- Email: [email protected]
High-strain ionomeric–ionic liquid electroactive actuators
Sensors and Actuators A 126 (2006) 173–181
High-strain ionomeric–ionic liquid electroactive actuators Barbar J. Akle ∗ , Matthew D. Bennett, Donald J. Leo Center for Intelligent Material Systems and Structures, Department of Mechanical Engineering, Virginia Polytechnic Institute and State University, Blacksburg, VA 24061, USA Received 6 January 2005; accepted 13 September 2005 Available online 17 October 2005
Abstract Ionomeric polymers are a class of electromechanical transducer consisting of an ionomeric substrate with metal-plated electrodes. Application of a low-voltage (<5 V) across the thickness of the membrane produces controllable strain. The advantage of ionomeric polymers compared to other types of electromechanical transducers (e.g. piezoelectric polymers) is low-voltage operation, High-strain capability, and high sensitivity to motion in charge sensing mode. Two of the primary limitations of ionomeric polymers for electromechanical transducers are unstable operation in air and solvent breakdown at low-voltage. This work focuses on overcoming these limitations through the development of an ionic liquid-ionomeric composite with a tailored electrode composition that maximizes strain output. It is becoming clear that charge accumulation at the polymerelectrode interface is the key to producing high-strain in ionomeric polymer transducers. In this work, we combine a previously developed process for incorporating ionic liquids into ionomer membranes with a new method for tailoring the electrode composition. The electrode composition is studied as a function of the surface-to-volume ratio and conductivity of the metal particulates. Results demonstrate that the surface-to-volume ratio of the metal particulate is critical to increasing the capacitance of the transducer. Increased conductivity of the metal particulates produces improved response at higher frequencies (>10 Hz), but this effect is small compared to the increase in strain produced by maximizing the capacitance. Increasing capacitance produces a transducer that is able to achieve >2% strain (ε) at voltage levels of ±3 V. © 2005 Elsevier B.V. All rights reserved. Keywords: Artificial muscle; Ionic liquid; Electroactive polymer; Nafion; Ionic polymer transducer
1. Introduction Ionic polymers membranes are materials that exhibit ionic conductivity and ion selectivity. Several researchers have shown that these polymer membranes can function as electromechanical actuators and sensors [1–3]. Typically, an ionic polymer transducer is made by forming electrodes on a Nafion membrane with platinum and gold layers and saturating the membrane with water. In this state, the membrane will bend towards the anode when a voltage (1–5 V) is applied across its thickness. Similarly, the polymer will generate charge when deformed. Ionic polymer transducers have the advantage of being able to generate large strains under small applied voltages, are compliant and thus compatible with conformal structures, and have a very high sensitivity to motion when used in charge sensing mode. Nafion, which is a product of DuPont, is an ion selective rubbery
∗
Corresponding author. E-mail address: [email protected] (B.J. Akle).
0924-4247/$ – see front matter © 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.sna.2005.09.006
membrane with a fluoro backbone and pending SO3 − anions on a short chain. Transducers fabricated from ionomeric polymers suffer from several key limitations; however, most notable is dehydration and the corresponding loss in performance of these materials when operated in air. Also, the process for plating the membranes with the platinum and gold electrodes is relatively expensive, time-consuming, and produces inconsistent transducers. Efforts to overcome these issues have been hindered by the fact that the performance of the membranes is intimately linked to the morphology of the metal electrodes. The authors have previously shown that the hydration problem can be overcome by using ionic liquids as the solvent in these transducers [4,5]. Furthermore, the use of a new method for plating metal electrodes on Nafion membranes has been shown to be faster and less expensive than the current method while improving the actuation performance [6]. The current work demonstrates that these two technologies can be effectively combined to capitalize on the benefits of each. This paper will discuss each technology individually and will compare the new approaches to baseline
174
B.J. Akle et al. / Sensors and Actuators A 126 (2006) 173–181
materials. Characterization of a new transducer fabricated using a combination of the new methods will then be presented. 2. Traditional Nafion transducers In their current state, Nafion polymer transducers consist of Nafion membranes (typically Nafion-117) that have been plated with interpenetrating platinum electrodes using an “impregnation/reduction” technique [7–9]. This deposition is performed by first saturating the membrane with Pt(NH3 )4 2+ ions from a tetraammineplatinum chloride solution. The adsorbed ions are then reduced to platinum metal by soaking the membrane in a solution of sodium borohydride. This process results in a surface platinum electrode that penetrates into the membrane up to a depth of 20 m. Penetration of the electrode can be enhanced by repeating the impregnation/reduction process several times. In order to improve the conductivity of the platinum interpenetrating electrodes, a thin layer of gold is typically electroplated on the outer surface of the membrane. Following the metal deposition process, the mobile cation in the membrane can be exchanged for any suitable ion by soaking the membrane in a salt solution of that ion. This process is done in aqueous solutions with the membrane in a water-saturated state. Because the platinum electrodes made in this process penetrate into the polymer, the interfacial area between the polymer and the metal is very large. This large interfacial area is the reason for the very large capacitance of these devices (1–5 mF/cm2 ). Several researchers have shown that the strain generation in Nafion actuators is proportional to the membrane capacitance and it is now generally accepted that the actuation mechanism in these membranes is driven by charge motion within the polymer and accumulation at the surface [10–17]. Therefore, in order for an electrode to be effective, it must offer the property of a large polymer/metal interfacial area. A new method for fabricating electrodes having a larger polymer/metal interfacial area has been developed by Akle and Leo [6]. In this process, a metal powder is mixed with a solution of Nafion polymer in alcohols. This mixture is painted onto the Nafion membranes and hot-pressed into the surface. Transducers made in this way have been shown to generate strains up to three times those of their traditional platinum-plated counterparts and are faster and less expensive to make. This conducting powder painting method allows better control on electrode morphology and enhances transducer manufacturing repeatability. In order for the transduction to occur in Nafion membranes, the cations within the membrane must be mobilized by saturation with an appropriate solvent. Typically, the solvent used has been water. The reasons for this are the favorable interaction of the Nafion polymer with water, and the low viscosity of water. Nafion membranes will absorb up to 38% of their dry weight in water and achieve an ionic conductivity of 83 mS/cm when fully hydrated [18]. The water-swollen membranes make excellent electromechanical transducers, but suffer from the limitation that they exhibit a dramatic loss in performance when dehydrated. Efforts to contain the water inside the polymer have met with little success. Another, less obvious problem associated with the use of water as a solvent is the small electrochemical
stability window of water. The water will separate into gaseous hydrogen and oxygen at an applied voltage greater than 1.23 V (versus platinum) and therefore the actuation voltage applied to the membranes must not exceed this value. Transduction has been demonstrated when using other, more stable solvents such as glycerol and ethylene glycol, but the speed of the response is limited by the high viscosity of these solvents [19,20]. Recent work by Bennett and Leo has shown that ionic liquids can be used as solvents for Nafion transducers with the benefit that the ionic liquids are non-volatile and will not evaporate like water [4,5]. Transducers made using these ionic liquids have been shown to be stable for over 250,000 cycles in air, as compared to about 2000 cycles for water-solvated materials. The goal of the current work is to demonstrate that these two approaches can be combined and to compare the transducers made using this new method to traditional platinum-plated, water-swollen Nafion membrane transducers. Performance metrics including generated strain, electrical impedance, and sensitivity will be presented. 3. Ionic liquids as solvents In this work, we demonstrate a method for making ionic polymer transducers using highly stable room temperature ionic liquids. Ionic liquids are salts containing only charged species that exist in their liquid state at room temperature. They have an immeasurably low vapor pressure, electrochemical stability windows of 4 V or more, and are thermally stable to temperatures as high as 400 ◦ C. Furthermore, ionic liquids have high ionic conductivities and can be used as electrolytes for a variety of applications, including electrochemical capacitors [21] and conducting polymer actuators [22,23]. The reasoning behind using ionic liquids as solvents for Nafion transducers is that they will be very stable and therefore will eliminate the problem of solvent evaporation, thus allowing for the use of Nafion transducers in a broader range of environments. Also, the ionic liquids are themselves ionically conductive and should therefore facilitate ionic motion in the Nafion membrane, and may possibly enhance the ion motion as compared to water. Furthermore, the electrochemical stability window of most ionic liquids is larger than that of water and therefore larger actuation voltages will be possible, thus increasing the available energy density of these devices. In order to utilize an ionic liquid as a solvent for an ionic polymer transducer, a Nafion membrane is typically first plated with platinum and gold electrodes by using the traditional impregnation/reduction process. As mentioned, this process is carried out in aqueous solution with the polymer fully saturated with water. Following the plating process the membrane is dried in a vacuum oven and placed into an appropriate ionic liquid and heated. The diffusion of the ionic liquid into the Nafion membrane is dependant on the ionic liquid used, but is typically slow and therefore high temperature (>80 ◦ C) and long soak times (>4 h) are used. The ionic liquids 1-butyl-3-methylimidazolium tetrafluoroborate (BMI-BF4) and 1-butyl-3-methylimidazolium trifluoromethanesulfonate (BMI-Tf) have been imbibed into Nafion membranes by Doyle et al. [24] and they have reported
B.J. Akle et al. / Sensors and Actuators A 126 (2006) 173–181
that the uptake of these ionic liquids into the membrane is proportional to the temperature at which the swelling is carried out. Furthermore, they have shown that increasing the ionic liquid content of the membrane will result in higher ionic conductivity. In their work, the Nafion membranes were not plated with metal electrodes and they used temperatures as high as 180 ◦ C. In the current work, a problem with electrode stability has been observed at high uptake of the ionic liquid into the membrane. It has been reported by Gebel et al. [25] that water uptake by the Nafion-117 membrane results in a dimensional expansion of 14 and 10% along the length and width of the membrane. The metal electrodes are plated with the membrane in its fully water-saturated state. It has been found that when the membrane is dried and re-swollen by soaking it in the ionic liquid at elevated temperature, the electrodes often develop cracks, especially in the layers of electroplated gold, that result in a loss of surface conductivity of up to six orders of magnitude. This damage is due to the large strains imposed in the electrodes during the drying and re-swelling process. In order to overcome this problem, a modification to the platinum and gold plating process has been developed to improve the robustness of the electrodes. In this modification, the platinum and gold plating process is carried out as before, except that 25% (by volume) ethanol or methanol is added to each of the solutions to be used for the plating. Because the Nafion membrane will swell by 45% in ethanol and 51% in methanol [25], the use of these as co-solvents will increase the size of the membrane during the plating step. It has been found that this will allow for the membrane to be dried and re-swollen with the ionic liquid without damage to the electrodes and a corresponding loss in conductivity. Membranes plated in this way have been swollen with the 1-ethyl-3-methyimidazolium trifluoromethanesulfonate (EmI-Tf) ionic liquid and characterized versus their water-swollen counterparts (see Bennett and Leo [4,5]). 4. Direct assembly plating method In order to overcome the problem of dehydration and electrode instability, a new method for fabricating Nafion transducers has been developed that combines the use of ionic liquids and metal powder painted electrodes in the same device. In this method, a bare Nafion membrane is first exchanged from the proton counterion form into one of the alkali metal (e.g. Na+ , Li+ ) forms. This is to prevent charring of the membrane during the drying step, which has been witnessed when Nafion membranes in the acid form are heated for extended periods of time. After ion exchange, the membrane is dried in an oven at 150 ◦ C under vacuum for 12 h. The dry membrane is then immersed in neat EmI-Tf ionic liquid and heated to 150 ◦ C for 4.5 h. At this temperature, the uptake of ionic liquid is 60% by the dry weight of the membrane. In order to evaluate the effect of several of these parameters on the performance of the transducers, a detailed study of different electrode types has been carried out. This method allows the variation of several parameters including the type of the conducting powder. As mentioned earlier, the maximum strain will be generated when the capacitance of the membrane is maxi-
175
mized. For this reason, a ruthenium(IV) oxide (RuO2 ) powder was used in the electrode. This powder was obtained from Alfa Aesar and has a particle size of 3–5 nm and a specific area of 45–60 m2 /g. Capacitance in ionomeric materials is an electric double layer type capacitance, which increase proportional to the ionomer/conductor interfacial area. After the ionic liquid has been imbibed into the membrane, the electrodes are applied using the metal powder painting technique. A polymer/metal solution is prepared typically containing 47% (by weight) of 5% Nafion solution, 47% glycerol, and 6% metal powder. These concentrations are chosen based on previous experiences in building fuel cell electrodes. This solution is painted directly onto each surface of the ionic liquid-containing Nafion membrane. The glycerol and alcohol solvents (in the Nafion solution) are removed by baking the membrane at 130 ◦ C under vacuum for 15 min between each electrode layer; typically a total of four layers are applied to each side of the membrane. Following the electrode painting process, the membrane is dried at 130 ◦ C under vacuum for an additional 10 min to ensure complete removal of all volatile components from the composite. Based on the weight change of the membranes, the loading of ruthenium dioxide is estimated to be around 5–10 mg/cm2 of membrane area. After the ruthenium dioxide layers have been painted onto the surfaces of the membrane and dried, the sample is sandwiched between two 100 nm-thick conductive gold foils and the five layers of the composite are intimately bound together by a hot-pressing process. In the previous method, this hot-pressing was carried out at 20 MPa and 210 ◦ C for 2 min. However, the membranes in that case contained no solvent during the hot-pressing process. For the membranes that are swollen with the ionic liquid, this high pressure and long duration causes the device to melt to the point that the polymer flows. In order to maintain the dimensional stability of the membrane during the hot-pressing process while still binding the three layers together, the pressure and pressing time have been reduced. It has been found that good results are obtained for hot-pressing carried out at 4 MPa and 210 ◦ C for around 20–120 s depending on the metal loading of the electrode. The gold foils used were Falcon-Brand 24 K gold leafs; they helped decrease the surface resistance from greater than 100 k/cm for the painted metal layers to less than 1 /cm. However, the conductivity of the electrodes is also important to obtaining wellperforming transducers. The gold leaf is used to enhance the surface conductivity of the devices, but high electrode conductivity through the thickness of the electrodes is also important in order to achieve rapid charging and fast motion of the transducer. Although the RuO2 powder has a very high surface area, it is not very conductive (Bhaskar et al. report that the resistivity of thin RuO2 films is greater than 290 cm [26]). In order to assess the importance of surface area-to-volume ratio and particulate conductivity on performance, electrodes were prepared using a mixture of gold (Au) powder and ruthenium(IV) oxide powder. The gold powder is in the form of flakes of ∼3 m in length and has a specific area probably lower than 1 m2 /g. However, the resistivity of gold is 2.2 cm, two orders of magnitude lower than RuO2 . Therefore, two metal particles are used, one with a high area and low conductivity and one with a low area and
B.J. Akle et al. / Sensors and Actuators A 126 (2006) 173–181
176
Fig. 1. SEM images of (a) electrode built using the traditional impregnation/reduction method and (b) electrodes built using the direct assembly process.
high-conductivity. By varying the relative proportions of these metals, we can correlate transducer properties to electrode composition. Representative scanning electron microscopy (SEM) images of electrodes assembled using the traditional impregnation/reduction and the direct assembly process are shown in Fig. 1(a) and (b), respectively. 5. Experimental procedure The materials fabricated in this work are characterized for their performance as electromechanical transducers using a single test setup that measures two actuation properties; the strain (ε) and the electrical impedance. The samples used for these tests are 3 mm × 30 mm rectangular transducers held in a cantilevered configuration in a clamp with fixed gold electrodes (see Fig. 2). The free length of the transducer is 19 mm. The thickness of the transducer varies between approximately 0.18 and 0.21 mm depending on the number of electrode layers and the duration of the hot pressing. Representation of the test setup is shown in Fig. 2. For the first property, a random voltage with a RMS magnitude of 1 V is applied to the sample using a Fourier analyzer and amplifier. The displacement x(t) at any point along the length is measured with a non-contact laser vibrometer. The current i(t) induced in the polymer due to the applied voltage v(t) is also measured by measuring the voltage drop across a small (0.1 ) resistor—see Fig. 2. Using the Fourier analyser, the fre-
quency response function between the excitation voltage and the output displacement and between the input voltage and induced current are measured over the frequency range 0.2–200 Hz. This setup is also used to measure the time domain response of the transducer to a step voltage input. The free deflection δ(t) (in the time domain) or the free deflection frequency response (δ(ω)/v(ω)) are used with Eqs. (1) and (2), respectively, to obtain the generated strain as a function of time or frequency. In Eqs. (1) and (2), ε is strain, h is the thickness of the sample, and Lf is the free length of the sample. These equations assume that the transducer bends with a constant curvature; experimental results demonstrate that this is an appropriate assumption for the experiments performed in this paper. ε(t) =
δ(t)h L2f
ε(ω) (δ(ω)/v(ω))h = v(ω) L2f
(1) (2)
The impedance Z(ω) of the samples is computed using Eq. (3), and the capacitance is computed using Eq. (4) by assuming the polymer is a network of capacitors and a resistor in series [27]. Z(ω) =
v(ω) i(ω)
(3)
C(ω) =
1 (Im(Z(ω))ωA)
(4)
where A is the area of the transducer. 6. Experimental results
Fig. 2. Setup for the free deflection and impedance tests.
Experimental results are obtained for three sets of experiments. In each experiment, the electrode composition is varied and the time response and frequency response of a transducer is measured. The goal of the experiments is to correlate the
B.J. Akle et al. / Sensors and Actuators A 126 (2006) 173–181
177
Table 1 Weights of Au and RuO2 mixed with 0.5 g of 5% NafionTM solution and 0.5 g glycerol Sample 0% Au 100% RuO2 25% Au 75% RuO2 50% Au 50% RuO2 75% Au 25% RuO2 100% Au 0% RuO2 Fig. 3. Strain response as a function of electrode composition by weight.
electrode properties with the electromechanical performance of the transducer. 6.1. Variation of metal content by weight In the first experiment, the electrode composition is varied according to the weight of RuO2 or Au contained within the electrode. The weight percentage of metal in the electrode is varied from 0% Au to 100% Au, with the remaining percentage of the electrode metal being RuO2 . Fig. 3 is a plot of the frequency response magnitude for the five cases studied. The results clearly demonstrate that the RuO2 electrode is superior to Au in producing strain for an equivalent applied voltage. The strain-per-voltage response of the polymer at frequencies less than 10 Hz is maximized by the pure RuO2 electrode and is minimized for the pure Au electrode. Using 1 Hz as a reference point, we see that the strain response drops from 273 /V to 6 /V when the electrode composition is changed from 100% RuO2 to 100% Au. At intermediate values, there is a clear trend of decreasing strain response for increasing Au loading. The trends in the electrical impedance clearly indicate the tradeoff in capacitance and resistance associated with the addition of gold or ruthenium oxide. A pure RuO2 electrode produces a higher capacitance electrode due to its larger specific surface area. This results in a lower impedance at low frequencies (electrical impedance is inversely proportional to capacitance). However, the device with the pure RuO2 electrode has a larger impedance at frequencies above approximately 10 Hz because of the higher resistivity of RuO2 as compared to Au (see Fig. 4). The addition of small amounts of Au to the electrode decreases
Fig. 4. Electrical impedance as a function of electrode composition by weight.
(2.5:1) (2.5:1) (2.5:1) (2.5:1) (2.5:1)
Au (g)
RuO2 (g)
Thickness (m)
0 0.0428 0.0855 0.1283 0.1711
0.0625 0.0469 0.0313 0.0156 0
205 180 180 180 180
It also presents the durations of sample hot-pressing. The densities of Au and RuO2 are assumed to be 19300 and 7050 kg/m3 , respectively.
the high-frequency resistance but also decreases the low frequency capacitance because of the lower specific area of the gold powder as compared to the RuO2 powder. Capacitance at 0.5 Hz for samples with 0, 25, 75, and 100% gold is 3.9, 3.7, 3.5, 1.6, and 1.0 mF/cm2 , respectively. Therefore, the pure Au electrode has the lowest capacitance but also has superior high-frequency resistance properties. Akle et al. [27] have demonstrated a linear relationship between capacitance and the strain response in ionic polymer actuators. This relationship explains the largest low frequency strain per volt in the 0% Au sample (see Fig. 3), which has the largest capacitance, while the 100% Au has the smallest strain per volt and capacitance. 6.2. Variation of metal content by volume The second experiment consists of varying the electrode composition while maintaining a consistent volume of metal in the electrode. In this set of experiments, the relative percentage of RuO2 and Au is varied while the total volume of metal in the electrode layers is held constant at approximately 40%. Table 1 summarizes the compositions studied for this set of experiments. The strain response of the transducers as a function of volume percent of metal exhibited the same trend as the trend obtained when varying the weight percentage (see Fig. 5). The electrode fabricated with pure RuO2 exhibits the largest strain per unit voltage at frequencies below approximately 8 Hz. Increasing the percentage of Au in the electrode reduces the strain output. At 1 Hz, the strain output is 272 /V for an electrode with 100% RuO2 while it is 24 /V for an electrode with 100% Au. A clear relationship between the slope of the strain-to-voltage frequency response at frequencies below the first mechanical resonance and electrode composition also emerges in this study.
Fig. 5. Strain response as a function of electrode composition by volume.
178
B.J. Akle et al. / Sensors and Actuators A 126 (2006) 173–181
Fig. 8. Strain output to the application of square wave potentials of ±2 V. Fig. 6. Electrical impedance as a function of electrode composition by volume.
The negative slope of the strain-to-voltage frequency response indicates limited charge mobility. This is at least partially due to the high resistivity of the RuO2 particles, as increasing the percentage of gold in the electrode decreases the slope of the frequency response. The decrease in slope is correlated with the variation in the electrical impedance as the electrode composition varies (see Fig. 6). The device with a 100% RuO2 electrode has the highest low-frequency capacitance, but exhibits a highfrequency impedance of approximately 5 . The addition of small amounts of gold maintains the low-frequency capacitance but decreases the high-frequency impedance to approximately 30 . Further increase in the gold content reduces the capacitance and causes a corresponding decrease in the strain generation. Differences in the strain response due to variations in the electrical impedance are reduced by normalizing the strain response to the charge induced in the polymer by the applied potential. The strain per unit electric displacement is illustrated in Fig. 7 for the variation in electrode composition by volume. The results demonstrate that low-frequency slope of the strain-to-voltage response is eliminated and the strain per unit charge is approximately flat as a function of frequency. Furthermore, the strain per unit electric displacement is between 5 and 15 /C/m2 for all samples considered. The time response of the transducer displacement to square wave potential waveforms are also measured and the strain is computed using Eq. (2). Results for square wave amplitudes of ±2 V demonstrate that strain output of approximately 4000 (0.4%) is achievable with the pure RuO2 electrode (see Fig. 8). The steady-state value of the strain decreases as the gold loading
Fig. 7. Strain per unit charge/area for the samples produced by variation in the metal content with consistent volume.
of the electrode is increased. This is because the capacitance of the transducers decreases as the content of ruthenium oxide is decreased. However, the conductivity of the electrode increases as the loading of gold is increased, leading to a faster response. Although the speed of the response increases notably from the 100% RuO2 electrode to the 100% Au electrode, this trend is not as evident for the intermediate electrode compositions. This is illustrated by comparing the normalized initial slope of the two samples, which is the initial slope of the step divided by the maximum strain attained. The sample with 100% RuO2 electrodes had a normalized initial slope of 1 s−1 as compared to 6.7 s−1 for the 100% Au electrode sample. This result is consistent with the frequency-domain analysis, as the pure ruthenium oxide sample exhibits a strongly negative slope in the strain-to-voltage response whereas the pure gold sample exhibits a relatively flat response. 6.3. Electrode optimization The results of the electrode analysis demonstrate that a metal particulate with high surface area-to-volume ratio (RuO2 ) will produce a superior electrode due to the increase in capacitance at low-frequencies. Increasing the content of the high-conductivity particulate (Au) decreases the impedance at frequencies above approximately 10 Hz, but does not produce an increase in the strain response because of the corresponding drop in capacitance associated with replacing the RuO2 particles with the low specific area gold particles. Three samples were prepared for the optimization and verification study; the first sample had the same metal to polymer ratio in the electrode as in the weight and volume experiments (2.5:1). However, the thickness of the electrode for this sample was increased by 50% to six layers as compared to four layers for the previous experiments. For the second sample, the RuO2 to polymer ratio was increased by 50% to 3.75:1 while holding the number of layers constant at four. The third sample was prepared by replacing the 50% increase in RuO2 by the equivalent volume of Au. Table 2 summarizes the electrode compositions studied for this set of experiments. Therefore, the three samples maintained the same total metal volume loading. The metal in sample 1 is dispersed compared to the other two samples, while gold accounted for the increase of metal in sample 3. In samples 1 and 2, the surface area of the electrode is theoretically maintained, while sample 3 should have a better conductivity
B.J. Akle et al. / Sensors and Actuators A 126 (2006) 173–181
179
Table 2 Weights of Au and RuO2 mixed with 0.5 g of 5% NafionTM solution and 0.5 g glycerol for electrode optimisation Sample
Au (g)
RuO2 (g)
Thickness (m)
0% Au 100% RuO2 (2.5:1) 0% Au 100% RuO2 (3.75:1) 33% Au 66% RuO2 (3.75:1)
0 0 0.0855
0.0625 0.0938 0.0625
210 180 190
Fig. 11. per volt frequency response for three pure RuO2 samples.
at maximum deflection, Rmax . The strain at maximum deflection is equal to: t εmax = (3) 2Rmax
Fig. 9. Frequency response of the samples with tailored electrodes.
through the electrode. Fig. 9 shows the frequency response functions of the strain per unit volt for the three samples. The frequency response functions of samples 1 and 2 have nearly the same slope with sample 1 providing better performance. This is believed to be due to the lack of control on the amount of electrode ink deposited using the brush painting technique. Theoretically, the two samples should provide similar capacitance and therefore similar performance. As for sample 3, it is shown to have the least low frequency response while it crossed sample 2 almost at 3 Hz and provided better high-frequency response. This is consistent with the expectation of a better high-frequency response from a more conductive, gold-containing electrode. The motion of the transducer to an applied potential is also measured in the time domain. Square wave potentials are applied to the polymer and the resulting displacement is measured using a digital camera due to the fact that the displacements are too large to be measured with the laser vibrometer. The measured deflected shapes are shown in Fig. 10. The pictures illustrate that the assumption of constant curvature is valid over the range of conditions tested. The range of motion increases as the amplitude of the square wave input is increased. The peak-to-peak strain in the sample is measured by determining the radius of curvature of the sample
The radius of curvature was determined on both sides of the neutral position and the peak-to-peak strain is the addition of the two values. The measured strains increase from 8195 (0.82%) at ±2 V to 24,380 (2.44%) at ±4 V. It is interesting to note that the increase in peak-to-peak strain is not linear as a function of input amplitude. The asymmetry in the deflection, which is especially significant in the ±3 V sample, is due to the pre-bend in some of the samples. Once again, we attribute this to non-linearity in the charge conduction as a function of voltage. 7. Method characterization In order to gauge the repeatability of the process described in this paper, three identical transducers are separately built and fully characterized using the methods described previously. The transducers were fabricated by painting four layers of 100% RuO2 ink with a weight ratio of metal to polymer of 2.5 to 1. As can be seen in Fig. 11, very little variation is observed in the per unit volt frequency response. Although more results are needed in order to obtain a standard deviation, this initial result shows that the new process is able to produce repeatable results. In order to determine the long-term stability of the transducers’ operation, a reliability test is also performed on these three identical samples. The three samples were actuated continuously with a 1 Hz sine wave potential and the free strain generated in each sample was measured periodically. The effect of the
Fig. 10. Deflected shapes as a function of amplitude for a square wave potential input.
180
B.J. Akle et al. / Sensors and Actuators A 126 (2006) 173–181
Acknowledgements This work was supported by the U.S. Army Research Laboratory and U.S. Army Research Office under contract/grant number DAAD19-02-1-0275 Macromolecular Architecture for Performance (MAP) MURI. Supplementary funding was provided by the Virginia Space Grant Consortium. References
Fig. 12. Reliability test on the three samples with ±1, ±2, and ±3 V 1 Hz sine wave applied on samples 1, 2, and 3, respectively.
peak amplitude of the applied potential on the long-term stability of the transducers is illustrated in Fig. 12. As can be seen, the samples that were actuated with a 1 and 2 V peak potential did not exhibit and degradation of the response after more than 250,000 cycles. It should be noted that the maximum generated strain in the samples is lower than in Fig. 10 because of the time scales involved. In order to achieve the large strains presented in Fig. 10, an input frequency of less than 0.04 Hz is required. This is due to the slow response speed of the devices fabricated with pure electrodes. However, the result demonstrates the improved long-term stability of ionic polymer transducers that utilize ionic liquids instead of water as a solvent. By contrast, Bennett and Leo [4] have shown that a water-swollen ionic polymer transducer would operate for about 2000 cycles in air. After 3600 cycles, the motion decreased by 96% due to solvent evaporation. Although the specific reason for the decrease in performance of the sample that was driven at ±3 V is unknown, a likely explanation is degradation of the ionic liquid. Cooper and Sullivan [28] and Bonhote et al. [29] have shown that the electrochemical stability window of the EMI-Tf ionic liquid is 4.1 V.
8. Conclusions Tailoring the electrode composition is an effective method for achieving high-strain transducers using ionomeric polymer materials. Our results demonstrate that surface area-to-volume ratio of the metal particulate is a critical property in achieving large capacitance at low frequencies. Increasing the low-frequency capacitance increases the strain output of the transducer. The electrical conductivity through the electrode is also proved to be critical to the speed of response of the transducers. A tradeoff between high speed for gold particles and large strains for ruthenium dioxide particles is observed. Results presented in this work yielded an ionomer-ionic liquid transducer that produced greater than 2% ε at peak-to-peak voltages of 6 V. Finally, fabricating air stable ionic polymer transducers with enhanced strain and strain rate will create a new range of application such as control and positioning actuators for conformal structures.
[1] K. Sadeghipour, R. Salomon, S. Neogi, Development of a novel electrochemically active membrane and ‘Smart’ material based vibration sensor/damper, Smart Mater. Struct. 1 (1992) 172–179. [2] K. Oguro, Y. Kawami, H. Takenaka, Bending of an ion-conducting polymer film-electrode composite by an electric stimulus at low-voltage, J. Micromach. Soc. 5 (1992) 27–30. [3] M. Shahinpoor, Y. Bar-Cohen, J. Simpson, J. Smith, Ionic polymermetal composites (IPMCs) as biomimetic sensors, actuators and artificial muscles—a review, Smart Mater. Struct. 7 (1998) R15-R30. [4] M. Bennett, D. Leo, Ionic liquids as solvents for ionic polymer transducers, Sens. Actuators A: Phys. 115, 79–90. [5] M. Bennett, D. Leo, Ionic liquids as novel solvents for ionic polymer transducers, in: Proceedings of the EAP Actuators and Devices. SPIE, paper 5385-27, 2004. [6] B. Akle, K. Wiles, D. Leo, J. McGrath, Effects of electrode morphology on the performance of BPSH and PATS ionic polymer transducers, in: Proceedings of the EAP Actuators and Devices. SPIE, paper 5385-73, 2004. [7] P. Millet, R. Durand, E. Dartyge, G. Tourillon, A. Fontaine, Precipitation of metallic platinum into Nafion ionomer membranes, J. Electrochem. Soc. 140 (1993) 1373–1379. [8] K. Kim, M. Shahinpoor, The synthesis of nano-scaled platinum particles (NSPP)—their role in performance improvement of ionic polymer-metal composite (IPMC) artificial muscles, in: Proceedings of the EAP Actuators and Devices. SPIE, vol. 4329, 2001, pp. 189–198. [9] T. Rashid, M. Shahinpoor. Force optimization of ionic polymermic platinum composite artificial muscles by means of orthogonal array manufacturing method, in: Proceedings of the EAP Actuators and Devices. SPIE, vol. 3669, 1999, pp. 289–298. [10] P. de Gennes, K. Okumura, M. Shahinpoor, K.J. Kim, Mechanoelectric effects in ionic gels, Europhys. Lett. 50 (2000) 513–518. [11] K.M. Newbury, D.J. Leo, Electromechanical modeling and characterization of ionic polymer benders, J. Intell. Mater. Syst. Struct. 13 (2002) 51–60. [12] K.M. Newbury, D.J. Leo, Linear electromechanical model of ionic polymer transducers. Part I. Model development, J. Intell. Mater. Syst. Struct. 14 (2003) 333–342. [13] K.M. Newbury, D.J. Leo, Linear electromechanical model of ionic polymer transducers. Part II. Experimental validation, J. Intell. Mater. Syst. Struct. 14 (2003) 343–358. [14] K. Farinholt, D.J. Leo, Modeling of electromechanical charge sensing in ionic polymer transducers, Mech. Mater. 36 421–433. [15] S. Nemat-Nasser, Micro-mechanics of actuation of ionic polymer-metal composites, J. Appl. Phys. 92 (2002) 2899–2915. [16] S. Nemat-Nasser, J. Li, Electromechanical response of ionic polymermetal composites, J. Appl. Phys. 87 (2000) 3321–3331. [17] K. Asaka, N. Fujiwara, K. Oguro, K. Onishi, S. Sewa, State of water and ionic conductivity of solid polymer electrolyte membranes in relation to polymer actuators, J. Electromech. Chem. 505 (2001) 24– 32. [18] DuPont, Nafion membranes NE-112, NE-1135, N-115, N-117. Product information, 2000. [19] S. Nemat-Nasser, Micromechanics of actuation of ionic polymer-metal composites, J. Appl. Phys. 92 (2002) 2899–2915. [20] S. Nemat-Nasser, S. Zamani, Experimental study of nafion and flemionbased ionic polymer-metal composites (IPMCs) with ethylene glycol as
B.J. Akle et al. / Sensors and Actuators A 126 (2006) 173–181
[21]
[22]
[23]
[24]
solvent, in: Proceedings of the EAP Actuators and Devices. SPIE, vol. 5051, 2003. pp. 233–244. A.B. McEwen, H.N. Ngo, K. LeCompte, J.L. Goldman, Electrochemical properties of imidazolium salt electrolytes for electrochemical capacitor applications, J. Electrochem. Soc. 146 (1999) 1687–1695. W. Lu, A.G. Fadeev, B. Qi, E. Smela, B.R. Mattes, J. Ding, G.M. Spinks, J. Mazurkiewicz, D. Zhou, G.G. Wallace, D.R. MacFarlane, S.A. Forsyth, M. Forsyth, Use of ionic liquids for -conjugated polymer electrochemical devices, Science 297 (2002) 983–987. F. Vidal, C. Plesse, D. Teyssi´e, C. Chevrot, Long-life air working conducting semi-Ipn/ionic liquid based actuator, Synth. Metals 142 (2004) 287–291. M. Doyle, S. Choi, G. Proulx, High-temperature proton conducting membranes based on perfluorinated ionomer membrane-ionic liquid composites, J. Electrochem. Soc. 147 (2000) 34–37.
181
[25] G. Gebel, P. Aldebert, M. Pineri, Swelling study of perfluorosulphonated ionomer membranes, Polymer 34 (1993) 333–339. [26] S. Bhaskar, P.S. Dobal, M. Majumder, R.S. Katiyar, Characterization of conductive RuO2 thin films as bottom electrodes for ferroelectric thin films, in: Proceedings of the Ferroelectric Thin Films IX, Materials Research Society, vol. 655, paper CC4.7, 2001. [27] B. Akle, M. Hickner, D. Leo, J. McGrath, Correlation of capacitance and actuation in ionomeric polymer transducers, accepted in the J. Mater. Sci., 2005. [28] E.I. Cooper, E.J. S O’Sullivan, New, Stable, Ambient-Temperature Molten Salts, In Proceedings of the Eighth Interational Symposium on Molten Salts, The Electrochemical Society, 1992, pp. 386–396. [29] P. Bonhˆote, A. Paula Dias, N. Papageorgiou, K. Kalyanasundaram, M. Gr¨atzel, Hydrophobic, highly conductive ambient-temperature molten salts, Inorg. Chem. 35 (1996) 1168–1178.
High-strain ionomeric–ionic liquid electroactive actuators Barbar J. Akle ∗ , Matthew D. Bennett, Donald J. Leo Center for Intelligent Material Systems and Structures, Department of Mechanical Engineering, Virginia Polytechnic Institute and State University, Blacksburg, VA 24061, USA Received 6 January 2005; accepted 13 September 2005 Available online 17 October 2005
Abstract Ionomeric polymers are a class of electromechanical transducer consisting of an ionomeric substrate with metal-plated electrodes. Application of a low-voltage (<5 V) across the thickness of the membrane produces controllable strain. The advantage of ionomeric polymers compared to other types of electromechanical transducers (e.g. piezoelectric polymers) is low-voltage operation, High-strain capability, and high sensitivity to motion in charge sensing mode. Two of the primary limitations of ionomeric polymers for electromechanical transducers are unstable operation in air and solvent breakdown at low-voltage. This work focuses on overcoming these limitations through the development of an ionic liquid-ionomeric composite with a tailored electrode composition that maximizes strain output. It is becoming clear that charge accumulation at the polymerelectrode interface is the key to producing high-strain in ionomeric polymer transducers. In this work, we combine a previously developed process for incorporating ionic liquids into ionomer membranes with a new method for tailoring the electrode composition. The electrode composition is studied as a function of the surface-to-volume ratio and conductivity of the metal particulates. Results demonstrate that the surface-to-volume ratio of the metal particulate is critical to increasing the capacitance of the transducer. Increased conductivity of the metal particulates produces improved response at higher frequencies (>10 Hz), but this effect is small compared to the increase in strain produced by maximizing the capacitance. Increasing capacitance produces a transducer that is able to achieve >2% strain (ε) at voltage levels of ±3 V. © 2005 Elsevier B.V. All rights reserved. Keywords: Artificial muscle; Ionic liquid; Electroactive polymer; Nafion; Ionic polymer transducer
1. Introduction Ionic polymers membranes are materials that exhibit ionic conductivity and ion selectivity. Several researchers have shown that these polymer membranes can function as electromechanical actuators and sensors [1–3]. Typically, an ionic polymer transducer is made by forming electrodes on a Nafion membrane with platinum and gold layers and saturating the membrane with water. In this state, the membrane will bend towards the anode when a voltage (1–5 V) is applied across its thickness. Similarly, the polymer will generate charge when deformed. Ionic polymer transducers have the advantage of being able to generate large strains under small applied voltages, are compliant and thus compatible with conformal structures, and have a very high sensitivity to motion when used in charge sensing mode. Nafion, which is a product of DuPont, is an ion selective rubbery
∗
Corresponding author. E-mail address: [email protected] (B.J. Akle).
0924-4247/$ – see front matter © 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.sna.2005.09.006
membrane with a fluoro backbone and pending SO3 − anions on a short chain. Transducers fabricated from ionomeric polymers suffer from several key limitations; however, most notable is dehydration and the corresponding loss in performance of these materials when operated in air. Also, the process for plating the membranes with the platinum and gold electrodes is relatively expensive, time-consuming, and produces inconsistent transducers. Efforts to overcome these issues have been hindered by the fact that the performance of the membranes is intimately linked to the morphology of the metal electrodes. The authors have previously shown that the hydration problem can be overcome by using ionic liquids as the solvent in these transducers [4,5]. Furthermore, the use of a new method for plating metal electrodes on Nafion membranes has been shown to be faster and less expensive than the current method while improving the actuation performance [6]. The current work demonstrates that these two technologies can be effectively combined to capitalize on the benefits of each. This paper will discuss each technology individually and will compare the new approaches to baseline
174
B.J. Akle et al. / Sensors and Actuators A 126 (2006) 173–181
materials. Characterization of a new transducer fabricated using a combination of the new methods will then be presented. 2. Traditional Nafion transducers In their current state, Nafion polymer transducers consist of Nafion membranes (typically Nafion-117) that have been plated with interpenetrating platinum electrodes using an “impregnation/reduction” technique [7–9]. This deposition is performed by first saturating the membrane with Pt(NH3 )4 2+ ions from a tetraammineplatinum chloride solution. The adsorbed ions are then reduced to platinum metal by soaking the membrane in a solution of sodium borohydride. This process results in a surface platinum electrode that penetrates into the membrane up to a depth of 20 m. Penetration of the electrode can be enhanced by repeating the impregnation/reduction process several times. In order to improve the conductivity of the platinum interpenetrating electrodes, a thin layer of gold is typically electroplated on the outer surface of the membrane. Following the metal deposition process, the mobile cation in the membrane can be exchanged for any suitable ion by soaking the membrane in a salt solution of that ion. This process is done in aqueous solutions with the membrane in a water-saturated state. Because the platinum electrodes made in this process penetrate into the polymer, the interfacial area between the polymer and the metal is very large. This large interfacial area is the reason for the very large capacitance of these devices (1–5 mF/cm2 ). Several researchers have shown that the strain generation in Nafion actuators is proportional to the membrane capacitance and it is now generally accepted that the actuation mechanism in these membranes is driven by charge motion within the polymer and accumulation at the surface [10–17]. Therefore, in order for an electrode to be effective, it must offer the property of a large polymer/metal interfacial area. A new method for fabricating electrodes having a larger polymer/metal interfacial area has been developed by Akle and Leo [6]. In this process, a metal powder is mixed with a solution of Nafion polymer in alcohols. This mixture is painted onto the Nafion membranes and hot-pressed into the surface. Transducers made in this way have been shown to generate strains up to three times those of their traditional platinum-plated counterparts and are faster and less expensive to make. This conducting powder painting method allows better control on electrode morphology and enhances transducer manufacturing repeatability. In order for the transduction to occur in Nafion membranes, the cations within the membrane must be mobilized by saturation with an appropriate solvent. Typically, the solvent used has been water. The reasons for this are the favorable interaction of the Nafion polymer with water, and the low viscosity of water. Nafion membranes will absorb up to 38% of their dry weight in water and achieve an ionic conductivity of 83 mS/cm when fully hydrated [18]. The water-swollen membranes make excellent electromechanical transducers, but suffer from the limitation that they exhibit a dramatic loss in performance when dehydrated. Efforts to contain the water inside the polymer have met with little success. Another, less obvious problem associated with the use of water as a solvent is the small electrochemical
stability window of water. The water will separate into gaseous hydrogen and oxygen at an applied voltage greater than 1.23 V (versus platinum) and therefore the actuation voltage applied to the membranes must not exceed this value. Transduction has been demonstrated when using other, more stable solvents such as glycerol and ethylene glycol, but the speed of the response is limited by the high viscosity of these solvents [19,20]. Recent work by Bennett and Leo has shown that ionic liquids can be used as solvents for Nafion transducers with the benefit that the ionic liquids are non-volatile and will not evaporate like water [4,5]. Transducers made using these ionic liquids have been shown to be stable for over 250,000 cycles in air, as compared to about 2000 cycles for water-solvated materials. The goal of the current work is to demonstrate that these two approaches can be combined and to compare the transducers made using this new method to traditional platinum-plated, water-swollen Nafion membrane transducers. Performance metrics including generated strain, electrical impedance, and sensitivity will be presented. 3. Ionic liquids as solvents In this work, we demonstrate a method for making ionic polymer transducers using highly stable room temperature ionic liquids. Ionic liquids are salts containing only charged species that exist in their liquid state at room temperature. They have an immeasurably low vapor pressure, electrochemical stability windows of 4 V or more, and are thermally stable to temperatures as high as 400 ◦ C. Furthermore, ionic liquids have high ionic conductivities and can be used as electrolytes for a variety of applications, including electrochemical capacitors [21] and conducting polymer actuators [22,23]. The reasoning behind using ionic liquids as solvents for Nafion transducers is that they will be very stable and therefore will eliminate the problem of solvent evaporation, thus allowing for the use of Nafion transducers in a broader range of environments. Also, the ionic liquids are themselves ionically conductive and should therefore facilitate ionic motion in the Nafion membrane, and may possibly enhance the ion motion as compared to water. Furthermore, the electrochemical stability window of most ionic liquids is larger than that of water and therefore larger actuation voltages will be possible, thus increasing the available energy density of these devices. In order to utilize an ionic liquid as a solvent for an ionic polymer transducer, a Nafion membrane is typically first plated with platinum and gold electrodes by using the traditional impregnation/reduction process. As mentioned, this process is carried out in aqueous solution with the polymer fully saturated with water. Following the plating process the membrane is dried in a vacuum oven and placed into an appropriate ionic liquid and heated. The diffusion of the ionic liquid into the Nafion membrane is dependant on the ionic liquid used, but is typically slow and therefore high temperature (>80 ◦ C) and long soak times (>4 h) are used. The ionic liquids 1-butyl-3-methylimidazolium tetrafluoroborate (BMI-BF4) and 1-butyl-3-methylimidazolium trifluoromethanesulfonate (BMI-Tf) have been imbibed into Nafion membranes by Doyle et al. [24] and they have reported
B.J. Akle et al. / Sensors and Actuators A 126 (2006) 173–181
that the uptake of these ionic liquids into the membrane is proportional to the temperature at which the swelling is carried out. Furthermore, they have shown that increasing the ionic liquid content of the membrane will result in higher ionic conductivity. In their work, the Nafion membranes were not plated with metal electrodes and they used temperatures as high as 180 ◦ C. In the current work, a problem with electrode stability has been observed at high uptake of the ionic liquid into the membrane. It has been reported by Gebel et al. [25] that water uptake by the Nafion-117 membrane results in a dimensional expansion of 14 and 10% along the length and width of the membrane. The metal electrodes are plated with the membrane in its fully water-saturated state. It has been found that when the membrane is dried and re-swollen by soaking it in the ionic liquid at elevated temperature, the electrodes often develop cracks, especially in the layers of electroplated gold, that result in a loss of surface conductivity of up to six orders of magnitude. This damage is due to the large strains imposed in the electrodes during the drying and re-swelling process. In order to overcome this problem, a modification to the platinum and gold plating process has been developed to improve the robustness of the electrodes. In this modification, the platinum and gold plating process is carried out as before, except that 25% (by volume) ethanol or methanol is added to each of the solutions to be used for the plating. Because the Nafion membrane will swell by 45% in ethanol and 51% in methanol [25], the use of these as co-solvents will increase the size of the membrane during the plating step. It has been found that this will allow for the membrane to be dried and re-swollen with the ionic liquid without damage to the electrodes and a corresponding loss in conductivity. Membranes plated in this way have been swollen with the 1-ethyl-3-methyimidazolium trifluoromethanesulfonate (EmI-Tf) ionic liquid and characterized versus their water-swollen counterparts (see Bennett and Leo [4,5]). 4. Direct assembly plating method In order to overcome the problem of dehydration and electrode instability, a new method for fabricating Nafion transducers has been developed that combines the use of ionic liquids and metal powder painted electrodes in the same device. In this method, a bare Nafion membrane is first exchanged from the proton counterion form into one of the alkali metal (e.g. Na+ , Li+ ) forms. This is to prevent charring of the membrane during the drying step, which has been witnessed when Nafion membranes in the acid form are heated for extended periods of time. After ion exchange, the membrane is dried in an oven at 150 ◦ C under vacuum for 12 h. The dry membrane is then immersed in neat EmI-Tf ionic liquid and heated to 150 ◦ C for 4.5 h. At this temperature, the uptake of ionic liquid is 60% by the dry weight of the membrane. In order to evaluate the effect of several of these parameters on the performance of the transducers, a detailed study of different electrode types has been carried out. This method allows the variation of several parameters including the type of the conducting powder. As mentioned earlier, the maximum strain will be generated when the capacitance of the membrane is maxi-
175
mized. For this reason, a ruthenium(IV) oxide (RuO2 ) powder was used in the electrode. This powder was obtained from Alfa Aesar and has a particle size of 3–5 nm and a specific area of 45–60 m2 /g. Capacitance in ionomeric materials is an electric double layer type capacitance, which increase proportional to the ionomer/conductor interfacial area. After the ionic liquid has been imbibed into the membrane, the electrodes are applied using the metal powder painting technique. A polymer/metal solution is prepared typically containing 47% (by weight) of 5% Nafion solution, 47% glycerol, and 6% metal powder. These concentrations are chosen based on previous experiences in building fuel cell electrodes. This solution is painted directly onto each surface of the ionic liquid-containing Nafion membrane. The glycerol and alcohol solvents (in the Nafion solution) are removed by baking the membrane at 130 ◦ C under vacuum for 15 min between each electrode layer; typically a total of four layers are applied to each side of the membrane. Following the electrode painting process, the membrane is dried at 130 ◦ C under vacuum for an additional 10 min to ensure complete removal of all volatile components from the composite. Based on the weight change of the membranes, the loading of ruthenium dioxide is estimated to be around 5–10 mg/cm2 of membrane area. After the ruthenium dioxide layers have been painted onto the surfaces of the membrane and dried, the sample is sandwiched between two 100 nm-thick conductive gold foils and the five layers of the composite are intimately bound together by a hot-pressing process. In the previous method, this hot-pressing was carried out at 20 MPa and 210 ◦ C for 2 min. However, the membranes in that case contained no solvent during the hot-pressing process. For the membranes that are swollen with the ionic liquid, this high pressure and long duration causes the device to melt to the point that the polymer flows. In order to maintain the dimensional stability of the membrane during the hot-pressing process while still binding the three layers together, the pressure and pressing time have been reduced. It has been found that good results are obtained for hot-pressing carried out at 4 MPa and 210 ◦ C for around 20–120 s depending on the metal loading of the electrode. The gold foils used were Falcon-Brand 24 K gold leafs; they helped decrease the surface resistance from greater than 100 k/cm for the painted metal layers to less than 1 /cm. However, the conductivity of the electrodes is also important to obtaining wellperforming transducers. The gold leaf is used to enhance the surface conductivity of the devices, but high electrode conductivity through the thickness of the electrodes is also important in order to achieve rapid charging and fast motion of the transducer. Although the RuO2 powder has a very high surface area, it is not very conductive (Bhaskar et al. report that the resistivity of thin RuO2 films is greater than 290 cm [26]). In order to assess the importance of surface area-to-volume ratio and particulate conductivity on performance, electrodes were prepared using a mixture of gold (Au) powder and ruthenium(IV) oxide powder. The gold powder is in the form of flakes of ∼3 m in length and has a specific area probably lower than 1 m2 /g. However, the resistivity of gold is 2.2 cm, two orders of magnitude lower than RuO2 . Therefore, two metal particles are used, one with a high area and low conductivity and one with a low area and
B.J. Akle et al. / Sensors and Actuators A 126 (2006) 173–181
176
Fig. 1. SEM images of (a) electrode built using the traditional impregnation/reduction method and (b) electrodes built using the direct assembly process.
high-conductivity. By varying the relative proportions of these metals, we can correlate transducer properties to electrode composition. Representative scanning electron microscopy (SEM) images of electrodes assembled using the traditional impregnation/reduction and the direct assembly process are shown in Fig. 1(a) and (b), respectively. 5. Experimental procedure The materials fabricated in this work are characterized for their performance as electromechanical transducers using a single test setup that measures two actuation properties; the strain (ε) and the electrical impedance. The samples used for these tests are 3 mm × 30 mm rectangular transducers held in a cantilevered configuration in a clamp with fixed gold electrodes (see Fig. 2). The free length of the transducer is 19 mm. The thickness of the transducer varies between approximately 0.18 and 0.21 mm depending on the number of electrode layers and the duration of the hot pressing. Representation of the test setup is shown in Fig. 2. For the first property, a random voltage with a RMS magnitude of 1 V is applied to the sample using a Fourier analyzer and amplifier. The displacement x(t) at any point along the length is measured with a non-contact laser vibrometer. The current i(t) induced in the polymer due to the applied voltage v(t) is also measured by measuring the voltage drop across a small (0.1 ) resistor—see Fig. 2. Using the Fourier analyser, the fre-
quency response function between the excitation voltage and the output displacement and between the input voltage and induced current are measured over the frequency range 0.2–200 Hz. This setup is also used to measure the time domain response of the transducer to a step voltage input. The free deflection δ(t) (in the time domain) or the free deflection frequency response (δ(ω)/v(ω)) are used with Eqs. (1) and (2), respectively, to obtain the generated strain as a function of time or frequency. In Eqs. (1) and (2), ε is strain, h is the thickness of the sample, and Lf is the free length of the sample. These equations assume that the transducer bends with a constant curvature; experimental results demonstrate that this is an appropriate assumption for the experiments performed in this paper. ε(t) =
δ(t)h L2f
ε(ω) (δ(ω)/v(ω))h = v(ω) L2f
(1) (2)
The impedance Z(ω) of the samples is computed using Eq. (3), and the capacitance is computed using Eq. (4) by assuming the polymer is a network of capacitors and a resistor in series [27]. Z(ω) =
v(ω) i(ω)
(3)
C(ω) =
1 (Im(Z(ω))ωA)
(4)
where A is the area of the transducer. 6. Experimental results
Fig. 2. Setup for the free deflection and impedance tests.
Experimental results are obtained for three sets of experiments. In each experiment, the electrode composition is varied and the time response and frequency response of a transducer is measured. The goal of the experiments is to correlate the
B.J. Akle et al. / Sensors and Actuators A 126 (2006) 173–181
177
Table 1 Weights of Au and RuO2 mixed with 0.5 g of 5% NafionTM solution and 0.5 g glycerol Sample 0% Au 100% RuO2 25% Au 75% RuO2 50% Au 50% RuO2 75% Au 25% RuO2 100% Au 0% RuO2 Fig. 3. Strain response as a function of electrode composition by weight.
electrode properties with the electromechanical performance of the transducer. 6.1. Variation of metal content by weight In the first experiment, the electrode composition is varied according to the weight of RuO2 or Au contained within the electrode. The weight percentage of metal in the electrode is varied from 0% Au to 100% Au, with the remaining percentage of the electrode metal being RuO2 . Fig. 3 is a plot of the frequency response magnitude for the five cases studied. The results clearly demonstrate that the RuO2 electrode is superior to Au in producing strain for an equivalent applied voltage. The strain-per-voltage response of the polymer at frequencies less than 10 Hz is maximized by the pure RuO2 electrode and is minimized for the pure Au electrode. Using 1 Hz as a reference point, we see that the strain response drops from 273 /V to 6 /V when the electrode composition is changed from 100% RuO2 to 100% Au. At intermediate values, there is a clear trend of decreasing strain response for increasing Au loading. The trends in the electrical impedance clearly indicate the tradeoff in capacitance and resistance associated with the addition of gold or ruthenium oxide. A pure RuO2 electrode produces a higher capacitance electrode due to its larger specific surface area. This results in a lower impedance at low frequencies (electrical impedance is inversely proportional to capacitance). However, the device with the pure RuO2 electrode has a larger impedance at frequencies above approximately 10 Hz because of the higher resistivity of RuO2 as compared to Au (see Fig. 4). The addition of small amounts of Au to the electrode decreases
Fig. 4. Electrical impedance as a function of electrode composition by weight.
(2.5:1) (2.5:1) (2.5:1) (2.5:1) (2.5:1)
Au (g)
RuO2 (g)
Thickness (m)
0 0.0428 0.0855 0.1283 0.1711
0.0625 0.0469 0.0313 0.0156 0
205 180 180 180 180
It also presents the durations of sample hot-pressing. The densities of Au and RuO2 are assumed to be 19300 and 7050 kg/m3 , respectively.
the high-frequency resistance but also decreases the low frequency capacitance because of the lower specific area of the gold powder as compared to the RuO2 powder. Capacitance at 0.5 Hz for samples with 0, 25, 75, and 100% gold is 3.9, 3.7, 3.5, 1.6, and 1.0 mF/cm2 , respectively. Therefore, the pure Au electrode has the lowest capacitance but also has superior high-frequency resistance properties. Akle et al. [27] have demonstrated a linear relationship between capacitance and the strain response in ionic polymer actuators. This relationship explains the largest low frequency strain per volt in the 0% Au sample (see Fig. 3), which has the largest capacitance, while the 100% Au has the smallest strain per volt and capacitance. 6.2. Variation of metal content by volume The second experiment consists of varying the electrode composition while maintaining a consistent volume of metal in the electrode. In this set of experiments, the relative percentage of RuO2 and Au is varied while the total volume of metal in the electrode layers is held constant at approximately 40%. Table 1 summarizes the compositions studied for this set of experiments. The strain response of the transducers as a function of volume percent of metal exhibited the same trend as the trend obtained when varying the weight percentage (see Fig. 5). The electrode fabricated with pure RuO2 exhibits the largest strain per unit voltage at frequencies below approximately 8 Hz. Increasing the percentage of Au in the electrode reduces the strain output. At 1 Hz, the strain output is 272 /V for an electrode with 100% RuO2 while it is 24 /V for an electrode with 100% Au. A clear relationship between the slope of the strain-to-voltage frequency response at frequencies below the first mechanical resonance and electrode composition also emerges in this study.
Fig. 5. Strain response as a function of electrode composition by volume.
178
B.J. Akle et al. / Sensors and Actuators A 126 (2006) 173–181
Fig. 8. Strain output to the application of square wave potentials of ±2 V. Fig. 6. Electrical impedance as a function of electrode composition by volume.
The negative slope of the strain-to-voltage frequency response indicates limited charge mobility. This is at least partially due to the high resistivity of the RuO2 particles, as increasing the percentage of gold in the electrode decreases the slope of the frequency response. The decrease in slope is correlated with the variation in the electrical impedance as the electrode composition varies (see Fig. 6). The device with a 100% RuO2 electrode has the highest low-frequency capacitance, but exhibits a highfrequency impedance of approximately 5 . The addition of small amounts of gold maintains the low-frequency capacitance but decreases the high-frequency impedance to approximately 30 . Further increase in the gold content reduces the capacitance and causes a corresponding decrease in the strain generation. Differences in the strain response due to variations in the electrical impedance are reduced by normalizing the strain response to the charge induced in the polymer by the applied potential. The strain per unit electric displacement is illustrated in Fig. 7 for the variation in electrode composition by volume. The results demonstrate that low-frequency slope of the strain-to-voltage response is eliminated and the strain per unit charge is approximately flat as a function of frequency. Furthermore, the strain per unit electric displacement is between 5 and 15 /C/m2 for all samples considered. The time response of the transducer displacement to square wave potential waveforms are also measured and the strain is computed using Eq. (2). Results for square wave amplitudes of ±2 V demonstrate that strain output of approximately 4000 (0.4%) is achievable with the pure RuO2 electrode (see Fig. 8). The steady-state value of the strain decreases as the gold loading
Fig. 7. Strain per unit charge/area for the samples produced by variation in the metal content with consistent volume.
of the electrode is increased. This is because the capacitance of the transducers decreases as the content of ruthenium oxide is decreased. However, the conductivity of the electrode increases as the loading of gold is increased, leading to a faster response. Although the speed of the response increases notably from the 100% RuO2 electrode to the 100% Au electrode, this trend is not as evident for the intermediate electrode compositions. This is illustrated by comparing the normalized initial slope of the two samples, which is the initial slope of the step divided by the maximum strain attained. The sample with 100% RuO2 electrodes had a normalized initial slope of 1 s−1 as compared to 6.7 s−1 for the 100% Au electrode sample. This result is consistent with the frequency-domain analysis, as the pure ruthenium oxide sample exhibits a strongly negative slope in the strain-to-voltage response whereas the pure gold sample exhibits a relatively flat response. 6.3. Electrode optimization The results of the electrode analysis demonstrate that a metal particulate with high surface area-to-volume ratio (RuO2 ) will produce a superior electrode due to the increase in capacitance at low-frequencies. Increasing the content of the high-conductivity particulate (Au) decreases the impedance at frequencies above approximately 10 Hz, but does not produce an increase in the strain response because of the corresponding drop in capacitance associated with replacing the RuO2 particles with the low specific area gold particles. Three samples were prepared for the optimization and verification study; the first sample had the same metal to polymer ratio in the electrode as in the weight and volume experiments (2.5:1). However, the thickness of the electrode for this sample was increased by 50% to six layers as compared to four layers for the previous experiments. For the second sample, the RuO2 to polymer ratio was increased by 50% to 3.75:1 while holding the number of layers constant at four. The third sample was prepared by replacing the 50% increase in RuO2 by the equivalent volume of Au. Table 2 summarizes the electrode compositions studied for this set of experiments. Therefore, the three samples maintained the same total metal volume loading. The metal in sample 1 is dispersed compared to the other two samples, while gold accounted for the increase of metal in sample 3. In samples 1 and 2, the surface area of the electrode is theoretically maintained, while sample 3 should have a better conductivity
B.J. Akle et al. / Sensors and Actuators A 126 (2006) 173–181
179
Table 2 Weights of Au and RuO2 mixed with 0.5 g of 5% NafionTM solution and 0.5 g glycerol for electrode optimisation Sample
Au (g)
RuO2 (g)
Thickness (m)
0% Au 100% RuO2 (2.5:1) 0% Au 100% RuO2 (3.75:1) 33% Au 66% RuO2 (3.75:1)
0 0 0.0855
0.0625 0.0938 0.0625
210 180 190
Fig. 11. per volt frequency response for three pure RuO2 samples.
at maximum deflection, Rmax . The strain at maximum deflection is equal to: t εmax = (3) 2Rmax
Fig. 9. Frequency response of the samples with tailored electrodes.
through the electrode. Fig. 9 shows the frequency response functions of the strain per unit volt for the three samples. The frequency response functions of samples 1 and 2 have nearly the same slope with sample 1 providing better performance. This is believed to be due to the lack of control on the amount of electrode ink deposited using the brush painting technique. Theoretically, the two samples should provide similar capacitance and therefore similar performance. As for sample 3, it is shown to have the least low frequency response while it crossed sample 2 almost at 3 Hz and provided better high-frequency response. This is consistent with the expectation of a better high-frequency response from a more conductive, gold-containing electrode. The motion of the transducer to an applied potential is also measured in the time domain. Square wave potentials are applied to the polymer and the resulting displacement is measured using a digital camera due to the fact that the displacements are too large to be measured with the laser vibrometer. The measured deflected shapes are shown in Fig. 10. The pictures illustrate that the assumption of constant curvature is valid over the range of conditions tested. The range of motion increases as the amplitude of the square wave input is increased. The peak-to-peak strain in the sample is measured by determining the radius of curvature of the sample
The radius of curvature was determined on both sides of the neutral position and the peak-to-peak strain is the addition of the two values. The measured strains increase from 8195 (0.82%) at ±2 V to 24,380 (2.44%) at ±4 V. It is interesting to note that the increase in peak-to-peak strain is not linear as a function of input amplitude. The asymmetry in the deflection, which is especially significant in the ±3 V sample, is due to the pre-bend in some of the samples. Once again, we attribute this to non-linearity in the charge conduction as a function of voltage. 7. Method characterization In order to gauge the repeatability of the process described in this paper, three identical transducers are separately built and fully characterized using the methods described previously. The transducers were fabricated by painting four layers of 100% RuO2 ink with a weight ratio of metal to polymer of 2.5 to 1. As can be seen in Fig. 11, very little variation is observed in the per unit volt frequency response. Although more results are needed in order to obtain a standard deviation, this initial result shows that the new process is able to produce repeatable results. In order to determine the long-term stability of the transducers’ operation, a reliability test is also performed on these three identical samples. The three samples were actuated continuously with a 1 Hz sine wave potential and the free strain generated in each sample was measured periodically. The effect of the
Fig. 10. Deflected shapes as a function of amplitude for a square wave potential input.
180
B.J. Akle et al. / Sensors and Actuators A 126 (2006) 173–181
Acknowledgements This work was supported by the U.S. Army Research Laboratory and U.S. Army Research Office under contract/grant number DAAD19-02-1-0275 Macromolecular Architecture for Performance (MAP) MURI. Supplementary funding was provided by the Virginia Space Grant Consortium. References
Fig. 12. Reliability test on the three samples with ±1, ±2, and ±3 V 1 Hz sine wave applied on samples 1, 2, and 3, respectively.
peak amplitude of the applied potential on the long-term stability of the transducers is illustrated in Fig. 12. As can be seen, the samples that were actuated with a 1 and 2 V peak potential did not exhibit and degradation of the response after more than 250,000 cycles. It should be noted that the maximum generated strain in the samples is lower than in Fig. 10 because of the time scales involved. In order to achieve the large strains presented in Fig. 10, an input frequency of less than 0.04 Hz is required. This is due to the slow response speed of the devices fabricated with pure electrodes. However, the result demonstrates the improved long-term stability of ionic polymer transducers that utilize ionic liquids instead of water as a solvent. By contrast, Bennett and Leo [4] have shown that a water-swollen ionic polymer transducer would operate for about 2000 cycles in air. After 3600 cycles, the motion decreased by 96% due to solvent evaporation. Although the specific reason for the decrease in performance of the sample that was driven at ±3 V is unknown, a likely explanation is degradation of the ionic liquid. Cooper and Sullivan [28] and Bonhote et al. [29] have shown that the electrochemical stability window of the EMI-Tf ionic liquid is 4.1 V.
8. Conclusions Tailoring the electrode composition is an effective method for achieving high-strain transducers using ionomeric polymer materials. Our results demonstrate that surface area-to-volume ratio of the metal particulate is a critical property in achieving large capacitance at low frequencies. Increasing the low-frequency capacitance increases the strain output of the transducer. The electrical conductivity through the electrode is also proved to be critical to the speed of response of the transducers. A tradeoff between high speed for gold particles and large strains for ruthenium dioxide particles is observed. Results presented in this work yielded an ionomer-ionic liquid transducer that produced greater than 2% ε at peak-to-peak voltages of 6 V. Finally, fabricating air stable ionic polymer transducers with enhanced strain and strain rate will create a new range of application such as control and positioning actuators for conformal structures.
[1] K. Sadeghipour, R. Salomon, S. Neogi, Development of a novel electrochemically active membrane and ‘Smart’ material based vibration sensor/damper, Smart Mater. Struct. 1 (1992) 172–179. [2] K. Oguro, Y. Kawami, H. Takenaka, Bending of an ion-conducting polymer film-electrode composite by an electric stimulus at low-voltage, J. Micromach. Soc. 5 (1992) 27–30. [3] M. Shahinpoor, Y. Bar-Cohen, J. Simpson, J. Smith, Ionic polymermetal composites (IPMCs) as biomimetic sensors, actuators and artificial muscles—a review, Smart Mater. Struct. 7 (1998) R15-R30. [4] M. Bennett, D. Leo, Ionic liquids as solvents for ionic polymer transducers, Sens. Actuators A: Phys. 115, 79–90. [5] M. Bennett, D. Leo, Ionic liquids as novel solvents for ionic polymer transducers, in: Proceedings of the EAP Actuators and Devices. SPIE, paper 5385-27, 2004. [6] B. Akle, K. Wiles, D. Leo, J. McGrath, Effects of electrode morphology on the performance of BPSH and PATS ionic polymer transducers, in: Proceedings of the EAP Actuators and Devices. SPIE, paper 5385-73, 2004. [7] P. Millet, R. Durand, E. Dartyge, G. Tourillon, A. Fontaine, Precipitation of metallic platinum into Nafion ionomer membranes, J. Electrochem. Soc. 140 (1993) 1373–1379. [8] K. Kim, M. Shahinpoor, The synthesis of nano-scaled platinum particles (NSPP)—their role in performance improvement of ionic polymer-metal composite (IPMC) artificial muscles, in: Proceedings of the EAP Actuators and Devices. SPIE, vol. 4329, 2001, pp. 189–198. [9] T. Rashid, M. Shahinpoor. Force optimization of ionic polymermic platinum composite artificial muscles by means of orthogonal array manufacturing method, in: Proceedings of the EAP Actuators and Devices. SPIE, vol. 3669, 1999, pp. 289–298. [10] P. de Gennes, K. Okumura, M. Shahinpoor, K.J. Kim, Mechanoelectric effects in ionic gels, Europhys. Lett. 50 (2000) 513–518. [11] K.M. Newbury, D.J. Leo, Electromechanical modeling and characterization of ionic polymer benders, J. Intell. Mater. Syst. Struct. 13 (2002) 51–60. [12] K.M. Newbury, D.J. Leo, Linear electromechanical model of ionic polymer transducers. Part I. Model development, J. Intell. Mater. Syst. Struct. 14 (2003) 333–342. [13] K.M. Newbury, D.J. Leo, Linear electromechanical model of ionic polymer transducers. Part II. Experimental validation, J. Intell. Mater. Syst. Struct. 14 (2003) 343–358. [14] K. Farinholt, D.J. Leo, Modeling of electromechanical charge sensing in ionic polymer transducers, Mech. Mater. 36 421–433. [15] S. Nemat-Nasser, Micro-mechanics of actuation of ionic polymer-metal composites, J. Appl. Phys. 92 (2002) 2899–2915. [16] S. Nemat-Nasser, J. Li, Electromechanical response of ionic polymermetal composites, J. Appl. Phys. 87 (2000) 3321–3331. [17] K. Asaka, N. Fujiwara, K. Oguro, K. Onishi, S. Sewa, State of water and ionic conductivity of solid polymer electrolyte membranes in relation to polymer actuators, J. Electromech. Chem. 505 (2001) 24– 32. [18] DuPont, Nafion membranes NE-112, NE-1135, N-115, N-117. Product information, 2000. [19] S. Nemat-Nasser, Micromechanics of actuation of ionic polymer-metal composites, J. Appl. Phys. 92 (2002) 2899–2915. [20] S. Nemat-Nasser, S. Zamani, Experimental study of nafion and flemionbased ionic polymer-metal composites (IPMCs) with ethylene glycol as
B.J. Akle et al. / Sensors and Actuators A 126 (2006) 173–181
[21]
[22]
[23]
[24]
solvent, in: Proceedings of the EAP Actuators and Devices. SPIE, vol. 5051, 2003. pp. 233–244. A.B. McEwen, H.N. Ngo, K. LeCompte, J.L. Goldman, Electrochemical properties of imidazolium salt electrolytes for electrochemical capacitor applications, J. Electrochem. Soc. 146 (1999) 1687–1695. W. Lu, A.G. Fadeev, B. Qi, E. Smela, B.R. Mattes, J. Ding, G.M. Spinks, J. Mazurkiewicz, D. Zhou, G.G. Wallace, D.R. MacFarlane, S.A. Forsyth, M. Forsyth, Use of ionic liquids for -conjugated polymer electrochemical devices, Science 297 (2002) 983–987. F. Vidal, C. Plesse, D. Teyssi´e, C. Chevrot, Long-life air working conducting semi-Ipn/ionic liquid based actuator, Synth. Metals 142 (2004) 287–291. M. Doyle, S. Choi, G. Proulx, High-temperature proton conducting membranes based on perfluorinated ionomer membrane-ionic liquid composites, J. Electrochem. Soc. 147 (2000) 34–37.
181
[25] G. Gebel, P. Aldebert, M. Pineri, Swelling study of perfluorosulphonated ionomer membranes, Polymer 34 (1993) 333–339. [26] S. Bhaskar, P.S. Dobal, M. Majumder, R.S. Katiyar, Characterization of conductive RuO2 thin films as bottom electrodes for ferroelectric thin films, in: Proceedings of the Ferroelectric Thin Films IX, Materials Research Society, vol. 655, paper CC4.7, 2001. [27] B. Akle, M. Hickner, D. Leo, J. McGrath, Correlation of capacitance and actuation in ionomeric polymer transducers, accepted in the J. Mater. Sci., 2005. [28] E.I. Cooper, E.J. S O’Sullivan, New, Stable, Ambient-Temperature Molten Salts, In Proceedings of the Eighth Interational Symposium on Molten Salts, The Electrochemical Society, 1992, pp. 386–396. [29] P. Bonhˆote, A. Paula Dias, N. Papageorgiou, K. Kalyanasundaram, M. Gr¨atzel, Hydrophobic, highly conductive ambient-temperature molten salts, Inorg. Chem. 35 (1996) 1168–1178.