Abstract
Electrochemical impedance spectroscopy (EIS) has been applied to the study of electrodeposited thin films of manganese dioxide as used in electrochemical capacitors. Conventionally EIS employs a relatively small AC excitation signal to provide valuable characterization on electrode features such as the series resistance, charge transfer and double layer charge storage processes, as well as mass transport. The small excitation signal is used so as to focus on the processes occurring within that potential domain allowing for considerable resolution across the full potential window. In this work we have compared the output from this conventional analysis with data from the application of a large amplitude AC excitation signal; i.e., an AC signal that spans the full potential window of the manganese dioxide electrode. Not only does this allow access to electrochemical data representative of the full range of domains within the manganese dioxide structure, it also facilitates performance analysis (determination of specific power and energy data) of the electrode in a much more efficient manner than conventional means, as well as enables separation of the total specific capacitance into its non-faradaic and faradaic components.
Export citation and abstract BibTeX RIS
Electrochemical capacitors
Energy storage is of critical concern to modern-day and future society. It enables portability for the myriad of electronic devices available to consumers, as well as being of considerable significance in the transportation sector, with the growing prevalence of electrified vehicles, and the grid (large scale) energy storage sector, as a means of load leveling. Electrochemical energy storage technologies represent an efficient and light weight approach to accomplishing the energy storage and delivery demands of modern consumers.
Electrochemical energy storage and conversion devices include batteries, electrochemical capacitors and fuel cells, the relative performance of each being described by a Ragone diagram,1 although other considerations such as cyclability, cost, safety and environmental impact also play a significant role in determining device viability.2 Electrochemical capacitors are a rapidly developing technology, with their performance being typified by favorable characteristics such high power densities and extended cyclability,1,3,4 and also less desirable characteristics such as low energy density and high cost per Wh.1,3,4 Recent developments in various electrochemical capacitor systems (e.g.,)5,6 have given some insight into what can be achieved by these systems, and so it is only a matter of time before these aforementioned less desirable characteristics are amended.
Electrochemical characterization
Central to the development of advanced electrochemical capacitor systems is the ability to characterize both material and system capabilities electrochemically. Many methods can be used to evaluate the outright performance characteristics of an electrochemical capacitor, but the most common are cyclic voltammetry and constant current charge-discharge cycling, both over the course of an extended number of cycles.6 Whilst these are by far the most common electrochemical characterization methods, others such as electrochemical impedance spectroscopy (EIS) (e.g.,),7 floating potential measurements (e.g.,),8 and electrochemical quartz crystal microbalance studies (e.g.,)9 have also been applied.
Of these other methods, the typical EIS experiment provides the researcher with extensive information on features of the system such as the series resistance, charge transfer resistance, any grain boundary impedances, double layer capacitance, and any mass transport characteristics.10 Additionally, when such EIS experiments are coupled with potentiostatic DC control, outputs from the EIS experiment can be mapped across the full potential window being employed.11 Central to the application of EIS is the premise that the small AC excitation potential used (typically less than 10 mV) enables reversible system behavior. This is particularly important in, for example, battery systems where the specific electrode behaviors of interest can be focused on in the narrow potential window, rather than being masked by overlapping competitive processes, as is the case with voltametric studies. Additionally, the behavior of battery systems is typified by structural and material changes that can limit cyclability over an extended period.12 However, for an electrochemical capacitor system, where cyclability over a relatively large potential window is known to be very good,1,3,4 there is the potential in an EIS experiment to increase the magnitude of the AC excitation signal to encompass a larger potential window with the goal of exploring how this type of experiment affects the electrochemical material and system performance.
This work
Here we apply large amplitude EIS to explore the technique and probe the behavior of a neutral aqueous manganese dioxide electrode system used for electrochemical capacitors. What we will demonstrate is that this approach can be used to examine not only fundamental material behavior, such as conventionally determined with EIS, but also performance characteristics of the electrode, such as obtained conventionally through the use of cyclic voltammetry or constant current charge discharge cycling, that ultimately leads to a Ragone diagram representation of material performance.
Experimental
Materials and electrode preparation
An electrodeposited manganese dioxide electrode was used to demonstrate the concepts in this work. A platinum disk electrode (geometric area = 0.785 cm2) mounted on the end of an epoxy shaft was firstly immersed into a solution of acidified H2O2 (0.1 M H2SO4 + 5% H2O2) to chemically clean the substrate. After removal, the platinum electrode was rinsed with Milli-Q ultra pure water (resistivity >18.2 MΩ.cm) before being polished by lightly rubbing on a cloth pad coated with moist 0.1 μm alumina powder for ∼3 min. After this the electrode was again rinsed thoroughly with Milli-Q water, patted dry with a lint-free tissue before being transferred to the electrodeposition cell (100 mL, three-necked flask). Also added to the cell were a saturated calomel reference electrode (SCE) and a carbon rod counter electrode (area ∼4.7 cm2). The electrolyte in the cell was aqueous 0.01 M MnSO4 + 0.1 M H2SO4, prepared from AR Grade MnSO4.H2O (Sigma Aldrich; >99.5%) and concentrated H2SO4 (Sigma Aldrich; 98%). Anodic electrodeposition was carried out using chronoamperometry, as this approach has been shown to produce manganese dioxide deposits with excellent electrochemical capacitor performance.13 In this case the potential of the platinum working electrode was stepped from the open circuit potential to 1.3 V vs SCE where it was held for 30 s. The manganese dioxide coated platinum electrode was then removed from the electrodeposition bath, again washed thoroughly with Milli-Q water, before being patted dry with a lint-free tissue. Note that the manganese dioxide produced here was not dried completely, and was used for electrochemical cycling as an electrochemical capacitor electrode almost immediately.
Electrochemical characterization
Electrochemical impedance spectroscopy (EIS) experiments were carried out using a Solartron 1287 Electrochemical Interface coupled with a Solartron 1245 Gain Phase Analyzer controlled using ZPlot software. The slew rate of the electrochemical interface is 107 V/s which is sufficiently fast for the experiments conducted here. Cyclic voltammetry and constant current charge-discharge cycling of the electrode materials were carried out using a Perkin-Elmer VMP 16 channel potentiostat/galvanostat controlled by EC-Lab software Version 6.80. The electrochemical cell was essentially the same as used for electrodeposition, except now the electrolyte being used was a solution of 0.1 M Na2SO4. The same saturated calomel electrode and carbon rod electrode were used as reference and counter electrodes, respectively.
Previous reports on the study of manganese dioxide electrodes for electrochemical capacitor applications have used a 0.8 V potential window (0.0–0.8 V vs SCE) to study the electrochemical performance of their active materials.13 In this study we therefore set the DC potential of the EIS experiment to 0.4 V vs SCE; i.e., the mid-point of the potential window, for a period of 15 min prior to starting. A series of EIS experiments were then conducted around this mid-point with progressively increasing amplitudes of the applied AC potential (0.01, 0.02, 0.05, 0.10, 0.20 and 0.40 V), over the frequency range from 20 kHz to 0.1 Hz, with ten frequency measurements per decade being employed. Of course the largest amplitude EIS experiment corresponds to a full potential window sweep; i.e., 0.8 V. In the set-up of the EIS experiments it was also specified that at every frequency considered, five full cycles would be applied to measure the impedance. This was done to ensure that there was no net charge applied to the electrode.
For the purposes of establishing a baseline for comparison, cyclic voltammetry and constant current charge-discharge cycling experiments were also conducted using the 0.0–0.8 V potential window. Cyclic voltammetry was conducted using scan rates of 10, 20, 50, 100, 250, 500, 750 and 1000 mV/s, while constant current cycling was carried out using currents of 8.47, 4.25, 1.74, 0.91, 0.12 and 0.09 mA, which equates to specific current densities of 5500, 2759, 1132, 594, 77 and 61 A/g. As with the EIS measurements, five full cycles were used to evaluate performance.
Results and Discussion
Electrochemical data
The synthesis, properties and behavior of the electrodeposited thin film of manganese dioxide prepared in this study have been examined extensively prior to this report, and so the reader is referred to these references for specific details.13–17 One feature of the chronoamperometric electrodeposition that is of key importance for this work is the mass of electrodeposited manganese dioxide.16 The mass was determined by integration of the current-potential data, and in this case was determined to be 1.54 μg (1.96 μg/cm2).
The EIS data collected in this study is shown in Figure 1, where it is clear that there are substantial differences in behavior depending on the amplitude of the AC excitation potential. Qualitatively these differences reflect changes in charge transfer processes at the electrode-electrolyte interface, changes in the double layer capacitance, and also changes in the mass transport phenomena present. These variations are due to changes in the electrochemical processes occurring at the electrode as a result of the applied AC potential accessing compositionally and structurally different domains with the manganese dioxide electrode.18,19 In essence, while small amplitude signals allow us to focus on very specific processes, the response from large amplitude signals essentially represents average electrode behavior over the entire potential window. In effect this is similar to measuring a cyclic voltammogram on the electrode over the same potential window.
Data from the cyclic voltammetry and constant current charge-discharge cycling analysis of the electrode is shown in Figure 2. In the case of cyclic voltammetry, the data presented here is compatible with that expected from a pseudo-capacitive electrode; i.e., a 'box'-like voltammogram over the potential range considered, with little evidence of an ohmic series resistance. Similarly, the constant current charge-discharge data exhibits as expected results, also with little ohmic resistance apparent even at high rates. Overall, all the electrochemical data collected is consistent with our expectations and previous literature data.
Equivalent circuit modeling
As is quite often done, an equivalent circuit has been developed and fit to the EIS data using complex non-linear least squares (CNLS) regression. The equivalent circuit used here is a modified Randles circuit, as shown in Figure 3. The equivalent circuit consists of:
- (i)A series resistance (Rs) which takes into account the resistivities of the electrolyte and electrode and their physical separation. This characteristic of the system is commonly interpreted as the high frequency intercept of the EIS data with the real (Z') axis.
- (ii)A charge transfer resistance (Rct) reflecting the faradaic processes the manganese dioxide electrode undergoes in 0.1 M Na2SO4. This has been reported to consist of concerted cation (from the electrolyte) and electron (from the conductive network) insertion into the manganese dioxide structure at intermediate frequencies; i.e., where in this case M represents either a H+ or Na+ ions.
- (iii)A constant phase element (CPE1) in parallel with Rct reflecting the double layer capacitance at the electrode/electrolyte interface. A constant phase element was used here, rather than an ideal capacitance, because the manganese dioxide surface is macroscopically and microscopically rough and porous, and as such it is anticipated that the double layer structure will be heterogeneous.20 The impedance of a constant phase element (ZCPE) is given by:21 where j is the complex number (), ω is the angular frequency, and K and P are the fitting parameters. The parameter K essentially represents the capacitance, while P (0 ⩽ P ⩽ 1) represents the divergence from a uniform capacitor (P = 1). For a rough solid-electrolyte interface P is typically ∼0.9.
- (iv)Another constant phase element (CPE2) in series with Rct to account for the mass transport features of the system at low frequencies. As before, a constant phase element was used here to account for heterogeneous electrode behavior, such as is encountered with a rough interface, and porous electrode behavior.20 In the ideal case where semi-infinite planar diffusion is apparent, P = 0.5, in which case the expression for the impedance of a constant phase element (Eqn 2) equates to that of a Warburg semi-infinite diffusion element.10 With the use of P in this case as a freely changing variable, the low frequency EIS diffusion tail can be accommodated.
The CNLS fitting procedure involves the simultaneous optimization of the real and imaginary components of the EIS data by modifying the equivalent circuit parameters so as to minimize the weighted difference between experimental and predicted EIS data (S);22,23 i.e.,
where Z = Z' + jZ'' is the experimental data, Z* = Z'* + jZ''* is the predicted data, and wi is a weighting factor; i.e.,
where |Z| is the modulus of Z (). The results of the fitting process are shown in Table I.
Table I. Equivalent circuit parameters for the EIS data modeling.
Capacitance Data | |||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|
CPE1 (Capacitance) | CPE2 (Diffusion) | Cdl | |||||||||
Amplitude (V) | Rs (Ω) | Rct (Ω) | K | P | K | P | (μF) | (F/g) | CT (F/g) | CF (F/g) | A√D (m3 s−1/2) |
0.01 | 11.34 | 27.83 | 2.71 × 10−3 | 0.885 | 5.44 × 10−4 | 0.678 | 74 | 48 | 1873 | 1825 | 6.61 × 10−8 |
0.02 | 11.46 | 47.94 | 2.46 × 10−3 | 0.870 | 4.22 × 10−4 | 0.704 | 82 | 53 | 1739 | 1686 | 6.68 × 10−8 |
0.05 | 11.57 | 79.78 | 2.45 × 10−3 | 0.797 | 3.47 × 10−4 | 0.730 | 92 | 60 | 1539 | 1479 | 8.27 × 10−8 |
0.1 | 11.55 | 106.7 | 2.69 × 10−3 | 0.440 | 3.26 × 10−4 | 0.740 | 100 | 65 | 1283 | 1218 | 1.33 × 10−7 |
0.2 | 11.23 | 228.7 | 2.04 × 10−3 | 0.905 | 5.97 × 10−4 | 0.680 | 234 | 152 | 1650 | 1498 | 2.06 × 10−7 |
0.4 | 10.81 | 63.38 | 1.35 × 10−3 | 0.921 | 2.35 × 10−3 | 0.546 | 483 | 314 | 2340 | 2026 | 7.15 × 10−8 |
In terms of the series resistance (Rs) there is very little change across the range of AC excitation signals used. If any variation was to be noted it was a slight decrease with the use of the largest excitation signal (0.4 V). Aside from this, the average over all experiments was 11.3 ± 0.3 Ω which indicates that the changing AC potential had little impact on the overall resistivity of the electrode system.
The charge transfer resistance (Rct) can be approximated from the diameter of the semi-circular arc in the EIS data in Figure 1. It is clear that the modeled data is consistent with this expectation, with a progressively increasing Rct value as the applied AC potential increases, at least until a 0.2 V AC signal was used. To explain this, consider that electrodeposited manganese dioxide is a very heterogeneous material, not only in terms of its chemical composition,24–26 but also in terms of its morphology27 and structural composition.18,19 This has been shown to lead to a material that contains micro-domains of different structures, also with different energies.19 As such, these micro-domains undergo redox cycling at slightly different potentials. In this case, with an increasing AC potential, not only will more of these domains be accessed, but the intrinsic charge transfer resistance may also change (increase), thus increasing the measured Rct value. The quite dramatic drop in Rct with the use of a 0.4 V AC signal was quite surprising given the essentially monotonically increasing values of Rct with the smaller applied AC potentials. In this case, the likely explanation is that the increased number of micro-domains being accessed (and hence increasing Rct) is counterbalanced by the increasing overpotential activating the redox processes, in which case Rct is expected to decrease.
While the use of a constant phase element to model experimental EIS data leads to a much better fit, its interpretation in terms of electrode capacitance is not a straightforward comparison due to changes in both the K and P parameters of the CPE. To alleviate this problem, a method has been devised28 to calculate the capacitance (Cdl) from a parallel arrangement of a resistance and CPE; i.e.,
where ωmax is the angular frequency (ω = 2πf, where f is the frequency (Hz)) when the imaginary impedance is a maximum in the semicircular arc of the EIS data. Using Eqn 5, the calculated double layer capacitance (Cdl; μF and F/g) data for the various AC potentials are shown in Table I. For the 0.01–0.10 V amplitude AC potentials the measured capacitance increased only slightly, indicating little change in the accessibility to stored charge with these AC potential windows. However, for the 0.20 and 0.40 AC potentials the capacitance increased dramatically, most likely due to an increase in access to micro-domains within the manganese dioxide structure as a result of the larger potential window. This data reported here is quite significant for a pseudo-capacitive electrode because it represents the portion of the total electrode capacitance due to non-faradaic, double layer charging processes.
Diffusional processes occurring within the electrode also need to be considered in terms of the nature of the impedance response under diffusion limited conditions and the modeled CPE2 parameter values. During the impedance experiment a frequency is reached at which point the impedance response becomes mass transport limited, rather than kinetically limited. As the experiment continues, and the frequency continues to decrease, this gives rise to a steadily increasing impedance response or diffusional tail (as shown in Figure 1). This diffusional tail reflects limitations in the diffusion of cationic species within the manganese dioxide structure, and under the assumption of semi-infinite diffusion is related to the modeled impedance response according to the following equation:10
where A is the electrochemically active surface area (m2), D is the proton diffusion constant (m2/s), R is the standard gas constant, T is the temperature, σ is the CPE2 pre-factor, which in this case is equal to 1/K, F is Faraday's constant, and [M+] is the effective solid-state cation concentration (mol/m3).
As the electrochemically active surface area is not known with certainty,14,29 and likely changes during cycling,29 mass transport within the electrode is best considered in terms of values. In order to calculate these values it is first necessary to determine the effective solid-state cation concentration ([M+]) at each step during cycling. Ideally to determine these values it is necessary to know both the degree of reduction and the unit cell volume of the manganese dioxide during each cycle. However, there are a number of complications associated with determining these values, including:
- (i)Firstly, to know the degree of reduction of the manganese dioxide requires us to know the starting oxidation state of the deposited material. As has been discussed in detail in previous work,13 because these electrodes are comprised of thin films there is insufficient material available to determine (by potentiometric titration30 which is the standard method) the average manganese oxidation state.
- (ii)Secondly, as has been already established, manganese dioxide is a pseudo-capacitive material in which charge is stored both in the double layer and via redox reactions. Due to this combination of charge storage processes, it remains unknown what amount of charge is associated with double layer charge storage and what is associated with redox processes.
- (iii)Again, because thin films of material are being considered here, the structure of these materials is yet to be determined.13 This means that the unit cell volume of the manganese dioxide is unknown, and so the cation concentration in the solid state is also unknown.
- (iv)Finally, the relative effects of H+ and M+ on the expansion of the manganese dioxide unit cell are not known, particularly since the amounts of H+ and M+ inserted into the structure is also yet to be determined.
Despite these uncertainties, we are able to make some reasonable assumptions regarding each of these effects so as to be able to evaluate the diffusional characteristics of the electrode. The first assumption is that the structure of the electrodeposited manganese dioxide thin film is that of γ-MnO2, which is the typical phase of manganese dioxide deposited via electrodeposition from these types of electrolytes, and for which we have considerable information available on its structure.18 This, then, allows us to determine the unit cell volume, and assuming a starting composition of MnO1.96 (with the counter charge being made up of protons), which is typical for electrodeposited manganese dioxide, the solid state proton concentration (mol/m3) can be determined from:31
where x is the value of x in MnOx (1.96 in this case). The solid state proton concentration was determined to be 4338 mol/m3 in this case. Now, since we do not know the extent of discharge of the material due to the different charge storage mechanisms, nor whether protons or Na+ ions are associated with pseudo-capacitance, we will assume at this point that the solid state proton concentration remains constant over the course of the experiment. While we know this is not true, estimating the change in cation concentration is not possible at this stage. Under these circumstances, the value of was determined using Eqn 6, with the resultant values also shown in Table I. What we see in this data set are progressively increasing values of as the amplitude of the excitation potential is increased. This is likely the result of an increase in the diffusion coefficient, implying that the rate of solid state mass transport is increasing under these circumstances. As has been described before, with an increasing AC excitation signal we are accessing more and varied manganese dioxide micro-domains within the material, which apparently under these conditions exhibit faster diffusional characteristics. The exception to this is at the largest excitation potential (0.4 V), in which case decreased slightly. Again this is likely due to accessing different domains within the manganese dioxide structure that now demonstrate poorer mass transport kinetics. If the assumption that the structure of the manganese dioxide is γ-MnO2, then with the largest excitation signal we should be accessing the pyrolusite domains within this intergrowth structure.18,19 Proton transport through these domains is known to be limited because of the relatively high activation barrier that exists between adjacent sites in this phase.32 Additionally it has also been reported that the electronic conductivity of manganese dioxide drops considerably when this depth of discharge is reached,33 meaning that mass transport may be limited not only by cation transport.
Further analysis of the EIS data
As with most electrochemical experiments, an EIS experiment involves applying a known potential profile to the working electrode, all the while measuring the current that flows in response. Almost invariably for EIS the potential and current are then combined to determine the impedance using Ohm's law.10 The resultant impedance (as a function of frequency) can then be analyzed using a number of different approaches, the common equivalent circuit model approach being described above.
As an alternative approach to modeling the EIS data, perhaps in a way that is more in line with the use of cyclic voltammetry to examine electrochemical capacitor electrodes, let us begin by taking the collected EIS data (i.e., real and imaginary impedance data as a function of frequency) and re-express it now as simply current versus potential data at different frequencies. To achieve this note that the potential applied to the working electrode during an EIS experiment (E; V) is given by:10
where Emax (V) is the maximum amplitude of the applied AC potential, ω is the angular frequency ( = 2πf; rad/s), and t is the time (s). The current (I; A) that flows in response to this applied potential is given by
where |Z| is the modulus of the impedance given by
and ϕ is the phase shift (radians) between the applied potential and the measured current, as determined using
It is important to remember that each of these expressions used here is frequency dependent, and so the relationship between potential and current can be determined at all frequencies covered in the EIS experiment. This is equivalent of course to conducting a series of cyclic voltammetry experiments with varying sweep rates. Figure 4 shows an example of the relationship between potential and current at selected frequencies with an applied AC potential of 0.40 V. Note in this figure as well that the potential sweep rates corresponding to the different frequencies, which are much higher than what is normally used in the study of electrochemical capacitor electrodes, but nonetheless allowing us to access important high rate (power density) cycling data. Additionally, at low frequencies we make use of what are more typical potential sweep rates, thus allowing us to determine important capacity (energy density) information.
Another very obvious difference between the data in Figure 4, and that which is normally determined from a cyclic voltammetry experiment, is the fact that the data in Figure 4 is elliptical rather than box-like as expected from a cyclic voltammetry experiment. This arises as a result of the use of a sinusoidal potential profile rather than a triangular saw-tooth pattern used in cyclic voltammetry. Otherwise the data in Figure 4 is as expected with fast cycling rates leading to high current densities.
Given that we now have expressed the obtained data as current versus potential, just like in a cyclic voltammetry experiment on electrochemical capacitor electrodes, the specific capacitance can be calculated. This was done by integrating the specific current flow (i; A/g) over the time taken for the discharge half cycle of the sinusoidal potential to determine the amount of charge passed (Q; C/g); i.e.,
The total specific capacitance (CT; F/g) for the electrode system is then determined by dividing the charge by the potential window (V; V) used; i.e.,
Figure 5 shows the calculated specific capacitance as a function of both frequency and implied scan rate for the various amplitude AC potentials applied. What these figures show is how the performance of the electrode improves as the frequency or sweep rate decreases, to the point where the electrode performance hits a plateau. From the data in this figure it is apparent that performance limitations of this thin film electrode become limited at frequencies higher than ∼0.2 Hz, which corresponds to a sweep rate of at most ∼1 V/s. At high frequencies or scan rates the specific capacitance is low due to the inability of charge on the electrode surface to be matched by the movement of charge in the electrolyte. Similarly, the kinetics of the charge transfer and mass transport processes associated with pseudo-capacitance are too slow to contribute significantly to the total capacitance. As the frequency decreases the contributions made by charge storage at the manganese dioxide – electrolyte interface increase as a result of both double layer and faradaic processes. These processes are still only apparent at the interface because the frequency is not low enough to for diffusional processes to occur; i.e., any H+ or Na+ ions inserted into the surface do not diffuse appreciably into the bulk of the manganese dioxide structure. Ultimately the frequency decreases to a point where the maximum specific capacitance is attained, at which point the double layer at the manganese dioxide – electrolyte interface is saturated, and where the kinetics of mass transport of inserted cations is sufficiently fast to attain maximum utilization (reduction) of the manganese dioxide structure. Interestingly, the maximum specific capacitance (CT; F/g) attained is different depending on the potential window used, as shown in Figure 6 and Table I. The reason for this variation is due to the capacity (or charge) able to be extracted from the manganese dioxide electrode, as is also shown in Figure 6. According to Eqn 13 the specific capacitance is dependent on the amount of charge extracted from the electrode due to the combination of double layer charging and pseudo-capacitive redox processes, as well as the potential window used. The decreasing maximum specific capacitance when the potential window was increased from 0.02 to 0.2 V is due therefore to the capacity being extracted from the electrode not increasing sufficiently to account for the increasing potential window. From a material perspective this is likely due to proportionately fewer Mn4+ domains being accessed in the manganese dioxide structure as a result of redox processes despite the increasing potential window. Additionally, there was only a slight increase in interfacial capacitance with potential window in this range (as indicated by the equivalent circuit modeling data in Table I), which would also have been a contributor. When the potential window was increased from 0.2 to 0.8 V there was a dramatic increase in the specific capacity which translated to an increase in the specific capacitance, despite the increasing potential window. With this larger potential window it is apparent that the manganese dioxide electrode is becoming more activated with a dramatic increase in interfacial double layer capacitance (as seen in Table I), and also apparently a greater activity toward the reduction of Mn4+ species in the solid state, and hence greater utilization of the active material. Overall this interpretation implies that the larger potential window is necessary to activate the pseudo-capacitance within the manganese dioxide, as well as interfacial charge storage.
An important outcome from this study is that we are now in the position of being able to determine the contributions made by double layer charging and the faradaic redox processes to the total specific capacitance. Table I already lists the double layer capacitance (Cdl) determined from the EIS experiments, as well as the overall specific capacitance (CT) determined from the expanded analysis of the EIS data. The difference therefore is due to the charge stored in faradaic redox processes at the electrode surface (CF), as also now shown in Table I. The data shown here indicates that the majority of the electrode capacitance is due to the redox processes, which is not surprising given the relatively low specific surface area of the manganese dioxide14 compared to activated carbon, which limits the achievable double layer capacitance, and the relatively high redox activity of manganese dioxide toward redox cycling. It is to be noted that the faradaic capacitance is in most cases larger than the theoretical value of 1386 F/g for a 0.8 V potential window. As has been mentioned previously14 the higher than theoretical faradaic capacitance is likely due to greater than a one electron reduction of the manganese dioxide present in the electrode.
As a final comment in this section it is definitely worth highlighting that the specific capacitance data collected using this approach to evaluate the performance of the manganese dioxide electrode is comparable to that obtained using the more conventional cyclic voltammetry and constant current charge-discharge cycling approaches, as shown in Table II. However, a distinct advantage of the EIS approach was firstly that the results were obtained over the course of a few minutes, rather than the many hours typically used for the cyclic voltammetry, and secondly, that considerably more information was available from the EIS data, in particular extensive information on the rate capability of the electrode, and the breakdown of the total capacitance.
Table II. Specific capacitance data from cyclic voltammetry and constant current charge-discharge cycling. Electrode potential window = 0.8 V.
Cyclic Voltammetry | Constant Current | ||
---|---|---|---|
Cycle Rate | Specific Capacitance | Rate | Specific Capacitance |
(mV/s) | (F/g) | (A/g) | (F/g) |
10 | 3281 | 61 | 4184 |
20 | 2523 | 77 | 3476 |
50 | 2015 | 594 | 2367 |
100 | 2003 | 1132 | 2153 |
250 | 1643 | 2759 | 1724 |
500 | 1548 | 5500 | 1650 |
750 | 1503 | ||
1000 | 1466 |
Now that the specific capacitance of the manganese dioxide has been determined, particularly as a function of frequency or scan rate, it can be quite readily converted into specific power and energy data to provide us with some insight into the more applied performance characteristics of the electrode that has been prepared. It is important to reinforce the point here that the performance we are reporting is related to that of an individual electrode, not a full cell. Therefore, the gravimetric energy (E; Wh/kg) contained within the electrode was calculated using
where CT is the specific capacitance (F/kg), and V is the potential window (V). The gravimetric power (P; W/kg) is then the duration (t; h) over which the energy was delivered during the discharge half cycle; i.e.,
In this case the time was determined from the frequency (f; Hz) of the applied AC potential; i.e.,
Figure 7 compares the specific energy and power performance of the electrode prepared in this work as a function of the different potential windows employed.
The first feature to note about Figure 7 is the considerable amount of information it contains, all collected using a series of experiments that lasted just a few minutes. Previous work we have completed based on a similar electrode and electrolyte combination13 exhibited a specific capacitance of ∼2000 F/g measured using a 0.8 V potential window (0.0–0.8 V vs SCE) and a cyclic voltammetry scan rate of 5 mV/s. Using Eqns 14 and 15 as before, the specific energy and power of this electrode can be calculated to be 178 Wh/kg and 4000 W/kg, respectively, which is comparable and consistent with the data determined using the large amplitude EIS experiments reported here. In particular it appears that this electrode system reported previously has reached its maximum attainable energy with a 0.8 V potential window, with the diminished power due to the fact that a slower scan rate was used in the cyclic voltammetry experiments. Similarly, the performance evaluation experiments conducted here using cyclic voltammetry and constant current charge-discharge cycling also results in similar performance characteristics. Therefore, we can be confident that the EIS-based method that we have proposed here is representative on the electrode behavior assessed using conventional methods.
Another feature apparent in Figure 7 is the increasing energy and power with potential window. This was certainly to be expected given that the large potential window allows access to a wider range of electrochemically active sites within the manganese dioxide structure for energy storage. Additionally, it would seem that the charge transfer kinetics associated with these sites is also facile enough to sustain a high power output. Quantitatively, the change in specific energy and power output from this electrode as a function of potential window is shown in Figures 8a and 8b, respectively. Essentially both undergo a dramatic increase with potential window. Also shown in this figure are data sets from different frequencies, which show clearly that the specific energy increases with a decrease in frequency, while the specific power increases with frequency, as has already been discussed.
Figure 7 is also very informative in the sense that it indicates quite clearly under what conditions both the energy and power become limited. Just like the limitations in specific capacitance, the energy becomes limited at frequencies less than ∼0.2 Hz, a value that is essentially independent of the potential window employed, as shown in Figure 9a. For almost all the different potential windows employed, the maximum specific power had been attained at a frequency of at least 2000 Hz, as shown in Figure 9b. Between these apparent limiting frequencies there is a compromise between energy and power, with an increasing frequency diminishing energy and increasing power in a roughly asymptotic fashion.
Conclusions
Electrochemical impedance spectroscopy (EIS) has been used to characterize the behavior of thin films of electrodeposited manganese dioxide for use in electrochemical capacitors. Conventionally, EIS is employed with a small AC excitation signal at a fixed DC potential so as to focus on electrode behavior at that specific potential. However, as the AC excitation signal is increased in amplitude increasingly different electrode behavior was observed as a result of the potential range used accessing both energetically and kinetically different domains within the manganese dioxide structure. By setting the applied DC potential to the mid-point of the electrode potential range, and then with an increasing applied AC excitation signal, we have essentially established a method for characterizing the electrochemical behavior of the manganese dioxide electrode akin to conventional cyclic voltammetry or constant current charge-discharge cycling, in a much shorter timeframe and in a much more convenient fashion. As a consequence, specific power and energy data for an individual electrode can be calculated in a timely fashion. Additionally, this EIS-based method has been shown to enable separation of the total specific capacitance into the non-faradaic double layer capacitance, as well as the capacitance associated with faradaic redox processes. As a final note, this approach can be used for any other electrode system.
Acknowledgments
MFD acknowledges the financial support provided by the University of Newcastle for a PhD scholarship.