Introduction

Porous organic polymers have great potential for various applications, such as adsorbents [1, 2], gas storage and separation [3], sensors [4], heterogeneous catalysts [5] and their supports [6], encapsulation for the controlled release of drugs [7], and electrodes for energy storage [8]. The synthesis of highly porous polymers has been developed using two dominant strategies: templating by porogens and molecular design for structural pore formation. Soft and hard templates are utilized in polymerization processes to fabricate porous polymers [2, 9,10,11]. For example, surfactant micelles and liquids that are miscible with monomers but insoluble with their polymers perform as soft templates. Commercial polystyrene adsorbents having a high specific surface area, such as Amberlite resins, are directly synthesized through the aqueous suspension polymerization of styrene as a monomer and divinylbenzene as a crosslinker. Mesopores of ~10 nm in diameter are formed in polystyrene after the removal of toluene or xylene as porogens [1, 2]. Pore size is controlled by a soft template of surfactant micelles in a range smaller than 50 nm [12, 13]. Porous phenolic resins with high specific surface areas of ~500 m2 g−1 were prepared using a triblock copolymer as a soft template [14]. However, suitable materials are restricted for soft templates as porogens [1]. Mesoporous phenol-formaldehyde resins with high specific surface areas of 60–1600 m2 g−1 were produced using mesoporous silica as a hard template [15]. This method has advantages in controlling morphologies by the selection of templates. Our research group reported hierarchical porous organic polymers using nanoscale crystal grains of biominerals as a template [16, 17]. Porous polypyrrole (PPy) with a specific surface area of ~500 m2 g−1 was also produced using synthetic nanocrystals, such as V2O5·nH2O fibrils [18]. In these cases, hard templates require complicated preparation processes and a method of removing the templates from the porous products [19].

The other strategy for synthesizing porous polymers is designing monomer molecules. Molecular-sized micropores that provide extremely high specific surface areas can be fabricated in polymer chains [20,21,22]. A porous polymer with an extremely high specific surface area of ~6000 m2 g−1 was reported with the polymerization of a designed monomer, tetrakis(4-bromophenyl)silane [23]. However, designed monomers are very expensive, and they are not offered commercially. Moreover, the application of molecularly designed porous polymers is limited with respect to molecular adsorption because of their exceedingly small pore sizes. As a consequence, it is still challenging to develop novel methods for simply synthesizing highly porous organic polymers.

In the present article, we report a novel strategy for fabricating highly porous organic polymers using nanometer-scale dendritic growth with oxidative polymerization in solution systems. Nanoscale dense branching morphologies of polymers are formed under a highly oversaturated condition for the polymer through a rapid oxidative polymerization reaction in a cosolvent containing high concentrations of monomers and oxidants. Since the polymer formation is regarded as the phase separation from a solution system, this behavior is similar to dendritic crystal growth under a diffusion-limited condition at a high degree of supersaturation or supercooling [24]. Here, the diffusion of the monomer and the oxidant limits the growth of polymer dendrites. We used pyrrole (Py) and its derivatives as monomers and FeCl3 as an oxidant for rapid oxidative polymerization. Py and its derivatives were selected as typical polymers obtained by oxidative polymerization. Their polymers are utilized for various applications, such as sensors [25], capacitors [26], and adsorbents of organic molecules [27]. Here, we obtained dense branching morphologies of PPy and its derivatives that have mesopores with diameters of 10–50 nm. Moreover, intragranular micropores derived from pendant groups were found in nanoscale grains of PPy derivatives. Finally, highly porous polymers with bimodal pore-size distribution and specific surface areas as high as ~900 m2 g−1 were successfully synthesized by a novel method using oxidative polymerization in concentrated solution systems.

Experimental procedure

Synthesis of polymers

All solvents, monomers, and oxidants were used as purchased without any further purification. A certain amount of FeCl3 (Junsei, Tokyo, Japan) as an oxidant was dissolved in 40 cm3 of chloroform (CHCl3, Kanto Chemical). We added Py (Tokyo Chemical Industry, Tokyo, Japan), 1-methylpyrrole (1MPy, Tokyo Chemical Industry), and 1-ethylpyrrole (1EPy, Wako Pure Chemical, Osaka, Japan) (Fig. 1) as monomers to the oxidant solution with vigorous stirring at a fixed molar ratio (monomer/oxidant = 1/4). CHCl3 was used as a cosolvent for the monomers and oxidants. The amount of monomers varied from 0.28 to 2.24 mmol (7–56 mmol dm−3). We confirmed that ~110 mmol cm−3 of FeCl3 was dissolved in CHCl3. However, an excess amount of the oxidant was dispersed in the liquid phase. Py and its derivatives were immediately polymerized by oxidation with FeCl3 in CHCl3. Resultant dispersions were filtrated, and products were washed with pure water, 1 mol dm−3 of HClaq, and ethanol. We used HClaq to remove iron ions in the nanometer-scale pores of resultant polymers. Washed samples were then dried at 60 °C for 12 h. For comparative experiments, we used pure water as a liquid medium for polymerization. The oxidant is miscible, and the monomers are immiscible in water.

Fig. 1
figure 1

Structural formula of monomers

Full size image

Characterization

Fourier transform infrared (FT-IR) absorption spectroscopy (Jasco FT/IR-4200) was used to characterize the molecular structure of the products. The morphologies of the products were observed with scanning electron microscopy (SEM, Hitachi S-4700 and JEOL JSM-7600F, operated at 5 kV). The polymer dispersion of ethanol was dropped on a copper grid covered with a collodion film for transmission electron microscopy (TEM) observation. Specific surface areas were evaluated using the Brunauer–Emmett–Teller (BET) method from nitrogen isotherms obtained by a Micromeritics 3Flex-3MP at 77 K. The pore-size distribution was calculated using non-local density functional theory (NLDFT) to estimate both micropores and mesopores in the polymers. BEL Master software was used for NLDFT calculations using a pore model of an oxygen-exposed surface with a slit-shaped pore structure and a log-normal model with double or triple peaks as the peak assumption. We also evaluated the size distribution of mesopores using the Barrett−Joyner−Halenda (BJH) method. The size distribution in the mesopore region obtained by NLDFT almost agrees with that obtained with BJH, whereas the mode values are slightly shifted (Figure S1).

Adsorption properties were studied by using aqueous solutions of p-cresol (4-methyl phenol, Kanto, 99%). Synthesized P1MPy and Amberlite XAD 2000 (SBET = 558 m2 g−1) were used as adsorbents. The adsorption experiments were performed in 1.5 cm3 of p-cresol aqueous solutions containing 10 mg of the adsorbents. The concentrations of p-cresol were measured by ultraviolet-visible (UV-vis) absorption spectroscopy using a Jasco V-670. The concentrations after 14 days of stirring were adopted as the equilibrium concentrations (Ce).

Results and discussion

Since CHCl3 is a cosolvent for the monomers and oxidants, Py, 1MPy, and 1EPy were immediately polymerized by oxidation with FeCl3. The yields, which were calculated by comparing the mass of dried products with the theoretical value, were >60% in a minute and > 90% in 24 h. Figure 2 shows typical FT-IR absorption spectra of the products. The presence of specific absorption bands indicates the formation of PPy (D-H), poly(1-methylpyrrole) (P1MPy), and poly(1-ethylpyrrole) (P1EPy) (A-H) from their monomers under all conditions in the present study [28,29,30,31]. Intense signals of C-H stretching vibrations at 2850–3000 cm−1 (B) originate from the methyl and ethyl groups of P1MPy and P1EPy. Large absorption bands assigned to O–H stretching vibrations (A) and C=O stretching vibrations (C) suggest the overoxidation of P1MPy and P1EPy [32, 33]. The variation of the conjugate structures between PPy and its derivatives caused intensity differences in C–H out-of-plane bending signals (G).

Fig. 2
figure 2

FT-IR absorption spectra of products synthesized from Py (i), 1MPy (ii), and 1EPy (iii) at [monomer] = 28 mmol dm−3. The absorption bands of polymers were assigned to the following vibrations: O–H stretching (a), C–H stretching of alkyl groups (b), C=O stretching (c), C=C stretching (d), C–N stretching (e), C–H in-plane bending (f), C–H out-of-plane bending (g), and C–C in-plane bending (h). The assignments of the absorption bands were referred to in previous reports [28,29,30,31]

Full size image

Figure 3 shows the morphologies of PPy, P1MPy, and P1EPy produced by oxidative polymerization in the solution system. Radially grown nanometer-scale ribbons (20–50 nm wide) were formed with submicron particles at [monomer] = 14 mmol dm−3 (Fig. 3a–d). The ribbons were composed of nanoscale grains. Micrometer-scale spherical particles consisting of nanoscale grains (10–50 nm in diameter) were obtained at [monomer] = 28 mmol dm−3 (Fig. 3e–g). SEM images of a fractured particle indicate that a dendrite having a dense branching morphology was formed in the spheres (Fig. 3h, i). As a consequence, the rapid polymerization of the concentrated solution system leads to a specific morphology that is not influenced by the sort of the monomers. An increase in the concentrations of the monomer and the oxidant induced nanoscale dendrites with dense branching.

Fig. 3
figure 3

SEM images of Py (a, e), P1MPy (b, f, h), and P1EPy (c, g) synthesized at [monomer] = 14 mmol dm−3 (a–c) and 28 mmol dm−3 (e–h). Schematic illustrations of radial ribbons (d) and a dendrite having dense branching morphology (i)

Full size image

Figure 4 shows the specific surface areas and pore-size distribution of three types of polymers prepared under various conditions. The specific surface areas of the polymers depend on their component and monomer concentrations. Here, we found P1MPy and P1EPy to have extremely high specific surface areas of ~800–900 m2 g−1. Assuming that dendrites are composed of spherical polymer grains (density: 1.5 g cm−3 [34], diameter: ~20 nm), their surface areas are estimated from their morphology to be ~200 m2 g−1. In fact, we observed the monomodal distribution of mesopores in a range of 5–50 nm, which was derived from the intergranular space for PPy dendrites (Fig. 4b). Thus, the variation of the specific surface area of PPy is ascribed to the morphological change from radial ribbons into nanoscale dendrites. Bimodal distributions with mesopores and micropores ~1 nm in diameter were found for P1MPy and P1EPy. Therefore, the extremely high specific surface area is ascribed to the presence of micropores that were not observed in TEM images (Figure S2). It is deduced that the micropores are attributed to the intragranular spaces originating from the pendant groups of P1MPy and P1EPy. On the other hand, an increase in the specific surface area of P1MPy with increasing monomer concentration is mainly ascribed to an increment of the mesopore volume (Fig. 4c). Thus, the bimodal pore structures derived from the nanoscale dendritic morphology and the intragranular spaces are essential for the extremely high specific surface area of ~900 m2 g−1. As shown in Fig. 2, the overoxidation reaction was deemed to occur during the polymerization of P1MPy and P1EPy from intense absorption signals assigned to O–H and C=O bonds. The high specific surface area originating from the micropores would enhance the oxidation of the polymer chains of PPy derivatives. (The high specific surface area would be derived from the micropores that are induced by the overoxidation of PPy derivatives.) The redox potential of 1MPy was reported to be lower than that of Py [35]. The overoxidation is also ascribed to the lower redox potentials of Py derivatives due to the electron donation effect of alkyl groups.

Fig. 4
figure 4

The BET specific surface areas SBET a and pore-size distribution b, c of PPy, P1MPy, and P1EPy as a function of [monomer]. The pore-size distribution of PPy and its derivatives at [monomer] = 28 mmol dm−3 (b). The pore-size distribution of P1MPy at [monomer] = 7, 14, and 28 mmol dm−3 (c)

Full size image

As shown in Fig. 4a, the specific surface area of the polymers did not increase when the monomer concentration increased above 28 mmol dm−3. We added quadruple the amount of FeCl3 to promote rapid polymerization in the solution system. However, the reaction rate was not increased in the highly concentrated solutions because the oxidant concentration was limited below its solubility (~110 mmol dm−3).

We changed the liquid medium to water to investigate the cosolvent’s effect on the morphology and specific surface area of P1MPy polymerized by FeCl3. Since the monomer is immiscible in water, the monomer drops were dispersed as an emulsion. Submicrometer-scale particles with rough surfaces aggregated in water (Fig. 5a). The specific surface area was estimated to be ca. 30 m2 g−1; however, the pore-size distribution varied only in micropores (Fig. 5b). The dendritic morphologies of the polymer were not produced in the aqueous system because polymerization proceeds gradually at the interface of the droplets and water containing the oxidant. The specific surface area of the polymer is mainly ascribed to the intragranular space.

Fig. 5
figure 5

SEM images (a) and pore-size distribution (b) of P1MPy prepared with FeCl3 in water at [monomer] = 28 mmol dm−3

Full size image

In the present article, we have reported a novel strategy for creating highly porous organic polymers using nanometer-scale dendritic growth with oxidative polymerization in concentrated solution systems. Nanoscale dense branching morphologies are formed during the rapid reaction with the diffusion of the monomer and the oxidant. Figure 6 shows schematic illustrations of the formation process of nanoscale dendritic polymers in a cosolvent of monomers and oxidants. Rapid polymerization is induced by the oxidation of a large number of monomers with abundant oxidants in the liquid medium. Nanometer-scale dendrites are formed by the rapid growth of polymers in the oversaturated solution. Since polymer formation is regarded as a phase separation from a solution system, this behavior is similar to dendritic crystal growth under a diffusion-limited condition at a high degree of supersaturation or supercooling [24]. Here, the diffusion of the monomers and oxidants limits the growth of polymer dendrites. Radially grown ribbons from a nucleus are observed with less branching growth under a moderate condition (Fig. 6a). Densely branching morphologies are generated under a diffusion-limited condition in a highly oversaturated solution for the polymer through a rapid oxidative polymerization reaction when the cosolvent contains high concentrations of monomers and oxidants (Fig. 6b). When the monomer is immiscible in the liquid medium, dendritic morphologies are not produced in the emulsion system. Polymerization proceeds gradually at the interface of the monomer droplets and water containing the oxidant (Fig. 6c). In the aqueous system, the monomer at the interface is quickly polymerized at the initial stage. However, the reaction rate decreases at the progressive stage because the polymerization requires penetration of the oxidant into the polymer film formed at the interface.

Fig. 6
figure 6

Schematic illustration of the formation of nanoscale polymer dendrites through rapid oxidative polymerization in a good cosolvent for monomers and oxidants at a moderate concentration (a) and under a highly concentrated condition (b). A non-dendritic morphology is formed when the monomer is immiscible in the liquid medium (c)

Full size image

Highly porous polymer dendrites of P1MPy were applied for the adsorption of aromatic compounds in water. We used p-cresol, a typical organic pollutant, as an adsorbate, which was determined quantitatively from UV-vis absorption spectra (Figure S3). As shown in Fig. 7, the adsorption amounts were higher on P1MPy dendrites than on a commercial polystyrene adsorbent (Amberlite XAD2000, SBET = 558 m2 g−1). The high specific surface area originating from bimodal pore structures is expected to be suitable for the adsorption of aromatic compounds due to π–π interactions.

Fig. 7
figure 7

Adsorption isotherms of p-cresol on P1MPy dendrites and a commercial adsorbent

Full size image

Conclusion

Highly porous polymers with extremely high specific surface areas up to 900 m2 g−1 were obtained via the simple oxidative polymerization of pyrrole derivatives, such as poly(1-methylpyrrole) and poly(1-ethylpyrrole), in a solution system. The bimodal distribution of micropores and mesopores is essential for extremely high specific surface areas. Rapid polymerization in the concentrated solution system provides nanoscale branching morphologies that exhibit mesopores in the range of 5–50 nm. Pendant groups of the derivatives lead to the formation of micropores of ~1–2 nm. High specific surface areas that are suitable for the adsorption of aromatic compounds originate from the bimodal pore-size distribution in the micropore and mesopore regions. Rapid polymerization in the solution system would be applicable to the fabrication of highly porous polymers composed of other components.