Multi-element mixing boosts exfoliation of layered hexaniobate single crystals

Abstract

Scalable and green methods to prepare two-dimensional materials are important in the fields of catalysis, electronics, and energy conversion and storage. Existing approaches can be classified as bottom-up and top-down. Mass production through bottom-up approaches is costly, inefficient, and poorly scalable. Herein, we demonstrate a top-down approach based on chemical exfoliation. Without using strong acids or bases, single-crystalline nanosheets were obtained through hydrothermal treatment of multi-element-substituted layered oxides. Layered K₄Nb₆O₁₇ single crystals with multi-element substitution were grown using a flux method. The homogeneous distribution of Ti, Ta, and Sb ions in the niobate framework was confirmed experimentally. Subsequent hydrothermal treatment caused spontaneous exfoliation to form approximately 2.3 nm-thick nanosheets. The weight-based exfoliation efficiency of single crystals after multi-element substitution was five times higher than that of pristine K₄Nb₆O₁₇. Theoretical calculations revealed that substituting Nb with Ti, Ta, and Sb ions induced structural distortion in the octahedral metalate units in the K₄Nb₆O₁₇ framework, facilitating spontaneous exfoliation. By mixing framework elements to exfoliate layered materials efficiently, this methodology can be used for preparing 2D ‘inks’ through solution-based fabrication techniques.

---

Introduction

Two-dimensional (2D) materials have drawn attention because of their unexpected physical and chemical properties and diverse applications.¹ Metal oxide nanosheets several atoms in thickness have been prepared from non-van der Waals layered solids.^2,3 Considered promising alternatives to graphene, these 2D materials have great potential in fields including (electro)catalysis, (opto)electronics, energy conversion, and environmental remediation.⁴ For example, the hybridisation of nanosheets with organometallic complexes, bulky organic cations, and polymers enables to Z-scheme overall water splitting,⁵ hierarchical super-structured systems,⁶ and anisotropic hydrogels.⁷ Notably, a large lateral size (or aspect ratio) of the nanosheets is crucial for certain functionalities such as liquid crystallinity and macroscopic alignment because of steric hindrance, as shown by Onsager's theory.⁸ Hence, the scalable preparation of large 2D nanosheets is critical for commercial applications.⁹

Methods for preparing 2D materials are generally classified as top-down or bottom-up approaches. Compared to bottom-up synthesis, top-down synthesis via liquid phase exfoliation (LPE) is probably the most powerful route to produce 2D inks from various layered materials.^9,10 The LPE process usually involves the intercalation of a guest species and subsequent exfoliation of the host compounds. Unfortunately, the process requires strong acids and organic bases including HCl, BuLi, and [Bu₄N][OH]. Thus, more environmentally benign techniques are required.

In this study, we exploited the high-entropy effect of host layered oxides to facilitate their exfoliation. High-entropy oxides (HEOs), a new type of ceramic based on the concept of high-entropy alloys (HEAs),^11,12 contain oxygen and five or more types of cations at a single site. In an equiatomic five-cation system, the maximum configuration entropy (S_config) value is 1.61R (where R is the gas constant). The definition of S_config is provided in the SI. Compounds with S_config ≥ 1.5R are classified as ‘high entropy’, whereas those with 1.5R > S_config ≥ 1.0R and S_config < 1.0R are classified as ‘medium-’ and ‘low entropy’, respectively.¹² Although various polycrystalline HEOs have been synthesised,^13–17 the growth of single-crystalline HEOs is limited to pyrochlore-, garnet-, and aluminate-type compounds;^18–20 and the cations are restricted to yttrium and f-block rare earth elements. Notably, there have been no reports of HEO single crystals based on d-block elements. High-entropy nanosheets represent a new class of two-dimensional materials that combine multiple metal cations within a single layered lattice. Their compositional complexity leads to unique physicochemical properties such as enhanced structural stability, tunable electronic structures, and versatile surface functionalities. These characteristics provide significant advantages for various applications including catalysis, energy storage, and environmental remediation. In particular, their high configurational entropy promotes uniform cation distribution and suppresses phase segregation, resulting in improved durability and multifunctional performance.

Layered hexaniobate (K₄Nb₆O₁₇, KNO) possesses two different galley spaces and photocatalytic activity.^21–23 KNO single crystals can be exfoliated to form colloidal nanosheets,⁸ but this process requires toxic organic reagents to overcome the relatively strong interlayer electrostatic interactions. Adding a high-entropy effect to the KNO system distorts the NbO₆ octahedra, reducing the electrostatic interactions between the interlayer cations and anionic framework to facilitate exfoliation.

Flux crystal growth is a liquid-phase approach that uses molten metals or metal salts as the solvent to enable efficient crystal growth at or below the melting points of the solutes.²⁴ In this study, we examined the flux growth conditions for preparing multicomponent KNO single crystals, and found that the appropriate flux method with rapid quenching produced multicomponent niobate-based HEO single crystals. Crucially, we achieved the exfoliation of the layered multicomponent niobate crystals without employing harsh acids and/or bases, i.e. the conventional exfoliating agents. Subsequent hydrothermal treatment of the single crystals produced multicomponent niobate nanosheet colloids.

Experimental

Crystal growth and exfoliation

All reagents were purchased from Wako Pure Chemical Industries unless otherwise stated and were used without further purification. K₂CO₃ (>99.5%), Rb₂CO₃ (>97.0%), Nb₂O₅ (99.9%), TiO₂ (anatase, 98.5%), NH₄VO₃ (>99.0%), ZrO(NO₃)₂·2H₂O (>98.0%), SnO₂ (>98.0%), Sb₂O₅ (>95.0%), HfO₂ (98.0%), Ta₂O₅ (99.8%), WO₃ (99.5%), Bi(NO₃)₃·5H₂O (>99.5%), and CeO₂ (>99.5%) were used as solutes, and K₂MoO₄ (>99.0%) was employed as the flux. In a typical procedure for synthesising pristine KNO, K₂CO₃ and Nb₂O₅ were mixed in a stoichiometric molar ratio of 2 : 3 at a solute concentration of 10 mol% (defined as the nominal molar ratio of KNO in the flux). The mixture was then placed in a platinum crucible with a lid. The crucible was heated in an electric furnace to a set temperature of 1100 °C, heated at a rate of 300 °C·h⁻¹, and held at this temperature for 10 h in air. After holding, the crucible was subjected to either flux cooling or quenching. In the former, the crucible was cooled to room temperature at a rate of 200 °C h⁻¹, and the product was washed with deionised water at room temperature and dried at 100 °C. In the latter, the crystal samples were quenched by immersion in water after taking it out for a short while and then washed and dried in the same manner. The samples prepared with Nb sites substituted by Ti, Ta, Sb, or Hf (each replacing 5 at% of Nb) are denoted K₄Nb_4.8Ti_0.3Ta_0.3Sb_0.3Hf_0.3O₁₇ (TiTaSbHf_0.3) and TiTaSbHf_0.3-Q for flux and quenched samples, respectively. In a preliminary experiment, (K₄Nb₆O₁₇)_0.5(Rb₄Ta₆O₁₇)_0.5 and K₄(Nb₃,Ta₃)O₁₇ crystals were prepared via the solid-state reactions shown in eqn (1) and (2) without using any flux. These samples are denoted KNORTO and KNTO, respectively.

2K₂CO₃ + 1.5Nb₂O₅ + 1.5Ta₂O₅ → K₄(Nb₃Ta₃)O₁₇ + 2CO₂ (2)

Configurational entropy (S_config) is the entropy associated with the arrangement of particles or components in a system. For HEOs, S_config is defined by eqn (3), where x_i represents the mole fraction of cationic or anionic components, R is the gas constant, and N is the number of corresponding constituent elements.

The exfoliation of flux-grown KNO-based crystals was achieved by hydrothermal treatment with and without the use of exfoliation reagent. Under ambient conditions, KNO (50 mg) and H₂O (15 mL) were added to a Teflon-lined vessel and kept at 120 °C for 3 days. Experiments with holding times ranging from 1 to 7 d revealed that the standard deviation of the yield was within 10%, and a 3 d treatment was sufficient compared to the 7 d treatment. The solution was then transferred to a glass beaker and allowed to stand for 5 min, and the supernatant was recovered as a colloidal solution for further analysis. For comparison, niobate-based nanosheets were also prepared using propylamine (0.2 mL). The exfoliation efficiency (EE) is given by eqn (4), measured on a weight basis in solution (ppm).

EE (%) = (W_exfoliated/W_total) × 100 (4)

Here, W_exfoliated and W_total are the masses of exfoliated nanosheets and starting crystals, respectively, as determined by inductively coupled plasma–optical emission spectroscopy (ICP–OES).

Characterisation

X-ray diffractometry (XRD; SmartLab, Rigaku, Japan) measurements were carried out using monochromatic Cu-K_α radiation (λ = 0.15418 nm, 40 kV, 30 mA). To analyse the crystal structures of KNO and TiTaSb_0.5-Q, synchrotron powder XRD patterns were recorded at beamline BL5S2 (Aichi Synchrotron Radiation Centre) at a beam energy of 12 keV (λ = 0.7 Å). Field-emission scanning electron microscopy (FE-SEM) images and energy-dispersive X-ray spectroscopy (EDX) data were obtained using a JSM-7600F (JEOL, Japan) at an acceleration voltage of 2–15 kV with and without use of STEM Microgrid (NP-C15, Okenshoji Co., Ltd). When using the microgrid, a small aliquot (typically 5–10 µL) of the sample suspension was dropped onto the microgrid, allowed to adsorb for 1–2 min, and then the excess liquid was gently wicked off using filter paper. Bright-field transmission electron microscopy (TEM) images and selected area electron diffraction (SAED) patterns were obtained using an 80 kV electron microscope (JEM-2100F, JEOL) equipped with a double-aberration corrector (CESCOR/CETCOR, CEOS). The TEM samples were prepared by dispersing a drop of crystal suspension diluted in ethanol on a carbon-covered copper grid. Atomic force microscopy (AFM; Agilent Technologies, 5500 Scanning Probe Microscope) was used to obtain thickness profiles of the nanosheets. To prepare samples for AFM observation, a 20 μL aliquot of the supernatant solution obtained by 10-fold dilution of an approximately 50 ppm niobate-based colloid dispersion was dropped onto a Si substrate. After deposition, the substrate was heated on a hot stirrer at 80 °C to evaporate the solvent. Raman spectra were collected using a visible-light laser (λ = 532 nm, 50 mW) and a neutral-density filter to prevent degradation (HR-Revolution, Horiba Jobin Yvon). For elemental analysis, the KNO-based sample (0.010 g) was dissolved in HF, and the solution was analysed using ICP-OES (ICPE-9800, Shimadzu Corporation, Japan). Particle size distributions were measured using a laser diffraction/scattering particle size analyser (SALD-7100, Shimadzu). The instrument was operated with the standard wet-circulation unit. Before each run, the dispersion medium (deionized water) was circulated to establish a stable baseline, and the optical path was aligned according to the manufacturer's procedure. Nanosheet samples (approximately 10 ppm_Nb) were added dropwise while magnetically stirred. Transmission UV-vis-NIR spectra were recorded using a single-monochromator double-beam instrument with dual gratings/detectors (V-670, JASCO, Tokyo). For nanosheet samples, quartz cuvettes (optical path = 10 mm) were used. Zeta potential measurements were obtained by monitoring the streaming potential during controlled titration and continuous sample flow (STABINO Zeta, Microtrac MRB, Germany). Before measurement, the suspensions were sonicated for 5 min to ensure homogeneous dispersion.

Computational methods and calculations

Density functional theory (DFT) calculations were carried out in the Vienna Ab Initio Simulation Package (VASP)^25,26 with the generalized gradient approximation Perdew–Burke–Ernzerhof functional modified for solids (PBEsol-GGA)²⁷ and the projector-augmented wave (PAW) method.²⁸ An energy cut-off of 500 eV and a 3 × 1 × 4 k-point mesh for the bulk were used for the unit cell of 108 atoms in a tetragonal lattice of K₁₆Nb₂₄O₆₈ (Z = 4) for K₄Nb₆O₁₇ (KNO) and K₁₆Nb₂₁TiTaSbO₆₈ for multicomponent KNO. To ensure charge neutrality in the system, the number of electrons (NELECT) was adjusted to +1 because one Ti species, which has a valence one lower than Nb is introduced into the unit cell. To investigate the thermodynamically stable structures, a random combination of 21 Nb species, one Ti, one Ta, and one Sb within the unit cell was generated using the Monte-Carlo method from 2024 possible configurations. From these, 100 models were selected, and their total energy was calculated and extracted. In this energy calculation, the effects of spin are not considered because multicomponent KNO is d⁰ type. The total energy of these models was calculated, and the 10 most stable (stable structure models) and 10 least stable models (unstable structure models) were examined for their total energies, structures, and interatomic distances. Relaxation of the crystal structure was allowed in all calculations, and the energies of the final optimised geometries were recalculated to correct for changes of the plane-wave basis during relaxation.

The distortion index, D, was calculated using the formula proposed by Baur (eqn (5)).²⁹

Here, d_i represents the M–O distance, and represents the average value of M–O bond lengths within the octahedron, and n is usually 6 for MO₆ octahedra.

The secondary elongation, 〈λ_oct〉, is a descriptor that measures the displacement (l_i) of the central atom towards the n vertices of a polyhedron, given a polyhedron of the same volume (length l₀).³⁰ This can be calculated using eqn (6).

The third distortion index is the bond angle variance, σ_oct², of the O–M–O bond within the octahedron, calculated using eqn (7).³⁰

In this case, we used the change in each internal angle (φ) of the octahedra within the crystal structure obtained through structural relaxation and compared it with that of a regular polyhedron of the same number of edges (m), i.e. a regular octahedron.

Results and discussion

Synthesis of multicomponent KNO

First, we prepared a KNO-Rb₄Ta₆O₁₇ (RTO) solid solution through solid-state reaction because RTO has the same crystal structure as KNO.²¹ Fig. S1 in the SI shows the XRD pattern of the as-synthesised KNO_0.5RTO_0.5 sample. KNO-related phases were not formed, but impurity phases such as Rb₄Ta₁₀O₂₇ and Rb_4.9Ta_11.02O₃₀, as well as the starting material (Ta₂O₅), were present. Further, K₄(Nb,Ta)₆O₁₇ could not be synthesised because of phase separation (Fig. S2). Next, we made substitutions at the Nb site of KNO to prepare K₄Nb₆O₁₇:M_I,M_II,M_III,M_IV single crystals (M_I–M_IV = Ti, Ta, Sb, Hf, etc.). Details of the synthetic conditions are summarised in Table S1, runs 2–14. For example, the nominal chemical composition of 5 at%-substituted KNO:Ti,Ta,Sb,Hf is K₄Nb_4.8Ti_0.3Ta_0.3Sb_0.3Hf_0.3O₁₇, abbreviated TiTaSbHf_0.3 (S_config ≈ 0.8R). Fig. 1a shows the XRD patterns of various 5 at%-substituted KNO:Ti,Ta,M_III,M_IV samples (M_III, M_IV = V, Zr, Sn, Bi, Ce, and Hf) and TiTaSb_0.3. The impurity phase KNbO₃ was observed at approximately 22° in the XRD patterns of TiTaSbSn_0.3, TiTaSbW_0.3, TiTaSbBi_0.3, and TiTaSbCe_0.3. The peaks corresponding to impurity phases in the XRD patterns of TiTaSbBi_0.3, and TiTaSbCe_0.3 were assigned to potassium niobates, Ce₂O₃, and Bi₂O₃. In contrast, single phases of K₄Nb₆O₁₇·3H₂O and K₄Nb₆O₁₇ were observed for TiTaSbHf_0.3 and TiTaSb_0.3. K₄Nb₆O₁₇·3H₂O contains intercalated water molecules between the layers, and its d₀₄₀ spacing is approximately 1.2 Å larger than that of K₄Nb₆O₁₇. Potassium molybdate, used as the flux, is not incorporated into the crystal framework as a cation because it is a bulky oxoanion during crystal growth and flux removal processes.³¹

Fig. 1 XRD patterns of pristine and multi-component KNO single crystals grown from a K₂MoO₄ flux at 1100 °C: (a) KNO, K₄Nb_4.8Ti_0.3Ta_0.3M_III0.3M_IV0.3, and K₄Nb_5.1Ti_0.3Ta_0.3M_III0.3 (M_III, M_IV = V, Zr, Sb, Sn, W, Bi, Ce, Hf), and (b) K₄Nb_4.0/3.6Ti_0.5/0.6Ta_0.5/0.6M_III0.5/0.6M_IV0.5/0.6 and K₄Nb_4.5Ti_0.5Ta_0.5M_III0.5. The suffix ‘-Q’ indicates the quenching method. Each 7.5 at%-substituted KNO:Ti,Ta,M_III sample is denoted TiTaM_III0.5. For reference, the patterns of K₄Nb₆O₁₇·3H₂O (PDF 00-021-1297), KNbO₃ (PDF 00-008-0212), Ce₂O₃ (PDF 01-074-1145), Bi₂O₃ (PDF 01-080-9185), K_2.6Nb_11.6O₃₀ (PDF 01-073-7471), and K₄Nb₆O₁₇ (PDF 00-031-1063) are shown (bottom). Synthetic conditions summarised in Table S1.

We subsequently examined the effects of different substitution levels and quenching on the growth of single-crystal KNO:Ti,Ta,Sb,Hf. Fig. 1b shows the XRD patterns of the 7.5- and 10-at%-substituted KNO-based crystals prepared using either flux cooling or quenching. The KNbO₃ impurity phase was observed in the XRD patterns of TiTaSbHf_0.5 and TiTaSbHf_0.6, as indicated by the diffraction peak located at approximately 22°. In contrast, a single KNO-related phase was observed in the pattern of TiTaSbHf_0.5-Q (where ‘-Q’ indicates the quenching method), showing that quenching suppresses the formation of KNbO₃ phase. However, STEM-EDX analysis (Fig. S3, SI) revealed an uneven distribution of Hf species in TiTaSbHf_0.5-Q, indicating the insolubility of this species in the NbO₆-based frameworks of TiTaSbHf_0.5-Q. Hereafter, we focus on TiTaSb_0.5-Q (S_config ≈ 0.8R), because higher levels of elemental substitution resulted in impurities (e.g. TiTaSb_0.6-Q in Fig. S4).

Characteristics of single crystal samples

Fig. 2a–c show the STEM, EDX mapping, and optical and SEM images of TiTaSb_0.5-Q. This sample has sharp edges in the STEM image, and the K, Nb, Ti, Ta, and Sb species are evenly distributed throughout the crystal (Fig. 2a). The optical image reveals single crystals with 100–500 µm in size (Fig. 2b), and most particles are larger than 20 µm in length (Fig. 2c). This is the first example of HEO single crystals based on d-block elements having S_config = 0.80R. Fig. S5 shows the FE-SEM image and corresponding EDX point analysis result for TiTaSb_0.5-Q. As shown, signals corresponding to the added Sb, Nb, and Ta can be observed. Fig. S6 shows the high-resolution TEM image of a KNO single crystal. SAED measurements along the 〈010〉 direction revealed that KNO is a single crystal. Additionally, the high-resolution TEM images were well reproduced by simulation, indicating that KNO grows along the ac plane.

Fig. 2 Microscopy and XRD results: (a) STEM image and EDX mapping, (b) optical microscopy image, and (c) SEM image of TiTaSb_0.5-Q prepared by quenching. SXRD patterns of (d) K₄Nb₆O₁₇ and (e) TiTaSb_0.5-Q prepared by quenching (R_wp = 6.25%, R_p = 4.82%, S = R_wp/R_e = 0.789 for K_3.8H_0.2Nb₆O₁₇ and R_wp = 9.35%, R_p = 6.65%, S = R_wp/R_e = 1.07 for K_3.3H_0.7Nb_4.5Ti_0.4Ta_0.5Sb_0.5). Crystallographic parameters summarised in Table S2.

The synchrotron radiation XRD (SXRD) patterns of KNO and TiTaSb_0.5-Q samples are depicted in the bottom panels of Fig. 2d and e, and their crystal structures were analysed by Rietveld refinement. Fig. S7 shows the structural model of TiTaSb_0.5-Q used for the Rietveld analysis. To maintain symmetry during the analysis, Ti, Ta, and Sb were assumed to substitute Nb sites. Additionally, a slight oxygen deficiency was introduced for charge balance (Table S2). The reported K₄Nb₆O₁₇ structure was used as the initial model, having lattice parameters a = 0.7830 nm, b = 3.3310 nm, and c = 0.6460 nm and crystallising in the orthorhombic space group P2₁nb.²¹ The results are summarised in Table S2. First, we explain the fitting results for KNO (Fig. 2d). The refinement smoothly converged to give the following lattice parameters: a = 0.78159(3) nm, b = 3.2962(2) nm, and c = 0.64734(3) nm, having reliability factors of R_wp = 6.25%, R_p = 4.82%, and S = 0.789 for SXRD. These lattice parameters are in good agreement with those reported for K₄Nb₆O₁₇. The lattice parameters of TiTaSb_0.5-Q were determined using the same method (Fig. 2e): a = 0.78222(4) nm, b = 3.2917(2) nm, and c = 0.64633(5) nm, having reliability factors of R_wp = 9.35%, R_p = 6.64%, and S = 1.07. The values of a and c in the in-plane direction only changed slightly after substitution (approximately 0.1%). The change in intensity ratios in the diffraction peaks can be attributed to the substitution of Nb by other atoms with different atomic scattering factors.

X-ray photoelectron spectroscopy (XPS) data of KNO and TiTaSb_0.5-Q are shown in Fig. S8. The peaks attributed to Mo 3p_3/2 and 3p_1/2 appear around 393 and 410 eV; however, neither KNO nor TiTaSb_0.5-Q produced signals in this region, indicating that Mo species from the flux were not incorporated in the KNO frameworks. Additionally, TiTaSb_0.5-Q shows a signal centred at 458.7 eV, characteristic of a +4 (TiO₂) oxidation state (Fig. S8b).³² For TiTaSb_0.5-Q, the Ta 4d_3/2, Ta 4d_5/2, and Sb 3d_3/2 peaks appeared at binding energies of 242.0, 230.8, and 540.0 eV (Fig. S8c and S8d), consistent with literature values.^33,34 Therefore, the elements in TiTaSb_0.5-Q are in the +4, +5, and +5 oxidation states, respectively.

Fig. 3 shows the Raman spectra of KNO and TiTaSb_0.5-Q. KNO displays the typical spectral pattern of orthorhombic K₄Nb₆O₁₇: the bands at 840 and 879 cm⁻¹ and 580 and 645 cm⁻¹ correspond to the stretching vibrations of short Nb=O bonds and those of NbO₆, both in the A₁ + A₂ + B₁ + B₂ mode.³⁵ The spectral intensity of TiTaSb_0.5-Q was slightly lower, but signals originating from the niobate framework were observed. In addition, two spectral features distinct from that of pristine KNO were observed. First, the two peaks corresponding to short Nb=O bond stretching vibrations were shifted to higher wavenumbers by 3–5 cm⁻¹. In layered materials, this behaviour can be attributed to weaker interactions in the direction perpendicular to the layers.^35,36 The second distinct feature is that the band around 570 cm⁻¹ was much weaker in the spectrum of TiTaSb_0.5-Q compared to that in KNO, possibly because of the substitution of Nb by other species. Thus, mixing heteroatoms into the NbO₆ octahedra significantly affected the local structure of the KNO system.

Fig. 3 Raman spectra of (a) pristine K₄Nb₆O₁₇ and (b) TiTaSb_0.5-Q. The bands at 840 and 879 cm⁻¹ correspond to stretching vibrations of short Nb=O bonds, and those at 580 and 645 cm⁻¹ correspond to stretching vibrations of NbO₆ octahedra in the A₁ + A₂ + B₁ + B₂ mode.

ICP-OES measurements revealed the compositions of KNO and TiTaSb_0.5-Q to be K_3.8H_0.2Nb₆O₁₇ and K_3.3H_0.7Nb_4.6Ti_0.4Ta_0.5Sb_0.5O_16.8, respectively. We assumed that protons were introduced into the interlayers during washing to remove the K₂MoO₄ flux to compensate for the charge of the anionic framework. These two compositions match the target nominal compositions approximately.

Exfoliation efficiency

Next, we studied the exfoliation of KNO and TiTaSb_0.5-Q under hydrothermal conditions with and without propylamine as an exfoliation reagent. Even without using propylamine, the resulting colloid exhibited the Tyndall effect (Fig. 4a), indicating that the exfoliation of niobate had occurred. Fig. 4b shows the FE-SEM image of a TiTaSb_0.5-Q nanosheet prepared with and without propylamine (Fig. 4b and c). The use of propylamine resulted in the formation of nanosheets approximately 10 µm in size with well-defined edges (Fig. 4b). In contrast, using our proposed reagent-free method, the niobate nanosheets were forcibly exfoliated (Fig. 4c), and because the surfaces were not stabilized or because the nanosheets reacted with water during the hydrothermal process, rounded nanosheet edges were frequently observed.

Fig. 4 Tyndall effect and morphologies: (a) Photographs of TiTaSb_0.5-Q nanosheet suspensions. FE-SEM image of TiTaSb_0.5-Q nanosheet prepared (b) with and (c) without use of exfoliation reagent.

Fig. S9 shows the particle size distribution of the TiTaSb_0.5-Q nanosheet colloid measured by dynamic light scattering (DLS). The results show an almost monomodal distribution centred at approximately 20 μm. According to Coleman et al.,³⁷ the hydrodynamic radius, a, is correlated with the length (diameter) of the nanosheet when approximating the nanosheets as disks. Fig. S10 shows the UV-Vis spectrum of TiTaSb_0.5-Q nanosheets, which was used to evaluate the electronic structure. Potassium hexaniobate crystals do not absorb in the visible-light region, showing absorption only below 380 nm (UV).³⁸ The high-entropy modification did not cause significant changes in UV spectrum, indicating that the electronic structure remained unchanged.

Fig. 5 show an AFM image and surface profile of a TiTaSb_0.5-Q nanosheet obtained without propylamine. The nanosheet has a uniform thickness of approximately 2.3 nm, almost identical to that reported for KNO,³⁹ suggesting that the TiTaSb_0.5 nanosheets consist of two or three layers of niobate-based building blocks. Note that instead of bilayer exfoliation, nanosheets with 4–10 nm in thickness were also observed (for examples, see the AFM images in Fig. S11). This is possibly because, under hydrothermal conditions, exfoliation is forced because strong acidic or basic exfoliating agents are not used.

Fig. 5 (a) AFM image of TiTaSb_0.5-Q nanosheet prepared without use of exfoliation reagent and (b) the corresponding thickness profile.

Furthermore, to evaluate how multi-element substitution affects the EE measured on a weight basis in solution (parts-per-million of Nb), we determined the concentrations of the KNO, TiTaSb_0.3, and TiTaSb_0.5-Q nanosheet suspensions obtained after hydrothermal treatment (Fig. 6a). The concentration of KNO nanosheets was 4.0 ppm_Nb, whereas those of TiTaSb_0.3 and TiTaSb_0.5-Q were 18.7 and 25.0 ppm_Nb, respectively. This sharp contrast indicates that a multicomponent niobium oxide framework greatly enhances the EE. Analysis of the exchangeability of the interlayer ions is shown in Fig. 6b. As shown, when KNO and TiTaSb_0.5-Q were placed in pure water at 25 °C, the solution pH increased for both samples but at significantly different rates: the pH of TiTaSb_0.5-Q suspension increased from 7.9 to 11.5, whereas that of the KNO suspension only reached 10.0. This increase in pH could be due to ion-exchange between protons from water and intercalated K⁺. The different rates of pH increase suggest that incorporating multiple elements significantly reduced the electrostatic interaction between ions in the interlayer and the host framework.

Fig. 6 Concentration and pH analysis. (a) Nb concentrations of K₄Nb₆O₁₇ (KNO), TiTaSb_0.3, and TiTaSb_0.5-Q nanosheet suspensions. (b) pH change during the immersion of KNO and TiTaSb_0.5-Q in water.

In addition, the surface charges were measured. Fig. S12 shows the ζ-potential profiles of KNO and TiTaSb_0.5-Q. The surfaces of both layered crystals were negatively charged, approximately −150 mV. Thus, high-entropy modification did not significantly affect the surface charge state, although the high-entropy modified sample was slightly more negatively charged than the parent sample.

Theoretical modelling

To understand the change in intercalation properties arising from substitution, stable and unstable crystal structures of TiTaSb_0.5-Q were constructed and analysed using DFT calculations. The stable pristine KNO was also analysed for comparison. Specifically, a multicomponent KNO random (MKNO-rdm) model was created by randomly placing three elements (Ti, Ta, and Sb) at each of the 24 Nb sites in the unit cell, and 100 models were generated for TiTaSb_0.5-Q. This corresponds to substituting 4.2 at% each of Ti, Ta, and Sb for the Nb species. After structural relaxation, the total energies of these 100 models were compiled, and a total of 2024 possible arrangements were used as models (Fig. S13). Generally, phase transitions at room temperature tend to occur within approximately 25 meV per atom from the most stable total energy. Structural models within this range (−847.31024 eV per atom) are designated MKNO-rdm-a1-10 (stable structures). In contrast, the 10 models with the highest total energy are designated MKNO-rdm-b91-100 (unstable structures).

Souza et al. reported⁴⁰ that the six types of NbO_x octahedra in KNO are distorted in increasing order Nb-3 ≈ Nb-4 < Nb-1 ≈ Nb-2 < Nb-5 < Nb-6, as summarised in Fig. S14. Fig. 7a shows the frequency histogram of the introduced Ti, Ta, and Sb atoms in the Nb site of MKNO-rdm-a1-10 and MKNO-rdm-b91-100 models. In the stable structures (MKNO-rdm-a1-10), Ti, Ta, and Sb occupied the less distorted Nb-3 and Nb-4 octahedra (Fig. 7a). In contrast, in the unstable structure models (MKNO-rdm-b91-100), Sb, which is not expected to exhibit octahedral distortion owing to the Jahn–Teller effect, occupied the most distorted Nb-6 site. This placement minimises additional structural distortions that would occur if a more distortion-prone ion occupied Nb-6. The Ti and Ta species occupies the Nb sites in several ways. A comparison of the histograms indicates that the introduced Ti, Ta, and Sb achieved stable states by occupying the less distorted Nb-3 and Nb-4 sites.

Fig. 7 (a) Frequency histograms of Ti, Ta, and Sb at Nb sites in the MKNO-rdm-a1-10 and MKNO-rdm-b91-100 models. (b) Schematic of four types of interlayers in MKNO-rdm-a1 and interlayer interactions. (c) Local structure of ion-exchange sites in interlayer-type-IV MKNO-rdm-a1. (d) Comparison of interlayer interactions in MKNO-rdm-a1 and K₄Nb₆O₁₇ (KNO). Arrows indicate which Nb-oxide framework the K⁺ ion is attracted to compared to KNO. Arrow length represents the magnitude of the attraction to one of the layers based on the distance between the K ion and the oxygen ions. Purple spheres: K, green spheres: Nb, blue spheres: Ti, brown spheres: Ta, yellow spheres: Sb, red spheres: O. Crystal structures in (b)–(d) is drawn using VESTA.⁴¹

Next, we focused on the local structure of ion-exchangeable K⁺ on the niobates, to understand the different ion-intercalation properties between KNO and TiTaSb_0.5-Q. In practice, there are several occupancy combinations for the introduced Ti, Ta, and Sb. However, here, we used the models of MKNO-rdm-a1 and KNO based on the DFT calculations because MKNO-rdm-a1 showed similar characteristics to TiTaSb_0.5-Q, and it is currently impossible to determine the specific locations of the introduced Ti, Ta, and Sb species experimentally. Note that the lattice parameters of KNO and TiTaSb_0.5-Q derived from DFT calculations match those from the SXRD analysis within a margin of approximately 0.5% (Table S3), and the local structure of MKNO-rdm-a1 is not significantly different from that of MKNO-rdm-a2-10, vide infra. Here, we use MKNO-rdm-a1 as an example, but similar multi-element substitution effects are expected in MKNO-rdm-a2-10 as well because the substitution trend does not change markedly in Fig. 7a. Considering the symmetry of MKNO-rdm-a1 (space group P1), the interlayers are named by their proximity to the origin as interlayers I, II, III, and IV; Fig. 7b shows the corresponding crystal structure of MKNO-rdm-a1 as a representative and these four types of interlayers. Whereas the conventional K₄Nb₆O₁₇·3H₂O possesses two distinct interlayer structures (type I and II), the introduction of high entropy effect generates four or more structurally different interlayers (Fig. 7b). We quantified the distortion level of MO₆ octahedra (M = Nb, Ti, Ta, and Sb) in both KNO and MKNO-rdm-a1-10 using three parameters: the distortion index (D, i.e. Baur's distortion index),⁴² quadratic elongation (〈λ_oct〉),³⁰ and change in the O–M–O bond angle within MO₆ (σ_oct²).³⁰ The average Nb–O bond length and D values for MO₆ of KNO and MKNO-rdm-a1-10 are summarised in Table S4. For KNO, the D value varies between the six Nb sites, and the order matches the previously reported order⁴⁰ of Nb-3 ≈ Nb-4 < Nb-1 ≈ Nb-2 < Nb-5 < Nb-6. In contrast, in MKNO-rdm-a1-10 the Nb–O bond length is shorter in over 90% of the cases, as highlighted in blue in Table S4. This could be because Ti⁴⁺ and Sb⁵⁺ have smaller ionic radii than Nb⁵⁺ (Ti⁴⁺: 0.605 Å, Sb⁵⁺: 0.60 Å, Nb⁵⁺: 0.64 Å).⁴³ In all 10 models of KNO-rdm-a1-10, over half of the 24 NbO₆ octahedra in each MKNO showed higher D values, indicating increased distortion. This increase can be attributed to the larger dispersion of M-O bond lengths caused by the substitution of Nb with Ti, Ta, and Sb. The 〈λ_oct〉 values for NbO₆ showed an increase in over half of the Nb sites in 8 out of 10 models, but the relative change was only 0.05% at most. The σ_oct² value for NbO₆ decreased in more than half of the cases but increased in at least 30% of the octahedra, indicating that the introduction of multiple elements affects the degree of distortion of NbO₆. Regarding the TiO₆ octahedra in MKNO models, all three distortion indices were significantly increased (11% to 64% for D, 0.4% to 1.0% for 〈λ_oct〉, and 8.8% to 37.0% for σ_oct²). However, all distortion indices for SbO₆ and over 90% of distortion indices for TaO₆ decreased.

We measured the distance between each K⁺ in the interlayers and its neighbouring oxygen atom for KNO and MKNO-rdm-a1. As an example, Fig. 7c illustrates the K–O bonds in interlayer type-IV of MKNO-rdm-a1, and Table 1 compares the bond lengths between KNO and MKNO-rdm-a1. The K–O bond lengths of 1 and 2 in Fig. 7c for MKNO-rdm-a1 are shorter than those of KNO whereas those of 5 and 6 are longer, indicating that the K⁺ ions are displaced towards the oxide layer with heteroatom substitution and away from the non-substituted oxide layer, as shown in Fig. 7d. Such substitution effects were similarly confirmed in the MKNO-rdm-a2-10 models.

Table 1 K–O bond lengths for K⁺ in interlayer type IV of MKNO-rdm-a1 and KNO

Sample	K–O bond length/Å
	1	2	3	4	5	6
MKNO-rdm-a1	2.80	2.78	2.92	2.93	3.04	3.06
KNO	2.98	2.98	2.94	2.94	2.86	2.87

---

Finally, we discuss why multicomponent KNO-based niobate single crystals can be easily exfoliated without acids or bases. From the results of SXRD structural analysis and DFT calculations, multicomponent and pristine KNO have very similar structural parameters but different local structures. For instance, the oxide framework layers of MKNO are distorted: the interatomic distance between the interlayer ions and multicomponent oxide layers (nearby oxygen ions) is shortened, whereas that with the opposite non-substituted oxide layer is lengthened (Fig. 7d). This local structural change could affect the electrostatic interactions between K⁺ ions and the framework layers, as well as the intercalation properties. In Fig. 7d, the length of the arrows represents the strength of electrostatic interaction by which a single K⁺ ion is attracted to one of the negatively charged frameworks. Because of element substitution, the ion is attracted to the negatively charged Nb-oxide framework (MKNO-rdm-a1 vs. KNO). The type-II interlayers of K₄Nb₆O₁₇ cannot intercalate water molecules, and ions in this interlayer undergo little exchange with other metal cations (Fig. S14).²² In contrast, the multicomponent oxide frameworks have different intercalation properties, that is, the type-II interlayer in K₄Nb₆O₁₇ arising from the distortion of the NbO₆ octahedron. In other words, the type-I-to-IV interlayers of MKNO can easily intercalate water molecules because of the distorted interlayer structure, facilitating swelling and subsequent exfoliation through proton replacement under hydrothermal conditions. Our immersion tests in water revealed that the K⁺ ions in TiTaSb_0.5-Q are more easily exchanged with protons than those in KNO, as shown in Fig. 6b. TiTaSb_0.5-Q with S_config = 0.80R has a higher proportion of Ti, Ta, and Sb than KNO-rdm-a1-10 and, thus, should show a higher degree of distortion in its niobate framework.

Conclusions

Hydrothermal exfoliation of layered single crystals is a promising method for mass producing 2D materials in a scalable manner. However, harsh reagents including hydrochloric acid, butyllithium, and tetrabutylammonium hydroxide are generally required for exfoliation because layered oxides including hexaniobates have relatively strong electrostatic interlayers interactions. Our proposed approach using framework distortion caused by the addition of multiple elements improved the exfoliation of the layered oxide crystals without using harsh acids or bases. Further, structural characterisation and theoretical analysis revealed that the substitution had little impact on the lattice parameters of KNO but a significant effect on the local structure, particularly that around the interlayer K⁺ ions. These findings provide a new concept for the exfoliation of layered materials to prepare 2D inks.

Author contributions

Conceptualisation, F. H.; materials synthesis, F. H. and U. K.; characterisation and analysis of data, F. H., K. U., T. S., and K. F.; theory and calculations, U. K. and H. S.; writing original draft, F. H.; writing, reviewing, and editing, all the authors; funding acquisition, F. H., T. S., and K. T.; supervision, F. H. and K. T.

Conflicts of interest

The authors declare no conflict of interests.

Data availability

The data supporting this article have been included as part of the supplementary information (SI). The data that support the findings of this study are available from the corresponding authors upon reasonable request.

Supplementary information is available. See DOI: https://doi.org/10.1039/d5nr01194a.

Acknowledgements

This work was supported by the JSPS KAKENHI (grant number 24K01234); the Council for Science, Technology and Innovation (CSTI), Cross-ministerial Strategic Innovation Promotion Program (SIP) in the 3rd period of SIP ‘Creating a materials innovation ecosystem for industrialization’ grant number JPJ012307 (Funding agency: NIMS); Research Grant from the TAKEUCHI Zaidan (grant number takeuchi2023-J-20); Research Grant from the Kurita Water and environment Foundation (grant number 23A030); and GIMRT Program of the Institute for Materials Research, Tohoku University (Proposal No. 202112-CRKKE-0038, 202103-CRKKE-0045). The SPXRD experiments were conducted at the BL5S2 beamline of the Aichi Synchrotron Radiation Center, Aichi Science & Technology Foundation, Aichi, Japan (Proposal No. 2020D6009).

References

1. C. Tan, X. Cao, X. J. Wu, Q. He, J. Yang, X. Zhang, J. Chen, W. Zhao, S. Han, G. H. Nam, M. Sindoro and H. Zhang, Chem. Rev., 2017, 117, 6225–6331.
2. K. Zhou, G. Shang, H. Hsu, S. Han, V. A. L. Roy and Y. Zhou, Adv. Mater., 2023, 35, e2207774.
3. R. Ma and T. Sasaki, Acc. Chem. Res., 2015, 48, 136–143.
4. C. N. R. Rao, H. S. S. R. Matte and U. Maitra, Angew. Chem., Int. Ed., 2013, 52, 13162–13185.
5. S. Nishioka, K. Hojo, L. Xiao, T. Gao, Y. Miseki, S. Yasuda, T. Yokoi, K. Sayama, T. E. Mallouk and K. Maeda, Sci. Adv., 2022, 8, eadc9115.
6. N. Miyamoto, M. Miyoshi, R. Kato, Y. Nakashima, H. Iwano and T. Kato, Sci. Adv., 2024, 10, eadk6452.
7. K. Sano, Y. Ishida and T. Aida, Angew. Chem., Int. Ed., 2018, 57, 2532–2543.
8. N. Miyamoto and T. Nakato, Adv. Mater., 2002, 14, 1267–1270.
9. S. Pinilla, J. Coelho, K. Li, J. Liu and V. Nicolosi, Nat. Rev. Mater., 2022, 7, 717–735.
10. R. Yang, Y. Fan, L. Mei, H. S. Shin, D. Voiry, Q. Lu, J. Li and Z. Zeng, Nat. Synth., 2023, 2, 101–118.
11. A. Sarkar, Q. Wang, A. Schiele, M. R. Chellali, S. S. Bhattacharya, D. Wang, T. Brezesinski, H. Hahn, L. Velasco and B. Breitung, Adv. Mater., 2019, 31, 1806236.
12. C. M. Rost, E. Sachet, T. Borman, A. Moballegh, E. C. Dickey, D. Hou, J. L. Jones, S. Curtarolo and J. P. Maria, Nat. Commun., 2015, 6, 8485.
13. Z. Lun, B. Ouyang, D. H. Kwon, Y. Ha, E. E. Foley, T. Y. Huang, Z. Cai, H. Kim, M. Balasubramanian, Y. Sun, J. Huang, Y. Tian, H. Kim, B. D. McCloskey, W. Yang, R. J. Clément, H. Ji and G. Ceder, Nat. Mater., 2021, 20, 214–221.
14. P. Edalati, Q. Wang, H. Razavi-Khosroshahi, M. Fuji, T. Ishihara and K. Edalati, J. Mater. Chem. A, 2020, 8, 3814–3821.
15. T. Maiti, R. Banerjee, S. Chatterjee, M. Ranjan, T. Bhattacharya, S. Mukherjee, S. S. Jana and A. Dwivedi, ACS Sustainable Chem. Eng., 2020, 8, 17022–17032.
16. J. Gild, M. Samiee, J. L. Braun, T. Harrington, H. Vega, P. E. Hopkins, K. Vecchio and J. Luo, J. Eur. Ceram. Soc., 2018, 38, 3578–3584.
17. K. P. Tseng, Q. Yang, S. J. McCormack and W. M. Kriven, J. Am. Ceram. Soc., 2020, 103, 569–576.
18. C. Kinsler-Fedon, Q. Zheng, Q. Huang, E. S. Choi, J. Yan, H. Zhou, D. Mandrus and V. Keppens, Phys. Rev. Mater., 2020, 4, 104411.
19. M. Pianassola, M. Loveday, B. C. Chakoumakos, M. Koschan, C. L. Melcher and M. Zhuravleva, Cryst. Growth Des., 2020, 20, 6769–6776.
20. M. Pianassola, M. Loveday, R. Lalk, K. Pestovich, C. L. Melcher and M. Zhuravleva, J. Am. Ceram. Soc., 2023, 106, 7123–7132.
21. M. Gasperin and M. T. Le Bihan, J. Solid State Chem., 1982, 43, 346–353.
22. N. Kinomura, N. Kumada and F. Muto, J. Chem. Soc., Dalton Trans., 1985, 2349–2351.
23. K. Domen, A. Kudo, A. Shinozaki, A. Tanaka, K. I. Maruya and T. Onishi, J. Chem. Soc., Chem. Commun., 1986, 356–357.
24. K. Teshima, K. Horita, T. Suzuki, N. Ishizawa and S. Oishi, Chem. Mater., 2006, 18, 3693–3697.
25. G. Kresse and J. Furthmüller, Phys. Rev. B: Condens. Matter Mater. Phys., 1996, 54, 11169–11186.
26. G. Kresse and J. Furthmüller, Comput. Mater. Sci., 1996, 6, 15–50.
27. J. P. Perdew, A. Ruzsinszky, G. I. Csonka, O. A. Vydrov, G. E. Scuseria, L. A. Constantin, X. Zhou and K. Burke, Phys. Rev. Lett., 2009, 102, 039902.
28. P. E. Blöchl, Phys. Rev. B: Condens. Matter Mater. Phys., 1994, 50, 17953–17979.
29. W. H. Baur, Acta Crystallogr., Sect. B, 1974, 30, 1195–1215.
30. K. Robinson, G. V. Gibbs and P. H. Ribbe, Science, 1971, 172, 567–570.
31. Y. P. Jimenez, F. Justel and J. M. Casas, J. Mol. Liq., 2022, 346, 117096.
32. F. Montilla, E. Morallón, A. De Battisti, S. Barison, S. Daolio and J. L. Vázquez, J. Phys. Chem. B, 2004, 108, 15976–15981.
33. M. Hellwig, A. Milanov, D. Barreca, J. L. Deborde, R. Thomas, M. Winter, U. Kunze, R. A. Fischer and A. Devi, Chem. Mater., 2007, 19, 6077–6087.
34. A. M. Volosin, S. Sharma, C. Traverse, N. Newman and D. K. Seo, J. Mater. Chem., 2011, 21, 13232–13240.
35. M. Ma̧czka, M. Ptak, A. Majchrowski and J. Hanuza, J. Raman Spectrosc., 2011, 42, 209–213.
36. J. Oi, A. Kishimoto and T. Kudo, J. Solid State Chem., 1993, 103, 176–185.
37. M. Lotya, A. Rakovich, J. F. Donegan and J. N. Coleman, Nanotechnology, 2013, 24, 265703.
38. K. Mori, S. Ogawa, M. Martis and H. Yamashita, J. Phys. Chem. C, 2012, 116, 18873–18877.
39. N. Miyamoto, R. Kaito and K. Kuroda, Chem. Commun., 2002, 2, 2378–2379.
40. J. K. D. Souza, T. M. Duarte, I. M. G. dos Santos, J. R. Sambrano, A. D. S. Maia and A. dos Reis Albuquerque, Phys. Chem. Chem. Phys., 2020, 22, 16562–16570.
41. K. Momma and F. Izumi, J. Appl. Crystallogr., 2011, 44, 1272–1276.
42. W. H. Baur, Acta Crystallogr., Sect. B, 1974, 30, 1195–1215.
43. R. D. Shannon, Acta Crystallogr., Sect. A, 1976, 32, 751–767.

---