The Promise of Dielectric Nanocomposites J. Keith Nelson Rensselaer Polytechnic Institute Troy, NY 280-3590 Abstract: Several research groups worldwide have now been able to document some significant improvements that can be made in the electrical, and other, properties of polymer composites through the incorporation of nanoparticulates. Although it is now becoming clear that the mechanisms responsible for these changes are by no means universal, some of the benefits are substantial and rely on the large interface areas which are inherent in the introduction of materials of nanometric dimensions. By examining a variety of nanomaterials, this contribution seeks to review the property changes that can be brought about and examines the possibilities for commercial applications. This involves not only the electrical properties, but the implications for the attendant mechanical characteristics and the polymer processing necessary for utilization of this emerging breed of dielectric material. In this context, the functionalization of the particulate surfaces to provide preferential coupling to the host polymer will be explored since, by this means, a degree of preferred assembly can be accommodated. Through experimental examples, the use of this technique to tailor the properties of nanodielectrics is illustrated. I. BACKGRUND Prior to a theoretical paper by Lewis [], interest in nanotechnology had been primarily centered on semiconductor, biological and sensor applications. owever, the experimental work of enk et al. [2] suggested that there may be advantages to be gained in the field of bulk electrical insulation. Early in this decade, some of these benefits were documented experimentally in joint US/European work [3], although a full understanding of the complex physics and chemistry was, as still is, lacking. In the last two years there has been a burgeoning interest in this technology by many research groups worldwide. The commercial impact is potentially very large since electrical insulation is a huge business segment and thus there has also been significant patent application activity. This contribution is limited to use of nanoparticles (defined loosely as material having one dimension less than about 00 nm) incorporated in a polymer matrix. owever, within this class of composites, there is a great deal of opportunity to tailor the properties of the resulting material to specific applications. Table provides an overview of some of the systems currently under investigation. From the viewpoint of insulating systems most of the activity has been on clays and inorganic oxides (particularly 2, Al 2 3, Zn and Ti 2 ). Although the interest here is primarily the electrical properties of this new class of material, it is likely that many of the applications will also take advantage of attendant changes in other attributes, particularly thermal conductivity, coefficient of the thermal expansion and thermal endurance [4]. TABLE I EXAMPLES F NANCMPSITE SYSTEMS UNDER INVESTIGATIN Base Polymers Polyolephins Epoxies/phenolics Elastomers Ethylene-vinyl copolymers Polyethylene terephthalate Polyamides Polyimides II. PRINCIPLES Nanomaterials Clays Inorganic oxides Carbon nanotubes Graphite It has become clear that the principal underlying reason for the changes in properties which are being seen is related to the plurality of interfaces introduced through the use of nanomaterials. Fig. indicates that for quite modest loadings, (Volume of interface/ total polymer volume).0 0.8 0.6 0.4 0.2 5 nm 00 nm 00 µm µm 0.0 0.0 0. Loading, (volume fraction) Fig.. Surface-to-volume ratios of nanocomposites as a function of nanoparticle loading the surface area associated with the internal interfaces is very large. In this way, properties associated with the interface may become dominant so that the new material can then display properties which are not necessarily provided by either of the phases from which it is derived. For this reason the mixing
rules for composite structures will no longer apply. Given that underlying philosophy, it is perhaps clear why the tailoring of material properties will often involve modifications of the internal interface. This may occur by physical means, such as tethered entanglement of the polymer chains, but can also be brought about chemically through the fuctionalization of the nanoparticle surface. In this way the bonding of the nanoparticles can be engineered to bring about a degree of self assembly and, in principle, provide an opportunity to affect the properties in desirable ways. III. FRMULATIN The preparation of nanocomposites has a marked effect on their eventual properties. It is clearly important to insure that there is proper dispersion of the nanophase material. Clays are an important class of filler, but the layered structure needs to be intercalated or exfoliated in order to make them effective. An example of this is the formation of layered silicates [5] which are rendered organophilic by means of ion exchange of inter-gallery sodium ions by protonated primary alkyl amines. Although exfoliation is not a complication with monodisperse inorganic oxides, processing is nevertheless pivotal if agglomeration is to be prevented for the high surface energies implicit in the system. The essential steps involved may generally be specified as: particle drying (after functionalization, if used) monitored by thermogravimetric analysis compounding with the base resin (and cross-linking agent if appropriate) casting or molding post curing protocol to remove unwanted byproducts which will prejudice the electrical properties. Clearly, the detail will depend on the nature of the material, but the compounding stage usually requires the application of a considerable amount of shear stress utilizing a twin-screw extruder, melt mixed and/ or ultrasonic wand. Characterization is imperative to ensure proper quality control. In addition to electron microscopy, differential scanning calorimetry is useful to determine the crystallinity (if applicable) and transition temperatures. Fig. 2 shows an example of a scanning electron micrograph taken of 23 nm 2 nanoparticles embedded in a cross-linked polyethylene (XLPE) resin system. Particle functionalization can also be carried out prior to compounding to provide preferred coupling sites. This involves the deposition of a thin layer of the desired agent (usually a form of silane treatment). A recent contribution by Roy et al. [6] provides an example of the use of vinylsilane for this purpose with good effect. Fig. 3 illustrates the use of IR spectroscopy to determine the way in which the components have assembled. Particle features such as free silanol groups at 3747 cm - and the broad peak centering around 3500 cm - are gone as the result of compounding, but the peak at 580-680cm - representing (-C=C-) double bond which was present in the vinylsilane treated particles (from the vinyl group) is replaced by the peak at 2860-2970 cm - representing the single bond of carbon (-C 2 -C 2 -.). Absorbance, a. u. 4 3 2 0 3747 -bonded Dried Vinylsilane particles Dried Vinylsilane in XLPE 630-4000 3500 3000 2500 2000 500 000 500 Wave Number, [cm - ] Fig. 3. FTIR spectrum of vinylsilane-treated 2 before and after incorporation into a XL-polyethylene polymer [6] IV. ELECTRICAL PRPERTIES 00 Fig. 2. Scanning electron micrograph of 23 nm 2 nanoparticles dispersed in a cross-linked polyethylene matrix A. Electric Strength Perhaps the most important property of nanocomposites is the change in electric strength which is found when the filler particles attain nanometric dimensions. This contrasts with the situation for conventional microparticles where substantial reductions in electric strength are typical as a result of the weak interfaces and defects which are involved. This is illustrated in Fig. 4 (a) where the DC quazi-uniform field electric strength of a biphenol epoxy resin system is depicted
with both micro- and nano-particulates of Ti 2 used to form the composites. The impact of the size of the filler, at the same nominal loading of 0% by weight, is clearly evident. and additional gains made through changing the interface conditions by functionalization. Probability of Failure Probability (%) 99 50 Breakdown Strength (MV/cm) (b (a 0. 0 00 000 Breakdown Strength (kv/mm) XLPE + TE- treated nanosilica XLPE + AEAPS-treated silica XLPE + MDS-treated nanosilica XLPE + untreated nanosilica XLPE + untreated microsilica XLPE only Fig. 4. Breakdown probability plots for nanocomposites using recessed specimens. (a) Epoxy-Ti 2, (b) XLPE- 2 Fig. 4(b) indicates more dramatic improvements in a polyolefin nanocomposite. In this case the base resin, depicted by the curve on the left, is seen to be substantially improved by the incorporation of nanoparticulates. The various curves on the right indicate that additional benefits can be obtained in electric strength by the use of appropriate functionalization which is used to affect conditions at the interfaces. In this instance the most effective agent is the vinylsilane which is known to couple well both to the 2 surface and to the XLPE chain (see Fig.2). B. Voltage Endurance Analogous results for the voltage endurance under nonuniform field conditions are depicted in Fig. 5 where the improvements are dramatic. These endurance curves are plotted in terms of the maximum field at the tip of the point/plane gap and show over 2 orders of magnitude improvement in the voltage endurance in both the Epoxy (a) and XLPE (b) systems. Again, it would appear from Fig. 5(b) that a substantial benefit is obtained from the nanoparticles 0 Tip Electric Field (kv/mm) Electrode-tip Stress, kv/mm 700 600 500 400 300 200 00 0 00 000 0000 500 400 300 200 00 000 900 800 700 600 500 400 XLPE Micro (.5µm) Untreated Nanosilica Life (hr) XLPE XLPE + untreated nanosilica XLPE + MDS-treated nanosilica XLPE + aminosilane-treated nanosilica XLPE + vinylsilane-treated nanosilica 00 000 0000 Time, min Treated Nanosilica C. Permittivity and Loss Much can be learned about the way these material operate through studies of thermally stimulated currents and dielectric Relative permittivity (a (b Nano (23 nm) Fig. 5. Voltage endurance characteristics for nanocomposites using 4 µ m tip/plane electrodes. (a) Epoxy-Ti2, (b) XLPE- 2 E+7 E+6 E+5 E+4 E+3 E+2 0% w/w Ti 2 38 nm particles in epoxy, 393K 00 0 0. E+ nano real nano imag E+0 micron real micron imag E- E-3 E-2 E- E+0 E+ E+2 E+3 E+4 E+5 E+6 Frequency/z tan delta vs. frequency 0.0 E-3 E- E+ E+3 E+5 Fig. 6. Real and Imaginary components of permittivity for nano- and micro- Epoxy/Ti 2 composites [3].
spectroscopy. Although the practical use of these materials usually involves frequencies of 50 z and above, it is the very low frequency domain where mechanistic evidence is to be found. For example the change of slope of the real part of the permittivity depicted in Fig. 6 for a thermoset nanocomposite is the result of the mitigation of the Maxwell-Wagner effect which is well known for conventional (micron) fillers. f more practical significance is the finding that at power frequencies and higher, the incorporation of a high permittivity nanoparticulate material results in a composite which may exhibit a bulk dielectric constant which is lower than either the base polymer or the nanoparticle introduced. This surprising result has been documented in several dielectric systems and contradicts the usual mixing laws. It is thought to result from the tethering of species giving rise to dispersion at the interfaces. owever, from the practical point of view it is important since it may be used as a technique for lowering the reactive current associated with the material which is an important factor in cable dielectrics, for example. D. Internal Charge Characteristics A common feature of many nanodielectric systems is the behavior of the internal space charge build-up. The recent development of systems to measure the internal charge in dielectrics has allowed both the magnitude and dynamics of the internal charge to be assessed and it appears that there are substantial differences when nanoparticles are incorporated into a polymer matrix. As an example, Fig. 7 depicts the differences seen with a polyethylene/ 2 nanocomposite in comparison with the base resin (XLPE). For many nanodielectric systems three features are usually dominant: (a) The magnitude of the internal charge is much less for nanocomposites (b) The dynamics of charge decay are much faster for nanocomposites (c) There is often a very different distribution of charge (and therefore of internal field) with nanocomposites. Again there are very important practical consequences of this finding. Applications, such as DC cables, that have design limitations dependent on internal charge can perhaps be reengineered using nanomaterials in the light of this finding. It 80.0 60.0 40.0 20.0 0.0-20.0-40.0-60.0-80.0-00.0-20.0 Calibrated Electric Field Calibrated gnal (C/m-3) Cathode -40.0 0.0 50.0 00.0 50.0 200.0 250.0 300.0 350.0 400.0 450.0 500.0 550.0 Calibrated potential ( kv) (MV/m) Time in Sec Electric Field Charge distribution potentia Time in hours Anode thickness of sample(um) 600.0 (a) 0.0 00.0 90.0 80.0 70.0 60.0 50.0 40.0 30.0 20.0 0.0 0.0-0.0-20.0-30.0-40.0-50.0 Calibrated Electric Field Calibrated gnal (C/m-3) Cathode -60.0 0.0 50.0 00.0 50.0 200.0 250.0 300.0 350.0 400.0 Calibrated potential ( kv) (MV/m) Time in Sec potentia Electric Field Charge distribution thickness of sample(um) Fig. 7. Typical PEA space charge measurements; (a) XLPE only, (b) nano-filled material [6]. Time in hours Anode 450.0 (b) is also likely that at least part of the enhanced electric strength may be due to the redistribution of internal charge. V. TE UNDERLYING PYSICS AND CEMISTRY In order to properly engineer future insulation systems using nanocomposites, it is clearly necessary to understand the way in which they function. Work conducted in a number of different laboratories worldwide has indicated that although there are common features, not all nanodielectric systems behave in the same way. If it is accepted that it is the internal interface that is dominating the behavior, then that is perhaps not too surprising. Although beyond the scope of this review, evidence for the part played by the internal interface has been provided by electroluminescence, photoluminescence, thermally stimulated currents, x-ray secondary emission spectroscopy, electron paramagnetic resonance, as well as the areas cited here. The picture which emerges is one in which the internal interfaces may create or influence: () scattering of electrons in high field regions (2) an interaction zone surrounding the nanoparticles (3) free volume within the polymer structure Although, there is evidence that all these factors can play a role, it is clear that the extent to which they will influence the properties does depend on the system concerned, and, more particularly, on the nature of the interface zone. Fig. 8. provides an example which suggests that, in high field regimes where electronic conduction is prevalent, the intervention of nanoparticle interfaces creates scattering which moderates the prevailing electron energies. This may be seen from the electroluminescence spectra where a red shift corresponding to an energy of about.3 ev is evident when 0% Ti 2 nanoparticles are added to an epoxy resin Electroluminescence (a.u.) 650 600 550 500 450 400 350 Frequency Shift 300 400 450 500 550 600 650 700 Wavelength (nm) (b) (a) (c) Fig. 8. Frequency resolved electroluminescence spectra for 0 % Ti 2 /epoxy composites [micro (b) and nano (c)] in comparison with the base resin (a).
nce the technique employed involved the use of calibrated interference filters and a sensitive photomultiplier tube, there is little fine structure available and no information from the absolute value of the output. Nevertheless, the shift in the major peak is quite reproducible. The divergent field used (~500 kvmm - ) is similar to that used in the dramatic results of Fig.5 strongly suggesting that scattering plays an important role in the improvement of voltage endurance. Another general feature of this class of materials stems from the modification of the internal space charge such as is shown in Fig. 7. This type of phenomenon has been observed in several nanodielectric systems [3, 5-6] and there is increasing evidence that this is also due to interface behavior. Lewis [7] has recently suggested that there may be a Guoy-Chapman layer associated with these interfaces similar to that which is well known in the electrochemistry of liquids. The dispersion of such small particles, even at loadings of a few percent, can cause these interaction zones to coalesce creating a pathway for local conduction without negatively affecting the bulk conductivity. Such a mechanism would provide an explanation for the changes in internal charge behavior and also would be expected to reduce the charge time constant seen in pulsed electroacoustic measurements and in light emission behavior [8]. Theoretical models are beginning to emerge based on such interaction zones [9] and are likely to form useful tools in the engineering of new materials based on this technology. The last major feature which is relevant here is the influence of free volume and defects. The free volume theory of electrical breakdown [0] is compelling on the basis of the close association between breakdown strength and free volume which is seen as a function of temperature. Furthermore, it is known that the free volume of a polymer is affected by the incorporation of nanoparticles [8]. owever, the changes in free volume are not always what might be expected and are likely to be dependent on the functionalization. Indeed, it may be that the advantages seen in Fig. 4(b) result primarily from the modification of the free Probability of Failure 0 00 000 0. Region-2 Region- 25 o C 60 o C 70 o C 80 o C 0 00 000 Breakdown Strength, kv/mm Fig. 9. Detailed breakdown statistics for an untreated 2 - XLpolyethylene nanocomposite with a 5% (by weight) oxide loading 0. volume through particle chemical coupling such as that depicted in Fig. 3. In addition to the inherent free volume, which can be thought of as a distributed defect structure, there will also be effects associated with the interfaces introduced. In the case of a conventional microcomposite, such interfacial defects will dominate and give rise to poor electric strength performance, as depicted, for example, in Fig. 4 (a). In this way, one might expect defects which are native to the host polymer (free volume) and those which are the result of the interface regions formed by the particles introduced. Careful investigation of the failure statistics of these materials indicates that there are two regions which can be identified as having two different (temperature dependent) breakdown distributions. Such characteristics are depicted for polyolefin nanocomposites in Fig. 9. Assuming a chain scission mechanism, the two segments of the line represent two ranges of defect distributions, each having its own width and maximum value that change independently over the temperature range studied (increase in maximum value but decrease in width of distribution with increasing temperature). In this way, one may speculate that, with increase in temperature, both distributions get narrower and the upper limits can finally merge giving rise to the characteristics seen in Fig. 9. VI. APPRAISAL While filled polymers have been common since the start of the plastics age, nanocomposites are intrinsically different in that they appear to be dominated by the characteristics of the internal interfaces. These properties appear to arise when the infilled material has a similar length scale to that of the polymer chain. ne possible reason for this lies in the relationship to the radius of gyration. Fig.0 illustrates this, again for a 2- polyethylene system. For the microparticles, the bond angles will allow hydrogen bonding of the groups which is not facilitated in the case of nanoparticles. Early crude attempts to engineer nanodielectrics by changing the conditions at the interface do suggest that indeed some degree of tailoring of dielectric properties and self assembly may be possible. From the commercial standpoint, the voltage endurance enhancements documented already should provide the encouragement to examine the properties of scaled-up systems using this technology. Dielectrics have long been regarded as mature science, but there are here real opportunities to provide substantial performance enhancements/cost savings. Furthermore, many insulation applications are not limited by electrical properties, but by mechanical and thermal considerations. For example, the nanocomposite giving rise to the breakdown characteristics shown in Fig. 4 (b) also has a concomitant increase in the melting temperature of about 6%. Indeed, an obvious application would be cryogenic dielectrics because of the criticality of some of the attendant properties. milarly, applications, such as DC cables, for which the mitigation of internal space charge is important, are clearly prime candidates for this technology.
Externally bonded water nanocomposite where electron trapping is indicated as the dominant mechanism. ACKNWLEDGMENT (a) d + d - Bulk-size particle ydrogen bonded - groups Finally, very recent results [], again in epoxy resin, have reiterated the importance of internal charge in nanodielectrics. Fig. shows clearly that, in the case of a microcomposite, heterocharge can build up in front of the cathode. This is in striking contrast to the case for a nanodielectric where, in Nanometric Particle (b) Fig. 0 Schematic illustration of the bond angles associated with (a) micro- and (b) nanoparticles Charge (C/m 3 ) Charge (C/m 3 ) 5 0 5 0-5 -0-5 -2.E-04 0.E+00 2.E-04 4.E-04 6.E-04 8.E-04 25 5 5-5 -5-25 Voltage Applied Thickness Voltage (m) Applied Voltage Applied -.E-04.E-04 3.E-04 5.E-04 Thickness (m) The author is grateful to the US Electric Power Research Institute, the UK Science and Engineering Research Council, sponsoring Companies of the Philip Sporn Chair at Rensselaer, and to a large number of collaborators and graduate students. REFERENCES. Lewis, T. J., Nanometric Dielectrics, IEEE Trans on Diel. and Elect. Ins., Vol., 994, pp 82-25 2. enk P.., Kortsen T.W. and Kvarts T., Increasing the electrical discharge endurance of acid anhydride cured DGEBA epoxy resin by dispersion of nanoparticle silica, igh Perf. Polym., Vol., 999, pp 28-296 3. Nelson J.K. and Fothergill J.C., Internal charge behaviour in nanocomposites, Nanotechnology, Vol. 5, 2004, pp 586-9 4. Motori A.,et al., Improving thermal endurance properties of polypropylene by nanosaturation, Ann. Rep. Conf. on Elect. Ins. & Diel. Phen., IEEE, 2005, pp 95-8 5. Montanari G.C., Modification of electrical properties and performance of EVA and PP insulation through nanostructure by organophilic silicates, IEEE Trans on Diel. and Elect. Ins., Vol., 2004, pp 754-6 6. Roy M., et al.. Polymer nanocomposite dielectrics the role of the interface, IEEE Trans EI. Vol. 2, 2005, pp 629-4 7. Lewis T.J., Interfaces are the dominant feature of dielectrics at the nanometric level, IEEE Trans EI. Vol., 2004, pp 739-53. 8. Nelson J.K. and u Y., Nanocomposite dielectrics - properties and implications, J. Phys. D (Appl. Phys.),Vol 38, 2005, pp 23-222 9. Tanaka T., Dielectric nanocomposites with insulating properties, IEEE Trans. DEI. Vol. 2, 2005, pp 94-28 0. Artbauer J., Electric strength of polymers, J. Phys. D., Vol. 29, 996, pp 446-56. Nelson J.K. and u Y. Candidate mechanisms responsible for property changes in dielectric nanocomposites, IEEE Int. Conf. Prop. & Appl. f Diel. Mater., Bali, Indonesia, 2006 (to appear) Fig.. Charge distribution profiles for 0 % Ti 2 /epoxy composites at an average stress of 20 kvmm - both after hr of stressing (indicated) and with the voltage removed. [Top: nano and bottom: micro; cathode on left] equivalent circumstances, a shielding homocharge is seen in front of a cathode in high electric fields situations. Not only would this have implications for both dielectric strength and voltage endurance, but also again points to the scattering action of nanofillers. The free paths in the micromaterial appear to allow impact ionization in contrast to the