Regulation of Two-Dimensional Lattice Deformation Recovery.

The lattice directly determines the electronic structure, and it enables controllably tailoring the properties by deforming the lattices of two-dimensional (2D) materials. Owing to the unbalanced electrostatic equilibrium among the dislocated atoms, the deformed lattice is thermodynamically unstable and would recover to the initial state. Here, we demonstrate that the recovery of deformed 2D lattices could be directly regulated via doping metal donors to reconstruct electrostatic equilibrium. Compared with the methods that employed external force fields with intrinsic instability and nonuniformity, the stretched 2D molybdenum diselenide (MoSe2) could be uniformly retained and permanently preserved via doping metal atoms with more outermost electrons and smaller electronegativity than Mo. We believe that the proposed strategy could open up a new avenue in directly regulating the atomic-thickness lattice and promote its practical applications based on 2D crystals.

Introduction

The lattice of a crystal is its most critical characteristic for determining the electronic structure and further enables controllably tailoring the performed properties (Lee et al., 2005, Weiss, 2010). There is sustained blooming of lattice deformation technology in the broad-range fields of electronics (Zhu et al., 2015, Wu et al., 2014), optics (Wu et al., 2016), magnetics (Levy et al., 2010), catalysis (Li et al., 2016), and energy conversion (Feng et al., 2012). Two-dimensional (2D) materials with great potential of widespread applications (Da et al., 2018, Huang et al., 2018) generally exhibited unique elasticity and flexibility (Frank et al., 2007, Liu et al., 2016), thus providing the most promising platforms to implement the strategy (Zeng et al., 2018, Naumis et al., 2017).

Owing to the unbalanced electrostatic equilibrium of the dislocated atoms, the deformed lattice appears as a thermodynamically unstable system and tends to relax for recovering the intrinsic structural parameters (Zhang et al., 2016). Balancing the additional inner electrostatic interaction requires persistent import of extra energy for suppressing the recovery (Conley et al., 2013, Hui et al., 2013). Consumed approaches in the previous works generally employed external force delivered by the substrates (Ahn et al., 2017, Yan et al., 2013, Li et al., 2015), in view of the great toughness in directly regulating the deformation recovery of atomic-thickness lattices. Therefore the present methods suffer from the intrinsic drawbacks of instability (Feng et al., 2017) and nonuniformity (Liu et al., 2014) of the deformed lattices that are inherited from these force fields.

Here, we demonstrate that the recovery of deformed 2D lattices could be directly regulated by doping metal donors. Metal atom dopants are employed as electron donors to provide extra electrons for reconstructing electrostatic equilibrium among the dislocated atoms. The recovery could be eliminated for achieving lattices with uniformly stabilized and permanently preserved deformation. In this regime, the thermally expanded and in situ elastically bent atomic molybdenum diselenide (MoSe2) lattices could be successfully maintained by doping metal donors, which are atoms owning more outermost electrons and smaller electronegativity than Mo, such as iron (Fe) and copper (Cu). We believe that the metal-donor doping (MDD) strategy could open up a new avenue in directly manipulating the 2D lattices and promote its practical applications based on 2D crystals.

Results and Discussion

Regulation of the Deformation Recovery via Doping Metal Donors

Figure 1A schematically shows the MDD strategy for regulating of MoSe2 lattice deformation recovery via doping metal donors. To investigate the proposition, experiments based on two kinds of metal-doping systems are carried out. The one is the metal-donor system including Fe, nickel (Ni) and Cu atoms, which featured more outermost electrons and smaller electronegativity than Mo atom. The other one is the metal-acceptor system including vanadium (V), niobium (Nb), gold (Au), and tungsten (W) atoms. The employed crystals are directly grown on liquid glass surface via chemical vapor deposition method (Xu et al., 2018, Ju et al., 2017), whereas a corresponding metal foil is utilized as the supporting substrate and the dopant source. During growth, the formed MoSe2 lattices are thermally expanded at the elevated temperature. The donor atoms doped in the deformed lattice could offer extra electrons to modify the electrostatic equilibrium. As a result, the thermal contraction of the donor-doped MoSe2 (D-MoSe2) lattice in the cooling process is inhibited owing to the improved electrostatic repulsion between the bonded atoms, thus realizing the regulation of deformation recovery. The Raman spectra of the as-obtained D-MoSe2 crystals exhibit a split of A1g in Figure 1B, which is in line with the stretched effect of MoSe2 crystals along the c axis (out of plane) (Yasuda et al., 2017). The results of other doping systems exhibit intrinsic Raman features (Gong et al., 2016), and it states the recovery of lattice expansion, thus demonstrating the proposition. The corresponding photoluminescence (PL) spectra (Figure 1C) of D-MoSe2 exhibit an obvious blue shift of the peak position compared with that of pristine MoSe2 crystals (Gong et al., 2016), whereas the spectra acquired in the other doping system are not. It is also thought to be caused by the preserved lattice expansion along all three dimensions of D-MoSe2. Notably, the variation of Raman and PL peaks of the W-doped MoSe2 crystal lines in the alloying effect of MoSe2 and tungsten diselenide (WSe2) (Xie, 2015), owing to their positions, are all located at the intermediate area between of pristine MoSe2 and pristine WSe2 (Liu et al., 2016). As plotted in Figure 1D, only the metal atoms located at the fourth quadrant could serve as powerful electron donors in MoSe2 to regulate the deformation recovery, thus demonstrating the proposed mechanism.

Figure 1

Regulation of the Deformation Recovery by Doping Metal Donors

(A) Schematic illustration of the MDD strategy.

(B and C) The Raman spectra (B) and PL spectra (C) of monolayer MoSe2 crystals with various metal dopants.

(D) The plot of the electronegativity and valence electrons of these atoms and their capability to regulate the deformation recovery.

(E) The Raman mapping of a D-MoSe2 crystal.

For investigating the uniformity of the MDD approach, the characteristic homogeneity of the D-MoSe2 crystals is probed. Taking the Fe-doped MoSe2 as an example, the Raman mapping of one A1g mode at ∼237 cm−1 of a triangle is presented (Figure 1E), where the uniform intensity over the whole crystal indicated the uniform regulation effect in a certain crystal. The corresponding optical microscopic image and PL mapping (Figure S1) also present a uniform contrast. Statistical analyses of the Raman peak difference between the two split A1g peaks and the PL peak position of 100 randomly selected D-MoSe2 single crystals on the same substrate further demonstrate the homogeneity (Figure S1). We notice that the split of A1g peak of monolayer MoSe2 crystals can also be caused by the S alloying (Feng et al., 2015) or Se vacancy (Mahjouri-Samani et al., 2016), which is denied by X-ray photoelectron spectroscopy (XPS) measurements, therefore no peak assigned to S element is detected (Figure S2) and the atomic ratio of Mo/Se is verified as 0.98 (Figure S3). It also suggested a doping level of ∼2%. Furthermore, the metal doping and monolayer nature of the crystal are also confirmed by XPS (Figure S3) and AFM measurements (Figure S4).

Atomic-Scale Investigations of Regulated D-MoSe2 Crystals

The MDD strategy enables the transfer of the crystals with preserved lattice deformation on the Cu grids. The characterizations of the as-grown monolayer D-MoSe2 crystals on the atomic scale are then probed by the transmission electron microscope (TEM) (Figure S5). Selected area electron diffraction (SAED) of the doped and pristine crystal is performed to evaluate the regulation of lattice deformation recovery. The obtained results (Figure 2A) suggest the different repeating intervals between two sets of SAED patterns, further revealing distinct lattice structures (Ahn et al., 2017). For the (100) planes, the separation distance between the diffraction points of the pristine crystal is longer than that of the MDD one (Figure 2B), which indicates a ∼101.5% times extension of the D-MoSe2 lattice. Notably, the microscope parameters camera length and lens aberrations are set as constant values for all subsequent measurements to ensure the reliability of the contrast experiments. The high-resolution TEM images of the two samples and the corresponding fast Fourier transformation patterns are also presented for further comparisons (Figure S6). The results also indicate similar observations of ∼1.5% persevered deformation for the doped lattice. The TEM-based energy dispersive X-ray (EDX) spectroscopy is employed to affirm the presence of Fe element, as displayed in Figure 2C. Here, the Cu element is contributed by the Cu grids.

Figure 2

Atomic-Scale Investigations of Regulated D-MoSe2 Crystals

(A) SAED patterns of the pristine MoSe2 and D-MoSe2 crystals.

(B) The corresponding profiles of the line marked in (A).

(C) The EDX spectra of the two samples.

(D) The false-color HAADF-STEM image of the D-MoSe2 crystal.

(E) The simulated atom structure of (D).

(F) The line profile of the atoms highlighted in (G).

(G) The corresponding Z-contrast image of the marked area in (D).

High-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) is also employed to probe the doping site of D-MoSe2 crystal at the atomic scale. The false-color Z-contrast image displayed in Figure 2D presents the typical atomic arrangement of MoSe2 crystal bonding with several heteroatoms (marked by circles), in which the Mo and Se2 sites exhibit similar intensity contrast in the HAADF measurement (Huang et al., 2014, Feng et al., 2014). The as-marked heteroatom sites manifest a weaker intensity compared with the surroundings, and they are believed to be the metal donor atoms that replaced the Mo sites, owing to the smaller atomic number of Fe (26) than Mo (42). The simulated atom structure (Figure 2E) illustrates a typical D-MoSe2 lattice architecture. The line profile (Figure 2F) of the relative contrast of the atoms highlighted in Figure 2G (the region marked in Figure 2D) suggests a visible intensity difference between the doped atom and Se2 or Mo atoms with a ratio of ∼0.4, which is close to the theoretically calculated value of 0.45 (Krivanek et al., 2010), thus further affirming the presence of Fe atoms. The theoretical intensity ratio between Se and Mo is 0.71, which is lower than the obtained value of ∼0.4, further excluding the presence of Se vacancy. Moreover, the geometric phase analysis (GPA) based on HAADF-STEM image is also performed to evaluate the uniformity at atomic scale. The results presented in Figure S7 reveal a homogeneous geometric phase over the whole region (Han et al., 2018, Xie et al., 2018), which intuitively demonstrates the uniformity of MDD strategy. Based on all these high-resolution structural characterizations, the bonding of metal donor atoms via substitution at Mo sites as well as the uniformly regulated deformation recovery of the lattice with ∼1.5% preservation is suggested (Figure S8).

Theoretical Calculations of the MDD Effect

The MDD effect is directly evaluated by the charge density difference based on the density functional theory (DFT) calculations (Yang et al., 2013). MoSe2 crystal structure with 7 × 7 × 1 cell size is constructed for simulation, whereas a doping level of ∼2% could be obtained by replacing one Mo atom by Fe atom. The top view of pristine MoSe2 crystal is illustrated in Figure 3A, and the corresponding charge density difference slices acquired along the as-marked (110) plane (slice MoSe), the (100) plane of Mo atoms (slice Mo), and the (100) plane of Se atoms (slice Se) are displayed in Figure 3B. The presented isosurfaces indicate a natural electron-trap state (colored by blue) around the Se atoms and electron-loss state (colored by red) around the Mo atoms, which are caused by the electronegativity difference. The scheme of D-MoSe2 crystal and the corresponding isosurfaces are also presented in Figures 3C and 3D for comparison, respectively. Considerable distinctions between two series of slices after the substitution of one atom are exhibited visually. By doping by one metal donor atom, the electron-loss effect (represented by the color intensity of red) of all the Mo atoms is much weaker than the pristine one, whereas that of all the Se atoms at the central region is totally converted into the electron-trap effect (represented by the color intensity of blue). This variation of charge density is believed to be caused by the doping of metal donor, which featured more valence electrons and smaller electronegativity than Mo, thus providing the additional electrons for the lattice and modifying the electrostatic equilibrium among the Mo and Se atoms. For further investigating the scope, the planar charge density difference slices along the side view of two crystals are analyzed and the same results along the entire metal-donor-doped cell could be confirmed (Figure S9). For further confirming the regulation effect of lattice deformation recovery, energy calculations of 7×7 MoSe2 crystals with four configurations are performed (Zhang et al., 2016). As exhibited in Figure S10, the enlarged pristine lattice is thermodynamically unstable and would recover to the original state, whereas the metal-donor-doped lattice would be more stable after expansion and the deformation recovery could be eliminated after a releasing process. These results theoretically illustrate the regulation ability of the lattice deformation recovery based on the MDD strategy.

Figure 3

Theoretical Calculations of the MDD Effect

(A and C) The top-view illustration of the analyzed pristine 7×7 MoSe2 crystal (A) and D-MoSe2 crystal (C).

(B and D) The corresponding isosurface images of the pristine MoSe2 (B) and D-MoSe2 (D) slices of the positions marked in (A) and (C), respectively.

In Situ Investigations of the MDD Strategy

In situ studies are carried out for directly confirming the regulation ability based on the MDD strategy. As schemed in Figure 4A, pristine monolayer MoSe2 crystals grown on Si/SiO2 substrates are transferred to the flexible polyethylene terephthalate (PET) films (stage 1). Then the lattices are deformed by bending the PET substrates (stage 2). Room temperature Fe-ion injection at 5 keV is performed to form the deformed D-MoSe2 crystals (stage 3) (Song et al., 2018). Finally, the bent PET films are relaxed to the plane configurations for releasing the delivered force field of the crystals (stage 4). Here, compared with the PL features of pristine monolayer MoSe2, an ∼15-nm blue shift of the PL peak position for the freshly transferred crystals on PET substrates (stage 1) is presented (Figure S11). Figure 4B displayed the PL acquisitions of a certain MoSe2 crystal on the flat PET substrate before (yellow, at stage 1) and after (cyan, at stage 4) Fe ion injection of 10 s. About 4-nm red shift of the peak position is observed in the normalized PL spectra of the same crystal. For the prolonged injection durations, the red shifts of the PL peak positions of the corresponding crystals exhibited correspondingly enlarged values, as presented in Figure 4C (20 s for 7 nm) and Figure 4D (25 s for 10 nm). Here, the variation of PL peaks agrees well with the shift tendency of monolayer MoSe2 lattice with in-plane deformation (Horzum et al., 2013), thus directly demonstrating the regulation ability of the lattice deformation recovery based on the MDD strategy. As a contrast test, the in situ experiments without stage 2 and injection for 25 s are investigated and no change is presented in PL peaks (Figure 4E), which reveals that directly doping metal donors would not cause the lattice deformation. These results affirm the realization of in situ regulating the lattice deformation recovery and further verify the proposed mechanism. However, further lengthening the injection time may deconstruct the crystal structure and eliminate the PL signal (Figure S12), which prevents the next step investigations. More in situ results at other stages are also offered in Figure S13, in which the statistical analyses of these PL peak position values demonstrate the same red shift tendency of D-MoSe2 crystals based on the MDD strategy.

Figure 4

In Situ Investigations of the MDD Strategy

(A) Schematic illustration of the in situ strategy.

(B−D) The PL spectra acquired from the corresponding monolayer crystals at stage 1 (yellow line) and stage 4 (cyan line) for injecting Fe ions for 10 s (B), 20 s (C), and 25 s (D) in stage 3.

(E) PL spectra of a MoSe2 crystal transferred on PET film (yellow line) and after injecting Fe ions for 25 s without bending the substrate (cyan line).

Conclusion

In summary, based on the MDD strategy, we achieved permanent and uniform regulation of 2D lattice deformation recovery. Metal donors are employed to offer additional electrons to the lattices and reconstruct the electrostatic equilibrium among the atoms. The expansion of monolayer MoSe2 could be then fixed after doping Fe, Cu, or Ni atoms. The proposition is perfectly confirmed by the theoretical calculations and in situ experiments. We believe that the developed strategy will open up a new avenue to directly manipulate the 2D lattice, thus promoting and accelerating the ongoing research efforts as well as practical applications of lattice deformation based on 2D crystals.

Limitations of Study

This work mainly investigates the regulation of 2D lattice deformation recovery by doping metal donors, in which most of the results originated from the Fe-doped samples. More studies need to concentrate on probing the difference brought about by doping different metal donors in the 2D crystals. Exploration of the robustness and the controllability of this strategy also warrants further studies.

Methods

All methods can be found in the accompanying Transparent Methods supplemental file.