1 – Online electrical conductance measured directly during mixing process of CNT and CR in dependence on mixing time.. Results and discussion composites filled with unmodified CNTs The o
Trang 1Kinetics of filler wetting and dispersion in carbon
nanotube/rubber composites
H.H Le a,*, X.T Hoang b, A Dasc, U Gohs c, K.-W Stoeckelhuber c, R Boldt c,
G Heinrich c,d, R Adhikari e,f, H.-J Raduscha
aCenter of Engineering Sciences, Martin Luther University Halle-Wittenberg, D-06099 Halle, Germany
bUniversity of Technology, National University, HCM City, Vietnam
cLeibniz-Institut fu¨r Polymerforschung Dresden e.V (IPF), D-01069 Dresden, Germany
d
Institut fu¨r Werkstoffwissenschaft, Technische Universita¨t Dresden, D-01069 Dresden, Germany
e
Central Department of Chemistry, Tribhuvan University, Kirtipur, Kathmandu, Nepal
fNepal Polymer Institute (NPI), P.O Box 24411, Kathmandu, Nepal
A R T I C L E I N F O
Article history:
Received 2 February 2012
Accepted 20 May 2012
Available online 27 May 2012
A B S T R A C T The effects of the surface modification of multi-walled carbon nanotubes (MWCNTs) by an ionic liquid, 1-butyl 3-methyl imidazolium bis(trifluoromethyl-sulphonyl)imide (BMI) on the kinetics of filler wetting and dispersion as well as resulting electrical conductivity of polychloroprene (CR) composites were studied Two different MWCNTs were used, Baytu-bes and Nanocyl, which differ in their structure, purity and compatibility to CR and BMI The results showed that BMI can significantly improve the macrodispersion of Baytubes, and increases the electrical conductivity of the uncured BMI–Baytube/CR composites up
to five orders of magnitude In contrast, the use of BMI slows the dispersion process and the development of conductivity of BMI–Nanocyl/CR composites Our wetting concept was further developed for the quantification of the bound polymer on the CNT surface We found that the bonded BMI on the CNT surface is replaced by the CR molecules during mix-ing as a result of the concentration compensation effect The de- and re-agglomeration pro-cesses of CNTs taking place during the subsequent curing process can increase or decrease the electrical conductivity significantly The extent of the conductivity changes is strongly determined by the composition of the bound polymer and the curing technique used
2012 Elsevier Ltd All rights reserved
Carbon nanotubes (CNTs) as new nanofiller in rubber
com-posites have got more and more attention in different areas
ranging from rubber hoses, tire components, sensing devices
to electrical shielding and electrical heating A significant
pro-gress in the area of the preparation and utilization of
nano-tube/polymer composite materials in recent years with
particular attention to their mechanical and electrical
proper-ties can be observed in a number of articles[1–9] However,
the effect of unmodified CNTs on the reinforcement and dy-namic mechanical performance as well as electrical proper-ties of rubber composites is not remarkable and much lower than expectation The main reason is related to the fact that CNTs are intrinsically bundled and heavily entangled due to van der Waals forces of attraction between adjacent tubes The fine dispersion of CNTs in rubber has been still the most challenging task for their practical application To overcome these problems, beside development of new preparation technologies [4,7–9] there has been much progress in the
0008-6223/$ - see front matter 2012 Elsevier Ltd All rights reserved
http://dx.doi.org/10.1016/j.carbon.2012.05.039
* Corresponding author: Fax: +49 3461463891
E-mail address:hai.le.hong@iw.uni-halle.de(H.H Le)
A v a i l a b l e a t w w w s c i e n c e d i r e c t c o m
j o u r n a l h o m e p a g e : w w w e l s e v i e r c o m / l o c a t e / c a r b o n
Trang 2functionalization of CNT surface with various organic
mole-cules for better compatibility and dispersion of CNTs in
rub-ber For example, the processability and mechanical
performance of rubber composites could be improved by
introducing carboxylic acid groups[10,11]or multifunctional
silane[12–14]onto the CNT surface Recently,
functionaliza-tion of CNTs with ionic liquids, a kind of molten salt with
nearly zero vapor pressure and high thermal stability, is an
interesting topic, because ionic liquids could provide a facile
and promising method to control the surface properties of
CNTs by means of cation–p interaction[5–17] Carbon
nano-tubes/ionic liquid mixture was mixed into silicone elastomer
by Sekitani et al.[18]and hydrogenated nitrile rubber (HNBR)
by Likozar and Major [19]to produce conductive rubberlike
stretchable composites Das et al.[20,21]used a series of ionic
liquids as surfactant for blends of styrene–butadiene rubber
(SBR) and polybutadiene rubber (BR) filled with CNTs in order
to determine the coupling activity of ionic liquids between
diene elastomers and CNTs Recently, Subramaniam et al
[22,23]described a new simple route to disperse CNTs in
poly-chloroprene (CR) using an ionic liquid, 1-butyl 3-methyl
imi-dazolium bis(trifluoromethyl-sulphonyl)imide (BMI) and
found that the usage of BMI and a low concentration (5 phr)
of CNTs in CR exhibited an electrical conductivity of 0.1 S/
cm with a stretchability of >500%
In this regard, the present work focuses on the
character-ization of the kinetics of CNT dispersion in CR
nanocompos-ites and its correlation to the electrical conductivity by means
of the method of the online measured electrical conductance
The effectiveness of BMI as surfactant for two different CNTs,
Baytubes and Nanocyl, at different conditions of composite
fabrication, i.e compounding and curing processes will be
characterized and discussed by taking into consideration
the selective wetting behavior of CNTs by BMI and rubber
The success of the work may encourage the application of
the method of the online conductance and the wetting concept
for the characterization and understanding of the mixing
mechanism of CNTs in rubber matrix in the laboratory scale
on the one hand, and for monitoring the production of CNT
nanocomposites in the industrial scale on the other
Polychloroprene (CR) Baypren 611 (Lanxess, Germany) with
Mooney viscosity MU ((ML 1 + 4) 100 C) of 43 ± 6 was used
as rubber matrix Ionic liquid BMI, which is basically made
of an asymmetric heterocyclic cation 1-butyl 3-methyl
imi-dazolium (BMI+) and an anion
bis(trifluoromethyl-sulpho-nyl)imide (BMI) (Sigma–Aldrich, Germany) was used as
surfactant According to the chemical structure of BMI a mass
ratio BMI+/BMI of 139/280 was calculated Multi-walled
carbon nanotubes Baytubes C150HP (Bayer MaterialScience,
Germany) and NanocylTMNC7000 (Nanocyl S.A., Belgium) were
used as fillers Nanocyl nanotubes were found to be longer
than Baytubes C150HP[24] In addition, Nanocyl possesses a
broad length distribution with several nanotubes up to
10 lm The amorphous carbon content of both CNTs is about
3 wt.% according to the thermogravimetric analysis made in the present work Some important data of both used fillers provided by the manufacturers are given inTable 1
For convenient admixing CNTs into the mixing chamber both CNTs were softly ground with BMI in different ratios
by weight of CNTs to BMI, till a black paste BMI/CNT was ob-tained The composites were prepared in an internal mixer (Rheocord 300p, ThermoHaake) by keeping the following mix-ing conditions: initial chamber temperature TAof 25 C, rotor speed of 70 rpm, fill factor of 0.72 The amount of the tubes used in the study was varied in parts per hundred rubber (phr) The black paste BMI/CNT was admixed into the cham-ber at 3 min mixing time The first curing package consists
of 2.5 phr ZnO, 1 phr stearic acid, 1.4 phr sulfur, and 1 phr CBS (n-cyclohexyl-2-benzothiazole-sulfenamide) The desig-nation of composites containing different CNT and BMI load-ings is given inTable 2 The mixing time was varied by taking into account the electrical conductance–time characteristic
as described in our previous work[26]
In order to characterize the effect of the preparation tech-nique, two composites were prepared according to Table 3 Baytubes and BMI were separately admixed into the mixing chamber For these samples 1.5 phr peroxide Luperox 101 (Atofina Chemicals) was used as the second curing agent
Optical microscopy was used to characterize the CNT macro-dispersion This method was described for CB filled com-pounds by Leigh-Dugmore[25]and modified by us[26] The macrodispersion D is calculated by the ratio of the surface
of non-dispersed agglomerates to that of the image A value
D = 0% indicates an image without any agglomerate larger than 6 lm
Ultra-thin sections with approximately 35 nm thickness cut from compression-molded plates with a diamond knife (35 cut angle, DIATOME, Switzerland) at 140 C on a
cryo-micro-Table 1 – Structural parameters of Baytubes and Nanocyl
diameter nm
Average length [24](lm)
Surface area BET (m2/g)
CB purity (%)
Table 2 – Designation of CNT/CR and BMI-Baytube/CR as well
as BMI-Nanocyl/CR composites
Composite Polychloroprene
(C) (phr)
Baytubes (Ba) (phr)
Nanocyl (Na) (phr)
BMI (I) (phr)
Trang 3tome were used for Transmission electron microscopy (TEM)
analysis The slices were collected on a copper grid with a
car-bon-hole-foil The specimens were investigated on a Zeiss
Li-bra200MC (Zeiss, field emission cathode, point resolution
0.2 nm) with an accelerating voltage of 200 kV
Scanning electron microscopy (SEM) investigations were
per-formed on an Ultra plus microscope (Zeiss, field emission
cathode) operated at 2 kV accelerating voltage To get a plain
surface the samples were prepared by cutting a cryo-section
using an ultra-microtome (Leica Ultracut UC7) with a
dia-mond knife at 140 C The cryo-sections were examined
without any additional coating to avoid masks over the
nano-tubes and to allow a charge contrast imaging between the
conductive CNT network and the rubber matrix
A conductivity sensor system was installed in the chamber of
the internal mixer to measure the electrical signal of the
con-ductive mixtures between the sensor and the chamber wall
The construction and position of the conductance sensors
has been described in our previous works[27–29]
As an example, Nanocyl-filled CR composites CNa5were
prepared for different mixing times of 7, 9, 12, 15, 22 and
32 min The online conductance was recorded during the
mixing process and is presented inFig 1as function of
mix-ing time The curves show a typical shape quite similar to that
of CB-filled rubbers[27,29]or CNT-filled SBR[30] The onset
time tonset of the conductance is observed at about 8 min
and the tGmaxat 20 min The reproducibility of the electrical
signal is very good
Measurement of electrical conductivity of uncured and cured samples was carried out at room temperature by means of a multimeter 2750 (Keithley) The shape of the conductive test specimens was a rectangular strip, whose ends were coated
by silver paste in order to receive a good contact with the electrodes
For extraction experiment 0.3 g of each raw mixture obtained directly from the mixing process was stored in 300 ml of an 80/20 toluene/acetone mixture for 4 days at room tempera-ture The presence of acetone significantly weakens the strong interaction between the cation and anion of the ionic liquid, and initiates ion pair dissociation[31,32] Once the ions are released, the anion is rapidly saturated with acetone and dissolved, while a part of cations remains bonded to the CNT surface through cation–p interaction[15–17] After the soluble part was entirely extracted from the raw mixture, the CNT– polymer gel was taken out and dried up to a constant mass
mG The insoluble polymer part is described by the so-called rubber layer L, which can be calculated according to Eq.(1) [33,34]
L ¼mGmcompcCNT
mG
ð1Þ
mcomp is the mass of the composite before extraction experiment, cCNTis the mass fraction of the filler in the com-posite mGis the sum of the insoluble part of BMI+mBMIþ
CR mCR
G as well as the mass of the filler:
mG¼mBMIþ
G þmCR
The collected solution was also dried up to receive the ex-tracted part mE, which consists of the soluble parts of BMI
mEBMI+, mEBMIand CR mECR:
mE¼mBMIþ
E þmBMI
E þmCR
According to[30]the rubber layer L is the sum of two con-tributions, LBMI þ
and LCR:
and their ratio can be calculated as follows:
LBMIþ
LCR ¼m
BMIþ G
mCR G
ð5Þ
analysis of the extracted part According to the wetting concept developed in our previous work for silica filled rubber blends[35], it is possible to quan-tify the composition of the rubber layer bonded to the silica surface by means of the FTIR analysis of the silica-rubber gel However, in the present work, due to the total absorption
of infrared beams by CNTs existing in gel the analysis of the gel by FTIR was not possible As an alternative, we indirectly characterized the gel composition by investigating the ex-tracted part by use of a FTIR spectrometer S2000 (Perkin–El-mer) equipped with a diamond single Golden Gate ATR cell (Specac)
The FTIR spectrum of the composite CNa5I10is shown in Fig 2 as an example The peaks at 1659 cm1, 740 cm1and
1
10-7
10-6
10-5
10-4
10-3
10-2
10-1
100
101
102
Mixing time (min)
tonset
Gonmax
Fig 1 – Online electrical conductance measured directly
during mixing process of CNT and CR in dependence on
mixing time
Table 3 – Designation of composites containing Baytubes
and BMI separately admixed
Composite Polychloroprene
(C) (phr)
Baytubes (Ba) (phr)
BMI (I) (phr)
Trang 41192 cm1are assigned to the C=C stretching mode of CR[36],
the vibration of C–H of cyclic BMI+ and –SO2 of BMI [37],
respectively The ratios of peak areas ABMI+/ACR and ABMI+/
ABMIwere calculated from the spectra of the extracted part
The ratio mEBMI+/mECRcan be determined from the ratio ABMI+/
ACRaccording to Eq.(6)
mBMIþ
E
mCR
E
fBMIþ=CR
ABMIþ
fBMI+/CR is a factor determined from the calibration curve,
which describes the correlation between ABMI+/ACRand the
gi-ven BMI+/CR ratio The procedure for generation of the
cali-bration curve was described in our previous work [35] In
the present work a value of 8.2 was determined for fBMI+/CR
from the calibration curve Using Eqs (3) and (6) the value
of mBMIþ
E and mCR
E can be calculated
The sum of BMI mass in gel and extracted part is
calcu-lated from mcompand the mass fraction of BMI in the
compos-ite cBMIas follows:
mBMIþ
G þmBMIþ
E þmBMI
mBMI
E can be calculated by taking into consideration the mass
ratio BMI+/BMIand the total BMI used cBMIÆmcomp
The sum of CR mass in gel and extracted part is similarly
determined from mcompand the mass fraction of CR in the
composite cCRas follows:
mCRþ
G þmCRþ
Setting the value of mBMIþ
E , mBMI
E and mCR
E into Eqs (7) and (8) the mass of BMI+and CR in gel mBMIþ
G and mCR
G, respectively can be determined
Setting the value of mBMIþ
G and mCR
G into Eq.(5)and combin-ing with Eq.(4)we can calculate LBMI þ
and LCR
analysis (SEM/EDX)
Scanning electron microscopy (SEM) (JSM 6300, Fa
JEOL)e-quipped with Energy dispersive X-ray analysis (EDX) (Voyager
1100, Fa Noran Instruments) was used for characterization of
the CNT-polymer gel of CBa5I10and CNa5I10in order to
charac-terize the presence of BMI in the bound layer of the
composites
electrons Samples were cured in a compression mold under 100 bar for
t90,at 160 C for sulfur-curing and at 190 C for peroxide-cur-ing In order to characterize the effect of cross-linking by high energy electrons on the electrical conductivity the sample CBa5I10 was treated with high energy electrons using an ELV-2 electron accelerator (Budker Institute of Nuclear Phys-ics, Novosibirsk, Russia) The samples were placed on the conveyor system of irradiation facility and passed under the electron beam exit window with well adjusted velocity in or-der to apply the desired dose to the samples The sample was irradiated with a dose of 50 kGy at room temperature in nitro-gen atmosphere The energy of the electrons and the beam current were 1.0 MeV and 4 mA, respectively
2.2.10 Swelling experiments of cured composites Swelling experiments were performed with cured samples by equilibrating them in toluene at room temperature for 48 h The swelling degree Q was calculated using Eq.(9):
Q ¼WswWi
Wdr
Wiis the weight of the rubber sample before immersion into the solvent, Wswand Wdrare the weights of the sample
in the swollen state and after dried in an oven at 80 C for
2 h from its swollen state, respectively
2.2.11 Thermogravimetric analysis (TGA) Thermogravimetric analysis of the neat and coated CNTs was carried out by a thermo-balance (Mettler Toledo) in the tem-perature range between 30 C and 900 C in air with a heating rate of 20 K/min The mass change in the temperature range between 310 C and 430 C as well as 500 C and 550 C is attributed to the degradation/oxidation of BMI and amor-phous carbon, respectively[38]
2.2.12 Determination of surface energies Wetting experiments (modified Wilhelmy method) were per-formed, using the dynamic contact angle meter and tensiom-eter DCAT 21, DataPhysics Instruments GmbH (Filderstadt, Germany) For the Wilhelmy measurements, the CNT parti-cles were put in a shallow plate In the filler powder a
2 · 1 cm2piece of a double-face adhesive tape (TESA 55733, Beiersdorf, Hamburg, Germany), was immersed and gently moved, until the tape was uniformly coated by filler particles The pellets of the granulated Baytubes were pulverized finely
in a mortar, before they were attached at the adhesive tape Surplus particles, which did not stick to the adhesive tape, were blown away by a stream of nitrogen The CNT particle covered tape was used for Wilhelmy contact angle measure-ments without further modification
Sessile drop contact angle measurements on a sheet of un-cured CR were conducted with the automatic contact angle meter OCA 40 Micro, DataPhysics Instruments GmbH (Filders-tadt, Germany) The surface energies were calculated from the results of these wetting experiments For this purpose a set of test liquids with different surface tension (and polarity) was used: water (Millipore Milli-Q-Quality), formamide (Merck, Darmstadt, Germany), ethylenglycol (Fisher Scientifiy,
0.10
0.15
0.20
0.25
0.30
SO2
1192 cm-1 BMI
-CR
C=C
1659 cm -1
C-H
740 cm-1
BMI+
Fig 2 – FTIR spectrum of the extracted part of CNaI
Trang 5Loughborough, UK), dodecane (Merck Schuchardt,
Hohenb-runn, Germany), n-hexadecane (Merck, Darmstadt, Germany)
and ethanol (Uvasol, Merck, Darmstadt, Germany) Surface
energy calculations were performed by fitting the Fowkes
equation[39]
3 Results and discussion
composites filled with unmodified CNTs
The online conductance curves of CR composites filled with
different loadings of unmodified Baytubes CBa5-20 are
pre-sented inFig 3a They show a typical conductance-time
char-acteristics with tonsetand tGmax At tonsetand tGmaxthe online
conductance starts to rise and reaches the maximum value,
respectively According to our previous works[27,28]the
mac-rodispersion of filler and the online conductance correlate
closely to each other The largest change of the size of filler
agglomerates, i.e the dispersion of filler agglomerate into
smaller aggregates or even individual tubes, is determined
in the period between tonsetand tGmax Thus, tonsetand tGmax
have been often used as a measure for characterization of
the filler dispersion kinetics Upon tGmaxthe online
conduc-tance decreases slightly that is related to the better
distribu-tion of small aggregates throughout the matrix as discussed
previously[27] For the composite CBa5the tonsetwas not
ob-served even till a mixing time of up to 200 min With
increas-ing loadincreas-ing of Baytubes the tonset and tGmaxshift to shorter
times due to higher shear stress, which results from the
increasing viscosity and accelerates the filler dispersion A
linear extrapolation by connecting Gonmax of the three curves
as shown in Fig 3a gives a theoretical tGmax of about
1000 min for CBa5
The online conductance of CR filled with different loading
of Nanocyl is presented inFig 3b For the composite
contain-ing 5 phr Nanocyl CNa5a tGmaxof 20 min was observed that is
much shorter compared to that of CBa5 In contrast, the tGmax
of the composites CNa2-5is nearly independent on the filler
loading The value of Gon
max of CNa5 is similar to that of CBa15 According to the online conductance measurement it
is easy to recognize that Nanocyl is dispersed much faster
than Baytubes and imparts higher values of conductivity of
the composites
Mu¨ller et al.[40] revealed that a lower mixing energy is
needed for dispersion of Nanocyl in PP composite than for
Bay-tubes By discussion of the different dispersion behavior of both
fillers they took into consideration the more compact structure
of Baytubes compared to the more loosely packed agglomerates
of Nanocyl Moreover, in our opinion the higher purity of
Baytu-bes causing better filler-filler interaction is surely essential for
the resistance of agglomerates against dispersing
The correlation between the offline conductivity and filler
loading of CR composites filled with Baytubes and Nanocyl is
presented inFig 4 The offline conductivity increases
expo-nentially with increasing filler loading For both series it is
obvious that the percolation threshold shifts to lower values
with increasing mixing time
Because of the strong conductivity-mixing time
depen-dence, the values determined at t should be chosen for
a reasonable comparison of conductivity properties between different systems A percolation threshold value of 8 phr for Baytube/CR and 3 phr for Nanocyl/CR composites, respec-tively, was determined at tGmax
conductivity
InFig 5a the online conductance of the Baytube composites without and with BMI CBa5and CBa5I10, respectively, is pre-sented in dependence on mixing time With addition of BMI the online conductance of CBa5I10 increases faster and reaches a tGmaxat about 55 min It is very short compared to the extrapolated 1000 min of the unmodified CBa5
The macrodispersion of filler in the CR matrix is studied by optical microscopy images The mixing time and correspond-ing macrodispersion D of each sample are given on the image
In the images of CBa5shown inFig 6a–d very large agglomer-ates of Baytubes are still visible even at a very long mixing time A macrodispersion D of 22% is determined at 220 min mixing time The addition of BMI leads to a significant improvement of dispersion as seen in the images shown in Fig 6e–h The largest change of the size of CNT agglomerates
is clearly determined in the range between tonset= 20 min and
t = 55 min The macordispersion D reduces from 37% to
1E-6 1E-5 1E-4 1E-3 0.01 0.1 1 10 100
5 phr
10 phr
on (mS)
Mixing time (min)
20 phr CNT
15 phr
max
1E-6 1E-5 1E-4 1E-3 0.01 0.1 1 10 100
2 phr
3 phr
on (mS)
Mixing time (min)
5 phr
max
(a)
(b)
Fig 3 – Online conductance of CR filled with different loading of Baytubes (a) and Nanocyl (b) in dependence on mixing time
Trang 620% after 50 min and 14% after 75 min mixing time Upon
tGmaxonly some small agglomerates are observed The better
macrodispersion of Baytubes by addition of BMI is attributed
to the physical cation-p interaction between BMI+ and the
tubes and/ or the perturbation of p–p stacking of multi-walls
of the tubes as discussed in literature[15–17,41]and evidently proved by Raman investigation by Subramaniam et al.[22] Fig 5b shows that the offline conductivity of the uncured composites CBa5 and CBa5I10 changes correspondently to the online conductance The one to one relationship between the online and offline conductive values cannot be expected, because the online conductance is measured through an undefined volume during the mixing at changing tempera-ture, while the offline conductivity has to be measured for definite sample geometry by keeping constant the tempera-ture and pressure The CNT network in the sample is in a steady state during the offline measurement Thus the online conductance can be used for determine the trend of the off-line conductivity The Goff
maxof 7.3 · 104S/cm is received for CBa5I10, while CBa5shows a conductivity of only 109S/cm all over the mixing time
InFig 7a the online conductance of Nanocyl composites without and with BMI CNa5 and CNa5I10, respectively, is shown with mixing time At mixing times up to tonsetthe on-line conductance of CNa5I10 is slightly higher than that of CNa5that may be related to the contribution of the ionic con-ductivity of BMI In the mixing period up to tGmaxthe online conductance of the unmodified composite CNa5 increases earlier and reaches tGmaxat 20 min, while tGmaxof CNa5I10is determined at 30 min In contrast to CBa5I10, an addition of BMI in CNa5I10decelerates the dispersion process of Nanocyl The optical microscopic images shown inFig 8reveal that the dispersion of CNa5is faster than that of CNa5I10 corre-sponding to the development of the online conductance After 30 min mixing time the macrodispersion of Nanocyl in CNa5is 10% and in CNa5I1017% That may be attributed to the fact that the as-produced Nanocyl inherently is a type
of CNT, which can be dispersed easily as discussed in [40], and thus, the dispersion of Nanocyl was not accelerated effec-tively by wetting it with BMI Furthermore, the shear stress determined by the torque during the mixing process reduces from 20 Nm to 15 Nm by adding the low-viscous BMI leading clearly to the slower dispersion process of Nanocyl in CR The offline conductivity of the uncured samples of CNa5 and CNa5I10 is presented inFig 7b as a function of mixing time It is obvious that the development of the offline conduc-tivity of both series corresponds well with that of the online conductance and optical microscopic observations, i.e the offline conductivity of CNa5I10increases more slowly in pres-ence of BMI The Goff
maxof CNa5and CNa5I10is in the same or-der of magnitude, i.e 103 S/cm, indicating the same dispersion degree of both composites at tGmaxregardless of BMI addition A closer look at the chart of the offline conduc-tivity of CBa5I10 and CNa5I10 reveals that the BMI addition makes the offline conductivity strongly dependent on the mixing time and input of mixing energy That may cause dif-ficulties for process control of the composite preparation As seen in Figs 5b and 7b the offline conductivity decays strongly after reaching tGmax That is attributed to the better separation/distribution of the filler aggregates throughout the matrix as discussed already in the case of CB filled com-posites [27,28] In CNT filled composites the shortening of the tubes during the mixing process becomes essential and should be taken into consideration The comparison between the pristine and the processed CNTs made by Fu et al.[42],
1E-5
1E-4
1E-3
0.01
0.1
1
Gon max
CBa5 CBa5I10
on (mS)
Mixing time (min)
Gonmax
1E-11
1E-10
1E-9
1E-8
1E-7
1E-6
1E-5
1E-4
1E-3
0.01
0.1
Mixing time (min)
Gon max
CBa5I10
CBa5
(a)
(b)
Fig 5 – Online conductance (a) and offline conductivity (b) of
the uncured composites CBa5and CBa5I10as function of
mixing time
1E-11
1E-9
1E-7
1E-5
1E-3
0.1
10
M
ing
me
=t
G a
M in
tim
= 0 in
time
ax
Mixi
ng t ime
= 100
min Baytube/CR
off (S/cm)
CNT loading (phr)
Nanocyl/CR
Fig 4 – Offline conductivity of the uncured composites CBa5–
20and CNa2–5in dependence on the filler loading and
mixing time
Trang 7Chen et al [43], and Lin et al [44]demonstrated a strong
shortening of CNTs up to 10% of the initial length According
to the work of Krause et al.[24]with the same CNTs as used in the present work, Nanocyl with initially longer nanotubes underwent a more significant shortening of the tube length
to about 30% (related to x50-value) The comparison of the tube length distribution of Baytubes before and after melt processing indicated a shortening to about 50% of their initial length (related to x50-value) However, the online conductance values measured upon tGmaxare slightly influenced by the shortening of CNT as shown in Figs.5a and7a, because the shortened tubes can still touch each other during the mixing and form a non-stationary filler network After mixing, the shortened tubes exist in a steady state, and they are separated from each other that affects the offline conductivity negatively
Fig 9 represents the extraction experiment and the rubber layer L as a measure for the bound polymer of different com-posites in dependence on mixing time In general, the rubber layer L increases with mixing time and reaches a plateau va-lue after a certain time According to the discussion made by Manas-Zloczower [45,46] and us[34]on the infiltration and dispersion processes, polymer first can infiltrate the outer layer of filler agglomerate and wets its surface This layer is then eroded and a new filler surface is created The new sur-face is progressively wetted by polymer Therefore wetting and dispersion take place simultaneously The wetting pro-cess of Baytubes in CBa5 was not measurable, because the gel was not formed In this sample due to the bad filler-rubber interaction and bad dispersion, Baytubes did not form a net-work, which acts as a skeleton to hold polymer in gel Thus the wetted agglomerates are separate and swim in the solvent making the flash black as seen inFig 9a In contrast, a clear solution observed for CBa5I10, CNa5and CNa5I10 demonstrat-ing a formation of the filler-polymer gel after extraction experiment
The rubber layer L of CBa5I10 shown inFig 9b increases slowly because of the slow dispersion of Baytubes The
pla-Fig 6 – Optical microscopic observations of CBa5(a–d), CBa5I10(e–h) in dependence on mixing time (image dimension
300 lm · 400 lm)
10 1E-5
1E-4
1E-3
0.01
0.1
1
Gonmax
Gonmax
on (mS)
Mixing time (min)
10 1E-8
1E-7
1E-6
1E-5
1E-4
1E-3
0.01
0.1
1
CR CNT
off (S/cm)
Mixing time (min)
max
(a)
(b)
Fig 7 – Online conductance (a) and offline conductivity (b) of
the uncured composites CNa5and CNa5I10as function of
mixing time
Trang 8teau value LPof CBa5I10of 0.62 is determined at about 50 min
indicating the end of the wetting process at this time, thus no
more free tube surface is available for further wetting
Compared to CBa5I10the plateau value LPof the Nanocyl
composites is obtained at much shorter mixing time thanks
to the faster dispersion process of Nanocyl Compared to
CNa5, the wetting process of Nanocyl in CNa5I10 becomes
more slowly due to the slower filler dispersion process as
dis-cussed above Moreover, Nanocyl has a surface area of 250–
300 m2/g compared to 140–180 m2/g of Baytubes, thus at the
same dispersion degree more surface is available for rubber
wetting in case of Nanocyl As a result, the plateau value LP
of CNa5and CNa5I10of 0.72 and 0.68, respectively, is higher than that of CBa5I10
A deeper insight into the selective wetting behavior of nanotubes by rubber and BMI can be obtained by taking into consideration the affinity between the components of the composites On the basis of the Z-model proposed in our pre-vious works[47,48]the CNT surface fraction wetted by CR and BMI in composites at a thermodynamic equilibrium state can
be predicted using Eqs (10)–(13)
SCR F
SBMIF ¼nCR=BMI
cBMI-F
cCR-F
¼nCR=BMI
cBMIþ cF2 ffiffiffiffiffiffiffiffiffiffiffiffiffipcBMIcF
cCRþ cF2 ffiffiffiffiffiffiffiffiffiffiffipcCRcF
ð10Þ
SCR
F þSBMI
SBMIF ¼ 1
with
x¼ cBMIþ cF2 ffiffiffiffiffiffiffiffiffiffiffiffifficBMIcF
p
cCRþ cF2 ffiffiffiffiffiffiffiffiffiffiffipcCRcF
ð13Þ
SCR
F and SBMI
F are the surface fractions of the CNT wetted by the CR and BMI molecules, respectively cBMI-Fand cCR-Fare the interfacial tension values between CNT and BMI or CR
cCR, cBMIand cFare the surface tension values of CR and BMI
as well as CNT, respectively nCR/BMIis the blend ratio CR to BMI In the present work cCR= 35 mN/m, cF= 28.5 mN/m for Baytubes and 30.5 mN/m for Nanocyl were experimentally determined cBMI= 33.6 mN/m was taken from Ref.[49] Setting the surface tension values of CR and BMI into Eqs (10)–(13) a master curve demonstrating the filler surface frac-tion wetted by the BMI molecules in dependence on the filler surface tension can be created as seen inFig 10 Fitting the surface tension of Baytubes and Nanocyl into the master curve with nCR/BMI+= 1, a CNT surface fraction SBMI
F of 0.72 was found for Baytubes and 0.82 for Nanocyl at a thermody-namic equilibrium state In this case, the selective wetting behavior of CNT is merely dependent on the filler-polymer affinity, and the affinity of both CNTs to BMI is better than
to CR
Fig 8 – Optical microscopic images of CNa5(a–d), CNa5I10(e–h) in dependence on mixing time (image dimension
300 lm · 400 lm)
1
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
CNa
5I
10
CBa
5I
10
Mixing time (min)
LP
(b)
(a)
Fig 9 – Extraction experiment (a) and rubber layer L of
different composites in dependence on mixing time (b)
Trang 9For nCR/BMI+= 33 as used in the present work a CNT surface
fraction SBMI
F of 0.07 was found for Baytubes and 0.11 for
Nano-cyl The CNT surface fraction wetted by BMI is much smaller
compared to that wetted by CR, although the affinity of both
CNTs to BMI is better than to CR According to Eq (10)the
CNT surface wetted by BMI is determined by the
thermody-namic driving force (cCR-F/cBMI-F) and the concentration
com-pensation effect (nCR/BMI+) In the present work at a high
value of nCR/BMI+the concentration compensation effect
dom-inates the thermodynamic effect and as a result, CR
expect-edly wets the large amount of CNT surface
The experimental characterization of the selective wetting
behavior of CNT by CR and BMI was done by means of FTIR of
the extracted parts The rubber layer L and its contribution
LBMI þ
and LCRof the sample CBa5I10, which were quantified
by Eqs (4) and (5) are presented inFig 11a in dependence
on mixing time It is obvious that in the first mixing period
up to 50 min both LBMI þ
and LCRincrease In this range BMI and CR concurrently infiltrate the CNT aggregates and wet
CNT surface After the wetting process is complete, LBMI þ
de-creases while LCRcontinuously increases Because the value
of L is constant in the second period, it can be concluded that
the free CR molecules replaced the bonded BMI+on the CNT
surface as a result of the concentration compensation effect
At 120 min mixing time BMI+ is completely replaced by CR
that is corresponding to the prediction made by the Z-model
The similar wetting behavior was also found for CNa5I10as
shown inFig 11b However, the first mixing period ends after
10 min and the second one after 60 min The faster wetting
and replacement process taking place in CNa5I10are related
to the faster dispersion process of Nanocyl compared to
Bay-tubes Such a replacement process between the blend
compo-nents on the filler surface was also found in our previous
works [30–35] for different rubber blends filled with silica,
CB and CNTs
The ratio ABMI+/ABMIof the extracted part is presented in
Fig 11c in dependence on mixing time In both composites
ABMI+/ABMIdecreases from 0.045, which is the value
deter-mined from the spectrum of the neat BMI, to a minimum va-lue in the first mixing period The decrease of ABMI+/ABMIof the extracted part is an evidence for the predominant bond-ing of BMI+to the CNT surface in the first mixing period In the second mixing period ABMI+/ABMI increases because more and more bonded BMI+are released from the CNT sur-face and at the end of the mixing process ABMI+/ABMIreaches the value of the neat BMI, i.e CNTs are wetted completely only by CR
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
0.9
replacement process
L
Mixing time (min)
wetting process
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
L
Mixing time (min)
replacement process wetting process
0.030 0.033 0.036 0.039 0.042 0.045 0.048
BMI-Mixing time (min) neat BMI
(a)
(b)
(c)
Fig 11 – Rubber layer L, LBMI þ
and LCRof CBa5I10(a) and CNa5I10(b) as well as the ratio ABMI+/ABMIof the extracted part of both composites (c) in dependence on mixing time
0.0
0.2
0.4
0.6
0.8
1.0
γBaytubes
γNanocyl γCR
γBMI
Fig 10 – CNT surface fraction wetted by BMI+predicted by
the Z-model in dependence on the filler surface tension and
blend ratio nCR/BMI
Trang 10The EDX spectrograms of the filler-polymer gel of two
composites, CBa5I10-120 m after 120 min mixing time and
CNa5I10-60 m after 60 min mixing time, are shown inFig 12
No signals of fluorine at 0.67 keV and nitrogen at 0.39 keV of
BMIand BMI+, respectively, were found, while a strong peak
of chlorine at 2.62 keV of CR can be observed for both
compos-ites That corresponds very well to our FTIR analysis
The unbound part of BMI forms an own phase as seen in
SEM images for both composites (Fig 13) A closer look at
the BMI phase does not reveal any CNTs InFig 13a beside
the BMI phase some non-dispersed Baytube aggregates are
still observed in submicron-scale, which appear in the SEM
micrographs as bright domains The wetting BMI+cations
in-side the non-dispersed aggregates could be difficultly
re-placed by the CR molecules, because CR needs time to
infiltrate the aggregates That is why the wetting and
replace-ment process take place slowly in CBa5I10as discussed above
InFig 13b CNTs were uniformly dispersed in the rubber
ma-trix and form a regular network that enables the fast
replace-ment of BMI+ by CR on the Nanocyl surface as observed in
Fig 11b
Concerning the interaction between ionic liquids and
CNTs as well as the molecular structure of their interphase
several works have been done recently Likozar[50]
investi-gated the adsorption kinetics of different ionic liquids into CNT/HNBR composites by immersing the cured composites
in the ionic liquid/chloroform solvent He observed a homo-geneous distribution of anions in the composites by use of the fluorine signal detected by scanning electron micros-copy/energy dispersive X-ray analysis (SEM/EDX) By means
of fully atomistic molecular simulations Frolov et al.[51] stud-ied the basic mechanisms of carbon nanotube interactions with several different room temperature ionic liquids in their mixtures with acetonitrile It was found that two distinct lay-ers of cations and anions are formed at the CNT surface In-crease of the length of the non-polar alkyl groups of cations increases the propensity of imidazolium-based cations to lay parallel to the CNT surface Wang et al.[41]carried out Ra-man and IR measurements on the mixtures of ionic liquids and single-walled carbon nanotubes (SWCNTs) and found that no strong interaction such as cation–p interaction exists between SWCNTs and imidazolium cations It could be seen that the fluorine atoms of anions and the hydrogen atoms
of the alkyl groups of cations are much closer to the SWCNTs than the nitrogen atoms and carbon atoms of the
imidazoli-um rings This indicates that the SWCNTs are surrounded
by the polar parts of anions and non-polar parts of cations simultaneously They proposed that the ionic liquids interact with SWCNTs through weak van der Waals interaction and
3000
6000
9000
12000
15000
18000
BMI+
F
Binding energy (keV)
CR BMI
-Fig 12 – EDX spectrograms of the filler-rubber gel of CBa5I10
-120 m and CNa5I10-60 m
Fig 13 – SEM images of CBaI -120 m (a) and CNaI -60 m (b)
1E-9 1E-7 1E-5 1E-3 0.1 10
5 phr Nanocyl + 10 phr BMI
high energy electrons cured
peroxidic -cured
5 phr Baytubes + 10 phr BMI
5 phr Baytubes + 10 phr BMI
20 phr Baytubes 5 phr Baytubes + 5 phr BMI
Samples
uncured cured
peroxidic cured
Fig 14 – Offline conductivity of different composites before and after curing process