Huge Absorption Edge Blue shifts of Layered α ... - ACS Publications

May 15, 2018 - Herein, we report experimental observations of huge blue shifts in the ... The much larger band gap changes upon thickness reductions o...
0 downloads 0 Views 10MB Size
Article Cite This: J. Phys. Chem. C 2018, 122, 12122−12130

pubs.acs.org/JPCC

Huge Absorption Edge Blue shifts of Layered α‑MoO3 Crystals upon Thickness Reduction Approaching 2D Nanosheets Hongfei Liu,* Coryl J. J. Lee, Yunjiang Jin, Jing Yang, Chengyuan Yang, and Dongzhi Chi Institute of Materials Research and Engineering (IMRE), Agency for Science, Technology and Research (A*STAR), 2 Fusionopolis Way, Singapore 138634, Singapore

J. Phys. Chem. C 2018.122:12122-12130. Downloaded from pubs.acs.org by KAROLINSKA INST on 08/17/18. For personal use only.

S Supporting Information *

ABSTRACT: Recent theoretical studies suggest none or minor changes in the band gap of two-dimensional (2D) α-MoO3 nanosheets as compared with that of the bulk because of the weak interlayer electronic interactions. Unfortunately, this suggestion is lacking positive support in the literature. Herein, we report experimental observations of huge blue shifts in the absorption edge of layered α-MoO3 as its thickness t is reduced approaching atomic layers. When t > 10 nm, every order of magnitude of thickness reduction gives rise to a blue shift of ∼0.29 eV without causing any Raman mode shifts. This blue shift, in terms of finite difference time domain calculations, is attributable to optical interferences at the crystal surfaces. However, when t is further reduced below ∼10 nm, an even larger blue shift, accompanied by a mode softening of the most strengthened Mo−O−Mo stretching phonon (Ag), has been observed. This observation is consistent with those of 2D α-MoO3 nanosheets produced by aqueous exfoliations and, based on our calculations of the electronic structures, can be explained as anisotropic inplane strain relaxations/redistributions. A gas-phase layer-by-layer etching of the layered α-MoO3 single crystals has also been demonstrated for consequent fabrications of novel electronic devices, as well as their integrations, based on α-MoO3 and other 2D nanosheets.

1. INTRODUCTION Orthorhombic MoO3, that is, α-MoO3 (JCPDF: 05-0508) with conventional cell axes and angles of a = 3.9628 Å, b = 13.855 Å, c = 3.6964 Å, and α = β = γ = 90°,1 has been widely investigated for applications in catalytic chemistry as well as in electronic/ optoelectronic devices. In recent years, α-MoO3 has been attracting increasing research interest concerning its twodimensional (2D) applications driven by ever-growing demand for transparent, flexible, and wearable electronics as well as energy storage devices.2−6 The van der Waals lattice structure of α-MoO3 along its [010] crystal axis makes it easily exfoliated into atomic layers. Carrier mobilities up to 1100 cm2/V s have been obtained for α-MoO3 with its van der Waals lattice intercalated by hydrogen, 7 which may have important consequences when fabricating high-speed nanoelectronics based on 2D materials.8,9 Compared with the tremendous research effort given to transition metal dichalcogenide (TMDC)-based semiconducting 2D materials such as MoS2 and WS2,10−15 researches on 2D MoO3 are still at the beginning stage.4,9,16 It has been wellknown that when reducing the thickness of TMDC crystals to a few atomic layers, their electronic band gap increases and eventually converts from indirect to direct at the thickness of single layer because of quantum confinements.10,17 Similar thickness reduction-induced band gap increases have also been © 2018 American Chemical Society

observed and well understood in 2D black-phosphorous (BP) and -arsenic (BA).18−20 In comparison, recent theoretical calculations suggest a minor or absence of such layer thicknessdependent band gap changes for 2D MoO3 because of its weaker van der Waals interactions than those of TMDCs and BP/BA.21 However, experimental evidence that can support this theoretical suggestion is missing in the literature.21,22 On the contrary, the absorption spectra collected from α-MoO3 nanosheets prepared by bovine serum albumin (BSA)-assisted aqueous exfoliations or by oxidation of few-layer MoS2 have shown remarkable blue shifts in the absorption edge with the decrease in the crystal thickness.4,23,24 On the other hand, in contrast to the absence of thickness reduction-induced phonon shifts of 2D α-MoO3 claimed by Wang et al.,22 a red shift has been observed by Cai et al. for the Raman feature at ∼818 cm−1 of 2D α-MoO3 as compared with that of α-MoO3 bulk, and the red shift was attributed to an in-plane lattice expansion of the few-layer α-MoO3.3 In this paper, we provide direct evidence that the optical band gap energy, EOpt, derived from the absorption edge of αMoO3 crystals synthesized by thermal vapor transport (TVT), Received: April 9, 2018 Revised: May 10, 2018 Published: May 15, 2018 12122

DOI: 10.1021/acs.jpcc.8b03340 J. Phys. Chem. C 2018, 122, 12122−12130

Article

The Journal of Physical Chemistry C

3. RESULTS AND DISCUSSION Figure 1 summarizes the synthesis [Figure 1a], morphological properties [Figure 1b−d], and structural properties [Figure

increases by up to 1.5 eV as the layer thickness decreases from a few microns approaching a few nanometers. The band gap increments have further been discussed in terms of two mechanisms. One is the interference effect caused by light interactions with the layered crystal at the surfaces, which appears more at larger layer thickness (i.e., t > 10 nm) and the other is related to lattice relaxations that occur at the thickness of a few atomic layers. The much larger band gap changes upon thickness reductions of the obtained 2D α-MoO3 than those reported by theoretical calculations are attributable to anisotropic in-plane strain relaxations/redistributions that manifested themselves as the formation of regular [001]oriented microcracks. We also demonstrate that high-quality αMoO3 belt crystals can be thinned down via a gas-phase layerby-layer etching, which provides an alternative routine to prepare 2D α-MoO3 nanosheets for their fundamental studies and device applications.

2. MATERIALS AND METHOD The α-MoO3 crystals, in the form of layered belts, were synthesized by TVT in a quartz tube reactor of a tube-furnace without employing intentional substrates. The source material, that is, MoO3 powder, was loaded in an alumina crucible located at the zone center. At 1000 °C, the vaporized MoO3 was carried by pure nitrogen, transferred to the downstream areas, and crystallized in the form of layered belts at the areas of ≤600 °C.4 The belt crystals are collected from the inner wall of the quartz tube reactor and disassembled on sapphire substrates for optical studies. Because of their belt shapes, the disassembled α-MoO3 crystals tend to lie on their belt surfaces, that is, with their (010) atomic planes lying on the substrates. To thin down the disassembled α-MoO3 belt crystals, argonplasma (AP) and oxygen-plasma (OP) treatments as well as gas-phase XeF2- and HF-etching were carried out at room temperature. The details of the crystal disassembling, plasma treatments, and gas-phase etching are supplied in the Supporting Information. A single-crystal X-ray diffractometer (SCXRD, KAPPA APEX) was used to identify the “single-crystallinity” characteristic of an individual α-MoO3 belt, whereas a general-area detector diffraction system (GADDS, Bruker-D8) was used to identify the crystal phases of gathered α-MoO3 belts that were pressed on a sample holder. Optical studies, that is, absorbance and Raman scattering, were carried out with the incident light along the surface normal direction of the individual α-MoO3 belt crystals. Their thicknesses were estimated from the interference fringes in the absorbance spectra, employing the refractive index of α-MoO3 available in the literature. The estimated values were further validated by a step-profiler and/ or atomic force microscopy (AFM). The absorbance spectra were collected using an UV−vis−NIR micro-spectrophotometer (QDI 2010, CRAIC Technologies), whereas the Raman spectra were collected in a backscattering configuration under a confocal micro-Raman system (Witec alpha 300 system equipped with a 532 nm wavelength argon ion laser). Morphological and spectral evolutions of the α-MoO3 belts upon plasma treatments and gas-phase chemical etching were also evaluated by optical microscopy, scanning electron microscopy (SEM), AFM, absorbance, and Raman scattering spectroscopies.

Figure 1. Synthesis and morphological and structural properties of αMoO3 single crystals. (a) Schematic diagram of the TVT setup; (b) typical SEM image recorded from the as-grown belt crystals lying on their belt surfaces; (c,d) microphotographs taken from the belt crystals disassembled on sapphire substrates with their color variations corresponding to their thickness changes; (e) XRD patterns collected by SCXRD and GADDS systems from a single belt and gathered belts, respectively, their comparisons indicate that the belts are typically lying on their (010) atomic planes; and (f) lattice structure calculated from the SCXRD measurements, which is indeed α-MoO3 that consists of “monolayers” stacking along their [010]-axis via van der Waals forces, the SCXRD measurements also revealed that the length direction of the belts is along their [001]-axis.

1e,f] of the MoO3 belt crystals. The typical SEM image recorded from the gathered crystals [Figure 1b] and the microphotographs recorded from the disassembled crystals [Figure 1c,d] show that the MoO3 crystals are of belt structures and they prefer to lie with their belt surfaces parallel to that of the substrate. The belt-to-belt contrasts in Figure 1c and the location-to-location ones of the single belt in Figure 1d provide clues for thickness comparisons, especially those of the single belt that have the advantages of being verified by a step-profiler and/or AFM. Figure 1e presents the XRD patterns collected by SCXRD from a single belt and those collected by GADDS from gathered belts [see Figure 1b]; their comparisons indicate that the MoO3 belts are typically lying on their (010) atomic planes. Figure 1f shows the lattice structure worked out from the SCXRD measurements, which is indeed α-MoO3 that consists of “monolayers”, stacking along their [010]-axis via van der Waals forces. The lattice parameters are a = 3.9527 Å, b = 13.8146 Å, c = 3.6903 Å, and α = β = γ = 90°. Their comparisons with the cell parameters reported in JCPDF 050508 reveal that the obtained α-MoO3 belt crystals are compressively stressed in their lattices; the inherent lattice strains are εa = −0.25%, εb = −0.29%, and εc = −0.17%, that is, anisotropic. Because of the parallel lying and the feasible thickness measurements of the disassembled α-MoO3 belt crystals, their absorption and Raman spectra were collected to address their thickness dependencies. Figure 2a presents the absorbance 12123

DOI: 10.1021/acs.jpcc.8b03340 J. Phys. Chem. C 2018, 122, 12122−12130

Article

The Journal of Physical Chemistry C

Figure 2. Absorption, Raman scattering, and EDX spectra collected from the α-MoO3 belt crystals as a function of their thicknesses. (a) Absorbance spectra; (b) absorption spectra in the form of (abs. × hν)2; (c) optical band gap energy EOpt derived from the absorption edge of (abs. × hν)2; (d) Raman spectra; (e) normalized Raman spectra addressing the constant intensity ratio of I282/I290; and (f) EDX spectra collected with reduced electron-beam accelerated voltage, the inset are the survey spectra showing the absence of aluminum from the sapphire substrate.

spectra collected from the belt crystals with their thickness monotonically decreasing from A to G. The fringes below the absorption edges originate due to light interferences at the crystal surfaces, from which the crystal thickness can be estimated with the optical refractive index of α-MoO3 available from the literature.25 The estimated thicknesses were further validated by a step-profiler and/or AFM measurements, especially for the thinner α-MoO3 layers, for example, samples F and G, where the interference fringes are not enough for the thickness estimations. One sees that the absorption edge in Figure 2a shifts to larger photon energies as the layer thickness decreases. To quantify the optical band gap energies (EOpt) from the absorption edges, we have replotted the absorbance spectra in the form of (abs. × hν)2 as a function of photon energy, hν, in Figure 2b, where the EOpt can be obtained by linear fitting to the increasing edge of the absorption features for the individual α-MoO3 belts; they are plotted as a function of their respective thicknesses in Figure 2c. It is seen that the EOpt increases by ∼0.29 eV when the layer thickness decreases by one order of magnitude in the range of 10−104 nm. In this thickness range, the quantum confinement effect, in terms of 2D-TMDCs, -BP/BA, and oxide quantum structures (e.g., MgO/ZnO/MgO quantum wells),10,17−19,26 is negligible and plays a minor role in the thickness reduction-induced EOpt blue shifts of the α-MoO3 belt crystals observed in Figure 2c. Presented in Figure 2d are the Raman spectra collected from the α-MoO3 samples A−G at the locations where the absorbance spectra were measured. One sees that as the layer thickness decreases from samples A to G, the Raman features monotonically decrease in their intensities, while their mode frequencies do not show any distinguishable shifts. These observations indicate that the lattice strains of the α-MoO3

belts are independent of their thickness in the studied range. Figure 2e presents a detailed spectral comparison, addressing the B2g mode, at about 282 cm−1, of the belt crystals A−G. It has been known that a B3g Raman active mode, at about 290 cm−1, is usually enhanced in its intensity in oxygen-deficient αMoO3 crystals. The intensity ratio of I282/I290, which has a linear relationship with respect to the O-to-Mo ratio, has been employed to evaluate the oxygen deficiencies of α-MoO3.27−29 In this regard, the comparisons in Figure 2e indicate an absence of detectable variations in the oxygen deficiencies if they are present in the studied α-MoO3 belt crystals. Energy-dispersive X-ray spectroscopy (EDX) installed in the SEM chamber was further employed to evaluate the stoichiometric variations among the α-MoO3 crystals. To minimize the influence of the oxide substrate, we have reduced the acceleration voltage of the electron beam to 5 keV from that of 15 keV for normal operations, so that the penetration depth of the electron beam in α-MoO3 is significantly reduced.5 The EDX spectra collected from the α-MoO3 belt crystals A−E at the locations where the absorbance and Raman spectra were collected are presented in Figure 2f with the spectra normalized to the peak intensity of oxygen; their survey spectra, shown in the inset, indicate that the effect of substrate is indeed excluded for the belts thicker than ∼100 nm. These comparisons reveal that the α-MoO3 belt crystals are oxygen deficient (i.e., MoO2.93) but their composition stoichiometry is constant and independent of layer thicknesses, supporting the Raman spectroscopy results in Figure 2e. A combination of the Raman spectroscopy and the EDX measurements provides evidence that both the lattice strain and the composition stoichiometry of the studied α-MoO3 belt crystals are constant and independent of their layer thicknesses. 12124

DOI: 10.1021/acs.jpcc.8b03340 J. Phys. Chem. C 2018, 122, 12122−12130

Article

The Journal of Physical Chemistry C

Figure 3. Morphological (a−l) and absorption spectral (n,o) evolutions of a SD (a−f) and a DF (g−l) α-MoO3 belt crystals upon sequential oxygenand argon-plasma (referred to as OP and AP) treatments. (m) Crystal orientations of the crystals in (a−l). The insets in (n) show a microphotograph and its schematic diagram of the plasma treatment-induced [001]-oriented linear structures on the surface of the SD crystals. (p,q) Schematic diagrams showing the SD and DF α-MoO3 belt crystals.

As a result, they play a minor role in the layer-thicknessdependent EOpt shifts in Figure 2c. Such thickness-dependent EOpt shifts have also been observed in other semiconductors, for example, γ-In2Se3, but were attributed to the quantum confinement effect because of the poly-crystallinity and limited grain sizes of the layered materials (Supporting Information, Figure S1).30,31 To understand the EOpt shifts of our singlecrystal α-MoO3 belts, we have carried out transmission/ absorption calculations for layered α-MoO3 on sapphire using the finite difference time domain (FDTD) method with the optical refractive index available from the literature (Supporting Information, Figure S2).25 The FDTD calculations reveal a clear semilinear relationship of EOpt as a function of the layer thickness in the range of 103−104 nm. However, the increase of EOpt tends to saturate as the layer thickness decreases below 103 nm. Taking the scattered refractive index values of α-MoO3 in the literature studies into account, we can attribute the observed thickness reduction-induced EOpt blue shifts to light interferences that occurred at the surfaces of the belt crystals rather than changes in the electronic structures of α-MoO3. Next, we have attempted to thin down the disassembled αMoO3 belt crystals via plasma treatments and gas-phase etching at room temperature. Detailed procedures are supplied in the Supporting Information. In general, the surface of the [010]oriented α-MoO3 single crystals could have uncompleted layer steps/edges because of their van der Waals lattice structure. Figure 3p,q shows a schematic comparison of the typical belt crystals with and without uncompleted surface steps/edges and, for the sake of brevity, they are referred to as surface-defective (SD) and defect-free (DF) crystals. Morphological evolutions recorded by optical microscopy from the SD and DF crystals

upon the plasma treatments are presented in Figure 3a−f,g−l, respectively. For the plasma treatments, oxygen- and argonplasma (referred to as OP and AP) were used in this study and the process consists of OP for 5 min (step 1), OP for 30 min (step 2), AP for 30 min (step 3), OP for 30 min (step 4), and AP for 30 min (step 5). The in-plane orientations of the belt crystals, revealed by SCXRD, are indicated in Figure 3m. The comparisons in Figure 3a−f,g−l reveal that the DF crystals are more resistive than the SD ones to the plasma treatments with the same process parameters. Optical absorbance spectra taken from the SD and DF crystals upon the sequential plasma treatments are presented in Figure 3n,o, respectively. They, together with the Raman spectra (Supporting Information, Figure S3), reveal two dominant effects of the plasma treatments. One is the bombardment of the energetic species and the other is the release of oxygen from the lattice sites, both occurred at the uncompleted surface steps/edges where the terminal oxygen atoms are readily knocked out.32 The former leads to a broadening and softening of the Raman features, whereas the latter results in apparent red shifts in the absorption edges.27 The Raman spectra (Supporting Information, Figure S3) also show that the AP treatment-induced lattice disorders, which manifested themselves as the appearance of the feature at 447 cm−1, during step 3 can be more or less recovered by the OP treatment during step 4 and regenerated by the AP treatment during step 5. This behavior suggests that these lattice disorders are most likely associated with oxygen vacancies (VO).33,34 So-generated VO tends to diffuse into the asymmetric bridging oxygen sites (i.e., in the asymmetric O−Mo−O bonds along the [100]-axis of αMoO3), leading to the [001]-oriented dark linear micro12125

DOI: 10.1021/acs.jpcc.8b03340 J. Phys. Chem. C 2018, 122, 12122−12130

Article

The Journal of Physical Chemistry C

Figure 4. Morphological and absorption and Raman spectral evolutions of a SD and a DF α-MoO3 belt crystals upon sequential gas-phase XeF2- and HF-etching. (a−e) microphotographs; (f,g) absorbance spectra; (h,i) Raman spectra; (j.k) AFM images and their surface height profiles. Spectra in (f,h) were collected from the SD crystal, while spectra in (g,i) were collected from the DF crystal; images in (j,k) were recorded from the DF crystal before and after the extended HF-etching, demonstrating a layer-by-layer etching mode.

structures initiated at the surface steps/edges.32,35,36 A typical microphotograph and a schematic diagram of such plasma treatment-induced surface microstructures are shown in the insets in Figure 3n. The plasma-treated α-MoO3 belt crystals were further treated by gas-phase XeF2- and HF-etching at room temperature; detailed procedures are supplied in the Supporting Information. The morphological evolutions recorded by optical microscopy in Figure 4a−e and the evolutions of the interference fringes in Figure 4f,g, as well as those of the Raman spectra in Figure 4h,i, provide evidence that the [010]-oriented α-MoO3 crystal is more reactive to HF than to XeF2 at room temperature, especially for the DF crystals. Similar to the plasma treatments, the XeF2-etching tends to red shift the absorption edge and broaden/soften the Raman features of the SD crystal [see Figure 4f,h] but has a minor effect on the absorption and Raman spectra of the DF crystal [see Figure 4g,i]. In comparison, the absorption edge is somewhat shifted back to larger energies (i.e., blue shift) by the HF-etching for both SD and the DF crystals. This can be attributed to an effective thickness reduction induced by the HF-etching, which is supported by the disappearance of the interference fringes from the absorption spectrum in Figure 4g; the nonshifted Raman features in Figure 4i; and the changed color of the belt in Figure 4e from those in Figure 4a−d. These comparisons also indicate that the gas-phase HF-etching can effectively thin down the high-quality [010]-oriented α-MoO3 (i.e., DF crystals) while keeping their crystal qualities undeteriorated. Presented in Figure 4j,k are typical AFM images, together with their surface height profiles, recorded from the high-quality [010]-oriented α-MoO3 crystal before and after the HF-etching, that is, corresponding to the samples imaged in Figure 4a,e, respectively. A comparison of the height profiles indicates that atomic steps of ≤2 MLs have been formed on the HFetched surface. This result, together with the thickness

reduction of >200 nm [derived from the disappearance of the interference fringes in the absorbance spectrum in Figure 4g], provides evidence that the gas-phase HF-etching of α-MoO3 was in a layer-by-layer mode. This gas-phase layer-by-layer etching method may serve as a compensation to the wet chemical route and have important consequences when producing 2D α-MoO3 nanosheets from their high-quality [010]-oriented bulk crystals.16 In fact, our polarized Raman spectroscopy revealed that the rotation symmetry of the αMoO3 nanosheets produced by the gas-phase layer-by-layer etching method (Supporting Information, Figure S4) remains intact when compared with those of the thick α-MoO3 belt crystal. Figure 5 presents the results collected from a high-quality [010]-oriented α-MoO3 crystal after an extended HF-etching (see the Supporting Information for the process). Shown in Figure 5a is a microphotograph with its contrasts corresponding to the layer thickness variations. Figure 5b shows an AFM image taken from the edge area indicated by the box in Figure 5a, which exhibits regular linear microcracks along the direction close to the [001]-axis as indicated by the arrows. The presence linear microcracks in the high-quality α-MoO3 layer upon the gas-phase etching, similar to those observed for the SD crystals upon plasma treatments discussed above (see Figure 3), suggests occurrences of anisotropic in-plane strain relaxations and/or redistributions in the thinned crystals approaching their 2D nanosheets. The height profile in Figure 5c, derived from the locations indicated by the straight line in Figure 5b, shows that the HF-etched α-MoO3 layer is about 6.4 nm (∼8 MLs) thick and has a thinner area of 3.2 nm (∼4 MLs) thickness at the edge. Raman mappings recorded from the edge are shown in Figure 5d−f, addressing the distributions of the intensity, mode frequency, and linewidth of the Raman feature at ∼818 cm−1, respectively. One sees that Raman scattering from the thinner area at the edge of the α-MoO3 layer exhibits lowered 12126

DOI: 10.1021/acs.jpcc.8b03340 J. Phys. Chem. C 2018, 122, 12122−12130

Article

The Journal of Physical Chemistry C

Figure 5. Morphology, Raman mapping, and absorption spectra of gas-phase etched α-MoO3 belt crystal. (a) Microphotograph; (b) AFM image recorded at the boxed area in (a), the arrows indicate the regular linear defect structures, along the direction close to the [001]-axis, created during the thinning down of the layered crystal approaching 2D nanosheets; (c) surface height profile derived from the locations indicated by the straight line in the (b) that shows a thinner area at the edge of the α-MoO3 layer on the sapphire substrate; (d−f) Raman mappings of the Mo−O−Mo (Ag) mode at 818 cm−1 addressing its intensity, mode frequency, and linewidth distributions, respectively; (g) mode frequencies as a function of their respective scattering intensities derived from the boxed areas in (d,e), the inset shows a spectral comparison derived from the thicker inner area and the thinner edge area of the α-MoO3 layer; and (h) absorption spectra collected from the thicker inner area and the thinner edge area the α-MoO3 layer together with that of a bulk crystal.

fact, the Cai et al. have already observed a slight red shift of the 818 cm−1 mode in few-layer MoO3 as compared with that of the bulk crystal and they attributed the red shift to an in-plane lattice expansion of the 2D nanosheet.3 It is thus reasonable that the observed Raman mode softening of our 2D α-MoO3 nanosheets processed from belt crystals is due to an overall effect of anisotropic relaxation/redistribution of the coherent compressive strains. Such anisotropic strain relaxations/ redistributions could tune the electronic structure of the 2D α-MoO3 nanosheets, leading to the absorption edge blue shifts. In this light, we have further studied the electronic structures of 2D α-MoO3 nanosheets concerning the strains relaxations and/ or redistributions. Figure 6a−e presents the electronic structures of a 4 MLs thick α-MoO3 with and without uniaxial in-plane strains (i.e., ε = ±6%). They were calculated based on the periodic density functional theory (DFT), and the details can be found in the Supporting Information. One sees that the electronic structures exhibit the typical indirect band gap character with the valence band maximum at the R point, with the conduction band minimum at the Γ point except for the case of ε = 6% along the [100]-axis [see Figure 6d]. Plotted in Figure 6f are the direct band gap and the indirect band gap energies as a function of inplane strain variations up to ε = ±6%. It is seen that the band gap changes induced by the strain variations along the [100]axis are larger than those induced by the same amount of strain

intensity [Figure 5d], softened frequency [Figure 5e], and broadened linewidth [Figure 5f]. In Figure 5e, we have further plotted the mode frequencies as a function of their respective intensities, derived from the boxed areas indicated in Figure 5d,e, together with a nonlinear fitting (i.e., the solid line), which clearly shows red shifts at lower scattering intensities. The red shift is more clearly seen in the spectral comparisons at ∼818 cm−1 of the 8 MLs thick layer and its 4 MLs thick edge presented in the inset of Figure 5g. These results provide evidence for the mode softening of the 818 cm−1 phonon when the thickness of α-MoO3 is reduced approaching 2D nanosheets. Likewise, the absorption spectra collected from the 8 MLs thick layer, the 4 MLs thick edge, and a bulk α-MoO3 are shown in Figure 5h. One sees that the Eopt of the 4 MLs thick α-MoO3 is significantly blue-shifted to ∼4.8 eV. This blue shift is much larger than that derived from the thickness-dependent semilinear relationship in Figure 2c. However, both Raman mode softening and huge absorption edge blue shift of the 2D α-MoO3 (∼4 MLs) obtained by the gas-phase HF-etching are consistent with those of α-MoO3 nanosheets exfoliated by the BSA-assisted aqueous method.4 Because the SCXRD measurements indicate that our αMoO3 single crystals are compressively stressed and the lattice strains are anisotropic, in this regard, anisotropic strain relaxations and/or redistributions might occur as the α-MoO3 belt crystal is thinned down approaching 2D nanosheets. In 12127

DOI: 10.1021/acs.jpcc.8b03340 J. Phys. Chem. C 2018, 122, 12122−12130

Article

The Journal of Physical Chemistry C

Figure 6. Electronic structures and band gap energies of a 4 MLs thick α-MoO3 calculated using the periodic DFT. For these calculations, in-plane uniaxial strain of up to ε = ±6% was introduced along either the [100]- or [001]-axis to evaluate the effect of strain relaxations/redistributions on the electronic structures and thus the band gap energies of the 4 MLs thick α-MoO3 nanosheet, which provides clear clues to the thickness reductioninduced huge absorption edge blue shifts. (a) Fully relaxed; (b) 6% compressive strain along the [100]-axis; (c) 6% compressive strain along the [001]-axis, (d) 6% tensile strain along the [100]-axis; and (e) 6% tensile strain along the [001]-axis. (f) Band gap energies of the α-MoO3 nanosheet as a function of in-plane strains.

variations along the [001]-axis; a strain variation from ε = −6 to 6% along the [100]-direction can result in a blue shift of ∼1.5 eV in the direct band gap. Taking the regular linear microcracks observed in Figure 5b into account, we believe that in-plane strain relaxations/ redistributions occurred during the layer’s thinning down approaching 2D nanosheets. The [001]-oriented linear microcracks are most likely created due to the relaxation of tensile strains along the [100]-axis. In this regard, the creak-free area of the thinner 2D nanosheets could be tensile stressed along their [100]-axis and that, in turn, resulted in the larger band gap energies. It has to be noted that the inherent VO and their diffusions upon the gas-phase etching of α-MoO3 towards 2D nanosheets may also affect their in-plane strain relaxations and/ or redistributions; unfortunately, the detailed mechanism is unclear at this stage and, therefore, not taken into account in the current calculations. Nevertheless, the combination between the anisotropic strain relaxations/redistributions and the calculated electronic structures provide clear clues to the band gap widening of the 2D α-MoO3 nanosheets and thus the observed absorption edge blue shifts.

together with the constant intensity ratios of I282/I290, supports the constant oxygen deficiency in the layered single crystals. The α-MoO3 layered crystal without surface defects is more resistive than the SD ones to plasma treatments at room temperature. The plasma treatment-induced linear structures along the [001]-axis at the edge of the layer steps, accompanied by red shifts in the absorption edge and softening/broadening of the Raman features, of the SD α-MoO3 belts rather than the DF ones suggest that the formation of the linear structures is related to the knocking-out of oxygen atoms as well as the diffusions of VO. This conclusion is further supported by the creation/recover/recreation of the Raman feature at ∼447 cm−1 of the α-MoO3 single crystal upon the sequential AP/OP/AP treatments. A novel gas-phase layer-by-layer etching of the layered α-MoO3 single crystals has been demonstrated and the HF-etching is more effective than that of XeF2-etching at room temperature. By extending the HF-etching, an α-MoO3 nanosheet of about 4 MLs thick has been obtained at the edge of a ∼8 MLs thick α-MoO3 layer; this 2D nanosheet exhibits apparent phonon softening of the Ag mode at ∼818 cm−1 and a huge Eopt blue shift. The phonon softening provides evidence for the in-plane strain relaxations/redistributions, while the huge EOpt blue shift, in terms of our periodic DFT calculations, is attributable to the anisotropic in-plane strain relaxations/redistributions. Although the effect of VO (either inherent or created during the film thinning), as well as their diffusions, along with the strain relaxations/redistributions was not taken into account, they could also play roles in varying the electronic structures of the 2D α-MoO3.

4. CONCLUSION In conclusion, we have synthesized high-quality layered αMoO3 single crystals by TVT of MoO3 powders. SCXRD revealed that the layered crystals are compressively stressed and the lattice strains are anisotropic. EDX indicated that the layered crystals are oxygen deficient; however, the oxygen deficiency (i.e., MoO2.93) is constant and independent of the layer thickness in the range of 10−104 nm. In this thickness range, we observed apparent absorption edge blue shifts of the layered crystals as their thickness is reduced; however, their Raman features do not exhibit any frequency shifts at all. The former, based on our theoretical FDTD calculations, is due to light interferences at the crystal surfaces, whereas the latter,



ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.jpcc.8b03340. 12128

DOI: 10.1021/acs.jpcc.8b03340 J. Phys. Chem. C 2018, 122, 12122−12130

Article

The Journal of Physical Chemistry C



Optical band gap of γ-In2Se3 as a function layer thickness; theoretical FDTD calculations of the transmissions, and thus the absorptions and optical band gaps, of layered α-MoO3 single crystals as a function of their thicknesses in the range of 10−104 nm; Raman spectra collected from a SD and a DF α-MoO3 single crystals upon sequential OP and AP treatments; disassembling of α-MoO3 belt crystals on sapphire substrate; polarized Raman spectra of α-MoO3 thick layer and thin nanosheet; plasma treatment procedures; gas-phase etching procedures; and details of the periodic DFT calculations (PDF)

(10) Liu, H. F.; Wong, S. L.; Chi, D. Z. CVD growth of MoS2-based two-dimensional materials. Chem. Vap. Deposition 2015, 21, 241−259. (11) Novoselov, K. S.; Jiang, D.; Schedin, F.; Booth, T. J.; Khotkevich, V. V.; Morozov, S. V.; Geim, A. K. Two-dimensional atomic crystals. Proc. Natl. Acad. Sci. U.S.A. 2005, 102, 10451−10453. (12) Manzeli, S.; Ovchinnikov, D.; Pasquier, D.; Yazyev, O. V.; Kis, A. 2D transition metal dichalcogenides. Nat. Rev. Mater. 2017, 2, 17033. (13) Liu, H.; Ansah-Antwi, K. K.; Ying, J.; Chua, S.; Chi, D. Towards large area and continuous MoS2 atomic layers via vapor-phase growth: Thermal vapor sulfurization. Nanotechnology 2014, 25, 405702. (14) Liu, H.; Ansah Antwi, K. K.; Chua, S.; Chi, D. Vapor-phase growth and characterization of Mo1‑xWxS2 (0 ≤ x ≤1) atomic layers on 2-inch sapphire substrates. Nanoscale 2014, 6, 624−629. (15) Liu, H.; Chi, D. Dispersive growth and laser-induced rippling of large-area singlelayer MoS2 nanosheets by CVD on c-plane sapphire substrate. Sci. Rep. 2015, 5, 11756. (16) Rahman, F.; Ahmed, T.; Walia, S.; Mayes, E.; Sriram, S.; Bhaskaran, M.; Balendhran, S. Two-dimensional MoO3 via a top-down chemical thinning route. 2D Mater. 2017, 4, 035008. (17) Mak, K. F.; Lee, C.; Hone, J.; Shan, J.; Heinz, T. F. Atomically thin MoS2: A new direct-gap semiconductor. Phys. Rev. Lett. 2010, 105, 136805. (18) Cai, Y.; Zhang, G.; Zhang, Y.-W. Layer-dependent band alignment and work function of few-layer phosphorene. Sci. Rep. 2014, 4, 6677. (19) Osters, O.; Nilges, T.; Bachhuber, F.; Pielnhofer, F.; Weihrich, R.; Schöneich, M.; Schmidt, P. Synthesis and identification of metastable compounds: Black arsenic-science or fiction? Angew. Chem., Int. Ed. 2012, 51, 2994−2997. (20) Liu, B.; Köpf, M.; Abbas, A. N.; Wang, X.; Guo, Q.; Jia, Y.; Xia, F.; Weihrich, R.; Bachhuber, F.; Pielnhofer, F.; et al. Black arsenicphosphorus: Layered anisotropic infrared semiconductors with highly tunable compositions and properties. Adv. Mater. 2015, 27, 4423− 4429. (21) Molina-Mendoza, A. J.; Lado, J. L.; Island, J. O.; Niño, M. A.; Aballe, L.; Foerster, M.; Bruno, F. Y.; López-Moreno, A.; VaqueroGarzon, L.; van der Zant, H. S. J.; et al. Centimeter-scale synthesis of ultrathin layered MoO3 by van der Waals epitaxy. Chem. Mater. 2016, 28, 4042−4051. (22) Wang, D.; Li, J.-N.; Zhou, Y.; Xu, D.-H.; Xiong, X.; Peng, R.-W.; Wang, M. Van der Waals epitaxy of ultrathin α-MoO3 sheets on mica substrate with single-unit-cell thickness. Appl. Phys. Lett. 2016, 108, 053107. (23) Sreedhara, M. B.; Matte, H. S. S. R.; Govindaraj, A.; Rao, C. N. R. Synthesis, characterization, and properties of few-layer MoO3. Chem.Asian J. 2013, 8, 2430−2435. (24) Ji, F.; Ren, X.; Zheng, X.; Liu, Y.; Pang, L.; Jiang, J.; Liu, S. 2DMoO3 nanosheets for superior gas sensors. Nanoscale 2016, 8, 8696− 8703. (25) Lajaunie, L.; Boucher, F.; Dessapt, R.; Moreau, P. Strong anisotropic influence of local-field effects on the dielectric response of α-MoO3. Phys. Rev. B: Condens. Matter Mater. Phys. 2013, 88, 115141. (26) Sun, C. W.; Xin, P.; Liu, Z. W.; Zhang, Q. Y. Room-temperature photoluminescence of ZnO/MgO multiple quantum wells on Si (001) substrates. Appl. Phys. Lett. 2006, 88, 221914. (27) Yano, T.-a.; Yoshida, K.; Hayamizu, Y.; Hayashi, T.; Ohuchi, F.; Hara, M. Probing edge-activated resonant Raman scattering from mechanically exfoliated 2D MoO3 nanolayers. 2D Mater. 2015, 2, 035004. (28) Dieterle, M.; Weinberg, G.; Mestl, G. Raman spectroscopy of molybdenum oxides Part I. Structural characterization of oxygen defects in MoO3‑x by DR UV/VIS, Raman spectroscopy and X-ray diffraction. Phys. Chem. Chem. Phys. 2002, 4, 812−821. (29) Yan, B.; Zheng, Z.; Zhang, J.; Gong, H.; Shen, Z.; Huang, W.; Yu, T. Orientation controllable growth of MoO3 nanoflakes: MicroRaman, field emission, and birefringence properties. J. Phys. Chem. C 2009, 113, 20259−20263.

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. ORCID

Hongfei Liu: 0000-0002-9905-7028 Dongzhi Chi: 0000-0001-9562-1595 Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS The authors would like to acknowledge S. Q. Bai for his help in SCXRD measurements. This research is supported by A*STAR Science and Engineering Research Council Pharos 2D Program (SERC grant no. 152-70-00012).



REFERENCES

(1) Berthumeyrie, S.; Badot, J.-C.; Pereira-Ramos, J.-P.; Dubrunfaut, O.; Bach, S.; Vermaut, P. Influence of lithium insertion on the electronic transport in electroactive MoO3 nanobelts and classical powders: Morphological and particle size effects. J. Phys. Chem. C 2010, 114, 19803−19814. (2) Guo, H.; Lan, C.; Zhou, Z.; Sun, P.; Wei, D.; Li, C. Transparent, flexible, and stretchable WS2 based humidity sensors for electronic skin. Nanoscale 2017, 9, 6246−6253. (3) Cai, L.; McClellan, C. J.; Koh, A. L.; Li, H.; Yalon, E.; Pop, E.; Zheng, X. Rapid flame synthesis of atomically thin MoO3 down to monolayer thickness for effective hole doping of WSe2. Nano Lett. 2017, 17, 3854−3861. (4) Liu, H.; Cai, Y.; Han, M.; Guo, S.; Lin, M.; Zhao, M.; Zhang, Y.; Chi, D. Aqueous and mechanical exfoliation, unique properties, and theoretical understanding of MoO3 nanosheets made from freestanding α-MoO3 crystals: Raman mode softening and absorption edge blue shift. Nano Res. 2018, 11, 1193−1203. (5) Liu, H.; Yang, R. B.; Yang, W.; Jin, Y.; Lee, C. J. J. Atomic layer deposition and post-growth thermal annealing of ultrathin MoO3 layers on silicon substrates: Formation of surface nanostructures. Appl. Surf. Sci. 2018, 439, 583−588. (6) Sreedhara, M. B.; Santhosha, A. L.; Bhattacharyya, A. J.; Rao, C. N. R. Composite of few-layer MoO3 nanosheets with graphene as a high performance anode for sodium-ion batteries. J. Mater. Chem. A 2016, 4, 9466−9471. (7) Balendhran, S.; Deng, J.; Ou, J. Z.; Walia, S.; Scott, J.; Tang, J.; Wang, K. L.; Field, M. R.; Russo, S.; Zhuiykov, S.; et al. Enhanced charge carrier mobility in two-dimensional high dielectric molybdenum oxide. Adv. Mater. 2013, 25, 109−114. (8) Balendhran, S. Devices and Systems Based on Two Dimensional MoO3 and MoS2; RMIT University, 2013. (9) Balendhran, S.; Walia, S.; Nili, H.; Ou, J. Z.; Zhuiykov, S.; Kaner, R. B.; Sriram, S.; Bhaskaran, M.; Kalantar-zadeh, K. Two-dimensional molybdenum trioxide and dichalcogenides. Adv. Funct. Mater. 2013, 23, 3952−3970. 12129

DOI: 10.1021/acs.jpcc.8b03340 J. Phys. Chem. C 2018, 122, 12122−12130

Article

The Journal of Physical Chemistry C (30) Ho, C.-H. Amorphous effect on the advancing of wide-range absorption and structural-phase transition in γ-In2Se3 polycrystalline layers. Sci. Rep. 2014, 4, 4764. (31) Ho, C.-H.; Chen, Y.-C. Thickness-tunable band gap modulation in γ-In2Se3. RSC Adv. 2013, 3, 24896−24899. (32) Smith, R. L. The Structural Evolution of the MoO3 (010) Surface during Reduction and Oxidation Reactions; Carnegie Mellon University, 1998. (33) Eda, K. Infrared spectra of hydrogen molybdenum bronze, H0.34MoO3. J. Solid State Chem. 1989, 83, 292−303. (34) Chen, D.; Liu, M.; Yin, L.; Li, T.; Yang, Z.; Li, X.; Fan, B.; Wang, H.; Zhang, R.; Li, Z.; et al. Single-crystalline MoO3 nanoplates: topochemical synthesis and enhanced ethanol-sensing performance. J. Mater. Chem. 2011, 21, 9332−9342. (35) Hsu, Z. Y.; Zeng, H. C. Generation of double-layer steps on (010) surface of orthorhombic MoO3 via chemical etching at room temperature. J. Phys. Chem. B 2000, 104, 11891−11898. (36) Zeng, H. C. Chemical etching of molybdenum trioxide: A new tailor-made synthesis of MoO3 catalysts. Inorg. Chem. 1998, 37, 1967− 1973.

12130

DOI: 10.1021/acs.jpcc.8b03340 J. Phys. Chem. C 2018, 122, 12122−12130