Hydrothermal Generation of Conjugated Polymers Using the Example of Pyrrone Polymers and Polybenzimidazoles

Abstract Various polyimides and polyamides have recently been prepared via hydrothermal synthesis in nothing but H2O under high‐pressure and high‐temperature conditions. However, none of the prepared polymers feature a truly conjugated polymer backbone. Here, we report on an expansion of the synthetic scope of this straightforward and inherently environmentally friendly polymerization technique to the generation of conjugated polymers. Selected representatives of two different polymer classes, pyrrone polymers and polybenzimidazoles, were generated hydrothermally. We present a mechanistic discussion of the polymer formation process as well as an electrochemical characterization of the most promising product.


Introduction
Hydrothermal (HT) reactions have been performed for decades in preparative inorganic chemistry to generate am ultitude of different materials ranging from synthetic gemstones to quartz crystals and zeolites. [1][2][3] However,ithas just been in recent years that they were also successfully applied to generate an umber of advanced organic highperformance polymers.T od ate,o nly two classes of macromolecular compounds,t hat is,v arious polyimides (PIs) and polyamides (PAs), have successfully been generated hydro-thermally. [4][5][6][7][8] TheH Tp reparation of these polymers is characterized by experimental simplicity as well as inherent environmental friendliness and low cost, since neither sophisticated synthetic skills nor toxic solvents nor harmful catalysts are needed. In hydrothermal polymerization (HTP), H 2 Oisnot only the sole solvent and catalyst, it is also the only reaction byproduct formed (PIs and PA sa re condensation polymers). Hence,t he handling and disposal of hazardous substances can be greatly minimized.
We have strong indication to believe that, besides the outstanding properties of high-temperature water (HTW), asubstantial energetic driving force-for example,generated through the formation of strong bonds (i.e.b onds of high dissociation energy), cyclization reactions,c onjugation, and/ or crystallization-is beneficial for ap olymerization to proceed under HT conditions.F or example,w hen as uitable aromatic anhydride group (or the corresponding dicarboxylic acid) such as in naphthalene bisanhydride (NBA) is reacted with ap rimary amine (e.g. aniline), ac yclic imide moiety is formed via cyclocondensation (Scheme 1A top). [9] Generally, the HT formation of imides has already been extensively investigated:B oth small molecules [9] and am ultitude of different PIs [4-6, 8, 10] can be efficiently synthesized. In the example introduced above,t he formed imide moieties are cyclic and six-membered. Furthermore,t he imidesC = O groups and its Natom are linearly conjugated. However,the entire imide group is only cross-conjugated with the naphthalene moiety.I ti sg enerally known that the electronic communication between cross-conjugated groups is significantly weaker than that in linearly conjugated groups. [11] Moreover,c ross-conjugation also leads to ad rastically reduced conductance and, hence,n egatively affects charge transport properties. [12] However,when instead of an aromatic monoamine group (to form an imide) an aromatic o-diamine moiety is reacted with an aromatic anhydride moiety (as in NBA; Scheme 1A bottom), ad ouble cyclization can occur. Consequently,a ni midazole group fused to ac yclic amide function is generated, which is an even stronger linking function than an imide. [13] Additionally,b yc onnecting the starting compounds via double cyclization, linear conjugation between the building blocks can be achieved. However,such aH Td ouble cyclization has not yet been reported for generating apolymer.There is just one report on synthesizing as mall molecule via HT double cyclization:r eacting ophenylene diamine (o-PDA) with NBAy ields an isomeric mixture of the carbonyl dye perinone (Scheme 1A bottom). [13] As aconsequence of perinoneshighly aromatic and conjugated structure,t he molecule is fully planar. Planarity, aromaticity,a nd conjugation impart properties such as extreme thermal and chemical stability and it would clearly also be highly interesting to implement these features in an organic polymer.G enerally,b ya pplying this perinone-type linking chemistry and reacting suitable aromatic bis(o-diamine)s with aromatic dianhydrides (or their tetracarboxylic acids), the corresponding pyrrone polymers (PPs,S cheme 1B,C) can be obtained. These materials were intensively investigated during the 1960s for their high T-a nd radiation resistance,w hich was considered as promising for the aerospace sector. [14,15] More recently,PPs have attracted attention for electronics and optics applications (especially for nonlinear optical properties) due to their conjugated nature. [16][17][18][19] Furthermore,c onsidering that PIs and quinone-decorated polymers have been studied as aqueous battery electrodes in neutral and acidic electrolytes, [20][21][22] we anticipated that PPs could exhibit useful electrochemical properties in aqueous electrolytes as well. Just like HTP,a queous batteries take advantage of the environmentally benign nature of H 2 O. [23] Taking inspiration from our recent reports on the HTP of various PIs as well as the HT preparation of perinone,w e have now set out to both expand the number of hydrothermally obtainable polymers and to generate polymers with ac onjugated backbone by HTP.W ec onsider this as at ruly crucial step towards lifting HTP to an ew level of broader applicability.T he two most prominent representatives of PPs are shown in Scheme 1B,C.Note that the type of dianhydride employed determines the size of the cyclic amide moiety linking the two comonomers.W hile NBAl eads to as ixmembered cyclic amide,p yromellitic dianhydride (PMDA) gives rise to af ive-membered cyclic amide moiety.T he corresponding semiladder-type PPs will herein be referred to as PP6 (aka BBB) and PP5, respectively.Ingeneral, arandom distribution of cis and trans configurations of repeating units (r.u.s) along the polymer chain is expected. Thetwo common synthetic procedures for PPs are:( i)c ontrolled and stepwise heating of the neat comonomers in polyphosphoric acid yielding PP particles, [14,15] which can be processed by high-T molding; [24] (ii)s tirring the neat comonomers at room temperature (rt) in an aprotic,p olar solvent such as N,Ndimethylformamide (DMF) to obtain ap rocessable poly(amide amino acid) solution, which can be processed into,f or example,f ilms and fibers before af inal thermal curing step at T ! 300 8 8Cy ields PP. [25,26] Clearly,t hese latter procedures are tedious and far from being environmentally benign. Therefore,wewere aiming at afacile,green synthetic strategy towards PPs in nothing but "hot water".

HTP for Synthesis of PP6
When investigating the general feasibility of hydrothermally synthesizing PPs,w ea ttempted to keep the initially used comonomers as similar as possible to the precursors of perinone,a nd also applied the reaction conditions that had been found to be optimal for HT perinone formation. [13] Hence,N BA and 3,3'-diaminobenzidine (DAB) were used to attempt PP6 synthesis under HT conditions.Since the use of monomer salts (MSs) as precursors instead of working with neat comonomers has proven to be highly beneficial for the HTP towards PIs, [4][5][6] we intended to prepare aM So fD AB and naphthalene tetracarboxylic acid (NTCA;f rom hydrolysis of NBA) as the precursor for PP6 (Scheme 2). MSs often intrinsically provide equimolar comonomer stoichiometry, which is of prime importance for obtaining high-molecularweight products by ap olycondensation (CarothersL aw). Moreover,c ompared to neat amines,M Ss feature-at least partially-protonated NH 2 groups,w hich are more stable towards oxidation and hence storable for much longer periods of time without special precautions.Additionally,MSs can be expected to exhibit altered reactivity and solubility compared to neat comonomers due to their ionic nature,p reorganization in as alt crystal, and hence close spatial proximity of reactive groups already in the solid state. [27] TheM Ss uitable for synthesizing PP6 (abbreviated as MS6) was generated by precipitation in H 2 O( see the Supporting Information). MS6 was obtained as apurple solid. Analysis via attenuated total reflectance Fourier transform infrared (ATR-FTIR) spectroscopy and 1 HN MR spectroscopy confirmed its successful formation. TheA TR-FTIR spectra of MS6 and its starting compounds NBAa nd DAB clearly differ from each other ( Figure 1A). Them odes characteristic for the starting compounds (highlighted by colored boxes; n C=O (NBA) % 1770 cm À1 , n N-H (DAB) % 3450-3200 cm À1 ) [28,29] are absent in the MS6 spectrum. Instead, several modes indicative for the formation of MS6 (highlighted by arrows; n N-H (NH 2 ) % 3375 cm À1 , % 3220 cm À1 ; n N-H (NH 3 + ) % 2885 cm À1 , % 2585 cm À1 ; n C=O (COOH) % 1710 cm À1 ; n C=O (COO À ) % 1525 cm À1 )a re observed. Subsequently,amicrowave (MW)-assisted HTP towards PP6 using MS6 as precursor was carried out. Fort his,a n aqueous dispersion of MS6 (c = 0.01 mol L À1 )w as placed in ag lass liner that was transferred into aP TFE-lined stirred MW autoclave,a nd heated via MW irradiation within ah eating time (t H )o f1 0min to the desired reaction temperature (T R )of250 8 8C. T R was kept constant for areaction time (t R )o f1 5min. After cooling back to rt, the glass liner contained two different distinct layers:asolid, black sediment at the bottom and at ranslucent, colorless,c lear liquid supernatant. Theb lack color of the solid was already indicative of PP6 formation. ATR-FTIR spectroscopy of the solid (Figure 2A)c onfirmed as uccessful double cyclization through the presence of several characteristic PP6 modes, [15,30] which are similar to those of perinone: [13] n C=O (PP6) % 1700 cm À1 ; n C=C/C=N (PP6) % 1620 cm À1 (combined mode); n(benzimidazole) % 1450 cm À1 (in-plane vibration); n C-N (PP6) % 1310 cm À1 .Yet, intense modes occurring at % 1780 cm À1 and % 1740 cm À1 (Figure 2A,brown box) point to the presence of anhydride end-groups. [31,32] Clearly,f ree anhydride groups imply that also unreacted NH 2 moieties must exist in the generated polymer.H owever,t he N-H stretching modes are expected to be inherently weaker than the anhydride endgroup modes.I ntermolecular H-bonding with PP6sC =O groups produces significant band broadening and additionally lowers the mean absorption frequency. [33] Hence,o nly the indicative and well-pronounced anhydride C=Om odes are used here and in the following to estimate the success of aH TP experiment. Based on the intensity of these C = O modes,weexpect PP6 generated in this initial test experiment to be of relatively low molecular weight. In further consequence,v arious attempts (see the Supporting Information) were made in order to obtain products of higher molecular weight:n either lowering c,a ltering the pH, increasing t R , extending t H ,n or the addition of an on-nucleophilic base (which had been shown to promote HT imide formation) [9] led to adecrease in anhydride end-group mode intensities in the ATR-FTIR spectra. All these HTP experiments were carried out using the conventional, stirred MW-assisted setup at T R = 250 8 8C( which is our setupsm aximum continuous operation temperature). However,investigating T R s ! 250 8 8Cseemed to be another promising strategy for obtaining products of higher molecular weight. Upon elevating T,t he static dielectric constant of H 2 Oc ontinuously decreases, [34] which allows for asignificantly better dissolution of various organic compounds (especially aromatics) that are virtually insoluble Scheme 2. Synthesis of PP6:First, NBA hydrolyzes to NTCA which reacts with DAB to give MS6 (position of charges in the depiction of MS6 is arbitrary). Second, when MS6 is subjected to HT conditions, the double cyclization to form PP6 readily takes place. in H 2 Ounder ambient conditions. [35][36][37] Furthermore,classical PP synthesis also uses T R s ! 300 8 8C. [26] Fore mploying higher T R s, we had to switch our experimental setup to anonstirred high-temperature high-pressure autoclave (referred to as HPA; see the Supporting Information for details). Thesample preparation procedure remained the same as for the MWassisted experiments plus 10 bar of Argon pre-pressure was applied prior to heating. [71] In the following,H TPs of MS6 at various T R s( 250 8 8C, 275 8 8C, 300 8 8C, 325 8 8C, 350 8 8C) and t R s(15 min, 2h,12h)were carried out. Initial experiments at T R = 250 8 8Ci mmediately showed that the absence of stirring in the HPAmakes longer t R sc ompared to the MW-assisted stirred setup necessary for achieving the same results.F urthermore,a taconstant t R the anhydride end-group modes in the ATR-FTIR spectra diminished in intensity with increasing T R ( Figure 2B). While at t R = 2hthe modes are still visible for T R supto325 8 8C, they have entirely vanished for T R = 350 8 8C. For T R sof250 8 8Cand 275 8 8C, respectively,prolonging t R sdid not lead to adecrease in intensity of the end-group modes.H owever,f or T R so f 300 8 8Cand 325 8 8Canextension of the t R to 12 hresulted in the complete disappearance of anhydride end-group modes. Moreover,f or none of the samples were any modes found in the ATR-FTIR spectra indicating incomplete cyclization or nonlinear branching (e.g.a mide or imide). [26,31] Based on these results,weconclude that fully condensed, linear PP6 of different degrees of polymerization can be intentionally synthesized hydrothermally without the need for any cosolvents or condensation catalysts simply by adjusting T R and t R .
We then investigated the thermal stabilities of such fully condensed PP6 samples via thermogravimetric analysis (TGA;F igure 2C). An initial mass loss between rt and 120 8 8C, present in all samples,i sa ttributed to physisorbed H 2 O. It is well known that PP6 always contains ac ertain amount of physisorbed H 2 Od ue to H-bonding to the lonepair-bearing heteroatoms (N,O )i nt he polymer. [38,39] Taking the initial mass loss due to physisorbed H 2 Ointo account, the temperature of 5% mass loss (T 95 % )i s6 64 8 8C, and the temperature of 10 %m ass loss (T 90 % )i s7 22 8 8C. These high values are in accordance with the literature data for fully condensed PP6. [15,26,31] Furthermore,i tw as also possible to dissolve PP6 in methanesulfonic acid, which allowed for the fabrication of thin films following ap rocedure recently described in the literature. [40] UV/Vis absorption spectra of diluted solutions as well as of the fabricated thin films (see the Supporting Information) agree well with previous reports and indicate the generation of af ully conjugated polymer backbone. [41] Forlinear PIs and PA networks,HTpreparation gives rise to outstandingly high crystallinity superior to that of conventionally prepared analogues. [4,7] To date,e xclusively amorphous PPs have been prepared. Powder X-ray diffraction (PXRD) measurements of fully condensed PP6 from HTP are entirely identical for all samples (representative diffractogram in Figure 2D). Thep attern only contains two broad features:o ne with am aximum at 11.68 8 (2q,C u-K a )( labeled M1), and one with am aximum at 25.88 8 (2q,C u-K a )( labeled M2). While M2 corresponds to the interchain distance (n = 1; d hkl (25.88 8) % 3.5 )w hich originates from interchain p-stacking between the completely planar r.u.s,M 1c an be assigned to the length of the r.u. (n = 2; d hkl (11.68 8) % 15.2 ). These observations agree with the literature. [42,43] In contrast to PIs and PA sfrom HTP, [4,7] PP6 features no improved crystallinity compared to classically prepared PP6. Attempts to increase PP6sc rystallinity by performing HTPs using lower c(MS6) were not successful (see the Supporting Information). However,t he highly ordered PA networks reported by Stewart et al. were obtained by devitrification of initially prepared amorphous precursor networks based on the reversibility of amide bonds under HT conditions. [7] PA devitrification requires T R so f2 40-250 8 8Ca nd t R so f3 -7 days. [7] Inspired by this report, we resubjected our PP6 samples to HT conditions at different T R s(250 8 8C, 300 8 8C) for 7days hoping that also for PP6 an increase in crystallinity could be realized through similar bond reversibility.U nfortunately,n oc hanges were observed in the PXRD patterns (see the Supporting Information). We believe that the lack of reversibility must be related to the fact that aperinone-type linking is much more stable,l inearly conjugated, and significantly less prone to hydrolysis than an amide linking.A sA TR-FTIR spectra of PP6 before and after devitrification treatment were virtually identical, we conclude that PP6 is fully stable towards prolonged exposure to "hot H 2 O".
Forhydrothermally prepared linear PIs,aMSdissolutionpolymerization-polymer crystallization mechanism has been shown, and thus the transformation from the corresponding MS precursor to the PI is accompanied by asignificant change in morphology. [4,44] In contrast, scanning electron microscope (SEM) images of MS6 and PP6 revealed that their morphologies are in fact strikingly similar:Both are mainly composed of microsheets and needles ( % 2-10 mmi nl ength) that are agglomerated into bigger structures ( Figure 3A,B). Most interestingly,t his shape retention from MS6 to PP6 is found for all samples irrespective of the reaction conditions.T hus, we conclude that the conversion of MS6 to PP6 must mainly take place in the solid state,w hile dissolution or melting do not seem to occur. It is well known for various types of condensation polymers (PAs,p olyesters,P Is) that the heat treatment of suitable MS precursors can lead to solid-state polymerization (SSP) with retention of the MSss hape. [45,46] Such SSPs are typically performed under solvent-free conditions by heating the MS to a T below its melting point. [47] Yet, reports on SSP in dispersed media exist, [48,49] and for PIs, SSP during HTP is apossible reaction pathway that occurs to varying extent depending on the MS solubility in HTW and its SSP temperature. [5,10] Here,h owever,t he observation that MS6 and PP6 morphologies are virtually identical strongly points to the transformation basically exclusively occurring via SSP.T his rather unexpected finding raised the question whether the presence of H 2 Oi sa ta ll necessary for the transformation of MS6 to PP6 or whether just heating MS6 to T R s ! 250 8 8Cwould be sufficient. Thus,weperformed several SSPs,h eating neat MS6 to different T R s( 250 8 8C, 350 8 8C, 600 8 8C) under N 2 atmosphere.I ntriguingly,n one of the SSP experiments yielded fully condensed PP6. (i)SSP at 250 8 8C (t R = 2h,1 2h)g ave low-molecular-weight imide intermediates with anhydride end-groups (n C=O (imide) % 1710 cm À1 , % 1675 cm À1 ; n C=O (anhydride) % 1780 cm À1 , % 1740 cm À1 ;F igure 3C and the Supporting Information). Interestingly,w e never found such imides in any of the HTPs towards PP6. Note that T R = 250 8 8Cw as already sufficient for HT PP6 formation. (ii)SSP at 350 8 8C( t R = 2h)l ed to some perinonetype linking,but significant amounts of imide and anhydrideend group modes were still present (see the Supporting Information). (iii)A TR-FTIR analysis of al ast experiment performed at 600 8 8C( t R = 30 min;F igure 3C)c learly confirmed the generation of aPP6 product no longer containing residual imide moieties.H owever,a nhydride end-groups are still visible,i ndicating al ow degree of polymerization. This nicely aligns with observations by Morgan and Scott, who found that full condensation of MS6 to PP6 was only reached at % 630 8 8C. [50] However,i nt his T-regime decomposition already sets in, which limits the attainable degrees of polymerization. This SSP study strongly suggest that H 2 Op lays ac rucial role during the transformation of MS6 to PP6. First, HTW significantly lowers the necessary T R sf or obtaining PP6; second, it makes it possible to generate products of higher molecular weight;a nd third, it additionally has as trong influence on the mechanism of the transformation and facilitates the double cyclization towards PP6. In principle, several possible explanations are conceivable:( i)The increased ionic product of H 2 Ou nder HT conditions [51] could promote the cyclocondensation. [52] This would have to involve am igration of H + (H 3 O + )a nd/or OH À into the solid MS particles.N ote that for many (organic) materials-for example,aqueous dispersions/solutions of solid biomass,graphene oxide,o rs ugars-it is known that HT treatment can significantly facilitate dehydration reactions. [53][54][55] (ii)The autogenously arising pressure when H 2 Oi sh eated in an autoclave could play ab eneficial role in the transformation. There are several literature reports demonstrating the feasibility of reactive hot pressing towards PPs starting from either properly mixed neat comonomers or MSs as precursors. [50,56] However, T R so f% 450 8 8Ca nd pressures of at least 275 bar had to be applied in order to achieve full cyclization and dense products.T his compares to pressures of % 165 bar in HTP at 350 8 8C.

HTP for Synthesis of PP5
After the HT generation of PP6 had been achieved, we decided to also attempt generating the more strained fivemembered amide linkages,t hat is,t he HTP of PP5 (Scheme 3). Therefore,w ef irst prepared as uitable MS (MS5;s ee the Supporting Information) of pyromellitic acid (PMA) and DAB. Thesuccessful formation of MS5 including 1:1s toichiometry of PMA:DAB was confirmed by ATR-FTIR and 1 H-NMR analysis (see the Supporting Information). Surprisingly,when MS5 was subjected to HT conditions using the conditions already applied for PP6 synthesis (MW, c = 0.01 mol L À1 , T R = 250 8 8C, t H = 10 min, t R = 15 min) the reaction yielded af ine,d ark orange powder topped by atranslucent liquid phase.This aspect lies in stark contrast to the deep black color reported for PP5. [25] Furthermore,t he corresponding ATR-FTIR spectrum ( Figure 4B;t op curve) lacked the intense C=Om ode at % 1755 cm À1 characteristic for PP5. Nevertheless,t he spectrum is still significantly different from that of MS5 (e.g.d isappearance of NH 2 and NH 3 + modes;s ee the Supporting Information). Yet, neither the well-defined ATR-FTIR spectrum, nor the products appearance indicate an uncontrolled decomposition. We considered it more likely that ad istinct chemical reaction had taken place that, however, did not yield PP5. In accordance with the literature on similar compounds, [57][58][59] three structural possibilities seemed probable ( Figure 4A): (i)Ani nitial condensation might generate ap oly(amide amino acid) (PAAA). In as econd condensation step two different species could form next:( ii)a poly(imide amine) (PI-NH 2 )or(iii)apoly(benzimidazole acid) (PBI-COOH). In theory,amixture of some or even all of these different species, as well as ac rosslinked network (necessarily implying amide linkages) is conceivable.Y et the productsA TR-FTIR spectrum shows neither amide nor imide modes.H ence,t he formation of PA AA, PI-NH 2 ,o ra na mide-crosslinked network can be excluded. Moreover,t he combined C = C/C = N ring vibration mode at % 1620 cm À1 ,t he benzimidazole inplane vibration mode at % 1445 cm À1 ,and the C-N stretching mode at % 1315 cm À1 are found. These modes are strongly indicative of polybenzimidazoles (PBIs), [30] and also in accordance with al ow-molecular-weight model compound synthesized from PMDAa nd o-PDA. [58] Therefore,w e expected the material obtained via HT treatment of MS5 to be PBI-COOH. Interestingly,t he ATR-FTIR spectrum does not feature aC = Om ode of -COOH (characteristic region indicated by orange box in Figure 4B), which we attribute to an intramolecular proton transfer generating H-bonded imidazolium and carboxylate groups (cf.s tructure in Figure 4B). [58] However,f or simplicity we will here refer to the obtained material as PBI-COOH.
To transform PBI-COOH to PP5, just one additional condensation-cyclization step is required. However,a ll our attempts to hydrothermally achieve this final ring closure to as ignificant extent were unsuccessful (see the Supporting Information). Interestingly,u pon elevating T R for HT PP5 synthesis,wefound that T R s ! 300 8 8C(experiment performed in HPA) lead to decomposition. This lies in stark contrast to PP6, where only HTPs at such T R sallow for synthesizing highmolecular weight products.I nf act, we speculate that the intramolecular H-bonds between imidazolium and -COO À (which we expect to be seven-membered cyclic motifs) energetically stabilize PBI-COOH (especially in ap rotic environment such as HTW) and hence make the final ring closure more difficult, if not hydrothermally impossible. Having demonstrated for the case of MS6 that different reactivities during HTP and SSP occur,w ed ecided to investigate post-polymerization solid-state heat treatment of PBI-COOH for triggering its transformation towards PP5. Luckily,t his solvent-free heat treatment indeed enabled the conversion to PP5:W hen PBI-COOH was heated to 400 8 8C under N 2 atmosphere,t he desired second ring closure could be achieved. ATR-FTIR analysis ( Figure 4B;b ottom curve) is nicely in accordance with the literature and confirms PP5 formation. [14,60] All characteristic modes are present: n C=O (PP5) % 1755 cm À1 ; n C=C/C=N (PP5) % 1620 cm À1 (combined mode); n(benzimidazole) % 1440 cm À1 (in-plane vibration); n C-N (PP5) % 1310 cm À1 ). Modes indicating incomplete ring closure (amide,i mide,a mino,c arbonyl/carboxyl) are absent.
Subsequently,aTGA experiment was performed to monitor the reaction from PBI-COOH to PP5 ( Figure 4C). An initial mass loss of % 7% is attributed to physisorbed H 2 O-a well-known phenomenon for various PBIs and PPs. [39,61] Thes econd mass loss step of % 8.9 %p erfectly matches the calculated m(H 2 O) for full condensation through the desired ring-closure (see the Supporting Information).
PXRD measurements (see the Supporting Information) evinced that PP5 as well as PBI-COOH are amorphous, featuring only one broad halo centered around % 268 8 (2q,Cu-K a )i ndicating low order yet intermolecular p-stacking interactions between the polymer chains.SEM measurements of MS5, PBI-COOH and PP5 ( Figure 5) revealed strong morphological similarities between PBI-COOH and PP5, indicating the absence of any softening phenomena during the corresponding transformation. Both contain mainly mats of fibroids and af ew spherical particles ( % 0.5-1 mmi nd iameter), while MS5 is exclusively composed of angular particles of % 5-10 mm. These morphologies imply that the HTP towards PBI-COOH must occur via ad issolution-polymerizationprecipitation mechanism, whereas the second solvent-free reaction step from PBI-COOH to PP5 proceeds in the solid state.  Overall, PP5 and PP6 differ significantly in their synthesis. Thef act that isolable PBI-COOH is obtained when MS5 is subjected to HT conditions suggested that also pure,n onfunctionalized PBIs might be available by HTP.

HTP for Synthesis of PBI
In analogy to the HTP of PIs [4][5][6]44] and PPs (this work), we first attempted to prepare aM Sf rom terephthalic acid (TA) and DAB. However,the preparation of the corresponding MS could not be achieved, which we explain by the fact that the pK a difference between TA (pK a (TA) = 3.49) and DAB (pK a (DAB) = 4.39) is too small for salt formation. [62] Therefore,w ea imed at preparing PBI via directly subjecting the neat comonomers TA and DABs uspended in H 2 Ot oH T conditions (MW, c = 0.01 mol L À1 , T R = 250 8 8C, t H = 10 min, t R = 15 min). Ab rown powder at the bottom of the liner was obtained. As ATR-FTIR measurements did not confirm PBI formation, we assumed that TA lacks sufficient reactivity. Hence,wedecided to replace TA by amore reactive carbonyl compound:i nspired by Neuse et al. we chose to use terephthalaldehyde (TDA). [63] They reported the preparation of an imine prepolymer generated under anaerobic conditions at T R 25 8 8Ci na na protic,p olar solvent, and its subsequent conversion to PBI upon heating to 60 8 8Ci nt he presence of at ransition metal catalyst under aerobic conditions.I nstead, our chosen synthetic strategy (Scheme 4) requires neither oxygen exclusion nor transition metal catalysts.T he only solvent and catalyst employed is H 2 O. When stirring an aqueous dispersion of TDAa nd DABa tr t, we observed arapid color change (from beige to red) of the solution as well as the formation of ar ed solid precipitate.A TR-FTIR spectroscopy ( Figure 6A)a llowed for identifying the precipitate as al ow-molecular-weight oligoimine intermediate.I n addition to the characteristic imine modes (arrows in Figure 6A; n C-H (imine) % 2860 cm À1 ; n C=N (imine) % 1615 cm À1 ; n C-N (imine) % 1295 cm À1 ), [63,64] also intense aldehyde (n C=O (aldehyde) % 1695 cm À1 ;b rown box) and amino (n N-H (NH 2 ) % 3450 cm À1 , % 3355 cm À1 ;brown box) modes are found. Such imine intermediates were already observed by Neuse et al. [63] Subsequent HT treatment of the aqueous red dispersion (MW, c = 0.01 mol L À1 T R = 250 8 8C, t H = 10 min, t R = 15 min) yielded an orange solid topped by at ranslucent, clear supernatant and ATR-FTIR measurements ( Figure 6A) evinced the transformation of the intermediate to PBI. Clearly,a ll modes indicating the presence of free NH 2 or aldehyde moieties (brown boxes) as well as imine linkages have completely vanished. Furthermore,s everal characteristic PBI modes ( Figure 6A,a rrows) are present. Finally, TGA measurements (see the Supporting Information) revealed the presence of physisorbed H 2 Oa nd high thermal stability (T 95 % = 584 8 8C, T 90 % = 659 8 8C). PXRD (see the Supporting Information) shows that PBI is completely amorphous only showing ab road and weakly pronounced halo at % 15-358 8 (2q,Cu-K a ). In terms of morphology,PBI from HTP contains some angular and sheet-like particles of broad size distribution ( % 20-300 mm, Figure 6B)a lways covered by significantly smaller spherical particles of quite narrow size distribution ( % 0.5-2.0 mm, Figure 6B,C).
In summary,w eh ave demonstrated here that in addition to the previously reported PIs and PA networks,H TP as an environmentally benign and experimentally simple strategy can also be used to obtain PPs and PBIs.W hile the major applications of PBIs lie in fireproof clothing, bearings,a nd fuel cell membranes,P Ps are interesting for optical and electronic applications owing to their conjugated nature. Especially since HTP does not require any catalysts,t he obtained products are very pure.T herefore,w ed ecided to investigate the electrochemical properties of PP6, which was considered most promising because of the naphthalene moiety in its backbone.

Electrochemical Characterization
PIs and quinone-decorated polymers have recently been studied as aqueous battery electrodes in neutral and acidic electrolytes. [20][21][22] While fabricating such abattery using PPs is way beyond the scope of this work, we decided to investigate PP6 from HTP by cyclic voltammetry (CV) in aqueous electrolytes as af irst step.F igure 7Ashows the CV curves recorded with PP6-carbon black (CB) electrodes at different pH values.In0.1m HClO 4 (pH 1), the redox feature centered at % 0.1 Vv s. the standard hydrogen electrode (SHE) is reversible,asconfirmed tby he symmetry of the cathodic and anodic waves and by the linear dependence of peak currents on the square root of the scan rate (inset of Figure 7B). The peak separation (DE p )i s2 5mV( see the Supporting Information for details). Themost cathodic part of the CV is partly tilted because of the presence of the hydrogen evolution reaction, which is readily catalyzed both by CB and by PP6, especially under acidic conditions (E 2H þ =H 2 = À0.059 Vv s. SHE at pH 1). We ascribe the redox feature to the reduction (and protonation)/ reoxidation (and deprotonation) of the PP6sC = Om oieties.S imilar behavior was indeed observed for molecular quinones and quinone-bearing polymers. [66] At as can rate of 10 mV s À1 ,t he specific concentration of redox centers (as extrapolated from the charge transferred in the CV) is 3.2 mmol g À1 .B ased on the comparison with the nominal concentration of C = Ogroups in PP6 (4.9 mmol g À1 ), we infer that the reduction and protonation of PP6 (i.e., H + intercalation) is not limited to the surface,but extends to the bulk, with ap rotonation efficiencyo f% 70 %( see the Supporting Information). Thep osition of the redox peak shifts using mild acidic and neutral electrolytes and the reversibility is partly lost (increase in DE p ). Thea symmetry and the overall changes in the electrochemical behavior may be caused by changes in the kinetics of the redox reaction at the C = Om oiety,w hich have been demonstrated to vary depending on the nature of the reaction media in buffered and unbuffered electrolytes. [67,68] Moreover,i ti sp lausible to expect that the redox behavior is given by the coexistence of the ion intercalation (K + from buffer) and the protonation reaction. A5 2mV/pH unit slope (inset Figure 7A)i s observed when the shift of the redox potential is monitored (E 1/2 ,calculated by averaging the potentials of the anodic and cathodic peaks,T able S1) as function of the pH of the electrolyte.T his agrees with ap redominant proton-coupled redox reaction. While several redox polymers have been characterized in organic electrolytes, [20,21,69,70] to the best of our knowledge,t here have been no studies on similar PPs in aqueous electrolytes.F or PIs and quinone-decorated polymers studied as aqueous battery electrodes,itwas shown that they can be used for the intercalation of ions (Mg + ,Na + ). PP6 seems to possess similar properties,which we plan to explore in more detail in future studies.

Conclusion
Herein, we demonstrate that in accordance with lowmolecular-weight perinone it is possible to generate the pyrrone polymer PP6 (in the literature commonly known as BBB) hydrothermally starting from as uitable monomer salt precursor MS6 (synthesized from naphthalene bisanhydride and 3,3'-diaminobenzidine). Upon elevating the reaction temperature,t he degree of polymerization increases.M S6 and PP6 are morphologically virtually identical. Thus,asolidstate transformation under shape retention occurring in dispersion is assumed. Yet, high-temperature water must play acrucial role for facilitating this transformation, but its exact role still remains unclear.F urthermore,P P6 is electrochemically active in aqueous electrolytes.U nder acidic conditions we ascribe the redox behavior, which was found to extend also to the bulk of PP6, to the reduction and protonation/ reoxidation and deprotonation of the C = Om oieties.T his characteristic could be exploited, for example,i na queous battery applications.
Interestingly,i nt he case of MS5 (a monomer salt synthesized from pyromellitic dianhydride and 3,3'-diaminobenzidine), hydrothermal treatment does not directly yield the corresponding pyrrone polymer PP5. Instead, apolybenzimidazole intermediate with -COOH side groups (PBI-COOH) is generated. Hence,the mechanistic pathway clearly differs from that of the hydrothermal polymerization towards PP6. Themost efficient and effective way to entirely convert PBI-COOH to fully condensed PP5 is ah eat treatment at 400 8 8C. Regarding the morphology,n os ignificant differences can be observed between PBI-COOH and PP5, which is in agreement with as olid-state transformation. However,t hey differ significantly from MS5. Therefrom, we propose ad issolution-polymerization-precipitation mechanism for the hydrothermal generation of PBI-COOH. This lies in stark contrast to the hydrothermal polymerization towards PP6 where no intermediate dissolution phenomena are observed.
In contrast to PP6 and PP5, for pure,n onfunctionalized polybenzimidazole (PBI) as uitable monomer salt precursor could not be prepared. Nevertheless,the desired product can be generated hydrothermally by reacting terephthalaldehyde with 3,3'-diaminobenzidine.Initially,animine intermediate is formed under ambient conditions which can subsequently be converted to PBI hydrothermally.
We consider the successful preparation of these two classes of high-performance polymers as as ignificant broadening of the scope of hydrothermal polymerization. These results are apromising starting point for future developments in the field. They open the door for avariety of new materials to be generated in af acile,e xperimentally simple,a nd environmentally benign way.