Two-Dimensional Oxides: Recent Progress in Nanosheets

3.1 Top-down synthesis: ion exchange and intercalation

In the 1990s first syntheses of oxide nanosheets were published, which used top-down techniques of ion intercalation for layered materials [25], [57], [58]. This method is still one of the most used synthesis routes for layered materials, although some modifications and improvements were established. Today, the common synthesis route for 2D oxidic nanosheets is a multistep top-down technique. The internal layers of the 3D material initially proceeds an ion exchange due to acid treatment followed by an ion intercalation caused by large organic molecules like the tetrabutylammonium cation (TBA⁺) [76], [100], [101]. Optionally, the exfoliation of the intercalated layers is subsequently assisted by sonication or shear forces. Adequate precursors for this method are layered materials with alkali or earth-alkali metals located inbetween the layers. The position of these metals can easily be exchanged by H⁺ cations trough immersing the layered materials in acid aqueous solution. A second cation exchange process occurs in basic aqueous solution with TBA⁺. Due to the large radius of the organic cation compared to H⁺, the distance of the interlayer increases (see Figure 2).

Fig. 2:

Schematic synthesis route for liquid exfoliation of layered oxides. In the first step, the alkali or earth-alkali cations (blue orbs) in the interlayers are partially or entirely exchanged caused by acid treatment (red orbs). Afterwards, the layers are separated by large organic cations like the TBA⁺ (green orbs), which function as an exfoliation agent, assisted by osmotic swelling or sonication. As a result, the free-standing nanosheets remain dispersed in the liquid medium. Inspired by Nicolosi et al. [10].

By applying further energy with sonication or shear forces, the exfoliation is completed with a positively charged surface on the oxidic nanosheets. Alternatively, further driving forces as the formation of insoluble solids during the exfoliation process can be used to improve the outcome [102]. Besides, the exchange of anions is applicable for layered double hydroxides (LDHs) in particular. Overall, the exchange of ions and subsequently intercalation provides high yields and upscaling possibilities [6]. Nevertheless, the chemical formula of the nanosheets deviates compared to the 3D bulk material due to occurred chemical reactions during the exfoliation.

3.2 Top-down synthesis: mechanical forces

The synthesis of graphene was initially performed by the use of Scotch tape, which provided a mechanical cleavage of the 3D graphite into 2D graphene [103], [104]. The method exploits the cleavage of relatively weak van der Waals bonds in layered bulk materials, but in-plane the stronger covalent bonds remain [105]. It provides a product with high crystallinity while the technique itself is easily and extensively applicable. Furthermore, the 2D material can be attached to many different surfaces after the cleavage. Although, several disadvantages make other techniques more viable. Since the Scotch tape has to be used several times until the nanosheets are obtained, the yield of the process is rather low. Furthermore, the process itself is quite slow and not applicable for upscaling. As the process is managed by hand, the controllability of thickness, size and shape of the nanosheets is lacking and repeatability is an issue. Another problem concerning layered oxides are the mainly covalent and ionic bonds, which require more appropriate techniques. Therefore, the Scotch tape lacks in applicability and is replaced by a wide range of various techniques.

Other possibilities of applying mechanical forces are sonication and shear forces. These methods require a liquid medium, which only disperses the layered material and does not dissolve it. The advantages of these techniques are high yields and the possibility of upscaling at rather low costs by utilizing a cheap solvent [106]. The idea of sonication is using bubbles and their implosions induced by acoustic cavitations to create tension between the layers [107]. Therefore, the surface energy of the layered material and the liquid medium have to match [106]. Furthermore, the medium should prevent the aggregation of nanosheets and their restacking. Thus, the selection and adjusting of the solvent is critical for a successful exfoliation. For using easily applicable and available solvents as water, the modulation of the surface tension is necessary by using ionic or nonionic polymers. As a downside, the attachment of residual polymer on the surface of the nanosheets is often an exclusion criterion for electrical applications. A further issue of sonication concerns the size of the nanosheets, because the mechanical force shatters sheets of large dimensions into smaller ones. The amount of defects in comparison to the Scotch tape is increasing as well due to the possible interaction with the liquid medium [6]. Hence, the controllability of the technique is limited to selecting the sonication time, solvent, additives, temperature and parameters of the sonication device. The utilization of shear forces contributes similar characteristics but with even bigger upscaling possibilities for industry standards. For example, Ti₅NbO₁₄ nanosheets were synthesized by Zhang et al. [108] with the help of a specialized milling machine. Hereby, additional parameters as rotor diameter and shear rate are determining factors [109].

3.3 Top-down synthesis: rapid heating

In recent years, a novel method with a potentially high production rate for oxidic nanosheets arised by Zhao et al. [47]. Simply by rapid heating of hydrous chloride compounds containing the desired metal species, freestanding nanosheets could be produced. The succeeded nanosheets consisted of Cr₂O₃, ZrO₂, Al₂O₃ or Y₂O₃, while the Cr₂O₃ nanosheets showed the most consistent results. The principle of this method is using the releasing gaseous molecules as water and hydrochloric acid at the heating process to exfoliate the sheets (see Figure 3).

Fig. 3:

Schematic principle of the exfoliation process realized by rapid heating of hydrous chlorides. The gaseous products are released in between the layers due to the rapid heating, leading to exfoliation and nanosheets consecutively. Reproduced with permission from [47]. Copyright 2016 Springer Nature. Creative Commons License CC BY 4.0.

A critical parameter is the heating rate, because rapid heating is necessary to effectively separate the layers and gain monolayers or few multilayer sheets. For example, direct heating in an alcohol lamp flame led to inhomogenous results, but placement in a preheated muffle furnace showed significant improvement. Since the non-layered or anhydrous materials did not lead to nanosheets, the technique seems limited to hydrous chlorides. Though, the simplicity and swiftness of the synthesis route is promising for producing nanosheets in industry orders of magnitude.

3.4 Top-down synthesis: oxidation

The aforementioned methods all require layered oxide materials, but gaining oxide products out of non-oxidic materials is another route. Therefore, the synthesis of non-oxide nanosheets and the oxidation afterwards is possible as shown by Bai et al. [94]. In this example, Rh nanosheets were oxidized chemically by HClO to Rh₂O₃ nanosheets. Other works showed the oxidation of large surfaces to exploit the properties of 2D sheets as shown for CuO/Cu₂O on a Cu surface [79]. Here, the nanosheet array was formed electrochemically via anodization. When using non-oxide nanosheets this method is only superior to other methods, if it is used for oxide compounds, which else can not be obtained. Otherwise it simply includes another oxidation step, which is already included or not necessary in the other methods. The modification of non-oxide surfaces on the other hand is well controllable if realized electrochemically by adjusting the anodization time, current density, temperature and viscosity of the electrolyte. Simple oxidation as in passivation does not enlarge the surface area and is not easily adjustable.

3.5 Bottom-up synthesis: chemical vapor deposition

A highly valuable synthesis for the bottom-up route is the chemical vapor deposition (CVD). Typically, one precursor is heated in a furnace and led to a reaction chamber via a gaseous flux. The gaseous flux may consist of transport gases and/or further gaseous precursors. Within the chamber, the gaseous precursors react on the surface of a specific substrate and form nanosheets or ultrathin films at proper experimental parameters [110]. In some cases, catalysts are necessary for enabling the desired reaction. The process itself provides a high level of control due to parameters as the choice of precursor, substrates and catalysts, temperature, atmosphere and the gaseous flow rate. Under appropriate experimental conditions, the amount of impurities and defects within the sheets or ultrathin films are negligible [6]. As a downside of this powerful technique, usually high temperatures and cost-intensive inert atmospheres are required. Furthermore, the nanosheets are developed on a specific substrate, which requires further transferring steps for applications or substrates with direct applicability.

3.6 Bottom-up synthesis: solvothermal method

The solvothermal synthesis is a typical bottom-up synthesis in a liquid medium. If water is used as a medium, the synthesis is called hydrothermal. Characteristic for these solvent-driven syntheses is the usage of closed autoclaves for reaching a higher reaction temperature than the boiling point of the selected solvent. The high temperature leads to an increasing pressure inside the vessel and promotes the formation of nanocrystals within the reaction chamber. For the formation of nanosheets commonly additives as Pluronic P123 or ethylene glycol are used [111], which function as a surfactant due to hydrophobic forces. The method enables a high yield at usually low costs, but the optimum experimental parameters for the desired size and thickness are difficult to determine and have to be adjusted to the used materials [6]. Critical parameters are the concentration of the precursors, the used solvent, surfactants, temperature and the oxygen partial pressure. Even the used oven and autoclave may have an impact on the gained product and may influence the experimental results. These circumstances are inconvenient for a production on industrial scale.

3.7 Bottom-up synthesis: templated method

A relatively new technique for synthesizing anisotropic nanosheets is the use of 2D templates. For example, graphene oxide (GO) can be used for nanosheets consisting of TiO₂, ZrO₂, Nb₂O₅, SnO₂ or Ta₂O₅ [52]. Therefore, dried GO was dispersed in cyclohexane together with the respective metal alkoxide, which resulted in adsorbed metal alkoxides on the surface of GO. In an autoclave, the GO was subsequently reduced and metal oxide nanofilms were formed. GO is especially suitable in this case, since the functional groups of GO hydrolyze the organic ligands of the adsorbed metal alkoxides. Free-standing nanosheets were finally obtained by calcining at 723 K in air [52]. Besides, multiple approaches using templates for the formation of nanosheets exist. Another route presented by Gao et al. [15] utilized a successive ion-layer adsorption and reaction (SILAR) technique, which was interrupted half-way to gain the nanosheets. The route also uses GO as a 2D template, where metal ions are adsorbed on the surface. The difference consists in the further treament, which starts with a stacking of the sheets via centrifugation. Afterwards, the sheets are disperged again, quickly freezed and freeze-dried to maintain the structure under mild conditions and not risk fissures or stacking of the sheets caused by drying. At last, the calcination removes the template and nanosheets of the metal oxide remain. Since the removal of 2D templates often requires high temperatures, the technique is especially viable for oxidic materials, because no inert gases are required. As a downside of this technique, 2D templates are required beforehand, which increases the expenses significantly. Therefore, another template-oriented technique is worth mentioning, where 3D structured salt is used (see Figure 4).

Fig. 4:

Salt-templated synthesis of 2D oxide nanosheets: (a–c) Schematic principle of salt-templated synthesis of nanosheets. (a, b) The first step includes the coating of the salt with the precursor solution, (c) followed by the growth and washing to gain the 2D oxide nanosheets. (d) In this case, the dispersion containing the nanosheets were filtrated to utilize the nanosheets on electrodes for their pseudocapacitory characteristics. (e) Photography of the salt mixed with the precursor solution in the front row and the dispersion with the gained MoO₃ nanosheets in the back row. (f) Cross-sectional SEM image of stacked 2D oxide nanosheets with a scalebar of 1 μm. Adapted with permission from [71]. Copyright 2016 Springer Nature. Creative Commons License CC BY 4.0.

Xiao et al. [71] used this technique to synthesize nanosheets consisting of MnO, MoO₃ or WO₃. An excess of cheap and easily obtainable salt as KCl or NaCl was used and mixed with a precursor solution containing the inorganic metal-salt. After annealing by heating, the oxidic nanosheets were formed with an intermediatic hydroxide phase and dissolving the salt in water led to the free-standing sheets. While in 2D templates the morphology is predefined, the mechanisms in the 3D salt are different. It is assumed, that the crystal geometry of the salt and the oxide have to be similar to allow Frank-van der Merwe film growth on the surface of the salt crystals and favor the lateral growth [71]. Though more investigations are necessary to fully understand the mechanism, not matching crystal geometries showed no formation of nanosheets. To control the thickness of the sheets, the ratio of salt to precursor volume can be adjusted and increases with larger volume. The advantage of this technique lies within the utilization of a cheap and reusable template with few reaction steps, which is favorable for large-scale productions.

3.8 Bottom-up synthesis: self-assembly of nanosheets

By utilizing non-covalent interactions like van der Waals or elecrostatic forces as in hydrogen bonds, the self-assembling of nanocrystals is a neat way to create nanosheets [111]. It is obvious, that for so-called self-assembling an accomodated driving force is still required, because the attractive forces have to operate in two dimensions or the repulsive forces in only one dimension. For this, the general synthesis contains structure-directing agents (SDAs) as surfactant molecules with amphiphilic structure properties (e.g. block copolymers) and a co-surfactant (e.g. alcohol). The metal species for the formation of nanosheets can be added as alkoxides or salts. The amphiphilic properties of the SDA provides attractive forces between one end of the molecule and the metal species as well as van der Waals forces between the long chains, thus leading to a stacking and formation of inverse lamellar micelles (see Figure 5). Hereby, the role of the co-surfactant is to ensure a stable lamellar phase. When the amount of lamellar phase is sufficient, a solvothermal treatment is added for crystallization and free-standing nanosheets are obtained after oxidative removal of the SDAs [111]. The advantages of this synthesis are a formation of a structured host material on a molecular basis and a scalable process. Although, the amounts of SDAs have to be carefully adjusted to the sensitive equilibrium of the lamellar phase.

Fig. 5:

Schematic principle of self-assembly for 2D oxide nanosheets. In the first step, the metal precursor gets orientated in two dimensions due to surfactant molecules to form a stacked lamellar structure. In the second step, the precursor oligomers get crystallized and the template is removed to gain the nanosheets. Adapted with permission from [111]. Copyright 2014 Springer Nature.

3.9 Bottom-up synthesis: interface-mediated methods

Another technique, similar to self-assembling, is the interface-mediated synthesis. In a typical approch, a metal salt is dissolved and supersaturated in an aqueous solution and an anionic organic molecule with a long hydrophobic tail is added, whose localization at the interface is essential for the nanosheet formation. To ensure a monolayer of the surfactant at the water/air interface, the polymer can first be dissolved in a hydrophobic medium and afterwards added to the water. Once the surfactants form a monolayer at the interface, the attraction of the negatively charged molecules with the positively charged metal cations is essential. In a supersaturated solution, the precipitation of the oxide is controlled at the interface, despite the formation of nanocrystallites elsewhere in the solution. Since the oxide is already crystallized, a calcination step can be avoided. For the formation of ZnO it was shown, that the reaction time and the cation concentration are crucial to the crystallinity of the sheets [112]. While this method enables especially the formation of large nanosheets due to missing calcination steps, it still contains a monolayer of the surfactant after extraction. Although, its removal may influence the stability of the sheets. Hence, at this state this method is limited to nanosheets where the monolayer of surfactant does not hinder the application fields.

A recently emerged synthesis route utilizes self-limiting thin interfacial oxides for the formation of atomically thin metal oxides [97]. It is especially useful for the synthesis of 2D nanosheets from compounds with non stratified crystal structures. In contrast to most other synthesis methods, a metal-based alloy is used as a solvent (e.g. eutectic gallium melts). In this case, the formation of an oxide skin at the liquid/air interface is mandatory (see Figure 6b).

Fig. 6:

Principles for the interface-mediated synthesis of 2D oxide nanosheets using metal alloys: (a) Gibbs free energy of several oxides in relation to Ga₂O₃, which is formed without co-alloying. (b) Schematic composition of a liquid-alloy droplet with a liquid core and an oxide skin at the liquid/air interface with possible crystal structures of displayed oxides. (c) Schematic principle for the attachment of the oxide skin to a solid substrate. Based on van der Waals forces, the oxide skin adheres to the solid surface. An optical image of the 2D nanosheets is presented at the right. Adapted with permission from [97]. Copyright 2017 The American Association for the Advancement of Science.

Zavabeti et al. [97] managed to gain nanosheets of HfO₂, Al₂O₃ and Gd₂O₃ to show the large compound variety of the technique by co-alloying the gallium melt with suitable metals. The suitability is proposed to be dependent on thermodynamic aspects, because the surface oxide with the highest absolute value of Gibbs free energy is preferentially formed at the interface (see Figure 6a). With no co-alloying, Ga₂O₃ is formed at the interface, therefore only oxides with enhanced Gibbs free energy can be synthesized by this method. To separate the oxide skin from the melt, two different approaches are presented. For the attachment on a specific surface, a substrate can be contacted to the liquid-alloy droplet as shown in Figure 6c. To gain the 2D nanosheets in suspension, water can be placed above the melt while gaseous air is injected into the melt below. By this, the amount of required interface is manually controlled and scalability is possible [97]. Although, due to the polycrystalline structure of the gained nanosheets, the method is limited to applications where single-crystals play a subordinated role.

3.10 Bottom-up synthesis: sol-gel method

The sol-gel method has its origins in the 1990s [58] and partly resembles the interface-mediated synthesis. Analogous to the interface-mediated synthesis, the metal species is dissolved in water and hydrolized. An amphiphilic chelating agent is placed in a monolayer at the water/air interface as well. The first reaction step includes the condensation of the metal species at the interface due to the complexation of the amphiphilic molecules with the metal centers. This process can be used to gain large nanosheets by using the Langmuir-Blodgett technique, where the singular oxide-based gel islands are merged by reducing the interface area. After sufficient condensation, an ultrathin gel is formed at the interface, which further can be extracted onto a substrate. Therefore, the Langmuir-Blodgett technique is convenient and the dipping of the substrate into the solution leads to a gel-film on both sides of the substrate surface. In the last reaction step of calcination, the complexing agent is removed and the gel is substituted by the oxide species. While this synthesis is in fact interface mediated, it is well established and the gel-formation as a characteristic step distinguishes it sufficiently. Moriguchi et al. [58] accomplished to gain heterolayers of TiO₂ and ZrO₂ with a thickness of 20 nm by this technique on a Si-substrate. The homogeneity of the films are superior to some other synthesis methods, but upscaling this process is unresolved.

3.11 Bottom-up synthesis: growth method

A simple method for growing an array of CuO nanosheets on a nickel foam substrate was shown by Wang et al. [80]. Therefore, a basic growth solution including the metal salt was prepared and the nickel substrate hung into it for 6 h at 90°C. It is similar to solvothermal methods, but does not require an autoclave or high temperatures. The gained nanosheets showed sufficient homogeneity on the substrate, but did not lead to dense sheets. Instead, they were formed by interconnecting nanoparticles, which allowed irregularities and pores. Nevertheless, the synthesis is easily managable and delivered beneficial properties regarding pseudo-supercapacitors. Determining parameters are temperature, concentration and growth time. The transferability for other metal species has yet to be shown though.

4.1 Free-standing nanosheets

Free-standing nanosheets can be synthesized via top-down and bottom-up routes as described in the Section Synthesis. The commonly used top-down route by exfoliation of layered 3D bulk compounds due to intercalation of an exfoliation agent (see Figure 2) is able to produce single-crystalline 2D oxide nanosheets. Often, the freestanding 2D nanosheets have similar structures as the parent material with slight in-plane expansion and eventually change in symmetry. In the following, structurally examined examples of exfoliated free-standing 2D nanosheets will be described and subsequently examples for bottom-up results.

The AFM-measured thickness of exfoliated 2D nanosheets is systematically higher than expected from crystallographic data, which is mainly caused by the absorption of water and other species from solution. Anyway, these measurements allow to elucidate if unilamellar or multiple-layered (e.g. bi- or tri-layer) nanosheets were obtained and moreover resolve differences in ionic radii of the center ion in structure-defining octahedra [113]. The example in Figure 7 exhibits clearly different thicknesses for perovskite-related Ca₂Na_m−3Nb_m O_3m+1⁻ nanosheets, which confirms the high precision of AFM measurements and the possibility to distinguish between unilamellar or multiple layer nanosheets [28].

Fig. 7:

AFM-analysis of Dion-Jacobson phases: (a) Theoretical structures of Ca₂Na_m−3Nb_m O_3m+1 2D nanosheets. (b–e) Morphology of 2D nanosheets, as received from the exfoliation of bulk Dion-Jacobson phases, measured with AFM in tapping mode under vacuum. Reproduced with permission from [28]. Copyright 2017 American Chemical Society.

Due to surface charges, unilamellar nanosheets of oxides like graphene oxide (GO), Ti_0.87O₂^0.52− and Ca₂Nb₃O₁₀⁻ can be well dispersed in solvents in the form of 2D unilamellar anionic, i.e. negatively charged nanosheets [114]. By the aid of AFM, Cai et al. [114] have shown, that 2D nanosheets of Ti_0.87O₂^0.52− have a uniform thickness of ca. 1.1 nm and that the surfactant polyethylenimine (PEI) which interacts with its oxidative functional groups to the 2D nanosheet, can increase its thickness uniformly to 2.9 nm (see Figure 8). This study also shows, that by using PEI it is possible to preserve the monodispersibility and avoid restacking-induced flocculation of nanosheets, which could be expected in the case of simple addition of cationic species.

Fig. 8:

Tapping mode AFM analysis of exfoliated Ti_0.87O₂^0.52− nanosheets: (a) Original 2D nanosheets and (b) nanosheets modified with polyethylenimine (PEI). (c) Thickness of original and modified 2D nanosheets at different PEI concentrations (50 and 125 g dm⁻³) and pH values (9 and 11) of the solution. (d) ζ potential of the original and the PEI-modified 2D nanosheets as function of pH value. Adapted with permission from [114]. Copyright 2015 American Chemical Society.

To observe expansions of certain crystallographic planes attributed to the exfoliation, SAED in a TEM can be used. Body-centered orthorhombic RbLaNb₂O₇ (a=5.4941 Å, b=21.9901 Å, c=5.4925 Å) and RbPrNb₂O₇ (a=5.4534 Å, b=22.012 Å, c=5.4549 Å) of double-layer Dion-Jacobson type were exfoliated into nanosheets using microwave heating in the presence of TBA⁺ [32]. It resulted in TBA⁺-attached 2D nanosheets of the type TBA⁺-LaNb₂O₇ and TBA⁺-PrNb₂O₇. The delamination of the crystalline structure occurred in the (a,c)-plane, so that SAED patterns along the former b-axis (i.e. [010] zone axis in the bulk material) could be observed as shown in Figure 9. These SAED patterns could still be indexed on a body-centered cell and gave values for the 2D lattice parameters a and c being close to those in the starting material: TBA⁺-LaNb₂O₇ (a=5.71 Å, c=5.71 Å), TBA⁺-PrNb₂O₇ (a=5.68 Å, c=5.74 Å). However, in both cases the nanosheets have expanded about 4%–5% in the (a,c)-plane [32].

Fig. 9:

TEM micrographs and SAED patterns of TBA-functionalized 2D nanosheets: (a) TBA⁺-PrNb₂O₇, (b) TBA⁺-LaNb₂O₇. Nanosheets are viewed along the corresponding b-axis in the orthorhombic bulk two-layer Dion-Jacobson phases. Reproduced with permission from [32]. Copyright 2017 Wiley-VCH.

Moreover, orthorhombic Rb₄W₁₁O₁₄ (a=14.64 Å, b=25.78 Å, c=7.64 Å) was exfoliated in the (a,c)-plane and SAED revealed a rectangular unit cell for the resulting 2D nanosheets (a=14.0 Å, c=7.8 Å), which dimensions are close to those of the corresponding bulk material [17]. The perovskite-type monoclinic RbTaO₃ (a=9.60 Å, b=8.426 Å, c=7.33 Å, β=94.2) was exfoliated in the (a,b)-plane and the 2D nanosheets (a=9.8 Å, b=8.7 Å) showed an in-plane expansion of 2%–3% compared to the corresponding bulk material. However, a particular feature of the nanosheets were open channels of the 1×1 width of a TaO₆ octahedron [98]. It is worth to note, that for the investigation of 2D nanosheets SAED and AFM are complementary to each other, as the former gives in-plane information and the latter gives out-of plane information.

Upon a tetramethylammonium hydroxide (TMA⁺OH⁻)-mediated delamination of Bi₂W₂O₉ to single-crystalline WO₃ 2D nanosheets, the nanosheets undergo a structural change from tetragonal symmetry in the parent material to monoclinic in their free-standing form, which resembles bulk WO₃ (a=7.297 Å, b=7.539 Å, c=7.688 Å, and β=90.85) and was elucidated by powder XRD, HRTEM and high-resolution STEM-HAADF [101]. In the WO₃ 2D nanosheets with the formal composition H₂W₂O₇, the WO₆ octahedral units were stacked in two levels, giving the sheets a thickness of only 0.75 nm while the average edge length was 90±38 nm [101].

For a rapid and reliable identification of unilamellar and up to five-layer oxide nanosheets, Kim et al. [100] have proposed an universal optical method by analyzing Ti_0.87O₂, Ca₃Nb₃O₁₀, and Ca₂NaNb₄O₁₃ nanosheets. The method is based on changes in the interference-based optical reflectivity of 2D nanosheets on SiO₂/Si substrates, which is dependent on the nanosheet thickness, SiO₂ film thickness and the optical wavelength. In an optical microscope, the contrast, which can be inverted in some cases, was carefully evaluated for varied parameters.

Considering the exfoliation behavior of protonated Ruddlesden-Popper phases H₂[A_n−1B_n O_3n+1], Schaak and Mallouk [30] mention their unpredictable tendency to curl and form scrolls rather than sheets and they give several possible explanations for why they do so. One explanation for the formation of scrolls is that, upon exfoliation, the individual sheets have asymmetric distributions of A- or B-site cations, creating a polarization that is relieved by curling. A second possibility is that the sheets curl as a result of cooperative distortions of the BO₆ octahedra, which is frequently observed in displacive ferroic phase transitions in perovskites. Curling may also occur due to the intrinsic bonding character of water or the intercalated base, which is used in the exfoliation process, with the interlayer atoms of the perovskite block. Anyway, there is still not a clear understanding of how intercalation affects the interlayer bonding and why certain Ruddlesden-Popper phases as well as some Dion-Jacobson and Aurivillus phases curl rather than remain as sheets [30].

For analyzing the impact of the chosen intercalating agent on the 2D nanosheets, a work from Takagaki et al. [35] should be mentioned. Several 2D nanosheets (TiNbO5−,Ti2NbO7−, and TiTaO5−) were prepared by TBA-mediated exfoliation of the corresponding H⁺-exchanged layered oxides (HTiNbO₅, HTi₂NbO₇, and HTiTaO₅) [35]. However, the addition of TBA⁺OH⁻ solution did not result in the exfoliation of layered HTiTaO₅, which was only achieved by adding ethylamine in an aqueous environment. SAED patterns of the titanium niobate nanosheets (TiNbO5− and Ti2NbO7−) exhibited single-crystalline nature (discrete spots pattern), while those of the TiTaO5− nanosheets showed polycrystalline nature (diffuse Debye-Scherrer rings). TEM revealed, that the TiTaO5− 2D nanosheets were decorated by equiaxial nanoparticles of 2–10 nm in diameter. This obviously shows, there is a definitive distinction between the exfoliation behavior of TBA⁺OH⁻ and ethylamine [35].

After describing multiple examples with their characteristics in top-down exfoliated 2D nanosheets, in the following nanosheets obtained from bottom-up routes will be discussed. Generally, bottom-up routes produce rather polycrystalline free-standing nanosheets as is described in the following examples. Although, single-crystalline nanosheets can also be obtained if the synthesis is appropriately adjusted.

The bottom-up hydrothermal method from Lu(OH)₃-based colloidal precursor produced Eu- or Tb-doped Lu₂O₃ square 2D nanosheets with thicknesses of ca. 40 nm and side lengths of 200–400 nm [115]. As revealed by HRTEM, the single 2D nanosheet has a polycrystalline structure and is composed of planar arrangement of equiaxial nanocrystals with 10–15 nm in diameter each. As a result, each individual Lu₂O₃ square free-standing 2D nanosheet is a nanoparticle 2D assembly. Similar results were observed for NiCo₂O₄ 2D nanosheets, prepared by a hydrothermal method and subsequent annealing. Planar 2D nanosheets with lateral dimensions of more than 300 nm were gained, which consisted of randomly arranged NiCo₂O₄ nanoparticles with diameters of less than 10 nm and of mesopores [39]. Thus, these free-standing nanosheets can be considered as nanoparticle assemblies. Subsequently, irregular arrays of stacked NiCo₂O₄ nanosheets were obtained by immersing stainless steel gauzes in the solution for hydrothermal method in polytetrafluoroethylene (PTFE)-lined autoclaves.

On the oher hand, single-crystalline ZnO nanosheets were gained by an interface-mediated synthesis. Surfactant monolayers were used as soft templates to produce 1–2 nm thick single-crystalline ZnO 2D nanosheets at the water-air interface with lateral sizes of up to tens of nanometers [112]. By collecting nanosheets at different reactions times (see Figure 10), HRTEM revealed that at first there was a continuous amorphous film at the interface with tiny crystallites embedded in it. Then, these crystallites grew in lateral size and were all oriented with the same hexagonal crystal plane exposed. However, their in-plane rotation appeared to be stochastic. As the crystallites grew larger, they merged at an aligned orientation into a contiguous, single-crystalline network coexisting with a decreasing amorphous region confined between the nanosheets. Eventually, the amorphous area was fully crystallized and the nanosheet became single crystalline with few dislocations that were probably formed by the misorientation of merged crystalline areas during the formation process [112]. This synthetic process, which has similar attributes as those found in biomineralization, has the potential to produce single-crystalline 2D nanosheets from a wide range of inorganic materials.

Fig. 10:

Plan-view HRTEM analysis of time-dependent evolution of ZnO 2D nanosheets obtained by an interface-mediated synthesis: (a) Mostly amorphous film with tiny crystallites, (b) more crystallized nanosheet with randomly in-plane oriented 2–3 nm crystallites, (c) in-plane crystallites had grown larger and aligned orientation, (d) large-area single-crystalline nanosheet. Insets show fast Fourier transforms (FFTs) of the respective HRTEM micrographs. In the structural models, regions with gold-colored spheres are amorphous and regions with deeper gold-colored spheres are crystalline. Scale bars are 2 nm each. Reproduced with permission from [112]. Copyright 2016 Springer Nature. Creative Commons License CC BY 4.0.

4.2 Stacked nanosheets

SEM has the capability to image material microstructures over six orders of magnitude from the mm- down to the nm-scale. So, it is of particular use in investigating assemblies of individual nanosheets. An example is given in Figure 11, which shows 2D Na_0.7CoO₂ nanosheets turbostratically stacked into macro-scale pellets [14]. The length of the nanosheets is exceeding 1.5 mm as can be seen from Figure 11e at low magnification. At the higher magnification in Figure 11a–c, it is revealed that lamellae made of the stacked nanosheets are partly wrinkled. In Figure 11c, individual nanosheets of less than 20 nm in thickness are resolved within a wrinkled lamella. SEM investigation covering macro-scale and nano-scale was made on water-stabilized, millimeter-length, stacked Ka_x CoO₂·yH₂0 nanosheets, which were obtained by an analogous synthesis method [116].

Fig. 11:

Millimeter-length 2D Na_0.7CoO₂ nanosheets fabricated from a bottom-up sol-gel process followed by autocombustion, alignment, calcination and electric field-induced kinetic demixing: (a–c) High-magnificiation SEM shows the thickness of individual sheets to vary from ca. 20 to 100 nm. (d) The bulk pellet consists of thousands of stacked nanosheets from which a single nanosheet stack was mechanically extracted with the individual nanosheets being aligned as drawn. (e) Low-magnification SEM shows the total length of nanosheets to be 1.8 mm and the stack thickness to be around 100 μm. Adapted with permission from [14]. Copyright 2012 Royal Society of Chemistry.

Apart from this, another type of stacked nanosheet assemblies is conceivable with focus on the transition of nanoscopic characteristics to macroscopic materials. For this, separately gained 2D oxide nanosheets can be aligned to each other by organic fibers to macroscopic fiber assemblies [117], [118]. As shown in Figure 12a, a dispersion of exfoliated Ti_0.87O₂^0.52− nanosheets in a liquid crystal were successfully assemblied with the biopolymer chitosan via a wet-spinning method using a coagulation bath to gain the fibers. The nanosheets within the fiber assemblies (see Figure 12b,c) were highly ordered and led to extraordinary mechanical enhancements competing with those of graphene. This is particularly important, because generally metal oxides only provide few hundredths of the intrinsic tensile strength in a single nanosheet compared to graphene [117]. The mechanical stability together with mechanical flexibility is e.g. useful for energy storage in flexible lithium-ion fiber batteries [119].

Fig. 12:

Stacked nanosheets in macroscopic fiber assemblies: (a) Schematic principle of a wet-spinning method for the assembly of titania nanosheets in a liquid crystal (LC) with a coagulation agent chitosan in a rotating bath. (b, c) SEM micrographs showing the macroscopic fibers with stacked nanosheets as building blocks and flexible characteristics. Adapted with permission from [117]. Copyright 2015 American Chemical Society.

4.3 Nanosheet arrays

SEM micrographs of mesoporous ZnCo₂O₄ nanosheet arrays, which were uniformly grown on the ridges of a Ni foam substrate by hydrothermal method followed by calcination [40] are shown in Figure 13. At low magnification, pores within the Ni foam having diameters of up to 100 μm can be seen. At higher magnification, individual upright standing 2D nanosheets with a thickness of less than 100 nm are resolved. Due to their deposition on the surface, they can obviously be distinguished from free-standing nanosheets. The high electrical conductivity of the Ni foam combined with the large accessible surface of the mesoporous ZnCo₂O₄ nanosheets give prospect to integrated electrodes for electrochemical supercapacitors for energy storage at high cycling rate.

Fig. 13:

Mesoporous ZnCo₂O₄ 2D nanosheet arrays, which were uniformly grown on Ni foam substrate by a bottom-up hydrothermal method followed by calcination: (a) Low magnification SEM micrograph showing the macroporous Ni-foam and the deposited nanosheets on the surface. (b–c) SEM micrographs at higher magnifications showing the nanosheet arrays. Adapted with permission from [40]. Copyright 2013 Royal Society of Chemistry.

Arrays of vertically standing CuO 2D nanosheets on Ni foam as a substrate were pepared via the bottom-up template-free growth method from aqueous copper nitrate NH₄NO₃/ammonia solution [80]. The thickness of the obtained 2D nanosheet film was about several μm, while the individual nanosheets were about 150 nm in thickness and parallely clustered into stacks.

Hierarchically crossed metal oxide 2D nanosheet arrays (Co₃O₄, NiO, MgO) were produced by bottom-up pyrolysis of a thin nitrate film on a FeO_x substrate, which was obtained by annealing an iron foil in air [23]. In the case of Co₃O₄ (see Figure 14), 2D nanosheets with an average thickness of about 40 nm were uniformly and vertically distributed on the iron foil after annealing. The hierarchical 2D nanosheet arrays showed promising catalytic activity for the elimination of soot from Diesel exhaust.

Fig. 14:

Hierarchical Co₃O₄ 2D nanosheet array on FeO_x substrate: (a) Schematic illustration of the preparation by oxidation of Fe foil in air to exhibit a Fe₃O₄ surface film with Fe₂O₃ nanosheet array to which cobalt nitrate was deposited and thermally decomposed to a crossed Co₃O₄ 2D nanosheet array, (b) plan-view SEM micrograph of FeO_x substrate, (c) TEM micrograph of FeO_x substrate, (d) plan-view SEM micrograph of Co₃O₄ nanosheet array on FeO_x substrate, (e) side-view SEM micrograph of Co₃O₄ nanosheet array on FeO_x substrate. Adapted with permission from [23]. Copyright 2016 Royal Society of Chemistry.

By an anodic process in a double-electrode cell, vertical standing nanosheet arrays composed of coexisting Cu₂O and CuO with a nanosheet thickness of ca. 30 nm were produced on a copper foil substrate [79]. High intensity diffraction peaks of CuO and low intensity diffraction peaks of Cu₂O were detected in XRD patterns of the anodization products obtained at temperatures of 60 and 70°C.

Moreover, arrays of vertical standing ZnO 2D nanosheets were deposited onto an indium tin oxide (ITO) coated transparent conducting glass substrate, using a galvanic deposition process in a Al/ZnSO₄//NaOH/ITO/glass cell with aqueous electrolytes. The ZnO 2D nanosheets were of hexagonal wurtzite structure and had lateral dimensions of up to 1 μm at thicknesses of less than 50 nm [84].

4.4 Ultrathin films

Technically, every film with a thickness lower than 10 nm can be considered ultrathin, which can be accomplished by several bottom-up routes as laser-deposition or molecular beam epitaxy. Another approach to obtain ultrathin films lies within utilizing nanosheets. As an example, bulk triple-layer Dion-Jacobson phase KCa₂Nb₃O₁₀ was exfoliated to a colloidal suspension of Ca₂Nb₃O₁₀ 2D nanosheets with lateral dimensions of about 3–10 μm from which, as shown in Figure 15, a multilayered ultrathin film was prepared by a layer-by-layer assembly using the Langmuir-Blodgett technique [120]. In the first step, the atomically-flat SrRuO₃ perovskite substrate is effective in obtaining an atomically uniform monolayer film with high dense characteristics. Repeated Langmuir-Blodgett deposition yielded (Ca₂Nb₃O₁₀)_n with n=3 (4.5 nm, shown in Figure 15c, d), n=5 (7.5 nm), n=10 (15 nm), n=15 (22.5 nm). Besides SrRuO₃, also quartz glass, Pt or SrTiO₃:Nb were used as substrates [120].

Fig. 15:

Ultrathin films gained by using the Langmuir-Blodgett method: (a) Layer-by-layer assembly of delaminated 2D metal oxide nanosheets into multilayer films using the Langmuir-Blodgett method. (b) Schematic crystallographic orientation of a (Ca₂Nb₃O₁₀)₃ Dion-Jacobson phase relative to a SrRuO₃ perovskite substrate. (c) AFM plan-view with color scale refering to altitude. (d) HRTEM cross-sectional view of a (Ca₂Nb₃O₁₀)₃ 2D film on a perovskite substrate. These 2D films are prospective high-κ dielectrics/ferroelectrics for use in ultrascaled electronics and post-graphene technology [28]. Reproduced with permission from [120]. Copyright 2010 American Chemical Society.

The Langmuir-Blodgett technique was also used to produce about 10 nm thin superlattices of double-layer LaNb₂O₇ (d=1.2 nm) and triple-layer Ca₂Nb₃O₁₀ (d=1.4 nm) Dion-Jacobson phases from their 2D nanosheets by a layer-by-layer assembly [121]. Interface coupling in these (LaNb2O7)nL (Ca2Nb3O10)nC superlattices gave rise to ferroelectricity regardless of the stacking sequence (nL/nC) of the two different 2D nanosheets. Cross-sectional HRTEM confirmed that wide-ranging uniform artificial superlattices could be produced on an atomically-flat SrRuO₃ substrate. Compositional modulation in alternating lamellae was elucidated by EELS of the La-M_2,3 (192 eV) and Ca-L_2,3 (347 eV) ionization edges in the STEM. From the viewpoint of crystal chemistry, unique intergrowth structures were obtained for the (LaNb2O7)nL (Ca2Nb3O10)nC superlattices, which do not naturally exist in the bulk form.

A 2D bottom-up sol-gel process has been described, in which the hydrolysis and polycondensation reactions occur at the air/water interface and the Langmuir-Blodgett technique is used to deposit a gel film on a substrate [122]. The 2D sol-gel process was further developed by involving repeated layer-by-layer Langmuir-Blodgett technique to produce (ZrO₂/TiO₂)_n hetero-multilayered nanofilms [58]. The multilayered nanostructure made up of TiO₂ and ZrO₂ ultrathin laminae with thicknesses of about 10–15 nm was observed throughout the Si substrate surface. A (ZrO₂/TiO₂)₅ film had a total thickness of about 130 nm. It exhibited alternate stacking of ZrO₂-lamina and TiO₂-lamina in the direction from the outer-most surface to the substrate with the diameter of the nanoparticles being 3 nm or less in the ZrO₂ layers and 5–10 nm in the TiO₂ layers as shown by HRTEM.

A more exotic example for ultrathin films is the using of a chemically driven self-assembly process in the gaseous phase, where 2D planar vanadium oxide [V₆O₁₂] clusters were produced on a Rh(111) surface and subsequently monitored in the ultra-high vacuum using the STM [67]. In a reducing environment (pH2≈10−8 mbar) at a substrate temperature of 250°C, the [V₆O₁₂] clusters assembled into nano-islands with a well-ordered 2D vanadium oxide monolayer structure. STM revealed a rectangular (5×33)-rect structure and incorporated star-shaped [V₆O₁₂] clusters at the boundary to the free Rh(111) surface [67]. The star-shaped [V₆O₁₂] clusters became mobile on the surface above 100°C.

4.5 Nanoparticle assemblies

As already mentioned for hydrothermally gained free-standing nanosheets, 2D nanosheets can also be obtained as nanoparticle assemblies. In contrast to the single-crystalline nanosheets, the functional properties may vary substantially, because the nanoparticle assemblies are polycrystalline and have a lower density with irregular pores between the particles. Nevertheless, they can be of significant interest as shown for a net-like SnO₂ 2D nanostructure in Figure 16. TEM is unrivaled in elucidating the arrangement of individual nanoparticles when assembled to 2D nanostructures. Here, particles are less than 10 nm in diameter with narrow size distribution. The HRTEM in Figure 16c reveals lattice plane distances, which enable the identification of the relative orientations of individual nanoparticles. However, both HRTEM and the SAED in the inset, which shows narrow Debye-Scherrer rings, suggest the individual particles to be oriented randomly. As an approach for complementary investigations, when TEM is operated in STEM mode, it provides with HAADF contrast, EDXS and EELS [123] powerful techniques to unequivocally distinguish different phases in hetero-nanostructures like SnO₂/ZnO as they were fabricated also by Fu et al. [53].

Fig. 16:

TEM analysis of net-like SnO₂ 2D homo-nanostructure as obtained from a wet-chemcial process with graphene as a 2D template: (a, b) TEM bright-field micrographs, (c) HRTEM micrograph. Inset shows SAED pattern with indexed Debye-Scherrer rings. Adapted with permission from [53]. Copyright 2015 Royal Society of Chemistry.

5.1 Electrochemical energy storage

A highly promising application field especially for 2D oxides is the utilization in rechargeable batteries and supercapacitors. The setups for both applications are roughly the same containing negative and positive electrodes, electrolytes and the separators [125]. However, they distinguish in functionality, properties and different requirements in materials. In rechargeable batteries, an electrochemical cell with an anode and cathode is formed to ensure the electrical current caused by redox reactions. The large usage in nowadays applications, as all sorts of mobile technical devices, shows the indispensable importance of this technology. On the other side, capacitors are designed to store potential energy by the alignment of opposing charges at two electric conductors in an electric field. In the case of supercapacitors, they nowadays are used as a replacement of batteries or complementarily to them to gain additional power as in electrical vehicles or hybrid electric vehicles by converting acceleration energy during regenerative braking [126]. The main advantage of supercapacitors versus batteries lies within their ability to store substantially more energy and have way more stable and faster cycling characteristics, although they are yet limited to low voltage applications due to lower energy densities. Usually, the life span of a battery is limited by non-reversible reaction products, which slowly decrease the energy density after each recharging. The replacement or combining of commonly used rechargeable batteries with supercapacitors may solve this issue satisfactorily. Supercapacitors can be separated in electrochemical double-layer capacitors (EDLCs) and pseudocapacitors [126]. The EDLCs contain electrolyte ions, whose adsorption on preferably large specific surface areas of porous electrodes is essential to the charging of the capacitor. For the energy storage of pseudocapacitors, reversible surface-faradaic redox reactions at the interface of electrode and electrolyte occur [126]. Therefore, pseudocapacitors are closer to the functionality of batteries than EDLCs, but their efficiency is diffusion limited. Crucial to the perfomance of a supercapacitor is the choice of materials and their design at the electrodes.

At this point, the 2D oxides come into play. Due to their faradaic behavior, metal oxides are prominent for their pseudocapacitive behavior rather than the utilization in EDLCs. In EDLCs various appearances of carbon are the currently state-of-the-art [126]. Prominent representative 2D oxides with very high capacities are RuO₂ and IrO₂, which suffer from their scarce availability though. Cheaper alternatives with likewise or slightly lower capacities are NiO, CuO, SnO₂, Co₃O₄, FeO_x , MoO₂, Cr₂O₃, MnO₂, V₂O₅, NiCo₂O₄, ZnCo₂O₄ and their doped deviates [38], [40], [81], [93], [99], [125], [126]. As proof of principle, the results of Rui et al. [66] gained with V₂O₅ nanosheets are qualified for the general utilization of nanosheets in hybrid Li-ion batteries. As shown in Figure 17, the usage of 2D nanosheets shows significant improvement in specific capitance, coulombic efficiency and power density compared to the 3D bulk V₂O₅. Possible explanations for this outcome are, that in contrast to the 3D bulk counterpart, the electrolyte can penetrate assembled nanosheets due to the large specific surface area, which leads to an increasing number of surface reactions. Furthermore, the small thickness of only 2.1–3.8 nm shortens the diffusion paths for the charge carriers, which results in much higher charge and discharge kinetics [66]. By this, hybrid Li-ion batteries with faster charging processes and longer cycle-lifetimes are possible.

Fig. 17:

Vanadium pentoxide V₂O₅ 2D nanosheets, as observed from direct exfoliation of bulk crystals in formamide solvent, were tested on coin cells for their lithium storage capacity as prospective cathode material in the lithium-ion battery (LIB). A metallic lithium counter electrode was used and the 2D nanosheet material was compared to bulk V₂O₅. (a) Charge-discharge curves show voltage over specific charge capacity. For 2D nanosheet material, the latter is distinctly enhanced over the bulk material. (b) At a rate capability of 0.2 C, whether the specific capacity nor the Coulomb efficency fade away for the 2D nanosheet material. The C-rate describes the rate of charge or discharge in a cell or battery with respect to the nominal rated capacity [127]. (c) The 2D nanosheet material shows a stable rate capability at different charge and discharge rates (up to 50 C). (d) The Ragone plot suggests that a LIB with V₂O₅ 2D nanosheet electrode might outperform over various energy storage and conversion devices with respect to both power density and energy density. Reproduced with permission from [66]. Copyright 2012 Royal Society of Chemistry.

Another promising example of utilized nanosheets, is the formation and characterization of a hybrid graphene/Cr₂O₃ hetero-nanostructured anode by a layer-by-layer approach [47]. Compared to graphene/Cr₂O₃ nanoparticles or other Cr₂O₃ based electrodes as a whole, the perfomance of the electrode could be significantly improved in terms of cycling stability and capacity. This approach exploits the very high theoretical capacity of 1.058 mAhg⁻¹ from Cr₂O₃ and solves problems with cycling perfomance and electron transport within the material. Again, this could be achieved by the 2D arrangement, which causes a better electronic contact of the active materials and led to a way more stable connection between graphene and Cr₂O₃ compared to the nanoparticles. Another attempt to utilize 2D oxides for batteries, is the intercalation of useful Li ions within the layered 3D bulk material by exfoliation and subsequent reassembly of the nanosheets to form an electrode with improved characteristics [76]. Overall, metal oxides appear indispensable for future electrochemical storage applications and the usage of 2D nanosheets show highly promising improvements in capacity, cycle-lifetime and energy density.

5.2 Electrochromic devices

Optical properties of a material are defined by their electronic band structures, which are on the other hand influenced by size effects. Moreover, when the thickness of a film is lower than the penetration depth of the wavelengths of the observed light, the absorption gets drastically reduced. For usual applications at visible and UV light, this occurs at a thickness of less than 10 nm [128]. Practically, this means materials may get more transparent with reducing thickness. This effect becomes even more interesting, if the transparency and absorbance are reversibly controllable, e.g. by applying an electrical current. Materials possessing this property are called electrochromic.

An attractive application for these materials are the so-called smart or switchable windows [129], which could help to reduce the energy consumption within buildings. The idea behind this is, to ensure normally transparent windows when people are within the room and add the possibility to cool and dim the brightness by applying a voltage to the electrochromic window. This leads to a doping of the electrochomic material and a change in color, light absorbance and reflection. Especially, the absorbance of light within the infrared spectrum may lead to an increased cooling effect. Therefore, the choice of the electrochromic material is one key factor within the whole electrochomic device alongside to the glass, transparent conductors and the electrolyte. Metal oxides as WO₃, MoO₃, Ta₂O₅, TiO₂, V₂O₅ and a few more are by far the most relevant electrochromic materials for this kind of application besides some polymers [129], [130]. Hereby, the tungsten oxides play an outstanding role compared to other metal oxides. In the application itself a film is structurally required, which can be attached to one side of the transparent conductor, respectively electrode (see Figure 18). The transparent conductor can likewise be composed of an ultrathin 2D oxide as indium-tin-oxide (ITO), which appears transparent due to its small thickness but does not show electrochromic characteristics.

Fig. 18:

Schematic setup with different thin layers for an electrochromic window application. Adapted with permission from [131]. Copyright 2016 Royal Society of Chemistry.

Regarding the influence of 2D materials, few attempts have been made to investigate WO₃ nanosheets as a component for building the film [17], [18], [132]. The nanosheets consisting of WO₃ showed intense UV absorption and switchable IR absorption, which is favorable application-wise. An actual film consisting of nanosheets was obtained with the layer-by-layer technique by Wang et al. [132], which showed good coloration efficiency together with nearly no reduced efficiency after over 800 cycles. The unique property of the film is a slow coloration process compared with a quick bleaching process, which can be attributed to the nanosheets implemented in the film. The suggested reason for this is the repulsion of charged intercalates at the top nanosheet layers during the coloring process, while at the opposing bleaching process the intercalates can easily be extracted from the top side without facing repulsion forces. Nevertheless, the calculated coloration time of 660 s is in reasonable limits and the bleaching time with 11 s [132] especially reliable. Therefore, further investigations of 2D materials within this research field could lead to even better results and pave the way for a market-ripe production of electrochromic devices.

5.3 (Bio-) chemical sensors

One important feature of 2D materials is their high surface-area-to-volume ratio in a chemical reaction. Due to this, the reactivity and accessibility of the material is drastically increased compared to their 3D bulk counterparts. Other competitive nanostructured appearances as nanoparticles or nanorods admittively have an even greater specific surface area, but the nanosheets often allow better or further manageability regarding stability and applying on surfaces. One research field with 2D oxides as candidates is the detection and immobilization of specific molecules. Due to their large specific surface area, nanosheets are especially suitable for detecting small amounts of molecules, which allows detectors with high sensitivity. Two important detectors utilizing 2D oxides are gas sensors and biosensors.

Regarding gas sensors, several 2D metal oxides show highly promising results for electrochemic detectors. As the materials have to be selective towards the analyzed gas, different materials are required for different gases. Typically, the sensoring of toxical or harming compounds as H₂S, NO_x , ethanol or formaldehyde is asked for. For example, SnO₂/ZnO hetero-nanostructures built by nanosheets reach a detection limit of 10 ppb for H₂S at a relatively low working temperature of 100°C [53]. Since this is lower than the acceptable concentrations of at least 20–100 ppb for humans, it is within applicable ranges [133]. Other highly toxic gases are nitric oxide NO and nitrogen dioxide NO₂, which can be summed as NO_x . NiO as a p-type semiconductor with a high chemical stability shows reasonable sensing properties regarding NO₂. Hereby, the N-doping within nanostructures of NiO nanosheets leads to a two-fold higher gas responsitivity and sensitivity [134]. A single-crystal nanosheet of In₂O₃ was also used for detecting NO₂ and NO at room temperature recently [135]. Concerning NiO, the difference of nanosheet-assembled hierarchical nanoflowers and nanoneedle-assembled structures was observed. The nanosheets showed a substantially higher gas response, but faster gas responses and recovery were gained with the needles [136]. These results show the importance of the used nanostructures and their significant influence on the sensing properties. For ethanol, ZnO nanosheets show high sensitivities and can be directly grown upon the electrode via a hydrothermal method [82]. Detection ranges of 25–1000 ppm at a working temperature of 400°C and a high gas response could be reached coupled with fast responce recovery. It has been shown, that even the type of surface-exposed crystal facets may influence the gas sensing properties substantially [137]. In the proposed study of Xu et al. the sensitivity of ZnO nanosheets towards low ethanol concentrations of 50 ppm was doubled with exposed (0001) planes in comparison to (101̅0) planes and an even greater perfomance gap at 1 ppm. Even lower detection limits of 0.127 nM for ethanol were achieved by a CdO/ZnO/Yb₂O₃ ternary oxide nanosheet system [138]. Another compound worth mention is the WO₃, which also plays a significant role in electrochromic devices. It was shown, that Cr-doped WO₃ nanosheets can be used for formaldehyde sensing [139].

In terms of biosensing, 2D metal oxides just earned growing attention over the last few years with enormous potential discovering further appropriate materials [11]. First successful attempts have been made with MnO₂ nanosheets for sensing ochratoxin A (OTA) within single-strand desoxyribonucleic acid or cathepsin D (Cat D) using peptide chains [69]. For OTA, low quantification limits of 0.02 ng mL⁻¹ were found, which are competitive towards other detection methods in aqueous solutions. Further attempts with MnO₂ nanosheets regarding a fluorescence polarization-based detection of Ag⁺ ions were accomplished with low detection limits of 9.1 nM [140]. For a non-enzymatic glucose sensor, Co₃O₄ nanosheets with incorporated NiO or Ni(OH)₂ [141] and Ni(OH)₂/NiO nanosheets [142] showed low detection limits of 1.08 μM and 5 μM, respectively. Further 2D metal oxides with proven value regarding biosensors are MoO₃ nanosheets for field-effect-transistor based biosensing [143], ZrO₂ nanolayers for improving the reliability of impedimetric biosensors [86] and Yb₂O₃ nanosheets for the detection of the biomolecule urea [96].

5.4 Photoelectrochemistry

An increasingly important research field concerns regenerative energy usage as in photovoltaic devices. For the last decades, solar cells based on silicon were the most prominent ones, but remaining high manufacturing costs make the search for alternatives attractive.

One greatly promising alternative using metal oxides are perovskite solar cells (PSCs), which offer low costs at possibly high efficiencies. While the active perovskite is not built of oxides but rather organic metal halides as CH₃NH₃PbI₃, an electron transporting layer (ETL) is additionally required for loading the perovskite, blocking holes, transmitting light and transporting electrons [144]. Therefore, the ETL actively effects the performance and the stability of the device. Nanosheets with small thicknesses of less than 10 nm usually provide high transmissions of light, which also is used in electrochromic devices. Furthermore, metal oxides are currently state-of-the-art forming these ETLs [145], [146].

Utilizing TiO₂ nanosheets instead of other nanostructures led to power conversion efficiencies of 6.99% [147] and 10.7% [148]. By looking at the rapid evolution of the power conversion efficiencies from 3.8% to roughly 20% since their introduction in 2009 [146], [149], the date of research seems to be essential when comparing different results. The best efficiencies are currently gained by using thin films of mesoporous TiO₂ as the ETL, but the long-term stability lacks due to fast degradation of the devices under UV illumination. As an alternative, SnO₂ shows high chemical and UV illumination stability [150], but slightly lower efficiencies. By using SnO₂ nanosheet arrays, power conversion efficiencies of 17.36% were reached and even further increased to stable 18.00% when coupled with a C₆₀ interlayer [144]. While the stability is not yet sufficient for application purposes with maintaining 90% of the original efficiency after storing at ambient conditions for 500 h, it could be further improved in the following years.

5.5 Thermoelectrics

In several kinds of chemical processes temperatures higher than room temperature are required. Although, there is a significant amount of heat loss in most of these processes. To gain a better energy efficiency, the heat energy can be transferred into a more usable form of energy, e.g. electrical energy. A promising way of converting heat loss into an electrical current is accomplished by thermoelectric materials. The efficiencies of such materials are determined by the dimensionless figure-of-merit zT, as given in Equation 1.

The figure-of-merit increases with a high Seebeck coefficient α, high electrical conductivity σ and low entropy conductivity Λ [151]. The numerator in the fraction is also described as the power factor and should be observed separately for specific applications where high electrical current densities are required. The problem in adjusting these parameters is typically their intercorrelation. Various different oxides as Ca₃Co₄O₉, CaMnO₃, SrTiO₃ or In₂O₃ are interesting due to their thermoelectric properties [152]. Although, they mainly show rather poor zT values, mostly in the range of 0.1–0.5 if being unmodified [153]. An usual attempt at increasing the zT value is made by doping, which may effect all three parameters incorporated in zT. Although, doping itself did not increase the thermoelectric properties of the aforementioned oxide materials enough to gain a significant improvement. As a novel material resembling classical oxides, the oxyselenide BiCuSeO is worth mentioning [154]. It is a layered compound with insulating Bi₂O₂ layers and conducting Cu₂Se₂ layers with zT values of up to 1.4 if doped properly. Nevertheless, it is properly observed not a classical oxide and only resistant to surface oxidation up to 573 K and to decomposition reactions up to 773 K under air [155]. In comparison to this, non-oxide based thermoelectrics, if adjusted properly, gain higher zT values between 1 and 2 near room temperature. Though, oxides maintain significant benefits as stability at higher temperatures or the working mode at ambient conditions.

It has been suggested to increase the thermoelectric figure-of-merit of certain materials by more than one order of magnitude by preparing them in 2D quantum-well structures [156], [157], [158]. In fact, the extremely high zT of 2.4 [159] at room temperature for a one unit-cell thin SrTi_0.8Nb_0.2O₃ layer with a 2D electron gas shows huge possibilities in utilizing the quantum confinement effect for thermoelectrics. The quantum confinement is described by the restriction of the movement for charge carriers in-plane and the quantization of energy out-of plane [160]. This leads to an enhanced electronic density of states at the Fermi level, which subsequently increases the Seebeck coefficient drastically [161]. The confinement of the electrons additionally increases the in-plane electrical conductivity. Nevertheless, superlattice thin films are not easily applicable and the effective zT=0.24 of nine unit-cell layers including barrier layers between the unit cells mentioned by Ohta et al. [159] is still quite low. Therefore, Koumoto et al. [153] proposed to utilize the quantum confinement effect in 3D bulk nanoceramics. For example, nanometer-sized grains would affect the phonon confinement and boundary-scattering to decrease the lattice thermal conductivity significantly. In a “brick-and-mortar”-type SrTiO₃ with nanostructured grain-boundaries and grain interiors these effects could theoretically be utilized. For non-oxide Bi₂Se₃ nanosheets it was exemplary shown by actual measurements, that about 10 times higher zT values compared to the 3D bulk material can be reached [162], [163], wherefore the nanosheets were spark-plasma-sintered to 3D bulk pellets.

Only a few attempts have been made to investigate the thermoelectric properties of stacked 2D oxide nanosheets. In the case of the oxyselenide BiCuSeO, 3–4 nm thin nanosheets were synthesized by a solvothermal reaction and afterwards restacked to dense pellets via hot pressing [164]. The entropy conductivity could be lowered due to enhanced phonon scattering at the nanoscaled grains and the power factor was increased due to a better electrical conductivity. This led to overall improved zT value of 0.2 for undoped BiCuSeO at 722 K compared to undoped 3D bulk material. Since values of up to 1.4 can be reached with modified BiCuSeO [154], the nanosheet restacking could easily be further optimized by using doped nanosheets. Ultrathin films of Na_x CoO₂ were also investigated by calculations within the Green-Kubo theory with rather poor outcomes regarding beneficial properties. Due to weak coupling of CoO₂ sheets within the layered structure, the in-plane lattice thermal conductivity seems to be insensitive to the thickness of nanosheets and stacks thereof [165]. Actual ultrathin films of Na_x CoO₂ were investigated and showed a decrease of the electrical conductivity for films below 10 nm and a constant Seebeck coefficient, but not measured thermal conductivities, which also does not recommend the utilization of Na_x CoO₂ in terms of nanosheets [166]. But, restacking of nanosheets is only one possible way to exploit the quantum confinement effect in 3D bulk oxides. Another possibility are self-assembled nanostructures within a matrix. A promising example is a nanocomposite of heavily doped Ca₃Co₄O₉, which contains stacked 2D hetero-oxide building blocks and exhibits a relatively high 3D bulk zT value for oxides at 1073 K, which is state-of-the-art concerning high-temperature ranges [167]. Thus, the assembly of 2D nanosheets for creating similar nanocomposites with other oxides is quite promising. Overall it can be stated, that the future of oxide-based thermoelectrics lies within utilizing the quantum confinement effect of nanostructures.

5.6 (Photo-) catalysis

Generally, catalysts lower the activation energy of chemical processes without being consumed. Due to this, they do not intervene in the chemical equilibrium but influence the reaction kinetics. Hence, they are indispensable for some of the most important industrially-sized productions of chemicals. Interesting catalyst types for 2D oxides are primarily electrocatalysts and photocatalysts.

Electrocatalysts are e.g. used for electrochemical water splitting for H₂ fuel production. The efficiency of the process is essential to ensure it is economically reasonable. Since the storage of electricity is a far from satisfactorily resolved problem, overproduced capacities of wind turbines are especially interesting for the electrochemical production of H₂. However, the slow kinetics of the essential oxygen evolution reaction (OER) within this process is a big issue, which usually requires the use of electrocatalysts as RuO₂ or IrO₂. The price of these compounds diminishes the manufacturing process in applicability though. The utilization of different nanosheets as an electrocatalyst for the OER has shown substantial improvements in the most recent years. First of all, nanosheets of IrO₂ with a thickness of 0.7 nm exhibit a 6 times higher mass activity for the OER than nanoparticles [99]. Serious competitors are CoOOH nanosheets with a thickness of 1.4 nm, which show 2.4 times higher electrocatalytic activity than currently used IrO₂ and 20 times higher activity in comparison to 3D bulk CoOOH [168]. Further examples are Rh₂O₃ nanosheet assemblies or spinel-structured NiCo₂O₄ nanosheets, which all overcome currently used electrocatalysts for OER due to their 2D structure [94], [169]. Furthermore, the oxygen reduction reaction (ORR) can also be improved by using 2D metal oxide nanosheets like CoO₂, Ti_1−x O₂ or RuO₂ as additives to graphene [63]. Especially by using CoO₂, the electrocatalytic activity could be greatly increased while the long-term stability and selectivity were improved as well. The superior benefits of nanosheets compared to nanoparticles are assumed to be linked to the surface expansion in nanosheets. Despite this, the transport and storage of produced H₂ is also an issue. Recently, urea has been shown to be a promising candidate as a solid hydrogen carrier, which simplifies transport and storage of gaseous hydrogen. For gaining the hydrogen out of urea, electrocatalysts are necessary. For improving the current systems, mesoporous NiCo₂O₄ 2D nanosheets were directly synthesized on collectors, which led to a decrease of the onset potential in comparison to 3D bulk Ni(OH)₂ electrodes and enhancement of the urea oxidation current simultaneously [39].

An even more promising way to supply the necessary energy for the production of H₂ is the usage of solar energy. Although, low quantum efficiencies in the spectral region of visible light of most suitable materials are an issue. Generally, quantum efficiency is strongly dependent on the recombination rate of electrons and electron holes, namely excitons. A high recombination rate prevents the reactions of the excitons with the water and conclusively the quantum efficiency. Short lifetimes of less than 1 μs [170] for the excitons lead to a transporting problem, because the reactions occur at the surfaces. An approach to shorten the transporting ways to the surface is by using nanocrystals. Since defects within the crystals may trap the excitons, the crystallinity also effects the efficiency. Simply enhancing the reactive surface area would recommend nanoparticles as the most effective way to modify the materials. Nevertheless, nanoparticles have to deal with increasing durations for absorbing photons, which for a diameter of 1 nm already means 4 ms [170]. A solution to this problem could be the utilization of 2D photocatalysts as displayed in Figure 19.

Fig. 19:

Schematic comparison of (a) nanocrystals and (b) 2D crystals as photocatalysts. Due to the large sheet area of the 2D crystals, the photons can be absorbed in a reasonable time and catalyze the water splitting more effectively as nanoparticles. (c) H₂ production rates for differently treated KCa₂Nb₃O₁₀ compounds with methanol as a sacrificial agent. The non-exfoliated compounds and the unmodified nanosheets do only have small photocatalytic activities, but the Rh-doped or loaded nanosheets have a significantly better performance. The Rh(0.03)-doped nanosheets have the highest production rate while subsequently doping and loading diminish the performance slightly. Adapted with permission from [170]. Copyright 2014 American Chemical Society.

The large areas of 2D nanosheets provide reasonable photon absorbing times, while the transport distance of the excitons to the surface remains short as in nanoparticles due to the small sheet thickness. Suitable semiconducting nanosheet oxides with a wide band gap are e.g. exfoliated Ruddlesden-Popper phases as KCa₂Nb₃O₁₀ or the Aurivillius phase Bi₂MoO₆ [46], [171]. Generally, there are two different options with co-catalysts and doping to enhance the photocatalytic activity of catalysts. While loading the surface with co-catalysts reliably improves the performance, doping usually does not a have huge impact in comparison [170]. The reason for this is the localization of the dopants within the catalyst and the resulting distance to the surface. In ultrathin 2D nanosheets the dopants are close to the surface, which means a direct involvement in catalytic activities of the dopants and a comparable improvement as caused by co-catalysts. The impact of doped but non-exfoliated compounds result in low photocatalytic activities confirming the influence of the 2D structure [171]. The adjustment of the doping amount is also important to gain high quantum efficiencies for KCa₂Nb₃O₁₀ nanosheets as it is shown in Figure 20.

Fig. 20:

Nanosheets as a photocatalyst for hydrogen production: (a) Amount of photocatalytically produced H₂ dependend on the Rh doping amount x in Ca₂Nb₃O₁₀⁻ 2D nanosheets. (b) Quantum efficiency for Rh(x=0.03)-doped nanosheets regarding the H₂ production and the absorbed wavelengths. Reproduced with permission from [171]. Copyright 2011 American Chemical Society.

By this, maximum H₂ production rates of 400 μmol h⁻¹ with 5 mg catalyst were reached. An additional possibility for further improvements of the photocatalytic performance may be the addition of non-expensive co-catalysts. It was shown, that loading niobate nanosheets with small amounts of MoS₂ and graphene with a molar ratio of 99.0:0.5:0.5 leads to significantly increased H₂ production rates [87]. As catalyst for the O₂ production, hetero-structured Bi₂MoO₆ nanosheets coupled with TiO₂ nanobelts were promoted [87]. Oxygen production rates of 0.668 mmol h⁻¹g⁻¹ were gained hereby. The remaining problem of these photocatalysts is their low activity under visible light irradiation. N-doped TiO₂ [172] and NbO_x [173] nanosheets improve the absorption under visible irradiation, but the absorption maximum continues to be in the ultraviolet (UV) spectrum and the doping amount is limited. Another problem for nanosheets in particular is attributed to their synthesis with ion intercalation. Organic intercalates may remain on the nanosheet surface even after washing and hinder the photocatalytic activity, which requires additional removing steps as UV light irradiation for several hours [170]. Therefore, alternative exfoliation techniques without the need of organic intercalates may be more appropriate.

Another application field for photocatalysts is the photodegradation of hazardous organics for water purification. Unmodified nanosheets as SnNb₂O₆ show higher selectivity and photocatalytic activity for positively charged pollutants in particular in comparison to the 3D bulk SnNb₂O₆, TiO₂ or N-doped TiO₂ [174]. Most recently, even better results were gained by creating 2D/2D heterojunctions between different nanosheets [175]. For niobate nanosheets Ca₂Nb₃O₁₀⁻ other oxide nanosheets as WO₃ or non-oxides as g-C₃N₄ are proven candidates. For both combinations, the photodegradation of tetracycline hydrochloride under simulated sunlight was investigated and both times led to a significantly enhanced perfomance due to improved charge carrier separations and transfers in strongly interacting hetero-interfaces. In terms of adding 20% WO₃, the optimum efficiency was gained with 5.1-fold and two-fold higher photocatalytic activity compared to WO₃ and Ca₂Nb₃O₁₀⁻ single nanosheets [176]. Similar results were accomplished with the combination of 20% g-C₃N₄, which led to 6.6 and 1.8 times higher degradation rates compared to bare g-C₃N₄ or Ca₂Nb₃O₁₀⁻ single nanosheets [175].

5.7 Separation technologies

A more exotic application field for 2D oxides comprises separation technologies as membranes or filters. While ceramic-based membranes maintain several disadvantages as brittleness or high costs, graphene oxide (GO) as a carbon-based material showed promising separation characteristics in the recent years [177], [178], [179]. Zeolites as a group of microporous aluminosilicates oxides are well-known for their applicability in separation technologies like membranes. The formation of 2D nanosheets with zeolites is also possible and shows auspicious results [180]. Because they are only oxide-related, further consideration is not envisaged within this review.

Several techniques for the arrangement of nanosheets to usable membranes have been tested as drop-casting, spray- and spin-coating, Langmuir-Blodgett method and vacuum filtration [178]. Vacuum filtration has also been shown to be suitable for 2D compounds as MXenes, which are mainly characterized as ultrathin layers composed of transition metal carbides and nitrides. Separating membranes built of MXenes recently showed substantially enhanced performances regarding permeability and selectivity as shown in Figure 21 [181], [182]. The nanosheet-composed membranes allow the separation of molecules with respect to their size, because larger gases as CO₂ have a longer transporting way moving along the nanosheet grains compared to smaller H₂.

Fig. 21:

Separation performance of graphene oxide (GO) membranes compared to other state-of-the-art membranes with highlighting MXene membranes. Abbrevations represent metal-organic-frameworks (MOFs), zeolitic-imidazolate-frameworks (ZIFs), compact-membrane-systems (CMS) and polybenzimidazole (PBI). Reproduced with permission from [181]. Copyright 2018 Nature Springer. Creative Commons License CC BY 4.0.

In comparison of GO-based to MXene-based membranes, the latter reaches higher permeabilities, but with lower selectivities. The disordered assembling of the nanosheets hindered better selectivities, which could be further improved by another assembling route.

2D metal oxides as membranes play only a minor role yet, because other materials show usually better performance. Nevertheless, exemplarily NbO nanosheets were assembled to a membrane by vacuum filtration and showed a high stability. The creation of nanochannels within the membranes allows the usage for high-flux nanofiltration as salts are retained [88].