The Electrochemistry of Fe₃O₄/Polypyrrole Composite Electrodes in Lithium-Ion Cells: The Role of Polypyrrole in Capacity Retention

Magnetite nanoparticles were synthesized using a co-precipitation method developed previously.^14,15 Briefly, stoichiometric amounts of iron(III) chloride hexahydrate (FeCl₃·6H₂O) and iron(II) chloride tetrahydrate (FeCl₂·4H₂O) were dissolved in H₂O under N₂(g) and added dropwise to aqueous triethylamine ((CH₃CH₂)₃N). The product was washed and dried in vacuo.

PPy was prepared by dissolving p-toluenesulfonic acid (HTos) monohydrate (CH₃C₆H₄SO₃H·H₂O) in H₂O. Under stirring, distilled pyrrole (C₄H₅N) and then iron(III) chloride hexahydrate (FeCl₃·6H₂O) to initiate oxidative polymerization were added and stirred for 18 hours. The Ppy was filtered and isolated as a black solid.

The magnetite-polypyrrole (Fe₃O₄-PPy) composite material was prepared as follows. As synthesized magnetite was suspended in H₂O. Then p-toluenesulfonic acid (HTos) monohydrate (CH₃C₆H₄SO₃H·H₂O) was dissolved in H₂O and added to the suspension and stirred for 15 hours. Distilled pyrrole (C₄H₅N) was added to the acidified magnetite suspension, ultrasonicated and stirred for 4 or 24 hrs yielding the Fe₃O₄-PPy product. The product was then washed with H₂O and dried in vacuo.

Crystallite size and sample purity were determined using X-ray powder diffraction (XRD) on a Rigaku Ultima-IV diffractometer with Cu Kα measured in a 2θ range from 5–90° with a step width of 0.04° and a scan rate of 5°/min. The sample purity was confirmed using the PDF card for magnetite (01-071-6336) and the peaks of interest (2θ = 30.2°) were fit using PeakFit version 4.12 by using a Person VII area function to find the full width at half max then applying the Scherrer equation to calculate crystallite size. The weight percent of polypyrrole was determined using thermogravimetric analysis (TGA) with a Thermal Advantage Release Q series 5.4.0 instrument using an alumina sample cup ramping from 25–800°C at a rate of 10°C/min under air. Raman spectroscopy was used to evaluate the polypyrrole coating on a Horiba Scientific XploRA instrument with a 532 nm laser between 100–2000 cm⁻¹. A JEOL JEM 1400 microscope was used for Transmission electron microscope (TEM) imaging. Images were obtained at 120 kV voltage using a Gatan ORIUS SC1000B CCD camera. Inductively coupled plasma optical emission spectroscopy (ICP-OES) on a Thermo Scientific iCAP 6000 series spectrometer was used to evaluate Fe dissolution. The material was pressed into a 13 mm in diameter pellet with the thicknesses ranging from 0.35–0.43 mm and placed between two electrodes under constant pressure to measure AC impedance on a Bio-Logic – Science Instruments EC-lab using a sine wave amplitude of 10 mV over a frequency range of 175 Hz–1 MHz. The impedance data was fit using equivalent circuits using the software ZView version 3.4c.

Ex situ XANES and EXAFS measurements of the Fe K-edge (7.11 keV) were conducted at the Cornell High Energy Synchrotron Source (CHESS) on beamline F3 with a Sagittal Si(111) crystal monochromator in transmission geometry with a Fe reference foil. Fe metal nanoparticles, FeO, α-Fe₂O₃, γ-Fe₂O₃, and Fe₃O₄ standards were measured with pristine samples of the materials. The data collected were aligned, merged, and normalized in Athena.

Electrodes were prepared using Fe₃O₄, Fe₃O₄-PPy, or PPy as the active material, 20% carbon, and 10% polyvinylidene fluoride (PVDF) in n-methyl-2-pyrrolidone (NMP). The slurry was spread onto copper foil with the final ratios of active material, ketjenblack carbon, and PVDF 7:2:1, respectively. The Fe₃O₄-PVDF electrode was constructed with a ratio of 9:1, Fe₃O₄:PVDF. Electrodes were also generated without additional carbon or binder using the same procedure. Cells were assembled in an Ar_(g) filled glovebox with polypropylene separator, Li metal counter electrode, and 1 M LiPF₆ in dimethylcarbonate:ethylene carbonate (7:3) electrolyte. All cells were cycled at a rate of 50 mA/g and 100 mAg/g of Fe₃O₄. The cycling data reported is an average capacity from each cell condition ranging from 2–5 cells. AC Impedance was completed on cells before and after cycling on a Bio-Logic – Science Instruments EC-lab using a sine wave amplitude of 10 mV over a frequency range of 1 mHz–1 MHz. and was fit using equivalence circuits on ZView version 3.4c.

For the XANES/EXAFS measurements, the cells were cycled to 10 cycles and recovered in the charged state or discharged state. The coin cells were opened in an Ar_(g) filled glove box where the electrode was extracted, washed with dimethylcarbonate, dried, then put between polyimide tape for analysis.

The parent Fe₃O₄ and Fe₃O₄-PPy composite were imaged by TEM, Figures 1A, 1B. The as prepared Fe₃O₄ material in absence of PPy is ∼9 nm in dimension with evidence of aggregation, Figure 1A. In the presence of 20% PPy, Figure 1B, the 9 nm Fe₃O₄ particles are closely affiliated with the PPy that is visible in the image consistent with the initial hypothesis. The preparation of the Fe₃O₄-PPy composite is illustrated in schematic form in Figure 1C.

Zoom In Zoom Out Reset image size

XRD patterns were collected after each synthesis to ensure purity of all materials. The background was subtracted and the patterns corresponded well to the magnetite structure (PDF # 01-071-6336) as shown in Figure 2A. After the PPy reaction resulting in Fe₃O₄-PPy composites containing 6 wt% or 20 wt% of PPy there are no additional peaks or significant peak shifting in the XRD indicating no major structural changes in the magnetite occurred. The crystallite size of the magnetite, 9 ± 1 nm along the (220) plane, remained constant during the polymerization process and no PPy peaks are present in the XRD after the polymerization, consistent with the XRD of pure PPy which shows no crystalline peaks and only a broad amorphous peak between 10–30°, Figure 2A.

Zoom In Zoom Out Reset image size

Thermogravimetric analysis (TGA) was used to quantify the amount of PPy in the Fe₃O₄-PPy composites. Pure Fe₃O₄ shows approximately 2.5 wt% loss below 250 °C attributed to surface water. In the samples with PPy added, an additional weight loss is observed between 250–625°C as shown in Figure 2B attributed to PPy decomposition consistent with previous literature and the mass loss profile of pure PPy.^10,16–18 The amount of PPy in the Fe₃O₄-PPy composites was determined from this mass loss in this region with Fe₃O₄ + 6 wt% PPy having an average weight percent of 6 ± 1 wt% PPy and the Fe₃O₄ + 20 wt% PPy had an average weight percent of 21 ± 1 wt% PPy.

Raman spectroscopy was used to characterize the Fe₃O₄, PPy, and Fe₃O₄-PPy composite materials. Specifically, Raman was used to confirm the presence of conducting polymer by looking for vibrational modes indicative of the oxidized state of PPy. The peaks associated with PPy in Figure 2C are 940, 977, 1054, 1241, 1334, 1399, and 1580 cm⁻¹. Peaks indicating short conjugation length of PPy are 1054, 1241, and 1334 cm⁻¹.¹⁹ The frequencies at 940, 977, 1399, and 1580 cm⁻¹ are associated with the δ_C-H, δ_ring, ν_C-C, and ν_{C = C}, respectively. The frequencies are within the range of previously reported polaronic and bipolaronic states of PPy indicating an intrinsic electrically conductive network.²⁰ The 6 wt% sample shows increased intensity of the PPy peaks with decreased intensity of the iron oxide peaks, with the 20 wt% PPy sample following the same trend. The Pure PPy spectrum shows a peak at 386.5 cm⁻¹ which could be residual iron hydroxide from the iron chloride used in the polymerization of the pure pyrrole monomer that was incorporated into the structure. In Figure 2C, the spectrum of pure magnetite shows the characteristic frequencies of the E_g, T_2g, and A_1g vibrations at 386, 505, and 666 cm⁻¹, respectively.²¹ The E_g and T_2g peaks appear to be shifted from the literature values of 310 and 540 cm⁻¹ and could suggest a hydrated surface based on the Raman of hydrated iron oxides which show E_g, T_2g, and A_1g vibrations at 370–385, 480–548, and 710 cm⁻¹, respectively.²²

AC impedance was measured for the as-synthesized Fe₃O₄, PPy and the Fe₃O₄-PPy composite materials pressed into pellets free from carbon or binder additives. The Bode plots of the pellets are shown in Figure 3 where the Fe₃O₄ has the largest overall impedance and the Fe₃O₄ + 20 wt% PPy and Pure PPy have the lowest overall impedance. The pure Fe₃O₄ and Fe₃O₄ + 6 wt% PPy show behavior consistent with an RC parallel circuit with resistor values of 5 × 10⁵ ± 2 × 10⁴ Ω and 2 × 10⁴ ± 2 × 10² Ω, respectively. Pure PPy and Fe₃O₄ + 20 wt% PPy show only a resistor component, with values of 99 ± 1 Ω and 62 ± 1 Ω, respectively. These values for the pure PPy material are comparable in magnitude to prior reports, however the morphology of the PPy and doping level can vary substantially between different syntheses altering the intrinsic conductivity of the polymer.^23–25 For the Fe₃O₄ + 20 wt% PPy sample, the elimination of the capacitor component and decrease in the resistor value suggest an electrically conductive network propagates electron transfer throughout the pellet.

Zoom In Zoom Out Reset image size

The electrochemistry of the composites were tested for various electrode compositions: as-synthesized Fe₃O₄ with 10 wt% PVDF binder (Fe₃O₄-PVDF), composites of Fe₃O₄ + 6 wt% PPy (Fe₃O₄ + 6 wt% PPy) and Fe₃O₄ + 20 wt% PPy (Fe₃O₄ + 20 wt% PPy) prepared with no additional carbon or binder. Then, those with electrically conductive carbon and binder for the electrodes (Fe₃O₄-C-PVDF, Fe₃O₄ + 6 wt% PPy-C-PVDF, Fe₃O₄ + 20 wt% PPy-C-PVDF). A cell with an electrode using polypyrrole, carbon, and binder was also constructed for comparison (PPy-C-PVDF).

Figure 4 shows the voltage profiles versus electron equivalents during reduction for Fe₃O₄ and the composites with and without added carbon in the electrode for the 1^st, 2^nd, 10^th, and 20^th cycles. Overall, cycle 1 and 2 capacities for the samples with carbon show the trend of Fe₃O₄-C-PVDF > Fe₃O₄ + 6 wt% PPy-C-PVDF > Fe₃O₄ + 20 wt% PPy-C-PVDF where the difference in delivered capacity between cycle 1 and cycle 2 is 603 mAh/g (5.2 electron equivalents), 445 mAh/g (3.9 electron equivalents) and 382 mAh/g (3.3 electron equivalents), respectively. The samples without carbon show the same trend of cycle 1 and 2 capacities for Fe₃O₄-PVDF > Fe₃O₄ + 6 wt% PPy-PVDF > Fe₃O₄ + 20 wt% PPy-PVDF with cycle 1 and 2 capacity difference values of 515 mAh/g (4.5 electron equivalents), 493 mAh/g (4.3 electron equivalents) and 205 mAh/g (1.8 electron equivalents), respectively. Notably, the samples containing added carbon show much higher cycle 10 capacities than those with no added carbon. The samples including electrically conductive carbon delivered cycle 10 capacities of 725 mAh/g (6.3 electron equivalents), 584 mAh/g (5.0 electron equivalents), and 424 mAh/g (3.6 electron equivalents) for Fe₃O₄-C-PVDF, Fe₃O₄ + 6 wt% PPy-C-PVDF, and Fe₃O₄ + 20 wt% PPy-C-PVDF, respectively. The electrodes without carbon additives delivered cycle 10 capacities of 26 mAh/g (0.2 electron equivalents), 80 mAh/g (0.7 electron equivalents), and 22 mAh/g (0.2 electron equivalents) for Fe₃O₄-PVDF, Fe₃O₄ + 6 wt% PPy and Fe₃O₄ + 20 wt% PPy, respectively. The PPy-C-PVDF electrode delivered a 10^th cycle capacity of 74 mAh/g (0.6 electron equivalents).

Zoom In Zoom Out Reset image size

The specific capacities versus cycle numbers to 20 cycles are shown in Figure 5A with the results for Fe₃O₄ free PPy electrodes provided for reference. The samples containing electrically conductive carbon retain capacity better than samples without electrically conductive carbon. Figure 5B shows data normalized to cycle 2 capacity to illustrate capacity retention for cycle testing under discharge of 50 mA/g and 100 mA/g. The Fe₃O₄-C-PVDF sample retained 78% the second cycle capacity when cycled at 50 mA/g. However, under 100 mA/g current, only 58% capacity retention was observed. In a similar fashion, the Fe₃O₄ + 6 wt% PPy-C-PVDF sample retained 83% and 51% capacity when cycled at 50 and 100 mA/g, respectively. In contrast, the Fe₃O₄ + 20 wt% PPy-C-PVDF sample showed capacity retention of 91% under both 50 and 100 mA/g discharge rates. The cycle number versus capacity plots at 100 mA/g are shown in Figure S1.

Zoom In Zoom Out Reset image size

The electrochemical impedance spectra (EIS) of the coin cells were measured before cycling, after 10 cycles in the charged state, and after 10 cycles in the discharged state after cycling at 50 mA/g. The EIS data were fit to an equivalent circuit model with a DC electrolyte resistance and ohmic resistances element (R1), charge-transfer resistor (R2) and constant phase element (CPE) in parallel with the R2, and a Warburg element as shown in Figure 6 with the values shown in Table I. The charge transfer resistor (R2) varied considerably among the electrodes enabling us to examine the effect of PPy. In the electrodes with added carbon all of the pristine materials showed a charge transfer resistance of less than 20 Ω. The pristine Fe₃O₄-PVDF electrode has a resistance of 426 Ω while the pristine Fe₃O₄ + 6 wt% PPy and Fe₃O₄ + 20 wt% PPy electrodes had significantly smaller value of 8.7 and 21 Ω, respectively. Therefore, the PPy does maintain electronic conductivity in the pristine electrode. The 6 wt% PPy composite had a similar resistance to the electrode containing carbon (Fe₃O₄ + 6 wt% PPy-C-PVDF) with a pristine R2 value of 8.1 Ω. The R2 of Fe₃O₄ + 20 wt% PPy-C-PVDF was 14 Ω, lower than the Fe₃O₄ + 20 wt% PPy only electrode illustrating the contribution of the electrically conductive carbon. Notably, the pristine sample PPy-C-PVDF showed a R2 of 15 Ω, similar to the pristine Fe₃O₄ + 20 wt% PPy-C-PVDF.

Zoom In Zoom Out Reset image size

After charge and discharge of all electrodes containing carbon the R2 was below 40 Ω where the highest R2 was observed for the charged Fe₃O₄-C-PVDF at 37 Ω (Table I). The Fe₃O₄ + 20 wt% PPy-C-PVDF and PPy-C-PVDF had similar R2 values when charged while the Fe₃O₄ + 6 wt% PPy-C-PVDF has the lowest R2 in the charged state of 19 Ω. In the discharged state, the Fe₃O₄-C-PVDF and Fe₃O₄ + 6 wt%-C-PVDF had similar values of 22 and 18 Ω, respectively.

Without PPy or electrically conductive carbon, the Fe₃O₄-PVDF sample showed a high resistance of 210 Ω in the charged state, Table I. The sample with 6 wt% PPy showed an increase of ∼130 Ω after cycling. This increase was even larger for 20 wt% composite which had an increase of ∼371 Ohms between the pristine and charged material after 10 cycles.

Ex-situ X-ray Absorption Near Edge Structure (XANES) was used to investigate the observed differences in the electrochemistry of the Fe₃O₄-PPy composites with and without additional carbon and binder. XANES was a useful tool for this study as it probes the oxidation state of the metal center and does not require high levels of crystallinity. In Figure 7, the absorption edge of the electrodes with carbon and binder are shown after 10 cycles in the discharged (Figure 7A) and charged (Figure 7B) state of the cells cycled at 50 mA/g. The spectra of reference materials magnetite (orange), Fe metal (green) and FeO (gray) are shown in the figure. In the discharged state, the samples all show edge positions at lower energy than Fe₃O₄ consistent with the samples having reduced iron centers, in agreement with previous XAS investigations of discharged Fe₃O₄.^1,26 Notably, the Fe₃O₄-C-PVDF and Fe₃O₄ + 6 wt% PPy-C-PVDF sample had spectra that were similar and indicating an oxidation state close to that of Fe⁰. In contrast, the Fe₃O₄ + 20 wt% PPy-C-PVDF sample indicated iron at an oxidation state less reduced compared to the other two samples with an intermediate oxidation state. These observations are consistent with the observed discharge capacities for each sample type.

Zoom In Zoom Out Reset image size

After 10 cycles, where the samples were isolated in the charged state, the edge positions of the electrode compositions containing carbon approach the magnetite edge, indicating a higher oxidation state of Fe present in the charged state of the electrodes. The Extended X-ray Absorption Fine Structure (EXAFS) k² weighted-R-space spectra for the charged electrodes are presented in Figure 7C. Even though the iron oxidation state is high, from the EXAFS spectra, it is evident that the local atomic structure of the charged products differed from that of the magnetite reference material, consistent with the moderate functional capacity for the carbon containing electrodes. In contrast, the electrodes that were constructed without carbon delivered less than 1 electron equivalent after 10 cycles and the XANES Fe edge as shown in Figure S2. The edge energies are summarized in Table S1. Both the discharged and charged state of the carbon free samples are similar to the Fe metal absorption edge, consistent with the small capacity observed during cycling for these samples.

The chemical oxidation of polypyrrole has been studied previously and is depicted as the oxidation step in the reaction scheme in Figure 8.²⁷ Variations in pH and solvent have also been investigated and showed that a higher acidity is favorable to prepare electrically conductive polypyrrole (PPy).^28,29 Therefore, p-toluenesulfonic acid (HTos) was used in excess to maintain the pH at ∼ 1.5, promoting Feⁿ⁺ dissolution from Fe₃O₄ and initiate oxidative pyrrole polymerization. Slow dissolution of magnetite may generate hematite (α-Fe₂O₃) and wustite (FeO) as products.^30,31 In this study neither of those phases were observed in the XRD or Raman of the solid Fe₃O₄-PPy composite product. Therefore, we propose that any dissolved magnetite results in the direct formation of the Fe^z+ in the acidic solution as shown in the dissolution step, Figure 8. The reduction step occurs as the dissolved Fe³⁺ is reduced to Fe²⁺. The balanced overall reaction indicates that for every 7 Fe³⁺ ions that are dissolved 3 pyrrole monomers should react to form the electrically conductive PPy, resulting in a 2.33 Fe/pyrrole ratio. After a 4 hr reaction 5.3 ± 1.1% of the magnetite dissolved in the reaction solution resulting in a measured ratio of Fe/py 1.5 ± 0.4, where the Fe content was monitored by inductively coupled plasma-optical emission spectroscopy (ICP-OES). After 24 hours 12.2 ± 0.7% of the magnetite dissolved and the Fe/py ratio was 2.2 ± 0.3, within error of the 2.33 nominal value based on the scheme shown in Figure 8.

Zoom In Zoom Out Reset image size

The overall reaction of Fe₃O₄ to Fe metal is an 8 e⁻ reaction and has been previously characterized with additional electron equivalents attributed to solid electrolyte interphase, SEI, formation.^32–34 The electron equivalents delivered in the 1^st, 2^nd, 10^th, and 20^th cycle for each sample configuration are shown in Figure 4. After 10 cycles, the Fe₃O₄-PVDF shows 0.9 e⁻ equivalents, indicating that the conversion of Fe₃O₄ to Fe⁰ is irreversible in that electrode condition. The Fe₃O₄-C-PVDF has far superior cycling capability delivering 6.3 e- equivalents after 10 cycles, showing the conversion efficiency of the electrode is higher with carbon present.

The addition of PPy to the Fe₃O₄ electrode is intended to provide additional electronic conduction pathways within the electrode. Interestingly, however the interaction of the electrode with the electrolyte was impacted by the presence of PPy. The difference in capacity between cycles 1 and 2 decreased with increasing PPy content in the electrode, indicating a lower irreversible capacity with more PPy present. For example, the first and second cycle discharge capacities of the Fe₃O₄-C-PVDF electrode were 1404 mAh/g (12.1 electron equivalents) and 801 mAh/g (6.9 electron equivalents) respectively, equivalent to an overall capacity difference of 603 mAh/g (5.2 electron equivalents), while the Fe₃O₄ + 6 wt% PPy-C-PVDF and Fe₃O₄ + 20 wt% PPy-C-PVDF electrodes showed smaller capacity differences from the first to second cycle of 445 mAh/g (3.9 electron equivalents) and 382 mAh/g (3.3 electron equivalents). The smaller difference in capacity between the first and second cycle with higher levels of PPy provides evidence that SEI generation decreases with an increased amount of PPy.

After 10 cycles, the Fe₃O₄ + 6 wt% PPy delivered only 0.6 e⁻ equivalents. The carbon containing sample, Fe₃O₄ + 6 wt% PPy-C-PVDF, delivered 5.0 e⁻ at the 10^th cycle. The increase in discharge capacity in the presence of the electrically conductive carbon indicates that the carbon improves the electrically conductive network and PPy alone is not as effective. This same trend is observed in the sample containing 20 wt% PPy where the carbon containing sample delivers higher capacity and shows the best capacity retention of the group over 50 cycles, Figure S3.

After cycling, the XANES absorption edge for the Fe₃O₄ + 20 wt% PPy-C-PVDF sample in the charged state is more similar to the oxidation state of magnetite edge than the other electrode compositions, Figure 7B. However, the Extended X-ray Absorption Fine Structure (EXAFS) k² weighted-R-space spectra, Figure 7C, indicate that the atomic structure of the recharged iron is significantly different than the Fe₃O₄ standard, which consists of major peaks centered at 1.3 Å (contributions from nearest neighboring oxygen atoms for octahedrally and tetrahedrally coordinated Fe atoms) and 2.6 Å with a broad shoulder at 3.1 Å (contributions from neighboring iron atoms and second coordination shell oxygen atoms). The recharged material, in contrast, has a single major peak a 1.2 Å, with the amplitude of the second shell peak greatly reduced. The spectra are indicative of the Fe atoms being in a highly disordered Fe-O like state where neighboring iron atoms are only a small contribution to the observed amplitude.³⁵ Thus, while the XANES data indicate that the charged electrodes have high oxidation state (between 2+ and 3+), EXAFS shows that the Fe atoms after recharge exist in a disordered oxide phase that is structurally different from Fe₃O₄. Although the Fe₃O₄ + 20 wt% PPy-C-PVDF composite charges to an oxidation state similar to that of magnetite, it does not fully discharge to Fe metal.

The capacity retention of the Fe₃O₄ + 20 wt% PPy-C-PVDF was the best for the group. High capacity retention of the Fe₃O₄ + 20 wt% PPy-C-PVDF sample can be rationalized by ion transport facilitated through lowered SEI formation resulting from electrolyte decomposition which has been implicated as a significant contributor to capacity fade for conversion materials.³⁵ The role of the PPy in providing favorable electrochemistry of composite electrodes containing conversion materials is related to its ability to provide favorable ion transport rather than its role in increasing electrical conductivity.