Advertisement

Nanofiller Dispersion in Rubber as Revealed by 3D-TEM

  • Shinzo Kohjiya
  • Atsushi Kato
  • Yuko Ikeda
Chapter
  • 121 Downloads
Part of the Springer Series on Polymer and Composite Materials book series (SSPCM)

Abstract

Firstly, it was found that visualization of three-dimensional structure of rubber vulcanizate by 3D-TEM observation necessitated a removal of rubber-soluble zinc compounds, which was derived from ZnO annexed as a vulcanization reagent. The Zn compounds deteriorated the resolution to give enough good resolution for 3-dimensional image construction. Secondly, an experimental method how to remove them on silica-loaded natural rubber vulcanizates was reported. Lastly, the nearest distance (dp) between the silica aggregates in the high-loading region was found 1.3 nm. It was suggested that this dp corresponds to the thickness of the silica/NR interaction layer, i.e., more or less intrinsic bound rubber on particulate silica surface, and the silica network structure in the rubber matrix was clearly visualized. Furthermore, in the case of carbon black (CB), dp in the high loading region was found 3 nm, and the CB network was also visualized. CB network formation was explained by the gelation theory: in the region of CB loading ≦ 20 phr pregel states, and the transition to gel state occurred in the region of CB loading  > 20 phr.

Keywords

Removal of rubber-soluble zinc compounds In situ silica Hydrophobic and hydrophilic silicas Silica network CB network 

4.1 Introduction

In this chapter, the results of 3D-TEM observation of nanofiller in the rubber matrix are to be described. Silica and carbon black (CB) are the representative nanofillers for rubber reinforcement among lots of known nanoparticles, and the visualized 3D images of particulate silica and CB are to be shown. Analysis of the images has afforded several structural features of value on their dispersion, and ultimately the visualized nanofiller network structure would be displayed.

From a specific interest of our research team on particulate silica generated in situ, the results of 3D-TEM on silica precede those of CB. Since CB has been the most significant reinforcer of rubber so far, the readers may be puzzled by the silica first. However, this is a matter of convenience, and please read through in situ silica as the first example of nanofillers. In particular, the pretreatment of the rubber vulcanizate samples has been found to be an indispensable process before the 3D-TEM measurements, which was first conducted on in situ silica sample as described in 4.2.1.

In the last section, fillers except silica and CB are described too. Together with Chaps.  2 and  7, the readers would understand that almost all fillers for rubber are qualified objects for 3D-TEM measurement.

4.2 Particulate Silica Dispersion as Revealed by 3D-TEM

4.2.1 Pretreatment of Rubber Sample Cross-Linked by Sulfur/Accelerator System

At the initial stage of our 3D-TEM studies of rubber vulcanizates, we encountered a few difficulties, among which the most serious was the low contrast of the slice TEM images. They did not have enough contrast to make a clear 3D image by the tomographic reconstruction treatments. We had intensive discussions on this problem, and finally found that it is ascribable not only to the problems of TEM and tomography but also to the rubber specimen itself. The specimens were silica-loaded cross-linked natural rubber (NR) samples; the cross-linking was done by a sulfur/accelerator system, i.e., they were the sulfur vulcanizates containing silica as a reinforcing filler [1, 2, 3, 4].

In Table 4.1 is shown the compounding recipes for the sulfur vulcanization (at 150 ℃ for 20 min) [1]. Here, stearic acid and zinc oxide are the activators and CZ-G is the accelerator for the sulfur vulcanization. DEG is used with commercial silica VN3 for improving the cross-linking efficiency. Two kinds of particulate silica, i.e., commercially available precipitated silica and in situ generated silica, were subject to the experiments for comparison at the same 33 phr (amount of silica 33 g per one hundred gram of rubber) compounding.
Table 4.1

Compounding recipes of silica for sulfur vulcanization (from Table 1 in Ref. [1])

Samples ingredient (phra)

NR-mix-V

NR-in situ-V

Raw NR (RSS#1)

100

0

Stearic acid (Accelerator activator)

1

1

ZnO (Accelerator activator)

5

5

Sulfur (Vulcanizing agent)

2

2

CZ-Gb (Accelerator)

1

1

Diethylene glycol (DG, Plasticizer)

2

0

Commercial silica (VN-3)c

33

0

Raw NR with in situ silica

0

133

aGrams of additive per 100 g of rubber

bCyclohexyl benzothiazyl sulfenamide

cNipsil VN-3 from Nippon Silica Ind. Co.

The slice TEM images, to be used for the reconstruction of 3D-TEM image of the two samples, are shown in Fig. 4.1 [3]. The contrast of the two-slice images was too low to give a good image after the reconstruction by the tomographic technique [3, 4, 5].
Fig. 4.1

3D-TEM slice images of silica-loaded NR vulcanizates before removal of Zn compounds

(from Fig. 3 in Ref. [3])

Because 3D-TEM measurements and tomographic treatments had been checked and it was found they were well functioning, the most probable reason of this low contrast was estimated to be the presence of a certain zinc compound solubilized in the rubber matrix of specimens (see Table 4.1), which might be the by-product or the residue of the vulcanization reaction [3, 4, 5, 6, 7]. In other words, zinc oxide in Table 4.1 was involved in the reaction and a portion of it might be converted to the rubber-soluble organic zinc compound via some reactions with sulfur and the organic accelerator (CZ-G), the exact mechanism of which has not been elucidated yet.

In order to dissolve this problem, several experimental works had been carried out, and finally the following method was found effective to remove the organic zinc compound from the rubber vulcanizate: The silica-loaded sulfur-cured sample was subject to extract by a mixture of diethyl ether, benzene, and concentrated hydrochloric acid (in volume ratio of 43/14/43), which we have named the NARC-AK method. Figure 4.2 shows energy dispersion X-ray spectra of NR-mix-V and NR-in situ-V before and after the zinc-removal pretreatment [5, 6, 7]. The five peaks from low to high energy in the figure, are assigned to 6C (Kab 0.280; Kα 0.277), 8O (Kab 0.532; Kα 0.525), 14Si (Kab 2.47; Kα 1.74), 16S (Kab 2.47; Kα 2.31), 30Zn (LIab 1.20; LIIab 1.05), respectively. Here, all Arabic numerals in the parentheses are in keV. In comparison with the intensity of Si peak, the decrease of Zn and S by the removal pretreatment was above 90% and approximately 50%, respectively. Majority of zinc has been successfully removed. The result on sulfur interestingly suggests that about half of S has been chemically connected to rubber. Of course, S has been consumed to form the sulfur cross-links, but some may possibly be connected as a pendant group to the rubber chain. Therefore, this result suggests that the cross-linking efficiency is 50% at most for the present systems shown in Table 4.1. Removed sulfur is estimated to include unreacted S, ZnS, and possibly some other not exactly known sulfur-containing materials. Further quantitative analysis on the rubber vulcanizates and the extracts from them may much contribute to elucidating the mechanism of sulfur vulcanization in general.
Fig. 4.2

3D-TEM slice images of silica-loaded NR vulcanizates after removal of Zn compounds

(from Fig. 4 in Ref. [3])

Thus, the clear seventy-one slice images per specimen were obtained on the pretreated sample, and two examples among them are shown in Fig. 4.3, which is assumed to be clear enough for the 3D image reconstruction. They were subjected to the tomographic reconstruction of the 3D-TEM image. The resultant two 3D images are shown in Fig. 4.4. Both commercial silica and in situ silica in the sample are successfully visualized in the figures. In other words, the described procedures have enabled us to obtain 3D-TEM image of the silica-loaded sulfur vulcanizate, which are reasonably assumed to be applicable to the other nanofillers too.

Fig. 4.3

3D-TEM slice images of silica-loaded NR vulcanizates after removal of Zn compounds

(from Fig. 4 in Ref. [3])

4.2.2 3D-Imaging of Silica Dispersion and Its Analysis

Three-dimensional images showing the dispersion of particulate silica in NR matrix have been obtained as in Fig. 4.4, which are subjected to various image analyses. One example is shown in Fig. 4.5, and on each silica, i.e., the commercial precipitated silica (VN3) and in situ silica, which displays 3D image. In the figure, silica particles which are separated from the nearest neighbors more than 1 nm distance are shown independently. The aggregate size of silica is larger in NR-in situ-V than in NR-mix-V, and the dispersion in the former seems more homogeneous than the latter.
Fig. 4.4

3D-TEM images of silica-loaded NR vulcanizates (from Fig. 7 in Ref. [4]). See Video 4.1 in Supplemental Electronic Material

Fig. 4.5

3D-image analysis results for NR-mix-V and NR-in situ-V (from Fig. 1b in Ref. [1]). See Video 4.2 in Supplemental Electronic Material

Based on the image in Fig. 4.5, radius of all the silica aggregates was calculated by assuming the perfect circle of the same volume to each aggregates. The distribution of radii for the two specimens is shown in Fig. 4.6 [3]. Here, the horizontal axis is the radius of the assumed equivolume sphere for the aggregate, and the vertical axis is the frequency in percentage. The size of aggregate of in situ silica was larger than that of VN3, and its distribution was broader. Also, it was found that the dispersion of the calculated aspect ratio of in situ silica was wider than that of VN3. In Table 4.2 are listed the densities by three methods, i.e., calculated by use of the Archimedes’ principle, obtained by thermos-gravimetric (TG) analysis, and obtained from 3D-TEM image [4]. The three values are reasonably in agreement.
Fig. 4.6

Radius distribution of conventional silica (VN3) and in situ silica measured by 3D-TEM/electron tomography (from Fig. 7 in Ref. [3])

Table 4.2

Comparison of the densities of CB-loaded NR vulcanizates after removal of Zn compounds: densities of CB-loaded NR vulcaizates obtained by the Archimedes’ principle, thermo-gravimetry (TG), and 3D-TEM observation (from Table 11 in Ref. [4])

Measurement method

Sample

CB-10

CB-20

CB-40

CB-60

CB-80

Density (g/cm3)

Archimedes’ principlea

0.972

1.03

1.09

1.14

1.19

Thermogravimetry (TG)b

0.951

1.06

1.07

1.12

1.17

3D-TEM observationb

0.960

1.01

1.06

1.08

1.15

aSolvent: Ethanol

bMixing law on densities of the components:

   d = d1φ1 + d2φ2

   d: Density of CB-filled NR vulcanizates

   φ1, φ2* Volume fractions of NR and CB

   (φ2 is obtained from TG measurements or 3D-TEM observation.)

   d1, d2: Densities of NR and CB (d1 = 0.913 g/cm3, d2 = 1.80 g/cm3)

4.2.3 Visualization of Hydrophilic and Hydrophobic Silica in Rubber Matrix

Nanofiller’s performance is much dependent on its surface is generally known, since its surface area increases with a square of its radius. In the case of the precipitated silica for rubber reinforcement, an additional factor is also very important. That is, polarity is influential due to the presence of the silanol group on the surface, which is not the case in CB. In order to elucidate the effect of polarity on the particulate silica, the silanol group of VN3 was substituted by trimethyl silyl group to produce nonpolar silica particle which is designated RX. In this subsection, comparative study of relatively hydrophilic VN3 and hydrophobic RX is presented in terms of silica dispersion elucidated by 3D-TEM.

Table 4.3 shows the recipes of peroxide cross-linking of silica/NR compounds [8, 9, 10]. The hydrophilic silica (VN; diameter, ca. 16 nm) carries silanol group on its surface. The hydrophobic one (RX; diameter, ca. 12 nm) carries trimethyl silyl group instead. As a cross-linker, dicumyl peroxide was used in 1 phr, and the amount of silica was changed from 0 to 80 phr. The cross-linking was carried out by compression molding at 155 ℃ for 30 min.
Table 4.3

Recipes for preparation of hydrophobic and hydrophilic silica-loaded peroxide cross-linked NRs (from Table 1 in Ref. [8])

Sample

NR-P-0RX

NR-P-10RX

NR-P-20RX

NR-P-30RX

NR-P-40RX

NR-P-60RX

NR-P-80RX

NR

100

100

100

100

100

100

100

DCPb (phrc)

1

1

1

1

1

1

1

Silica RXd (phr)

0

10

20

30

40

60

80

Sample

NR-P-0VN

NR-P-10VN

NR-P-20VN

NR-P-30VN

NR-P-40VN

NR-P-60VN

NR-P-80VN

NR

100

100

100

100

100

100

100

DCPb (phrc)

1

1

1

1

1

1

1

Silica VN3e (phr)

0

10

20

30

40

60

80

aCross-linking conditions: 30 min at 155 ℃ under 100–150 kg/cm2

bDicumyl peroxide

cGrams per one hundred grams of rubber

dAEROSIL RX200 (trimethyl silyl group treated silica, average radius = ca. 12 nm) from EVONIK DEGUSSA JAPAN CORPORATION

eNipsil VN-3 (average primary diameter = ca. 16 nm) from TOSOH SILICA CORPORATION

In Fig. 4.7 is shown the 3D-TEM results of three specimens both from RX and VN series [8, 9, 10]. The hydrophobic silica (RX series) shows relatively homogeneous distribution in the rubber matrix within the compounding range from 10 to 80 phr. On the other hand, hydrophilic series shows a distinctive inhomogeneity compared with the corresponding hydrophobic ones. The image of VN10 clearly shows larger sizes of the aggregates than that of RX10. These results are fully consistent with a traditional interpretation, i.e., in hydrophilic VN silica, silica-to-silica interaction is more enhanced than silica-to-rubber one. In contrast, silica-to-rubber interaction may be comparable to silica-to-silica interaction in hydrophobic RX silica.
Fig. 4.7

3D-TEM images of hydrophobic silica-loaded peroxide-cured NR (NR-P-10RX, -30RX and -80RX) and hydrophilic silica-loaded peroxide-cured NR (NR-P-10VN, -30VN and -80VN) (from Fig. 5 in Ref. [8]). See Video 4.3 in Supplemental Electronic Material

For the sake of further discussion on 3D-TEM results, the nearest distance (dp) between the silica aggregates is defined as shown in Fig. 4.8 [11]. In this figure, the silica aggregate is approximated by a perfect sphere of the equivalent volume, and the distance is defined on the line connecting the centers of gravity (x in the figure) of the adjacent two aggregates. The dp is calculated from the 3D-TEM images shown in Fig. 4.7. In Fig. 4.9 are shown dp and its standard deviation (STD (dp)), which are plotted against the silica amount shown in phr. Figure 4.9a suggests that both dps decrease by the increase of compounding amount and are gradually converging to a constant value. The larger dp of hydrophobic silica up to 30 phr is the reflection of its larger aggregate’s size, but the two kinds of silica seem to have converged to one common constant value of 1.3 nm approximately [12].
Fig. 4.8

Definitions of the closest distance (dp) between two neighboring aggregates of nanofillers as 3D parameters

(from Fig. 3 in Ref. [11])

Fig. 4.9

Dependence of closest distance (dp) between the two nearest silica aggregates and its standard deviation (STD (dp)) on silica loading

(from Fig. 9 in Ref. [12])

This value, dp = 1.3 nm, suggests that this is a specific rubber layer preventing the silica aggregates from the direct contact and assigned to the bound rubber or the immobilized layer (see  2.5.2). The fact that the two kinds of silica of different surface polarity have showed the same dp value seems to be due to the maximal high filling up of the particles. The smaller decrease of VN silica up to 40 phr (see Fig. 4.9a) may suggest that the stronger filler-to-filler interaction due to the hydrogen bonding by the presence of surface silanol group has promoted the formation of nanofiller network structure intermediated by the bound rubber, which is described later, at the lower filling regions.

In terms of the loaded silica amount dependency of STD (dp) (Fig. 4.9b), both VN and RX showed approximately the same decreasing tendency. This is assumed to be the reflection of a general trend, i.e., the more amount of compounded filler is to result in the more homogenized distribution of the fillers in rubber matrix.

Since polarity is due to an electrical interaction, electrical resistivity difference between VN and RX series is an interesting subject to be elucidated. Figure 4.10 shows the volume resistivity behaviors of the two silica series [8, 10]. Nonpolar RX silica showed much higher resistivity than VN silica, and the value remained almost constant at an electrical insulator level (1015 Ω cm) even with the increase of loading amount. On the contrary, the volume resistivity of VN silica showed marked decrease to reach a constant value of much smaller level (1012 Ω cm) at the loading of 40 phr VN silica. This behavior is well known as an electrical percolation phenomenon [13, 14, 15]. Silica itself is not electron conductive, but the surface of hydrophilic silica may absorb some moisture and possibly a few polar impurities, which are assumed to be the electron carrier. The similar silica amount dependencies of the two independent quantities, the resistivity and the distance between the two neighboring primary aggregates, suggest that electron can hop over the 1.3 nm layer of insulating rubber to conduct electricity. It is estimated that the silica aggregates have percolated to form a three-dimensional network structure at 40 phr of silica loading in the rubber matrix. Additionally, the bound rubber or immobilized rubber layer of 1.3 nm thickness between the silica aggregates has not inhibited the electron conduction. This distance, 1.3 nm, may allow hopping of electron or permit the permeation of electron by the quantum mechanical tunneling effect [14, 15, 16, 17].
Fig. 4.10

Dependence of volume resistivity (ρv) at room temperature on silica loading of hydrophilic silica- and hydrophobic silica-loaded peroxide-cured NR

Thus, the results so far obtained, i.e., the similar loading amount dependences of the resistivity and dp suggest that the structuring of particulate silica leads to the formation of a silica network, which is due to further clustering of the primary aggregates of silica to higher aggregates, and ultimately to an agglomerate. The visualization of this nanofiller network from the 3D-TEM image is carried out. The procedure is to be explained: Toward the higher aggregate (including the agglomerate), the primary aggregates encircled by bound rubber are forming a cluster. Namely, they are connected via rubber layer of 1.3 nm thickness as suggested by dp in Fig. 4.9. (In the case of carbon black, the thickness is different. See next section.) As shown in Fig. 4.8 defining dp, the two center of gravities of the adjacent aggregates of 1.3 nm distance are connected by a line and continue this lining on every neighboring pair in a higher aggregate. The lining is, of course, to be done only on the adjacent pair of 1.3 nm distance. Figure 4.11 shows this lining to result in the skeletonization [8], where jungle gym-like skeletons of various sizes are seen.
Fig. 4.11

Visualization of the silica networks by skeletonizing the 3D-TEM images. Hydrophilic silica; NR-P-10, -30, and -80: Hydrophobic silica; NR-P-10RX, -30RX, and -80RX (from Fig. 9 in Ref. [8]). See Video 4.4 in Supplemental Electronic Material

In the images of lower silica contents, color difference indicates an independent cluster. While almost all clusters are connected at 30 phr in the RX series, hydrophilic VN series shows the presence of local clusters, i.e., higher aggregates at 30 phr. At 80 phr, both RX and VN series suggest the agglomeration stage. In other words, gelation of the primary aggregates is completed to give a nanofiller network. Qualitatively speaking, this finding is in conformity with the results shown in Fig. 4.7, and both are explainable by the higher filler-to-filler interaction of VN silica than RX silica. Detailed observation of the images in Fig. 4.11 reveals the nanofiller networks being composed of cross-linking chains, branching chains, and isolated chains (though only a few in Fig. 4.11) just like cross-linked rubber networks.

Figure 4.12 schematically represents a network structure of the nanofiller network in general [8, 11]. The black arrow shows connection to the overall network structure. Here, numbers of the cross-linked chains, the branched chains and the isolated chains of the silica aggregates are designated as N.NdNd, N.NdTm, and N.TmTm, respectively, and their fraction as Fcross, Fbranch, and Fisolate are defined as follows:
Fig. 4.12

Silica aggregate network and its parameters

(from Fig. 4 in Ref. [11])

$$ \begin{gathered} F_{{{\text{cross}}}} = {\text{N}}{\text{.NdNd}}/({\text{N}}{\text{.NdNd}} + {\text{N}}{\text{.NdTm}} + {\text{N}}{\text{.TmTm}}) \hfill \\ \hfill \\ \end{gathered} $$
(4.1)
$$ \begin{gathered} F_{{{\text{branch}}}} = {\text{N}}{\text{.NdTm}}/({\text{N}}{\text{.NdNd}} + {\text{N}}{\text{.NdTm}} + {\text{N}}{\text{.TmTm}}) \hfill \\ \hfill \\ \end{gathered} $$
(4.2)
$$ F_{{{\text{isolate}}}} = {\text{N}}{\text{.TmTm}}/({\text{N}}{\text{.NdNd}} + {\text{N}}{\text{.NdTm}} + {\text{N}}{\text{.TmTm}}) $$
(4.3)
The three fractions are calculated from the skeletonized 3D-TEM images in Fig. 4.11, and the resulted Fcross and Fbranch are plotted against the amount of compounded silica in Fig. 4.13 [8, 11]. Even though the results are scattered, temporarily the following tendencies are noted: Fraction of the cross-linked chain (Fcross) increased with silica amount, but RX silica seemed to show maximum at 60 phr. Fraction of branched seemed to decrease with silica amount, but RX silica showed minimum at 60 phr. Hydrophobic RX silica’s behaviors are similar to those of carbon black (see the next section), and the behavior is in accordance with the gelation theory to be described later. In other words, 60 phr may be assumed to be the gelling point where the gel has fully extended. Hydrophilic VN silica’s behavior giving maximum or minimum is explained by the gelling being completed earlier at 60 phr, and hence 80 phr data were post-gelling ones. After the gelling point, branching is predominant over networking due to the steric hindrance.
Fig. 4.13

Fractions (Fcross and Fbranch) of cross-linked and branched chains of silica aggregates as a function of silica loading

Still more on the different behavior between the hydrophilic VN silica and the hydrophobic RX silica, optical transparency was distinctively found and the disparity was explained by the 3D-TEM results based on the gelation theory [8, 9, 11]. If interested in the gelation theory of nanofiller in rubber matrix, refer to Sect. 4.3.3.

4.3 Carbon Black Dispersion as Revealed by 3D-TEM

4.3.1 Importance of Carbon Black in Rubber Reinforcement

As explained in Chap.  2, carbon black (CB) has been keeping the most important position among so many fillers for more than one hundred years now. Accordingly, CB has been the target of huge number of studies on rubber reinforcement. Those, of course, included several TEM studies. For example, the book edited by Kraus [18] contains an excellent chapter by W. H. Hess on filler dispersion by using a conventional TEM. However, the majority of rubber products contain at least 40 phr filler (in terms of volume fraction, 0.173 for HAF). When CB in rubber is projected on a two-dimensional plane, the resultant image only displays the piled up image of CB, that is, CBs are overlapped on top of one another, and no exact 3D distribution is obtained.

Technical importance of CB in rubber reinforcement would continue for a while in spite of the most recent trend of decarbonization or carbonlessness (see Chap.  9). In consequence, the CB utilization in rubber industry is going to continue to future, and CB cannot be missed for the study on rubber reinforcement.

4.3.2 3D-Imaging of Carbon Black Dispersion and Its Analysis

Table 4.4 shows the compounding recipes of CB-loaded NR vulcanizates [19, 20, 21], which are supposed somewhat standardized ones using a delayed action accelerator, CBS, with an activator system consisting of stearic acid and zinc oxide for vulcanization of NR by sulfur. As the reinforcing filler, HAF grade CB is employed which is a popular grade for tire rubbers. Nine compounds of the varied amount of HAF were press cured at 140 ℃ for 5 min. As described in 4.2.1, the vulcanizates were subject to the pretreatment in order to remove the zinc compounds in the rubber sample.
Table 4.4

Recipes of carbon black-loaded NR compounds (from Table 1 in Ref. [21])

Compound

Ingredients/phra

CB0

CB5

CB10

CB20

CB30

CB40

CB50

CB60

CB80

NR (RSS#1)

100

100

100

100

100

100

100

100

100

Stearic acid (ST)

2

2

2

2

2

2

2

2

2

Active ZnO

1

1

1

1

1

1

1

1

1

CBSb

1

1

1

1

1

1

1

1

1

Sulfur

1.5

1.5

1.5

1.5

1.5

1.5

1.5

1.5

1.5

Carbon black (CB)c

0

5

10

20

30

40

50

60

80

aPer 100 g of rubber

bN-cyclohexyl-2-benzothiazole sulfenamide

cHAF, dried at 120 ℃ for 2 h

In Fig. 4.14 is shown the 3D-TEM images of four specimens from the nine vulcanizates (cured at 140 ℃ for 15 min) from the compounds shown in Table 4.4 [21, 22]. In these images, CB is represented by the white. An aggregate separated more than 1 nm from the adjacent is differentiated by the degree of whiteness or color as shown in Fig. 4.15 [21, 22]. From Figs. 4.14 and 4.15, dispersion of the aggregates is not homogeneous and the presence of them is much localized at the lower CB loading. With the increase of CB-loading amount, clustering of the aggregates is more and more observed up to the apparently full packing at 80 phr.
Fig. 4.14

3D-TEM images of CB-10, 20, 40, and 80 (from Fig. 3 in Ref. [21]). See Video 4.5 in Supplemental Electronic Material

Fig. 4.15

3D-image analysis results for CB-10, -20, -40, and -80 (from Fig. 4 in Ref. [21]). See Video 4.6 in Supplemental Electronic Material

Those 3D images enabled us to calculate dp and STD (dp), which are plotted in Fig. 4.16 against the CB-loading amount [22, 23, 24]. As observed in the case of silica (see Fig. 4.9), increase of CB amount resulted in a drastic decreasing of dp, and more than 40 phr gave an almost constant value. The convergent value is approximately 3 nm, which suggests the presence of the rubber layer inhibiting the direct contact of CB aggregates. In other words, bound rubber on the surface of CB aggregates is ultimately compressed to 3 nm thickness by the increasing CB amount up to 80 phr, where presumably the maximum packing of CB is observed, exactly as observed in particulate silica. This estimation is somewhat in conformity with an experimental observation that mixing of more CB than 100 phr into rubber is often difficult in operating to get a good-looking rubber compound.
Fig. 4.16

Dependence of closest distance (dp) between the two nearest CB aggregates and its standard deviation (STD (dp)) on CB loading

(from Fig. 7 in Ref. [22])

In addition, the volume resistivity (ρv) of the CB-loaded NR vulcanizates is plotted against the CB amount in Fig. 4.17. Since CB is electron conductive, addition of CB lowered the resistivity, and after 40 phr, it tends to show an asymptomatic value. This trend is in accord with the behavior of dp shown in Fig. 4.16, and is suggesting an electrical percolation. These behaviors are exactly corresponding to those of particulate silica shown in Figs. 4.9 and 4.10. The two nanofillers for rubber, CB and particulate silica, show percolation behavior, and both seem ultimately to form the network structure in rubber matrix.
Fig. 4.17

Dependence of volume resistivity (ρv) on CB loading

(from Fig. 8 in Ref. [22])

The thickness value of 3 nm estimated from Fig. 4.17 is the distance which is most probably minimal between the CB aggregates observed at higher CB loading than 40 phr. This rubber layer is due to the bound rubber or the immobilized rubber layer (see  2.5.2). A number of papers have reported various values of the thickness of the bound rubber. Among them, most cited and hence dependable ones have been evaluated by NMR technique, and they are between 5 and 20 nm [25, 26, 27, 28]. The value 3 nm obtained here is considered to be minimal, reasonably the lowest among the reported, since the 3 nm thickness is obtained under a highly packed condition of CB aggregates in the rubber matrix. The 3 nm may be suggesting larger filler-to-rubber interaction of CB than that of silica, whose minimal value is 1.3 nm. This comparison is exactly compatible with a traditional concept of much larger filler-to-filler interaction of silica than CB, too.

In order to visualize the 3D networks of CB, the center of gravity of CB aggregate of 3 nm distance with the neighboring one is mutually connected by a line [19], which is explained in 4.2.3 on the silica networks. The resultant CB network skeletons are displayed in Fig. 4.18 [19, 23]. The scale bar shows 100 nm length. In the skeletonized images, presence of the branched chains of CB aggregates is confirmed, but the isolated CB aggregates (probably equivalent to the primary CB aggregates) are not observed. On the other hand, some isolation is recognized in the case of silica. In CB-10 and CB-20, localized networks are seen, but at the higher CB loadings, CB networks are seemingly expanded to cover the whole area. The higher conductivity of CB-40 and CB-80 is reasonably ascribed to the percolation of the conductive networks of CB, which is confirmed by the parallel relationship of dp and ρv as shown in Figs. 4.16 and 4.17. Thus, these skeletonized images have successfully visualized the percolation phenomenon of CB-loaded conductive composites as well as the network structure formation of CB in rubber matrix.
Fig. 4.18

Visualization of CB network structures and their characteristics for CB-10, 20, 40, and 80. The centers of gravity of CB aggregates are connected by a line. The aggregates are linked by a dp of 3 nm. The unconnected patterns are shown in different colors (from Fig. 17.17 in Ref. [23]). See Video 4.7 in Supplemental Electronic Material

As done on the silica networks, the parameters to describe the CB networks are to be defined on the basis of Fig. 4.12. The difference of CB from silica is lack of the isolated aggregates as mentioned earlier. Therefore, two parameters, fraction of the network chains (Fcross) and that of the branching chains (Fbranch) are defined as follows:
$$ \begin{gathered} F_{{{\text{cross}}}} = {\text{N}}{\text{.NdNd}}/\left( {{\text{N}}{\text{.NdNd}} + {\text{N}}{\text{.NdTm}}} \right) \hfill \\ \hfill \\ \end{gathered} $$
(4.4)
$$ \begin{gathered} F_{{{\text{branch}}}} = {\text{N}}{\text{.NdTm}}/\left( {{\text{N}}{\text{.NdNd}} + {\text{N}}{\text{.NdTm}}} \right) \hfill \\ \hfill \\ \end{gathered} $$
(4.5)
The results of calculation of the two parameters are shown in Fig. 4.19 [21]. The fraction of the cross-linked chains (Fbranch) initially increases up to 40 phr of CB and decreases after that. On the other hand, the fraction of the branched chains (Fbranch) shows quite contrary behavior. These results suggest that the connection of the branched chains (Fcross) to the networks is predominant up to 40 phr and above 40 phr, branching is predominant. The reason of the latter is estimated due to the steric hindrance which has delayed connection between the grown network aggregates [19, 20, 21, 23, 24, 29].
Fig. 4.19

Fractions (Fcross and Fbranch) of cross-linked and branched chains of CB as a function of CB loading. At CB loadings larger than 40 phr, the fraction of cross-linked chains decreases linearly and the fraction of branched chains increases linearly with increasing CB loading (from Fig. 13 in Ref. [21])

4.3.3 Network Formation of Carbon Black in Rubber Matrix

From the so far obtained results, formation of the CB networks is to be discussed. The process is assumed to be a kind of gelation, and the Charlesby’s theoretical treatment [30, 31, 32] is applied to the present system [4, 23, 24, 33]. In this theory, it is assumed that the structure of the overall CB network is formed by radiation-induced polymerization of monomer (the primary aggregate is here assumed to be a monomer, whose mass is m). Also, the monomer contains functional group which is able to combine with the other and the functionality is F. The F is equal to or larger than 2 (F ≥ 2). The Charlesby’s theory treats the formation of sols and insoluble polymers, gels, here interpreted as the 3D CB networks. As seen in Fig. 4.20, before the formation overall network, Ny number (y is from 1 to n) of CB aggregates of various length are present. They contain functional groups in order to convert themselves to the final CB network as mentioned. During the polymerization (here, clustering or linking) process, a certain CB aggregate (Xn) is presumed to be formed by linking of n monomers (primary aggregate whose basic mass is m, as having assumed) and to have functionality F. By assuming that there is no linking within this Xn-mer (equivalent to no intramolecular reaction assumption), the percentage of linked F (q) is equal to the probability that one F forms a link, and (1 – q) is the probability that F remains unlinked (not reacted).
Fig. 4.20

CB network formation process: Propagation to a network is by connection of CB aggregates which are from X1 to Xn units

If the total number of the primary CB aggregates that become visible under TEM is A, A and the number of linking (N.Nd), which is called as the number of cross-linking points from now on, are expressed as follows:
$$ A = \sum\limits_{y = 1}^{y = n} {N_{y} x_{y} } $$
(4.6)
where Ny is the number of xy-mer (y is between 1 and n)
$$ \begin{gathered} {\text{N}}{\text{.Nd}} = qA \hfill \\ \hfill \\ \end{gathered} $$
(4.7)
Further, the amount of loaded CB is expressed by
$$ W_{{{\text{CB}}}} = \xi m\sum\limits_{y = 1}^{y = n} {N_{y} x_{y} } $$
(4.8)
where ξ is a calibration coefficient for the CB amount under the assumption of a homogeneous distribution of CB. From Eqs. (4.6), (4.7) and (4.8), the cross-link density (N.Nd/TV) is expressed by
$$ {\text{N}}{\text{.Nd}}/{\text{TV}} = \left( {q/\xi m{\text{TV}}} \right)\,W_{{{\text{CB}}}} $$
(4.9)
where TV is the total volume under the TEM, and TV is assumed not much dependent on the CB-loading amount. Thus, the cross-link density would be proportional to the loaded amount of CB, WCB, if the Charlesley’s theory is applicable to this system.
Figure 4.21 displays the change of (N.Nd/TV) as a function of WCB [34]. In the region of WCB up to 20 phr, linearity is approximately observed with a slope of q/ξmTV passing through the origin in the figure. Consequently, this indicates that the formation of the CB network in small CB-loading region follows the Charlesby’s gelation theory. (As will be explained in Chap.  5, it is noted that in the region between 20 and 30 phr, a structural transition occurs with respect to the CB aggregates.) Moreover, a linear relationship (shown in dashed line) is observed at the region of larger CB loading than 40 phr, too. However, the slope is larger than that at the smaller CB-loading region. When the m (mass of the primary CB aggregate) and the TV are thought to be constant in Eq. (4.9), the q (hence F) and the ξ become larger and smaller at the higher CB, respectively. On the former, the larger is the CB aggregate, the larger is the F. On the latter, the aggregates are more packed and the coefficient seems to be nearly unity, i.e., both the changes seem to be reasonable at least qualitatively.
Fig. 4.21

Relationship between cross-link density (N.Nd/TV) and CB loading (WCB) (from Fig. 7 in Ref. [34])

So far, the formation of CB networks has been detailed from the observation results by 3D-TEM. These results are to be more discussed in the next chapter, in relation with the mechanistic considerations on rubber reinforcement.

4.4 Dispersion of the Other Fillers as Revealed by 3D-TEM

Recently, nanofillers other than silica and CB have been subjected to 3D-TEM or electron tomographic observation. Particularly, information on orientation is possibly obtained from the 3D images, too, on non-particulate fillers, and hence the technique is more and more widening its applicability. Some stimulating examples are briefed in this section, including those of non-rubber arena.

Koster et al. [35] reported that the 3D-TEM images of silver/natural zeolite (modelnite) crystal (Ag/NaY) afforded the exact position of particulate Ag (diameter, 10–40 nm), and the image of acid-treated modelnite suggested a 3D mesoporous structure (diameter, 3–20 nm). They further carried out the exact visualization of Ag in the meso pore of zeolite crystallite, and have concluded that the 3D-TEM technique is unparalleled for characterizing the nanostructure of solid catalysts [36]. Tanaka [37] has compared the 3D tomographic image and the topographical one of zinc oxide crystallites, and suggested that the tomography may possibly involve various artifacts to result in less resolution. Selection of the most suitable method matching with the morphological feature of the specimen may be of utmost importance in elucidating the functional properties of a complex chemical species.

Jinnai et al. [38] showed that the distance between two closest plate-like clays (about 2 nm thickness) is 0.83 nm from a 3D-TEM image of montmorillonite/ethylene vinyl acetate copolymer (EVA). Nishioka et al. [39] examined the dispersion and aggregation of organophilic montmorillonite (MMT) in MMT/EVA composite using transmission electron tomography (TEMT). As a result, it became clear that the volume fraction of the clay phase evaluated from the TEMT image is equal to the value calculated from the recipe. They also showed that the anisotropy of the organophilic MMT can be estimated by using one of three semi-axes of the approximate ellipsoid with the same volume as the MMT in EVA. Sinkler et al. [40] showed that 3D-TEM observation is the best method to quantitatively evaluate the shapes of pores and particles.

Jinnai [14] showed that comparing three-dimensional reconstruction images of zirconia (ZrO2)/polymer nanocomposites at tilt angles ±90° and ±60°, the former is much clearer than the latter. Sato et al. [41] examined an image reconstruction method for quantitatively determining the particle size, shape, and location of FePd nanoparticles in uniaxial tilt tomography using atomic number contrast (contrast depending on atomic number) of high-angle annular dark-field scanning transmission electron microscopy. Bai et al. [42] made clear that carbon nanotubes (CNTs) do not aggregate in laser-sintered CNT/polyamide 12 moldings.

Das et al. [43] reported that based on the result of 3D-TEM observation of soft graphene-filled solution-polymerized styrene-butadiene rubber (S-SBR), a graphene sheet is present in the oligomer layer of S-SBR and showed that the graphene sheets form a complex network structure in S-SBR. Natarajan et al. [44] three-dimensionally observed CNT/epoxy nanocomposites using energy filtering electron tomography. From this result, they showed that the information on CNT morphology and dispersion state with increasing volume fraction of CNT in this composite can be obtained quantitatively.

Various examples described above suggest that 3D-TEM has already been applied to various species as well as sphere nanoparticles like CB or silica. Particularly, non-sphere functional materials are potentially to be one of the next main targets for the quantitative and higher structural studies. Also to be emphasized is the advanced design of new softwares to be used in structural evaluations, which can be of crucial importance for nanomaterials in most of these applications.

Supplementary material

Video 4.1a: Movie for Fig. 4.4a 3D-TEM image of NR-mix-V. © NISSAN ARC, LTD. ALL RIGHTS RESERVED

Supplementary file1 (MPG 5263 kb)

Video 4.1b: Movie for Fig. 4.4b 3D-TEM image of NR-insitu=V. © NISSAN ARC, LTD. ALL RIGHTS RESERVED

Supplementary file2 (MPG 4610 kb)

Video 4.2a: Movie for Fig. 4.5a NR-mix-V. © NISSAN ARC, LTD. ALL RIGHTS RESERVED

Supplementary file3 (AVI 75265 kb)

Video 4.2b: Movie for Fig. 4.5b NR-insitu-V. © NISSAN ARC, LTD. ALL RIGHTS RESERVED

Supplementary file4 (AVI 75265 kb)

Video 4.3a: Movie for Fig. 4.7a RX10. © NISSAN ARC, LTD. ALL RIGHTS RESERVED

Supplementary file5 (MPG 21879 kb)

Video 4.3b: Movie for Fig. 4.7b RX30. © NISSAN ARC, LTD. ALL RIGHTS RESERVED

Supplementary file6 (MPG 28083 kb)

Video 4.3c: Movie for Fig. 4.7c RX80. © NISSAN ARC, LTD. ALL RIGHTS RESERVED

Supplementary file7 (MPG 23851 kb)

Video 4.3d: Movie for Fig. 4.7d 10VN. © NISSAN ARC, LTD. ALL RIGHTS RESERVED

Supplementary file8 (MPG 5594 kb)

Video 4.3e: Movie for Fig. 4.7e 30VN. © NISSAN ARC, LTD. ALL RIGHTS RESERVED

Supplementary file9 (MPG 5774 kb)

Video 4.3f: Movie for Fig. 4.7f 80VN. © NISSAN ARC, LTD. ALL RIGHTS RESERVED

Supplementary file10 (MPG 6037 kb)

Video 4.4a: Movie for Fig. 4.11a NR-P-10RX. © NISSAN ARC, LTD. ALL RIGHTS RESERVED

Supplementary file11 (AVI 75265 kb)

Video 4.4b: Movie for Fig. 4.11b NR-P-30RX. © NISSAN ARC, LTD. ALL RIGHTS RESERVED

Supplementary file12 (AVI 75265 kb)

Video 4.4c: Movie for Fig. 4.11c NR-P-80RX. © NISSAN ARC, LTD. ALL RIGHTS RESERVED

Supplementary file13 (AVI 75265 kb)

Video 4.4d: Movie for Fig. 4.11d NR-P-10VN. © NISSAN ARC, LTD. ALL RIGHTS RESERVED

Supplementary file14 (AVI 75265 kb)

Video 4.4e: Movie for Fig. 4.11e NR-P-30VN. © NISSAN ARC, LTD. ALL RIGHTS RESERVED

Supplementary file15 (AVI 75265 kb)

Video 4.4f: Movie for Fig. 4.11f NR-P-80VN. © NISSAN ARC, LTD. ALL RIGHTS RESERVED

Supplementary file16 (AVI 75265 kb)

Video 4.5a: Movie for Fig. 4.14a CB-10. © NISSAN ARC, LTD. ALL RIGHTS RESERVED

Supplementary file17 (MPG 5225 kb)

Video 4.5b: Movie for Fig. 4.14b CB-20. © NISSAN ARC, LTD. ALL RIGHTS RESERVED

Supplementary file18 (MPG 3502 kb)

Video 4.5c: Movie for Fig. 4.14c CB-40. © NISSAN ARC, LTD. ALL RIGHTS RESERVED

Supplementary file19 (MPG 5315 kb)

Video 4.5d: Movie for Fig. 4.14d CB-80. © NISSAN ARC, LTD. ALL RIGHTS RESERVED

Supplementary file20 (MPG 8627 kb)

Video 4.6a: Movie for Fig. 4.15a CB-10. © NISSAN ARC, LTD. ALL RIGHTS RESERVED

Supplementary file21 (AVI 75265 kb)

Video 4.6b: Movie for Fig. 4.15b CB-20. © NISSAN ARC, LTD. ALL RIGHTS RESERVED

Supplementary file22 (AVI 75265 kb)

Video 4.6c: Movie for Fig. 4.15c CB-40C. © NISSAN ARC, LTD. ALL RIGHTS RESERVED

Supplementary file23 (AVI 75265 kb)

Video 4.6d: Movie for Fig. 4.15d CB-80. © NISSAN ARC, LTD. ALL RIGHTS RESERVED

Supplementary file24 (AVI 75265 kb)

Video 4.7a: Movie for Fig. 4.18a CB-10. © NISSAN ARC, LTD. ALL RIGHTS RESERVED

Supplementary file25 (AVI 75265 kb)

Video 4.7b: Movie for Fig. 4.18b CB-20. © NISSAN ARC, LTD. ALL RIGHTS RESERVED

Supplementary file26 (AVI 75265 kb)

Video 4.7c: Movie for Fig. 4.18c CB-40. © NISSAN ARC, LTD. ALL RIGHTS RESERVED

Supplementary file27 (AVI 75265 kb)

Video 4.7d: Movie for Fig. 4.18d CB-80. © NISSAN ARC, LTD. ALL RIGHTS RESERVED

Supplementary file28 (AVI 75265 kb)

References

  1. 1.
    Y. Ikeda, A. Kato, S. Kohjiya, Macromol. Rapid Commun. 25, 1186 (2004)Google Scholar
  2. 2.
    A. Kato, Y. Ikeda, S. Kohjiya, Nippon Gomu Kyokaishi 78, 180 (2005) (in Japanese)CrossRefGoogle Scholar
  3. 3.
    S. Kohjiya, A. Kato, J. Shimanuki, T. Hasegawa, Y. Ikeda, Polymer 46, 4440 (2005)CrossRefGoogle Scholar
  4. 4.
    A. Kato, S. Kohjiya, Y. Ikeda, Rubber Chem. Technol. 80, 690 (2007)CrossRefGoogle Scholar
  5. 5.
    Y. Ikeda, A. Kato, S. Kohjiya, Y. Nakajima, Rubber Science: A Modern Approach (Springer, Singapore, 2017)Google Scholar
  6. 6.
    A. Kato, Y. Ikeda, Y. Sato, E. Nagano, Nippon Gomu Kyokaishi 87, 203 (2014) (in Japanese)CrossRefGoogle Scholar
  7. 7.
    A. Kato, Y. Isono, K. Nagata, A. Asano, Y. Ikeda, in Characterization Tools for Nanoscience & Nanotechnology, Chap. 4, ed. by C.S.S.R. Kumar (Springer, Berlin, 2014)Google Scholar
  8. 8.
    A. Kato, Y. Ikeda, Nippon Gomu Kyokaishi 87, 351 (2014) (in Japanese)CrossRefGoogle Scholar
  9. 9.
    A. Kato, Y. Kokubo, R. Tsushi, Y. Ikeda, Hydrophilic and hydrophobic silica-filled cross-linked natural rubber: Structure and properties, in Chemistry, Manufacture and Applications of Natural Rubber, Chap. 7, ed. by S. Kohjiya, Y. Ikeda (Woodhead/Elsevier, Cambridge, 2014)Google Scholar
  10. 10.
    A. Kato, A. Tohsan, S. Kohjiya, T. Phakkeeree, P. Phinyocheep, Y. Ikeda, Manufacturing and structure of rubber nanocomposites, in Progress in Rubber Nanocomposites, Chap. 12, ed. by S. Thomas, H.J. Maria (Woodhead/Elsevier, Amsterdam, 2016)Google Scholar
  11. 11.
    A. Kato, Y. Ikeda, Y. Kasahara, J. Shimanuki, T. Suda, T. Hasegawa, H. Sawabe, S. Kohjiya, J. Opt. Soc. Am. B 25, 1602 (2008)CrossRefGoogle Scholar
  12. 12.
    A. Kato, Y. Ikeda, R. Tsushi, Y. Kokubo, J. Appl. Polym. Sci. 130, 2594 (2013)CrossRefGoogle Scholar
  13. 13.
    D. Stauffer, Introduction to Percolation Theory (Taylor & Francis, London, 1985)CrossRefGoogle Scholar
  14. 14.
    M. Okano, M. Fujii, S. Watanabe, Appl. Phys. Lett. 111, 221902 (2017)CrossRefGoogle Scholar
  15. 15.
    T. Blythe, D. Bloor, Electrical Properties of Polymers (Cambridge University Press, Cambridge, 2005)Google Scholar
  16. 16.
    R.H. Norman, Conductive Rubbers and Plastics (Appl. Sci. Publishers, London, 1970)Google Scholar
  17. 17.
    V.E. Gul, Structure and Properties of Conducting Polymer Composites (VSP, Utrecht, 1996)Google Scholar
  18. 18.
    G. Kraus (ed.), Reinforcement of Elastomers (Interscience, New York, 1965)Google Scholar
  19. 19.
    S. Kohjiya, A. Kato, T. Suda, J. Shimanuki, Y. Ikeda, Polymer 47, 3298 (2006)CrossRefGoogle Scholar
  20. 20.
    S. Kohjiya, A. Kato, J. Shimanuki, T. Hasegawa, Y. Ikeda, J. Mater. Sci. Lett. 40, 2553 (2005)CrossRefGoogle Scholar
  21. 21.
    A. Kato, J. Shimanuki, S. Kohjiya, Y. Ikeda, Rubber Chem. Technol. 79, 653 (2006)CrossRefGoogle Scholar
  22. 22.
    S. Kohjiya, A. Kato, Kobunnshi Ronbunshu 62, 467 (2005) (in Japanese)CrossRefGoogle Scholar
  23. 23.
    A. Kato, Y. Ikeda, S. Kohjiya, Carbon black-filled natural rubber composites: Physical chemistry and reinforcing mechanism, in Polymer Composites Volume 1: Macro- and Microcomposites, Chap. 17, ed. by S. Thomas, K. Joseph, S.K. Malhotra, K. Goda, M.S. Streekala (WILEY-VCH, Weinheim, 2012)Google Scholar
  24. 24.
    A. Kato, Y. Ikeda, R. Tsushi, Y. Kokubo, N. Kojima, Colloid Polym. Sci. 291, 2101 (2013)CrossRefGoogle Scholar
  25. 25.
    K. Fujimoto, Nippon Gomu Kyokaishi 37, 602 (1964)CrossRefGoogle Scholar
  26. 26.
    S. Fujiwara, K. Fujimoto, Rubber Chem. Technol. 44, 1273 (1971)CrossRefGoogle Scholar
  27. 27.
    T. Nishi, J. Polym. Sci. Polym. Phys. 12, 685 (1974)CrossRefGoogle Scholar
  28. 28.
    J. O’Brien, J.E. Cassshell, G.E. Wardell, V.J. McBriety, Macromolecules 9, 653 (1976)CrossRefGoogle Scholar
  29. 29.
    S. Kohjiya, A. Kato, Y. Ikeda, Prog. Polym. Sci. 33, 979 (2008)CrossRefGoogle Scholar
  30. 30.
    A. Charlesby, J. Polym. Sci. 11, 513 (1953)CrossRefGoogle Scholar
  31. 31.
    A. Charlesby, Proc. Roy. Soc. A22, 542 (1954)Google Scholar
  32. 32.
    A. Charlesby, J. Polym. Sci. 14, 547 (1954)CrossRefGoogle Scholar
  33. 33.
    A. Kato, Y. Ikeda, Netw. Polym. 33, 267 (2012) (in Japanese)Google Scholar
  34. 34.
    A. Kato, Y. Ikeda, S. Kohjiya, Polym.-Plast. Technol. Eng. 57, 1418 (2017)Google Scholar
  35. 35.
    A.J. Koster, U. Ziese, A.J. VerkLeij, A.H. Janssen, K.P. de Jong, J. Phys. Chem. B 104, 9368 (2000)Google Scholar
  36. 36.
    A.J. Koster, U. Ziese, A.J. Verkleij, A.H. Janssen, J. de Graaf, J.W. Geus, K.P. de Jong, Stud. Surf. Sci. Catal. 130, 329 (2000)Google Scholar
  37. 37.
    N. Tanaka, Kenbikyo 39, 26 (2003) (in Japanese)Google Scholar
  38. 38.
    K. Jinnai, T. Nishikawa, T. Nishi, Kobunshi Ronbunshu 62, 488 (2005) (in Japanese)CrossRefGoogle Scholar
  39. 39.
    H. Nishioka, K. Niihara, T. Kaneko, J. Yamanaka, T. Inoue, T. Nishi, H. Jinnai, J. Compos. Interfaces 13, 589 (2006)CrossRefGoogle Scholar
  40. 40.
    W. Sinkler, S.A. Bradley, U. Ziese, P.K. de Jong, Microsc. Microanal. 12, 52 (2006)CrossRefGoogle Scholar
  41. 41.
    K. Sato, K. Aoyagi, T.J. Konno, J. Appl. Phys. 107, 024304 (2010)CrossRefGoogle Scholar
  42. 42.
    J. Bai, R.D. Goodridge, R.J.M. Hague, M. Song, J. Mate. Res. 29, 1817 (2014)CrossRefGoogle Scholar
  43. 43.
    A. Das, R. Boldt, R. Jurk, D. Jehnichen, D. Fisher, K.W. Stöckelhuber, G. Heinrich, RSC Adv. 4, 9300 (2014)CrossRefGoogle Scholar
  44. 44.
    B. Natarajan, N. Lachman, T. Lam, D. Jacobs, C. Long, M. Zhao, B.L. Wardle, R. Sharma, J.A. Liddle, ACS Nano 9, 6050 (2015)CrossRefGoogle Scholar

Copyright information

© Springer Nature Singapore Pte Ltd. 2020

Authors and Affiliations

  • Shinzo Kohjiya
    • 1
  • Atsushi Kato
    • 2
  • Yuko Ikeda
    • 3
  1. 1.Kyoto UniversityKyotoJapan
  2. 2.Department of Automotive AnalysisNISSAN ARC, LTD.YokosukaJapan
  3. 3.Faculty of Molecular Chemistry and Engineering, Center for Rubber Science and TechnologyKyoto Institute of TechnologyKyotoJapan

Personalised recommendations