**3. Substrates**

A typical characteristic of 2D materials is the very weak interlayer Van der Waals forces, in contrast with the stronger in-layer chemical bonds, which are ionic or covalent. This is the reason why bulk crystals of TMD materials, and MoS<sup>2</sup> in particular, can be easily cleaved and exfoliated to produce few-layer flakes. As a consequence, the dangling bonds of the 2D plane are relatively inactive, and few interactions occur out-of-plane. For this reason, one could argue that in the 2D synthesis, the substrate's role should be less important with respect to standard epitaxy, in which lattice and thermal matching conditions between the substrate and the epilayer are of paramount importance, determining whether defects are generated at the interface. The weak interactions between the substrate and the 2D layer reduce the importance of lattice mismatch, but the choice of the substrate remains an important parameter, not to be overlooked. The reason should be sought in the processes still occurring at the substrate surface that involve the reactive species, such as adsorption, diffusion, and nucleation. It was indeed observed that the substrate played an important role in influencing both the growth process and the properties of 2D layers, as evidenced in the excellent review on this topic in [36].

SiO2/Si substrate is the most commonly used for CVD synthesis of MoS<sup>2</sup> because of: (i) the ease of observing and characterizing flakes on it through optical microscopy (from this point of view, the thickness of the oxide layer is of paramount importance [37]); (ii) its high melting point; (iii) its compatibility with silicon processing. On the other hand, for further characterization and/or applications, it is necessary to transfer the 2D material from the SiO2/Si substrate on other substrates or TEM grids. This implies a delicate chemical process composed of various steps: (i) coating of the material with poly(methyl methacrylate) (PMMA) film on the MoS<sup>2</sup> layers by spin coating, (ii) etching of the SiO<sup>2</sup> layer to separate it from the Si substrate in a solution (KOH, HF, NaOH), (iii) transfer of

the floating PMMA/MoS<sup>2</sup> film from the surface of the solution to the substrate of interest, and (iv) removal of the PMMA coating by rinsing in acetone and isopropanol [38].

The possibility of growing MoS<sup>2</sup> on rigid metals has been investigated, in particular for the case of gold, as it was demonstrated that excellent nanostructures can be obtained on Au foils [39]. If one considers surfaces of Au(111), very large nanoflakes can be obtained [40] thanks to the lattice matching between Au(111) and MoS<sup>2</sup> and the fact that the step edge of Au(111) favours the unidirectional nucleation of the film. The advantage of this substrate was confirmed recently in [41], where a unidirectional growth ratio over 99% of MoS<sup>2</sup> flakes was achieved on Au(111) films sputtered on c-sapphire. The growth on Ag(111) was also demonstrated, opening up interesting perspectives on the realization of metal contacts on this 2D material [42].

Liquid metals are interesting, as their uniform and smooth surface can inhibit the inhomogeneous nucleation of 2D materials. Although convincing results have been published on other 2D materials, the growth of MoS<sup>2</sup> on these substrates has been reported only for the MoS2/h-BN heterojunction grown on liquid metal Ni–Ga alloy [43].

MoS<sup>2</sup> has also been grown with considerable success on c-sapphire, as it has a stable hexagon single crystal structure, matching the lattice symmetry of 2D material [27,44,45]. Sapphire surface can be prepared with a thermal treatment at 1100 ◦C before MoS<sup>2</sup> growth to obtain a clean and atomically flat surface. Scaling up to a Metal–Organic Vapour Phase Epitaxy (MOVPE) process for 6-inch wafers was obtained, and the possibility of transferring the obtained flakes to a secondary carrier substrate opens up the possibility to reuse the sapphire substrate, making this choice an interesting alternative to the common SiO2/Si approach [46]. Very recently, the growth of monolayer flakes as large as 50 µm on a Cplane sapphire was demonstrated by designing the miscut orientation towards the A axis. This resulted in a break in the degeneracy of nucleation energy for the antiparallel MoS<sup>2</sup> domains [47].

The growth of a 6-inch uniform monolayer of MoS<sup>2</sup> on solid soda-lime glass was achieved because of the advantageous effect of sodium in enhancing the formation of flakes that were homogenously distributed in glass [48]. Na acts as a growth promoter; its effects are discussed in depth in a subsequent section

Recently, the possibility of growing MoS<sup>2</sup> on flexible substrates such as polyimide was reported [49]. The issues related to the stability of the substrate were solved thanks to a CVD growth protocol allowing for low temperatures (<300 ◦C), and the feasibility of developing flexible gas sensors taking advantage of the properties of MoS<sup>2</sup> was demonstrated.

In an interesting work, MoS<sup>2</sup> monolayers were reproducibly deposited on a variety of substrates, amorphous (SiO2/Si and fused quartz), crystalline (bare Si and sapphire), and layered-flexible (mica), using the same growth conditions with solid precursors and mixed NaCl [50].

CVD growth of MoS<sup>2</sup> on gallium nitride (GaN) was also reported, a very relevant step towards the realization of heterostructures for many applications [51]. This possibility is of particular interest because it would eliminate the delicate step of detachment and transfer of 2D flakes, allowing realizing MoS<sup>2</sup> heterojunctions directly on a nitride template.

Following the ideas widely explored with III-N and III-V semiconductors, in which AlGaAs–InGaAs or AlGaN–InGaN heterostructures are used to exploit the possibility of bandgap engineering to design advanced devices, an emerging topic of research is the realization of so-called 2D Van der Waals heterostructures. By vertically stacking layers of different 2D materials coupled by very weak forces, it would be possible to avoid the presence of defects due to lattice mismatch observed in conventional 3D heterostructures [52]. A considerable effort has thus been devoted to exploring the synthesis of MoS<sup>2</sup> and other 2D compounds onto graphene and other 2D materials. First results highlighted how it was possible to obtain high-quality 2D flakes by adding some gases to gas carriers (N or Ar) for the treatment of graphene surfaces, such as H<sup>2</sup> [53] or ozone [54]. This allows for the preparation of high-quality nanoflakes of MoS<sup>2</sup> because of the reduction in the oxidation of graphene during the growth process.

In [55,56], few-layer MoS<sup>2</sup> was grown on graphene oxide films or flakes. These results suggested that carbon-based materials can significantly promote the growth rate and yield of MoS2. In [57], the differences in using graphene, sapphire, and SiO2/Si substrates were investigated, highlighting the role of surface diffusion mechanisms in determining different properties of MoS<sup>2</sup> flakes. The authors remarked that the growth on graphene was very stable, resulting in the realization of a strain-free 2D layer.

A topic of paramount importance for the development of nanodevices is the growth on patterned substrates to increase the spatial control of MoS<sup>2</sup> nanoflakes and to obtain a controlled nucleation only on certain substrate zones. The first pioneering work on this issue was the one by Najmaei et al. [58], who observed a catalytic effect of the edges of the substrate. Based on this effect, authors proposed and demonstrated a method to control the growth of 2D flakes using lithography to pattern the substrate. Later, it was demonstrated that, by using patterned seeds of molybdenum, it was possible to obtain flakes of MoS<sup>2</sup> at predetermined locations with a good spatial resolution [59]. Another approach to control the nucleation site of MoS<sup>2</sup> flakes was to use droplets of (NH4)2MoS<sup>4</sup> in dimethyl-formamide suspended from the tip of a micropipette that is dragged across a sapphire single-crystalline substrate by a controlled substrate movement. This resulted in MoS<sup>2</sup> films with alternating mono- and few-layer regions that had distinct optical properties [60].

Other metal seeds proven as very valuable for the spatial control of the nucleation of flakes include Pt/Ti and Au [61–63]. More recently, the synthesis of patterned MoS<sup>2</sup> nanoflakes using an industrial inkjet fed with an aqueous solution of ammonium molybdate tetrahydrate as liquid precursor was demonstrated (Figure 3). The possibility of obtaining large-area patterned 2D films in centimetre size with good controllability of the thickness and good reproducibility opens up interesting possibilities for the development of complex devices based on this material [64].

**Figure 3.** Synthesis of MoS<sup>2</sup> flakes from inkjet-printed aqueous precursors: (**a**) a photo of the customized inkjet printer; (**b**) schematic of the synthesis process with the fast annealing process; (**c**–**f**) growth of flakes at different growth temperatures; (**g**–**l**) corresponding optical images. Adapted from [64]. Copyright 2021 John Wiley & Sons.
