Suppporting Information for Electrochemical Effects in Thermoelectric Polymers

Transcription

1 Suppporting Information for Electrochemical Effects in Thermoelectric Polymers William B. Chang 1, Haiyu Fang 2, Jun Liu 3, Christopher M. Evans 2, Boris Russ 4, Bhooshan C. Popere 2, Shrayesh N. Patel 5, Michael L. Chabinyc 5, Rachel A. Segalman 2, * 1 Department of Materials Science and Engineering, University of California, Berkeley, CA Departments of Chemical Engineering and Materials, University of California, Santa Barbara, CA Department of Mechanical and Aerospace Engineering, North Carolina State University, Raleigh, NC Molecular Foundry, Lawrence Berkeley National Lab Berkeley, CA Department of Materials, University of California, Santa Barbara, CA Measurement Details The PEDOT:PSS and PEDOT:Ag:PSS films are drop-cast onto glass or sapphire substrates measuring 1 cm in width and 2 cm in length. The resistance of the PEDOT:PSS films was 2 kω, and the resistance of the PEDOT:Ag:PSS film was 3 kω. This was accomplished by casting a thicker film of PEDOT:Ag:PSS as compared to PEDOT:PSS. Electrodes corresponding to the set of three electrodes presented in Figure 1 are deposited onto the substrate 1.5 cm apart. The sample is placed onto two Peltier heaters, one which is kept at room temperature at all time, the other which is heated to the temperature interval required. Type K thermocouples were placed directly on the electrodes to accurately gauge the temperature difference. The thermoelectric voltage and current are measured through a resistor in series to the film. The load resistance is impedance matched to the resistance of the film, as it was shown that the maximum thermoelectric power can be observed at this point. This entire measurement was performed in a humidity controlled setup using water bubblers and nitrogen gas, and humidity was measured via a silicon-based humidity sensor. Electrical DC conductivity was measured using I-V sweeps with Au electrodes, PEDOT:PSS DC proton conductivity was measured using I-V sweeps with Ag:Ag-Nafion:Nafion electrodes, while PEDOT:Ag:PSS DC Ag + conductivity was measured with I-V sweeps with Ag:Ag-Nafion electrodes. [36] The maximum voltage bias during I-V sweeps was 50 mv with 1 mv steps per 1 millisecond. All measurements were done using Keithley 2400s interfaced to MATLAB. Ag:PSS was prepared with 40 µl of 18 wt% polystyrenesulfonic acid in water purchased from Sigma Aldrich added to 1 ml of deionized water with 5.7 mg of AgNO3 for 1:1 silver to sulfonate molar ratio solutions. The Ag:PSS solutions were covered in foil and stirred over-night. PEDOT:Ag:PSS was made by adding 25 µl of a 10 mg/ml solution Ag:PSS to 50 µl of PEDOT:PSS (Clevios PH1000, Lot 2014P0344), and casting on a 1 cm 2 glass substrate at 50 C, resulting in a film with 12 wt% silver. Silver electrodes were thermally evaporated or painted using silver paint. SEM and EDS was performed on the FEI XL40 Sirion FEG Digital SEM. Electrical conductivity was measured using platinum electrodes and running a cyclic voltammetry sweep on a Biologic VSP300 at 0.1 mv/sec to a peak voltage of +/- 1mV, [36] and checked by four point Van de Pauw measurements on gold electrodes using two Keithley 2400s. Film thickness was measured using a Dektak profilometer. Ionic conductivity was measured by a specialized cell which allows an electrode stack of film, Nafion (Ag/H) and silver foil to block out electronic contributions. A transfer length method was used to

2 determine the contact resistances between the samples and ion selective electrodes, where the contact resistance was obtained by measuring the resistances at three different electrode spacings (4.2mm, 7.0 mm and 11.8 mm) and extrapolating to zero. The humidity chamber was custom built, using dry compressed air flowed through two water bubblers to achieve saturation. Humidity was equilibrated for 15 to 30 minutes before measurement. Seebeck coefficient measurements were conducted using two Peltier heaters in the humidity chamber with two thermocouples. The voltage was measured with a Keithley We used time-domain thermoreflectance (TDTR) to measure thermal conductivity of PEDOT:Ag:PSS thin film samples. Details of TDTR methods can be found elsewhere. 1 Prior to the TDTR measurements, 80nm-thick Al thin films were deposited on the samples by magnetron sputtering. We used a modulation frequency f = 9.2 MHz, with a 1/e 2 radius of the focused laser beams of w 0 = 11.7 µm. For each sample, we fit the TDTR ratio data 1 with thermal diffusion model to extract the thermal conductivity of the polymer thin films. In the fitting model, the heat capacities of Al and Si are adopted from literature values. 2,3 The thickness of Al thin film was obtained from picosecond acoustics using a longitudinal speed of sound 6.42 nm/ps. 4 The thickness of the polymer films was measured by atomic force microscope (AFM). The thermal conductivity of the Al thin film was calculated using the Wiedemann-Franz law and the electrical resistance of the same transducer layer deposited on a 315 nm SiO 2 on Si reference sample. Density of the polymer films were determined by measuring the mass and volume of the material. Specific heat capacity was measured using Perkin Elmer DSC We set the interface thermal conductance between Al and polymer thin film G 1, and interfacial thermal conductance between polymer thin film and Si substrate G 2, as 100 MW//m 2 K. The sensitivity of the data analysis to the interfacial thermal conductance is small when the thermal conductivity Λ is low. The error bars of the TDTR measurements can be obtained by taking into account the individual uncertainties and sensitivities of the parameters in the thermal model. 1 The thermal conductivity of 50 nm thick PEDOT:Ag:PSS thin film was measured to be / W/m K. The electrical and ionic conductivity were measured using electron blocking electrodes and ion blocking electrodes as described by Shetzline et al., and are illustrated in Figure 1. [36] By using silver Nafion and Nafion membranes to block electron flow through PEDOT:PSS, the pure proton current can be measured. A silver Nafion/silver foil interface allows extraction of Ag + charge carriers while blocking electron flow. Platinum electrodes were used to measure the electronic conductivity, as protons and Ag + do not react with platinum. The entire setup was run in a humidity controlled potentiostat setup, and experimental details are described in the methods section. Control experiments for PEDOT:PSS and PEDOT:Ag:PSS In order to determine the extent of the ionic Seebeck effect enhancement in mixed conductors, control experiments were performed on PEDOT:PSS at 20% relative humidity to measure the

3 electronic Seebeck coefficient, and on Ag:PSS at 100% relative humidity to measure the thermoelectric voltage and current in a pure ion conductor. The electronic Seebeck coefficient in PEDOT:PSS is well-established, and can be seen in Figure S1-1. Figure SI-1 PEDOT:PSS measured at RH=20% in the open circuit condition. The temperature profile applied was as follows: T = 0K, T = 1K, T = 0K, T = 2.5K, T = 0K, T = 3.5K, T = 0K. The Seebeck coefficient was measured to be 18 µv/k, and there is no evidence of ionic Seebeck enhancement. A key feature of mixed conductors is that the electrical conductivity improves the thermoelectric current measured across a resistor, when compared to the pure ionic conductor. This is shown in SI-2, where silver poly-styrenesulfonate is measured as the control. The thermoelectric current is seen to be orders of magnitude lower than that observed for PEDOT:Ag:PSS, while the thermoelectric voltage is still high. This indicates that mixed conductors can have improved thermoelectric performance compared to pure ion or electronic conductors.

4 Figure SI-2 Ag:PSS with silver electrodes thermoelectric measurement, with the current and voltage being measured across a 1 kω resistor at a relative humidity of 100%. Temperature profile applied is as follows: T = 0K, T = 3K, T = 0K, T = 2.5K, T = 0K, T = 1K, T = 0K. While the thermoelectric voltage is very large due to the ionic Seebeck effect, pure ionic conductivity is low, resulting in a small thermoelectric current. An important feature of the ionic effect in mixed conductors is the behavior of the open circuit voltage under a temperature gradient. With a typical Seebeck coefficient measurement, such as that of PEDOT:PSS at 0% relative humidity (SI-1), the open-circuit voltage at zero temperature gradient is constant. This is not observed with mixed conductors, as shown in SI-3, where there is a shifting baseline when measured in the open-circuit condition. By measuring the thermoelectric current and voltage output by the film across a resistor, the changing baseline is mitigated, as is shown in Figure 2. Even when in the open circuit condition, PEDOT:PSS at 100% relative humidity still displays the exponential decay. In direct comparison, PEDOT:Ag:PSS also has a shifting baseline in SI-4, and also has a much prolonged ionic Seebeck effect. Since PEDOT:Ag:PSS still has residual protons, there is still an observable exponential decay observed upon temperature relaxation.

5 Figure SI-3 PEDOT:PSS Open circuit thermoelectric measurement at RH=100%. Temperature profile applied as follows: T = 0K, T = 1K, T = 0K, T = 1K, T = 0K, T = 2.5K, T = 0K, T = 2.5K, T = 0K, T = 3.5K, T = 0K, T = 3.5K, T = 0K. The shifting baseline at zero temperature gradient is minimized by measuring across a resistor with similar impedance to the film. Figure SI-4 PEDOT:Ag:PSS thermoelectric measurement in the open circuit geometry with RH=100%. Temperature profile applied is T = 0K, T = 1K, T = 0K, T = 2.5K, T = 0K, T = 3.5K, T = 0K. Residual protons in the film result in an exponential decay as observed in PEDOT:PSS. Finally, it is important to know the rise of the temperature gradient over time, and correlate this rise to the change in thermoelectric voltage. SI-5 demonstrates these two experiments for PEDOT:PSS and PEDOT:Ag:PSS. From these experiments we see that the

6 temperature rise is on the same time-scale as the voltage change, and that there is no overshooting of the temperature, which was posited as a potential factor in the exponential decay of the ionic Seebeck effect. Figure SI-5 Temperature and voltage measurements as a function of time, upon application of a temperature gradient of 1K at the 1700 second mark. These measurements are done to show the time dependence of the voltage drop, and that there is no over-shooting of the temperature gradient that results in the exponential decay observed in PEDOT:PSS. As has been shown in the literature, the ionic Seebeck voltage of PEDOT:PSS decays exponential. Through the addition of an electrochemical species, we observe a change in the decay of the thermoelectric voltage, as seen in SI-5. With PEDOT:Ag:PSS, the decay in thermoelectric voltage is better modeled using a linear trend, as seen in SI-6, which shows the absolute adjusted voltage of PEDOT:Ag:PSS upon addition of a temperature gradient T = 1.

7 Figure SI-6 Fitting of absolute adjusted thermoelectric voltage for PEDOT:Ag:PSS with a temperature gradient of 1K. An exponential fit is not valid to predict the decay of the ionic Seebeck effect; rather, a linear model is better suited to describe the decrease in thermoelectric voltage. Figure SI-7 Thermoelectric current complement from Figure 2 of PEDOT:PSS (green) and PEDOT:Ag:PSS (black). The inset shows the measurement setup, with a 2kOhm resistor. Material Lifetime Considerations Since PEDOT:Ag:PSS requires an electrochemical reaction for an enhanced Seebeck coefficient over time, special consideration must be made for the deposition and consumption of silver from the electrodes. Figure SI-8 is an optical micrograph of PEDOT:Ag:PSS after 1 hour of operation under a temperature gradient of 5K. Silver deposition can be observed on the painted electrode surface, which is roughly planar and not dendritic in nature. Additionally, in the right hand side of the micrograph small wire shaped bundles were observed. To further investigate these wire shaped bundles, SEM and EDS were taken of the wires. EDS revealed the composition of the bundles to be 68 wt% silver. These wires are hypothesized to be another form of failure mechanism for PEDOT:Ag:PSS.

8 Figure SI-8 Deposited silver growing on painted silver electrode/pedot:ag:pss interface during thermoelectric measurements. Black bar = 100 µm. Proper device cycling would be necessary to maintain the ionic Seebeek effect over prolonged usage. Micron sized nanowires can be observed in the right-hand portion of the image, and are discussed in the main text and SI- 6. Figure SI-9 EDS on microwires observed in PEDOT:Ag:PSS during thermoelectric operation confirm that they are metallic silver.

9 To determine the effect of silver incorporation into the film thermal conductivity, TDTR was performed on PEDOT:Ag:PSS and compared to literature values of PEDOT:PSS. The experimental geometry is shown in SI-10, along with the best-fit line to the measured data. From this, the thermal conductivity is calculated to be 0.24 W/(m K), which is the same literature value of PEDOT:PSS. The addition of silver does not significantly enhance the thermal conductivity of PEDOT:Ag:PSS. Figure SI-10 Thermal conductivity measurement of PEDOT:Ag:PSS via time-domain thermoreflectance method (the inset figure shows the schematics of this method and sample configuration) results in a phonon-dominated thermal conductivity of 0.24 W/(m K). The measurement data (open circle) and the best-fit (red-line) for measuring a 50-nm thick PEDOT:Ag:PSS thin film. The thermal conductivity upon addition of silver does not significantly change the thermal conductivity of PEDOT:PSS. The morphology of PEDOT:Ag:PSS is of great interest, as two percolated networks are required for ion transport and electron transport. In an attempt to observe the double network, TEM images were acquired of PEDOT:Ag:PSS, as it was expected the silver would enhance the contrast of the polymer film. The micrographs are presented in SI-11, and show clear regions that are silver rich, and regions that are polymer rich, as they lack contrast. However, these images are inconclusive since they are taken in a high-vac environment, while the bulk of the measurements are done in a hydrated state, which would have a significantly different morphology.

10 Figure SI-11 TEM images of PEDOT:Ag:PSS dehydrated spun coat films on silicon nitride windows. The morphology upon hydration is of great future interest. The dark regions are silverrich, and the light regions are polymer-rich. Finally, a cycle lifetime calculation is included to predict the overall lifetime of a potential device. The calculations are done using known experimental geometries and measured values. Calculation of cycle lifetime In a system where the electrical conductivity is six orders of magnitude higher than the ionic conductivity, the ion transference number is essentially zero, resulting in no ion Seebeck enhancement. The ionic conductivity observed in the PEDOT:Ag:PSS system is near 1 ms/cm, and perhaps with future optimization the ionic conductivity can be pushed as high as 10 ms/cm with an electrical conductivity of 10 S/cm. An ionic conductivity of 10 ms/cm is considered the limit, as this is the proton conductivity of Nafion, widely considered the best polymer ion conductor. [21] This high ionic conductivity was achieved through electrospinning aligned nanofibers of Nafion, and potential electrospinning of PEDOT:Ag:PSS could lead to higher ionic conductivity. Assuming that the Seebeck coefficient is 1 mv/k (near the peak value observed for PEDOT:PSS) this would yield a system with a peak zt = 1.5, which will last for seconds before decaying. The device lifetime is intrinsically related to a continuous electrochemical reaction. Traditional thermoelectric devices have an infinite theoretical lifetime; however, because the electrochemically active system here requires consumption of Ag+, the gradual depletion of silver under operation can be calculated and is found in the SI-12. The silver electrodes are approximately 100 µm in thickness compared to the polymer layer which is 3µm think, and this geometry would lead to 813 days of continuous operation if dendrite growth was suppressed and no mechanical separation of the electrode from the polymer is present. However, in the worst case scenario of dendrite formation, a lifetime of 2.5 days is predicated. Once the cycle has been

11 exhausted, the thermal gradient must be reversed, current must be run in the opposite direction, or else the device must be physically flipped. A simple geometry of two electrodes on a PEDOT:Ag:PSS film is considered, using values similar to experimental data. If all of one electrode is consumed through thermoelectric activity, the ionic enhanced Seebeck effect will end. Dimensions of one electrode: 100 µm x 1 cm x 0.2 cm = cm 3 Density of Ag = 10.5 g/cm 3 Therefore we would expect a total mass of silver at electrode = 1.9e -4 grams Now, with a temperature gradient of 5K, a Seebeck coefficient and resistance of a PEDOT:Ag:PSS film at 100% relative humidity, we can calculate the appropriate thermoelectric current. T = 5K, S = 500 µv/k, V T = V, Resistance of film = 10 kω and I = SG T I thermoelectric = 2.5e- 7 A =2.5e -7 C/s. This value has also been confirmed experimentally, by measuring the current through a resistor. Since the ionic conductivity of the film is orders of magnitude smaller than the electrical conductivity, we find that the ionic resistance of the film is 1,000 kω (1 MΩ). Using the Joule- Thomson equation, this will translate to an ionic thermoelectric current of I ionic = 2.5e -9 A =2.5e -9 C/s. Faraday s constant = C/mol, Molar mass of Ag = 107 g/mol Rate of consumption = 2.7e -12 g/s At this rate of consumption, the entire electrode should be consumed in 70,370,200 seconds. This is approximately 800 days, in this through plane geometry, if the ionic current does not result in dendrite growth. However, if dendritic growth does occur, then the dendrite limiting case (1 dendrite, 1 um radius), has a SA ratio is 314. We would expect a dendrite to result in failure within 2.5 days.

12 Power Output (W) Power Output versus Time Comparision 5E E-11 4E E-11 3E E-11 2E E-11 1E-11 5E-12 0 PEDOT:PSS Integrated Power over 30 mins = 8.67e-9 J PEDOT:Ag:PSS Integrated Power over 30 mins = 2.086e-8 J Break-even time = 519 s Time (s) PEDOT:Ag:PSS PEDOT:PSS Figure SI-12 Instantaneous power output of PEDOT:PSS and PEDOT:Ag:PSS over time for T = 1K. Over the entire experiment of 30 minutes, the total output useable energy of PEDOT:PSS was 8.6 nj, while the output energy from PEDOT:PSS was 20.8 nj. The break-even time for when PEDOT:Ag:PSS out-performs PEDOT:PSS is at 519 seconds. This means that if the temperature gradient fluctuates or reverses within 519 seconds, it would be advantageous to use PEDOT:PSS as a ionic thermoelectric material. If the temperature gradient is stable for a time period greater than 519 seconds, then PEDOT:Ag:PSS should be selected as the active thermoelectric material.

13 Figure SI-13 The contact resistances between the samples and ion selective electrodes was measured by a transfer length method at 100% relative humidity. The measurement was done using the same protocol as the two electrodes measurement described in the experimental section, but the spacing between the two electrodes was varied by from 4.2mm, 7.0 mm and 11.8 mm. By measuring the resistance at each spacing and extrapolating to a theoretical zero electrode spacing, the contact resistance was determined. The results shown in (a) and (b) are for PEDOT:PSS and PEDOT:AgPSS, respectively. The contact resistance between Ag/Ag- Nafion/Nafion electrodes and PEDOT:PSS was estimated to be 3.33 kω, which is higher than the contact resistance between Ag/Ag-Nafion electrodes and PEDOT:AgPSS is 0.82 kω. The larger contact resistance for the PEDOT:PSS setup could be caused by the extra Nafion membrane in the electrode setup, but the smaller slope of the linear fit in (a) indicates that PEDOT:PSS has a higher ion conductance.

14 Figure SI-14 The magnitude of the applied voltage was found to have a notable effect on the resistance, as expected. In a different measurement protocol conducted at 100% relative humidity, higher voltages of 0.5 V, 0.35 V, 0.15 V, V, V and -0.5V, were applied on PEDOT:AgPSS through Ag/Ag-Nafion electrodes for a prolonged time of 30 min to extract steady DC currents which were then used to calculate the sample resistance. The results show that the calculated resistances are smaller than the ones measured using the smaller voltage range of 0.05V, V as in Figure SI-13 (b). It may be related to the electrochemical reaction of Ag/Ag + at the electrodes which is facilitated at higher voltages. The contact resistance was determined to be 1.5 kω with the transfer length method under this new voltage range. In the case of PEDOT:PSS, this new protocol cannot be simply applied because a steady DC current cannot be extracted with blocking electrodes. Even a proton source and sink in the form of Pd electrodes and H 2 gas only yield a transient current before depletion of protons occurs in the Pd film.