1):39-43
[20] Wang K, Wu J, Ye L, Zeng H. Composite Part A: Applied Science and Manufacturing
2003; 34(12):1199-1205
[21] Okuno K, Woodhams R T. Journal of Cellular Plastics 1974; 10(6):295-300
19
[22] Alter H. Journal of Applied Polymer Science 1965; 9(4):1525-1531.
[23] Turcsnyi B, Puknszky, Tudos F. Journal of Materials Science Letters 1988; 7:160-
162
[24] Shen YL, Finot M. Acta Metallurgica et Materialia 1994; 42(1):77-97
[25] Demjn Z, Puknszky B, Nagy J. Composite Part A: Applied Science and
Manufacturing 1998; 29(3):323-329
[26] Fu Q, Wang G. Polymer Engineering and Science 1992; 32(2):94-97
[27] Bartczak Z. Polymer 1994; 42(1):2347-2365
[28] Wu S. Polymer 1992; 26(12):1855-1863
[29] Wu CL, Zhang MQ, Rong MZ, Friedrich K. Composites Science and Technology
2002; 62(10-11):1327-1340
[30] Bonati A, Merusi F, Bochicchio G, Tessadri B, Polacco G, Filippi S, Giuliani F.
Contruction and Building Materials 2013; 47:990-1000.
[31] Qin H, Zhang S, Zhao C, Feng M, Yang M, Shu Z, Yang S. Polymer Degradation and
stability 2004; 85:807-813.
[32] Samyn F, Bourbigot S, Charafeddine J, Bellayer S. Polymer Degradation and Stability
2008; 93:2019-2024
[33] Laachachi A, Cochez M, Leroy E, Gaudon P, Ferriol M, Lopez-Cuesta JM. Polymer
for Advanced Technologies 2006; 17(4) :327-334
20
LIST OF FIGURES
Fig. 1. Standard tests specimens (a) tensile (ISO 527-2 1A), thickness 4mm (b)
Charpy impact tests
Fig. 2. SEM micrographs of PA with 30%wt of K2, scale factors (a) 100 (b) 1000 (c)
20000
Fig. 3. Image analysis on SEM micrographs (a) initial image (b) selected areas
Fig. 4. Diameter size distribution for the PAK1 sample: histogram of the normalized
frequencies, cumulative distributions (experiments (full line), identification
(dotted line))
Fig. 7. Youngs modulus of composites, effect of the surface treatment, prevision via
Mori Tanaka modelling p=15%
Fig. 8. (a) Yield stress as a function of median diameter (b) Yield strain as a function
of agglomeration rate
Fig. 11. Heat Release Rate (HRR) curves of pristine PA6 and PA6 composites vs time
Fig. 12. Peak of Heat Release Rate (PHRR) vs median diameter (100 magnification)
21
22
LIST OF TABLES
23