Experimental Studies of Ion Transport in PPy(DBS) - The Journal of

Jeffrey S. HarrisonDean A. WaldowPhillip A. CoxRajiv GiridharagopalMarisa ... Hyacinthe Randriamahazaka , Thomas Bonnotte , Vincent Noël , Pascal Mar...
0 downloads 0 Views 1MB Size
J. Phys. Chem. C 2009, 113, 369–381

369

Experimental Studies of Ion Transport in PPy(DBS) Xuezheng Wang and Elisabeth Smela* Department of Mechanical Engineering, UniVersity of Maryland, College Park, Maryland 20742 ReceiVed: October 14, 2008

Lateral ion transport in conjugated polymer films is studied using a special experimental geometry in which the top surface of the film is covered by a transparent ion barrier. Because of the barrier, when the oxidation level of the polymer is switched electrochemically, charge-compensating ions can only enter and leave the polymer from the edges. Since conjugated polymers are electrochromic, the color of the film changes during switching, and this can be monitored to provide information on the oxidation level of different parts of the film. Since the oxidation level cannot change until the cations arrive, the color also directly maps the positions of the cations. This geometry was employed to study cation transport in polypyrrole doped with dodecylbenzenesulfonate, PPy(DBS). Upon reduction, the ions travel in a front from the edges to the center of the film. During the first-ever reduction, this cation front stays sharp, but in subsequent reduction scans the front moves 20-30 times faster and broadens as it moves. The higher the applied voltage, the faster the front moves, with a linear dependence. The velocity of the front also strongly depends on the initial oxidation level of the polymer. During oxidation, on the other hand, the entire film gradually darkens, with no front and no dependence of the switching speed on the applied potential. 1. Introduction This paper characterizes lateral ion transport in polypyrrole doped with dodecylbenzenesulfonate, PPy(DBS), using a sample geometry that consists of a thin conjugated polymer film sandwiched between an electrode on one face and a transparent ion barrier on the opposite face. Ions are able to enter and exit the polymer during electrochemical cycling only at the edges of the film, at the electrolyte interface. However, holes are able to enter and exit from one entire face, at the metal interface. This geometry ensures that ion transport is the rate-limiting step in the electrochemical reactions, since the path for ions is much longer than the path for holes, and it also allows one to track ion transport inside the film through electrochromic color changes that can be seen through the transparent ion barrier. (Cation transport is rate-limiting in the following sense: although the electrochemical reaction is energetically possible under the applied potential, it cannot proceed in a given region of the polymer until the cations arrive or depart from there. However, it should be borne in mind that the state of the polymer matrix controls how rapidly the cations can move through it. Freevolume generation by electrochemical stimulation of chain conformational movements1,2 therefore plays an indirect role in the results since higher overpotentials provide more energy to drive conformational changes, thereby opening the matrix and increasing ion mobility/diffusivity. The ion mobility and diffusivity thus depend on the local oxidation level of the polymer as well as the applied voltage. This is further discussed in the Supporting Information.) In prior work, we showed that during reduction cations enter the PPy(DBS) in a front, moving from the edges to the center,3,4 as seen by a color change from dark red to transparent. The associated volume change is linearly correlated with the color change and occurs simultaneously. In this work, the color change, which also corresponds to the ion concentration as shown in ref 4, is studied as a function * Corresponding author. Tel.: 301-405-5265. Fax: 301-314-9477. E-mail: [email protected].

of time during both reduction and oxidation in the first cycle and in subsequent cycles and as a function of applied potential. The findings have important implications for controlling actuator speed. We show that during reduction the ion front moves with a velocity that is linearly proportional to the overpotential (until it reaches a limiting value) and that the front position advances with a time dependence between t and t, depending on the sample. The first-ever reduction step is 30 times slower than in later cycles, and in later cycles, the front is broader. If the film is switched from a partially reduced state to the fully reduced state, then the front velocity increases considerably, with the velocity related to the charge that was consumed in reaching the partially reduced state. When the PPy(DBS) is oxidized, on the other hand, there is no front. Instead, the entire film changes color all at once, although this change is faster at the edges. Furthermore, the oxidation speed is insensitive to the applied potential. 1.1. Motivation for the Ion Transport Studies. This work is motivated by the need to better understand and predict actuator movement. Mathematical models that thoroughly describe material behavior during actuation, i.e., the constitutive equations, have not yet been formulated and validated; although some aspects are now understood, such as polymer chain relaxation,1,2,5 other aspects, such as the role of electric fields, are not yet understood. As a result, rate-limiting processes have not been clearly identified, and trade-offs between strain, stress, and speed have not been mapped. The experimental work presented in this paper is thus part of a longer-term effort to develop models of the coupled chemical, electrical, and mechanical processes that occur during reduction and oxidation (redox) of conjugated polymers (Figure 1). To build predictive models, experiments need to be designed in which only a single effect is dominant, so that it can be isolated and characterized without confounding, and unknown, contributions from the other effects. For example, Otero et al. (see for example refs 1, 2, and 5) have carried out a thorough series of experiments in which polymer chain conformational

10.1021/jp809092d CCC: $40.75  2009 American Chemical Society Published on Web 12/11/2008

370 J. Phys. Chem. C, Vol. 113, No. 1, 2009

Wang and Smela

Figure 1. Given electrical input signal resulting in an electrochemical state with particular mechanical, chemical, and electrical properties, which in turn result in particular actuator metrics. Changing the oxidation level requires inter-related charge and mass transport, as well as polymer chain conformational changes; several of these interrelations are indicated. The final state also depends on the deposition and cycling conditions.

changes were the rate-limiting step in the electrochemical reaction and could thus be characterized. Conformational relaxation in PPy(DBS) was studied in ref 6. In this paper, ion transport is similarly isolated and examined in a comprehensive series of experiments. (That is not to say that in any given set of experiments the other effects do not occur or contribute to the results indirectly, but that the experiments are designed to elucidate the role of the variable in question without being confounded by the other effects.) We have focused on ion transport since that is directly responsible for actuation strain.7-9 1.2. Background. We have previously used PPy(DBS) in microactuators10-13 and a number of basic studies8,9,14-17 and continue to work with it here because it is important to understand one system well to build good, predictive models based on the dominant physical/chemical/electrochemical effects. PPy(DBS) is a cation-transporting material in which the DBS- counteranions are immobile.7,18,19 When a sufficiently negative electrochemical potential is applied to the polymer, the positive charge carriers on the polymer chains (“holes”) are removed (electrons are added, the polymer is reduced). Since the negatively charged DBS- counterions are immobile, hydrated cations enter the polymer to maintain charge neutrality. (This process is described in more detail in the Discussion section below.) Likewise, a sufficiently positive potential adds the holes back onto the polymer chains (the polymer is oxidized), and the cations exit. The oxidation and reduction processes are not symmetric, since the oxidized state is electronically conducting while the reduced state is resistive; this affects the electric fields within the polymer. In addition, in the reduced state PPy(DBS) is swollen with ions and solvent, which disrupts hydrogen bonding and π-π stacking between chains, while in the oxidized state, the polymer matrix is more compact and contains less water, and the chain segments are straighter and stiffer (Figure 2).20 As we show below, ion transport rates are much higher in the open matrix. 1.3. Experimental Configuration. To develop constitutive equations for charge transport during redox, each of the terms in the transport equations needs to be identified and the form of the coefficients determined, without confounding influence from other processes that occur simultaneously. The device configuration used here (Figure 1 in ref 4) ensures that ion transport is much faster than electron/hole transport. However,

Figure 2. Schematic representation of PPy(DBS) undergoing reduction. In the oxidized state (red), the polymer is more compact and contains less water than in the reduced state (yellow).

as shown in prior work by Otero et al.,1,2,5 ion transport is strongly affected by the state of compaction of the polymer matrix. Therefore, the experimental protocols were designed to ensure that the initial state of the polymer was always the same. In other experiments, the effect of the initial state was explicitly examined by comparing the behavior during the first cycle vs later cycles and by examining the effect of changing the initial oxidation level. The polymer matrix is also affected by the applied reduction potential since higher overpotentials provide more energy for chain conformational changes. These electrochemically stimulated conformational relaxation (ESCR) effects cannot be so readily separated from the effects of the electric field on ion transport. Fortunately, in PPy(DBS), the ESCR effects have been measured to be small, affecting only ∼20% of the reduction charge.6 One can also make use of differences in voltage dependences in interpreting the results. The experimental configuration used here can be compared to those of Kaneto et al.21 and Tezuka et al.,22-25 who made an electrical connection to one end of a strip of conjugated polymer, the face of which was exposed to the electrolyte. In that configuration, however, the path of the holes was long compared to that for the ions, the opposite of our configuration. In ref 24, a film of PPy(ClO4), an anion-conductor, was used. When the polymer was switched from the reduced to the oxidized state, the oxidized state grew outward from the electrical connection at constant velocity. There was a clear boundary between the two states, which could be observed by the color and which was tracked using an array of photodiodes. Speeds increased with potential. Upon reduction,25 the film changed color simultaneously over the whole area, stopping at a doping level

Experimental Studies of Ion Transport in PPy(DBS)

J. Phys. Chem. C, Vol. 113, No. 1, 2009 371

Figure 3. (a) Illustration of the 10th normalized intensity level, used to determine the front velocity, and the 5th and 15th levels, used to determine the front width. (b) Front position vs time. The slope of the linear part of the curve was used to determine the velocity.

of ∼13%, when the percolation threshold was crossed, leaving electrically disconnected islands of charge in the film. In these experiments, the hole and ion transport speeds may have been comparable, so it is not clear which of the two was rate-limiting. Similar experiments have also been reported by Ingana¨s et al. with anion-transporting poly(3,4-ethylenedioxythiophene), PEDOT,26 and poly(3-hexylthiophene), P3HT.27 2. Methods The experimental methods were detailed in ref 4, and the reader is referred to that paper. In brief, the samples were electrochemically cycled in 0.1 M aqueous NaDBS, and color changes were recorded from directly overhead through a microscope. (The fabrication steps for creating the ion-barrier covered PPy(DBS) stripes were shown not to affect the behavior of the conjugated polymer.) Front positions were obtained by tracking the color intensity across the film width vs time, i.e., from profiles such as the one shown in Figure 4 of ref 4. The intensity of the oxidized state was used as a baseline. The intensity difference between the fully oxidized and fully reduced states was arbitrarily divided into 20 levels, and the position of three of these levels was tracked, as shown in Figure 3a. The 10th level defined the front positions. The front velocity was determined by tracking the position of the 10th level versus time and then taking a linear fit to the center portion of this position-time curve (Figure 3b). To obtain the center portion, early-stage data before fronts were formed (such as between 0 and 0.4 s in Figure 3b), and approximately the first 10% of the reduction time (e.g., between 0.5 and 1 s) was disregarded. Data from late in the process, after the two fronts reached the center and started interacting (between 3.5 and 4 s in the figure), were also disregarded. The remaining data were fit to a straight line using least-squares regression. The slope of this line was taken as the velocity of the front. The width of the front was defined as the distance between the 5th and 15th levels. Two factors limited the spatial resolution of the data analysis: (1) the number of pixels in the images and (2) the resolution of the data read into Matlab, which determines the number of intensity levels. The camcorder created video with a resolution of 640 × 480 pixels. The fronts propagated along the direction of 480 pixels, and the sample covered approximately half of the picture, so approximately 200 pixels were used in tracking the front positions. Considering that the sample width was 300 µm, the minimum spatial resolution was 1.5 µm. The second

factor is the 256 intensity levels. If two pixels had the same intensity, the tracking program chose one of them, but the real position could have been the average of the two. Such cases occurred when the front was close to the fully reduced area, where the intensity plateaus. Also, when the pixel did not have exactly the same intensity as the threshold, the program chose the pixel with the closest intensity, and this occurred frequently. Therefore, the tracking method was only accurate to (1 pixel ((3 µm). Errors of this magnitude are small for tracking the front position, since that changes from 0 to 150 µm, creating an uncertainty of 2%. However, these errors are large in front broadening analyses: for a front broadening rate of 5 pixels/s, an increase by a factor of 1.5 from 4 to 6 pixels is within the noise. 3. Results: Reduction We first examine the ion ingress process in detail. In the experiments in this section, the potential was stepped from 0 V (vs Ag/AgCl), unless otherwise noted, to various reducing potentials. The oxidizing potential had been held for at least 30 s so that in every reduction step the Na+ cations encountered the same compacted matrix, in which the chains had undergone conformational relaxation to eliminate free volume. The work required to open the matrix was therefore essentially the same in every case, eliminating the initial conformational state as a variable. Referring to Figure 1, the experiments allowed us to probe mass transport into the polymer with minimal confounding effects from electron transport and chain movement. (As shown in section 3.5, when the initial state of the matrix is changed, the effects of chain conformation can be seen to be substantial.) We begin by focusing on cation transport during reduction. This section starts with an overview of the color changes during the first reduction cycle (3.1.1) and then during the steady-state behavior of later reduction cycles (3.2.2). The front position versus time (3.2) and the velocity of the front as a function of reduction potential (3.3) are examined, and then the broadening of the front (3.4). These results are used in other publications for model building.28 This is followed by an examination of the effect of the initial oxidation potential (3.5). In section 4, we move to ion transport during oxidation. The oxidation behavior is shown to be completely different than the reduction behavior (4.1, 4.2). The effect of the oxidation potential is determined (4.3), and then broadening is examined (4.4). 3.1. Overview of Color Change During Reduction. It is well-known that the first reduction scan of a conjugated polymer

372 J. Phys. Chem. C, Vol. 113, No. 1, 2009

Figure 4. Overhead images taken during the first reduction potential step from 0 to -1 V vs Ag/AgCl at (a) 0 s, (b) 60 s, (c) 120 s, and (d) 180 s. (The PPy was 300 nm thick, and the SU8 2 µm thick.) A video image of this process, speeded up 10×, is available in AVI format in the Supporting Information: jp809092d_si_002.avi.

Figure 5. Intensity profiles at 30 s intervals during the first-ever reduction at -1 V vs Ag/AgCl. The arrow indicates the direction of front movement over time.

is anomalous and that it can take several cycles to “break in” the polymer before it starts to show steady state behavior. This is due to a high degree of compaction of the chains (see, e.g., ref 29 and references therein) and because a considerable amount of water enters the polymer during the first reduction scan that does not readily leave in subsequent scans.30,31 We therefore start with an examination of the first-cycle behavior and in the next section show the steady-state behavior. To ensure the same initial state (and thus hold polymer matrix effects constant), a fresh film was used for each scan. 3.1.1. First Cycle. Overhead images of a section of the PPy(DBS) stripe during the first reduction scan are shown at 60 s intervals in Figure 4. Ions traveled in a front from both edges of the stripe to the center, and the fronts stayed sharp and parallel to the edges. The intensity profiles showed nearly vertical steps (Figure 5). Front velocities were on the order of 1 µm/s. (This velocity was independent of the amount of time that the sample spent immersed in the electrolyte prior to electrochemical reduction: velocities were the same for samples immersed for 10 min and 24 h.) 3.1.2. Later Cycles. Later reduction steps were recorded after at least three cycles to ensure steady-state behavior (and thus the same initial polymer chain conformational state, although

Wang and Smela not the same final state). For V < -0.75 V (Figure 6c and d), fronts were again observed, but they moved much more rapidly than during the first scan (see below). They remained parallel to the film edges but were no longer sharp, broadening over time. At potentials less negative than V ) -0.75 V, no fronts were discernible to the eye; instead the films gradually lightened from the edges inward (Figure 6a and b). The higher ion mobility in the film in later cycles, compared to the first cycle, has consistently been observed (see, for example, refs 30 and 32-35). This has been attributed to the uptake of solvent that subsequently remains within the film, as well as to only slowly reversible chain conformational changes. Volume change measurements have also revealed irreversible swelling during the first cycle.4,17 Intensity profiles at several time intervals are shown in Figure 7 for the two later-cycle reduction cases in Figure 6. The curves at low overpotential η, where we here define η as the magnitude of the difference between the applied voltage and the onset of the peak in the cyclic voltammogram, η ) |V - V0|, were smooth and u-shaped (Figure 7a), lifting up from the center of the stripe. Profiles at potentials more negative than -0.8 V (Figure 7b) were stepped, although the step walls were slanted because the fronts were not perfectly sharp. Over time, the slopes of the steps decreased as the fronts became increasingly diffuse. Behind the fronts were regions with lower slope extending right to the edges of the film. The absolute intensity values in Figure 5 and Figure 7 differ, but the absolute numbers on the y-axis are not meaningful since they depend on the lighting conditions. The only meaningful parameter is the normalized difference between the fully oxidized state (i.e., the lowest intensity value) and the fully reduced state (highest intensity value). 3.2. Front Position vs Time. Modeling predicts that the front velocity V should be proportional to t under these experimental conditions.28 This is not due to classical Fickian diffusion, but rather occurs due to the drop in potential caused by the conducting-to-insulating transition during reduction. (Diffusion is movement of species down a concentration gradient, whereas drift (also known as migration) is movement under a force, such as that of a charged particle in an electric field; this is discussed in ref 28.) However, under other conditions, the model shows that fronts can propagate with a linear dependence on t. To experimentally characterize the front position vs time, we again start with an examination of the first-cycle behavior and in the next section compare this with the steady-state behavior. 3.2.1. First Cycle. The position of the front (defined as being at the 10th intensity level, and corresponding to approximately 50% doping, as indicated by the vertical position of the arrow in Figure 5) during the first-ever reduction of the film is shown versus time in Figure 8. The position goes from 150 (at the edge) to 0 (the center). The slopes of these curves, i.e., the front velocities V, varied from sample to sample, ranging from V ∼ t to V ∼ t, with Figure 8 illustrating a typical case intermediate between these two limits. (Cases at both extremes are shown in the Supporting Information.) The reasons for the variations are not entirely clear, but sample preparation conditions play a role (see the Supporting Information). During the first cycle, there is no rapid decrease in intensity as the fronts from either edge approach each other (as there is during later cycles; see Figure 9) since the fronts remain sharp throughout the first reduction. 3.2.2. Later Cycles. The front position versus time during later reduction cycles is shown for two potentials in Figure 9. The two primary differences from Figure 8 are the shorter time

Experimental Studies of Ion Transport in PPy(DBS)

Figure 6. Fronts during later reduction steps to -0.7 V (vs Ag/AgCl) at (a) 10 s and (b) 22 s and to -1.5 V at (c) 0.3 s and (d) 1.3 s (PPy 300 nm, SU8 2 µm). Videos of these processes are available in AVI format in the Supporting Information: jp809092d_si_004.avi and jp809092d_si_003.avi.

it takes for the fronts to reach the center and the apparent increase in velocity when the fronts meet since they are no longer perfectly sharp. The general shapes of the curves did not change with potential, but the (negative) slopes increased. Just as for the first reduction cycle, the curves from different samples had time dependences between t and t, and Figure 9 shows an intermediate example. (See the Supporting Information for further information.) 3.3. Effect of Reduction Potential. When the ions move with a front, they are not transported by classical Fickian diffusion36 (a phase front can, however, result during diffusion if the diffusion coefficient is concentration-dependent; this is one form of Case II diffusion). To ascertain whether they are moving by migration, the dependence of the ion velocity V on the reduction potential V was measured. 3.3.1. First Cycle. The velocity as a function of potential during the first reduction is shown in Figure 10. Each point represents a separate sample (since the first cycle can only be run once). The PPy(DBS) did not reduce at all at potentials less negative than -0.6 V. Upon switching to potentials e -0.7 V, a front was launched. There was a linear relationship between the velocity of the front and the applied voltage (V ∼ 3.25η). This is as would be expected if the ions were moving under

J. Phys. Chem. C, Vol. 113, No. 1, 2009 373

Figure 8. Front position vs time during the first-ever reduction step (-1.0 V). The inset shows the position vs the square root of time. Dashed lines are linear fits to the data between 17 and 180 s.

Figure 9. Front position vs time during later reduction steps from 0 V to -1.0 and -2.0 V (vs Ag/AgCl). The inset shows the same data versus the square root of time.

drift (migration) with a field-independent mobility µ, in which case V ) µE, where E is the electric field (which increases linearly with the applied potential). (If the front velocity were controlled by ESCR effects, one would instead expect an eη dependence.) The cyclic voltammogram of an uncoVered film of the same thickness during the first-ever reduction scan is shown in Figure 10 for reference. The intercept to zero velocity occurred near the potential V0 corresponding to the onset of the

Figure 7. Intensity profiles during later-cycle reductions corresponding to the images in Figure 6 at (a) -0.7 V, with curves separated by 6 s intervals, and (b) at -1.5 V with 1 s intervals. Arrows indicate the direction of change. In (b) at 2.3 s, the fronts from the two edges have met.

374 J. Phys. Chem. C, Vol. 113, No. 1, 2009

Figure 10. Effect of applied potential on velocity (points) in the first reduction cycle. The dashed line is a linear fit to the data (y ) -2.0 3.25x, R ) 0.976). A cyclic voltammogram (solid line) from the first cycle of a film without an ion barrier is shown for reference (scan rate 20 mV/s).

reduction current in the cyclic voltammogram. This is consistent with an interpretation that once the potential is sufficiently cathodic to reduce the film the ions enter the film and travel through it by migration. Although the front velocity increases with potential, it is important to bear in mind that in practical applications device speed cannot always be increased simply by applying large overpotentials. In aqueous systems, for example, hydrogen evolution and delamination occur at high voltages. 3.3.2. Later Cycles. Front velocities vs reduction potential during later cycles are shown in Figure 11. The data points represent multiple reductions of 7 samples. Between -0.8 and -1.6 V, the velocity V increased linearly with applied potential V (V ∼ 62η), but for more negative potentials, the velocity saturated at between 70 and 80 µm/s. This shows that the ratelimiting process changed, perhaps from cation transport in the polymer to another process, not yet identified (but not ESCR, since that would result in exponentially increasing speed with overpotential). (Recall that in this geometry the rate-limiting step is expected to be ion transport. However, if another process becomes rate-limiting, then the experiments no longer provide unambiguous information on ion transport.) As was the case for the first-ever reduction cycle, the velocity extrapolated to zero near the onset of the reduction process.

Wang and Smela

Figure 12. Front and back of the ion front versus time upon reduction (-1 V vs Ag/AgCl). The arrow indicates the width of the front, and the numbers the intensity levels.

The difference between the first and later cycles is therefore only in the slope, which is related to the mobility. The slope was -3.3 for the first cycle and -62, or 20 times higher, in later cycles. This is consistent with the more open, hydrated configuration of the polymer matrix after the first reduction. The reproducibility of the results can be judged from Figure 11. Differences in velocity for the various samples were primarily caused by small differences in the PPy and SU8 thicknesses.37 For comparison, in PEDOT samples with the Kaneto/Tezuka geometry (an uncovered film contacted electrically on one side), the reaction front moved at a rapid 10 cm/s (10 000 times faster than in our experiments) upon oxidation (anion ingress), suggesting that in this experiment electron transport was ratelimiting.26 In P3HT,27 the velocity of the fronts, higher in the center of the film than at the edges, was proportional to t-γ, where γ was less than 0.5 and depended on the applied potential. (Fickian diffusion will result in a front velocity proportional to t-0.5, while migration under a constant electrical field results in a front velocity proportional to t0, i.e., a constant velocity.) 3.4. Front Broadening. There was no broadening during the first reduction step, but during steady-state reduction steps the front got wider as it traveled to the center. The positions of the 5th and 15th levels, at the front and back of the ion front, respectively, are shown vs time in Figure 12 for a sample

Figure 11. (a) Front velocities during later cycles vs reduction potential. A later-cycle CV of a film without an ion barrier is shown for reference. The line shows a linear fit to the data between -0.8 and -1.6 V. (b) Data shown over a wider potential range. Different samples are indicated by different symbols. The range in (a) is indicated by the dashed vertical line.

Experimental Studies of Ion Transport in PPy(DBS)

Figure 13. Front width as a function of time during electrochemical reduction at four different voltages (400 nm PPy, 2 µm SU8).

reduced at -1 V after oxidation at 0 V. The two lines are superimposed at the edge of the film but diverge as the front propagates to the center. If the front broadening was due to classical Fickian diffusion, the broadening would go as t and would not depend on the applied potential. Figure 13 shows the time dependence of the front width for several potentials. These data came from scans on the same sample. (Additional data are presented in the Supporting Information.) The broadening did not depend on V, but the front width increased linearly with time in the vast majority of experiments. This inconsistency with Fickian diffusion is not unexpected, given the change in the state of the polymer during reduction (Figure 2). Ion mobility has consistently been shown to depend on oxidation level,1,2,5,29,38-48 as have the polymer’s mechanical properties.49,50 As a result, diffusion is in general not Fickian, although it can be Fickian if small enough potential steps are applied. 3.5. Effect of Initial Oxidation Potential. In this section, we describe experiments designed to ascertain the role of the state of the matrix on the ion mobility. This variable had been held constant in the sections above by starting with the polymer always in the fully oxidized state at 0 V. The ion velocity and front sharpness during the first reduction cycle compared to later cycles, described above, showed that diffusivity and mobility in PPy(DBS) depend strongly on the compaction of the polymer. Thus, a series of experiments was conducted in which reduction (during steady state cycling) was initiated from different oxidation potentials to vary the state of the film at the onset of switching to the fully reduced state. So whereas in prior sections the applied potential was stepped from 0 V to various negative voltages, in this section the applied potential was instead stepped from various initial voltages to the same final voltage of -1.1 V. This changed the initial state of the polymer while keeping the same final, fully reduced state. The relationship between front velocity and voltage provides qualitative insight into the dependence of the ion mobility on the doping level, which is linked to the ion concentration, C (see ref 4). To find the quantitative dependence, however, requires knowing the electric field E throughout the material, which is found in ref 28 and subsequent modeling papers. One must also take into consideration the fact that the initial potential varies not only the hydration and mechanical state (i.e., factors such as chain packing and stiffness) of the polymer matrix but also its electrical conductivity, σe. To obtain the electric field correctly therefore requires a knowledge of the dependence of

J. Phys. Chem. C, Vol. 113, No. 1, 2009 375

Figure 14. Effect of initial oxidation potential on front velocity (points). Different samples are represented by different symbols. The same symbols are used for multiple data points taken from the same sample. A CV from an uncovered PPy(DBS) film is shown for reference (gray line), as is the cathodic Gaussian 1 peak (dashed line, see ref 4).

the conductivity on the doping level, σe(C), which changes by at least 5 orders of magnitude between the conducting and insulating states. Figure 14 shows the front velocity plotted as a function of the initial potential, which was held for 30 s prior to switching to -1.1 V to ensure that the film had uniformly reached the given oxidation level. (The experiments were done on four samples, with initial potentials chosen in the order of increasing cathodic voltage, from +0.4 down to -0.63 V, although individual samples did not span the entire range, and each sample was taken over its particular range of initial voltages at least twice, giving multiple points at each initial potential for each sample.) This was verified by monitoring the stabilization of the color of the film prior to reduction. Between -0.4 and -0.65 V, the front velocity shot up almost an order of magnitude. This potential range corresponds to the onset of the reduction peak in the cyclic voltammogram and thus the onset of (solvated) cation entry into the film and a more open polymer matrix. This result is not unexpected, given the previous observations on first versus later cycles. At initial potentials more negative than -0.65 V, no fronts were discernible, and color variations could not be tracked because the polymer was already essentially fully reduced. The velocity data consistently showed a small peak at approximately +0.1 V. This is also where the peak for anion transport occurs in electrolytes with mobile anions, but it is not clear why this small increase in velocity is seen in the NaDBS electrolyte. The front velocity data below -0.4 V can be fit to an exponential with y ∼ e9x to e12x, where x is the overpotential η. This is a very strong dependence of velocity on the state of the polymer (ion content, hydration, chain conformation, etc.), and it is clear that the asymmetry sketched in Figure 2 should be considered in any modeling. The dependence of velocity on voltage was converted to a dependence of velocity on charge density by integrating the current in a cyclic voltammogram to obtain a charge-voltage curve (Figure 15 inset). (Using instead the charge consumed during voltage steps to various potentials yielded similar curves.) The dependence of velocity on the consumed reduction charge, Q, is approximately V ∼ e2Q beyond the first 2 mC/cm2 (Figure 15a). The dependence on the number of ions in the film,

376 J. Phys. Chem. C, Vol. 113, No. 1, 2009

Wang and Smela

Figure 15. Dependence of front velocity on consumed charge. Samples are represented by the same symbols as in Figure 14. (a) Velocity versus reduction charge density Q (Qtot ) -5.5 mC/cm2). The inset shows the Q vs V curve used to create this plot. The gray line is an exponential curve y ) 30 + 0.006e(2-2Q), where 30 is the velocity in the fully oxidized state. (b) The data for one of the samples versus QG1 (QG1,tot ) -1.0 mC/cm2), the charge in the Gaussian 1 peak, with a linear fit y ) 48 - 419x (R ) 0.949), and the charge density vs potential curve used for creating this plot.

Figure 16. Overhead images of color change during oxidation, stepping from -1.1 to 0 V, at (a) t ) 0 s, (b) 2 s, and (c) 4 s (300 nm thick PPy, 2 µm thick SU8). A video image of this process is available in AVI format in the Supporting Information: jp809092d_si_002.avi.

Figure 17. Red channel intensity profiles of the images in Figure 16 and also of intermediate times. The arrows labeled “1” indicate changes at early times and “2” subsequent changes.

estimated by the charge under Gaussian 1 (see ref 4; this is the charge under the more cathodic pair of peaks in the cyclic voltammogram), is shown in Figure 15b, and this relationship is linear. These relationships are used in ref 28 and subsequent papers for model building. 4. Results: Oxidation We now examine the oxidation process, in which the cations are expelled and the film goes from insulating (and transparent) back to conducting (red). Since the polymer matrix starts in

the fully expanded state and is switched to the fully compacted state, polymer chain conformational changes play less of a role. Because the electrical, chemical, and mechanical states of the polymer are different at the two end points, reversing the voltage does not result in a simple reversal of the cation front. Instead, the behavior is substantially different upon oxidation than upon reduction. 4.1. Overview of Color Change During Oxidation. The color changes undergone by the PPy(DBS) during the course of oxidation upon stepping from -1.1 to 0 V are shown in a series of images in Figure 16 and as intensity profiles in Figure 17. The first thing to note is that color change was greatest and fastest at the edges. This is unsurprising since cations nearest the electrolyte interface are the first that can leave. Unlike during reduction, however, the color at the center of the film also changed rapidly. (See the arrows labeled “1”.) We postulate that this is due to the high diffusivity/mobility of the Na+ in the reduced state (Figure 14, Figure 15). The average intensity change over the whole film width was thus initially more rapid than during reduction for the same overpotential. The second thing to note is that although diffuse frontlike intensity changes were seen in the earliest stages, these areas broadened rapidly to more closely resemble the behavior seen at low voltages during reduction (compare Figure 6c and d). Note that the color change direction was again from the edges inward, although the intensity dropped rather than rose (compare Figure 7a). After the initial intensity drop in the center of the film, further changes took increasingly longer, as evident from the increasingly closer line spacing in Figure 17.

Experimental Studies of Ion Transport in PPy(DBS)

J. Phys. Chem. C, Vol. 113, No. 1, 2009 377

Figure 18. Positions vs time of the 10th intensity level (solid line, 50% doped) and the 5th and 15th levels (dotted lines) upon oxidation from -1.1 to 0 V. (Same sample as in Figure 16 and Figure 17.)

Figure 20. Oxidation front broadening at different oxidation potentials (400 nm PPy, 2 µm SU8).

The rapid initial color change occurring over the whole film bears a resemblance to the reduction of anion-transporting films with the Kaneto or Tezuka geometries.25 In that system, the phenomenon was attributed to the film starting in the conducting state, so that the electrical signal had control authority over the entire film. In our case, however, the film started in the highresistivity state at t ) 0. We examine the reasons for the behavior in Figure 7 and Figure 17 using modeling in ref 28 and subsequent papers. 4.2. Position vs Time. Even though there was no actual front propagation during oxidation, since the entire film was affected more or less immediately upon application of the potential step, oxidation level positions were determined using the same method as previously to allow comparison with the behavior during reduction. The positions of the 5th, 10th, and 15th color levels versus time are shown in Figure 18. The shapes of the curves are almost identical to the ones for reduction (Figure 9) but are upside-down. However, the positions moved as the square root of time (see inset) until the oxidation levels from either edge met (at that point, the lines go almost straight up.). However, the broadening was considerably greater than during reduction, as seen by comparing the 5th and 15th levels (approximately 25% and 75% doping, respectively) to those in Figure 12. The 15th levels from either edge met each other and disappeared very rapidly ( -0.75 V, there are no film (Figure 16, Figure 17) -0.6 V; no reduction takes fronts: reduction occurs over the place at less cathodic potentials entire film (Figure 6, Figure 7) (Figure 4, Figure 5) varies from V ∼ t to V ∼ t (Figure 8, Figure 9) goes as V ∼ t (Figure 18) front velocity V ) 1 µm/s at -1.1 front velocity V ) 30 µm/s at NA V (Figure 10) -1.1 V; it increases if the scan is begun with the film in a partially reduced state: V ∼ e2Q where Q is the total consumed reduction charge and V ∼ QG1, where QG1 is the charge in the most cathodic reduction peak (Figure 11, Figure 14, Figure 15) oxidation speed is independent of front velocity V ∼ V, going to front velocity is linear with potential; oxidation is faster than zero at V0 (Figure 11a); slope of voltage, V ∼ V, going to zero at reduction for V > -0.9 V, but V vs V is 20 times higher than V0 (Figure 10) reduction is faster if V < -0.9 for the first reduction, and (Figure 19) velocities saturate at ∼70 µm/s for potentials below -1.6 V (Figure 11b) fronts broaden linearly with time, and broadening does not depend on no broadening: fronts remain voltage (Figure 13, Figure 20) sharp throughout reduction (Figure 4)

indicating that the cation mobility is greater by that factor. The mobility is increased even further if the polymer is stepped from a partially reduced state, which has an even more open structure, rather than from the fully oxidized state. For small overpotentials, however, the color lightens over the whole film with a u-shaped profile, indicating that the process is dominated by diffusion. The broadening of the ion fronts is linear with time but does not depend on potential, consistent with non-Fickian diffusion. The oxidation and reduction processes are not symmetric in two very significant respects. First, upon reduction, the polymer starts as a conductor and becomes an insulator, and upon oxidation it goes from an insulator to a conductor, so the potential drops seen by the ions are different. In addition, in the oxidized state, the polymer matrix is more compact and contains less water, so ion mobility is significantly lower than in the reduced state. These differences are reflected in the front behavior and are the subject of the modeling work in ref 28 and later papers. Referring back to Figure 1 and Figure 23, what additional experimental research is needed to understand the redox process? (Let us put aside, for the moment, structure-property relationships that link the deposition conditions to the material properties and these properties to the behavior. These issues still require a substantial amount of experimental study.) Additional work is required on the role of the electrolyte. The relationship between the solvent type and content and the polymer’s electrical and mechanical properties must be determined. Also, the solvent plays a role in determining whether anions vs cations are transported,55,56 and how fast,20 and this is not yet adequately understood. Further, the effective volume of the solvated ions in the polymer must be known to predict the magnitude of the actuation. Lastly, for working in electrolytes with a mixture of ions, such as biofluids, the factors that determine the amount of compensation by different cations (or anions) and by anions versus cations needs to be better understood.

Acknowledgment. We would like to acknowledge funding through DuPont’s Young Professor Grant and the Laboratory for Physical Science. Supporting Information Available: Includes additional experimental results on reduction, oxidation, and the effects of temperature, ion barrier thickness, and PPy thickness. This material is available free of charge via the Internet at http:// pubs.acs.org. References and Notes (1) Otero, T. F.; Grande, H.-J.; Rodriguez, J. Reinterpretation of polypyrrole electrochemistry after consideration of conformational relaxation processes. J. Phys. Chem. B 1997, 101, 3688–3697. (2) Otero, T. F.; Boyano, I. Comparative study of conducting polymers by the ESCR model. J. Phys. Chem. B 2003, 107 (28), 6730–6738. (3) Wang, X.; Shapiro, B.; Smela, E. Visualizing ion transport in conjugated polymers. AdV. Mater. 2004, 16 (18), 1605–1609. (4) Wang, X.; Smela, E. Color and volume change in PPy(DBS) J. Phys. Chem. C 2009, 113, 359-368. (5) Otero, T. F.; Grande, H.; Rodrı´guez, J. Role of conformational relaxation on the voltammetric behavior of polypyrrole. Experiments and mathematical model. J. Phys. Chem. B 1997, 101, 8525–8533. (6) West, B. J.; Otero, T. F.; Shapiro, B.; Smela, E. Chronoamperometric study of conformational relaxation in PPy(DBS) J. Phys. Chem. B 2008, submitted. (7) Pei, Q.; Ingana¨s, O. Electrochemical applications of the bending beam method; a novel way to study ion transport in electroactive polymers. Solid State Ionics 1993, 60, 161–166. (8) Shimoda, S.; Smela, E. The effect of pH on polymerization and volume change in PPy(DBS). Electrochim. Acta 1998, 44 (2-3), 219–238. (9) Wang, X.; Smela, E. Cycling conjugated polymers with different cations; SPIE 13th Annual Int’l. Symposium on Smart Structures and Materials, EAPAD, San Diego, CA, 2006. (10) Smela, E.; Ingana¨s, O.; Lundstro¨m, I. Controlled folding of micrometer-size structures. Science 1995, 268 (23 June), 1735–1738. (11) Smela, E. Microfabrication of PPy microactuators and other conjugated polymer devices. J. Micromech. Microeng. 1999, 9, 1–18. (12) Smela, E.; Kallenbach, M.; Holdenried, J. Electrochemically driven polypyrrole bilayers for moving and positioning bulk micromachined silicon plates. J. Microelectromech. Syst. 1999, 8 (4), 373–383. (13) Smela, E. A microfabricated movable electrochromic “pixel” based on polypyrrole. AdV. Mater. 1999, 11 (16), 1343–1345.

Experimental Studies of Ion Transport in PPy(DBS) (14) Maw, S.; Smela, E.; Yoshida, K.; Sommer-Larsen, P.; Stein, R. B. The effects of varying deposition current density on bending behaviour in PPy(DBS)-actuated bending beams. Sens. Actuators, A 2001, 89, 175–184. (15) Wang, X.; Smela, E. Fast switching of conjugated polymer films; SPIE 13th Annual Int’l. Symposium on Smart Structures and Materials, EAPAD, San Diego, CA, 2006. (16) Christophersen, M.; Shapiro, B.; Smela, E. Characterization and modeling of PPy bilayer microactuators. Part 1: Curvature. Sens. Actuators, B 2006, 115, 596–609. (17) Smela, E.; Gadegaard, N. Volume change in polypyrrole studied by atomic force microscopy. J. Phys. Chem. B 2001, 105 (39), 9395–9405. (18) De Paoli, M. A.; Peres, R. C. D.; Panero, S.; Scrosati, B. Properties of electrochemically synthesized polymer electrodes-X. Study of polypyrrole-dodecylbenzene sulfonate. Electrochim. Acta 1992, 37 (7), 1173–1182. (19) Matencio, T.; De Paoli, M.-A.; Peres, R. C. D.; Torresi, R. M.; Cordoba de Torresi, S. I. Ionic exchanges in dodecylbenzenesulfonate doped polypyrrole Part 1. Optical beam deflection studies. Synth. Met. 1995, 72 (1), 59–64. (20) Mao, H.; Pickup, P. G. Ion transport in a polypyrrole-based ionexchange polymer. J. Phys. Chem. 1989, 93 (17), 6480–6485. (21) Kaneto, K.; Ura, S.; Yoshino, K.; Inuishi, Y. Optical and electrical properties of electrochemically doped n- and p-type polythiophenes. Jpn. J. Appl. Phys. 1984, 23 (3), L189–191. (22) Tezuka, Y.; Aoki, K. Direct demonstration of the propagation theory of a conductive zone in a polypyrrole film by observing temporal and spatial variations of potentials at addressable microband array electrodes. J. Electroanal. Chem. 1989, 273 (1-2), 161–168. (23) Tezuka, Y.; Aoki, K.; Ishii, T. Alternation of conducting zone from propagation-control to diffusion-control at polythiophene films by solvent substitution. Electrochim. Acta 1999, 44, 1871–1877. (24) Tezuka, Y.; Ohyama, S.; Ishii, T.; Aoki, K. Observation of propagation speed of conductive front in electrochemical doping proces of polypyrrole films. Bull. Chem. Soc. Jpn. 1991, 64, 2045–2051. (25) Tezuka, Y.; Kimura, T.; Ishii, T.; Aoki, K. Concentration distribution of conducting species with time resolution in electrochemical undoping process at the polypyrrole-film-coated electrode in the light of electric percolation. J. Electroanal. Chem. 1995, 395, 51–55. (26) Carlberg, J. C.; Inganas, O. Fast optical spectroscopy of the electrochemical doping of poly(3,4-ethylenedioxythiophene). J. Electrochem. Soc. 1998, 145 (11), 3810–3814. (27) Johansson, T.; Persson, N. K.; Inganas, O. Moving redox fronts in conjugated polymers studies from lateral electrochemistry in polythiophenes. J. Electrochem. Soc. 2004, 151 (4), E119-E124. (28) Wang, X.; Shapiro, B.; Smela, E. Ion transport in conjugated polymers: Model development J. Phys. Chem. B 2008, submitted. (29) Otero, T. F.; Padilla, J. Anodic shrinking and compaction of polypyrrole blend: electrochemical reduction under conformational relaxation kinetic control. J. Electroanal. Chem. 2004, 561 (1), 167–171. (30) Yang, H.; Kwak, J. Mass transport investigated with the electrochemical and electrogravimetric impedance techniques 1. Water transport in PPy/CuPTS films. J. Phys. Chem. B 1997, 101 (5), 774–781. (31) Baker, C. K.; Qiu, Y.-J.; Reynolds, J. R. Electrochemically induced charge and mass transport in polypyrrole/poly(styrenesulfonate) molecular composites. J. Phys. Chem. 1991, 95, 4446–4452. (32) Gandhi, M. R.; Murray, P.; Spinks, G. M.; Wallace, G. G. Mechanism of electromechanical actuation in polypyrrole. Synth. Met. 1995, 73 (3), 247–256. (33) Duffitt, G. L.; Pickup, P. G. Enhanced ionic conductivity of polypyrrole due to incorporation of excess electrolyte during potential cycling. J. Chem. Soc., Faraday Trans. 1992, 88 (10), 1417–1423. (34) Li, Y. On the large overpotential of the first reduction of polypyrrole perchlorate films in organic solutions. Electrochim. Acta 1997, 42 (2), 203– 210.

J. Phys. Chem. C, Vol. 113, No. 1, 2009 381 (35) Lizarraga, L.; Andrade, E. M.; Molina, F. V. Swelling and volume changes of polyaniline upon redox switching. J. Electroanal. Chem. 2004, 561 (1), 127–135. (36) Lacroix, J. C.; Fraoua, K.; Lacaze, P. C. Moving front phenomena in the switching of conductive polymers. J. Electroanal. Chem. 1998, 444 (1), 83–93. (37) Wang, X.; Smela, E.; Shapiro, B. Understanding ion transport in conjugated polymers; SPIE 12th Annual Int’l. Symposium on Smart Structures and Materials, EAPAD, San Diego, CA, 2005. (38) Otero, T. F.; Angulo, E. Oxidation-reduction of polypyrrole filmskinetics, structural model and applications. Solid State Ionics 1993, 63-5, 803–809. (39) Otero, T. F.; Grande, H.; Rodriguez, J. A new model for electrochemical oxidation of polypyrrole under conformational relaxation control. J. Electroanal. Chem. 1995, 394, 211. (40) Otero, T. F.; Grande, H.; Rodriguez, J. Conformational relaxation during polypyrrole oxidation: from experiment to theory. Electrochim. Acta 1996, 41 (11/12), 1863–1869. (41) Otero, T. F.; Grande, H.; Rodriguez, J. An electromechanical model for the electrochemical oxidation of conducting polymers. Synth. Met. 1996, 76, 293. (42) Otero, T. F.; Grande, H.; Rodriguez, J. Electrochemical oxidation of polypyrrole under conformational relaxation control. Electrochemical relaxation model. Synth. Met. 1996, 76, 285. (43) Otero, T. F.; Grande, H. Thermally enhanced conformational relaxation during electrochemical oxidation of polypyrrole. Electroanal. Chem. 1996, 414, 171. (44) Otero, T. F. In ConductiVe Polymers: Transport, Photophysics, and Applications; Nalwa, H. S., Ed.; John Wiley & Sons: New York, 1997; Vol. 4, pp 517-594. (45) Otero, T. F.; Grande, H.; Rodriguez, J. A conformational relaxation approach to polypyrrole voltammetry. Synth. Met. 1997, 85, 1077. (46) Otero, T. F.; Bengoechea, M. UV-visible spectroelectrochemistry of conducting polymers. Energy linked to conformational changes. Langmuir 1999, 15, 1323–1327. (47) Otero, T. F.; Boyano, I. Potentiostatic oxidation of polyaniline under conformational relaxation control: experimental and theoretical study. J. Phys. Chem. B 2003, 107 (18), 4269–4276. (48) West, K.; Jacobsen, T.; Zachau-Christiansen, B.; Careem, M. A.; Skaarup, S. Electrochemical synthesis of polypyrrole: influence of current density on structure. Synth. Met. 1993, 55-57 (2-3), 1412–1417. (49) Murray, P.; Spinks, G. M.; Wallace, G. G.; Burford, R. P. In-situ mechanical properties of tosylate doped (pTS) polypyrrole. Synth. Met. 1997, 84 (1-3), 847–848. (50) Smela, E.; Lu, W.; Mattes, B. R. Polyaniline Actuators, Part 1: PANI(AMPS) in HCl. Synth. Met. 2005, 151 (1), 25–42. (51) Streetman, B. G.; Banerjee, S. Solid State Electronic DeVices, 5th ed.; Prentice-Hall: Upper Saddle River, 2000. (52) Lin, X.; Li, J.; Smela, E.; Yip, S. Polaron-induced conformation change in single polypyrrole chain: an intrinsic actuation mechanism. Int. J. Quantum Chem. 2005, 102, 980–985. (53) Aime, J. P.; Bargain, F.; Schott, M.; Eckhardt, H.; Miller, G. G.; Elsenbaumer, R. L. Structural study of doped and undoped polythiophene in solution by small-angle neutron scattering. Phys. ReV. Lett. 1989, 65 (1), 55–58. (54) Bay, L.; Jacobsen, T.; Skaarup, S.; West, K. Mechanism of actuation in conducting polymers: osmotic expansion. J. Phys. Chem. B 2001, 105 (36), 8492–8497. (55) Vidanapathirana, K. P.; Careem, M. A.; Skaarup, S.; West, K. Ion movement in polypyrrole/dodecylbenzenesulphonate films in aqueous and non-aqueous electrolytes. Solid State Ionics 2002, 154-155, 331–335. (56) Skaarup, S.; Bay, L.; Vidanapathirana, K.; Thybo, S.; Tofte, P.; West, K. Simultaneous anion and cation mobility in polypyrrole. Solid State Ionics 2003, 159 (1-2), 143–147.

JP809092D