Compressible sponge electrodes by oxidative molecular layer deposition (oMLD) of polyethylenedioxythiophene (PEDOT) onto open-cell polyurethane sponges

The foundational hypothesis motivating our effort here to employ oMLD to deliver PEDOT coatings on PU sponges is that the sequential self-limiting precursor exposures used in oMLD will allow for the formation of a uniform and contiguous network of electrically conducting PEDOT within PU sponges. We expect that this will enhance electrical conductivity relative to wet-chemical methods for delivering PEDOT onto PU foams. As an initial test, we coated PU sponges with 150 oMLD cycles of PEDOT using 10 s EDOT dose time, 100 s MoCl₅ dose time, and 100 s Ar purge after each precursor dose. These growth conditions were established in previous work and found to provide saturating doses on flat substrates [48]. After forming the PEDOT-coated PU sponge using oMLD, we employed EIS to measure the electrical impedance (Z) of the sponge as a function of AC frequency (f), as depicted in figure 1. These measurements were performed under ∼10% strain using 1'' diameter graphite rod contacts as described in the experimental methods section.

Zoom In Zoom Out Reset image size

The Bode plot of the EIS data in figure 1(a) for the 100 s MoCl₅ dose time condition in figure 1(a) indicates a high impedance of Z > 10⁶ Ω at low f ≤ 100 Hz. We expect that this high electrical impedance arises because MoCl₅ precursor delivery is insufficient to saturate the full surface area of the PU sponge. The MoCl₅ vapor pressure is <10 mTorr, requiring a ≥100 s dose time for complete saturation on flat substrates [48]. By increasing the MoCl₅ exposure using 250 s and 500 s MoCl₅ dose times, we observe a decrease in Z to 10³ Ω and <10 Ω, respectively, at low f ≤ 100 Hz, confirming that an increased MoCl₅ dose time provides a more complete PEDOT coating within the PU foam substrate. We note that ideal resistive behavior (phase equals zero) is observed for f < 10³ Hz for the solution cast PEDOT, and 250 and 500 s MoCl₅ dose time conditions, and the total impedance plateaus to a constant value for f ≤ 100. As such, we employed a frequency of 100 Hz for further electrical impedance characterization below, unless otherwise noted.

In figure 1(b), we compare the electrical impedance as measured at f = 100 Hz for each of the three samples in figure 1(a) relative to a PU sponge coated using solution cast PEDOT. Here, the solution casting of PEDOT onto the PU sponge was performed by mixing EDOT monomers and the MoCl₅ oxidant in an ethanol solution containing a PU sponge following procedures described previously [57]. Despite the solution-cast PEDOT forming a visible amount of excess PEDOT on the PU sponge, the electrical impedance we measured for this coating approach was still high at 8.9 × 10⁴ Ω. By contrast, the 500 s MoCl₅ dose condition yielded an electrical impedance of 5.2 Ω when measured under the same conditions. The 500 s MoCl₅ dose condition produced a resistance 1.7 × 10⁴ times smaller than using solution casting. This provided an initial positive indication that improving the conformality and uniformity of PEDOT coatings using oMLD could enhance the electrical conductivity of the PU sponges beyond solution casting and may provide a pathway for us to generate porous, electrically conductive electrodes with controlled microstructure.

In a next step, we set out to confirm that the mechanical properties of the PU were not significantly altered by the oMLD PEDOT coating. For this, we employed a mechanical compression tester to measure stress (σ) versus strain (ε) up to 50% compression (ε = 0.5) for sponges formed using the same oMLD conditions as in figure 1. Unfortunately, here we identified a significant deficiency in mechanical properties when using the 500 s MoCl₅ dose time to deliver oMLD PEDOT. Presented in figures 2(a) and (b) are photographs taken during the mechanical compression of sponges coated using 100 s and 500 s MoCl₅ dose times, respectively. In each panel (a) and (b) of figure 2, we depict photographs of the (i) initial sponge before compression followed by (ii)–(iv) sequential compression up to 50% strain, and then (v) a final photograph of the sponge after removing compression. The PU sponge coated using a 100 s MoCl₅ dose time behaved qualitatively like an uncoated sponge—returning to its original shape after compression. By contrast, the PU sponge coated using a 500 s MoCl₅ dose time crumbled under applied stress, disintegrating into a course powder after mechanical compression, as depicted in figure 2(b).v. The highest measured pressures under 50% compression were 1.52 psi and 4.72 psi in figures 2(a) and (b), respectively.

Zoom In Zoom Out Reset image size

To further quantify this mechanical compression behavior, in figure 2(c) we plot σ versus ε data collected during mechanical compression of PU sponges with PEDOT coatings delivered by oMLD using 100, 250, and 500 s MoCl₅ dose times, and compare this versus mechanical compression data collected for uncoated PU sponge and PU sponge coated with PEDOT using solution casting. In these graphs, ε = 0 corresponds to the uncompressed state. As the sponge is compressed (ε increases) at a constant strain rate of 0.1 in min⁻¹, the increase in the necessary stress required to compress the sponge is measured using a force gauge. The uncoated PU sponge reflects typical behavior for an open-cell compressible foam, where the stress initially rises as the sponge fibers undergo elastic deformation. Under sufficient stress, the fibers yield, leading to bulk collapse of the porous foam network, producing a region of relatively constant stress over a wide range of strain (e.g. from 0.1 < ε < 0.4). After reaching 50% strain, the strain is gradually removed at the same strain rate of 0.1 in min⁻¹, and the measured stress reflects the force that the sponge delivers to the piston under decompression. Except for the PU foam coated with 500 s MoCl₅ dose time, we observe qualitatively similar σ versus ε curve shapes for all of the sponges in figure 2(c), consistent with open-cell foam collapse. The PU sponges coated with oMLD PEDOT using 100 s and 250 s MoCl₅ dose times exhibit average stresses of 1.52 ± 0.05 psi and 1.94 ± 0.05 psi, respectively over the range in strains from 0.1 ≤ ε ≤ 0.4, compared with 0.96 ± 0.08 psi for the uncoated sponge and 3.2 ± 0.2 psi for the PU sponge with a solution-cast PEDOT coating. Likewise, the slope in σ versus ε on onloading indicates a Young's modulus of 27 ± 1 psi and 35 ± 2 psi, for the 100 s and 250 s MoCl₅ dose times, respectively, compared with 12.7 ± 0.6 psi for the uncoated sponge and 63 ± 3 psi for the PU sponge with a solution-cast PEDOT coating. By contrast, the PU sponge coated with oMLD PEDOT using a 500 s dose time exhibits a linear increase in stress on compression consistent with the disintegration of the foam fibers. After reaching the maximum strain of 0.5, this sponge exerts no force on the pistons during decompression, which is expected considering the crumbled powder remnants we observe in figure 2(b).v after compression.

We expect that PU becomes brittle under long MoCl₅ because of oxidative chemical decomposition [58–60]. PU is known to degrade under oxidizing environments and elevated temperatures. Under these conditions, urethane groups oxidize to form isocyanate and aldehyde products [58]. We expect a similar pathway proceeds here, where MoCl₅ oxidizes urethane groups according to R₁-NH-CO₂-C-R₂ + MoCl₅ → R₁-N=C=O and O=CH-R₂ + 2HCl + MoCl₃, breaking down the PU chemical structure into volatile decomposition byproducts. We note that Cl₂ is known to decompose PU [61], and the MoCl₅ oxidant undergoes a two-step reduction of MoCl₅→MoCl₄→MoCl₃ with oxidation potentials of 3.34 and 1.46 V versus SHE [48], both of stronger oxidation potential than Cl₂ (1.36 V versus SHE). To confirm that MoCl₅ is leading to decomposition of the PU chemical structure, we exposed 1/16'' thick dense PU sheets to a prolonged (1000 s) MoCl₅ exposure (no EDOT dose) and performed XPS and Raman spectroscopy to measure the chemical composition before and after MoCl₅ exposure. After MoCl₅ exposure, we observed a macroscopically visible roughening and discoloration of the PU sheet, indicating that MoCl₅ is chemically degrading PU. Using XPS and Raman spectroscopy, we expected to see a decrease in C–O and C=O groups comprising the ester-like group within the urethane based on the mechanism proposed above. Indeed, XPS results in figure 3(a) indicate a decrease in peak intensity at C 1s binding energies of ∼286 and ∼289 eV, consistent with the removal of C–O and C=O, respectively. Likewise, the Raman spectra in figure 3(b) indicate a decrease in the ester C=O feature at 1732 cm⁻¹ after MoCl₅ exposure [62–64]. We also observe a decrease in the Raman intensity at 1530 cm⁻¹ after MoCl₅ exposure, corresponding to loss of C–N and N–H groups [62–64]. We note that we observe equivalent peak intensity for C–H modes at 1300 cm⁻¹ and 2800–3000 cm⁻¹, and CH₂ modes at 1443 cm⁻¹ before and after MoCl₅ exposure indicating that these decreases in ester and N–H do not arise from different sampling volumes [62–64]. These data support that MoCl₅ oxidatively decomposes PU, leading to the mechanical brittleness we observe under high MoCl₅ exposure in figure 2.

Zoom In Zoom Out Reset image size

Unfortunately, the chemical reaction of MoCl₅ with PU makes it difficult to coat PU sponges with PEDOT by oMLD because MoCl₅ or another strong chemical oxidant is needed to polymerize EDOT [48]. We can decrease the MoCl₅ dose to limit PU degradation and preserve the PU sponge mechanical properties as indicated in figure 2, but this leads to higher electrical resistance through the sponge, as shown in figure 1. Considering these constraints, we hypothesized that decreasing the purge times following each precursor dose may overcome these challenges. Our rationale for this approach arose from previous studies examining the effect of purge time on MLD growth processes [53, 65] and the effect of purge time on ALD within polymer substrates [43–46, 66]. In general across these processes, vapor phase precursors are known to dissolve and infiltrate into bulk polymers, requiring 100 s of seconds or more of purge time to fully remove loosely bound precursors from the bulk polymer structure. Even on rigid dense substrates (e.g. Si), MLD purge times below 100 s lead to an increase in polymer growth rate, either from a similar dissolution effect into the bulk MLD film or from the formation of multilayers of monomer on the growth surface [53, 65]. Considering these known phenomena, we expect that an excess of EDOT monomer accumulates either on the surface of the PU or dissolved within the PU bulk during the EDOT dose, and that this EDOT slowly dissipates during the purge step. By decreasing the purge time after the EDOT dose, our strategy was to use this excess EDOT as a transient protective layer that would react with the MoCl₅ oxidant and rapidly form a surface PEDOT layer, preventing MoCl₅ from arriving at the underlying PU substrate. In other words, our goal was to intentionally perform controlled CVD to direct the polymerization to the surface of the PU fibers rather than within the PU fibers.

To evaluate whether this approach would be viable, we reduced the EDOT purge time from 100 s to 60, 30, 20, and 15 s, keeping the number of oMLD cycles constant at 30 for deposition onto PU. The mechanical properties for these PU sponges treated using different purge times during PEDOT oMLD are reported in figure 4. The mechanical testing data under compression (σ versus ε) is depicted in figure 4(a) for each of these purge conditions, compared with an uncoated PU sponge and a PU sponge coated with PEDOT using solution casting. We find that the PU sponge coated with oMLD PEDOT using a 20 s EDOT purge time exhibits a qualitatively similar σ versus ε curve, where a roughly constant value of stress is measured with increasing strain. This is consistent with bulk foam pore collapse similar to the mechanical behavior observed for uncoated PU and PU coated with PEDOT using solution casting in figure 2(a). In contrast, we find that sponges formed using 60, 30, and 15 s purge times exhibit increasing stress with increasing strain, which may indicate a gradient in mechanical properties throughout the foam. In figure 4(b) we plot the average stress (σ_avg) measured over the range 0.1 ≤ ε ≤ 0.4 for each of these sponges. We measure σ_avg = 0.96 ± 0.08 psi for uncoated PU and σ_avg = 3.2 ± 0.02 psi for PU coated with PEDOT using solution casting. The values of σ_avg fall between these bounds for all the oMLD-coated PEDOT sponges, with the highest value of σ_avg= 2.1 ± 0.1 psi measured for the 20 s EDOT purge time. We expect this higher value of σ_avg for the 20 s EDOT purge time arises from an increased PEDOT thickness on the fibers from CVD reactions (vide infra). Even for the 20 s purge condition, the value of stress to collapse the foam is 38% lower than the stress required to collapse the PU foam coated with PEDOT using solution casting. The error bars in figure 4(b) are the standard deviation in σ averaged over the range 0.1 ≤ ε ≤ 0.4, and we observe a higher variance in sponges coated using 15, 30, and 60 s EDOT purge times, consistent with the graded σ versus ε curve measured in figure 4(a) for these growth conditions.

Zoom In Zoom Out Reset image size

In figure 4(c) we plot the Young's modulus, E, on decompression for each of the sponges depicted in figure 4(a). We measure E = 12.7 ± 0.6 psi for uncoated PU and E = 63 ± 3 psi for PU coated with PEDOT using solution casting. The values of E fall between these bounds for all the oMLD-coated PEDOT sponges, starting at with a value of E = 30 ± 1 psi using a 15 s EDOT purge time up to a value of E = 61 ± 4 psi using a 60 s purge time. We measure a value of E = 37 ± 2 psi for the 20 s EDOT purge time condition, which is a factor of ∼3 higher than the uncoated PU. The value of error bars in figure 4(c) represent the standard error in the slope of σ versus ε on decompression from least squares linear regression. The trend in Young's modulus versus EDOT purge time in figure 4(c) is consistent with our expectations based on the reaction of MoCl₅ with the PU fibers as outlined in figure 3. At higher EDOT purge times, more of the EDOT is removed from the PU fibers, and the MoCl₅ readily degrades the PU, leading to increased brittleness of the fibers. As the purge time is decreased, the EDOT molecules remaining on/within the PU act as a reactive barrier preventing MoCl₅ from reacting with and degrading the PU. These results confirm our hypothesis that decreasing the purge time following the EDOT dose will prevent reaction of MoCl₅ with the PU fibers, making the PEDOT-coated PU foam softer.

As a side-effect of decreasing the EDOT purge time to prevent MoCl₅ from reacting with the PU fibers, we would expect that this will also lead to an increased PEDOT film thickness due to a greater extent of vapor-phase (CVD) reaction between EDOT and MoCl₅. To examine this, we measured the thickness of PEDOT formed on Si wafers placed adjacent to the foam pieces within the reactor, as well as the electrical resistance across the foam pieces under ∼10% mechanical compression, as plotted in figure 5. Indeed, we find that starting from an EDOT purge time of 120 s down to a purge time of 30 s, the PEDOT film thickness on Si increases from 19 ± 2 nm at 120 s EDOT purge time to 120 ± 20 nm at 30 s EDOT purge time. For an EDOT purge time of 20 and 15 s, the film thickness drops to 50 ± 8 and 14.0 ± 1 nm. Here the error bars in thickness represent the standard deviations measured on two Si wafers positioned upstream and downstream of the PU sponge in the oMLD reactor. We note that the EDOT and MoCl₅ precursors do not only react at the surface of the PU sponge fibers, but infiltrate into the bulk PU fibers down to a depth of 8 μm (vide infra). Therefore, the thicknesses on Si wafers adjacent to the PU sponges are not a direct indication of the amount of PEDOT deposited on/into the PU sponge. We report the PEDOT thicknesses on Si wafers placed adjacent to the PU here to help describe the qualitative effect of EDOT purge time on PEDOT oMLD growth.

Zoom In Zoom Out Reset image size

Considering this, the high variation in film thickness upstream and downstream of the PU for the 30 s purge time may arise from either (a) a thickness gradient due to CVD-type reactions or (b) decomposition of the PU sponge, leading to outgassing of decomposition byproducts that may condense on the Si adjacent to the PU sponge. Raman spectroscopy performed on Si wafers placed both upstream and downstream of the PU sponge coated using a 30 s purge time (not shown) does not provide evidence of PU residue present on either sample, indicating that the higher thickness value downstream of the PU is related to a thickness gradient due to CVD-type reactions. We emphasize that the absence of PU Raman signal in the Si downstream of the PU sponge for the 30 s purge time does not imply that the PU did not decompose, but only that PU decomposition byproducts did not incorporate into the film that formed on the downstream Si. A longer EDOT purge time is expected to yield more PU decomposition and less CVD reaction, while a shorter EDOT purge time is expected to yield less PU decomposition and more CVD reaction. Furthermore, the higher PEDOT thickness and larger variation in thickness for the 30 s purge time versus the 20 s purge time suggests a higher vapor phase EDOT concentration after 30 s of purge than 20 s of purge. This indicates that more EDOT has escaped from the PU at the longer 30 s purge time, which is consistent with less PEDOT deposited on the PU for the 30 s purge time versus the 20 s purge time, consistent with the lowest electrical resistance measured for a 20 s purge time in figure 5.

We attribute the decrease in PEDOT thickness at a 15 s purge to the gas-phase mixing of precursors upstream of the sponge, limiting precursor delivery to the sponge. The trend in thickness versus EDOT purge time is generally consistent with the electrical properties, where the lower electrical resistance is measured on sponges where a thicker PEDOT film is observed on Si witness wafers. We interpret that the lowest electrical resistance measured using a 20 s EDOT purge arises from effectively balancing PEDOT deposition with limited PU decomposition, consistent with the largest increase in mechanical properties using a 20 s EDOT purge as reported in figure 4, where the 20 s EDOT purge yields the highest average stress during bulk foam collapse.

Taken together, the results in figures 4 and 5 suggest that a 20 s EDOT purge time is most viable for forming PEDOT films onto PU sponges using oMLD of the purge times tested—this purge time resulted in mechanical properties consistent with bulk foam collapse in figure 4(a), the lowest electrical resistance in figure 5, and moderate increases in the average stress and Young's Modulus values of ∼2 times and ∼3 times higher, respectively, than uncoated PU foam in figures 4(b) and (c). However, to be viable in generating porous electrodes, the oMLD PEDOT coatings must uniformly coat the PU sponge through to the center of each foam piece.

To examine the depth to which PEDOT was formed within the PU foams, we sliced each foam piece in half and employed Kelvin-probe micropositioners to measure the electrical resistance from the center of the sponge to the outside edge (figures 6(a), (b)) versus from one outside edge to the opposite outside edge (figures 6(c), (d)). We plot the resistance measured using each of these configurations as a function of the number of EDOT/MoCl₅ oMLD cycles in figure 6(e). For these, and all further experiments below, we employed the optimized growth conditions of a 20 s EDOT dose, 20 s EDOT purge, 500 s MoCl₅ dose, and 20 s MoCl₅ purge based on the data in figures 1, 4, and 5, and we varied the number of oMLD cycles from 10 to 100. The center-to-edge and edge-to-edge resistance values measured for 1'' foams formed using each of these conditions are plotted in figure 6.

Zoom In Zoom Out Reset image size

Here, as we increase the number of oMLD cycles from 10 to 50, we observe a decrease in the electrical resistance by ∼3 orders of magnitude for both the center-to-edge and edge-to-edge measurement conditions. We attribute this decrease in resistance to an increase in PEDOT thickness/loading with an increasing number of cycles. A greater amount of PEDOT material will increase the electrically conductive network throughout the PU fibers, and thereby reduce the electrical resistance of the sponge. We observe roughly the same electrical resistance for both 50 and 100 oMLD cycles, suggesting that additional material provides diminishing returns in electrical connectivity after ∼50 oMLD cycles. This may arise in part from residual oxidation of the PU as described in figure 3 as the number of oMLD cycles increases (i.e. higher total MoCl₅ exposure). To first order, one would expect that the electrical resistance for center-to-edge would be ∼1/2 of the electrical resistance for edge-to-edge, neglecting differences in the foam network that may impact electron conduction pathways. However, the center-to-edge electrical resistance is 2–8 times higher than the edge-to-edge electrical resistance for all numbers of oMLD cycles in figure 6. This suggests that the electrical connectivity pathways may not be as robust to the center of the sponge as around the external perimeter of the sponge, consistent with the partial CVD growth at this purge time as discussed above surrounding figure 5. Nonetheless, the general trend in electrical resistance versus number of oMLD cycles and the order-of-magnitude values of electrical resistance for center-to-edge tracks closely with the values measured for edge-to-edge, indicating that the PEDOT film is delivered all the way to the center of the sponge, and that increasing number of oMLD cycles increases the thickness/loading of PEDOT throughout the whole volume of the PU sponge.

To further confirm PEDOT loading at the center of each 1'' foam cube, we performed SEM measurements on the center area of the PU sponges in figure 6 after cutting them in half. In figure 7, we show SEM micrographs of PU sponges coated with 30 oMLD cycles (a)–(c), 50 oMLD cycles (d), and 100 oMLD cycles (e). All samples in figure 7 were formed using a 20 s EDOT dose, 20 s EDOT purge, 500 s MoCl₅ dose, and 20 s MoCl₅ purge based on the studies in figures 1, 4, and 5 above. The SEM images in panels (a), (d), and (e) of figure 7 were collected using backscatter electron detection (BSD) while carefully maintaining identical conditions (beam current, working distance, acquisition time, and detector brightness and contrast). By controlling the portion of the signal yield attributable to variable system parameters, the resulting intensity (brightness) is more strictly correlated to absolute material density and can thus be compared across samples. In figure 7(a) we see that the PU fibers contain a high-intensity lining on the external surface of each fiber, indicating that a higher density PEDOT coating was formed around each lower density PU fiber. To confirm this interpretation, we performed EDS analysis on this same fiber as indicated in figure 7(b) (note some beam damage after prolonged electron beam exposure) and mapped the Mo and S composition using EDS. Here we show both Mo and S together because these elements are unique to the PEDOT coating. We note that oMLD PEDOT films grown using EDOT and MoCl₅ are known to contain residual MoCl_x at a near stoichiometric quantity with EDOT monomers [47, 48]. Comparing the intensity data in figure 7(a) to the EDS data in figure 7(c) we observe that, indeed, the increased intensity near the edge of each PU fiber corresponds to an increased quantity of S and Mo consistent with a PEDOT layer. This SEM/EDS data confirms that the EDOT and MoCl₅ precursors are reaching the fibers within the center of each 1'' sponge piece and reacting to form a conformal PEDOT layer on these fibers.

Zoom In Zoom Out Reset image size

To confirm that the decrease in resistance with increasing number of oMLD cycles shown in figure 6 corresponds to an increase in the thickness/loading of PEDOT, we compare the BSD SEM micrographs after 30 oMLD cycles (figure 7(a)), 50 oMLD cycles (figure 7(d)) and 100 oMLD cycles (figure 7(e)). Qualitatively we observe an increase in the average intensity of each SEM image with an increasing number of oMLD cycles, consistent with an increase in PEDOT thickness/loading. We note that the thickness of PEDOT on Si wafers placed adjacent to each PU sponge analyzed in figure 7 were measured to be 32 nm, 88 nm, and 120 nm for 30, 50, and 100 oMLD cycles, respectively. We also measured the mass change of the PU sponges after oMLD growth. The initial mass of uncoated sponges was 363 ± 6 mg, and the mass of the sponges depicted in figure 7 after oMLD growth were 270 mg, 320 mg, and 367 mg, following 30, 50, and 100 oMLD cycles, respectively. These final masses correspond to −25%, −12%, and +1% changes in mass, respectively, relative to the initial PU foam mass. We attribute the net decrease or neutral mass change to the chemical decomposition and release of volatile byproducts from the PU fibers during MoCl₅ exposure, as described above surrounding figure 3. We note that the decomposition and removal of mass from PU by MoCl₅ exposure is reminiscent of molecular layer etching [67] and atom layer etching processes [68] reported in recent years, but the initial indications in this work do not suggest that this etching process is self-limiting for MoCl₅ exposure to PU. Further insights into the deposition of oMLD PEDOT onto PU foam fibers is gained by examining higher resolution SEM images of individual PU fibers in the insets of figures 7(a), (d), and (e). In each of these three panels of figure 7, we show an individual PU fiber and plot the pixel intensity along the cross-section of each fiber as indicated. In general, we observe that the average intensity within each fiber increases with an increasing number of oMLD cycles as indicated by the horizontal dashed lines in each inset. This is consistent with the qualitative observation of increasing intensity with increasing number of oMLD cycles across the wide field in figures 7(a), (d), and (e), and supports an increased mass loading with increasing oMLD cycles. However, these SEM images also reveal additional insights into the nature of the oMLD growth process on the PU fibers. First, we observe rough/torn edges on each of the PU fibers after oMLD growth. We attribute this to the disintegration and etching of PU upon MoCl₅ exposure as described above surrounding figure 3. We also note that there is a distinct increase in intensity with maxima at the peripheries of each fiber, to a depth of up to ∼8 μm. This dimension is ∼100 times larger than the thickness of 50 nm measured on Si wafers adjacent to the PU sponge and suggests that the MoCl₅ and EDOT precursors are infiltrating into the bulk PU fibers, providing both a primary external PEDOT coating and secondary PEDOT layer infiltrated into the PU fibers. This behavior is consistent with prior observations for the infiltration of organometallic precursors into bulk polymers [43–46, 66].

These combined processes of (1) PU etching by MoCl₅ and (2) vapor infiltration of EDOT and MoCl₅ into the PU fibers are expected to give rise to nonlinear changes in mechanical properties versus the number of PEDOT oMLD cycles. In figure 8(a) we report the σ versus ε curves during mechanical compression and decompression for PU sponges coated with 10, 30, 50, and 100 oMLD cycles. In figure 8(b) we report the average σ_avg measured over 0.1 ≤ ε ≤ 0.4 for each of these sponges. The value of σ_avg increases from 0.96 psi for uncoated PU up to 2.1 ± 0.1 psi measured for 30 oMLD cycles, and then decreases to 1.9 ± 0.1 psi at 50 oMLD cycles and 1.6 ± 0.4 psi at 100 oMLD cycles in figure 7(b). Likewise, in figure 8(c) we plot E on decompression for each of the sponges depicted in figure 8(a). The value of E increases with an increasing number of oMLD cycles up to a value of 47.2 psi for 50 oMLD cycles and then a decrease by 14% to a value of 40.5 psi after 100 oMLD cycles. These trends in σ_avg and E are consistent with our expectations based on the combined effects of (1) the infiltration and incorporation of more PEDOT with a higher number of oMLD cycles and (2) the reaction of MoCl₅ with the PU fibers as outlined in figure 3. Rigid, dense PU has a Young's modulus ∼500 psi [69], whereas PEDOT has a Young's modulus of ∼200 000 psi [70]. This ∼400-fold higher Young's modulus of PEDOT relative to PU means that a greater mass fraction of PEDOT within a PU sponge will give rise to higher values of σ_avg and E. However, under extended MoCl₅ exposure, more PU will be etched away, leading to a decrease in σ_avg and E. We attribute the increase and levelling off that we observe for σ_avg and E versus the number of oMLD PEDOT cycles in figure 8 to the superposition of these effects.

Zoom In Zoom Out Reset image size

To be useful in electrochemical devices, these sponge electrodes must effectively balance (1) high electrical conductivity, (2) low compressive stress for pore collapse, and (3) low Young's modulus for more elastic recovery of foam pores. Based on the composite data collected above, we identified that, of the conditions explored in this work, 50 oMLD PEDOT cycles grown using a 20 s EDOT purge time provide the best balance of electrical conductivity and mechanical compressibility. However, to be useful as porous compressible electrodes, these PU foams must maintain these electrical and mechanical properties under repeated mechanical compression and exposure to electrolyte solutions. Based on this, we performed extended mechanical cycling of PU foams coated with PEDOT using these conditions and examined changes in the electrical conductivity and mechanical properties in figure 9. Here, we control the strain (figure 9(a)) using the compression tester described above and simultaneously measure the stress (figure 9(b)) and electrical resistance (figure 9(c)) during extended mechanical cycling for up to 400 mechanical compression/decompression cycles. We report the mechanical stress versus strain data for these PEDOT-coated PU sponges as-formed (figure 9(d)) and after repeated rinsing in tap water, manually wringing out the liquid, and then air-drying (figure 9(e)). We also report changes in the electrical resistance as a function of mechanical compression cycles (figure 9(f)) for these sponges as-formed (i), and after repeated rinsing and wringing (ii).

Zoom In Zoom Out Reset image size

Panels (a)–(c) of figure 9 are data for an as-formed PEDOT coated PU sponge loaded with 50 oMLD cycles of PEDOT using the optimal growth conditions identified above. In figures 9(b) and (c), we observe that the stress and strain undergo dynamic changes during the first 20 mechanical compression cycles. In figure 9(d), we process the data from figures 9(a) and (b) and plot the σ versus ε curves for select cycles, with similar data shown for a PEDOT-coated PU sponge after rinsing and wringing in figure 9(e). We observe that the mechanical response reaches steady state after 40 compression cycles, with minimal difference between the σ versus ε curves after 40 and 200 compression cycles for the as-formed PEDOT-coated PU, and minimal differences between the σ versus ε curves after 40 and 160 compression cycles for the PEDOT-coated PU after repeated rinsing in tap water and manual wringing. We note that this repeated mechanical cycling was performed in such a way that the strain was reduced on decompression until zero force was measured, at which point compression was reinitiated, starting the next compression cycle. Because of this, the minimum strain on decompression reached a steady state value of ∼0.1 for the as-formed PEDOT-coated PU (figure 9(d)), and a value of ∼0.2 for the PEDOT-coated PU after rinsing/wringing (figure 9(e)). As above, these data were collected at a strain rate of 0.1 in min⁻¹. This steady state mechanical compression data in figures 9(d) and (e) is analyzed in more detail below, but the convergence to repeatable steady-state mechanical compression behavior after >100 cycles generally suggests that repeated mechanical cycling does not lead to significant degradation of the mechanical properties of these sponges.

Although the PEDOT-coated PU sponges both as-formed and after rinsing/wringing reach steady state behavior, the mechanical properties we observe vary between the two. For the as-formed PEDOT-coated PU sponge, the Young's modulus on decompression during the first cycle was 29.1 psi, and decreased to 22.5 psi after 40 compression cycles and 22.1 psi after 200 compression cycles. For comparison, after repeated rinsing and mechanical wringing the Young's modulus on decompression during the first cycle was 41.0 psi, and decreased to 18.8 psi after 40 compression cycles and 17.0 psi after 160 compression cycles. Similarly, the average energy required to compress the as-formed PEDOT-coated PU foam from a strain of 0.2 to a strain of 0.5 was 26.9 mJ after 40 compression cycles and 26.4 mJ after 200 compression cycles. The energy recovered on decompression was 13.2 mJ after 40 compression cycles and 13.4 mJ after 200 compression, corresponding to ∼50% energy recovery at this strain rate of 0.1 in min⁻¹. For comparison, the average energy required to compress the PEDOT-coated PU foam after rinsing/wringing from a strain of 0.2 to a strain of 0.5 was 13.8 mJ after 40 compression cycles and 13.4 mJ after 160 compression cycles. The energy recovered on decompression was 6.5 mJ after 40 compression cycles and 6.7 mJ after 160 compression cycles, corresponding to ∼50% energy recovery. These data indicate that the rinsing and wringing process may remove some of the PEDOT-coated PU sponge, and/or damage the integrity of the polymer network within the PU foam, leading to a smaller amount of energy required to compress the sponge.

In figure 9(f) we report the minimum electrical resistance measured at ε = 0.5 versus the number of compression/decompression cycles for both the as-formed PEDOT-coated PU, and PEDOT-coated PU following rinsing and wringing. As-formed PEDOT-coated PU exhibits a low electrical resistance of 20 Ω during initial mechanical cycling, that only increases by 0.02 Ω per compression/decompression cycle on average over >400 mechanical compression cycles. After rinsing and wringing, the PEDOT-coated PU exhibits a higher electrical resistance of 30 Ω during initial mechanical cycling, which increases by 0.3 Ω per compression/decompression cycle on average over >150 mechanical compression cycles. Both of these sponges exhibit mean electrical resistances of ≤100 Ω or less during the extent of mechanical testing shown in figure 9. The higher electrical resistance observed following rinsing and ringing may arise from either (a) changes in ionic dopant concentration and state of charge or (b) mechanical degradation of the PEDOT sponge network following exposure to the manual rinsing and wringing procedure. These results are qualitative demonstrations that high electrical conductivity can be maintained even after exposure to electrochemically relevant conditions.