Report

A liquid metal reaction environment for the room-temperature synthesis of atomically thin metal oxides

See allHide authors and affiliations

Science  20 Oct 2017:
Vol. 358, Issue 6361, pp. 332-335
DOI: 10.1126/science.aao4249

Expanding the world of 2D materials

Two-dimensional (2D) materials have a wide variety of potential applications in the electronics industry. However, certain compositions of 2D materials are difficult to obtain owing to the challenges in exfoliating thin sheets from bulk crystals. Zavabeti et al. exploited liquid metals to synthesize 2D Ga2O3, HfO2, Gd2O3, and Al2O3. The 2D sheets appear as a surface layer in gallium-based liquid metals after the Hf, Gd, or Al is dissolved into the bulk alloy. The 2D oxide that appears on the surface is the oxide with the lowest energy, suggesting that it should be possible to make other 2D oxides by using the same process.

Science, this issue p. 332

Abstract

Two-dimensional (2D) oxides have a wide variety of applications in electronics and other technologies. However, many oxides are not easy to synthesize as 2D materials through conventional methods. We used nontoxic eutectic gallium-based alloys as a reaction solvent and co-alloyed desired metals into the melt. On the basis of thermodynamic considerations, we predicted the composition of the self-limiting interfacial oxide. We isolated the surface oxide as a 2D layer, either on substrates or in suspension. This enabled us to produce extremely thin subnanometer layers of HfO2, Al2O3, and Gd2O3. The liquid metal–based reaction route can be used to create 2D materials that were previously inaccessible with preexisting methods. The work introduces room-temperature liquid metals as a reaction environment for the synthesis of oxide nanomaterials with low dimensionality.

Room-temperature liquid metals have a number of interesting surface and bulk properties that make them intriguing for a variety of engineering applications, including flexible electronics and microfluidics (1). Gallium-based eutectic alloys such as EGaIn (containing gallium and indium) and galinstan (containing gallium, indium, and tin) are liquid at room temperature, are nontoxic, and feature metallic bonds. Unlike molecular and ionic liquids, liquid metals are rarely used as reaction solvents. Here, we used eutectic gallium melts as a reaction environment to create a variety of low-dimensional metal oxides.

Most metals, including gallium-based alloys (2, 3), feature a self-limiting thin oxide layer under ambient conditions at the metal-air interface (4). This atomically thin interfacial oxide is considered a naturally occurring two-dimensional (2D) material. The concept of using a liquid metal as a reaction environment is based on the observation that the self-limiting oxide layers of gallium alloys, such as EGaIn and galinstan, are exclusively composed of gallium oxide, despite indium content of 22 to 25 weight percent (wt %) and tin content of as much as 10 wt % in these alloys (2, 3). Co-alloying suitable reactive metals allows us to form the co-alloyed metal oxides at the metal-air interface. We identified suitable co-alloyed metals on the basis of thermodynamic considerations, which require that the composition of the surface oxide is determined by the reactivity of the individual metals within the melt. Thus, the oxide resulting in the greatest reduction of Gibbs free energy (ΔGf) will dominate the surface (Fig. 1A). If room-temperature liquid gallium alloys are the solvent, all lanthanide oxides and a sizable portion of the transition metal and post–transition metal oxides should be accessible as 2D nanostructures (table S1) (5, 6).

Fig. 1 Fundamental principles and synthetic approach.

(A) Gibbs free energy of formation for selected metal oxides (5, 6). Oxides to the right of the red dashed line are expected to dominate the interface. (B) A cross-sectional diagram of a liquid metal droplet, with possible crystal structures of thin layers of HfO2, Al2O3, and Gd2O3 as indicated. (C) Schematic representation of the van der Waals exfoliation technique. The pristine liquid metal droplet is first exposed to an oxygen-containing environment. Touching the liquid metal with a suitable substrate allows transfer of the interfacial oxide layer. An optical image is shown at the right. (D) Schematic representation of the gas injection method (left), photographs of the bubble bursting through the liquid metal (center), and an optical image of the resulting sheets drop-cast onto a SiO2/Si wafer (right). See movie S1.

We synthesized three different 2D oxides using galinstan as a reaction environment. To show the versatility of the method, we selected oxides of a transition metal (HfO2), a post–transition metal (Al2O3), and a rare earth metal (Gd2O3) with high melting points. Of these oxides, hafnium oxide is particularly important because of its use as an insulating oxide in the electronics industry. For galinstan, we expected gallium oxide to form on the metal surface because it features the most negative ΔGf of the possible oxides (3, 5). We alloyed ~1 wt % of elemental hafnium, aluminum, or gadolinium into the liquid metal, which served as the precursors for the formation of their respective oxides (fig. S1) (7). These oxides fulfilled the requirement of featuring a lower ΔGf than gallium oxide, and as a result we expected them to form surface oxides on their respective alloys (Fig. 1, A and B) (5). All three oxides that we synthesized as 2D materials have nonstratified crystal structures and cannot be obtained using standard exfoliation techniques (8, 9). Nonetheless, either the investigated oxides or closely related structures have been predicted to be stable as 2D monolayers (1012). We prepared alloys containing ~1 wt % copper and silver for control experiments.

We developed two synthesis approaches for the isolation of the surface oxide. The first method was a van der Waals (vdW) exfoliation technique similar to the method for obtaining monolayer graphene pioneered by Novoselov et al. (Fig. 1C) (13). The technique entails touching the liquid metal droplet with a solid substrate. The liquid nature of the parent metal results in the absence of macroscopic forces between the metal and its oxide skin, allowing clean delamination (14). This technique is suitable for the production of high-quality thin oxide sheets on substrates. The second technique relies on the injection of pressurized air into the liquid metal (Fig. 1D and movie S1) (7). We expected the metal oxide to form rapidly on the inside of air bubbles as they rose through the liquid metal. The released air bubbles passed through a layer of deionized water that we had placed above the liquid metal, allowing the produced oxide sheets to be dispersed into an aqueous suspension. This technique is highly scalable and is therefore suitable for creating high-yield suspensions of the target oxide nanosheets.

To directly compare the different oxide nanosheets, we used the exfoliation method on all the alloys we investigated. The optical images show the large lateral dimensions of the products (Fig. 1, C and D). To further characterize the nanosheets, we used atomic force microscopy (AFM) on samples deposited on top of SiO2/Si wafers and transmission electron microscopy (TEM) on samples directly deposited onto TEM grids (Fig. 2). Materials produced by the exfoliation method featured a smooth appearance under AFM imaging. Each 2D oxide sheet we produced had a different thickness: 2.8 nm for gallium oxide, 0.6 nm for hafnium oxide, 1.1 nm for aluminum oxide, and 0.5 nm for gadolinium oxide. The ultrathin nature of the oxides is apparent from the translucent appearance and occasional wrinkles in our TEM images (Fig. 2).

Fig. 2 Characterization of materials derived from the exfoliation method.

Left: AFM images, with thickness profile (inset) determined at the red line as indicated. Center: TEM characterization, with selected-area electron diffraction (SAED) (top inset) and HRTEM images (bottom inset; scale bar, 0.5 nm). Right: Elemental composition determined by XPS (2429). (A) Results obtained from a eutectic gallium-indium-tin alloy. (B to D) Alloys containing approximately 1% of added hafnium, aluminum, and gadolinium, respectively. See (7) for characterization of the alloys before use. The lattice parameters in SAED and HRTEM images were indexed using literature reports (1517). The lack of crystallinity in the gallium oxide sample might be beam-induced. See (7) for XPS spectra used to determine the oxide composition. The sample derived from pure galinstan features metallic inclusions, which are visible as dark dots and elongated nodules. The other materials feature no inclusions. The lateral dimensions of the 2D sheets are extraordinarily large and exceed the AFM field of view.

We found that the nanosheets derived from unalloyed galinstan were amorphous when characterized by electron diffraction and high-resolution TEM (HRTEM). By contrast, the 2D oxides produced from alloys containing hafnium, aluminum, and gadolinium were polycrystalline, with characteristic lattice parameters of m-HfO2, α-Al2O3, and cubic Gd2O3 (Fig. 2) (1517). Using x-ray photoemission spectroscopy (XPS), we found that the exfoliated oxide of unalloyed galinstan is exclusively composed of gallium oxides (mostly Ga2O3 with a small Ga2O content) (2, 3). The 2D materials derived from our alloys were composed entirely of the oxide of the added element (Fig. 2 and fig. S2) (7). Our control experiments with copper and silver alloys resulted in nanosheets predominantly composed of gallium oxide (fig. S3) (7). The results confirm our hypothesis that the oxide with the lowest ΔGf dominates the interfacial surface oxide.

Galinstan and hafnium-containing galinstan were investigated using the gas injection synthesis method. The aluminum- and gadolinium-containing alloys visibly reacted with water and were not investigated further. Replacing water with an inert solvent was expected to enable the synthesis of aluminum and gadolinium oxide with this gas injection method. The materials we derived from the gas injection synthesis method were similar in appearance to the oxides we produced with the exfoliation method (Fig. 3). The material synthesized using unalloyed galinstan had a thickness of ~5.2 nm, about double that of the exfoliation method. We found that the HfO2 nanosheets had a thickness of 0.5 nm, which is similar to the thickness of exfoliated HfO2. The gallium oxide nanosheets contained small, spherical metallic inclusions that we discovered during TEM imaging. We confirmed the presence of ~15-nm spherical inclusions with high-resolution AFM and TEM imaging (fig. S4) (7). The nanosheets have low crystallinity, which we examined using HRTEM and electron diffraction measurements. This indicates that the gas injection synthesis may have a tendency to create amorphous oxides. The short reaction time, attributable to a limited residence time for the gas bubble within the liquid metal, may be the origin of the poor crystallinity. The reaction time frame available for the exfoliation method is considerably longer, because the metal droplet may rest for several minutes prior to the vdW transfer, allowing for the reorganization of the crystal structure within the interfacial oxide layer.

Fig. 3 Characterization of materials derived from the gas injection method.

Left: AFM images, with thickness profile (inset) determined at the red line as indicated. Center: TEM characterization, with SAED (top inset) and HRTEM images (bottom inset; scale bar, 0.5 nm). Right: Raman spectra of the resulting oxides. (A) Results obtained from a eutectic gallium-indium-tin alloy. (B) Alloy containing approximately 1% of added hafnium. The Raman spectra match well with literature reports for Ga2O3 (A) (30) and HfO2 (B) (15).

Using Raman spectroscopy, we found that the nanosheets derived from the gas injection method are also predominantly composed of the oxide with the lowest ΔGf (Fig. 3). The successful gas injection synthesis of both gallium and hafnium oxides demonstrates the suitability of the approach for the high-throughput synthesis of 2D oxides for applications where high crystallinity is not a primary concern. The 2D materials we synthesized hold promise for applications in energy storage, such as supercapacitors and batteries that require large quantities of materials with high ratios of surface area to volume (18).

One major application of the materials we developed is their use as ultrathin insulator dielectrics for the fabrication of field-effect transistors. HfO2 is the material of choice for the electronics industry, with a relative dielectric constant of >20 and a band gap of >5 eV (19). Because of the lack of a stratified crystal structure and the resulting absence of amenable exfoliation techniques, HfO2-based dielectrics are traditionally deposited using atomic layer deposition or other chemical and physical vapor-phase deposition techniques (19). These techniques rely on an island-mediated film growth mechanism that results in a minimum film thickness of 3 to 5 nm (20). Our liquid metal–derived HfO2 had a minimum thickness of approximately 0.5 nm.

We directly deposited 2D HfO2 onto a wafer coated with Pt to evaluate our exfoliation methodology for the synthesis of dielectrics. Using conductive AFM characterization, we found that the deposited 2D HfO2 sheet is completely insulating and pinhole-free, despite being only ~0.5 to 0.6 nm thick (Fig. 4, A and B) (7). The breakdown electric field, defined as the field at which the current rises above the noise level, is ~5.3 GV cm−1, as determined by current-voltage (I-V) characterization of the 2D nanosheet. This value is three orders of magnitude higher than the breakthrough field for chemical vapor deposition (CVD)–grown multilayer h-BN (21, 22), highlighting the excellent quality of the dielectric. Fitting the I-V curve to the Schottky emission model allowed us to determine a dielectric constant of ~39 for the oxide sheets (7). We analyzed the low-loss region of the electron energy loss spectrum (EELS) of exfoliated ultrathin HfO2 sheets and found that the band gap of the material is of a direct nature with a ~6-eV gap (Fig. 4D) (7, 23).

Fig. 4 Characterization of HfO2 as a dielectric.

(A) Schematic of the peak force tunneling AFM (PF-TUNA) setup. (B) AFM height (top) and current (bottom) maps for the edge region of an HfO2 sample directly deposited onto a Pt-coated wafer by the exfoliation method. Both maps are 300 nm wide. The profiles at the right correspond to the regions indicated by the red lines. The HfO2 and Pt sides are labeled. (C) Current-voltage curve measured through the HfO2 layer. Inset: Fit to the Schottky emission model and the determined dielectric constant (7). (D) Plot of the low-loss EELS spectrum, which provides an estimate of the band gap. Inset: Analysis of the nature of the band gap, indicating a direct gap (7).

Our findings show that oxide layers on liquid metals can be manipulated by selecting appropriate alloying elements on the basis of the Gibbs free energy for oxide formation. The two methods we used to recover the 2D nanosheets are both scalable, do not require complex equipment, and provide nanosheets either directly deposited onto substrates or as an aqueous suspension. The liquid metal serves as a solvent during the reaction. Our methodology facilitated the isolation of atomically thin layers of metal oxides that do not naturally present themselves as stratified systems, providing a synthetic pathway toward an important class of 2D materials that was previously inaccessible. The underlying principles suggest that this should apply to a sizable fraction of metals, giving access to 2D crystals of many transition metal, post–transition metal, and rare earth metal oxides. Several of these metal oxides are of exceptional importance because of their various electronic, magnetic, optical, and catalytic properties.

Supplementary Materials

www.sciencemag.org/content/358/6361/332/suppl/DC1

Materials and Methods

Supplementary Text

Figs. S1 to S5

Table S1

Movie S1

References (3136)

References and Notes

  1. See supplementary materials.
  2. Acknowledgments: Supported by Australian Research Council Centre of Excellence FLEET (CE170100039). We thank RMIT University’s Microscopy and Microanalysis Facility, a linked laboratory of the Australian Microscopy and Microanalysis Research Facility, for scientific and technical assistance, and the RMIT MicroNano Research Facility for associated technical support. Additional data are available in the supplementary materials. The authors declare no conflict of interest.
View Abstract

Navigate This Article