Polymer blend directed anisotropic self-assembly toward mesoporous inorganic bowls and nanosheets

See allHide authors and affiliations

Science Advances  12 Aug 2020:
Vol. 6, no. 33, eabb3814
DOI: 10.1126/sciadv.abb3814


Anisotropic mesoporous inorganic materials have attracted great interest due to their unique and intriguing properties, yet their controllable synthesis still remains a great challenge. Here, we develop a simple synthesis approach toward mesoporous inorganic bowls and two-dimensional (2D) nanosheets by combining block copolymer (BCP)–directed self-assembly with asymmetric phase migration in ternary-phase blends. The homogeneous blend solution spontaneously self-assembles to anisotropically stacked hybrids as the solvent evaporates. Two minor phases—BCP/inorganic precursor and homopolystyrene (hPS)—form closely stacked, Janus domains that are dispersed/confined in the major homopoly(methyl methacrylate) (hPMMA) matrix. hPS phases are partially covered by BCP-rich phases, where ordered mesostructures develop. With increasing the relative amount of hPS, the anisotropic shape evolves from bowls to 2D nanosheets. Benefiting from the unique bowl-like morphology, the resulting transition metal oxides show promise as high-performance anodes in potassium-ion batteries.


Anisotropic mesoporous inorganic materials are an emerging class of porous materials, which have attracted notable attention in recent years because the shape anisotropy and high complexity break the symmetry of particles, thereby creating remarkable physical properties that cannot be attained by conventional isotropic materials (1, 2). Of particular importance are (i) bowl-like morphologies with fascinating new functions, e.g., superhydrophobicity of array film (3, 4), special optical properties (5), and stacking phase behavior (6, 7); and (ii) two-dimensional (2D) nanosheets with unique optical, chemical, and electronic properties and a high packing density and an improved mass/ionic transport capability (812). These intriguing features of anisotropic morphologies could add a new dimension to the range of potential applications for mesoporous inorganic materials.

Block copolymer (BCP) self-assembly allows the fabrication of mesoporous inorganic materials with tunable pore structure and size (1315). Conventional BCP-directed approaches for preparation of mesoporous inorganic particles mainly rely on the multistep procedures that use sacrificial silica templates (e.g., inverse opal) (16) or the solution-based systems such as emulsion and direct precipitation methods, which favorably generate isotropic spheres (or ill-defined shapes) because of a natural tendency to minimize total interfacial energy (1720). Thus, anisotropic morphologies can only be realized by breaking such intrinsic centrosymmetric arrangements; this breaking usually requires additional macroscale templates (for bowls, collapsed hollow polymers and concave substrates; and for 2D nanosheets, graphene and flat substrates/interfaces) (2125). The kinetically controlled anisotropic growth on such templates has been also proposed (26, 27). Specific apparatus or techniques, e.g., lithography, mechanical cleavage, and (electro)chemical exfoliation, are often further combined for 2D nanosheet syntheses (28). However, these processes are either tedious or difficult to control the interaction with BCPs, so attempts to integrate the macro-templates with BCPs often yield poorly defined mesostructures that are composed of irregularly attached nanoparticles or nanoflakes (21, 22). Furthermore, solution-phase growth methods are usually materials-specific to polymer, carbon, and silica because the used aqueous condition is unsuitable for most inorganic precursors such as transition metal alkoxides, which are highly reactive toward hydrolysis and condensation (26, 29). These methods lack the ability to control pore sizes and structures because of limited applicability of high–molecular weight BCPs in aqueous solution. To date, synthesis approach that enables simultaneous control over various anisotropic shape and pore sizes/structures has never been realized. Therefore, it still remains a crucial challenge to develop a facile and versatile strategy to synthesize anisotropic inorganic ordered mesoporous materials with tunable large pores >15 nm and controllable particle dimension and sizes.

Multiphase polymer blends generally show complex yet unique phase behaviors that may provide an opportunity for a new approach to the synthesis of anisotropically shaped materials. In particular, the ternary-phase polymer blends—e.g., two minor phases dispersed in one major phase—can be useful components in design of various intriguing polymer morphologies, such as core-shell and Janus particles (30, 31). The phase migration and blends morphology are determined by the formation of polymer-polymer interfaces, which are greatly affected by interfacial tension between the components. Although the underlying structure evolution mechanism of multiphase polymer blends is relatively well understood, the exploitation of such phase behavior for synthesis of functional nanomaterials (e.g., inorganic porous materials) has been rarely explored because of the absence of effective system that can integrate complex polymer physics and inorganic materials chemistry in a controlled manner. Here, we bridge the gap between physics of multiphase polymer blends and BCP-directed inorganic materials self-assembly to provide a new synthetic approach for anisotropic mesoporous inorganic bowls and 2D nanosheets.

In the multiphase polymer blends, the blend morphologies can be predicted using spreading coefficients λijk, which are calculated asλijk=γij(γik+γjk)(1)where γ represents the interfacial tension for each component pair (32, 34). Each spreading coefficient shows the tendency of one phase located at another phase. In ternary blends in which two minor components (phases 1 and 3) are dispersed in a continuous matrix (phase 2), three types of phase morphologies can be identified: nonwetting (dispersed particles), partial wetting (stack formations), and complete wetting (core-shell) formations (fig. S1). When all three spreading coefficients are negative, each phase partially spreads at the interfaces of the other two, so the two minor phases can self-organize anisotropically into Janus droplets in the matrix (fig. S1B). The partially encapsulated domains of Janus droplets correspond to concavities. Here, we combine macroscale polymer phase behavior with BCP-directed microphase separation to provide simple access to anisotropically self-assembled particles (ASAP) with tailored macro- and mesostructures (Fig. 1). The balanced interfacial tension between immiscible phases leads to phase migration into anisotropic macroscopic structures (bowls and 2D nanosheets), while BCP self-assembly controls mesostructures (pore sizes and structures). The sizes of bowls’ concavities and anisotropic shapes from bowls to 2D nanosheets are tuned by changing the relative amount of minor phases. This strategy is applicable to various materials including aluminosilicate (AS), carbon, and transition metal oxides (TiO2 and Nb2O5/C).

Fig. 1 Schematic representation of ASAP approach to prepare bowl-like mesoporous inorganic particles and 2D mesoporous nanosheets.


Materials synthesis and characterization

We designed multiphase blend system consisting of homopolymer(s)/BCP/inorganic precursors for ASAP approach (Fig. 1 and fig. S2). To fabricate bowl-like mesoporous AS (bm-AS) particles with large mesopores (>15 nm), we used poly(ethylene oxide)-b-poly(styrene) (PEO-b-PS) as a structure-directing agent, excess homopoly(methyl methacrylate) (hPMMA) (350 kg mol−1) as an organic matrix, homopoly(styrene) (hPS) (350 kg mol−1) as one of the minor components, and prehydrolyzed AS sol (35) dissolved homogeneously in tetrahydrofuran (THF). The solvent evaporation at 50°C initiates macrophase separation via spinodal decomposition (SD) among hPS, hPMMA, and BCP/AS phases, which forms independent and dispersed hPS and BCP/AS macrodomains in hPMMA matrix (stage 1). During the solvent evaporation, the transparent solution becomes opaque, indicating macrophase separation–induced scattering of visible light (fig. S3). PEO-b-PS is immiscible with both hPS and hPMMA, as a result of their high molar mass (350 kg mol−1) and unfavorable thermodynamic changes upon mixing (fig. S4) (36). hPS and hPMMA are highly immiscible with each other as well because of their unfavorable enthalpic interaction and low entropy of mixing (37). The mutual immiscibility of all three phases can lead to SD by solvent removal from homogeneous solution. Two minor phases (BCP/AS phase and hPS phase) migrate to reduce interfacial tension, resulting in closely stacked domains that are confined in the major hPMMA matrix (stage 2). Because PS blocks of BCP phase are partially compatible with hPS domains, the formation of BCP/hPS interfaces is preferred over that of BCP/hPMMA and hPS/hPMMA interfaces. As a result, BCP phase partially covers the hPS domains to reduce enthalpically less favorable contact area between hPS and hPMMA (31). This phase behavior induces the formation of anisotropically self-assembled macrodomains from bowls to 2D nanosheets that are dispersed in the hPMMA matrix, depending on the relative mass ratio of the minor phases (Fig. 1). During the further solvent evaporation, microphase separation of BCP/AS phase drives the formation of cylindrical micelles (stage 3) (38). Because AS sols have a number of hydroxyl groups, they can selectively incorporate into PEO blocks by hydrogen bonding. The increase in the micelle concentration leads to self-assembly of BCP/AS micelles, thereby forming ordered mesostructures within the particles. Subsequent annealing at 100°C freezes the AS frameworks by inorganic condensation (stage 4). After pyrolysis to remove organic species, exposed PS block microdomains are converted into open accessible mesopores, and discrete bm-AS particles are obtained (stage 5).

This unique phase behavior can be further understood by introducing the spreading coefficients (λijk) in ternary immiscible blends (fig. S1B). To generate anisotropically stacked phases, there should be no specific high interfacial tension that overwhelms the sum of the other two (λijk< 0 in Eq. 1). The interfacial tensions among the immiscible polymer phases were obtained through measuring contact angle of two liquids (water and diiodomethane) on the solid polymer surfaces (fig. S5 and Supplementary Text) (39). The spreading coefficients for the ternary system of hPMMA/hPS/BCP were calculated using Eq. 1 (table S1). The interfacial tension between high molar mass pair of hPS/hPMMA (γPS/PMMA = 1.63 mN m−1) is higher than that of hPMMA/BCP (γPMMA/BCP = 1.60 mN m−1) (37). Because the PS blocks of BCP and hPS are chemically identical, hPMMA matrix has similar enthalpic interactions with both BCP and hPS phases. Therefore, γPMMA/PS could be only slightly higher than γPMMA/BCP, and the difference is not substantial (γPS/PMMA ≳ γPMMA/BCP). Because PS blocks of BCPs can partially lower the interfacial tension between hPS and BCP through favorable enthalpic interaction, γPS/BCP should be the smallest (γPS/BCP = 0.63 mN m−1). Therefore, the relation γPS/PMMA ≳ γPMMA/BCP > γPS/BCP can be derived, which leads to negative λijk in all cases according to the Harkin’s theory (λPMMA/PS/BCP = −0.66 mN m−1; λPMMA/BCP/PS = −0.60 mN m−1; λPS/PMMA/BCP = −2.60 mN m−1) (table S1). As a result, each phase shares its surface with both of the other phases and is adjacent to the other two. Because of a high value of λPMMA/BCP/PS, the tendency for BCP phase to spread over hPS was slightly more favored, leading to bowl-like mesoporous structures (fig. S2B).

The formation of bowl-like mesostructured BCP/AS hybrids is mainly based on anisotropic self-assembly of ternary polymer blends. To elucidate the morphology formation and phase behaviors of blends, as-made hybrids were microtomed at a thickness of 100 nm for transmission electron microscopy (TEM). TEM images of as-made bm-AS show the stacked morphology of hPS (dark area) and BCP/AS domain confined in the hPMMA matrix (bright area). The as-made samples cut horizontally exhibit that hPS domains are placed on BCP/AS domains (Fig. 2A). In an obliquely cut image, the BCP/AS domain partially covers the hPS domain (Fig. 2B). The BCP/AS domains form bowl-like particle structures with the internal mesostructures oriented to hPMMA matrix. Notably, hPMMA matrix can act as an enthalpically neutral surface for both AS/PEO blocks and PS blocks of BCP, i.e., the interfacial energy between each of the BCP microdomains and the PMMA matrix is well balanced and neutralized (40). Thus, both BCP microdomains tend to be exposed to hPMMA surfaces and oriented perpendicularly. The formation of such highly accessible pore orientation can be further confirmed by the control experiment in the absence of hPS phase (fig. S6) (17). Therefore, bm-AS particles with highly accessible open cylindrical pores were generated after calcination. Scanning electron microscopy (SEM) and TEM images reveal the uniform, anisotropic particle morphology of bm-AS (Fig. 2, C to E, and fig. S7). Close observation of bm-AS particles shows the bowl-like particle shapes and the mesopores, which are clearly exposed on the surfaces (Fig. 2D). Average particle size and concavity sizes are about 410 and 240 nm, respectively. Small-angle x-ray scattering (SAXS) pattern shows well-resolved diffraction peaks, indicating the 2D hexagonal mesostructures (p6mm) (Fig. 2F). Nitrogen physisorption isotherm exhibits typical type IV characteristics with a capillary condensation at 0.84 P/P0, indicating that large pores are dominant (Fig. 2G). bm-AS particles have a specific Brunauer-Emmett-Teller (BET) surface area of 300 m2 g−1 and a pore size of 15.5 nm calculated by the Barrett-Joyner-Halenda (BJH) method (Fig. 2H).

Fig. 2 Characterization of bm-AS.

(A) Top and (B) oblique views of particle cross-sectional TEM images of as-made bm-AS. (C) Low and (D) high magnification of SEM images, and (E) TEM image of bm-AS particles after calcination at 550°C. (F) Small-angle x-ray scattering (SAXS) pattern of calcined bm-AS consistent with a 2D hexagonal structure. (G) N2 physisorption isotherm and (H) the resulting pore-size distribution. a.u., arbitrary units; STP, standard temperature and pressure.

We further investigate the effect of hPS amount on bowl particle shape and concavity size by changing the relative mass ratio of hPS to BCP (rM = MhPS/MBCP; 0.2 ≤ rM ≤ 2). The sizes of the particle and concavities increase simultaneously as the rM increases. During anisotropic self-assembly, the hPS domain acts as an in situ generated template for the concavity. When a small amount of hPS was added (rM = 0.2), bm-AS-0.2 particles have an average particle size of 180 nm and a shallow concavity size of 70 nm (Fig. 3A and fig. S8). As the rM increased to 1 and 2, the concavity/particle size increased to 240/410 and 370/600 nm, respectively (Fig. 3, B and C; fig. S9A and S10A; table S2). Because BCP phase partially covers hPS phase to reduce thermodynamically more unfavorable contact area between hPS and hPMMA, an increase in size of hPS phase leads to an increase in size of BCP phase (i.e., bm-AS particles) accordingly. Regardless of the amount of hPS, bm-AS-rM particles had the same 2D hexagonal mesostructures (Fig. 3D) and the similar mesopore size (15.1 nm for bm-AS-0.2; 15.3 nm for bm-AS-2) (fig. S9, B and C; fig. S10, B and C). These results supported that hPS barely involves in mesostructured formation but selectively controls the macrostructures only (e.g., concavity and particle size).

Fig. 3 Tunable size of the concavities.

(A to C) SEM images and (D) SAXS patterns of bm-AS-rM particles with different concavity size, depending on relative mass ratio of hPS to BCP [rM = 0.2 (A), 1 (B), and 2 (C)].

The bowl-like mesoporous inorganic particles with uniform and tunable large pores above 15 nm, while highly desirable, still remain challenging to prepare. We successfully synthesized bm-AS particles with tunable pore size ranging from 15 to 39 nm by changing BCP molar mass (fig. S11 and table S2). Moreover, ASAP approach is highly versatile for the preparation of various compositions including carbon (fig. S12) and transition metal oxides that are difficult to prepare via solution-based methods, such as TiO2 (fig. S13) and Nb2O5/C (fig. S14).

Our ASAP approach can be extended to the preparation of the intriguing anisotropic 2D morphology such as mesoporous nanosheets. As the amount of hPS increases, the BCP-rich domains tend to be thinner and wider to cover more hPS-hPMMA interfaces. Therefore, at high rM value, BCP-rich domains become extremely thin with monolayered micelles, resulting in mesoporous inorganic nanosheets (Fig. 4A). We prepared 2D mesoporous AS nanosheets (mns-AS) at rM = 40. TEM images and electron energy-loss spectroscopy (EELS) elemental mapping of the as-made sample corroborate the location of each component in the blends after anisotropic self-assembly; the bright, gray and dark domains are assigned to the hPMMA, hPS, and BCP/AS domains, respectively (Fig. 4B and fig. S15). The oxygen distributes in the ester and ether groups of hPMMA matrix and PEO block, as well as siloxane and silanol groups of AS frameworks, but not in hPS domains. Thus, the dark, oxygen-poor area corresponds to hPS domains (fig. S15B). The silicon and aluminum maps clearly show that the hPS domains are partially covered by mns-BCP/AS domains whose thickness is much thinner than that of bm-AS. TEM and SEM images reveal that mns-AS clearly has well-developed large mesopores and ultrathin thickness (Fig. 4, C and D). mns-AS has a surface area of 268 m2 g−1 and a pore size of 16.2 nm (Fig. 4, E and F). The thickness of mns-AS was 8.5 nm, as determined by atomic force microscopy (AFM) (Fig. 4G).

Fig. 4 Characterization of mns-AS.

(A) Schematic representation of the preparation of 2D mesoporous AS nanosheets (mns-AS). (B) TEM images of as-made mns-AS. (C) TEM, (D) SEM images, (E) N2 physisorption isotherm, and (F) the corresponding pore-size distribution of mns-AS after calcination. (G) AFM image and the corresponding height profile of mns-AS.

BCPs are commonly used as compatibilizers to lower interfacial energy between incompatible components (41). Here, PEO-b-PS acts as the compatibilizer and the structure-directing agent. In a control experiment without PEO-b-PS, AS sol and hPS weakly interact as a result of their high enthalpy of mixing, leading to separated macrodomains without any interfacial contact area between AS and hPS (fig. S16A). The AS nanoparticles were completely separated from hPS (fig. S16, B and C), and nonporous particles were only obtained after calcination (fig. S16, D and E). In contrast, in the presence of PEO-b-PS, a morphology transition is observed from separately dispersed phases to partially encapsulated droplets. The AS phase migrates toward the hPS phase and is in contact with both the hPS phase and the hPMMA matrix. This phenomenon is attributed to the reduced interfacial tension by the compatibilizing effect of PS blocks of BCPs. PEO blocks of BCPs interact with AS sol by hydrogen bonding, which induces the ordered mesostructures. Mesostructured BCP/AS domains, therefore, are located in partial compatibility with hPS to decrease the interfacial energy.

The success of ASAP approach requires a number of design criteria. (i) Independent and discrete macrodomains should be generated for controllable multiphase assembly. To this end, homopolymers with relatively high molar mass compared to BCP (Mhomo/MBCP > 5) are required, because they can spontaneously macrophase-separate from BCP phase to minimize the conformational entropy loss during mixing and thus do not involve/interfere BCP-directed mesostructure formation. (ii) To fabricate ordered mesostructures, inorganic precursors should selectively mix with the hydrophilic blocks of BCP through favorable enthalpic interaction (e.g., hydrogen bonding), but not with the homopolymers. In this regard, hydrophobic homopolymers—which are immiscible with inorganic precursors—are highly desired. (iii) The interfacial tension between two minor phases against major matrix (γPS/PMMA and γPMMA/BCP) should not differ substantially, while that between two minor phases (γPS/BCP) should not be too large compared to the other two. Otherwise, morphology transition can occur into core-shell or totally separated particles.

Electrochemical performance of bm-Nb2O5/C as KIB anodes

Potassium-ion battery (KIB) has recently received considerable attention for next-generation energy storage system due to high abundance and low cost of potassium resource. In addition, the standard redox potential of K/K+ is similar to or even lower than that of Li/Li+ in nonaqueous electrolytes (42, 43). The poor rate capability and unstable cyclability of typical graphite anodes have stimulated intensive research for development of high-performance anodes in KIBs (44). Among metal oxides, orthorhombic Nb2O5 can accommodate K+ via the large (001) interplanar lattice spacing (0.39 nm) and exhibits pseudocapacitive reaction that achieves superior rate capability (45).

Here, we demonstrated the bowl-like mesoporous Nb2O5/carbon composite (bm-Nb2O5/C; Fig. 5) as a promising anode material in KIB. The crystalline bm-Nb2O5/C was obtained by pyrolysis of the as-made polymer-Nb2O5 hybrid (Fig. 5A and fig. S14A) at 700°C under an inert atmosphere. The part of PS blocks was converted to amorphous carbon on the mesostructured frameworks. After pyrolysis, the bm-Nb2O5/C particles maintained the well-developed bowl-like morphology (Fig. 5, B and C). N2 physisorption analysis and SAXS pattern show that bm-Nb2O5/C have a surface area of 50 m2 g−1 and the large mesopores of 21.5 nm with the regular pore structures (Fig. 5, D and E, and fig. S14B). X-ray diffraction (XRD) pattern of bm-Nb2O5/C corresponds orthorhombic phase Nb2O5, and the average crystallite size calculated by the Debye-Scherrer equation is 14.5 nm (Fig. 5F). The carbon content in the composite was determined to be 12 weight % (wt %) by using thermogravimetric analysis (fig. S14C).

Fig. 5 Structural characterization of bm-Nb2O5/C particles.

(A) Cross-sectional TEM image of as-made bm-Nb2O5/C. (B and C) SEM images, (D) N2 physisorption isotherm, (E) corresponding pore-size distribution, and (F) x-ray diffraction (XRD) pattern of bm-Nb2O5/C after pyrolysis. JPCDS, Joint Committee on Powder Diffraction Standards.

The half-cell was fabricated with bm-Nb2O5/C and K metal as a working and counter electrode, respectively. In the cyclic voltammetric curves at a scan rate of 0.1 mV s−1, the cathodic peak is located at 0.46 V and disappears in subsequent cycles, which is mainly attributed to the formation of solid-electrolyte interphase layer (Fig. 6A). Figure 6B shows the 1st, 2nd, 3rd, 5th, and 10th voltage profiles of bm-Nb2O5/C at a current density of 10 mA g−1 in the voltage range of 0.01 to 3 V versus K/K+. The initial discharge and charge capacities are 420 and 260 mA·hour g−1, respectively, with remarkable initial coulombic efficiency (CE) of 62%. In the subsequent cycles, the bm-Nb2O5/C anode delivers a highly reversible capacity. bm-Nb2O5/C also shows the outstanding rate performance in galvanostatic charge/discharge tests at various current densities (Fig. 6C). bm-Nb2O5/C exhibits reversible specific capacity of 245, 220, 198, 176, 157, and 127 mA·hour g−1 at 10, 25, 50, 100, 200, and 500 mA g−1, respectively (Fig. 6D). For comparison, bulk mesoporous Nb2O5/carbon composites (meso-Nb2O5/C) and commercial Nb2O5 (c-Nb2O5) were tested (figs. S17 and S18; table S3). At all current densities, bm-Nb2O5/C shows about two and four times higher specific capacity than meso-Nb2O5/C and c-Nb2O5, respectively (Fig. 6D). Even at extremely high current density of 1000 mA g−1, bm-Nb2O5/C retains 60% of specific capacity (105 mA·hour g−1) at 100 mA g−1, suggesting excellent rate capability (fig. S19). Notably, bm-Nb2O5/C exhibits the highest rate capability among the previously reported metal oxide anodes (fig. S20A and table S4) (4657). After galvanostatic cycles, the specific capacity could be recovered to 201 mA·hour g−1 when the current density was reset to 50 mA g−1. Moreover, bm-Nb2O5/C shows good reversible cycle performance and CE (Fig. 6E). After 100 cycles at a current density of 50 mA g−1, the specific capacity of ~189 mA·hour g−1 and CE of ~99% were still maintained without considerable capacity fading. The long-term cyclability test was further performed at 500 mA g−1 for 1000 cycles (fig. S21). The reversible capacity slightly decreases from 132 mA·hour g−1 to 110 mA·hour g−1 with average CE of 99.6%, which exhibits higher electrochemical performance after long-term cycling than even conversion-type anode materials (fig. S20B).

Fig. 6 Electrochemical characterization of bm-Nb2O5/C particles.

(A) Cyclic voltammetry curves of bm-Nb2O5/C electrode at a scan rate of 0.1 mV s−1. (B) Charge/discharge curves of bm-Nb2O5/C at a current of 10 mA g−1. (C) Charge/discharge voltage profiles and (D) rate performance at various current densities from 10 to 1000 mA g−1. (E) Cycling performance of bm-Nb2O5/C electrode at 50 mA g−1 and the corresponding coulombic efficiency (CE).

The outstanding electrochemical performance is mostly attributed to unique morphological features of the bm-Nb2O5/C. The uniform concavities can contain large amount of electrolyte and facilitate its access to the anode material. The regular mesopores allow K+ to easily penetrate internally via comprehensive electrode/electrolyte contact, thus leading to decrease in activation barrier and enhancement in reaction kinetics (58). In addition, the nanosized Nb2O5/C frameworks can increase the rate of K+ intercalation/deintercalation remarkably, because they shorten the diffusion pathway for the K+ transport within the crystalline Nb2O5. The mesoporous structures also provide the sufficient space to accommodate the large volume strain during the potassium ion intercalation, particularly at high rates, which is beneficial to the stable cyclability. Importantly, bowl-like morphology and uniform particle size/shape allow higher packing density than irregular bulk particles (fig. S22, A and B), thereby resulting in high volumetric capacity (fig. S22C) (24). Furthermore, anisotropic morphology provides the short path length for electron/mass transport and increases the contact area between adjacent mesoporous bowls leading to enhanced charge transfer (6, 20, 25). All results suggest that the distinctive structural features of bm-Nb2O5/C greatly contribute to the excellent rate performance and the cycling stability in KIBs.


In conclusion, we have developed a facile strategy to prepare anisotropic bowl-like mesoporous inorganic materials with controllable particle shape, particle/concavity size, pore size, and chemical composition. To induce macroscopically anisotropic morphologies, we designed the asymmetric polymer-polymer interfacial tension–mediated multiphase assembly in the ternary blends; the asymmetric self-organization of two minor phases enables the formation of bowl-like particles and 2D nanosheets, while BCP-directed co-assembly generates ordered mesostructures. Benefiting from the unique structures and morphologies, the bowl-like mesoporous metal oxides exhibited highly enhanced specific capacity and excellent cycle stability as anodes in KIB. Our approach provides new insights into the rational design of inorganic porous materials with complex morphologies that are desired in a myriad of potential applications.


Synthesis of AS sol

A total of 2.703 g of (3-glycidyloxypropyl-)trimethoxysilane (Glymo), 0.313 g of aluminum sec-butoxide [Al(OBus)3] (Si:Al molar ratio is 9:1), and 0.011 g of KCl were added in a vial placed in an ice bath. After 5 min of stirring, 0.135 g of 0.01 M HCl was added to the solution and stirred for 15 min at 0°C. The ice bath was removed, and the solution was further stirred at room temperature. After 15 min, 0.765 g of HCl was added slowly for 7 min with stirring. After 20 min, the solution was filtrated using a 0.2-μm Polytetrafluoroethylene (PTFE) syringe filter to remove KCl (35).

Synthesis of anisotropic bm-AS

In a typical synthesis of ms-AS with hexagonally ordered mesopores, PEO-b-PS, hPMMA (350 kg mol−1), and hPS (350 kg mol−1) were dissolved in THF to make a 5 wt % polymeric solution. Then, proper amount of AS sol was added in sequence. The homogeneous solution consisting of PEO-b-PS:hPMMA:hPS:THF:AS sol:H2O (0.1:0.6:0.02 to 0.2:13.7 to 17.1:0.27:0.08 g) was stirred for 1 hour. Subsequently, the solution was poured into a glass dish placed on a hot plate at 40°C to evaporate THF and H2O. After solvent evaporation, polymer/inorganic hybrids were further annealed at 100°C for 24 hours. The as-made samples were calcined at 550°C for 4 hours under air with a heating rate of 1°C min−1.

Synthesis of 2D mns-AS

To prepare mns-AS, we use the mixed solution at rM of 40 consisting of 0.025 g of PEO-b-PS, 1 g of hPMMA, 1 g of hPS, and 0.08 g of AS sol. The following procedures were identical to those used for the synthesis of bm-AS.

Synthesis of phenol-formaldehyde resin (resol)

2 g of Phenol (26.5 mmol) was melted at 40°C and 0.55 g of 20 wt% NaOH aqueous solution (2.65 mmol) was added slowly. After 20 min of stirring, 4.3 g of formaldehyde solution (53 mmol) was mixed with the phenol solution. The phenol-formaldehyde solution was reacted at 70°C for 1 hr with slow stirring, and then cooled at room temperature. The reacted solution was neutralized by 2 M HCl solution. After the water removal under vacuum condition below 50°C, THF was poured into resol to precipitate NaCl produced during neutralization. NaCl was removed by filtration, and then THF was removed using vacuum evaporator below 50°C.

Synthesis of bm-carbon

The mixed solution of PEO-b-PS:hPMMA:hPS:THF:AS sol:resol:water (0.1:0.6:0.1:15.2:0.15:0.1:0.046 g) was prepared. A total of 0.5 g of 20 wt % resol solution in THF was added, and the solution was stirred for 1 hour. The casting and annealing procedures were identical to those described for bm-AS particles. For carbon, the as-made hybrids were carbonized at 900°C for 2 hours under Ar with the heating rate of 1°C min−1. After carbonization, the AS species was removed by using 5 wt % HF solutions.

Synthesis of bm-transition metal oxides

To prepare bm-transition metal oxides, we used highly acidic solution by using concentrated HCl for controlled sol-gel reaction of reactive metal alkoxide precursors. The highly acidic condition prevented rapid inorganic condensation by protonation of metal hydroxyl groups (M-OH), resulting in the formation of small inorganic oligomers; they can favorably mix with PEO-blocks of BCP, thereby leading to ordered mesostructures without macrophase separation. For bm-Nb2O5, 0.1 g of PEO-b-PS, 0.6 g of hPMMA, and 0.1 g of hPS were dissolved in 15.2 g of THF. A total of 0.57 g of niobium ethoxide and 0.28 g of concentrated HCl were added dropwise to the solution in sequence. The mixture was stirred for 1 hour and poured into the glass dish. The solvent was evaporated at 40°C for several hours, and the as-made Nb2O5/organic hybrids were annealed at 100°C for 24 hours. Pyrolysis at 700°C under inert atmosphere leads to the formation of the discrete bm-Nb2O5/C particles.

To synthesize bm-TiO2, the mixed solution containing 0.1 g of PEO-b-PS, 0.6 g of hPMMA, 0.1 g of hPS, and 15.2 g of THF was prepared. A total of 0.45 g of titanium isopropoxide and 0.24 g of concentrated HCl were added slowly to the solution in sequence. The remaining procedures were identical with those used for bm-Nb2O5/C. To remove organic species and crystallize TiO2, the as-made TiO2 hybrids were calcined at 450°C under air for 2 hours.

Materials characterization

Gel permeation chromatography (GPC; Waters) was performed using THF as the eluent, and the molar mass was calibrated on the basis of PS standards. The structure of bowl-like mesoporous inorganic particles was characterized by TEM (JEM-1011, JEOL Ltd.) and SEM (S-4200 field-emission SEM, Hitachi). The nitrogen physisorption was carried out at 77 K using a TriStar II 3020 (Micromeritics Instrument Co.). Powder XRD patterns were obtained using Rigaku D/MAX-2500/PC x-ray diffractometer (Cu Kα). SAXS measurements were carried out on the 4C SAXS beamline at the Pohang Light Source II. AFM measurements were conducted using a Veeco Dimension 3100 instrument in tapping mode.

Electrochemical characterization

For half-cell test, K metal and bm- or meso-Nb2O5/C were used as counter and working electrode, respectively. The battery performance of half-cell was evaluated in a half-cell configuration using CR2032-type coin cells, which were assembled in Ar-filled glove box. The separator was GF/F glass microfiber filters (Whatman, USA). The electrolyte was 1 M potassium bis(fluorosulfonyl)imide (KFSI) dissolved in a mixture (1:1, v/v) of ethylene carbonate and dimethyl carbonate. The working electrode was prepared via typical slurry and casting method. The slurry was made using mixture of 70 wt % bm-Nb2O5/C, 20 wt % super-p, and 10 wt % carboxyl methyl cellulose and then pasted on Cu foil. The galvanostatic electrochemical test was assessed by a WBCS-3000 battery cycler (WonATech Co., Korea) in the potential range of 0.01 to 3 V (versus K/K+). Electrochemical impedance spectroscopy was conducted via the potentiostat (Reference 600, Gamry Instruments, USA) with an amplitude of 5 mV from 105 to 0.001 Hz.


Supplementary material for this article is available at

This is an open-access article distributed under the terms of the Creative Commons Attribution-NonCommercial license, which permits use, distribution, and reproduction in any medium, so long as the resultant use is not for commercial advantage and provided the original work is properly cited.


Acknowledgments: Funding: This research was supported by the National Research Foundation of Korea (NRF) grant funded by the Korean government (MSIT) (2018M1A2A2061987, 2019M3E6A1064706, and 2019M3D1A1079306). Author contributions: S.K., J.H., and Jinwoo Lee conceived the idea. S.K. and J.H. performed the synthetic experiments and the characterization. Jisung Lee carried out the electrochemical test. S.K., J.H., and Jinwoo Lee wrote the manuscript. All authors discussed the results and commented on the manuscript. The project was planned, directed, and supervised by Jinwoo Lee. Competing interests: The authors declare that they have no competing interests. Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper and/or the Supplementary Materials. Additional data related to this paper may be requested from the authors.
View Abstract

Stay Connected to Science Advances

Navigate This Article