Nano-scale morphological analysis of graphene–rubber composites using 3D transmission electron microscopy

Amit Dasab, Regine Boldta, René Jurka, Dieter Jehnichena, Dieter Fischera, Klaus Werner Stöckelhuber*a and Gert Heinrichac
aLeibniz-Institut für Polymerforschnung, Dresden e. V., Hohe Straße 6, D-01069 Dresden, Germany. E-mail: stoeckelhuber@ipfdd.de; Fax: +49 351 4658 362; Tel: +49 351 4658 579
bTampere University of Technology, Korkeakoulunkatu 16, Fi-33101 Tampere, Finland
cTechnische Universität Dresden, Institut für Werkstoffwissenschaft, D-01069 Dresden, Germany

Received 26th November 2013 , Accepted 21st January 2014

First published on 23rd January 2014


Abstract

In this work three-dimensional transmission electron microscopy (3D-TEM) is exploited to characterize a soft graphene based nano-composites structure and the constituted morphology in a qualitative way. The reconstruction of the two dimensional slides into a three dimensional tomographic image is a powerful tool, when the images of the nano-object are reflected into a quasi-distinguishable object due to superposition effect. By using this technique it is possible to mark the contour area of the nano-sized object inside the soft rubber matrix. To extract information about the filler network, the clustering process of the fillers or the existence of single or multiple graphene sheets, a solution polymerised styrene butadiene rubber was selected as a soft matrix which was filled with carbon black (CB) and graphene nano-platelets (GnP). The dispersion/exfoliation of the stacked graphene sheets into individual single sheets was facilitated by the presence of carbon black in the system as understood from TEM, X-ray diffraction and Raman spectroscopic studies. The existence of oligo-layer graphene sheets was detected by this 3D-TEM, especially when the rubber matrix exists in a complex morphology arisen from filler–filler networks in all spatial dimensions.


Introduction

3D tomographic reconstructed imaging is a rapidly growing technology that intersects the fields of chemistry, physics, engineering and biomedical applications. As a rapid growth of the nano-material usage in material science has taken place, a detailed insightful characterization of the microstructure of the nano-material is necessary. Specially, for soft materials that are filled with nano-particles, the microstructures developed by nano-particles are mainly governing the ultimate performance of the materials. In this respect 3D image analysis at the molecular dimension is a very interesting tool to explore the structure–property relationship of these nano-materials. For instances, Ikeda et al. published a detailed 3-dimensional nano-structure of the aggregated carbon black (CB) in natural rubber matrix.1 They noted that most of the CB particles are existing in aggregated and agglomerated form, even at that condition when the volume fraction of the CB was very low. In another work, the same group compared the 3-dimensional structure of precipitated silica with in situ synthesized nano-silica in the soft natural rubber matrix.2 The in situ silica, synthesized in their work was found to be with larger primary particle size as compared with externally added precipitated silica. The size of the silica particles synthesized by in situ sol–gel method, however, could be controlled by the experimental condition like concentration of the catalyst, type of the catalyst, temperature of the reaction, and other external conditions.3–5 In an another work dealing with 3D tomography of CB filled natural rubber, Kohjiya et al. showed that the critical minimum distance which is necessary to form a percolating network of the CB particles is nearly 3 nm.6 So electrons can hop through the rubber layers of 3 nm thickness and in this way the aggregates of carbon black can form electron-conductive pathways, to give rise to a conductive path of network. Even they also observed that this critical distance of 3 nm between the carbon black aggregates was not decreased even at higher concentration of carbon black. Describing the basis entity of CB as form of aggregates they proposed the bound rubber, the gel like non-extractable portion of the rubber, can only cover the aggregates but not all primary particles. In a 3D-TEM study of a hybrid filler system of silica (Si) and CB particles in a mixture of natural rubber and polybutadiene, Jinnai et al. reported that they were able to distinguish the two different filler systems; additionally they observed that each aggregate was made of only one species – not a single aggregate contained both the CB and Si nanoparticles.7 A detailed review article on 3D tomography of rubber composites also can be found in literature.8 However, publications related with the characterization of graphene rubber composites by 3D tomography are not reported yet.

Though TEM is supposed to be a tool to visualize the dispersion quality of various nano-sized filler materials inside a polymer matrix without any special sample preparation methods (like staining with electron rich materials), it is rather difficult to apply simple TEM technique to visualize graphene in rubbers.9 Owing to the similar atomic composition of carbon both in the host polymer and graphene itself there are a small contrast between them. Particularly, graphene with extremely thin molecular layer cannot be easily detected by TEM technique especially when it is embedded in the polymer matrix in the form of a wrinkled and cramped paper like pattern.9 Also other characterization techniques, such as scanning electron microscopy SEM and transmission optical microscopy (TOM) can be used to detect the graphene fillers in polymer matrix. However, in the case of SEM, the detection of graphene filler particles is limited to the characterisation of superficial objects at the sample or at a fracture surface, where the dispersion of the graphene particles can be very different from inside the polymer matrix. TOM provides valuable information of the macro dispersion of filler particles, i.e. the occurrence of agglomerates in micrometre range, but cannot deliver information on the dispersion of particles in nanometre range due to the limited resolution of light microscopy.

Experimental

In this work 4 rubber samples based on solution styrene butadiene rubber (SSBR) were prepared. The solution styrene butadiene rubber (S-SBR) used in this work was Buna VSL 5025-0 HM kindly provided by Lanxess Deutschland GmbH. The Mooney viscosity ML(1 + 4) of this solid rubber was 65 with styrene and vinyl content 25% and 50% respectively. The graphene nanoplatelets GnP of the grade xGnP-M-5 were bought from XG Sciences Inc, USA. The average diameter and bulk density of the graphene nano-platelets were 15 μm and 0.05 g cm−3 respectively. These graphene nano-platelets were consisting with a number of graphene sheets (at least several hundreds) attached like graphite rather than single sheet. The carbon black used in this study was N330 provided by Orion Engineered Black. The rubber compounds were prepared following the recipe given in Table 1. The rubber compounding was done in two stages. In the first stage rubber was mixed with zinc oxide, stearic acid and the desired amount of fillers in an internal mixer (Thermo Fisher Rheomix 3010 OS). For mixing, the rotor speed was set at 70 rpm; mixing temperature was 70 °C. All fillers were added in 2–3 min and the mixing was further proceeded for 5 min. After that the rubber compound was taken out from the mixer. In the second stage the curatives were added to rubber compound using a two-roll mill (Servitec Polymix 110L). In the two-roll mill the temperature was set at 40 °C and rubber compound was processed for 10 min. The curing time (t90 + 10% at 160 °C) for the rubber compounds was determined using a rubber process analyzer (Scarabaeus SIS V-50).
Table 1 The position of different bands in the Raman spectra of CB, GnP and its composites
Sample Material D band (cm−1) G band (cm−1) D* band (cm−1) D/G ratio
GnP Graphene nano-platelets (GnP) 1355 1582 2720 0.11
N-330 Carbon black (CB) 1355 1598   1.12
SBR–N330 Styrene butadiene rubber–CB composites 1350 1598   0.89
SBR–N330–GnP (dark area) Styrene butadiene rubber–GnP–CB composites 1350 1598 2720 1.00
SBR–N330–GnP (light area) Styrene butadiene rubber–GnP–CB composites 1350 1587   0.64


A hybrid compound consisting of 35 phr (parts of hundred parts rubber) CB and 5 phr graphene nano-platelets (GnP) was used for the 3D-TEM study of the graphene filler visualisation. As control standard compounds also a gum SSBR rubber and compounds filled with 5 phr GnP, and 40 phr CB were considered. The main objective of this study was to find out, if it is possible, to visualize very thin graphene platelets either in the form of single sheets or in the form of stacks of multiple sheets embedded in the three dimensional filler network of CB.

The samples were sliced with a diamond knife (35° cut angle, DIATOME, Switzerland) into ultra-thin sections of 30 nm thickness at −150 °C on a Leica ultracut UC7 cryo microtome. A tilt series of 69 BF-TEM images was recorded on a TEM Philips CM200FEG by tilting the specimen from −68° to +68° with an increment of 2° at a magnification of 39.000×, resulting in a pixel size of 1.5 nm.

For alignment and 3D reconstruction the most used publicity available software IMOD was used. The reconstruction algorithm used is W-SIRT a combination of Weighted Back Projection and Simultaneous Iterative Reconstruction Technique developed by Wolf.10 Isosurfaces rendering of the 3D volume were carried out by AMIRA software.

Raman spectra were taken using the Raman Imaging Microscope alpha300R (WITec GmbH, Ulm, Germany) with a laser wavelength of 532 nm and a laser power of 0.1 mW. The samples were measured with an objective with magnification 20 times and an integration time of 0.5 s for one single spectrum, which was accumulated 200 times. The spectra were smoothed by the Savitzky–Golay method and baseline corrected (subtraction of the fluorescence background).

XRD experiments were executed by means of a 2-circle diffractometer XRD 3003 Θ/Θ (GE Sensing & Inspection Technologies GmbH, Seifert-FPM, Freiberg/Sa., Germany) using Cu Kα radiation in the measuring region of 2θ = 0.5 to 50° in steps of 0.05°.

Results and discussion

An earlier study11 using commercially available graphene in solution styrene-butadiene rubber (SSBR) revealed that the mechanical reinforcement was rather poor, even if the loading of the graphene nano-platelets (GnP) was 20 phr. The hybrid composite showed stronger mechanical and dynamic mechanical properties: a tensile strength of 21 MPa, 100% modulus of 3.4 MPa and an elongation at break values of 460%, whereas for the 40 phr CB composite the corresponding values are 20 MPa, 2.5 MPa and 410%, respectively, were obtained. That means that this hybrid composition can offer a higher reinforcement than a composite of 40 phr CB (N330). Whereas an addition of 5 phr GnP to SSBR alone, without the addition of other fillers, did not improve the mechanical properties at all, compared to the unfilled material.

An explanation for this effect might be that the presence of CB facilitates the mechanical dispersion of GnP by a delamination process. Therefore, the composite comprised with 35 phr CB and 5 phr GnP was further considered for in depth TEM investigation, aiming to visualize few layer graphene sheets directly, if there are some. At first, conventional 2D images of these composites were taken, which are shown in Fig. 1.


image file: c3ra47050d-f1.tif
Fig. 1 Transmission electron microscope images of SSBR filled with 35 phr CB and 5 phr GnP.

At lower magnification (Fig. 1a) the typical structure of carbon black in aggregated form could be easily detected. Additionally, some grey areas are also observed, due to the presence of GnP. At higher magnification (Fig. 1b and c) the existence of aggregates of the primary particles of CB can be observed,12 rather than the aggregated morphology of the CB. It is well known, that a agglomeration of the CB aggregates can occur due the so-called filler flocculation.13 But, a distinct filler–filler network structure due to this flocculation process was not clearly visualized in all these 2D micrographs, although the concentration of CB used in the formulation was above the percolation threshold.

So, a simple two dimensional image cannot explain a 3-dimensional filler–filler network in the present case. In addition to the typical CB structure one dark grey shade could be observed at higher magnification (Fig. 1c). From these figures conclusions on the dispersion state of the graphene – thin layers or stacks of a couple of layers – cannot be drawn.

However, in Fig. 1b the thick dark line marked by an arrow is the laminar structure of a graphitic structure, as it was further magnified (Fig. 1d). But, after careful observation of this images (Fig. 1b), one can find grey shades around the dark lines of GnP. Owing to the very less electron density the ultrathin exfoliated graphene layers, these could show such grey layers near the dark thin phase. So a complete exfoliation of the GnP could not be achieved by simple mechanical mixing of GnP. However, typical grey area, which can be clearly observed from Fig. 1b and c resemble the possible existence of exfoliated and crumpled thin flakes of graphene.

Another point should be discussed here about the virtual distribution of the CB. A lot of spherical CB particles with different shades (from black to very light grey) are seen to be well distributed over the total sight field under the focusing area. But here we do not know whether these particles were remained in the same plane parallel to the examined surface or they are far below from other particles. The same comment can be made for graphene layers.

Additionally, if very thin graphene layers remain parallel to the surface of the visualized plane of the specimen, it is nearly impossible to trace it by this contrast image. But if it is orientated perpendicular to the surface of the specimen under examination, it shows a higher contrast image with an impression of a dark line. So, these 2D images cannot explain the above details. However, more information could be gathered, if the morphology is examined in 3D.

Fig. 2 displays images from reconstructed images at 4 different angles. A movie with a full rotation around a tilted axis from +68° to −68° can be found in the ESI.


image file: c3ra47050d-f2.tif
Fig. 2 Transmission electron microscope images of a tilt series of SSBR filled with 35 phr CB and 5 phr GnP. The images were taken at different angles: (a) 2°, (b) 10°, (c) 18° and (d) 24°. The dashed line in (a) shows the tilting axis of this series of TEM micrographs. For a more detailed view please see also the video clip in the ESI TEM_tomo_ts.mpg.

The aggregates of CB particles can be found with a mean single particle diameter nearly 30 nm. Though a range of aggregate sizes is seen, it is very difficult to measure its sizes, even if the total rotation about 130° is taken into consideration. Two aggregates can appear – due to the superposition of their images – as one large, even if they are located in depth rather far away from each other. In the movie the presence of very thin layer of graphene with an appearance like a saddle rectangle could be observed. The diameter of this particle is ∼500 nm. Most probably, this particle is consisting of a very few layers (1–4) of graphene sheets. Further study of this 3D object exhibited that a couple of delaminated layers are formed by the mechanical dispersion of the GnP particles and remained in the vicinity of each other. It is also observed that owing to the comparable surface energy of CB with graphene the CB particles and the graphene sheets show a co-flocculation behaviour, because the work of adhesion between CB and GnP is high, compared to the filler/polymer matrix adhesion.14,15 Particles with different surface energies, as CB and silica do not show this co-flocculation effect, which can be calculated by the difference of the work of adhesion values between fillers and rubber matrix,14,15 and is also found experimentally by 3D-TEM.7 An upwards movement of a small dark spot can also be found during the rotation of the sample (Fig. 2). Obviously, this is originated from more transmitted electrons or the generation of higher electron density from two thin sheets in a spatial position like two blades of a scissors. As the rotation of the specimen takes place the junction of the two graphene layers are seen from different angles and the resulting interference diffraction of the electrons can be observed.

Fig. 3 gives an schematic depiction of the reconstruction of 3D tomography image by backprojection of images, recorded at different tilt angles.


image file: c3ra47050d-f3.tif
Fig. 3 Schematic depiction of 3D-TEM tomography by reconstruction via backprojection: the composite is sampled by projection from several angles and then reconstructed by backprojection of these projections at the original sampling angles into the object space.

Fig. 4 shows the 3D reconstruction of the composite material, filled with 35 phr CB and 5 phr GnP. Hereby, the yellow and green regions represent CB and GnP filler particles respectively. In Fig. 4b and c the resultant volume on the left side and a slice in z-direction on the right side is depicted. A video to visualize the image at different depth can be also found in the ESI TEM_tomo_rec.mpg.


image file: c3ra47050d-f4.tif
Fig. 4 (a) Visualization of the 3D reconstruction of SSBR filled with 35 phr CB and 5 phr GnP, yellow and green regions represent CB and GnP, respectively, (b and c) present the resultant volume on the left side and a slice in z-direction on the right side.

It can be seen from Fig. 4b and c that the filler particles – both CB and GnP – are present in the form of agglomerated structures. Obviously, the distribution of the agglomerated particle size is very broad. Mostly, this inhomogeneous structure of the filler was formed by a co-flocculation of the different filler particles, caused by the similar surface energies of GnP and CB, which leads then to a hybrid CB–GnP agglomerated structure. A slice from upper part of the reconstructed cube is shown in Fig. 4b. From this figure the uneven distribution of CB particles is observed. However, the distribution of the CB particles seems to be better in the 2D images (Fig. 1). So, a localization of different CB particles at different depth yields the 2D image as better distribution (virtual) of the CB particles, but the real fact is different as observed from the 3D tomography. Though the reconstructed image (Fig. 4a and c) did not show any platelet character of GnP, but the thin slice from bottom of this tomographic image a plate like geometrical object can be observed (Fig. 4c). Here it is important to note that the total volume fraction of the fillers used here is ca. 20% (35 phr CB and 5 phr GnP); a 3D network structure of the fillers cannot be observed. This suggests that the ultrathin structure of GnP could not produce a detectable network structure by imaging technique which is unlike to the work of Ikeda where CB was used as reinforcing agent in NR matrix.16 However, a colour image of the mass density distribution (Fig. 4a) confirms the existence of diffused graphene layers in the rubber matrix.

The presence of sp2 hybridized planar structures in both CB and GnP were indicated by Raman spectroscopy by the G band at 1582 cm−1 and 1598 cm−1 respectively (see Fig. 5a–c). Both, graphene and also CB possess graphitic layers. After incorporation of these fillers a slight change of this band towards higher wavenumbers was found, when the spectra were taken from the light area of the micrographs. However, in the dark area the band appears at the same position as pure CB, whereas CB filled systems do not show changes of this band. The D band, which indicates defects and disorder structures of the carbon network (sp3 hybridization) with a wavenumber of 1355 cm−1, can be designated for both CB and GnP. In both of the materials this band is shifted to 1350 cm−1, when they were embedded in rubber matrix. On the other hand, GnP exhibited a second order band of the D band at 2720 cm−1, designated here as D*. This D* band is a characteristic peak of a graphene like structure. This band is found to be very sharp with a symmetrical appearance, when the graphene exists in monolayers. A plenty of layers make the band broader and asymmetric. After mechanical dispersion of GnP, however, the D* band does not show any changes in our case. Additionally, compared to the CB, the relatively increased intensity of the disorder mode suggested that after incorporation of GnP the degree of order in the hexagonal framework of graphene is decreased. The intensity ratio ID/IG, which indicates the disorder in a sample, is 0.11 for GnP. Hereby a higher intensity ratio indicates a higher degree of disorder. In rubber filled with CB and GnP, this ratio was 1.00 in dark areas, and 0.64 measured in bright areas. In both cases, the increase of the D band intensity clearly indicates a larger amount of defects on the GnP surfaces. The bright areas have obviously a higher content of GnP and therefore a higher order than the dark areas, which probably consist mostly of rubber and CB and a lower content of GnP in comparison to the brighter areas.


image file: c3ra47050d-f5.tif
Fig. 5 Raman spectra of (a) pure carbon black (CB, blue) and graphene nano-platelets (GnP, red), (b) SSBR filled with CB and GnP bright area (red) and dark area (black), (c) SSBR filled with CB (green) and filled with CB and GnP (black), (d) XRD of the CB, GnP and the SSBR composites at different compositions.

Fig. 5d shows the XRD patterns of GnP, CB and its related rubber composites. S-SBR, being an amorphous polymer, itself also gives rise to diffuse X-ray scattering. This diffuse halo is found in our case with a maximum at 2θ = 19.65° in both cases filled with either hybrid filler or only CB. On the other hand pure GnP showed a strong and sharp (002) peak at 2θ = 26.60° with a d-value of 0.335 nm. This peak appears due to the graphitic structure of GnP, as this commercial materials is comprised with a large number of stacked layers of graphene. This peak is still present in the sample prepared by the use of GnP in rubber even at low concentration at different loading (see ESI). Since the concentration of GnP was only 5 phr, the associated peak appears with a weak intensity, but without altering its position. However, after incorporation of GnP along with CB the signature of this graphitic diffraction is missing. The disappearance of this graphitic peak in the hybrid filler system could be an indication of the existence of single graphene layers without any ordered structure. Most probably, the presence of CB could enhance the dispersion of multilayers GnP into mono layers of graphene during high energy mechanical mixing. Moreover, owing to similar work of adhesion between CB and GnP and the high energy mixing process, the peeling of graphene sheets took place by overcoming the van der Waals attraction and π bonds between the graphitic structures of GnP. It should be mentioned here that only a tiny portion of the rubbers was taken in the XRD experiment and it did not convey the message of the global scale. In the 2D-TEM the presence of multiple graphitic layers of graphene was detected. That means the composites prepared here contains a bimodal characteristic of graphitic as well as few layers of graphene structures.

Conclusions

The existence of very thin oligo-layers of graphene was clearly observed by the 3D TEM technique. We also found that in our hybrid filler rubber system the graphene layers are surrounded by CB and other thin layers of GnP resulting into a rather complex morphology of the filler network structure. The presence of CB helps the delamination process of GnP (a multiple stacked layers of graphene) into individual sheets of graphene. Appearance of dark spots or lines in the 2D TEM does not always mean a laminated structure of graphite but it may appear due to interference from different single layers, embedded in different depth of the rubber matrix. Hereby, the 3D tomography can provide a useful method to detect and recognise individual objects, i.e. to distinguish different filler platelets from each other.

It can be finally concluded that for in-depth understanding of the morphology of complex three dimensional nano-sized fillers 3D tomography is a valuable tool and can yield some qualitative and potentially quantitative data that can be used in prediction of the materials behaviour under working conditions.

Further studies are required to understand, how the presence of CB and the type of CB are influencing the dispersion behaviour of GnP.

Acknowledgements

The authors thank the German Federal Ministry of Education and Research (BMBF) for financial support of the project ELAGRA within the framework of the BMBF program “Werkstoffinnovationen für Industrie und Gesellschaft” – WING. Grant number: 03X0110B. We are also grateful to Dr Daniel Wolf, TU Dresden, Speziallabor für Höchstauflösende Elektronenmikroskopie und Elektronenholographie Triebenberg for the valuable support. Manfred Klüppel, Markus Möwes and Christian W. Karl, DIK Hanover, are acknowledged for helpful and fruitful discussions.

Notes and references

  1. S. Kohjiya, A. Kato and Y. Ikeda, Prog. Polym. Sci., 2008, 33, 979–997 CrossRef CAS PubMed.
  2. Y. Ikeda, A. Kato, J. Shimanuki, S. Kohjiya, M. Tosaka, S. Poompradub, S. Toki and B. S. Hsiao, Rubber Chem. Technol., 2007, 80, 251–264 CrossRef CAS.
  3. L. Dewimille, B. Bresson and L. Bokobza, Polymer, 2005, 46, 4135–4143 CrossRef CAS PubMed.
  4. Y. Ikeda, A. Katoh, J. Shimanuki and S. Kohjiya, Macromol. Rapid Commun., 2004, 25, 1186–1190 CrossRef CAS.
  5. W. Stöber, A. Fink and E. Bohn, J. Colloid Interface Sci., 1968, 26, 62–69 CrossRef.
  6. A. Kato, J. Shimanuki, S. Kohjiya and Y. Ikeda, Rubber Chem. Technol., 2006, 79, 653–673 CrossRef CAS.
  7. H. Jinnai, Y. Shinbori, T. Kitaoka, K. Akutagawa, N. Mashita and T. Nishi, Macromolecules, 2007, 40, 6758–6764 CrossRef CAS.
  8. S. Kohjiya, Mater. Sci. Forum, 2006, 514–516, 353–358 CrossRef CAS.
  9. M. Tosaka, R. Danev and K. Nagayama, Macromolecules, 2005, 38, 7884–7886 CrossRef CAS.
  10. D. Wolf, A. Lubk and H. Lichte, Ultramicroscopy, 2014, 136, 15–25 CrossRef CAS PubMed.
  11. A. Das, G. R. Kasaliwal, R. Jurk, R. Boldt, D. Fischer, K. W. Stöckelhuber and G. Heinrich, Compos. Sci. Technol., 2012, 72, 1961–1967 CrossRef CAS PubMed.
  12. T. C. Gruber, T. W. Zerda and M. Gerspacher, Rubber Chem. Technol., 1994, 67, 280–287 CrossRef CAS.
  13. G. G. A. Böhm and M. N. Nguyen, J. Appl. Polym. Sci., 1995, 55, 1041–1050 CrossRef.
  14. K. W. Stöckelhuber, A. Das, R. Jurk and G. Heinrich, Polymer, 2010, 51, 1954–1963 CrossRef PubMed.
  15. M.-J. Wang, Rubber Chem. Technol., 1998, 71, 520–589 CrossRef CAS.
  16. S. Kohjiya, Y. Matsumura, S. Yamashita, H. Yamaoka and T. Matsuyama, J. Mater. Sci. Lett., 1984, 3, 121–123 CrossRef CAS.

Footnote

Electronic supplementary information (ESI) available: (a) Movie from the TEM images recorded at different tilting angles (TEM_tomo_ts.mpg) and (b) from the reconstructed 3D-tomography (TEM_tomo_rec.mpg). See DOI: 10.1039/c3ra47050d

This journal is © The Royal Society of Chemistry 2014
Click here to see how this site uses Cookies. View our privacy policy here.