Rheological properties and irreversible dispersion changes in Carbon nanotube/epoxy systems.
In the recent years, the use of carbon nanoiubes (CNT) in different matrixes such as polymer melt or epoxy systems has become a popular subject of research due to the electrical, mechanical, and thermal properties of the composites now achievable (1-5),
In combination with conventional reinforcements (e.g., glass fibers), the composites show excellent properties for multifunctional engineering applications. The physical properties of CNT composites and solutions strongly depend on the network microstructure of the fillers, which is generally controlled by the thermorheological history and the processing conditions (6-10). Better dispersion quality usually results in an increase in viscosity (9-11). However, higher viscosity generally makes processing and manufacturing more complicated. When the resin is being stored, moved, infused, or wicked, filtration effects and re-agglomeration may occur (12), (13). Because of this, the rheological properties are of special importance.
Ideally, fillers dispersed in a matrix are isolated from each other at low filler concentrations fulfilling the required events to treat them as dilute suspensions. At higher filler concentrations and due to van der Waals forces, they start to interact with each other, entangle loosely, and finally form a network of agglomerates (14-16). Flow-induced ilocculation was observed and successfully modeled in a variety of fiber suspensions (17-21). Apart from the particles, the process depends strongly on the applied shear forces as well as on the temperature-dependent matrix viscosities (22).
Agglomeration can already lake place at low shear rates for sufficient long shear times. After high-shear rate treatment, the agglomeration at low shear rates is strongly accelerated. This effect is correlated with strong demixing leading to phase separation due to high shear forces (23). This initial high shear strongly changes the electrical and rheological properties and is irreversible in the investigated shear rate regime. In this study, we want to demonstrate that such shear-induced physical changes are also present at lower temperatures for higher filler concentrations. Additionally, we want to show how they affect the properties of the suspension.
Fan et al. (11) defined a good dispersion regime as a system featuring micro CNT agglomerates, which are uniformly distributed in the suspending liquid at the lower end, and, in the ideal case, single CNTs distributed in the suspending matrix. In reality, many dispersions labeled "very good" feature both kinds of microagglomerates and isolated CNTs distributed in the resin. Hence, depending on the dispersion quality, the suspension may behave like a microparticle/liquid or like a flexible fiber/suspension system.
Huang et al. (9), (10) reported on the rheological properties of MWCNT/polymer suspension and showed the relationship between viscosity, mixing time, and dispersion state. Their suspensions were produced by mechanical stirring. Stirring with high speeds for an adequately long time may lead to sufficient dispersion quality. Nevertheless, the type and shape of the used stirrer, the suspension volume, and sedimentation effects can lead to inhomogeneities within the suspension (9), (10). Better and more homogeneous dispersions are reached by calandering (7), (13). Both methods are nowadays widely used. We used both methods to specifically create differences in dispersion quality and to prove if high preshear is able to erase the sample's history as it is often argued.
Our experiments were performed with multiwall CNTs/epoxy suspensions at several CNT concentrations below 0.5 wt%. The results are meant to contribute to the understanding of nanofiber agglomeration in general. The aim of this work is to study the role of shear direction (increasing or decreasing shear rate sweeps) and the influence of different shear treatments on the rheological properties.
Epoxy suspensions filled with different contents of multiwalled carbon nanotubes (MWCNT) were prepared by mixing the epoxy resin and carbon nanotubes. The polymer matrix consisted of bisphenol-A-based epoxy resin (Araldite LY 556) obtained from Huntsman Advanced Materials (Belgium). MWCNTs "NC 7000" grown by catalytic chemical vapor deposition were supplied by Nanocyl S.A. (Belgium) with an average length of 1.5 [micro]m and average diameter of about 9.5 nm. The CNTs have a purity of 90% and surface area of about 250-300 [m.sup.2]/g according to the supplier.
All suspensions were prepared using a dissolver disk. They were stirred at 20' C for 2 h with the maximum possible rotation speed of 2000 rpm (24-26). In the following step, half of the stirred material was taken and dispersed using a three-roll mill (Exact 120S, Exact GmbH). The gap size between the rolls was 5 [micro]m, and speed was set to 20/60/180 rpm. The material was milled for 2 min and tested directly after mixing (max. 1 h between mixing and testing). No masterbatch dilution was used to avoid different mixing times.
Rheological measurements with different filler contents were carried out using a stress-controlled Ares rheometer (Rheometric Scientific) in shear flow mode with 40 mm parallel plates and 1 mm gap. The gap size was chosen to be big enough to avoid constriction of shear-induced agglomerates. All measurements were performed at 25 [degrees] C.
The measurements were done directly one after another to keep relaxation time as short as possible (sequence testing). A minimum delay time of 10 s was necessary for the instrument to establish the correct shear rate. A measurement time of 10 s was chosen. In preceding experiments, it was found out that shorter times led to erratic values while longer measurement times do not improve the measured values. Following, measurements of sequences with increasing shear rates (0.1-100 [s.sup.-1]) will be named "up" or "increasing," while measurements with decreasing shear rates (100-0.1 [s.sup.-1]) are named "down" or "decreasing."
RESULTS AND DISCUSSION
Influence of Filler Content
The interactions between CNTs are strongly affected by their concentration. The higher the concentration, the smaller the distance between the filler particles. As a result, smaller distances would lead to more and eventually stronger entanglement between the tubes (agglomeration) and in this way to an increased flow resistance. To measure the viscosity of stirred solutions containing different amounts of CNTs, a forward shear rate sweep test (0.1-100 [s.sup.-1]) was performed at room temperature. Additionally, the pure resin was measured for comparison. The results are shown in Fig. 1. The pure resin shows Newtonian behavior for shear rales up to 50 [s.sup.-1] at a constant viscosity of about 15 Pa s. A very small decrease in viscosity is visible at higher shear rates that can be explained with shear-induced heating of the system. Hence all further rheological changes can be attributed to the presence of CNTs. Suspensions with a CNT content of 0.04 and 0.1 wt% show a significantly higher viscosity of around 20 Pa s at low shear rates. Both concentrations show only little shear thinning behavior at high shear rates, reaching the same values for 100 [s.sup.-1] as the pure resin.
As one can see in Fig. 1, the slopes of viscosity values for 0.15, 0.2, and 0.3 wt% are pretty similar. Adding a higher amount of filler particles results in a more pronounced viscosity enhancement at low shear rates. As shown before by optical micrographs. CNTs tend to form agglomerates (14), (16), (23), (27). At a certain loading fraction, the agglomerates start interacting with each other and form a network. This network shows an increased resistance to shear deformation resulting in a higher viscosity. Low-shear rate enhancement was also reported for CNT/polymer melt systems previously (28), (29)
Interestingly, the viscosity of these three suspensions increases even more when shearing at low shear rates. A maximum of viscosity was found at 0.2 [s.sup.-1]. Further increase of the shear rate leads to decreasing viscosity values (shear thinning). It is believed that a shear-induced growth of the agglomerates is the reason for this increase. A balance between growth and erosion or rupture will occur place at each shear rate (2), (30). At low shear rates, agglomerates generally build up due to the dominant constructive part until the relatively weak structure ruptures again or erode due to the particle depleted matrix. With increasing shear rates, a reduction of agglomerate size starts to set in. Agglomeration is ceased, and a steady state is achieved when constructive and destructive effects cancel each other out. When the shear rate is increased further (critical shear rate), the destructive part prevails and the agglomerate size is reduced. This results in a strong viscosity decrease.
Usually, the pseudoplastic behavior of polymer solutions can be explained by stretching and arrangement of long polymer chains. For highly filled CNT/PC, Potschke et al. (29) suggested a combined model with interaction between the CNTs and the polymer chains. In an epoxy system, no long polymer chains exist. Thus, we assume the rheological properties of the suspension are mainly influenced by the fillers. The agglomerate size on the other hand is influenced by matrix viscosity and thus temperature as well as the applied shear rate.
Rahatekar et al. (16) showed similar behavior for a 0.3 wt% containing CNT/epoxy suspension. Because of the different form of the CNTs (different aspect ratio as well as entanglement state), they reached higher viscosity values even at the same loading fraction. Nevertheless, they also measured a maximum viscosity in the same shear rate regime for 0.3 wt% suspension with a similar curve shape as our 0.5 wt% suspension.
If we assume a fast changing system, the rheological properties are strongly influenced by the build up and destruction process of the agglomerates in the matrix. This means that the way of exact measurement method is very important to get comparable results. Thus, it becomes clear that a detailed description of the measurement and material may be necessary to make comparison possible.
We performed a loop-test at certain concentrations to get more information about the rheological properties under shear: the suspension was first sheared with increasing shear rates (0.1-100 [s.sup.-1]) immediately followed by a backward sweep with decreasing shear rates (100-0.1 [s.sup.-1]). The results are presented in Fig. 2. As expected, the pure resin shows no hysteresis behavior. Even for 0.1 wt% suspensions, the up and downsweep are almost identical. For higher concentrations such as 0.3 and 0.5 wt%, a different rheological response is measured for each shear sweep. The sweep with decreasing shear rates in both cases shows a constant increase of viscosity. In the case of the backward sweep, the increase in viscosity can be explained by shear induced network formation. A steady agglomerate and network build-up takes place as agglomerates coalesce while the shear rate is lowered continuously.
As reported before (23), this different behavior is supposed to be correlated with strong demixing. This can be seen in a decrease in numbers of exfoliated particles due to likely and strong particle interaction at high shear forces. Additionally, the dispersion state in the beginning of the shear sweep is assumed to be totally different from the state obtained after applying shear treatment. Directly after the dispersion process, we assume a mixture of single nanotubes, bundles of tubes forming very small agglomerates, and some residual agglomerates within the resin. Under the very first shear sweep, these particles form bigger agglomerates under low shear conditions. With increasing shear rates, these agglomerates will be reduced in size as their mechanical integrity becomes insufficient to resist the forces acting on their volume. So, in the end of the sweep ([100.sup.-1]), there will be very small, relatively dense, and robust agglomerates, so that this dispersion state is totally different from the initial one. This is assumption likely explanation, because the applied shear is able to induce a lasting change of the rheological properties, as shown in the measurements. These changes are irreversible in the investigated shear rate regime.
Shear-induced physical changes have been shown for lower filler contents at higher temperatures (23) but are obviously also present at lower temperatures for higher filler contents.
Measuring with either increasing or decreasing shear rates seems to have a strong influence on higher-tilled systems, and so further experiments were performed. A sequence of sweeps with different shear "directions" was performed for 0.5 wt% CNT/epoxy suspensions. In a first sequence A, an increase-decrease shear rate sweep was performed three times (six runs). In a second sequence B, only shear sweeps with increasing shear rates alternating with abrupt halts were performed six times. The results are shown in Fig. 3a and h.
At 0.5 wt%, which is well above the rheological percolation threshold, agglomerates of CNTs are present at every shear rate. The size and the interconnections of the agglomerates strongly influence the rheological properties. As shown by optical analysis, the size of the agglomerates is very sensitive to the applied shear rate (14), (16), (23).
The first runs of both sequences A and B show the already known curve with a maximum in viscosity at a shear rate of 0.2 [s.sup.-1]. In sequence A, the second run repeats the behavior described for loop-testing. The interesting difference occurs during the third run: starting at a lower viscosity, the suspension shows only shear thinning behavior indicated by decreasing viscosity values. The slope of the third forward sweep differs significantly from that of the first one. When performing the fourth run (decreasing shear rate sweep), the two curves superimpose very well and the same viscosity values as during the second run are reached. The viscosities of the increasing shear rate sweeps stay little below the viscosities reached under the decreasing ones. Sequence A was extended to six loops to prove if the decrease in initial viscosity depends on the number of measurements. After the third loop, no further noticeable decrease was measured and the curves were well reproducible within the measurement accuracy.
In contrast to sequence A, sequence B looks quite different. As one can see from Fig. 3b, already the second run shows a completely different shape. Any further runs show the same rheological behavior in sequence B: the initial viscosity lies around 20-25 Pa s. This is only one third of the initial value of the repeated runs in sequence A. While shearing, the viscosity increases and reaches a maximum at a shear rate around 0.3 [s.sup.-1]. Thereafter, the viscosity decreases to a value of 15-20 Pa s. Again, after some runs a repealable rheological behavior occurs, but the first run is not reproducible. For sequence A, we assume large agglomerates or coarse structures at the end of the second run, because a continuous decrease of the shear rate allows a steady agglomerate build-up. The agglomerate size increases constantly reaching a maximum at the lowest shear rate. When the third run is started, shear rate increases, and thus the agglomerate size is decreased again. Typical shear thinning behavior is observed as shown in Fig. 3a for the third and fifth run as well. Every applied shear rate induces a maximum size of agglomerates. Because the rheological properties of the CNT/epoxy depend on the size of the fillers, viscosity changes when agglomerate sizes change at different shear rates.
In sequence B, we assume that small agglomerates at the end of the first run are due to high shear rates. The second run starts with very low shear rates that induce agglomerate growth. When the shear rates rise above a critical value, the size of the agglomerates is decreased again.
The measurement itself thus affects the size of the agglomerates at the start of the measurement and is not negligible. Furthermore, it is possible to switch between these two different rheological systems. In Fig. 4, a combination of the two sequences is presented. Two up-down loops as in sequence A were performed followed by two up-up runs as in sequence B. As one can see, a combination of the rheological response shown in Fig. 3a and b occurs. This illustrates that it is very important to clarify the shear history of a sample and the way of measuring when publishing results of such systems.
The importance of these results becomes clear when looking at the literature. We could not find any literature reporting on a similar effect for CNT-filled systems. Additionally, detailed information like measurement times and measurement direction are often not available. The exact measurement procedure is unclear; this makes it difficult to compare research results quantitatively. Allaoui and El Bounia (3) recently reported on the rheological properties of CNT/epoxy suspensions giving detailed information about the measuring procedure. They mentioned that they used the forth sweep of two up-down cycles. This means that their results correspond to a backward sweep similar to those shown in Fig. 3a.
In a previous work, we showed the existence of an "initial shear effect" on a CNT/epoxy suspension with 0.1 wt% filler content at 60 degrees] C (23), Once the suspension is sheared with a high shear rate, the optical, electrical, and rheological properties changed drastically. This change is irreversible in the used shear rate regime. Obviously, this effect is also valid at higher filler concentrations and lower temperatures as seen in the irreversible first shear sweep measurement for 0.5 wt%. Because of the high concentration, it was not possible to make optical observation during the measurement with the used gap of 1 mm. To determine if there are structural changes induced by shearing, a small amount of the suspension was dropped on a glass holder. A second glass slide was pressed evenly until a very narrow gap was reached (d ~ 50 [micro]m), and light microscopy imaging was possible. As shown in Fig. 5a, the suspension shows a very good dispersion quality with only some visible small agglomerates. The glass plates were then manually sheared against each other for some seconds (v [approximately equal to] 1 mm/s). A significant more coarse structure is visible (Fig. 5b) after the shear treatment. We assume that the white areas seen on the image consist mostly of pure resin while the deep black areas are strongly entangled CNT agglomerates. These observations support our assumptions that shear is able to induce demixing even in higher-filled systems.
Influence of Production Method
The application of high shear rate treatment on the suspension is a common way to delete the history of the sample. We warned to prove if this is applicable for different dispersion techniques. Therefore, some material was additionally calandered in a three-roll-mill to achieve a finer dispersion. It has been shown that calandering leads to a very good dispersion quality (13), (25). When there are microagglomerates left in the suspension, one would expect different resistance to shear flow compared to suspensions in which the CNTs are separated (11). The suspensions were tested in a three sweep measurement, first with increasing shear rates (0.1-100 [s.sup.-1]) followed by decreasing shear rates (100-0.1 [s.sup.-1]) and finally again with increasing shear rates. The results are shown in Fig. 6.
One can directly observe a difference between the two mixing methods. For only stirred suspensions, the same rheological behavior as previously reported occurs. Calandered suspensions show a much lower maximum in viscosity for the very first shear sweep. The starting viscosity is around 28 Pa s. Nevertheless, an increase up to 50 Pa s at a shear rate of 0.4 [s.sup.-1] is found. The maximum in viscosity was shifted to higher shear rates compared to the stirred sample (maximum at 0.15 [s.sup.-1]). If two suspensions, one well dispersed and one worse dispersed, have the same time under steady shear to form agglomerates, it is obvious that build-up of big agglomerates is more complicated for the line dispersion. Thus, the agglomerates build up at low shear rates in milled suspensions are smaller compared to the agglomerates in the stirred suspension. In a bad dispersion like in the stirred suspension, the already existing agglomerates only have to stick together and thus grow faster. As mentioned before, at a certain agglomerate size, the shear rate becomes destructive. This critical shear rate is reached earlier for stirred suspensions explaining the left shift of the viscosity maximum. Additionally, the viscosities of the calandered suspensions at low shear rates lie clearly above those for stirred suspensions (second and third run). A liner and homogeneous dispersion enhances the viscosity more effectively.
The strong dependence of the rheological properties on the production method in steady shear measurements may be used for the characterization of the dispersion stale. It is known by optical and mechanical measurements that calandering provides a better and liner dispersion than stirring. If the rheological behavior of a CNT/epoxy system is known, the results of steady shear sweep tests can give information on the processing method or, to be more precise, the dispersion state achieved in this system for the method.
Additionally, the results show that the history of a suspension cannot be erased simply by taking a subsequent shear sweep or by moderate shear treatment (<100 [s.sup.-1]). Even then, differences between the different production methods, and thereby their dispersion states are measurable.
This study highlights the importance of shear treatment for viscosity-related properties. A key finding of this work is that a sample's shear history cannot be erased by pre-shearing with shear rates < 100 [s.sup.-1]. It is, however, possible to determine the dispersion quality of a suspension by rheological testing. One can clearly distinguish between different dispersion methods leading to different dispersion qualities by steady shear experiments, which are easy and fast to perform.
With these, it was specifically shown how the Theological properties of different contents of carbon nanotubes dispersed in epoxy were affected by steady shear measurements. The influence of increasing or decreasing shear rates on agglomeration and thus the viscosity was considerable. Furthermore, in the first measurement of a sequence, a nonrecurring and significantly different behavior with regard to subsequent measurement occurs. Literature supports (31), (32) the notion that the mechanism for this singular behavior after the first measurement is related to structural changes and depletion of exfoliated fillers from the matrix. As a consequence, coalescence of agglomerates leads to a pronounced increase of viscosity.
Optical observations indicate that the "initial shear effect" reported for systems with low filler content is also valid for system with higher filler content. This effect correlates with a lasting change in the microstructure directly after shear and after shear-induced agglomeration of the nanotubes.
(1.) W. Bauholcr and J.Z. Kovacs, Compos. Sci. Technol., 69. 1486 (2009).
(2.) W. Bauhofer, S.C. Schulz, A.E. Eken, T. Skipa. D. Lellinger, I. Alig, E.J. Tozzi, and D. Klingenberg, Polymer. 51, 5024 (2010).
(3.) A. Allaoui and N.E. Bounia, Chit. Nenrwci., 6, 158 (2010).
(4.) J.K.W. Sandler, J.E. Kirk, LA. Kinloch, M.S.P. Shaffer, and A.H. Windle, Polymer, 44, 5893 (2003).
(5.) E.T. Thostenson, Z. Ren, and T.W. Chou, Compos. Sci. Technol., 61, 1899 (2001).
(6.) M. Chapartegui, N. Markaide, S. Florcz, C. Elizetxea, M. Fernandez, and A. Santamar, Compos. Sci. Technol., 70. 879 (2010).
(7.) I.D. Rosea and S.V. Hoa, Carbon. 47, 1958 (2009).
(8.) G. Faiella, F. Piscitelli, M. Lavorgna, V. Antonucci, and M. Giaordano, App. Phys. Lett., 95, 153106 (2009).
(9.) Y.Y. Huang and E.M. Tcrcntjcv, Adv. Funct. Mater., 20, 4062 (2010).
(10.) Y.Y. Huang, S.V. Ahir, and E.M. Terentjev, Phys, Rev. B, 73, 125422 (2006).
(11.) Z.H. Fan and S.G. Advani J. Rheol., 51, 585 (2007).
(12.) F.H. Gojny, M.H.G. Wichmann, U. Kopke, B. Fiedler, and K. Schulte, Compos. Sci. Technol., 64, 2363 (2004).
(13.) M.H.G. Wichmann, J. Sumrlcth, B. Fiedler, F.H. Gojny, and K. Schulte, Mech. Compos. Mater., 42, 395 (2006).
(14.) A.W.K. Ma, M.R. Mackley, and F. Chinesta, Inter. J. Mater. Form., 1, 75 (2008).
(15.) A.W.K. Ma, M.R. Mackley, F. Chinesta, and A. Ammar, J. Rheol., 52, 1311 (2008).
(16.) S.S. Rahatekar, K.K.K. Koziol, S.A. Butler, J.A. Elliott, M.S.P. Shaffer, M.R. Mackley, and A.H. Windle, J. Rheol., 50, 599 (2006).
(17.) E.K. Hobbie, Rheol. Acta, 49, 323 (2010).
(18.) I. Alig, T. Skipa, D. Lellinger, and P. Potschke, Polymer, 49, 3524 (2008).
(19.) S.B. Kharchenko, J.F. Douglas, J. Obrzut, E.A. Grulke, and K.B. Migler, Nat. Mater., 3, 564 (2004).
(20.) L.H. Switzer and D.J. Klingenberg, Int. J. Multiphase Flow, 30, 67 (2004).
(21.) A.F. Schmid, L.H. Switzer, and D.J. Klingenberg, J. Rheol., 44, 781 (2000).
(22.) T. Skipa, D. Lellinger, W. Bohm, M. Saphiannikova, and I. Alig, Polymer 2010, 51, 201 (2010).
(23.) S.C. Schulz, J. Schlutter, and W. Bauhofer, Macromol. Mater. Eng., 295, 613 (2010).
(24.) J.Z. Kovacs, R.E. Mandjarov, T. Blisnjuk, K. Prehn, M. Sussiek, J. Midler, K. Schulte, and W. Bauhofer, Nanotechnology, 155703 (2009).
(25.) J.Z. Kovacs, B.S. Velagala, K. Schulte, and W. Bauhofer, Compos. Sci. Technol., 67, 922 (2007).
(26.) C.A. Martin, J.K.W. Sandler, M.S.P. Shaffer, M.K. Schwarz, W. Bauhofer, K. Schulte, and A.H. Windle, Compos. Sci. Technol., 64, 2309 (2004).
(27.) S.C. Schulz and W. Bauhofer, Polymer, 51, 5500 (2010).
(28.) F. Du, R.C. Scogna, W. Zhou, S. Brand, J.E. Fischer, and K.I. Winey, Macromolecules, 37, 9048 (2004).
(29.) P. Potschke, M. Abdel-Goad, I. Alig, S. Dudkin, and D. Lellinger, Polymer, 45, 8863 (2004).
(30.) T. Skipa, D. Lellinger, M. Saphiannikova, and I. Alig, Phys. Status Solidi B, 246, 2453 (2009).
(31.) J.C. Gimel, T. NicoJai, and D. Durand, Phys. Rev. E, 66, 061405 (2002).
(32.) K.G. Soga, J.R. Melrose, and R.C. Ball, J. Chem. Phys., 108, 6026(1998).
Correspondence to: Sonja Carolin Schulz; e-mail: email@example.com
Published online in Wiley Online Library (wileyonlinelibrary.com).
[c] 2011 Society of Plastics Engineers
Sonja Carolin Schuiz, (1) Jana Schlutter, (1) Samuel T. Buschhorn, (2) Karl Schulte, (1) Wolfgang Bauhofer (1)
(1) Institut fur Optische und Elektronische Materialien, Technische Universitat Hamburg-Harburg, EiBendorfer StraBe 38, D-21073 Hamburg, Germany
(2) Institut fur Kunststoffe und Verbundwerkstoffe, Technische Universitat Hamburg-Harburg, DenickestraBe 15, D-21073 Hamburg, Germany
|Printer friendly Cite/link Email Feedback|
|Author:||Schuiz, Sonja Carolin; Schlutter, Jana; Buschhorn, Samuel T.; Schulte, Karl; Bauhofer, Wolfgang|
|Publication:||Polymer Engineering and Science|
|Date:||Apr 1, 2012|
|Previous Article:||Shear-induced crystallization of isotactic polypropylene studied by simultaneous light intensity and rheological measurements.|
|Next Article:||Investigation on the motion of different types of fibers in the vortex spinning nozzle.|