Layer-controlled fractal growth of vanadium-doped molybdenum disulfide

Abstract

Vanadium doping can effectively modify the physical properties of transition metal dichalcogenides (TMDCs), broadening their application prospect in electronic, spintronic, and valleytronic devices. However, vanadium dopants always suppress the growth of TMDCs and lower the growth controllability. Here, we report the morphology-controlled growth of vanadium-doped MoS₂ (V-MoS₂). Both the layer-controlled growth and fractal growth of monolayer V-MoS₂ were realized, and the mechanism was analyzed, including contributions from the localized concentration of Mo atoms, the etching effect, and formation energies of different terminations. It will shed light on the morphology-controlled growth of other vanadium-doped TMDCs, promoting the construction of novel devices.

GRAPHICAL ABSTRACT

IMPACT STATEMENT

Controll vanadium dopants enables layer-controlled and fractal growth of V-MoS2 in this work with contributions from the localized concentration of Mo atoms, the etching effect, and formation energies of different terminations. It will help to realize the morphology-controlled growth of other vanadium-doped TMDCs and thus promotes the construction of novel devices.

KEYWORDS:

• Molybdenum disulfide
• doping engineering
• layer controlled
• fractal growth
• mechanism analysis

Introduction

Doping engineering has become one of the most powerful methods to precisely modify the physical properties of two-dimensional transition metal dichalcogenides (2D TMDCs) [1,2]. It can further boost the performance of electronic devices based on these atomically thin materials and thus continue Moore's law [3–5]. Complicated electronic and optoelectronic devices have been fabricated based on TMDCs [6–9], which can integrate sensing, data storage, and computing [10–12]. Novel spintronic and valleytronic devices are also constructed owing to the unique coupled spin and valley physics of TMDCs, including the strong spin–orbit coupling [13], large valley spin splitting [14], Zeeman spin polarization [15], and valley Hall effect [16]. However, these complicated or novel devices raise many requirements to 2D TMDCs, such as the demand for regular morphology, p-type transfer characteristic, and magnetism, to realize the easy fabrication of nanodevices [17], the construction of complementary inverters [18], and the integration of spintronic and magnetoelectric devices [19].

However, the concurrence of p-type transport behavior and magnetism is rarely observed in 2D semiconductors. The only several p-type 2D semiconductors, e.g. WSe₂, black phosphorous (BP), and b-AsP are all non-magnetic [20]. Substitutional vanadium doping provides an effective approach to solve this problem. It can not only convert n-type TMDCs into p-type but also endow them with magnetism [21–24]. Such magnetic semiconductors are the ideal platform for the construction of spintronic and valleytronic devices [25–27]. All these effects make vanadium superior to other elements, such as manganese and niobium [28,29]. Moreover, vanadium doping can also extend the functionality of TMDCs. For example, synaptic behaviors can be well mimicked by the transistor based on the vanadium-doped MoS₂ (V-MoS₂) without any complex device design [30]. V-MoS₂ and other vanadium-doped TMDCs (V-TMDCs) have also become an ideal platform for the study of spintronic, valleytronic, and other multifunctional devices [25,31,32]. V-MoS₂ can even extend its application in bio-electronic devices with further surface modifications just like some existing material systems [33–35]. The controlled growth of V-MoS₂ and other V-TMDCs is the prerequisite for these novel applications. However, vanadium dopants usually suppress the chemical vapor deposition (CVD) growth of TMDCs, reduce growth controllability, and cause the formation of dendritic or irregular flakes. So, although the doping concentration of vanadium in V-MoS₂ has been well controlled [30], the morphological control of heavily doped V-MoS₂ has not been systematically studied yet, which will bring some challenges to the fabrication of novel devices.

In this work, the morphology-controlled growth of V-MoS₂ has been realized by the careful adjustment of vanadium dopants. First, by adjusting the composition of vanadium dopants, monolayer and bilayer V-MoS₂ flakes were grown with mono- and bi-compositional dopants, respectively. Three effects, including the adsorption of V atoms, the localized concentration variation of Mo, and the etching effect, contributed to the layer-controlled growth of V-MoS₂. Second, by adjusting the mass ratio of Mo precursor and V dopants, monolayer V-MoS₂ flakes with different morphologies were synthesized following a fractal growth model, which is caused by the change in the formation energy of different terminations. The morphology-controlled growth will facilitate the device application of this vanadium-doped 2D semiconductor.

Materials and methods

Growth of V-MoS₂

V-MoS₂ flakes were grown by using the method mentioned in our previous works [21,30]. MoO₃ and sulfur powder were used as the growth precursors while VCl₃, NH₄VO₃, and V₂O₅ were adopted to provide the dopant atoms. All the V-MoS2 flakes were grown at 750 °C for 5 min with the protection of Ar (80 sccm). The morphology and layer number of the grown V-MoS₂ flakes were obtained with the optical microscopes. The atomic structure of the typical V-MoS₂ was characterized by a high-angle annular darkfield scanning transmission electron microscopy (HAADF-STEM). Raman and photoluminescence (PL) characterizations were also performed to analyze the doping effect of the V-MoS₂. Besides, the spin-polarized DFT calculations were conducted with the Cambridge sequential total energy package (CASTEP) using the Perdew–Burke–Ernzerhof (PBE) generalized gradient approximation (GGA) functional to calculate the edge formation energy of MoS₂ and V-MoS₂ [36].

Results and discussion

V-MoS₂ flakes with different morphologies were grown on the silicon substrate with the usage of sulfur, vanadium dopants, and MoO₃ powders as schematically shown in Figure 1(a), and their crystal structure is illustrated in Figure 1(b). The substituted vanadium atoms are clearly distinguished at Mo sites due to their low contrast as compared to the surrounding Mo atoms in the HAADF-STEM image (Figure 1(c)). The corresponding atom-by-atom map reveals the uniform distribution of V atoms (Figure 1(d)). Figure 1(e) shows the magnified region marked in Figure 1(a) and the corresponding line profile of the Z-contrast. The observed two low peaks confirm the substitution of two neighboring V atoms at Mo sites.

Figure 1. Growth and characterization of V-MoS₂. (a) Schematic of the growth setup. (b) Atomic structure diagram of V-MoS₂. (c) HAADF-STEM image of VCl₃-doped V-MoS₂ and (d) the corresponding atom-by-atom map to show the distribution of V atoms. (e) The magnified HAADF-STEM image of the area marked in (c) and the corresponding profile of Z-contrast intensity. (f) Raman spectra of undoped and VCl₃-doped MoS₂. (g) PL spectra of undoped and VCl₃-doped MoS₂.

Raman spectroscopy, which has been confirmed as an effective method to analyze the V-doping effect in TMDCs [30], was used to analyze the obtained V-MoS₂ flakes. For undoped MoS₂, only two Raman active modes, E_2g and A_1g, are observed in Figure 1(f), whereas many new active modes appear for V-MoS₂, including the defect-related peaks at the low wavenumber region (100–250 cm⁻¹), peaks split from the E_2g and A_1g modes (350–425 cm⁻¹), and the characteristic peak of V-MoS₂ at 323 cm⁻¹. These strong newly emerged peaks indicate the high doping concentration in the V-MoS₂ as all these peaks enhanced with the increasing doping concentration (Figure S1a). The high doping concentration is also confirmed by the quenched photoluminescence as PL intensity decreases with the doping concentration (Figure 1(g) and Figure S1b). Meanwhile, p-type transport behavior emerges and is enhanced with the increasing doping concentration (Figure S2).

The layer number of as-grown V-MoS₂ flakes was regulated by changing the composition of the V precursor. Monolayer V-MoS₂ flakes were grown using mono-compositional V precursor, either V₂O₅, NH₄VO₃, or VCl₃ (Figure 2(a)). Similar results have been reported in our previous works [21,30]. Bilayer V-MoS₂ flakes were synthesized with a bi-compositional dopant — the mixture of VCl₃ and V₂O₅, and their morphology was modified by changing the mass ratio of VCl₃ and V₂O₅. A hexagonal top layer was formed with a ratio of 2:1 (Figure 2(b)), whereas a three-bladed top layer grew when the mass ratio exceeded 4:1 (Figure 2(c)). The HAADF-STEM image depicts the crystal structure of a bilayer V-MoS₂ flake with a hexagonal top layer (Figure 2(e)). Substituted V atoms can be clearly distinguished in the right monolayer region. The observed edge of the top layer on the left is terminated with Mo-zz, and it is located above the bottom layer with its lattice being translated by about half the side length of the hexagon along the armchair direction, i.e. 30° left to the vertical direction as marked by the arrow. All these features are depicted in Figure 2(e). The dramatic change of the Z-contrast, from overlapped two Mo atoms (Mo2) to a single Mo atom (Mo1), confirms the sharp edge of the top layer (Figure 2(f)). There is only one S atom neighboring to Mo2, which confirms the Mo-zz termination of the top layer. However, the other neighboring edge of the top layer is terminated with S-zz as there are two S atoms neighboring to Mo2 (Figure S3). Therefore, it can be concluded that the hexagonal top layer is terminated with three Mo-zz and three S-zz terminations.

Figure 2. Layer-controlled growth of V-MoS₂. (a) Typical monolayer V-MoS₂ flake. (b) Bilayer V-MoS₂ flake with a hexagonal top layer. (c) Bilayer V-MoS₂ flake with a three-bladed top layer. (d) HAADF-STEM image to show one edge of the hexagonal top layer, (e) the corresponding atomic structure, and (f) the corresponding Z-contrast intensity profile of the marked region in (d). (g) Schematic to show the growth process of the bilayer V-MoS₂ flake. The blue line and yellow line correspond to the Mo-zz and S-zz edges, respectively.

Based on these observations, we put forward the growth process of the bilayer V-MoS₂ flakes. As the sulfur powder is over-added, the bottom V-MoS₂ layer with S-zz terminations is synthesized in the S-rich condition at first (step 1), and then the hexagonal second layer (top layer) nucleates (step 2) and grows up (step 3), which is terminated with Mo-zz and S-zz (Figure 2(g)). The growth of the top layer is motivated by the active gas-phase VOCl₃ (melting point of 127 °C) and MoO₂Cl₂ adsorbed on the surface of the bottom layer (Figure 2(b)). VOCl₃ and MoO₂Cl₂ are formed through the following reactions [37–39]: (1) $2\text{VC}{\text{l}}_3\left(\text{s}\right)\to 2\text{VC}{\text{l}}_2\left(\text{s}\right)+\text{ C}{\text{l}}_2\left(\text{g}\right)$(1) (2) $2{\text{V}}_2{\text{O}}_5\left(\text{s}\right)+6\text{C}{\text{l}}_2\left(\text{s}\right)\to 4\text{VOC}{\text{l}}_3\left(\text{g}\right)+3{\text{O}}_2\left(\text{g}\right)$(2) (3) $2\text{Mo}{\text{O}}_3\left(\text{s}\right)+2\text{C}{\text{l}}_2\left(\text{s}\right)\to 2\text{Mo}{\text{O}}_2\text{C}{\text{l}}_2\left(\text{g}\right)+O_2\left(g\right)$(3)

The adsorbed VOCl₃ facilitates the nucleation of the top layer while MoO₂Cl₂ increases the localized Mo concentration and causes the growth of the hexagonal top layer. To verify this, a simple experiment was performed. 5 or 10 mg pure MoO₃ were heated at 750 °C for 5 min. 2.2 or 4.7 mg residue remained with the addition of 10 mg VCl₃ (MoO₃ and VCl₃ were separately loaded in two neighboring tiny corundum boats) while 3.4 or 7.2 mg residue remained without VCl₃. The weight loss of MoO₃ reaches 53–56 wt% with the introduction of VCl₃ while it is only 28–32 wt% without VCl₃ (Figure S4). This result reveals that VCl₃ can promote the evaporation of MoO₃, the formation of MoO₂Cl₂, and thus increase the localized Mo concentration.

The released Cl₂ reacts with the formed V-MoS₂ by etching V-MoS₂ from the side of Mo-zz terminations (step 4) until being hindered by S-zz terminations (step 5) because of the lower stability of Mo-zz terminations. The etching process follows the reaction equation [40,41]: (4) $\begin{array}{rl}& 2\text{Mo}{\text{S}}_2\left(\text{s}\right)+7\text{C}{\text{l}}_2\left(\text{g}\right)\to 2\text{MoC}{\text{l}}_5\left(\text{s}\right)+2{\text{S}}_2\text{C}{\text{l}}_2\left(\text{l}\right)\end{array}$(4) To confirm this etching reaction, the as-grown monolayer V-MoS₂ flakes were put above VCl₃ and faced down. Then, they were heated up to 500 °C and maintained at this temperature for 5 min. Even the triangle V-MoS₂ flakes terminated with S-zz were etched into a similar three-bladed morphology along the Mo-zz terminations (Figure S5). A start-up time (≥ 1 min) is needed to get a certain concentration of Cl₂ and thus to initiate the VCl₃-induced etch of V-MoS₂. Then, the etch depth, ratio, and rate increased monotonously with the prolonged etch time (Figure S6), which facilitates the growth of the three-bladed top layer (Figure 2(c)). In contrast, V-MoS₂ flakes heated without the addition of VCl₃ remained unchanged. This difference supported that the competition between etching and growth causes the formation of the three-bladed top layer observed in Figure 2(d).

Fractal growth of monolayer V-MoS₂ flakes with different morphologies is realized by changing the additive amount of VCl₃. As shown in Figure 3(a–c), with the increasing amount of VCl₃, the shape of the obtained V-MoS₂ flakes evolves from a regular triangle to a three-pointed star at first, and then a branch grows from one corner of the three-pointed star with an angle of 60°. With the further increasing additive amount of VCl₃, more branches emerge, and the overall shapes of these flakes change from triangle to hexagram (Figure 3(d,e)). This is because that the formed MoO₂Cl₂ increases the localized Mo concentration and turns the S-rich condition into a medium or even localized Mo-rich condition, which facilitates the growth of hexagonal flakes. More importantly, all the branches form 60° or 120° angles with the trunk, which shows the fractal growth model of V-MoS₂. Fractal dimension (D) is calculated to evaluate the complexity of the flake morphology by using the box-counting method (Figure S7). D increases monotonously with the increasing mass ratio of VCl₃:MoO₃ (Figure 3(g)). This result quantitively reflects that the morphological complexity of V-MoS₂ flakes can be well controlled by precisely adjusting the added amount of V precursors.

Figure 3. Fractal growth of monolayer V-MoS₂ flakes. (a–f) Morphology evolution of monolayer V-MoS₂ flakes grown with the increasing mass ratio of VCl₃:MoO₃ from (a) 1:4, (b) 2:5, (c) 1:2, (d) 3:5, (e) 4:5, and finally to (f) 1:1. (g) Relationship between D and the mass ratio of VCl₃:MoO₃.

We then analyzed the mechanism behind the fractal growth of monolayer V-MoS₂ flakes. As S-zz termination is the most stable in S-rich conditions, a triangle nucleus with S-zz terminations forms at first and then grows up into a triangle flake when the mass ratio of VCl₃:MoO₃ ≤ 0.4 (Figure 4(a) 1→2). However, the substituted V atoms change the edge formation energy of different terminations and facilitate the growth of Mo-S-ac terminations. So, V-MoS₂ with Mo-S-ac edges grows out from S-zz terminations of the triangle nucleus (Figure 4(a) 1→3) to form the three-pointed star, and then thinner branches stretch out from one corner of the three-pointed star to start the fractal growth model (Figure 4(a) 3→4).

Figure 4. Fractal growth of V-MoS₂ flakes. (a) Schematic of the fractal growth process. (b) HAADF image of the folded edge of a V-MoS₂ flake, in which the folded bilayer region forms a Moiré pattern. (c) Schematic to show the corresponding arrangement of Mo atoms. (d) The corresponding lattice structure of the V-MoS₂. The right region shows the bottom layer while the left region shows the top layer with a well-distinguished Mo-S-ac termination. The orange line and green dash line correspond to the S-zz and Mo-S-ac edges, respectively.

Two main positive evidences can support this growth mechanism. The first is the observed Mo-S-ac termination. The edges of the three-pointed-star V-MoS₂ flake were folded during the transfer process (Figure S8) to form a Moiré pattern (Figure 4(b)). The edge is tilted by about 30° left to the vertical direction. The corresponding arrangement of Mo atoms is shown in Figure 4(c). After adding sulfur atoms and completing the lattice structure, the 30°-oriented top layer, i.e. the edge of the three-pointed-star V-MoS₂ flake, can be identified to be ended with a Mo-S-ac termination (Figure 4(d)).

The second evidence is the edge formation energy (E_e) caused by vanadium doping. For undoped MoS₂, Mo-zz terminations have the lowest edge formation energy of −0.39 eV/Å, followed by the Mo-S-ac termination (−0.36 eV/Å), while the S-zz termination has the highest edge formation energy (−0.35 eV/Å) in the Mo-rich condition (Figure 5(a–c)). As a result, Mo-zz terminations form with the high localized Mo concentration, causing the growth of the hexagonal top layer observed in Figure 2(b). This is consistent with the results reported in other works [42,43]. For V-MoS₂, E_e of Mo-S-ac and S-zz terminations become more negative, and Mo-S-ac terminations become energetically optimum (Figure 5(d–f) and Figure 5(g–i)). This causes the growth of Mo-S-ac terminations, promotes the fractal growth of V-MoS₂, and facilitates the formation of three-pointed-star flakes. Besides, the E_e of Mo-zz terminations increased from −0.39 eV/Å for undoped MoS₂ to −0.26 eV/Å for V-MoS₂ with V atoms at the edge and then to −0.23 eV/Å for V-MoS₂ with V atoms at the edge and inside the bulk phase. This result indicates that the Mo-zz termination becomes unstable with the high localized Mo concentration, which causes the etching of V-MoS₂ and the formation of the three-bladed flake mentioned above. Similar etching and termination-dominated growth of 2D TMDC is also reported by other researchers [44,45]. In general, the substituted vanadium atoms change terminations from S-zz in the sulfur-rich condition to Mo-S-ac in the localized Mo-rich condition and cause the formation of both the three-bladed and three-point-star flakes. It provides an effective and universal approach to fulfill the morphology-controlled growth of other doped TMDCs.

Figure 5. DFT calculation results of the edge formation energies of the undoped MoS₂ and V-MoS₂. (a–c) Undoped MoS₂ with (a) S-zz, (b) Mo-zz, and (c) Mo-S-ac terminations. (d–f) MoS₂ with V atoms at the edge terminated with (d) S-zz, (e) Mo-zz, and (c) Mo-S-zz. (g-i) MoS₂ with V atoms at the edge and inside the bulk phase terminated with (g) S-zz, (h) Mo-zz, and (i) Mo-S-ac.

Conclusion

In summary, we have realized the layer-controlled fractal growth of V-MoS₂ by controlling the vanadium dopants. Monolayer and bilayer V-MoS₂ flakes were grown with the use of mono- and bi-compositional dopants, respectively. Besides, the fractal growth of monolayer V-MoS₂ flakes was realized by adjusting the mass ratio of MoO₃:VCl₃. Vanadium dopant (VCl₃) can cause the change of the localized Mo concentration and the etching of MoS₂, and the substitutionally doped vanadium atoms can change the edge formation energy of different terminations. These effects prompt the growth of Mo-S-ac terminations and the etching of Mo-zz terminations, contributing to the morphology control of bilayer flakes and the fractal growth of monolayer flakes. This work will shed light on the morphology-controlled growth of V-TMDCs and provide appropriate platforms for the study of spintronic, valleytronic, and other multifunctional devices.

Acknowledgments

The authors acknowledge the support from the National Natural Science Foundation of China (No. 52202044), the Suzhou Science and Technology Program for Industrial Prospect and Key Technology (No. SYC2022018), the Startup Program of Suzhou University of Science and Technology (No. 332114704 and 342234701).

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by the National Natural Science Foundation of China: [Grant Number No. 52202044]; the Suzhou Science and Technology Program for Industrial Prospect and Key Technology: [Grant Number No. SYC2022018]; the Startup Program of Suzhou University of Science and Technology: [Grant Number No. 332114704 and 342234701].