Optimization of Mechanically Assembled Van Der Waals Heterostructure Based On Solution Immersion and Hot Plate Heating

Layers of two-dimensional material are bonded together by van der Waals force, as a result, there is no need to take into consideration of the lattice mismatch in the formation of heterojunction, which is endowed with the characteristics of simple stacking in method, free of limitation to the type of materials and diverse changes. However, although the Van Der Waals heterojunction is relatively easy to stack, it is still difficult to generate inter-layer coupling between the thin crystal layers that form the Van Der Waals heterojunction. In most cases, the stacked heterojunction is simply stacked together without any new effects. Therefore, the realization of heterojunction coupling is a difficult problem to be considered in the process of preparing Van Der Waals heterojunction. In this paper, a method based on solution immersion and hot plate heating is proposed to optimize the mechanical stacking of Van Der Waals heterojunctions. It is found that the heterojunctions prepared by normal mechanical stacking method are usually uncoupled before treatment, but they can be stably coupled after treatment. Our method, simple, fast with low-cost, has been repeatedly verified to have a high success rate of coupling, which is suitable for most experimental groups to use and reproduce.


Introduction
Two-dimensional materials have become a hot area of research in condensed matter physics after graphene is discovered. The two-dimensional material based inter-layers are bonded together by Van Der Waals force, as a result, it is easy to form many thin layers and monolayers by mechanical stripping method. At the same time, these thin layers and monolayers can also be stacked together using a micro-optical manipulator by mechanical stacking method, which are bonded together by Van Der Waals forces to form a van der Waals heterojunction [1]. Since there is no need for lattice matching between layers of the Van Der Waals heterojunction, so any two kind of materials can be bonded together in theory.
At present, the achievements pertinent to Van Der Waals heterojunction emerge one after another in an endless manner. The Van Der Waals heterojunction has become one of the most productive research objects in the field of two-dimensional materials from enhancement of graphene properties on  [2], to Interlayer excitons in double layer transition metal chalcogenides [3], and then to twisted Angle transition metal chalcogenide heterojunctions and magic angle grapheme [4][5][6] in recent years. However, it is not easy to achieve the coupling of the heterojunction in the experiment, and many experimental groups failed to achieve the coupling of the heterojunction in the process of studying the heterojunction. Therefore, how to realize the coupling of heterojunction is a long-standing experimental problem.
Currently, in order to enable the heterojunction coupling, there have been some processing methods, such as to remove the organic defective gum on the material surface in order to get a clean surface by using acetone [2], to get a better adhesion of the heterojunction through long time annealing [7], and to use hydrophilic/hydrophobic transfer of samples [8].These methods have been popularized and used to a certain extent with good results.
We have improved the traditional stacking method using PDMS as the stacking medium [9]. In the process of sample preparation using mechanical stacking method, the monolayer sample is prepared on PDMS and then transferred to silicon wafer. The sample is then cleaned with acetone, isopropyl alcohol and deionized water. After initial drying, the coupled heterostructures are stabilized by heating. Our improved method, with simple processing, is operable for all laboratories with low cost and easy repetition, which saves time and efforts. It is suitable for general experimental sites and can greatly improve the coupling probability of heterojunction. Taking the transition metal sulfur compounds (TMDs) as an example, this paper describes double transition metal sulfur compounds heterojunction by using our preparation method, and a new interlaminar exciton emission peak and a new interlaminar Raman model have been measured in the region of heterojunction, confirming that interlayer couplings exist in heterojunction after processed .We stack different types of heterojunctions by changing the types of transition metal chalcogenides, and the interlayer coupling phenomenon can be found in all of them, which proves that our experimental method is universal.

Experimental Results and Discussions
TMDs is a two-dimensional semiconductor material with direct band gap luminescence in a single layer. The chemical formula is Mx2, where M is a transition metal element, M=Mo, W;X is a chalcogenide element, X=S, Se, which is characterized with a hexagonal lattice structure ( Fig. 1  (a)).Coupling may occur when different TMDs monolayers are stacked using the mechanical stacking method, intercambium excitons [3] are made by overlapping of energy bands, as shown in Fig. 1 (c).However, in actual experimental operations, due to insufficient stacking or excessive fluctuation of the sample surface, the layer spacing between the upper and lower TMDs is too large, and the electronic states between the upper and lower TMDs fail to overlap, so the failure in coupling is generated. At this time, the heterojunction is in an uncoupled state [10]. The distortion and undulation of the thin layer samples during mechanical stacking are the important reasons for no coupling. As shown in Figure 2 (a), in the process of transferring samples to the silicon wafer by PDMS, the samples on the PDMS are generally intact and flat. However, in the process of fitting with the silicon wafer, there is a gap between the thin layer sample in general conditions and the substrate because PDMS fails to reach a complete and uniform contact with the silicon wafer [9].This effect persists as other thin layers of samples are transferred to the wafer for stacking, so that the thin layers of samples on the wafer do not touch each other well. The result is a heterojunction with poor coupling and poor contact. Our method is an optimization of the above mentioned mechanical stacking method using PDMS as the transfer medium. After preparing the thin layer sample on PDMS, we have to transfer the samples to the silicon chip, and it is soaked with acetone for 10 minutes to remove the residual glue brought by PDMS on the thin layer surface. Isopropyl alcohol was used for 3 minutes to remove acetone, and deionized water was used for 3 minutes to remove isopropyl alcohol. After soaking, air was used to gently blow the surface of the silicon wafer as soon as possible to slightly remove the deionized water of the silicon wafer to prevent residual water stains. After rapid blow-drying, there is still a certain amount of solution in the gap between the thin layer sample and the substrate, the sample and the substrate will adhere closely to the solution due to the presence of the solution. The solution gradually evaporate and dissipate by heating the hot plate. In this process, the material and the substrate will always maintain good contact with the solution until the solution evaporates, and the material can be combined with the substrate more evenly and closely. In order to verify the reliability of our experimental method, we initially prepared a relatively thick (hundreds of nm) but clean and flat hBN sample on the silicon wafer, as shown in Figure 3 (a).We stripped a single layer of WSe2 samples on the PDMS and stacked them on the hBN samples, as shown in Figure 3 (b).In can be found that there are areas of different contrast on the surface of hBN, which is the result of the bubbles and wrinkles due to stress, uneven bonding and other factors.We then soaked it with the three solutions described above, as shown in Fig. 3 (c), and we found that bubbles still existed on the surface of WSe2, with no visible difference from Fig. 3 (b).
However, after being heated by a hot plate at 150℃ for 3 minutes, the bubbles and wrinkles on the surface of WSe2 disappeared, as shown in Figure 3 (d), which indicated that the surface of WSe2 was greatly stretched and evenly fitted to the flat surface of hBN, so there were no bubbles and wrinkles. Then we continued to prepare single-layer MoS2 samples in PDMS and repeated the mechanical stacking process to stack MoS2 on the surface of WSe2, as shown in Figure 3 (e).We found that there were not many bubbles and folds in the WSe2-MoS2 heterojunction region at this time. As shown in Figure 4, we measured the PL spectrum (yellow line) of theWSe2-MoS2 heterojunction at this time, and it is discovered that the signal of the yellow line was equal to the simple superposition of the PL spectrum of the monolayer MoS2 (red line) and the PL spectrum of the monolayer WSe2 (blue line), and no PL peak of interlayer excitons was found, nor any other phenomena were found.
We soaked the prepared heterojunction in solution, as shown in Fig. 3 (f), and we found that the bubbles and folds increased, indicating that the solution immersion made the bonding between WSe2 and MoS2 closer. We repeated the measurement in Fig. 3 (e) for the sample in Fig. 3 (f), and there was no significant difference in PL spectra before and after immersion in the solution. We put the soaked  Fig. 3 (g). it is shown as: Figure 4. MoS2, WSe2 and PL spectra of WSe2-MoS2 heterojunction before and after heating. Among them, MoS2 is 1/3 of the actual luminescence intensity, and WSe2 and WSe2-MoS2 heterojunction before heating is 1/40 of the actual luminescence intensity The bubbles and wrinkles are slightly darker in color, indicating that the the solution (deionized water) is removed in heating process, which allows the upper and lower layers of the sample to bond more closely. According to the characteristics of heterojunction coupling that have been mentioned, a large number of bubbles and wrinkles tend to appear at the interface of heterojunction coupling. We re-measured the PL spectrum of the WSe2-MoS2 heterojunction region, as shown in the green line in Figure 4, and found that the the following changes occurred in PL spectrum: 1.The luminescence intensity is equal to 1/40 of that of WSe2-MoS2 heterojunction before heating, and the luminescence intensity is greatly reduced.2. Interlayer exciton signals are generated, and the luminescence intensity is much higher than that of monolayer WSe2 and MoS2.These phenomena are consistent with the reports of interlaminar excitons in the double-layer TMDs heterojunction. To further verify our conclusion, we also measured the Raman spectra of monolayer MoS2, monolayer WSe2, WSe2-MoS2 heterojunction before heating, and WSe2-MoS2 heterojunction after heating, as shown in Fig. 5.The single-layer WSe2 enjoys a degenerate 2g E' and 1g A characteristic Raman mode (255cm -1 ).Single layer MoS2 is characterized with a 2g E' mode (387cm -1 ) and a 1g A mode (405cm -1 ).We found that the Raman mode of WSe2-MoS2 heterojunction before heating was basically the same as the simple superposition of Raman mode of single-layer WSe2 and MoS2 in the whole wave-number range, while the WSe2 E' mode of WS2-MoS2 heterojunction after heating showed an obvious red shift (about 5cm -1 ) compared with that of single-layer WSe2, as shown in Fig. 5 (a).At the same time, a 2LA(M) pattern near 260cm -1 exists for WSe2, which produces a red shift (about 7cm -1 ) in the heated WSe2-MoS2 heterojunction. This shows evidence that WSe2 is affected by MoS2.
Most notably, two new Raman modes appeared in the WSe2-MoS2 heterojunction at the 280-320cm -1 wave-number segment. The new mode with a wave number of 284cm -1 is a E '' mode that only appears in the odd-numbered multilayer MoS2 [11]. In the even-numbered multilayer MoS2, this mode will be transformed into a 1 E' g mode due to the change of symmetry, which will not appear in the single-layer MoS2.The new mode with a wavenumber of 309cm -1 is the Raman 1 2 B g mode of WSe2 [12]. It is an out-of-plane Raman pattern that only exists in two or more layers of WSe2.Therefore, we can confirm that inter-layer coupling does exist betweenWSe2 and MoS2. For the wave-number segment of single-layer MoS2 characteristic Raman mode, the wave-number of single-layer MoS2 is the same as that of WSe2-MoS2 heterojunction before heating, but after heating, both 2g E' modes and modes 1g A appear red shift. The change of 2g E' mode of MoS2 after heating also verifies the existence of inter-layer coupling between MoS2 and WSe2. We also changed different TMDs materials to stack MoS2-WS2 heterojunction and WS2-WSe2 heterojunction for repeated experiments, as shown in Figure 6.We found the presence of interlayer excitons in the low energy region of PL spectrum in the final heterojunction region. Therefore, our experimental method is universal and effective for different types of materials. A hot plate is heated and treated in the same way as annealing, but the physical process is completely different. According to the reported work, annealing treatment is generally carried out in the environment of 100-300℃ and high vacuum, and the duration lasts usually 3-12 hours, and the strategy of low temperature/long time natural cooling is usually adopted. Annealing treatment can be considered to provide energy to the heterogeneous junction which is not completely stacked. In the continuous heating process, the lattice slowly stretches, passes the energy barrier from one system to another, and slowly changes from a disordered stacked state to an orderly flat contact state. Therefore, long heating and high vacuum environment are required. But the physical process of combination of solution immersion and hot plate heating are quite different.
The solution acts as a adhesive, transforming the thin layers of the upper and lower layers from untightly bonding to tightly bonding. It can quickly evaporate from the gap between the upper and lower layers during a brief heating process. But, until it evaporates, it continues to hold the upper and lower layers together through the tension of the liquid. Therefore, high vacuum is not required in heating process, and nor several hours of heating is required. Only a short period of low temperature heating is sufficient to complete the process of heterojunction uncoupling to coupling.  Figure 6. (a) MoS2-WS2 heterojunction and its interlayer excitons after processed;(b)WS2-WSe2 heterojunction and its interlayer excitons after processed

Conclusion
This paper introduces a simple mechanical stack based optimization method, the methods of solution soaking and hot plate heating is combined with the PDMS based mechanical stacking method. Taking the two-layer TMDS heterojunction as an example, it is proved that our method can effectively realize the interlayer coupling of heterojunction, and the time and consumption are made with relatively low cost, so it is suitable for the ordinary heterojunction stacking process.