All reagents, such as FeCl₃·6H₂O, NaOH, and H₃BO₃, were of analytical grade and used as received without further purification. Monodisperse α-Fe₂O₃ particles were synthesized via a coprecipitation of FeCl₃ and NaOH solutions at room temperature, followed by a facile hydrothermal treatment of the slurry in the presence of H₃BO₃ as the additive. In a typical procedure, 1.281 g of H₃BO₃ was poured into 10.1 mL of deionized (DI) water, then 9.3 mL of FeCl₃ (1.5 mol·L⁻¹) solution was added, and finally 7.0 mL of NaOH (4 mol·L⁻¹) solution was dropped into the above mixed solution under vigorous magnetic stirring at room temperature, with the molar ratio of FeCl₃/H₃BO₃/NaOH as 2:3:4. After 5 min of stirring, 26.4 mL of the resultant brown slurry was transferred into a Teflon-lined stainless steel autoclave with a capacity of 44 mL. The autoclave was sealed and heated to 90°C to 210°C (heating rate 2°C·min⁻¹) and kept under an isothermal condition for 1.0 to 24.0 h, and then cooled down to room temperature naturally. The product was filtered, washed with DI water for three times, and finally dried at 80°C for 24.0 h for further characterization. To evaluate the effects of the molar ratio of the reactants, the molar ratio of FeCl₃/H₃BO₃/NaOH was altered within the range of 2:(0–3):(2–6), with other conditions unchanged.

The electrochemical evaluation of the Fe₂O₃ NPs and nanoarchitectures as anode materials for lithium-ion batteries were carried out using CR2025 coin-type cells with lithium foil as the counter electrode, microporous polyethylene (Celgard 2400, Charlotte, NC, USA) as the separator, and 1.0 mol·L⁻¹ LiPF₆ dissolved in a mixture of ethylene carbonate, dimethyl carbonate, ethylene methyl carbonate (1:1:1, by weight) as the electrolyte. All the assembly processes were conducted in an argon-filled glove box. For preparing working electrodes, a mixed slurry of hematite, carbon black, and polyvinylidene fluoride with a mass ratio of 80:10:10 in N-methyl-2-pyrrolidone solvent was pasted on pure Cu foil with a blade and was dried at 100°C for 12 h under vacuum conditions, followed by pressing at 20 kg·cm⁻². The galvanostatic discharge/charge measurements were performed at different current densities in the voltage range of 0.01 to 3.0 V on a Neware battery testing system (Shenzhen, China). The specific capacity was calculated based on the mass of hematite. Cyclic voltammogram measurements were performed on a Solartron Analytical 1470E workstation (Farnborough, UK) at a sweep rate of 0.1 mV·s⁻¹.

The crystal structures of the samples were identified using an X-ray powder diffractometer (XRD; D8-Advance, Bruker, Karlsruhe, Germany) with a Cu K_α radiation (λ = 1.5406 Å) and a fixed power source (40.0 kV, 40.0 mA). The morphology and microstructure of the samples were examined using a field-emission scanning electron microscope (SEM; JSM 7401 F, JEOL, Akishima-shi, Japan) operated at an accelerating voltage of 3.0 kV. The size distribution of the as-synthesized hierarchical architectures was estimated by directly measuring ca. 100 particles from the typical SEM images. The N₂ adsorption-desorption isotherms were measured at 77 K using a chemisorption-physisorption analyzer (Autosorb-1-C, Quantachrome, Boynton Beach, FL, USA) after the samples had been outgassed at 300°C for 60 min. The specific surface area was calculated from the adsorption branches within the relative pressure range of 0.10 to 0.31 using the multipoint Brunauer-Emmett-Teller (BET) method, and the pore size distribution was evaluated from the N₂ desorption isotherm using the Barrett-Joyner-Halenda method. The optical properties were examined using a UV–vis spectrophotometer (Cary 300, Varian, Palo Alto, CA, USA), with absolute alcohol as the dispersive medium.

Figure 1 shows the influences of the molar ratio of FeCl₃/H₃BO₃/NaOH on the compositions and morphologies of the hydrothermal products obtained at 150°C for 12.0 h. When changing the molar ratio of FeCl₃/H₃BO₃/NaOH within the range of 2:(0–3):(2–6), all products were composed of pure-phase hematite (α-Fe₂O₃, JCPDS No. 33–0664), with a detectable slight difference of the crystallinity (Figure 1a). With the molar ratio of FeCl₃/H₃BO₃/NaOH changed from 2:0:6 to 2:0:4 and to 2:0:2, the crystallinity of hematite decreased slightly (Figure 1a ₁,a₂,a₃). In contrast, the morphologies of the obtained products varied significantly with the change of the molar ratio of reactants. Quasi-spherical hematite NPs with a diameter of 30 to 150 nm were obtained when the molar ratio of FeCl₃/H₃BO₃/NaOH was 2:0:6 (Figure 1b,b₁), similar to the so-called α-Fe₂O₃ nanopolyhedra synthesized in the ammonia-water system at 180°C for 8.0 h 23 . With the molar ratio decreased to 2:0:4 and 2:0:2, hierarchical pod-like (with elliptical ends and relatively uniform diameter along the long axial direction, Figure 1c) and peanut-type nanoarchitectures (with relatively sharp elliptical ends and saddle-shaped middle part, Figure 1d,d₁) were acquired, respectively. The pod-like architectures contained 1D or linear chain-like assemblies of smaller nanoparticles or rod-like subcrystals within the body (as shown in red dotted elliptical and rectangular regions in Figure 1c), with distinct cavities on the surfaces (Figure 1c). The peanut-type nanoarchitectures (Figure 1d,d₁) also comprised small nanoparticles within the body whereas with not so distinct cavities on the surfaces owing to the relatively compact assembly. Similar 1D assemblies, such as rod-like subcrystals and linear chains of interconnected primary particles, have also been found to exist as the subunits of peanut-type 45 and double-cupola 46 hematite, respectively. Obviously, the molar ratio of 2:0:6 (FeCl₃/H₃BO₃/NaOH) led to nearly monodisperse hematite NPs, whereas the molar ratio of 2:0:4 and 2:0:2 resulted in porous hierarchical architectures with different morphologies. According to Sugimoto’s research 45 47 48 , size control is generally performed by controlling the number of nuclei during the nucleation stage, and nucleation occurs during the addition of NaOH solution into FeCl₃ solution. In the present case, the molar ratio of FeCl₃/H₃BO₃/NaOH as 2:0:6 is the stoichiometric ratio for the formation of colloidal Fe(OH)₃ at room temperature, which led to the greatest degree of supersaturation of Fe(OH)₃ and further resulted in the largest number of nuclei and ultimately brought the quasi-spherical α-Fe₂O₃ NPs.

However, when H₃BO₃ was introduced into the reaction system, e.g., the molar ratio of FeCl₃/H₃BO₃/NaOH was designed as 2:0.3:4 (Figure 1a ₄,e,e₁) and 2:1.5:4 (Figure 1a ₅,f,f₁), relatively uniform porous pod-like hematite nanoarchitectures were obtained. For the ratio of 2:0.3:4, 90% of the nanoarchitectures have an aspect ratio (ratio of longitudinal length to latitude diameter) within 1.4 to 1.8 (Figure 1e ₁). For the hematite obtained from a molar ratio of FeCl₃/H₃BO₃/NaOH as 2:1.5:4, 95% of the nanoarchitectures have an aspect ratio within 1.4 to 1.8 (Figure 1f ₁). Therefore, the introduction of H₃BO₃ not only preserved the shape of hematite particles, but also improved the morphology uniformity of the nanoarchitectures. This situation was different from that of the formation of peanut-type hematite, which evolved from pseudocubic particles via an ellipsoidal shape with the increasing concentration of the additive such as sulfate or phosphate 49 . On the other hand, compared with those organic surfactant-assisted solvothermal or other solution-based calcination methods, the introduced H₃BO₃ in the present case could be easily removed via DI water washing and then reused, indicating the environmentally benign characteristic.

The compositions and morphologies of the hydrothermal products obtained at various temperatures for 12.0 h were tracked so as to further understand the corresponding evolution, as shown in Figure 2. Obviously, the hydrothermal temperature had significant influences on the compositions as well as the morphologies of the products. The sample hydrothermally treated at 90°C was composed of relatively poor-crystallinity and low-aspect-ratio akaganeite (β-FeOOH, JCPDS No. 34–1266, Figure 2a ₁) nanorods or nanofloccules (Figure 2b). When hydrothermally treated at 105°C, the product gradually changed into poor-crystallinity α-Fe₂O₃ (Figure 2a ₂, JCPDS No. 33–0664) of pod-like and pumpkin-like nanoarchitectures (Figure 2c). Moreover, the local details showed that the nanoarchitecture consisted of short 1D nanostructured subunits and tiny NPs (Figure 2c ₁). When treated at 120°C, α-Fe₂O₃ nanoarchitectures with greatly improved crystallinity (Figure 2a ₃) and uniform compact pod-like morphology (Figure 2d) were formed, 87% of which had a longitudinal length of 2.2 to 2.5 μm (Figure 2d ₁). Notably, the compact nanoarchitecture contained numerous tiny NPs onto the surfaces. With the temperature increasing up to 150°C and keeping it constant for 12.0 h, the products comprised uniform porous pod-like α-Fe₂O₃ with higher crystallinity (Figure 2a ₄) and multitudinal cavities on the surfaces (Figure 2e,f), 84% of which had a longitudinal length of 2.6 to 3.2 μm 44 . The morphology of the present pod-like α-Fe₂O₃ nanoarchitectures was somewhat similar to that of the melon-like microparticles by the controlled H₂C₂O₄ etching process 25 . With the temperature further going up to 180°C, porous pod-like α-Fe₂O₃ nanoarchitectures with further improved crystallinity (Figure 2a ₅) and more and larger cavities on the surfaces were obtained (Figure 2g), 84% of which had a longitudinal length of 2 to 2.4 μm (Figure 2g ₁). When hydrothermally treated at 210°C for 12.0 h, the product evolved into high-crystallinity whereas entirely loose porous α-Fe₂O₃ nanoarchitectures (Figure 2a ₆,h), 84% of which had a longitudinal length of 2.1 to 2.7 μm (Figure 2h ₁).

It was worth noting that when treated at a temperature from 90°C to 210°C for 12.0 h, the overall crystallinity of the products became higher (Figure 2a ₂,a₃,a₄,a₅,a₆), and the NPs and cavities within the α-Fe₂O₃ nanoarchitectures grew larger. The product evolved from compact pod-like nanoarchitectures (Figure 2c,d) to loose (Figure 2e,f) and to looser (Figure 2g,h) pod-like nanoarchitectures. As a matter of fact, with the temperature going up from 120°C to 150°C, to 180°C, and to 210°C, the crystallite size along the [104] direction, i.e., D ₁₀₄, calculated by the Debye-Scherrer equation also increased from 23.3 to 27.3, to 28.0, and to 31.3 nm, respectively. This was in accordance with the direct observation on the gradual increase in the NP size within the nanoarchitectures (Figure 2d,e,f,g,h), thus accounted for the gradual sharper tendency for the XRD patterns of the corresponding hydrothermal products (Figure 2a ₃,a₄,a₅,a₆) obtained from 120°C to 210°C. Analogous to those obtained previously (Figure 1c,e,f), the nanoarchitectures obtained at 150°C to 210°C for 12.0 h were speculated to be constituted of 1D assemblies (Figure 2e,f) or NPs (Figure 2g,h).

Figure 3 shows the N₂ adsorption-desorption isotherms and corresponding pore size distributions of the hydrothermally synthesized α-Fe₂O₃ nanoarchitectures with typical morphologies. Influences of the temperature on the porous α-Fe₂O₃ nanoarchitectures are summarized in Table 1. As listed, the selected nanoarchitectures 1, 2, 3, and 4 corresponded with those obtained at 120°C (Figure 2d), 150°C (Figure 2e,f), 180°C (Figure 2g), and 210°C (Figure 2h) for 12.0 h, respectively. All N₂ adsorption-desorption isotherms of the nanoarchitectures exhibited type IV with an H3-type hysteresis loop. The compact pod-like nanoarchitecture 1 (Figure 2d, D ₁₀₄ = 23.3 nm) had a relatively large adsorbance of N₂ (Figure 3a ₁) with a broad hysteresis loop at a relative pressure P/P ₀ of 0.45 to 0.95 and a very narrow pore diameter distribution concentrating on 3.8 nm (Figure 3a ₂). In contrast, the relative loose pod-like nanoarchitecture 2 (Figure 2e,f, D ₁₀₄ = 27.3 nm) showed a relatively small adsorbance of N₂ (Figure 3b ₁) with a typical H3-type hysteresis loop at a relative pressure P/P ₀ of 0.45 to 1.0 and a bimodal pore diameter distribution concentrating on 3.8 and 17.5 nm (Figure 3b ₂). The characteristic N₂ adsorption-desorption isotherms (Figure 3a ₁,b₁) and pore size distributions (Figure 3a ₂,b₂) revealed that both nanoarchitectures 1 and 2 are of mesoporous structures.

Comparatively, the looser pod-like nanoarchitecture 3 (Figure 2g, D ₁₀₄ = 28.0 nm) demonstrated a similar adsorbance of N₂ (Figure 3c ₁) whereas with a narrow hysteresis loop at a relative pressure P/P ₀ of 0.40 to 0.95 and a quasi-bimodal pore diameter distribution (Figure 3c ₂). Very similarly, the loosest pod-like nanoarchitecture 4 (Figure 2h, D ₁₀₄ = 31.3 nm) exhibited a relatively low adsorbance of N₂ (Figure 3d ₁) with also a narrow hysteresis loop at a relative pressure P/P ₀ of 0.25 to 0.95 as well as a quasi-bimodal pore diameter distribution (Figure 3d ₂). It was worth noting that the broad hysteresis loop (Figure 3a ₁) and relative narrow one (Figure 3b ₁) were due to the strong and weak capillarity phenomena existing within the compact (Figure 2d) and relatively loose nanoarchitectures (Figure 2e), respectively. Moreover, the characteristic H3-type hysteresis loop (Figure 3b ₁) indicated the existence of dominant slit pores and channels with a relatively uniform shape and size within the relatively loose pod-like nanoarchitectures (Figure 2e,f). This was in accordance with the SEM observation (Figure 1c) and literature results 45 46 . The thin hysteresis loops (Figure 3c ₁,d₁) were due to the slight capillarity phenomenon existing within the very loose nanoarchitectures (Figure 2g,h).

As shown in Table 1, with the temperature increasing from 120°C to 150°C, to 180°C, and to 210°C, the corresponding multipoint BET specific surface area of the nanoarchitecture decreased from 21.3 to 5.2, to 2.6, and to 2.0 m²·g⁻¹, respectively. Meanwhile, the total pore volume changed from 3.9 × 10⁻² to 2.9 × 10⁻², to 2.9 × 10⁻², and to 2.1 × 10⁻² cm³·g⁻¹, with a roughly decreasing tendency; the average pore diameter changed from 7.3 to 22.1, to 44.7, and to 40.3 nm, with a roughly increasing tendency. Thus, according to the general recognition of the porous materials 50 , nanoarchitectures 3 and 4 were determined as the mesoporous structures, whereas the pore diameters were near the macropores category. As a matter of fact, with the temperature increasing from 120°C to 210°C, the evolution of the BET specific surface area, total pore volume, and average pore diameter of the various-morphology pod-like α-Fe₂O₃ nanoarchitectures agreed with the variation of the D ₁₀₄ calculated by the Debye-Scherrer equation, also in accordance with the SEM observation (Figure 2d,e,f,g,h).

Since the compact pod-like nanoarchitecture obtained at 105°C for 12.0 h (Figure 2c) bridged 1D β-FeOOH nanostructures and pod-like α-Fe₂O₃ nanoarchitectures, the composition and morphology of the products hydrothermally treated at 105°C for various times were monitored, as shown in Figure 4. All hydrothermal products obtained at 105°C for 1.0 to 12.0 h exhibited relatively poor crystallinity (Figure 4a ₁,a₂,a₃). When treated for 1.0 h, the product was composed of β-FeOOH (JCPDS No. 34–1266) and detectable trace amount of maghemite (γ-Fe₂O₃, JCPDS No. 25–1402) in a nearly amorphous state (Figure 4a ₁,b). With the time extending to 3.0 h, the product was only β-FeOOH with improved crystallinity, and γ-Fe₂O₃ no longer existed (Figure 4a ₂,c). Notably, β-FeOOH at that period exhibited very tiny primary 1D morphology (i.e., fibrils, Figure 4c ₁), and a rudimental pod-like aggregate was also observed (denoted as yellow dotted elliptical region in Figure 4c). When treated for 6.0 h, the hydrothermal products containing trace amount of β-FeOOH and majority of newly formed α-Fe₂O₃ (Figure 4a ₃ were acquired, exhibiting pod-like or ellipsoidal-shaped aggregates entangled with 1D nanostructures (Figure 4d). The enlarged image (Figure 4e) corresponding to the red dot-dashed rectangular region in Figure 4d clearly showed that the selected developing pod-like aggregate was assembled by 1D β-FeOOH nanowhiskers. In other words, the pod-like aggregate did not simply coexist or was not simply coated with, but constructed by 1D β-FeOOH nanostructures. With the time prolonged to 12.0 h, as mentioned previously, the pure phase of α-Fe₂O₃ nanoarchitectures consisted of very tiny NPs with compact pod-like and pumpkin-like morphologies acquired (Figure 2a ₂,c). The crystallite size D ₁₀₄ calculated by the Debye-Scherrer equation was 20.5 nm, smaller than that of the compact pod-like α-Fe₂O₃ nanoarchitectures obtained at 120°C for 12.0 h (Figure 2d) due to a relatively lower temperature hydrothermal treatment.

From the phase conversion and morphology evolution of the hydrothermal products, formation of the monodisperse pod-like α-Fe₂O₃ phase could be further clarified, which experienced a two-step phase transformation from Fe(OH)₃ to β-FeOOH and from β-FeOOH to α-Fe₂O₃ 51 52 . The room-temperature coprecipitation of FeCl₃ and NaOH solutions and hydrolysis of excessive Fe³⁺ ions can be expressed as

Fe 3 + aq + 3 OH − aq → Fe OH 3 amorphous gel

Fe 3 + aq + 3 H 2 O → Fe OH 3 amorphous gel + 3 H + aq

Hydrothermal conversion of amorphous Fe(OH)₃ gel can be expressed as

Fe OH 3 gel → β − FeOOH fibrils + H 2 O

2 β − FeOOH fibrils → α − Fe 2 O 3 mesoporous + H 2 O

As known, iron oxyhydroxides (FeOOH) can be crystallized as goethite (α-FeOOH), lepidocrocite (γ-FeOOH), and akaganeite (β-FeOOH), and an environment rich of Cl⁻ was favorable for the formation of β-FeOOH phase 53 . In the present case, a molar ratio of the reactants as FeCl₃/H₃BO₃/NaOH = 2:(0–3):4 led to a surrounding rich of Cl⁻ and thus promoted the formation of β-FeOOH. Tiny β-FeOOH fibrils with poor crystallinity formed at the early stage of the hydrothermal treatment (e.g., 90°C, 12.0 h, Figure 2a ₁; 105°C, 1.0 to 3.0 h, Figure 4a ₁,a₂) tended to agglomerate with each other owing to the high surface energy, leading to quasi-amorphous agglomerate bulks of irregular shape (Figures 2b and 4b,c). Undoubtedly, the conversion from β-FeOOH to α-Fe₂O₃ was crucial to the formation of mesoporous pod-like hematite nanoarchitectures. Sugimoto et al. reported a preparation of monodisperse peanut-type α-Fe₂O₃ particles from condensed ferric hydroxide gel in the presence of sulfate 49 and found that ellipsoidal hematite turned into a peanut-like shape with the increase in the concentration of sulfate 51 . In the present case, although quasi-spherical α-Fe₂O₃ NPs were obtained in due case (Figure 1b), the mesoporous hematite nanoarchitectures (Figures 1c,d,e,f and 2d,e,f,g,h) were not directly assembled by those NPs, taking into consideration the remarkable differences of the morphology especially size between the NPs and subunits of nanoarchitectures. It was worth noting that the hydrothermally formed hematite particles exhibited a peanut-like shape at the molar ratio of FeCl₃/H₃BO₃/NaOH as 2:0:2 (Figure 1d) and a pod-like shape at the molar ratio of FeCl₃/H₃BO₃/NaOH as 2:(0–3):4 (Figures 1c,e,f and 2d,e,f,g,h). Moreover, with the content of H₃BO₃ increasing, the pod-like α-Fe₂O₃ nanoarchitectures tended to be uniform in size distribution. Consequently, the morphology evolution of the hydrothermally synthesized α-Fe₂O₃ nanoarchitectures in the presence of boric acid, from a peanut-type to a pod-like shape, was obviously different from that of the peanut-type α-Fe₂O₃ particles that originated from condensed ferric hydroxide gel in the presence of sulfate 49 .

Thus, based on the present experimental results (Figures 1, 2, 3, and 4), the overall formation mechanism of mesoporous pod-like hematite nanoarchitectures in the presence of boric acid was illustrated in Figure 5. Firstly, the amorphous Fe(OH)₃ gel derived from room-temperature coprecipitation was hydrothermally treated under an environment rich of Cl⁻, leading to poor-crystallinity β-FeOOH fibrils (Figure 5a) 53 . Secondly, with the hydrothermal temperature going up and time going on, β-FeOOH fibrils were organized into a peanut-type assembly, and at the same time, β-FeOOH fibrils began to dissolve, resulting in α-Fe₂O₃ NPs. As a consequence, peanut-like β-FeOOH/α-Fe₂O₃ assemblies were obtained (Figure 5b). This process was very analogous to the ‘rod-to-dumbbell-to-sphere’ transformation phenomenon, which had been found in the formation of some other hierarchical architectures, such as carbonates (CaCO₃, BaCO₃, SrCO₃, MnCO₃, CdCO₃) 8 54 55 , fluoroapatite (Ca₅(PO₄)₃OH) 56 , etc. Like the dumbbell transition structure, the present peanut-type assembly was also believed to be formed due to the reaction-limited aggregation. Thirdly, with the hydrothermal treatment further going on, remanent β-FeOOH fibrils were further dissolved and the peanut-like β-FeOOH/α-Fe₂O₃ assemblies were converted into relatively compact pod-like α-Fe₂O₃ nanoarchitectures, consisting of 1D or linear chain-like assemblies of rod-like subcrystals or tiny NPs within the body (Figure 5c). No proof convinced that the peanut-type β-FeOOH/α-Fe₂O₃ assemblies were thoroughly dissolved and reorganized into the pod-like nanoarchitectures with almost unchanged external shape and size. In other words, peanut-like β-FeOOH/α-Fe₂O₃ assemblies were in situ transformed into α-Fe₂O₃ NPs within the peanut-like aggregates owing to the hydrothermal treatment. However, the in situ converted tiny α-Fe₂O₃ NPs bore high surface energy. This promoted the aggregation, instead of the segregation, of those tiny NPs so as to reduce the overall surface energy, leading to relatively compact pod-like α-Fe₂O₃ nanoarchitectures due to a slight expansion of the entire volume. Finally, with the hydrothermal treatment going on, the compact pod-like α-Fe₂O₃ nanoarchitectures became looser and looser owing to the coarsening 57 58 of the constitutional NPs controlled by the traditional Ostwald ripening, i.e., dissolution-reprecipitation mechanism (Figure 5d) 58 . The constitutional α-Fe₂O₃ subcrystals grew into larger NPs, with 1D assembly behavior disappeared largely.

It is notable, however, that the boric acid played a significant role in the formation of the present mesoporous pod-like α-Fe₂O₃ nanoarchitectures with uniform morphology and size, confirmed by the above experimental results (Figures 1 and 2). Also, as confirmed to improve the uniformity, the amount of boric acid or molar ratio of FeCl₃/H₃BO₃/NaOH should be tuned within a certain composition range. As known, as a weak acid, H₃BO₃ could form sodium borate (i.e., borax) after the introduction of NaOH, giving rise to the buffer solution. This could tune the release of hydroxyl ions and further control the mild formation of amorphous Fe(OH)₃ gel, leading to subsequent β-FeOOH fibrils with relatively uniform size. This was believed to contribute to the further formation of the peanut-like β-FeOOH/α-Fe₂O₃ assemblies and ultimate occurrence of the pod-like α-Fe₂O₃ nanoarchitectures.

Hematite NPs have been widely used as ultraviolet absorbents for their broad absorption in the ultraviolet region from the electron transmission of Fe-O. Figure 6 shows the optical absorbance spectra of the α-Fe₂O₃ particles with the photon wavelength in the range of 350 to 650 nm. For sample a₁, it revealed two absorption edges around 380 to 450 and 540 to 560 nm, which were consistent with the reported hematite NPs 59 60 61 . When the α-Fe₂O₃ clustered into samples b₁ and c₁, the size of α-Fe₂O₃ agglomerates was around 500 to 800 nm. The absorbance spectra showed two absorption peaks around 520 to 570 and 600 to 640 nm. The change in the degree of transition depended on the shape and size of the particles. When the hematite particles aggregated to pod-like nanoarchitectures, the size became larger, and then the scattering of visible light was superimposed on the absorption of as-prepared architectures.

It was well illustrated that three types of electronic transitions occurred in the optical absorption spectra of Fe³⁺ substances: (a) the Fe³⁺ ligand field transition or the d d transitions, (b) the ligand to metal charge-transfer transitions, and (c) the pair excitations resulting from the simultaneous excitations of two neighboring Fe³⁺ cations that are magnetically coupled. According to 62 63 , the absorption bands near 390 and 430 nm corresponded to the ⁶A₁ → ⁴E(⁴G) and ⁶A₁ → ⁴E, ⁴A₁(⁴G) ligand field transitions of Fe³⁺ 59 60 . The observed edge at around 520 to 570 and 600 to 640 nm could be assigned to the ⁶A₁ → ⁴ T₂(⁴G) ligand field transition of Fe³⁺. As revealed by Figure 6, the electronic transition for the charge transfer in the wavelength region 380 to 450 nm dominated the optical absorption features of the NPs, while the ligand field transitions in the range of 520 to 640 nm dominated the optical absorption features of the architectures. This indicated that the absorption could be modulated by controlling the size and shape of the hematite, which was quite important for the enhancement of the photoelectrocatalytic activity.

The electrochemical behavior of the hematite electrodes was evaluated by cyclic voltammetry and galvanostatic charge/discharge cycling. As shown in Figure 7a, a spiky peak was observed at 0.65 V with a small peak appearing at 1.0 V during the cathodic polarization of the hematite NPs (presented in Figure 1b) in the first cycle. This indicated the following lithiation steps 43 64 65 :

α − Fe 2 O 3 + 2 Li + + 2 e − → Li 2 Fe 2 O 3 cubic

Li 2 Fe 2 O 3 cubic + 4 Li + + 4 e − → 2 Fe 0 + 3 Li 2 O

With lithium ions inserted into the crystal structure of the as-prepared α-Fe₂O₃, the hexagonal α-Fe₂O₃ was transformed to cubic Li₂Fe₂O₃. The peak at 0.65 V corresponded to the complete reduction of iron from Fe²⁺ to Fe⁰ and the decomposition of electrolyte. A broad anodic peak was recorded in the range of 1.4 to 2.2 V, corresponding to the oxidation of Fe⁰ to Fe²⁺ and further to Fe³⁺ 66 67 . The curve of the subsequent cycle was significantly different from that of the first cycle as only one cathodic peak appeared at about 0.8 V with decreased peak intensity, while the anodic process only showed one broad peak with a little decrease in peak intensity. The irreversible phase transformation during the process of lithium insertion and extraction in the initial cycle was the reason for the difference between the first and second cathodic curves 24 . After the first discharge process, α-Fe₂O₃ was completely reduced to iron NPs and was dispersed in a Li₂O matrix. The decrease of the redox peak intensity implied that the capacity was decreased during cycling.

The charge–discharge curves of the α-Fe₂O₃ NP (shown in Figure 1b) electrode during the first and second cycles are shown in Figure 7b. In the first discharge curve, there was a weak potential slope located at 1.2 to 1.0 V and an obvious potential plateau at 0.9 to 0.8 V. The capacity obtained above 0.8 V was 780 mAh·g⁻¹ (4.6 mol of Li per mole of α-Fe₂O₃). After discharging to 0.01 V, the total specific capacity of the as-prepared α-Fe₂O₃ reached 887 mAh·g⁻¹, corresponding to 5.3 mol of Li per mole of α-Fe₂O₃. During the second cycle, the discharge curve only showed a slope at 1.0 to 0.8 V, and the capacity was reduced to 824 mAh·g⁻¹. Usually, the slope behavior during the discharge process of metal oxide anode materials was considered to be related with the irreversible formation of a nanocomposite of crystalline grains of metals and amorphous Li₂O matrix.

The comparison of the rate as well as cycling performances between Fe₂O₃ NPs and nanoarchitectures were also conducted, which were obtained by a 12.0-h hydrothermal treatment at 150°C with a molar ratio of FeCl₃/H₃BO₃/NaOH as 2:0:4 (Figure 1b) and 2:3:4 (Figure 2e), respectively. The discharge and charge capacities in the first cycle at a current of 0.1 C were 1,129 and 887 mAh·g⁻¹ for Fe₂O₃ NPs (Figure 7c) and 1,155 and 827 mAh·g⁻¹ for Fe₂O₃ nanoarchitectures. For the second cycle, the discharge and charge capacities were 871 and 824 mAh·g⁻¹ for Fe₂O₃ NPs and 799 and 795 mAh·g⁻¹ for Fe₂O₃ nanoarchitectures. The Li-ion storage capacitance of the current Fe₂O₃ NPs/nanoarchitectures reported in this work is higher than that of hematite nanorod (ca. 400 mAh·g⁻¹ at 0.1 C) 68 , nanoflakes 69 , hierarchial mesoporous hematite (ca. 700 mAh·g⁻¹ at 0.1 C) 65 , hollow nanospindles (457 mAh·g⁻¹ at 0.2 mA cm⁻²) 37 , hollow microspheres (621 mAh·g⁻¹ at 0.2 mA cm⁻²) 37 , and dendrites (670 mAh·g⁻¹ at 1 mA cm⁻²) 70 . When the current increased, both the discharge and charge capacities decreased, especially for Fe₂O₃ NPs (Figure 7c,e). The discharge and charge capacities of Fe₂O₃ nanoarchitectures were larger than those of Fe₂O₃ NPs. For instance, when the current rate increased to 2.0 C, the charge and discharge capacities of Fe₂O₃ nanoarchitectures were 253 and 247 mAh·g⁻¹, while those of Fe₂O₃ NPs were only 24 and 21 mAh·g⁻¹. This indicated that the Fe₂O₃ nanoarchitectures presented much improved rate performance for the reason that the porous nature of Fe₂O₃ nanoarchitectures allow a fast Li-ion diffusion by offering better electrolyte accessibility and also accommodate the volume change of NPs during Li insertion/extraction.

However, similar to many Fe₂O₃ nanostructures reported in literatures, the α-Fe₂O₃ nanoarchitectures exhibited a rapid capacity fading within the potential range of 0.01 to 3.0 V, suggesting that the crystalline structure of the electrode materials was destroyed by the insertion/extraction of lithium ions and the electrode decomposed the electrolyte. The Fe₂O₃ nanoarchitectures presented superior charge/discharge stability to the Fe₂O₃ NPs, e.g., the charging capacities of Fe₂O₃ nanoarchitectures (Figure 7f) and NPs (Figure 7d) of the tenth cycle were 503 and 356 mAh·g⁻¹, respectively. This indicated that the mesoporous structure of Fe₂O₃ nanoarchitectures provided more space for Fe₂O₃ volume change and avoided severe pulverization. Such an improvement could also be confirmed by the cycling performance of mesoporous hematite 67 , which maintained a good stability attributed from the small Fe₂O₃ size (ca. 10 nm) and abundant pores. The introduction of conductive carbon into the hematite electrode is an effective way to improve the cycle performance 68 . It is highly expected that the hierarchical Fe₂O₃ nanoarchitectures with ultrafine Fe₂O₃ building blocks and interconnected pores afford shorter Li-ion diffusion way, fast diffusion rate, and large-volume changes during the charge/discharge process, which can serve as potential anode materials for Li-ion storage.