Research Article www.acsami.org
Surface Analysis of WSe2 Crystals: Spatial and Electronic Variability Rafik Addou* and Robert M. Wallace* Department of Materials Science, Engineering, The University of Texas at Dallas, Richardson, Texas 75080, United States S Supporting Information *
ABSTRACT: Layered semiconductor compounds represent alternative electronic materials beyond graphene. WSe2 is one of the two-dimensional materials with wide potential in optoand nanoelectronics and is often used to construct novel threedimensional architectures with new functionalities. Here, we report the topography and the electronic property of the WSe2 characterized by means of scanning tunneling microscopy and spectroscopy (STM and STS), X-ray photoelectron spectroscopy (XPS), and inductively coupled plasma mass spectrometry. The STM images reveal the presence of atomic-size imperfections and a variation in the electronic structure caused by the presence of defects and impurities below the detection limit of XPS. Both STS and photoemission reveal a spatial variation in the Fermi level position. The analysis of the core levels indicates the presence of different doping levels. The presence of a large concentration of defects and impurities has a strong impact on the electronic properties of the WSe2 surface. Our findings demonstrate that the growth of controllable and high quality two-dimensional materials at nanometer scale is one of the most challenging tasks that requires further attention. KEYWORDS: surface defects, impurities, tungsten diselenide, scanning probe microscopy, photoemission
■
INTRODUCTION Tungsten diselenide, WSe2, is another promising compound exhibiting properties that drastically differ from monolayer to bulk form, as is typical for MoS2 and other transition-metal dichalcogenides (TMDs). Both group VI compounds MoS2 and WSe2 have the same indirect bandgap of ∼1.2 eV in the bulk form.1,2 But in monolayer regime, the bandgap increases to ∼1.6 eV for WSe21,3,4 and ∼1.7 eV for MoS2.4,5 Depending of the types of contact metals, transport properties of exfoliated monolayer WSe2 can be tuned to be either p-type or ambipolar.6 Mechanically exfoliated monolayer WSe2 has been used for diverse electronic and optoelectronic devices such as field effect transistors, photodetectors, light-emitting diodes, and photovoltaic devices.6−10 The WSe2 crystal is a semiconductor layered material belonging to the transition metal dichalcogenides family with the MX2 formula, where M is a transition metal and X a chalogen (S, Se, or Te).11 The WSe2 bulk structure consists of three atomic planes: two atomic layers of close-packed Se atoms separated by one close-packed W atomic sheet (Se−W− Se). Within each single layer, the atoms are strongly bonded through covalent forces, while weak van der Waals forces dominate the interaction between Se−W−Se layers (Supporting Information, Figure S1a).11 The hexagonal crystal structure of WSe2 belongs to the space group D6h4-P63/mmc with the following lattice constants: a = b = 0.3282 nm and c = 1.2937 nm.12,13 The crystals of such layered materials can be cleaved down to a single monolayer (ML). The 2H polytype is the most stable WSe2 phase 2H, where H means that this compound has a hexagonal crystal structure where the W © XXXX American Chemical Society
atoms are coordinated in trigonal prismatic geometry and 2 is the number of Se−W−Se layers by unit cell (Figure S1b). WSe2 and other novel semiconducting two-dimensional (2D) materials are now a primary focus of the scientific community for the outstanding potential of fabricating a 3D building block by layer-by-layer stacking with any required sequence to create a new device physics and in-demand functionality.14 In the effort to understand the properties of TMDs, several studies were performed to measure the surface characteristics of bulk and thin film materials such as MoS2, WSe2, and HfSe2, among others.2,4,15−18 Recent reports showed that synthetic and geological MoS2 bulk crystal reveals large spatial variability, where the surface characteristics (stoichiometry, work function, and doping) varies from a sample to another and even across the same surface.2,15,16 Recent studies showed that the transition metal sulfides (MoS2 and WS2) are more susceptible for passivation after superacid treatment than transition metal selenides (WSe2 and MoSe2). This difference in response toward the chemical treatment was explained by the difference in the density and the nature of surface defects.19 Contacts and interfaces of semiconductors plays a major role for nanoelectronic device performance.16 The overall properties at the interface are affected by the imperfections nature and density, since the intrinsic defect states and change in the work functions have an effect in determining the Schottky barrier height. Interface-induced gap states alter substantially the band offsets and Schottky barrier heights of TMDs has been Received: July 19, 2016 Accepted: September 7, 2016
A
DOI: 10.1021/acsami.6b08847 ACS Appl. Mater. Interfaces XXXX, XXX, XXX−XXX
Research Article
ACS Applied Materials & Interfaces
Figure 1. Imaging of surface imperfections on freshly exfoliated WSe2 crystal surface (sample A). (a) STM image (voltage bias (Vb) = 2.0 V, tunneling current (It) = 0.5 nA) shows contrast variation at large scale as indicated in the attached line profile. (b) AFM image showing very flat surface with very low roughness as represented by the attached line profile. (c) STM image (Vb = 1.3 V, It = 0.5 nA) shows hillock-like features appearing as “protrusions”; the line profile across those bright defects is drawn in the inset. (d) STM image (Vb = −1.5 V, It = 0.7 nA) recorded on different WSe2 crystal (sample B) showing bright defect and depression with corresponding line profiles. (e) STM image (Vb = 1.25 V, It = 0.5 nA) showing topographic depression. (f) STM image (Vb = 1.5 V, It = 0.5 nA) showing an atomic structure with the expected lattice constant and 2H phase.
reported.20 The goal of this study is to characterize at atomic scale the surface properties and outline the spatial variation of WSe2 basal plane. We studied the surface of a bulk crystal of WSe2 via room temperature (RT) scanning tunneling microscopy (STM), scanning tunneling spectroscopy (STS), X-ray photoelectron spectroscopy (XPS), and inductively coupled plasma mass spectrometry (ICPMS).
morphological height (e.g., protrusions or hillhocks), but corresponding to variations in the tunneling current. Such surface features are representative of this exfoliated sample surface and are detected across different areas (another large STM image exhibiting similar surface characteristics is presented in Figure S3). Atomic force microscopy is an appropriate method to differentiate between the electronic effect and the real surface topography. Figure 1b shows the AFM image recorded in air on freshly exfoliated surface indicating very smooth surface with low roughness measured at 0.10 ± 0.02 nm. This AFM vs STM comparison demonstrates that the large local variation in the STM image is purely electronic. The high magnification STM image in Figure 1c shows different surface characteristics in comparison to Figure 1a when the imaging conditions were changed. The bias dependence provides further evidence that the STM “roughness” on this surface is not morphological in nature. The step edge height is measured at ∼0.7 nm, which corresponds to a WSe2 single layer (c/2).
■
RESULTS AND DISCUSSION Surface Topography and Electronic Effects. Figure 1a shows a STM image recorded on freshly cleaved surface of WSe2 (sample A). The average height is measured at about 0.48 ± 0.02 nm, which indicates a significant roughness is detected for a pristine surface of such a layered material in comparison to the well-known flatness measured on graphite (Figure S2). The first explanation of such high “roughness” is caused by an electronic effect of the imaging technique, since the scanning tunneling microscopy image portrays the local surface state density. In other words, the bright contrast in the STM image in Figure 1a is not directly interpreted as an indicator of B
DOI: 10.1021/acsami.6b08847 ACS Appl. Mater. Interfaces XXXX, XXX, XXX−XXX
Research Article
ACS Applied Materials & Interfaces
Figure 2. Imaging of atomic size imperfections on WSe2 (sample A). (a) STM image (Vb = 1.5 V, It = 1.5 nA) shows a single Se vacancy with the corresponding line profile. (b) STM image (Vb = 1.5 V, It = 1.5 nA) showing two types of point defects: single vacancy (noted as “V”) and local depression (noted as “A”) caused by the presence of an acceptor at this area. (c) STM image (Vb = 1.5 V, It = 0.5 nA) shows an atomic bright spot (noted as “D”) induced by the presence of donor atom at the vicinity of the surface.
On a different WSe2 crystal (sample B), the imaged surface defects are similar to those found on MoS2,2 where the surface is dominated by the presence of dark and bright defects (Figure 1d). The depth of the hole (“h”) is measured at 0.4 nm, and the height of the bright defect “protrusion” (“p”) is 0.6 nm. The presence of such defects near the surface causes electron depletion that appears as a depression as highlighted by the gray zone in the line profile in Figure 1d. Such defects could be either morphological in origin or a tunneling current contrast change induced by foreign atoms located near the surface.2,22 Figure 1e presents a real topological “pinhole” with a depth of ∼0.3 nm and a diameter of ∼1.5 nm, caused by missing material (Se and W atoms), surrounded by a hillock features with a lateral size of ∼1 nm. The atomic order appears to be maintained within the defective area. Interestingly, similar defects have been reported after Ar ion bombardment of WSe2 at a beam energy of 5 keV.21,22 However, in this study, the sample was transferred immediately to the STM chamber after exfoliation in air without any sputtering or thermal treatment. The atomic resolution of the WSe2(0001) surface is presented in Figure 1f; the expected hexagonal 2H structure is imaged with a unit cell of 0.32 ± 0.03 nm. This is in agreement with the ideal crystal model, and reported STM images for WSe2 thin film and bulk crystal, as well as the lattice-resolution using AFM.12,13,21−23 Atomic Size Imperfections. The WSe2 surface exhibits three main local imperfections. Figure 2a shows an STM image of a single vacancy (V) caused by a missing of chalcogen atom. The line profile measured across the vacancy shows a depth of ∼0.15 nm, which corresponds to half of the pinhole depth in Figure 1e. The point defect corresponds to a single Se vacancy, and the pinhole corresponds to vacancy in that atomic plane. In addition to single Se vacancies, outlined features in Figure 2b show a local depression (A) without disrupting the hexagonal structure. The local depression is attributed to the presence of an acceptor impurity in the vicinity of the surface. The line profile across the depression shows a depth of 0.07 ± 0.02 nm smaller than the single vacancy depth. The hexagonal atomic order remains preserved within the depression (Figure S4). The dark point defect (V) can be also attributed to W vacancy and O or Ni impurities.28 Dopants such as Au, Ni, Cr, Mo, O, and Fe are not active and do not essentially contribute to the charge carriers in WSe2.24,25
Similarly, when a donor impurity is present, the same area is imaged as high contrast (D) as indicated in Figure 2c. Such local variations in the tunneling image contrast were well studied on conventional semiconductors and also reported for TMD surfaces and wurtzite-structure compounds.2,25−29 Theory showed the impact of subsurface impurities scattering on the STM imaging.27 Sommerhalter et al. used current imaging tunneling spectroscopy and photoassisted scanning tunneling spectroscopy to study impurities in p-type WS2. They demonstrated that the nanoscale depression is induced by the local charge distribution at ionized acceptors.28 The change in the local imaging contrast (on the depression “A” or on the hillock site “D”) were attributed to the spatial depth of the dopant site; i.e., the contrast is correlated to the localization of the impurity in the first, second, or third atomic layer.25,29 In order to quantify the quality of the WSe2 surface, we have estimated the density of point defects on several atomically resolved STM images recorded at different bias voltages and tunneling currents. The density of the atomic-size defects is found to be ∼1.22 × 1012 cm−2, in agreement with the defect density measured on WSe2 film grown by MBE (2.8 × 1012 cm−2).30 The density of bright defects (∼1.72 × 1012 cm−2) is higher than dark defects (∼1.19 × 1012 cm−2). The lowest defect density was found for V defects (∼0.74 × 1012 cm−2). Electronic Properties. The electronic structure of the WSe2 surface was characterized using STS at room temperature and under white light. Figure 3 summarizes several dI/dV vs V spectra showing the most representative electronic behaviors observed on the exfoliated WSe2 crystal surface. Figure 3a shows the expected p-type behavior with a Fermi level located close to the valence band maximum (VBM), with a VB edge located at −0.2 eV and the conduction band minimum (CBM) at 1.08 eV giving a band gap of ∼1.28 eV. On a different region (but on the same sample surface) and under the same imaging conditions, the STS shows the CBM located at ∼1.2 eV, but the current is zero at reverse bias (Figure 3b). A feature in the band gap region appears at ∼0.65 eV above the Fermi level. A similar feature was observed on n-MoS2 at 0.65 eV but below the Fermi level.2 Defects are thought to be responsible for this local surface state detected in the band gap.2,31 On sample B (Figure 3d), the Fermi level position indicates that this sample is n-type with a VBM (CBM) located at −1.04 eV (0.24 eV). In a different area on sample B (Figure S5), the Fermi level was C
DOI: 10.1021/acsami.6b08847 ACS Appl. Mater. Interfaces XXXX, XXX, XXX−XXX
Research Article
ACS Applied Materials & Interfaces
Similar behavior of zero tunneling current detected in Figure 3b at reverse bias has been reported on other semiconductors: synthetic and geological MoS2, WS2, WSe2, H-terminated Si surfaces, and III−V compounds.2,16,28,32−39 It is worth noting that on semiconductor materials with a low surface state density one should consider an unbalanced effect of band bending occurring between the applied voltage and the work function difference between the metallic tip and the semiconductor. The standard STS interpretation is invalid with this nonequilibrium condition corresponding to a flat VB electronic structure. For a p-type semiconductor such as WSe2, a drastic change in the I− V curves will be measured at negative applied voltage, where a saturation current is observed. This behavior indicates that the tunneling of the electrons from the valence band of the sample into the unoccupied states of the tip is not possible. Several tunneling spectroscopy studies were performed on p-WSe2 surface, as well as p-WS2, under dark conditions and under laser illumination.32−36 Sans illumination, no tunneling current was detected at negative bias; the flat VB behavior was explained by the nonequilibrium between the minority charge carrier concentration at the WSe2 surface and the bulk concentration. This means that electrons at the surface conduction band of WSe2 drift to the bulk region and, thus, are not able to tunnel effectively from the valence band of WSe2 to the tip. Under continuous laser illumination, the backward tunneling current (Vb < 0) is increased due to the incident photons and can be divided into two distinctive regions: photovoltaic and photoamperic.28,32−36 Elemental Analysis and Spatial Variation. As noted in the above discussion, the detection of STM image contrast variations implicates the presence of impurities as well as defects. To establish the presence of such impurities, ICP mass spectrometry was performed on a WSe2 bulk crystal. A search for 39 elements typically of interest for Si-based integrated circuit (IC) technologies was conducted, and 16 elements were found with a concentration greater than 0.1 parts per billion by weight (ppbw). In this sample, Ge and Mo are detected with an equivalent impurity concentration levels estimated to be higher
Figure 3. dI/dV vs V spectra recorded on two different freshly exfoliated WSe2 samples (A, B) showing different behaviors: (a) p-type conductivity, (b) band bending effect (It ∼ 0 when Vb = 0), (c) defect state in the gap, and n-type conductivity.. The STS in (a−c) was recorded on sample “A”, and the STS in (d) was recorded on sample “B”.
found to be close to the midgap with a band gap of ∼1.25 ± 0.05 eV with the VBM (CBM) located at −0.56 eV (0.69 eV).
Figure 4. Periodic table shows the impurity concentrations obtained from ICPMS measurements of WSe2 bulk sample. Elements in blue were not measured, elements in gray were measured but not detected, and elements in yellow and orange were detected with different concentration levels. D
DOI: 10.1021/acsami.6b08847 ACS Appl. Mater. Interfaces XXXX, XXX, XXX−XXX
Research Article
ACS Applied Materials & Interfaces
Figure 5. Photoemission study of freshly exfoliated WSe2 crystal. (a) W 5p and W 4f XPS core levels recorded on two different samples. Inset showing overlapping spectra to highlight the additional component at lower binding energy. Similar comparison is performed for Se 3d and valence band edge as shown in (b), and (c), respectively.
than 2 × 1010/cm2 (concentrations reported here are normalized to a Si host matrix to enable comparison to IC starting material specifications). All other detected impurities are below 1 × 1010/cm2 as indicated in Figure 4. Typical impurity specifications for Si-based IC technology limit concentrations to ∼5 × 1010/cm2.40 This study shows the synthetic WSe2 crystal has a significantly lower impurity concentration relative to geological MoS2.15 Nevertheless, the detected impurities may explain the variation in the electronic behavior observed using XPS and STS. Moreover, the local variation in the contrast observed by STM can be caused by the incorporation of foreign atoms (substitutional or intercalated) in the vicinity of the surface. High resolution XPS (energy step = 0.05 eV) was used to search for all elements detected by ICPMS, but all impurities are below the detection limit of XPS (
