Degradation patterns of silicone-based dielectric elastomers in electrical fields

ABSTRACT Silicone elastomers have been heavily investigated as candidates for the flexible insulator material in dielectric elastomer transducers and are as such almost ideal candidates because of their inherent softness and compliance. However, silicone elastomers suffer from low dielectric permittivity. This shortcoming has been attempted optimized through different approaches during recent years. Material optimization with the sole purpose of increasing the dielectric permittivity may lead to the introduction of problematic phenomena such as premature electrical breakdown due to high leakage currents of the thin elastomer film. Within this work, electrical breakdown phenomena of various types of permittivity-enhanced silicone elastomers are investigated. Results showed that different types of polymer backbone chemistries lead to differences in electrical breakdown patterns, which were revealed through SEM imaging. This may pave the way towards a better understanding of electrical breakdown mechanisms of dielectric elastomers and potentially lead to materials with increased electrical breakdown strengths.


Introduction
Silicone elastomers are gaining more and more attention for dielectric elastomer applications, mainly due to their inherent reliability which arises from the covalently crosslinked silicone polymer chains. This stands in contrast to the commonly used acrylic adhesive by 3M, namely VHB. VHB has proven to be an excellent material to showcase the possibilities of dielectric elastomers, for example through extreme extensibilities [1,2], but VHB does not possess the required reliability to become a viable, commercial dielectric elastomer transducer material [3,4]. Usually, VHB loses its performance after 100-1000 actuation cycles at large strains, and this lifetime is far from the 10-100 million cycles that is desired for most transducers [5]. Silicone elastomers, conversely, suffer from low dielectric permittivity (ε r =~3) compared to acrylics (ε r =~5) and polyurethane-based (ε r = > 5) elastomers and will thus usually result in a lower actuation response at a given voltage when used as a dielectric elastomer actuator [6]. For some applications, the achievable strain from unmodified silicone elastomers is sufficient such as lenses [7] and valves [8] but for example for wave energy harvesting [9] larger strains will also lead to larger energy generation and thus increased strains are desired. Numerous CONTACT Anne L. Skov al@kt.dtu.dk Polymer Centre, Department of Chemical and Biochemical Engineering, Technical University of Denmark, Søltofts Plads Building 227 Room 122, Kgs. Lyngby 2800, Denmark approaches to increase dielectric permittivity of silicone elastomers exist, which include mixing in metal oxides [10][11][12] or high permittivity liquids [13], covalent grafting of dipoles [14][15][16] and the use of functional copolymers [17,18]. All methods introduce complex changes in mechanical and electrical properties such that many elastomers with increased permittivity either suffer from increased stiffness, increased leakage currents or decreased electrical breakdown strengths. For more details on the effects of specific methods see the recent review on silicone dielectric elastomer transducers by Madsen et al [5]. Most methods for dielectric permittivity enhancement lead to a reduction in the electrical breakdown strength of the resulting elastomer. Therefore these elastomers will need to be operated at a lower voltage for reliability. This is highly unfavorable, but currently no universal method to increase the electrical breakdown strength of dielectric elastomers exists since the electrical breakdown processes are not fully understood. This circumstance is complicated further by the interplay between electrical and mechanical processes, which may lead to electro-mechanical breakdown processes, usually denoted electro-mechanical instability (EMI) [19][20][21][22]. Thus, no guideline on how to formulate soft dielectric elastomers with high electrical breakdown strength exists apart from the generic guidelines on what to avoid such as avoiding percolation of added fillers, which potentially leads to the introduction of conductive paths [23,24]. One approach to increase the electrical breakdown strength is, however, to increase the Young's modulus of the elastomer. This can be done, for example, by adding highly reinforcing, nonconductive metal oxide particles. Nevertheless this method usually causes a significant decrease in the actuation performance at a given voltage due to the increased stiffness of the resulting elastomer.
So-called voltage stabilization of high voltage cable insulation materials has been heavily investigated with the main focus being on thermoplastic or lightly crosslinked polyethylene (PE). The principle relies on the addition to the PE matrix of low-molecular weight compounds such as aromatic substances which then trap electrical charges [25,26]. This requires that the aromatic substances absorb strongly at the energy levels of the labile electrons in the PE material. Thereby avalanche breakdown is avoided or inhibited to some extent. This method is versatile for PE since the required quantities of voltage stabilizers are minute (usually significantly less than 1%) and they can be blended into the polymer matrix with littleif anyphase separation. The attained increases in electrical breakdown strengths are significantby the inclusion of diversified aromatic substances optimized for a given PE insulator matrix the electrical breakdown strengths have been improved by almost 50% [25,27,28]. This concept should, in theory, be relatively easily transferred to silicone elastomers. However, suchusually aromaticsubstances are not miscible in the silicone matrix even in minute concentrations, and thus grafting of the additives at a molecular level is required in order to obtain reliable and reproducible elastomer films. Most silicone elastomer formulations with non-grafted lowmolecular weight aromatic substances (i.e. additives) cause an easy detectable macroscopic phase separation with clear aromatic regions leading to local loss of insulating properties. Therefore, voltage stabilization of silicone elastomers becomes a complex task because it is crucial that the elastomer remains insulating and without conductive zones that could lead to build up of space charge [29]. This means, that synthetic approaches to incorporation of additives into the silicone polymer backbone are required and that the final polymer structure must be tailored in such a way that the microscopic phase separation favors voltage stabilization [18] rather than electrical conduction. This renders the design of the silicone elastomer system rather complex.
Recent studies have shown that the use of for example simple dipolar moieties in small amounts can increase the electrical breakdown strength. Such studies include synthetic incorporation of aromatic compounds [18] and chloropropyl groups [30,31] in the silicone polymer backbone. In this study we investigate the electrical breakdown patterns of two chloropropyl-functionalized silicone elastomers as prepared in Madsen et al [30] which break down electrically in rather different ways and we compare them to a silicone-based reference. The chloropropyl-functionalized elastomers are interesting candidates for dielectric elastomer transducers since their overall properties provide great figures of merit [5]. This is mainly due to combination of low Young's modulus, and relatively high dielectric permittivity and electrical breakdown strength, but also due to very low dielectric losses. Thermogravimetric analysis (TGA) and scanning electron microscopy (SEM) are used to evaluate the elastomers before and after electrical breakdown. Furthermore, the TGA potentially provides a basis for evaluation of the energies of chemical degradation (i.e. bond cleavage) involved in the electrical breakdown process.

Materials
Vinyl-terminated PDMS, DMS-V31 ( M w ≈ 28,000 g mol −1 ), and a hydride-functional crosslinker, HMS-301, were acquired from Gelest Inc. The platinum cyclovinylmethyl siloxane complex catalyst (511) was purchased from Hanse Chemie, while silicon dioxide amorphous hexamethyldisilazane-treated particles (SIS6962.0) were purchased from Fluorochem. The inhibitor Pt88 was acquired from Wacker Chemie AG, and all other chemicals were acquired from Sigma-Aldrich and used as received, unless otherwise stated.
Two types of copolymers (Co-1 and Co-2) with two different spacer lengths between the chloropropyl-groups were synthesized according to previous work [30].

Film preparation
Co-1, Co-2, or DMS-V31, respectively, and an 8-functional cross-linker HMS-301 were mixed with treated silica particles (25 wt%) and inhibitor (1 wt%, Pt88) and then mixed on a FlackTek Inc. DAC 150.1 FVZ-K SpeedMixer™. The catalyst (511) (1.5 ppm) was added thereafter and the mixture was speed-mixed once more. The uniform mixtures thus made were coated on a glass substrate using a film applicator (3540 bird, Elcometer, Germany) with 150 μm blade. Films were fully cured in the oven for 1 hour at 115°C.

Electrical breakdown strength determination
Electrical breakdown tests were performed on an in-house-built device based on international standards (IEC 60,243-1 (1998) and IEC 60,243-2 (2001)), while film thicknesses were measured through the microscopy of cross-sectional cuts, and the distance between the spherical electrodes was set accordingly with a micrometer stage and gauge. An indent of less than 5% of sample thickness was added, to ensure that the spheres were in contact with the sample. The polymer film was slid between the two spherical electrodes (diameter of 20 mm), and the breakdown was measured at the point of contact by applying a stepwise increasing voltage (50-100 V step −1 ) at a rate of 0.5-1 steps s −1 . Each sample was subjected to 12 breakdown measurements, and an average of these values was given as the breakdown strength of the sample. The electrical breakdown experimental setup is shown in Figure 1.

Young's modulus and tensile strength measurements
Uniaxial extensional rheology was performed on the series of elastomer films in order to determine the Young's modulus and tensile strength. The stress-strain curves of films were tested at room temperature by ARES-G2 rheometer using the SER2 geometry. A sample of 20 mm length and 6 mm width was placed between two drums and initially separated by a distance of 12.7 mm. The test specimen was elongated uniaxially at steady Hencky strain rate of 0.01 s −1 until sample failure at the middle part. Each composition was subjected to four tensile measurements which were then averaged. Young's moduli were obtained from the tangent of the stress-strain curves at 5% strain.

Dielectric properties test
Dielectric relaxation spectroscopy (DRS) was performed on a Novocontrol Alpha-A highperformance frequency analyser (Novocontrol Technologies GmbH & Co) operating in the frequency range 10 -1 -10 6 Hz at room temperature and low electrical field (~1 V/ mm). The diameter of the tested 0.5-1 mm thick samples was 25 mm.
2.6. Thermogravimetric analysis (TGA) and differential scanning calorimetry (DSC) measurements Thermogravimetric analysis (TGA) was performed on a Discovery TGA from TA Instruments in a nitrogen atmosphere with a heating rate of 10°C min −1 from RT to 900°C. Differential scanning calorimetry (DSC) measurements were performed on a Discovery DSC from TA Instruments in a nitrogen atmosphere with a heating and cooling rate of 10°C min −1 from −150 to 200°C.

Scanning electron microscopy (SEM) and microanalysis
The morphology of the elastomer film breakdown was investigated with an FEI Quanta 200E-SEM environmental scanning electron microscope, equipped with a field emission gun. The surface was visualized in low vacuum, using water vapors as auxiliary gas at a pressure of 150 Pa. A mixture of secondary and back scattered electrons, generated by the sample surface, was detected with the large field detector for an incident electron beam of spot 3 accelerated to 10 keV.
The elemental composition of the elastomer films was determined by energy dispersive X-rays (EDX) with an Oxford Instruments 80 mm 2 X-Max silicon drift detector Mn Kα resolution at 124 eV, also at low vacuum (150 Pa) with a 500 µm pressure limiting aperture X-ray cone. The microanalysis data acquisition and quantification was performed with the Oxford Instruments Aztec program version 3.1.

Chemical and mechanical analysis of the prepared elastomers
Crosslinked chloropropyl-functional silicone elastomers, denoted Co-1 and Co-2, and the reference DMS-V31 were successfully prepared using 8-functional crosslinker, HMS-301, and reinforcing silica particles, giving elastomers with almost identical molecular weight spacing between each crosslink. Details of the prepared elastomers are shown in Table 1. The structure and preparation of the chloropropyl-functional silicone elastomers are illustrated in Scheme 1. The difference between Co-1 and Co-2 is the different spacer lengths between the chloropropyl-groups. Co-1 have 1200 g mol −1 dimethylsiloxane spacers between each chloropropyl group, and Co-2 have 580 g mol −1 dimethylsiloxane spacers between the functional chloropropyl groups. Co-2, with the shorter spacer, thus contains approximately double the amount of chloropropyl groups than Co-1 prepared with the long spacer (at comparable copolymer lengths) [25]. A summary of the elastomer properties is given in Table 2.
The effect of the chloropropyl-functionalities on the mechanical properties of the elastomers was determined through tensile testing and the results are shown in Table 2. It is evident that the samples possess differences in their elastic properties. The Young's modulus of the reference elastomer is ten times higher than elastomer Co-1 and three times higher than elastomer Co-2. If interpreted entirely from the Young's moduli the electrical breakdown strengths should then follow the trend  Table 2 this is apparently not correct and will be discussed in further detail later. With respect to tensile strengths, the reference sample outperforms the two chloropropyl-functional elastomers almost proportionally to the Young's moduli since the elastomers behave more or less as ideal Hookean solids in the entire strain regime. All samples have similar maximum extensions (314-429%). For silicone elastomers the temperature dependency on mechanical properties is weak [32] so no full temperature range characterization has been performed.

Electro-mechanical evaluation
The electromechanical properties of the Co-1 and Co-2 elastomers are evaluated through a comparison with the properties of the commercially available silicone elastomer film from Wacker Chemie AG whose properties resemble that of Elastosil RT625 (likewise from Wacker Chemie AG). The evaluation and comparison is done by normalizing with the electro-mechanical properties of the commercial elastomer film in a comparable thickness. The synthesized reference DMS-V31 is not included in the evaluation since it is solely used as a reference with respect to the thermal and electrical properties.  The figure of merit for actuation is given by [33]: where E breakdown is the electrical field at which electrical breakdown occurs, ε r is the relative dielectric permittivity, ε o is the permittivity of free space (8.85 × 10 -12 F m −1 ) and Y is the Young's modulus of the elastomer, The figure of merit for energy generation is given by [34]: where Φ is the energy density function of the investigated elastomer. Due to lack of data for most silicone elastomers it is usually regarded as a constant and is thus ignored in relative comparisons [5]. The properties of Elastosil RT625 for the comparison are regarded as ε r @0.1Hz = 3, Y@5% strain = 1 MPa and E breakdown = 80 V μm −1 [5]. The normalized F om (DEA) of Co-1 and Co-2 elastomers with respect to the commercial reference become 8.9 and 4.5, respectively. In other words, actuation at ultimate conditions of the developed elastomers is 9 and 4.5 times better than the commercial reference elastomer. The normalized F om (DEG) of Co-1 and Co-2 elastomers are 1.3 and 2.3, respectively, i.e. the developed elastomers are interesting to a greater extent from an actuation point of view than for energy generating since only minor improvements in F om (DEG) are achieved. For the figures of merit, the normalized numbers state the improved performance at ultimate conditions, i.e. just before the electrical breakdown of the given insulating film compared to the reference film at its ultimate conditions. Here it is assumed that the dielectric elastomer will always fail electrically before a mechanical rupture takes place, which from Table 2 can be seen as a reasonable assumption since all films survive more than 300% strain in a pure mechanical deformation.
It is evident from the calculated figures of merits that the elastomers with chloropropylfunctional groups show improved dielectric elastomer performance for both actuation and energy generation applications. However, the main improvement was found for actuation due to the low Young's modulus of the elastomers. Usually a low Young's modulus causes low electrical breakdown strength [5,35] but due to the voltage stabilizing effect of the chloropropyl groups the developed elastomers do not follow this trend and thus the materials are from a fundamental point of view interesting to investigate further.

Thermal properties and electrical breakdown
Electrical breakdown may occur through thermal breakdown mechanisms, i.e. heating takes place locally and the heating results in increased conductivity which then again leads to further Joule heating. Thus an accelerating coupling arises since conductivity scales exponentially with temperature. The heating may lead to instantaneous breakdown due to an increased field but may also a priori lead to degradation of the elastomer and thus loss of mass, which then subsequently leads to electro-mechanical breakdown due to decreased thickness. When the elastomer starts degrading it may even experience reduced mechanical properties locally in such a way that the elastomer collapses mechanically. With these scenarios in mind the thermal breakdown patterns are investigated in order to elucidate at which temperatures the elastomers degrade in a nitrogen atmosphere (which is assumed a comparable atmosphere to that of a dielectric elastomer sealed within metallic electrodes).
The thermogravimetric data of the chloropropyl-functional elastomers and the reference DMS-V31 are shown in Figure 2. The thermal degradation temperatures for the samples prepared with chloropropyl-functional copolymers decreases slightly compared to the reference DMS-V31 but remained within the same range. Evidently the synthesized chloropropyl-functional elastomers are very stable with respect to temperature and are not fully degraded before a temperature of around 660 o C is reached, as seen in Table 2. A reduction of T d3% (the temperature at which 3% by mass is degraded) is observed when the amount of chloropropyl groups is increased indicating that the chloropropyl groupas expecteddoes not stabilize the silicone elastomer thermally compared to the reference elastomer based solely on PDMS. However, it is unexpected that the Co-1 elastomer is the least thermally stable composition and the Co-2 elastomer resembles the reference elastomer the most with respect to thermal degradation behaviour. A gradual change from the reference over Co-1 to Co-2 was expected due to the increased chloropropyl content. The results are reproducible and the different behaviour between the two chloropropyl-functional elastomers is most likely a result of the spacing between the chloropropyl groups of Co-1 which allows for an easier ring formation of the degradation products, namely cyclic siloxanes [36]. This difference will be discussed later in connection with the electrical breakdown characterization.

Breakdown analysis
Typical breakdown zones in silicone-based elastomers consist of a pinhole with a surrounding region of solidification, both within and on top of the elastomer. This is illustrated in Figure 3. Three different breakdown zones (pinholes and surrounding area) of identical elastomers (reference DMS-V31 sample) are shown. The key observation is that there is a large extent of solidified matter around the pinhole as well as solidified 'threads' spreading from the pinhole outwards. These threads are unfortunate since they lead to stress inhomogeneities, which may potentially lead to tearing of the elastomer film if the breakdown experiment (or actuation) is performed in a non-zero stress configuration. The threads are not solely surface phenomena and span into the bulk of the material. Usually congruent patterns on both back and front side of the film are observed if the elastomer is inserted in the breakdown equipment in an unstretched configuration. If small stretches occur, the top and bottom sides of the film usually look different due to stresses not necessarily in the film area plane. Stretching during mounting is of course attempted avoided in every single experiment but small stresses during the mounting are inevitable. The observed 'threads' may be similar to that of breakdown in solid dielectrics where treeing is a common breakdown mechanism. However, at the current state it is not conclusive whether the threads are a result of the combination of excessive heating and the softness of the material leading to a mechanical phenomenon or if the threads are of electrical nature, i.e. due to charge delocation (or a combination thereof).
For the prepared elastomers varying degree of damage to the material upon electrical breakdown was observed. Two different types of breakdown patterns can be seen for the two chloropropyl-functional elastomers even though their chemical structure is rather similar with the only difference being the concentration of the chloropropyl groups. Examples of such patterns for all three elastomers, Co-1, Co-2 and the reference DMS-V31, are shown in Figure 4. The breakdown zones of the three different elastomers do not look identical but the appearance is rather consistent within one elastomer type. This gives a clear indication that minor changes to the backbone of the silicone elastomer may result in significant changes to electrical breakdown behaviours. The tearing process originating from the breakdown zones is a major challenge in siliconebased dielectric elastomer transducer products since the inherent tear strength of silicone elastomers is usually rather low [37].
The presence of relatively large amounts of chloropropyl groups stabilizes the silicone elastomer electro-mechanically since the Co-2 elastomer has significantly larger electrical breakdown strength than the Co-1 elastomer (Table 2). This effect can partly be attributed to increased stiffness of the elastomer since Co-2 has a significantly larger Young's modulus than the Co-1 elastomer. However, with respect to the reference elastomer, increased stiffness cannot be the explanation for increased breakdown strength of Co-2 since it is softer than the reference. Still the increase in breakdown strength is significant. The Co-2 elastomer possess an average electrical breakdown strength of 20 V μm −1 higher than the Co-1 elastomer and break down in what seems to be a more commonly observed pattern, namely through solidification of the elastomer around the pinhole and subsequent partial tearing due to the developed stress gradients around the solidified threads. This gives a clear indication that a certain concentration of chloropropyl groups has a positive effect on the electrical properties, either because the given concentration leads to a better morphology and/or better electrical stabilization. Decoupling the two effects is at current not possible but will be pursued in the future.
As is clearly seen in the SEM images, Co-1 and Co-2 decompose differently during the imposed electrical breakdown. The breakdown of Co-1 is seen to involve substantial boiling and subsequent condensation of volatiles evident in more detail in Figure 5. In order to get a better idea of the involved breakdown phenomena, SEM/microanalysis was performed as well. The composition of the solidified zones in the Co-2 elastomer could not be detected due to lack of difference in elemental composition. But very interestingly, as is shown in Figure 6, it is apparent that during the breakdown of Co-1 elastomer silicon-containing substances are burnt off and a chlorine-rich region remains. This could indicate that the siloxane spacer units between the chloropropyl groups are released as cyclics (which are common degradation products of polysiloxanes [38]) during the electrical breakdown process.
For the Co-2 elastomer the elemental composition is comparable throughout the breakdown zone. Interestingly, one would expect HCl to form as a degradation byproduct (such as for polyvinyl chloride) but this does not seem to be the case for any of the synthesized chloropropyl-based elastomers (at least it takes place to less extent than the formation of volatile siloxane-based degradation products). The formation of HCl would lead to an auto-degradative process since a highly acidic environment is well  known to cause degradation of silicone polymers [39]. From the thermal gravimetric analysis of the investigated elastomers (shown in Figure 2), it is obvious that in terms of energies involved in the electrical breakdown correspond to the energies from a solely thermal degradation process of more than 500°C.
Another explanation for the variations in the breakdown zones may be the energy dissipated during the electrical breakdown. Since the Co-1 elastomer breaks down at lower electrical fields than the two other elastomers, Co-2 and DMS-V31, there will also most likely be the smallest excessive energy involved. The burnt volumes (determined from the outer dimensions of the breakdown zones, which equal the pinhole volume for the elastomers with no tendency to boiling) and breakdown areas (from top views) have been determined by scanning through breakdown zones by means of SEM. The resulting data is shown in Figure 7 and in Figure 8 tilted SEM images are shown to illustrate how the volume was determined by measuring out distances for various tilting angles. Figure 7. Breakdown zone analysis of Co-1 and Co-2 silicone elastomers for four different samples as shown in Figure 5. The breakdown area is determined from a top view rather than a scan through the elastomer thickness as done for the pinhole volume determination. As can be observed from Figure 7, the average burnt volume for Co-2 elastomer is smaller than for Co-1 elastomer despite the significantly larger electrical field at breakdown. In other words, for the Co-1 elastomer the energy is efficiently transferred to boiling rather than into solidification and subsequent tearing. For the Co-2 elastomer full evaporation is hindered and thus excessive energies are transferred through the elastomer leading to solidification of the silicone part of the elastomer in the traditional way. It means that the two chloropropyl-functional elastomers undergo very different breakdown mechanisms despite the sole difference being the concentration of the chloropropyl groups.
Summing up, the Co-1 elastomer breaks down prematurely in a static electrical breakdown experiment (compared to the two other elastomers) but in contrast the breakdown zones are all without possible tear propagation from solidified threads. This can be explained by electrical energy being consumed in the evaporation process of the volatiles rather than in solidification. However, despite the premature breakdown the Co-1 elastomer may be a better candidate for dielectric elastomer products since the tendency of local (microscopic) electrical breakdown with subsequent macroscopic tearing will be reduced.
Yet another interesting aspect will be to investigate if pre-stretch possibly change the electrical breakdown patterns. In recent work from our group it was proven that the electrical breakdown strength of silicone elastomers strongly improves with pre-stretch [40] even when accounting for the decreased elastomer thickness. This investigation is, however, complicated by the fact that the electrical breakdown of the highly stretched films leads to tearing of the film and thus quantitatively measuring the breakdown zones is difficult.

Conclusions
It was shown that chemically very similar silicone elastomers break down electrically in very different ways. These observations emphasize that the modification of the silicone backbone may open up for completely new possibilities for stabilizing the silicone elastomer electrically. In order to tailor the dielectric elastomers, more knowledge is needed but the results of these chloropropyl-functional elastomers pave the first path towards a better understanding of the complex connection between electrical stability and thermal stability and morphology. Minor changes in the polymer backbone structure and chemistry result in changes in electrical breakdown patterns and understanding why is crucial for the design of extraordinarily stable elastomers and thus ultimately reliable dielectric elastomer-based products.

Funding
The Danish Research Council is greatly acknowledged for the funding provided for Frederikke Bahrt Madsen.