π‐Conjugation Enables Ultra‐High Rate Capabilities and Cycling Stabilities in Phenothiazine Copolymers as Cathode‐Active Battery Materials

In recent years, organic battery cathode materials have emerged as an attractive alternative to metal oxide–based cathodes. Organic redox polymers that can be reversibly oxidized are particularly promising. A drawback, however, often is their limited cycling stability and rate performance in a high voltage range of more than 3.4 V versus Li/Li+. Herein, a conjugated copolymer design with phenothiazine as a redox‐active group and a bithiophene co‐monomer is presented, enabling ultra‐high rate capability and cycling stability. After 30 000 cycles at a 100C rate, >97% of the initial capacity is retained. The composite electrodes feature defined discharge potentials at 3.6 V versus Li/Li+ due to the presence of separated phenothiazine redox centers. The semiconducting nature of the polymer allows for fast charge transport in the composite electrode at a high mass loading of 60 wt%. A comparison with three structurally related polymers demonstrates that changing the size, amount, or nature of the side groups leads to a reduced cell performance. This conjugated copolymer design can be used in the development of advanced redox polymers for batteries.


Introduction
Organic cathode materials have been identified as promising candidates for next-generation battery systems. [1][2][3][4][5][6][7][8][9][10][11][12] Their advantages include a high structural diversity and design flexibility as well as a low toxicity. In addition, they are accessible from less-limited resources compared to their inorganic counterparts. [13,14] Redox polymers are among the best organic potentials due to the well-defined phenothiazine redox centers as well as ultra-high rate capability, and long-term cycling stability at 100C rate in the case of P1a, because of the semiconducting nature of the polymers. With 3.6 V versus Li/Li + , the discharge potentials in composite electrodes lay close to the operation potential of commercial Li-ion battery cathodes. [25] In addition, they contained a high mass-loading of 60 wt% active material. P1b-d served to elucidate the influence of side chains on the battery performance of the polymers, and P3 was used to assess the influence of π-conjugation ( Figure 1).
The concept of a π-conjugated redox polymer as battery electrode material has been demonstrated before using the well-known n-type NDI-bithiophene copolymer P(NDI2OD-T2) with a discharge potential of about 2.4 V versus Li/ Li + , [26] as well as the corresponding fluorene copolymer. [27] In this work, a p-type conjugated polymer with phenothiazine was chosen based on its reversible oxidation chemistry, high oxidation potential of 3.6 V versus Li/Li + , and excellent performance in battery cathode materials. [28][29][30][31][32][33] We recently demonstrated high cycling stabilities at 10C rate using poly (3-vinyl-N-methyl-phenothiazine) due to supramolecular hole-transport. [28,29,34] In order to separate the redox-active phenothiazine moieties and to allow for good hole conductivity, we used bithiophene and fluorene as comonomers in P1 and P2, respectively. An aryl ether group was chosen as substituent R to obtain highly reversible oxidation processes of the phenothiazine group upon oxidation, [35] and alkyl groups were attached in P1a, P2, and P3 to ensure solubility of the polymers for purification and preparation of composite electrodes. Conjugated copolymers of phenothiazine with fluorene and bithiophene have been reported before for use in organic light-emitting diodes or photovoltaic cells, [36][37][38][39][40][41] but not for battery applications.

Synthesis of Polymers P1a, P2, and P3
The required phenothiazine monomer 1 was synthesized in three steps starting from phenothiazine (2, Scheme 1). Buchwald-Hartwig reaction allowed introducing the aryl ether moiety on the nitrogen atom, followed by twofold bromination to yield 3. The bromo substituents were then replaced by boronic ester moieties to obtain 1. Suzuki-Miyaura polycondensations of phenothiazine monomer 1 with the corresponding dibromides 4-6 yielded polymers P1a, P2, and P3, respectively (Scheme 1). 4 [42] and 5 [43] were synthesized according to the literature, and the synthesis of 6 can be found in Supporting Information. In order to obtain P1-P3 with high molecular weights and in high yields, we performed an extensive optimization of reaction conditions, in particular concerning the catalyst and reaction mode (see Supporting Information for details). High molecular weights are important to ensure insolubility of the polymers in the battery electrolyte. Tris(dibenzylideneacetone)dipalladium with the ligand SPhos and sodium hydroxide as base in toluene/ water (3/1) with Aliquat 336 as phase-transfer catalyst was the method of choice. We furthermore found that polymerization in a microwave synthesizer using SPS mode provided superior results and shorter reaction times compared to oil bath heating.

Electrochemical and Optical Properties of Polymers P1a, P2, and P3
Cyclic voltammetry (CV) measurements and UV-vis absorption and emission spectroscopy provided information about the electrochemical and optical properties of P1a, P2, and P3, as summarized in Table 1. In solution, P1a, P2, and P3 featured reversible oxidations at half-wave potentials of 0.20, 0.23, and 0.13 V versus Fc/Fc + , respectively (see Figure 2a-c). The second oxidations of the phenothiazine units in P1a, P2, and P3 occurred at 0.77, 0.90, and 0.69 V versus Fc/Fc + , respectively (see Figure S45, Supporting Information). P1a showed two further oxidation waves at potentials of 0.69 and 0.92 V versus Fc/ Fc + , likely because of the bithiophene units. Combining defined redox centers, as present in aliphatic redox polymers, with a conjugated polymer backbone leads to π-conjugated redox polymers P1a-d and P2 with well-defined oxidation potentials and hole transport along the polymer backbone. P3 with broken conjugation served for comparison.
The absorption spectra of P1a, P2, and P3 (see Supporting Information) displayed two main bands with maxima between 307 and 410 nm ( Table 1). The optical band gaps lay between 2.44 and 2.66 eV. All three polymers were fluorescent with emission maxima of 519, 483, and 474 nm for P1a, P2, and P3, respectively (see Table 1 and Figure S56, Supporting Information).

Electrochemical Testing of Polymers P1a, P2, and P3 as Cathode-Active Battery Materials
For the investigation of P1a, P2, and P3 as cathode-active materials in batteries, we fabricated composite electrodes containing a high ratio of 60 wt% polymer, 35 wt% Super C65 as conductive additive, and 5 wt% PVdF binder. Metallic lithium was used as the counter and reference electrode and 1 m LiPF 6 in EC:DMC 1:1 as electrolyte. Scanning electron microscopic (SEM) measurements showed that the porous, percolated network structure of the carbon black Super C65 was maintained in the composite electrodes. The surface of the carbon network was evenly coated with the polymers, as confirmed by energy-dispersive X-ray spectroscopy (EDS) measurements (see Figures S62-S70, Supporting Information, for SEM and EDS images). As seen from the CVs (Figure 2d,e), the oxidations of P1a, P2, and P3 in composite electrodes occurred at potentials of 3.68, 3.70, and 3.59 V versus Li/Li + , respectively, around 0.2 V higher than would be expected from the values in solution (assuming 3.25 V [45] for Fc/Fc + versus Li/Li + ). For P1a and P2, the oxidation peaks were narrow, indicating well-defined, faradaic redox processes, while P3 showed broader peaks. This stands in contrast to polymers solely containing phenothiazine units with broad oxidation peaks. [31,46,47] At 3.9/4.0 V versus Li/Li + , the second oxidations started setting in, visible in the cathodic scans (see Figure S45, Supporting Information).
Constant current cycling measurements at 1C rate (Figure 3a-d) in the range of 3.2-3.9 V versus Li/Li + showed stable cycling behavior. All three polymers reached a value close to their theoretical capacity for one oxidation of each subunit, which amount to 36.4 mAh g −1 for P1a and P2 and 35.1 mAh g −1 for P3 (see also Scheme 1). [48] After 100 cycles, capacities of 34.7, 33.4, and 31.0 mAh g −1 were obtained for P1a, P2, and P3, respectively. For P1a, no loss in capacity was measured, while for P2 the capacity had dropped by only 2%, demonstrating the high cycling stability of both polymers. The coulombic efficiencies lay between 97.6% and 99.6% (after cycle ten). The Adv. Funct. Mater. 2019, 29,1906436 Scheme 1. Synthesis of polymers P1a, P2, and P3 and theoretical specific capacities. Table 1. Electrochemical and optical data for P1a, P2, and P3 in solution and in composite electrodes.  [44] ; d) In CH 2 Cl 2 ; e) Optical band gap from the onset of the longest wavelength absorption band.
charge/discharge curves (Figure 3a-c) possessed flat plateau potentials for all three polymers at discharge potentials of 3.6 V for P1a and P2 and 3.4 V for P3 (all vs Li/Li + ). This shows that the phenothiazine units in all three polymers, even in conjugated P1a and P2, acted electrochemically separated, and that the oxidation potentials of the polymers were not affected by the charging state of the battery. This is a clear advantage in comparison to π-conjugated polymers, where usually sloping potential curves are obtained. [4,23] The self-discharge, measured for P1a-based cells, was low with only 11% capacity loss within 3 days (see Figure S51, Supporting Information). This is comparable to or even better than published values for organic cathode materials. [18,49] Extending the potential range to 4.1 V versus Li/Li + in the upper region allowed accessing a higher capacity for P1a (see Figure S54a, Supporting Information). However, a lower cycling stability resulted. The conjugated polymer backbone in P1a and P2 allowed for an ultra-high rate capability, as can be seen in the C-rate test in Figure 3e. Even at 100C, corresponding to a current density of 3.6 A g −1 (1.5 mA cm −2 ), a specific capacity of 32.7 and 30.9 mAh g −1 was accessible for P1a and P2, corresponding to 90% and 85% of the theoretical value, respectively. In P3, on the other hand, where conjugation is interrupted, rates higher than 10C provided no capacity of the active material, and even at 10C rate, the capacity significantly declined within six cycles. Longterm cycling at 10C rate (for charge and discharge) showed that bithiophene copolymer P1a was superior to fluorene-based copolymer P2 and non-conjugated polymer P3 (Figure 3f   33.1 mAh g −1 . For P2, the capacity dropped to 72% of the initial value after 2200 cycles (23.0 mAh g −1 ), while P3 did not show any capacity of the active material at 10C rate. The same result was obtained from measurements of different cells. The higher cycling stability of bithiophene copolymer P1a compared to fluorene copolymer P2 at 10C rate was likely due to the electrondonating and stabilizing effect of the bithiophene units on the oxidized phenothiazine groups in P1a compared to the fluorene units in P2.
As visible from electrochemical impedance spectroscopy (Figure 4 and Figure S55, Supporting Information), P3-based composite electrodes showed a significantly higher chargetransport resistance in the discharged state (0% SOC) of 259 Ω compared to P1a (24 Ω). With increasing state-of-charge, this resistance further increased in P3 as did the double-layer capacitance due to the poor hole conduction of the material (965 Ω at 100% charge). In P1a-based composite electrodes, on the other hand, the charge-transport resistance decreased with increasing state-of-charge (14 Ω at 100% charge), demonstrating good hole conduction of the material.
Encouraged by these results, a long-term cyclization at the ultra-fast rate of 100C was performed with P1a-based composite electrodes. An excellent cycling stability was observed ( Figure 5 and Figure S52, Supporting Information, for an average of three different cells). A maximum capacity of 30.1 mAh g −1 (83% of the theoretical value) was reached after 15 000 cycles, which only dropped by 2.5% up to cycle 30 000. The charge/discharge curves showed that even at this high rate, plateau potentials were obtained with a value of 3.6 V versus Li/Li + for the discharge (Figure 5b). The differential capacity plots (see Figure S49, Supporting Information) showed well-defined peaks for charge and discharge. The specific energy of the final discharge cycle 30 000 amounted to 108.4 Wh kg −1 , corresponding to a specific power of 10 836 W kg −1 of the active material. This is remarkable in comparison to established inorganic cathodes for lithium batteries [50] with specific energies of 80-250 Wh kg −1 and specific power values of 200-4500 W kg −1 , which often times have a cycle life of only up to 2000 full charge/discharge cycles. [50,51] To the best of our knowledge, such a fast and long-term cycling stability at a potential of 3.6 V versus Li/Li + is unprecedented for organic cathode materials. A literature survey (see Supporting Information for details) showed that organic electrodes with the highest reported rate capabilities employing at least 25 wt% active material and with discharge potentials ≥3.4 V were based on TEMPO (2,2,6,6-tetramethylpiperidinyl-N-oxyl), [52,53] phenothiazine [28,29] or triarylamines [54][55][56] as redox-active groups. However, a long-term cyclization at 100C has been reported only once with 5000 cycles and a material loading of 40 wt%. [54] We ascribe these results to the conjugated copolymer structure, where defined redox processes take place, but at the same time hole conduction along the polymer backbone is possible. As we will show later through substituent modification, the aryl ether group on the phenothiazine and the alkyl groups on the bithiophene units provided an ideal compromise between well-defined phenothiazine redox centers and a sufficient degree of conjugation along the polymer backbone.

Mechanistic Investigation of Charge/Discharge Processes in P1a and P2
We next investigated whether the observed redox processes in P1a-and P2-based cells were faradaic or capacitive in nature because their ultra-high rate capability was reminiscent of capacitors. The CVs, featuring distinct oxidative and reductive peaks (Figure 2), and the charge/discharge curves with well-defined plateaus (Figures 3 and 5) provided evidence for a battery-type behavior. [57] The differential capacity plots (see Figure S49, Supporting Information) showed narrow and well-defined peaks for P1a and P2, even at the fast rate of 100C for P1a, which further underlines the faradaic nature of the redox processes. Furthermore, the change of the current i with the scan rate ν at a fixed potential, as extracted from the CVs at different scan rates, was investigated. The relationship between the peak current i and ν can be expressed as where k and b are adjustable parameters. [58] It has been suggested that if b has a value of 0.5, the process can be considered  faradaic, while if b equals 1, the process is capacitive. [58] Evaluating the peak current for oxidation and reduction (plotting log(i) vs log(ν)) resulted in b-values of 0.55 for P1a and 0.51 for P2, demonstrating the faradaic nature of the redox processes ( Figure S48, Supporting Information).
With the calculated k-values in hand, the capacitive contribution to the total current at a fixed potential V was evaluated using the equation [59] i V k k ν ν As can be seen in Figure 6, the capacitive contribution was very low for the redox process of both P1a and P2. In particular in P1a-based cells, 91% of the capacity stemmed from faradaic processes.
Calculations served to rationalize the different cycling behavior of polymers P1a and P2. In benchmark calculations, we chose oligomer model systems O1a and O2 for polymers P1a and P2, respectively, containing two repeating units of the respective polymer with symmetric endings (see Figure 7 and Supporting Information for details). Calculations were performed at the wB97XD [60] /6-31G(d) [61,62] theory level for structure optimizations followed by single-point energy calculations at the wB97XD/6-311G(d,p) [63][64][65] theory level for the density of states (DOS, obtained using the AOMix package [66,67] ) and spindensity (for further detail see Supporting Information).
The DOS (Figure 7a) shows that for O1a, both bithiophene and phenothiazine moieties contributed to the HOMO composition (DOS between −6 and −8 eV). For O2, on the other hand, the major contribution stemmed from the phenothiazine units, while the fluorene moieties played a negligible role. This indicated a higher conjugation degree for O1a than for O2, which could explain the higher rate capability of P1a compared to P2. The spin densities obtained for the first oxidized state are shown in Figure 7b. Both are mostly localized on the phenothiazine units.
In all three oligomer model structures O1a, O2, and O3, the phenothiazine unit planarized upon oxidation from a "butterfly" angle of about 149° to close to 180° (see Supporting Information for details). In addition, in O1a, the torsional angle between the phenothiazine groups and the adjacent thiophenyl groups became smaller (54° to 35°). This is in line with the enhanced conjugation in polymer P1a.

Influence of Side Chains on the Cell Performance of P1
Based on the excellent rate performance and cycling stability of bithiophene-based polymer P1a, we next aimed to increase its   In polymer P1b, all alkyl chains were removed apart from a methyl group at the nitrogen atom. In P1c and P1d, the aryl ether unit on the nitrogen was maintained, but the branched alkyl chain was replaced with a methyl group. In P1c, the bithiophene groups remained unsubstituted, while in P1d, methyl substituents were attached. This furthermore allowed us to assess the influence of substituents on the cycling behavior of the polymers, resulting from electronic and conformational differences.

Synthesis of Polymers P1b-d
The syntheses of P1b-d employed phenothiazine monomers 7 and 8 together with bithiopene monomers 9 and 10, as shown in Scheme 2. Polycondensations were performed using the optimized Suzuki-Miyaura conditions, as discussed above for P1a, P2, and P3. The resulting polymers P1b-d were insoluble in common organic solvents due to the lack of alkyl side chains and precipitated from the reaction mixture. Hence, molecular weights could not be determined. Consecutive Soxhlet extraction with methanol, acetone, ethyl acetate, and dichloromethane allowed obtaining the polymers in sufficient purity for battery cell measurements.

Investigation of Polymers P1b-d as Cathode-Active Battery Materials
To investigate P1b-d as cathode-active materials in batteries, we fabricated composite electrodes as described for P1a, P2, and P3 above in the ratio of 60 wt% polymer, 35   In cyclic voltammograms in composite electrodes, P1b showed a somewhat narrow oxidation peak but a broad re-reduction peak on the cathodic scan and a large peak separation (Figure 8). This likely resulted from the stronger conjugation between the phenothiazine and bithiophene units due to the absence of alkyl chains. Similar electrochemical behavior has been observed in phenothiazine homopolymers. [31] A calculation on the radical cation of the corresponding model compound O1b revealed that the spin density was mostly localized on a bithiophene unit and only half of the adjacent phenothiazine groups (Figure 9b). As a result, the phenothiazine units did not planarize upon oxidation (Figure 9a). This stands in contrast to aryl ether-substituted O1a, O1c, and O1d, where the radical cation was centrally localized on the phenothiazine units, which, as a consequence, planarized upon oxidation (see Figure 9 and Figure S80, Supporting Information). Hence, in polymer P1b, the bithiophene groups significantly partook in the oxidation reaction. [68] In P1c and P1d, containing an aryl ether substituent on the phenothiazine group, the oxidations were more defined (Figure 8). This had also been the case for P1a (see Figure 2) and is in line with the calculated planarization of the phenothiazine units upon oxidation in all three model compounds O1a, O1c, and O1d (see Figure 9). Introducing alkyl substituents on the bithiophene units led to a further narrowing of the redox peaks in P1d (see also P1a in Figure 2). This localization of charges due to the presence of alkyl substituents in thiophene copolymers has been demonstrated in other contexts before. [69,70] In constant current cycling measurements, P1b, featuring the highest theoretical capacity, provided the highest specific capacity at 1C rate, while P1d showed the best rate capability (Figure 10). At 1C rate, all three polymers demonstrated stable cycling behavior. For P1b, a reversible capacity of 56.9 mAh g −1 was accessible, which amounted to 80% of the theoretical value for a one-electron oxidation (the theoretical capacities can be found in Scheme 2). For P1c, 41.0 mAh g −1 were accessible (72% of the theoretical value), and for P1d, 38.9 mAh g −1 (72% of the theoretical value). The incomplete capacity use in these polymers was likely either due to the presence of polymer agglomerates in the composite electrode caused by their insolubility during processing (for SEM images, see Figures S71-S76, Supporting Information) or to the fact that the polymer chains were short and electrochemically inactive end groups reduced the specific capacities. Extending the potential range to 4.1 V versus Li/Li + in the upper region allowed accessing higher capacities for all three polymers P1b-d with stable cycling at 1C rate for P1b and P1c (see Figure S54b-d, Supporting Information).
The charge/discharge curves (Figure 10d sloping potential curves were obtained, while P1d showed more defined plateaus. This again can be explained by the presence of the methyl groups on the bithiophene units in P1d, which had a pronounced influence on the localization of charges on the phenothiazine units. P1d featured the best rate performance followed by P1c and P1b ( Figure 10). For P1d, a capacity of 31.5 mAh g −1 was accessible at 100C rate (Figure 10c). The rate capability of P1c was also good, which again can be traced back to the presence of the aryl ether groups on the phenothiazine units. The poorer rate performance of P1b, on the other hand, might be due to 1) the missing planarization of the phenothiazine groups upon oxidation and 2) the larger torsional angle between the thiophene units in the oxidized form compared to P1c and P1d (see Figure 9).
In summary, in spite of the larger amount of inactive mass in P1a compared to P1b-d, this polymer showed the best performance regarding accessible capacity, rate capability, and cycling stability. This was due to the electronic effects of the aryl ether group, the steric influence of the bithiophene alkyl groups on the polymer conformation, as well as its high solubility, which allowed for good processing of the composite electrodes. The long alkyl groups in P1a likely favor formation of an amorphous phase in the composite electrode, which accelerates counter ion diffusion, a phenomenon, which has been observed in other redox-active polymers before. [71]

Conclusions
In conclusion, we have shown that using a π-conjugated copolymer structure enabled ultra-high rate capability and cycling stability at 100C rate in phenothiazine-based polymers as cathode-active battery materials. Best results were achieved with P1a, possessing alternating phenothiazine and bithiophene units. We ascribe this to the semiconducting nature of P1a, which allowed for fast charge transport in the composite electrode. Yet, due to the copolymer structure, localized phenothiazine redox centers were maintained along the polymer chain, which resulted in well-defined plateau potentials on charge and discharge at 3.6 V versus Li/Li + , close to the operating potential of commercial Li-ion battery cathodes. The redox processes were mostly faradaic in nature (>90%). Calculations showed that a slightly higher degree of conjugation in P1a compared to fluorene-based copolymer P2 was likely responsible for its superior rate performance. A comparison with polymers P1c-d carrying fewer and smaller substituents demonstrated that both the aryl ether as well as the alkyl substituents in P1a were necessary to obtain this outstanding behavior. We believe that this concept of using a π-conjugated copolymer structure is attractive for applications, where a high rate performance of the electrode is required.

Experimental Section
Synthetic Procedures: Further details on materials and methods as well as synthetic procedures for compounds 1-10 and computational details can be found in Supporting Information.
Microwave reactions were conducted in a Discover S-class microwave oven from CEm. NMR spectra were recorded at room temperature on a Bruker Avance III HD 500, Bruker Avance Neo 400, Bruker Avance II 400, or Bruker Avance III HD 300 spectrometer. Chemical shifts were reported in parts per million (ppm, δ scale). The 1 H and 13 C spectra were calibrated against the residual proton and natural abundance 13 C resonances of CDCl 3 ( 1 H: 7.26 ppm, 13 C: 77.16 ppm) or against TMS ( 1 H, 13 C: 0 ppm). The following abbreviations were used: s = singlet, d = doublet, t = triplet, m = multiplet, and br = broad. The coupling constants (J) were indicated in Hertz (Hz). Analytical gel permeation chromatography was performed on a SECcurity GPC System from PSS Polymer Standard Service using components of the 1260 Infinity series from Agilent Technologies (IsoPump: G1310B, auto sampler: ALS G1329B, UV-detector: MWD VL G1365D, RI-detector: RID G1362A). Three columns were used (PSS SDV, 8 mm × 300 mm with a porosity of 10 2 , 10 3 , and 10 5 with integrated pre-column). As eluent, degassed THF was used with a flow rate of 1 mL min −1 . A polystyrene standard by PSS was used for calibration. For TGA, a STA 409 or a STA 449 F5 by Netzsch and for DSC a Netzsch DSC 204 F1 Phoenix were used. UV-vis absorption and emission spectra were recorded on a Tidas I or Tidas II diode array spectrometer from J&m Analytik AG, and on a Perkin Elmer LS55 in a 10 mm fused quartz cuvette. Cyclic voltammetry in solution was performed inside of a glovebox using a Metrohm Autolab PGSTAT 128N. As working electrode, a glassy carbon disc electrode (2 mm diameters) was used. A platinum rod served as counter electrode, and as reference electrode, a Ag/AgNO 3 electrode containing a silver wire immersed in an inner chamber filled with 0.1 m AgNO 3 containing 0.1 m n-Bu 4 NPF 6 in the outer chamber was used. For the internal reference, the ferrocene/ferrocenium (Fc/Fc + ) redox couple was used.
General Method for the Suzuki Polycondensation (Microwave Assisted): The following optimized method was used for polymers P1a, P2, P3, and P1b (the amounts of reagents and solvents are detailed for each polymer below). Phenothiazine diboronic ester 1 or 7, the respective dibromo-compound 4, 5, 6, or 9, Aliquat 336, NaOH, Pd 2 dba 3 , and SPhos were dissolved in degassed toluene. Argon-saturated H 2 O was added. The vessel was placed in the microwave reactor with the following settings: SPS-mode, 60 W, 90 °C, 30-90 min (depending on solvent volume) with air cooling on. After completion of the reaction time, 4,4,5,5-tetramethyl-2-phenyl-1,3,2-dioxaborolane was added and left to react for 4 min and bromobenzene was added, and the reaction was stirred for another 4 min (same microwave conditions as before). Then the mixture was precipitated from cold acetone. The filtered polymer was redissolved and precipitated from methanol and dried in vacuum.