Simon J.
Dünki
ab,
Frank A.
Nüesch
ab and
Dorina M.
Opris
*a
aSwiss Federal Laboratories for Materials Science and Technology Empa, Laboratory for Functional Polymers, Ueberlandstr. 129, CH-8600, Dübendorf, Switzerland. E-mail: dorina.opris@empa.ch
bÉcole Polytechnique Fédérale de Lausanne (EPFL), Institut des matériaux, Station 12, CH 1015, Lausanne, Switzerland
First published on 18th October 2016
Novel electroresponsive silicone elastomers modified with nitrile groups are presented, whose dielectric permittivity (ε′) is tuned from ε′ = 4.3 to ε′ = 17.4. They are prepared in a one-step process starting from a high molecular weight poly(methylvinylsiloxane) to which polar nitrile groups and cross-links are introduced in thin films. Different ratios of butanethiol/3-mercaptoproprionitrile are used to vary the amount of nitrile groups in these elastomers, while 2,2′-(ethylenedioxy)diethanethiol is used as a cross-linker. Because of the systematic nature of this investigation, we not only present promising elastic materials with remarkable dielectric, mechanical, and electromechanical properties but also provide a guideline for materials design aimed at dielectric elastomer actuator applications.
The actuation voltage can be reduced by decreasing the thickness and the elastic modulus of the elastomer film, as well as by increasing its dielectric permittivity.10 Actuators with low operation voltage and high energy density would be available if high dielectric permittivity elastomers were used as a dielectric layer. Any elastic material can in principle be used as a dielectric.11,12 Silicones are however one of the most investigated materials in the DEA field.13 They are easy to process in thin films with excellent mechanical and dielectric properties. They have however a rather low ε′ of about 3. Attempts to increase their permittivity rely on either blending with highly polarizable organic or inorganic fillers or chemical modification with polar groups.14 Significant progress has been achieved in regard to polar silicones with enhanced permittivity.4 Unfortunately, only very few polar silicones have been investigated in electromechanical tests. One common problem faced is that these polar silicones show only a moderate increase in permittivity15,16 or meet only partially the requirements for actuators, and therefore are often used as high permittivity fillers in a silicone matrix with excellent mechanical properties.17–20 It would be an obvious advantage if blending could be obviated and the cross-linked polar silicones themselves would have attractive mechanical properties and high permittivity.
We have recently shown that it is possible to prepare high molar mass polysiloxanes containing different mol% of polar CN side groups.21,22 The permittivity of such highly viscous polymers was tuned from 4.7 to 18.4. Several attempts from our laboratory to combine such attractively high permittivity with good elastic properties (large and reversible elongations) failed during the cross-linking step. In a first attempt to overcome this problem, a one-step process to prepare silicone elastomers was developed which allowed the synthesis of a material with a permittivity of ε′ = 10.1 at high frequencies.23 Even though this achievement is substantial, the chemical structures and the concept involved suggest that a permittivity of approximately 18 would be within reach. The likely reason for the obvious failure to reach this high value is seen in the photo-curing step to which thin films of this material are subjected to. The films are prepared on a substrate leaving their upper side exposed to air. If such films are now irradiated in order to bring about cross-linking and nitrile addition, the volatile thiol may evaporate leaving some vinyl groups unreacted, which then causes the permittivity to be lower than anticipated. IR investigations of the film surface exposed to air confirmed the presence of unreacted vinyl groups.
Here, we report on a novel polysiloxane which not only combines a permittivity of 17.4 with excellent mechanical properties but also does not require blending with other materials anymore. Thus, this novel polysiloxane recommends itself for direct use as the final material in DEA devices. Furthermore, we describe a way to systematically tune the permittivity in the range of 4.3–17.4 while maintaining the same cross-link density in the elastomers. This provides a unique insight into how the nitrile content impacts critical materials properties such as permittivity, dielectric loss, conductivity, and elastic modulus. Since all these properties have an impact on the electromechanical characteristics we believe that our findings are an important basis for rational materials design.
To tune the amount of CN groups, P was reacted with 1-butanethiol (1) and 2 separately and in various ratios x:
y, as well as with the dithiol 3 cross-linker so as to create materials M(x
:
y) with different contents of CN groups (y). The amount of cross-linker 3 used has a direct impact on the mechanical properties of M(x
:
y). In a typical experiment a solution of P, an approximately equimolar amount of functional thiols (1, 2 or a mixture of 1
:
2), the selected amount of dithiol cross-linker 3, and DMPA were placed between a glass plate and a Teflon substrate. The distance between the plates was adjusted by spacers. During UV-irradiation, the dipole attachment and the cross-linking in thin films occur simultaneously. The top glass plate was immediately removed after irradiation, and the films were carefully peeled off from the glass plate (Fig. 1). All films yielded elastic materials upon cross-linking.
Since we wanted to keep the amount of cross-linker constant for all materials, the optimum concentration that allowed the formation of elastomers irrespective of the CN group content had to be found. This concentration was optimized for the most challenging material, which turned out to be the one which had only butyl thioether groups M(1:
0). As mentioned in the Introduction, the synthesis of a material fully functionalized with CN groups was recently described by us, and therefore the reported recipe was used as the starting point for our investigations. When material M(1
:
0) was prepared using the reported recipe, except that thiol 2 was replaced by 1, a rather brittle material with low strain at break formed. Therefore, we gradually decreased the amount of 3 from 23 μl to 10 μl per g of P. The stress–strain curves of the resulting materials are shown in Fig. 2, while the amounts of reagents used are given in Table 1. All materials showed a strain at break of more than 100%. As expected, the materials turned softer with the decreasing amount of cross-linker since the cross-link density is decreased. The best material of composition M(1
:
0) had a strain at break of 190% and a tensile strength of 122 kPa (Fig. 2). A small improvement in the mechanical properties of M(1
:
0) was achieved when 5 wt% of hexamethyldisilazane treated silica particles were used.
![]() | ||
Fig. 2 Stress–strain curves of material M(1![]() ![]() |
Name | P [g] | 1 [g] | 2 [g] | 3 [μl] | SiO2 [wt%] | THF [ml] | Spacer μm |
---|---|---|---|---|---|---|---|
M(1![]() ![]() |
1 | 1.00 | — | 23 | — | 4 | 300 |
M(1![]() ![]() |
1 | 1.00 | — | 19 | — | 4 | 300 |
M(1![]() ![]() |
1 | 1.00 | — | 14 | — | 4 | 300 |
M(1![]() ![]() |
1 | 1.00 | — | 10 | — | 4 | 300 |
M(1![]() ![]() |
1 | 1.00 | — | 14 | 4.8 | 5 | 300 |
M(3![]() ![]() |
2 | 1.52 | 0.49 | 28 | 4.8 | 10 | 300 |
M(1![]() ![]() |
2 | 1.01 | 0.98 | 28 | 4.8 | 10 | 300 |
M(1![]() ![]() |
2 | 0.50 | 1.45 | 28 | 4.8 | 10 | 300 |
M(0![]() ![]() |
2 | — | 1.96 | 28 | 4.8 | 10 | 300 |
For the synthesis of materials M(x:
y) the amount of 3 was kept constant (14 μl per g P) (Table 1) and a dispersion of hexamethyldisilazane treated silica particles in THF was used, as it is known that silica particles can reinforce the silicones and increase their strain at break. The amount of silica was kept below 5% so that the elastic moduli are not much affected.
The conversion of the vinyl groups in the films was investigated by FT-IR measurements conducted on the two surfaces as well as with the pellets obtained by pressing the powders of ground films in KBr. The spectra were normalized to the highest peak assigned to the Si–O–Si st vibration at 1025 cm−1. For the vinyl group three characteristic signals at ν = 3055 cm−1, 3020 cm−1 and 2965 cm−1 with the one at 3055 cm−1 having the highest intensity can be easily identified. The signal at ν = 2250 cm−1 is characteristic of the CN groups. With the increasing dipole content, the intensity of the CN st band at 2250 cm−1 increased, while that of the C–H st bands at 2956 and 2929 cm−1 decreased. When covering the thin film with a glass plate during irradiation, the IR spectra clearly showed an elevated conversion of the vinyl groups. The IR spectra of the pellets obtained by pressing the powders of ground films in KBr show the presence of less than 4% unreacted vinyl groups. Some residual vinyl groups were expected, since a slightly less than stoichiometric amount of thiols to vinyl groups was used. However, as compared with our previous report23 where the film surface exposed to air had most of the vinyl groups unreacted, here, the top surface covered by the glass plate had most of the vinyl groups reacted (Fig. S1, ESI†).
The densities of M(x:
y) were measured via hydrostatic weighting in ethanol. The materials get denser with the increasing content of nitrile groups from 1.084 g cm−3 for M(1
:
0), to 1.118 g cm−3 for M(1
:
1), and reached a maximum value of 1.180 g cm−3 for M(0
:
1).
The swelling–extraction tests show that all materials contain about 10% extractable species, except for M(1:
0) for which the amount of extractable species was 13.4%. The crosslink density was determined by the Mooney–Rivlin equation using the data obtained from the stress–strain experiments. In principle, the Flory–Rhener model could also be used, if the Flory–Huggins interaction parameter of materials M(x
:
y) were known. It was found that M(1
:
0) has the lowest cross-link density of 18 mol m−3, which explains the slightly higher extraction fraction observed for this material, while the cross-link density for all other materials was between 21 and 26 mol m−3 (Table 2).
Sample |
M(1![]() ![]() |
M(3![]() ![]() |
M(1![]() ![]() |
M(1![]() ![]() |
M(0![]() ![]() |
---|---|---|---|---|---|
a Calculated using the Mooney–Rivlin equation [mol m−3]. b Calculated using the amount of starting materials [mol m−3]. | |||||
Density [g cm−3] | 1.084 | 1.097 | 1.118 | 1.154 | 1.180 |
Extraction loss [%] | 13.4 | 10.4 | 8.8 | 10.0 | 11.1 |
XL densitya | 18 | 24 | 25 | 21 | 26 |
XL densityb | 63 | 63 | 65 | 67 | 68 |
The thermal stability of M(x:
y) was investigated by TGA (Fig. S2, ESI†). All materials are stable up to 160 °C where a small amount of volatiles is removed, while above 300 °C the materials lose more than half of their weight. Differential scanning calorimetry (DSC) measurements show an increase in Tg with the increasing mol% of nitrile groups from −92.6 °C for M(1
:
0) to −67.7 °C for M(1
:
1), and reached a maximum value of −49.2 °C for M(0
:
1) (Fig. 3 and Fig. S3, ESI†). No melting points and no other transitions were observed. Table 3 summarizes the transition temperatures observed for M(x
:
y) as well as the ΔCp of the transitions. Despite the shift of Tg towards higher values due to the chemical modification of P with CN groups, the Tg of all materials was well below room temperature. The increase in Tg and ΔCp with the CN content suggests the presence of some dipole–dipole interactions which restrict the mobility of the polysiloxane segments.
Material | T g [°C] | ΔCp [J g−1 °C−1] | ||
---|---|---|---|---|
2nd Heating | Cooling | 2nd Heating | Cooling | |
M(1![]() ![]() |
−92.6 | −102.3 | 0.30 | 0.31 |
M(3![]() ![]() |
−78.3 | −88.3 | 0.35 | 0.37 |
M(1![]() ![]() |
−67.7 | −75.2 | 0.38 | 0.46 |
M(1![]() ![]() |
−58.2 | −65.5 | 0.44 | 0.43 |
M(0![]() ![]() |
−49.2 | −56.5 | 0.44 | 0.45 |
Tensile and DMA investigations were conducted on all materials M(x:
y). Fig. 4 shows the stress–strain curves of different materials. It can be seen that the elastomers get stiffer with the increasing amount of CN groups incorporated and the strain at break and the tensile strength increase. All materials were prepared using the same batch of starting polymer and the same amount of cross-linker, the only parameter changed was the ratio of thiols 1 and 2. Therefore, the change in the mechanical properties of M(x
:
y) can be attributed to the presence of CN groups. The reason behind this is the dipole–dipole interactions which slightly hinder the slipping of the segments, and therefore more stress is needed to deform the materials. The hindered freedom of segment motion is reflected by the increase in the Tg and ΔCp and also reported in the literature.27,28 Sample M(1
:
0) is the softest very likely due to its lower cross-link density and the plasticising effect of butyl groups. Because M(1
:
0) does not contain CN groups large segments can move over each other under a mechanical stress. The elastic moduli at different strain levels are summarised in Table 4. The moduli increase from 89 kPa for M(1
:
0) to 97 kPa for M(1
:
1) and reach a maximum value of 155 kPa for M(0
:
1). The strain at break of the films rose from about 210% for M(1
:
0) to 290% for M(1
:
1) up to about 320% for M(0
:
1). The positive increase in the strain at break observed for M(0
:
1) is mostly due to the dipole–dipole interactions which hinder to some extent the crack propagation.29
![]() | ||
Fig. 4 Stress–strain curves for materials M(x![]() ![]() |
Stress–strain curves obtained from standard tensile tests at various initial strain rates (0.463 s−1, 0.0463 s−1, 0.00463 s−1, and 0.00463 s−1) overlapped and an increase in the strain at break with the increasing strain rate was observed for all materials (see ESI†). The increased strain at break at higher initial strain rates might be related to increased creeping in the materials at higher rates. To elucidate this, dynamic mechanical frequency sweep measurements between 0.05 and 10 Hz for M(x:
y) were also conducted (Fig. 5). The measurements turned noisier above 3 Hz, indicating that the elastomer cannot follow the modulated amplitude. Nevertheless, the max tan(δ) values at about 6 Hz were below 0.4, a sign that the materials did not yet reach the viscoelastic regime and are still elastic. For all materials, the storage moduli E′ at 0.05 Hz range between 100 kPa and 110 kPa and they only slightly increase with an elevated content of nitrile groups. M(1
:
0) shows only a minor dependence of E′ and tan(δ) on the strain rate, while M(0
:
1) shows a significant dependence of E′ on the strain rate. For example, an increase in E′ from 110 kPa at 0.05 Hz to 195 kPa at 5 Hz was observed for material M(0
:
1) that has the highest content of nitrile groups. An increase of tan(δ) with frequency and with the increasing content of nitrile groups was noted. For example, the tan(δ) of M(0
:
1) changed from 0.08 at 0.05 Hz to 0.39 at 5 Hz. The storage modulus at low frequencies and the Young's modulus at low strains from the tensile test are quite similar (Table 4).
The permittivity of M(x:
y) was tuned by reacting different ratios of mercaptopropionitrile/butanethiol with P. As expected, a linear increase in permittivity ε′ with the increasing content of nitrile groups was observed. The ε′ values at high frequencies increased from 4.3 for M(1
:
0)via 9.7 for M(1
:
1) to the maximum of 17.4 for M(0
:
1), which has a CN group at every repeat unit (Fig. 6).
The improved process to form thin films has a direct influence on their dielectric properties. For example, when we aimed at elastomers with the maximum content of CN groups, M(0:
1), the film where one surface was exposed to air showed a permittivity value of 10.1 at 10 kHz and a relaxation peak at 500 Hz that resulted in a further increase in permittivity,23 while the film cured by the optimized method allowed the formation of M(0
:
1) with permittivity as high as ε′ = 17.4 at 10 kHz.30 A surge in permittivity at low frequencies for all materials modified with CN groups was observed. This increase is very likely due to electrode polarization, when ion impurities accumulate at the electrode. This effect is higher for the materials that have a high content of CN groups and is shifted towards higher frequencies. Unfortunately, the dielectric losses ε′′ increase at higher dipole contents. A similar behaviour was observed for the conductivity σ. For example, material M(1
:
0) has a σ = 3.5 × 10−11 S cm−1 at 10 kHz, while M(0
:
1) reached a value of σ = 6 × 10−9 S cm−1, values that are still typical for insulators. The increase in conductivity with the CN content is mainly due to increased ion mobility which is favoured by the presence of CN groups. It is still unclear how the materials got contaminated with ions and if it is possible to avoid them by using ultra-pure reagents, solvents, and glassware. Further work is required to elucidate this aspect.
All materials M(x:
y) as well as two commercial reference materials silicone Elastosil and VHB foil were evaluated regarding their actuation strain at different electric fields using circular actuators having 8 mm diameter carbon black electrodes. A biaxial prestrain of 28.6% was applied to all samples to avoid wrinkling of the electrode during actuation, except for VHB where 300% biaxial prestrain was used. The changes in the electrode diameter were measured while the voltage was increased (see the Experimental section). Fig. 7 shows the lateral actuation strain as a function of electric field for different materials M(x
:
y). The actuation as a function of electric field and the stress–strain curves of Elastosil, VHB, and M(x
:
y) are given in the ESI.†
Table 5 summarizes the dielectric permittivity, Y30%, electromechanical sensitivity ε′/Y10%, the actuation at different electric fields, some dielectric characteristics of materials M(x:
y) as well as of two commercial materials. For all materials two independent actuation tests were conducted except for M(3
:
1) for which only one actuator was measured. All materials, except for VHB actuated at electric fields lower than 20 V μm−1 whereby M(0
:
1) showed the largest actuation of 21% at the lowest field of 11.9 V μm−1. Despite the slight increase in the elastic moduli at higher nitrile contents in M(x
:
y), a rise of the actuation strain at a certain voltage with the increasing content of polar groups was observed. For example, at an electric field of 10 V μm−1, M(1
:
0) exhibits 2.7% strain, M(1
:
1) gave 6.4% strain, and M(0
:
1) reached a maximum actuation strain of 14.1%. The reason behind the low electric field required for actuating M(x
:
y) is the strong increase in permittivity for these materials, and therefore an increase in electromechanical sensitivity. It is generally accepted that the actuation strain increases with increasing ε′/Y values. Despite the lower elastic modulus of the two commercial materials, the silicone Elastosil and the acrylic VHB foil as compared to M(x
:
y), their actuation strain is rather low. This is caused by low permittivity, and therefore low ε′/Y values. Novel, chemically modified silicone based materials reported in the literature also show reduced driving voltages at increased ε′/Y values which are mainly caused by a moderate increase of permittivity and a decrease of elastic modulus.17–19,31–33 This decrease with the increasing content of polar groups in the material is somewhat surprising as one would normally expect an increase in stiffness caused by the dipolar interactions.29 The reason for this unexpected observation is a reduced cross-linking degree caused by the polar groups which resulted in a softer material.
Material | CN groups [mol%] | ε′ @ 1 kHz | Y 10% [N mm−2] | ε′/Y10% | s [%] | E [V μm−1] | s max [%] | E b,act [V μm−1] | E b [V μm−1] | β | η | R 2 |
---|---|---|---|---|---|---|---|---|---|---|---|---|
a Lateral actuation strain at 10 V μm−1. b Electric field used to induce 9.7% lateral actuation strain. c E b was determined by placing the dielectric film between electrodes of 1 mm2 area. d wt% of DR19 in materials is given. e The elastic moduli at small strains.34 | ||||||||||||
M(1![]() ![]() |
0 | 4.3 | 0.089 | 47.8 | 2.7 | 17.3 | 9.7 | 23 | 33.8 ± 4.3 | 9.4 | 35.6 | 0.87 |
M(3![]() ![]() |
25 | 6.8 | 0.096 | 68 | 4.6 | 13.8 | 17.5 | 18 | 27.0 ± 2.5 | 12.0 | 28.1 | 0.96 |
M(1![]() ![]() |
50 | 9.8 | 0.097 | 81.7 | 6.4 | 11.9 | 13.5 | 13 | 21.2 ± 2.5 | 9.4 | 22.3 | 0.94 |
M(1![]() ![]() |
75 | 12.9 | 0.128 | 92.1 | 9.6 | 10.0 | 12.0 | 11 | 19.4 ± 2.7 | 8.1 | 20.5 | 0.91 |
M(0![]() ![]() |
100 | 17.4 | 0.155 | 116 | 14.1 | 8.7 | 21.0 | 12 | 15.6 ± 1.9 | 8.6 | 16.4 | 0.96 |
Elastosil | — | 3 | 0.09 | 33.3 | 0.7 | — | 6 | 25 | — | — | — | — |
VHB | — | 4.4 | 0.2 | 22 | 0 | 68 | — | — | — | — | — | — |
E-DR19 | 0%d | 2.72 | 0.268e | 10.1 | <1.5% | — | 4.3 | 53.1 | — | — | — | — |
E-DR19 | 1.2%d | 3.05 | 0.296e | 10.3 | <1.5% | — | 7.2 | 58.8 | — | — | — | — |
E-DR19 | 4%d | 3.95 | 0.309e | 12.8 | <1.5% | 63.4 | 10.3 | 64.7 | — | — | — | — |
E-DR19 | 7.1%d | 4.64 | 0.325e | 14.1 | <1.5% | 57.8 | 17.6 | 68.2 | — | — | — | — |
E-DR19 | 10.3%d | 4.82 | 0.367e | 13.1 | <1.5% | 54.2 | 11.7 | 56.8 | — | — | — | — |
E-DR19 | 13.3%d | 4.88 | 0.725e | 6.7 | <1.5% | NA | 6.2 | 53.9 | — | — | — | — |
A recent comprehensive overview regarding the performance of different polar silicone materials by Madsen et al. is available.4 Here, we compare our results with materials reported by Zhang et al. which behave similar to ours and where actuation tests were performed. Their silicone elastomers were modified with Disperse Red 19.34 An increase in permittivity from 2.72 to 4.88 and an increase in elastic modulus from 268 kPa to 725 kPa with the increasing content of DR19 were observed (Table 5). The ε′/Y values increase with the content of DR19 up to a concentration of 10.3 wt% when a decrease of the ε′/Y value was observed. Above this concentration, the actuation performance deteriorated. In our materials, no such saturation was observed, e.g. the ε′/Y values gradually increase with the nitrile content which is reflected in a linear increase in the actuation strain. In addition, our materials show significantly higher ε′/Y values. An actuation strain as high as 21% at an electric field as low as 11.9 V μm−1 was achieved, while the DR19 modified materials reach a maximum actuation strain of 17% at significantly higher electric fields of about 68.2 V μm−1.
The dielectric breakdown field (Eb) of the films was also measured. As expected, the higher the dipole content in M(x:
y), the lower the dielectric breakdown. This trend was also observed in actuator tests, except that the Eb values in actuators are slightly lower given the fact that the electrode area was significantly larger (∼50 mm2), and therefore the probability of failure increased.
The influence the CN content in the materials M(x:
y) has on the dielectric breakdown field was further investigated using Weibull statistics using eqn (1)
![]() | (1) |
![]() | ||
Fig. 8 Weibull plot for different materials M(x![]() ![]() |
Since the points are in good agreement with the linear fit, the Weibull location parameter γ was neglected. Sample M(1:
0) is an exception (R2 = 0.87) because the points might fit better to a concave curve which might suggest that γ has to be considered. The corresponding calculations of sample M(1
:
0) for which γ ≠ 0, can be found in the ESI.† The Weibull shape parameter β corresponds to the slope of the linear fit and gives information about the shape of the distribution. The β values lie in a narrow range between 8.1 and 12.0 indicating a similar shape of the distribution for all materials. The Weibull scale parameter η was determined for F = 0.632, were η is equal to Eb and can therefore easily be calculated using eqn (2):
![]() | (2) |
Footnote |
† Electronic supplementary information (ESI) available: IR spectra, TGA, DSC curves, and the calculation of the Weibull parameter for M(1:0), stress–strain curves at different strain rates, stress–strain curves and the actuation strain as a function of electric field for Elastosil and VHB. See DOI: 10.1039/c6tc01731b |
This journal is © The Royal Society of Chemistry 2016 |