Microscopic evidence of strong interactions between chemical vapor deposited 2D MoS2 film and SiO2 growth template

Two-dimensional MoS2 film can grow on oxide substrates including Al2O3 and SiO2. However, it cannot grow usually on non-oxide substrates such as a bare Si wafer using chemical vapor deposition. To address this issue, we prepared as-synthesized and transferred MoS2 (AS-MoS2 and TR-MoS2) films on SiO2/Si substrates and studied the effect of the SiO2 layer on the atomic and electronic structure of the MoS2 films using spherical aberration-corrected scanning transition electron microscopy (STEM) and electron energy loss spectroscopy (EELS). The interlayer distance between MoS2 layers film showed a change at the AS-MoS2/SiO2 interface, which is attributed to the formation of S–O chemical bonding at the interface, whereas the TR-MoS2/SiO2 interface showed only van der Waals interactions. Through STEM and EELS studies, we confirmed that there exists a bonding state in addition to the van der Waals force, which is the dominant interaction between MoS2 and SiO2. The formation of S–O bonding at the AS-MoS2/SiO2 interface layer suggests that the sulfur atoms at the termination layer in the MoS2 films are bonded to the oxygen atoms of the SiO2 layer during chemical vapor deposition. Our results indicate that the S–O bonding feature promotes the growth of MoS2 thin films on oxide growth templates. Supplementary Information The online version contains supplementary material available at 10.1186/s40580-021-00262-x.


Introduction
Transition metal dichalcogenides, such as MoS 2 , have attracted much interest because of their remarkable electrical, mechanical, thermal, and optical properties. Therefore, they are considered novel materials and suitable for application in optoelectronic devices, water-splitting catalysts, sensors, field-effect transistors, capacitors, and energy storage devices [1][2][3][4][5][6][7][8][9][10]. Recent studies have focused on the large-scale growth of MoS 2 films, which is mainly carried out on Al 2 O 3 and SiO 2 /Si substrates [11][12][13][14]. Through chemical vapor deposition (CVD) and pulsed laser deposition (PLD), MoS 2 films are well deposited when the Si wafer has a SiO 2 layer on the top. In contrast, the growth of MoS 2 fails or MoSi 2 is synthesized instead of MoS 2 when CVD growth is carried out on a bare Si wafer even if Si substrate has a native oxide layer on top because the thin native oxide layer can be removed or penetrated during the high temperature synthesis process [4,15]. Moreover, there are a few studies to show the formation of MoSi 2 through chemical reactions between MoS 2 and Si. When the Mo film is sulfurized on a bare Si wafer, the formation of a gas phase SiS 2 rather than MoS 2 is presumable. This indicates that making large-scale growth of MoS 2 directly on Si is hardly achievable. However, when Mo is sulfurized on SiO 2 substrates, S-O bonding is preferable rather than S-Si bonding, which helps Mo-S bonding followed by growth of a MoS 2 film. The different tendency of CVD MoS 2 growth on bare Si and SiO 2 /Si substrates was found in previous studies [4,15]. To date, it has been reported that some substrates such as SiO 2 /Si and sapphire enable the largescale deposition of MoS 2 films. However, microscopic evidence showing the interaction between the chemicalvapor-deposited MoS 2 and the SiO 2 growth template has not been observed. Microscopic studies are mainly focused on the atomic and electronic structures of MoS 2 , using transmission electron microscopy (TEM) and aberration-corrected scanning transmission electron microscopy (Cs-corrected STEM), which can directly observe materials on an atomic level. Crystal and atomic structures with various defects have been discovered using a combination of ab-initio calculations [4,[11][12][13][14][16][17][18][19][20][21][22].
To investigate the atomic and electronic structure and chemical state of the MoS 2 /SiO 2 heterostructure, an atomic-scale study using Cs-corrected STEM and EELS with a cross-sectional view should be carried out. The use of Cs-corrected STEM with a probe corrector would enable the direct observation of heterointerfaces at the atomic resolution. Moreover, the changes in the electronic and chemical states would be available in unit-cell resolution. Interfaces between dissimilar compounds provide unusual properties that are attributed to lattice mismatch, strain, chemical bonds, and the formation of secondary interfacial layers [23][24][25][26][27]. The observation of the atomic and electronic structures of MoS 2 / SiO 2 interfaces not only provides insights on the growth mechanisms and bonding states, but also suggests the possibility of application in various devices. Understanding the nature of the interfaces can provide insights on how the MoS 2 film is synthesized via CVD and how the MoS 2 `film on the SiO 2 /Si template can be utilized in electronic devices.
In this work, we investigated the atomic and electronic structures of MoS 2 (more than 20 layers)/SiO 2 interfaces on an Si substrate. First, we observed and compared the atomic structure of the AS and TR-MoS 2 films on the SiO 2 /Si substrates and plotted the (001) plane distance from the interface to the interior of the film. Second, we investigated the chemical bonding state at the AS and TR-MoS 2 /SiO 2 hetero interfaces. The final aim of this study is figuring out the main reason why MoS 2 is deposited only on oxide substrates such as SiO 2, Al 2 O 3 and understanding how the SiO 2 layer affects the structures of MoS 2 films when Mo is sulfurized through CVD [11][12][13]. We used SiO 2 for a growth template mainly because other oxides such as Al 2 O 3 and SrTiO 3 do not dissolve in HF, disabling transfer process. For this purpose, we carried out a TEM-based analysis with Cs-corrected high annular dark field (HAADF) imaging, and annular bright field (ABF) STEM imaging and EELS with a combination of ab-initio calculations. To the best of our knowledge, this is the first reported paper investigating the influence of the SiO 2 template on the growth of MoS 2 films using Cs-corrected STEM and EELS with a cross-sectional view.

MoS 2 film deposition
SiO 2 (300 nm)/Si wafers were cleaned with a standard piranha solution (3:1 mixture of H 2 SO 4 and H 2 O 2 ) using conventional cleaning procedures followed by ultrasonication in acetone, isopropyl alcohol, and deionized (DI) water. To obtain hydrophilic surfaces on the SiO 2 /Si wafers, O 2 plasma and UV-O 3 surface treatments were sequentially performed for 15 min. A 10 nm-thick Mo thin film was deposited on a SiO 2 /Si substrate using an E-beam evaporator. (Rocky Mountain Vacuum Tech, Englewood, CO, USA). The base pressure, E-beam voltage, and current were 10 -6 Torr, 7.3 kV, and 70 mA, respectively, and the deposition rate was approximately 0.1 Å/s. Sulfurization of the Mo thin film was performed by CVD at 900 °C for 30 min. After sulfurization, the thin films were annealed for the crystallization of the MoS 2 films.

TEM sample preparation
Due to the weak interaction between the MoS 2 film and SiO 2 layer, MoS 2 films are peeled off from SiO 2 when cross-section TEM specimen preparation using polishing. To minimize mechanical damage, we carried out FIB TEM sample preparation followed by nanomilling (Fischione 1040). Nanomill is similar to precision ion polishing, except that the milling area can be selected during nanomill process. Thus, nanomill is the optimal method for the preparation of the FIB TEM specimens.

Raman, XPS, TEM and STEM/EELS characterization
Raman and X-ray photoelectron spectroscopy were carried out to analyze vibration modes and surface chemical states of the MoS 2 films using LABRAM HR Evolution and AXIS-Hsi. The Mo to S ratio of the MoS 2 films are calculated by dividing intensity or the area of each elements by factor of Mo or S elements [4].
The TEM analysis was divided into 2 steps. First, we obtained bright-field TEM and high-resolution TEM images to confirm the quality and thickness of the TEM specimen. In this step, we analyzed the MoS 2 film using JEOL JEM-2100F. In the next step, to investigate the atomic structure of the interfaces, Cs-corrected highresolution STEM images were obtained with a Cs-probe corrected TEM instrument (JEOL JEM ARM 200F). STEM imaging with a spherical aberration corrector provided clearer images with a spatial resolution of 80 pm. Thus, Cs-corrected HR-STEM imaging enabled the identification of elements as well as the location of atomic positions with high accuracy.

Theoretical calculation
For the calculations, the MoS 2 -SiO 2 heterostructure supercell was composed of 8 Si layers of SiO 2 , a monolayer of MoS 2 , and a 15 Å vacuum layer to prevent the interaction between layers. Both sides of the SiO 2 slab were reconstructed surfaces, which have a lower surface energy compared to a pristine surface according to a previous study [29]. Before constructing the heterostructure, the unit cells of SiO 2 and MoS 2 were first relaxed; the lattice parameters were a = b = 4.896 Å for SiO 2 and a = b = 3.161 Å for MoS 2 . Subsequently, along the x-y plane, SiO 2 in 2 × 2 lateral periodicity and MoS 2 in 3 × 3 lateral periodicity was stacked together. Considering that the monolayer MoS 2 is much more prone to deformation than bulk SiO 2 , the x-y plane lattice parameter of 2 × 2 SiO 2 was employed as that of a heterostructure supercell. This allowed the MoS 2 layer to expand along the x and y directions with a lattice mismatch of approximately 3.4%. In addition, the position of the SiO 2 layer was fixed during the relaxation of the heterostructure for the same reason mentioned above. All structural relaxations and free energy calculations were performed with the Vienna ab initio simulation package (VASP) [31,32] based on density functional theory (DFT) [33,34]. For the replacement of the core electrons, a projector augmented wave (PAW) [35,36] scheme was implemented. The exchangecorrelation energy was described through the generalized gradient approximation (GGA) using the Perdew-Burke-Emzerhof (PBE) functional [37]. The kinetic cutoff for the plane-wave basis was 400 eV. The Brillouin zone for 3 × 3 × 1 k-point sampling was constructed with a gamma-centered grid. For electronic self-consistency and force tolerance, criteria of 5-10 eV and 0.01 eV/A were applied.

The growth of the MoS 2 film and imaging of the MoS 2 / SiO 2 interface
The growth and transfer process of the MoS 2 film is shown in Fig. 1(a). The experimental details are explained in the Sect. 2.1. The growth of the AS-MoS 2 film was demonstrated using Raman spectroscopy, as shown in Additional file 1: Figure S1. The Raman spectrum showed two characteristic Raman vibration modes, E 1 2g and A 1g , which indicated that the MoS 2 film grew laterally on the Si/SiO 2 substrate.
Due to the weak interaction between the TMD film and the substrate, which hinders the observation of the structure of the interfaces when the TEM sample is prepared using mechanical polishing, we prepared a TEM specimen using a focused ion beam (FIB) followed by nanomill using a Fischione 1040 nanomill. Additional file 1: Figure S2 depicts the result using the prepared AS and TR-MoS 2 films, which are in line with the Raman spectrum. Figure 1(b) and (c) is the HAADF STEM image and shows that the MoS 2 film was laterally aligned on the amorphous silicon oxide layer. Since the SiO 2 growth template is not atomically smooth, the AS-MoS 2 /SiO 2 interface seems to be rough. Figure 1(c) is a contrastinverted ABF STEM image. As ABF STEM imaging is efficient in detecting light elements, especially sulfur, the atomic configuration in Fig. 1(d) shows clearer image contrast than that in Fig. 1(c). The TR-MoS 2 film also showed comparatively well laterally aligned MoS 2 sheets at the surface and in the interior of the layer as shown in Fig. 1(e-g). In addition, the TR-MoS 2 film showed no significant difference in the overall lattice structure and interlayer distance.

Interlayer distance of the AS and TR-MoS 2 films depending on number of layers
To quantify the effect of the amorphous SiO 2 layer on the atomic structure of the MoS 2 film, we plotted the changes in the out-of-plane distance at 16 different positions and calculated the average interlayer distance with error bars. Each measured distance of the AS and TR-MoS 2 films is depicted in Additional file 1: Figure S3. As shown in Fig. 2(a) and (b), at the AS-MoS 2 /SiO 2 interface, the interlayer distance was up to 3.7% shorter than that of the AS-MoS 2 film, whereas there was no significant change in the interlayer distance in the TR-MoS 2 film, as shown in Fig. 2(c) and (d). The huge error bar is not originated from the roughness of the of the SiO 2 substrate, but irregular S-O bond at the MoS 2 /SiO 2 interface, making the shape of the MoS 2 surface wavy. Despite of this roughness, the difference in tendency of interlayer distance of AS and TR-MoS 2 films is non-negligible. This suggests that when MoS 2 films are grown on SiO 2 via the sulfurization of the Mo film, SiO 2 not only plays a significant role as a growth template but also influences the atomic structure of the AS-MoS 2 film at the MoS 2 /SiO 2 interface. Considering that only van der Waals interactions occur between the MoS 2 layers in the bulk, the change in interlayer distance is attributed to the chemical bonding between AS-MoS 2 and SiO 2 .
Next, we carried out ab-initio calculations to suggest an atomic model based on STEM images. We compared the equivalent distance between SiO 2 (single crystal)-MoS 2 and MoS 2 -MoS 2 with the assumption that the MoS 2 /SiO 2 interface is S-O-terminated. As shown in Fig. 3(a), the shortest distance between SiO 2 -MoS 2 was calculated to be 3.16 Å, which was slightly longer than the MoS 2 -MoS 2 distance, 3.04 Å. Figure 3(b) illustrates the formation  Figure 3(b) shows that 3.16 Å provides the most stable formation energy in our calculation. Since the calculation assumed that SiO 2 was a single crystal and considered only the van der Waals interaction, the result of the calculation deviated from that of the experiment. This discrepancy suggests that there is a strong chemical interaction other than the van der Waals force between the AS MoS 2 film and the SiO 2 template. This is in contrast with the previous DFT study, which argued that the structure of MoS 2 is not affected by SiO 2 [28,29].

Chemical bonding at the MoS 2 /SiO 2 interface
MoS 2 films were analyzed using X-ray photoelectron spectroscopy (XPS) to determine the chemical composition and atomic ratios of the films on the SiO 2 /Si substrate. The core level spectra of Mo 3d and S 2p were recorded for the AS-and TR-MoS 2 films (Additional file 1: Figure S4a-d). Figure 4 shows that the AS-and TR-MoS 2 films were both deposited in a stoichiometric composition (Mo:S = 1:2) without any significant chemical shift. In addition, the XPS characteristics of the AS-MoS 2 film were not significantly different from those of the To identify the chemical bonding state at the MoS 2 / SiO 2 interface, we obtained and analyzed the EELS SL and O K-edge spectra. Unlike XPS, EELS can provide chemical information in a local area with a spatial resolution in the sub-nanometer range, thus facilitating the analysis of the MoS 2 /SiO 2 heterointerfaces. The TEM sample was thin enough for the noise in the EELS spectra to be minimized, and each spectrum was acquired at a distance of 0.6 nm. Figure 4(a) and (e) show the STEM HAADF image of both samples from which the EELS spectra were acquired; both films were found to be well attached to the substrate through the TEM sample preparation. Next, we compared the S L-and O K-edge spectra in the two above-mentioned samples to identify the bonding state of AS-MoS 2 on the SiO 2 /Si substrate. Figure 4(b) shows the changes in the S L-edge at 10 different positions, and Fig. 4(c) shows the difference in the S L-edge spectra at the MoS 2 /SiO 2 interface (red solid line) and the MoS 2 film (blue solid line). The two spectra showed differences not only in the intensity, but also in the overall edge structure. In addition, considering that the MoS 2 /SiO 2 interface was S-O-terminated, the difference of the S L-edge at the interface.Please provide complete details for the References The appearance of the peak at 168 eV indicates that the sulfur atoms were partially bonded to the oxygen atoms. The deviation from the previous study is due to the defects at the AS-MoS 2 / SiO 2 interface. In other words, at the MoS 2 /SiO 2 interface, the sulfur atom of the MoS 2 film had four nearestneighbor oxygen atoms from the amorphous SiO 2 growth template. Figure 4(d) shows the O K-edge at 10 different positions. The O K-edge also showed a peak shift at the AS-MoS 2 /SiO 2 interface compared to that of the interior of SiO 2 . Figure 4(c) shows that along with the S-O bond, several defects were also formed at the AS-MoS 2 /SiO 2 interface over the range of 1.2-1.5 nm. In addition, the negative peak shifts of the SL and O K-edges at the AS-MoS 2 /SiO 2 interface were due to an increase in the negative charge around the sulfur and oxygen ions.
The EELS edge spectra of the TR-MoS 2 /SiO 2 /Si heterostructure were also observed. Figure 4(g) and (i) present the S L-and O K-edges at 10 different positions. In contrast to those for the AS-MoS 2 sample, the sulfur and oxygen edge spectra exhibited no significant difference, as shown in Fig. 3(h) and (j). Since the TR-MoS 2 film was detached and transferred onto the SiO 2 /Si substrate, only van der Waals interactions existed between the MoS 2 and SiO 2 template.
Based on the EELS spectra of the AS-MoS 2 sample, we prepared an electron transition diagram (Fig. 5). The EELS core loss spectra showed the transition of electrons from the core state to the unoccupied state [30]. As shown in Fig. 5(a) and (b), the peak positions near 540 eV and 560 eV in the O K-edge spectra are attributed to the  [30]. Figure 5(a) shows negative peak shifts at the interface and the formation of an energy loss peak at 542 eV, which were due to an increase in the negative charge at the interface and the formation of the S-O bond, respectively. However, there is no peak shift in O K edge at the TR-MoS 2 /SiO 2 interface, as shown in Fig. 5(b). This indicates that when sulfur atoms are chemically bonded with oxygen in SiO 2 , excess negative charges accumulate at the interface, causing a negative peak shift of the S L-and O K-edges.
Combining all experimental results, the summary of our study is illustrated in Fig. 6. At the AS-MoS 2 /SiO 2 interface, the interlayer distance decreases due to the formation of S-O bonding, whereas there is no significant change in the interlayer distance at the TR-MoS 2 /SiO 2 interface, 12 which can explain the role of oxide templates such as SiO 2 and Al 2 O 3 on the large-scale growth of the MoS 2 film. In addition, our key findings play a role in enhancing the carrier mobility of the MoS 2 film, which can lead to the improved performance of devices [38][39][40][41][42][43].

Conclusions
We prepared MoS 2 films on SiO 2 /Si substrates and studied the effect of the amorphous SiO 2 layer on the atomic and electronic structure of the MoS 2 films. The interlayer distance of the AS-MoS 2 film exhibited a change at the AS-MoS 2 /SiO 2 interface, which was attributed to the formation of S-O chemical bonding at the interface. Through theoretical calculations, we confirmed the existence of a bonding state in addition to the van der Waals force, which was the dominant interaction between MoS 2 and SiO 2 . The formation of S-O bonding at the AS-MoS 2 /SiO 2 interface layer suggested that during CVD, the Mo thin film was not only sulfurized, but the sulfur atoms at the termination layer were also bonded to the oxygen atoms of the SiO 2 layer, preventing the formation of Si-S bonding and MoSi 2 (Fig. 6). Our key findings in the study are consistent regardless of the deposition techniques. In other words, the formation of S-O bonding occurs and interlayer distance between the AS-MoS 2 film and the substrate is affected by the SiO 2 growth template even if MoS 2 is deposited by MOCVD or other deposition techniques. This study not only provides a guideline on the relationship between the interfacial structure and electrical properties of MoS 2 thin film-based heterostructures and explains the role of oxides on the growth of MoS 2 films, but also shows that this kind of interfacial interaction is prominent when it comes to single layer MoS 2 which is generally used for a wide variety of devices.