SUPPLEMENTARY INFORMATION

Transcription

1 Surface functionalization of two-dimensional metal chalcogenides by Lewis acid base chemistry Sidong Lei, Xifan Wang, Bo Li, Jiahao Kang, Yongmin He, Antony George, Liehui Ge, Yongji Gong, Pei Dong, Zehua Jin, Gustavo Brunetto, Weibing Chen, Zuan-Tao Lin, Robert Baines, Douglas S. Galvão, Jun Lou, Enrique Barrera, Kaustav Banerjee, Robert Vajtai, Pulickel Ajayan NATURE NANOTECHNOLOGY 1

2 Theoretical simulations on InSe, MoS 2, WS 2 and MoSe 2 electron orbitals: Figure S1. DFT calculations show the highest occupied crystalline orbital (HOCO) shape for an atomic layer of (a) InSe, (b) MoS 2, (c) WS 2 and (d) MoS 2, respectively. All the orbital surfaces are plotted considering an iso-surface of It is obvious that all the chalcogen ions on the 2D surfaces are capped with lone pair electron orbitals. The DFT simulation supports the phenomenological explanation by molecular orbital theory and orbital hybridization theory. 2 NATURE NANOTECHNOLOGY

3 SUPPLEMENTARY INFORMATION Control experiment on non-protic solution: Figure S2. (a) The molecular structure of TiCl 4 and (b) control experiment with InSe treated with TiCl 4 toluene solution. Due to the steric hindrance of TiCl 4, as demonstrated in Figure S2 (a), Ti 4+ can hardly reach the InSe surface without ionization, because of the impendence from surrounding Cl -. So protic solution has to be applied to ionize TiCl 4, otherwise, the coordination complex will not form. To prove this, a TiCl 4 toluene solution was prepared and was applied to InSe. Photoresponse measurements were performed before and after the treatment, as shown in Figure S2 (b). It can be found that there is not obvious charge, indicating no InSe-Ti complex was formed. HAADF Z-contract profile measurement orientation. Figure S3. The lattice constant is measured along the red line, which is 0.4 nm. While the HAADF profile is mapped along the green line. After a simple geometric calculation, it is easy to find that the length of two In atoms along the green line is nm, which agrees well with the distance between two In atoms in Figure 1d and 1f. NATURE NANOTECHNOLOGY 3

4 High-resolution TEM and selected area electron beam diffraction study on InSe and InSe-Ti. 4 NATURE NANOTECHNOLOGY

5 SUPPLEMENTARY INFORMATION Figure S4. TEM image (a) and diffraction pattern (b) of pristine InSe. TEM image (c) and diffraction pattern (d) of InSe-Ti. (e) InSe-Ti lattice model and unit cell. (f) The reciprocal vectors and theoretical diffraction model of InSe-Ti. As mentioned in main text, every selenium atom on the InSe surface has lone pair electrons and can serve as ligand. These ligands are close to each other and periodically ordered, so that it is possible that each Ti 4+ is shared with several selenium ligands and forms a large area of ordered titanium chelation pattern on top of InSe lattice. Such a phenomenon is rarely encountered in traditional metal coordination complex. Figure S4a shows the lattice plane of non-functionalized InSe which has a lattice constant of 0.4 nm, matching very well with previous reports. Figure S4b shows the selected area electron beam diffraction pattern of the un-treated InSe, which shows a clear hexagonal structure. After Ti treatment, the TEM image shows significant change, as shown in Figure S4c, demonstrating a Ti 4+ decorated surface, with an atomic spacing of 0.22 nm. Figure S4d shows the selected area electron beam diffraction pattern associated with the area shown in Figure S4c. It clearly shows that the diffraction spots associated with,,,, and lattice planes strongly extinct, while,,,,, lattice planes get strongly enhanced due to the electron beam scattering and interference from titanium cations on InSe surface. The TEM image and electron beam diffraction pattern, together, indicate a hexagonal Ti assembling configuration on InSe surface, as shown in Figure 1g. Ti is captured by each selenium triangles and sits in the centre. This configuration leads to a complex unit cell highlighted by the light blue area in Figure S4e. In this case, the atomic displacement between Ti and neighbouring In or Se should equal to the displacement between selenium atoms ( 0.4 nm = 0.23 nm). This matches very well with the measured value (0.22 nm) from TEM image. In addition, this configuration is supported by the diffraction pattern. Along the two neighbouring sides of the unit cell, two base vectors can be defined as and, then the coordinates of the selenium is (0,0), indium is (1/3, 1/3) and titanium cation is (2/3, 2/3). Correspondingly, in the reciprocal space, two reciprocal base vector can be defined as and with respective to two base vector in lattice space. Then the coordinates of reciprocal vectors can be labelled as Figure S4f. Because of the bijective relationship between reciprocal vectors and diffractions pattern, these vectors shown in Figure S4f can assist the analysis of electron beam diffraction pattern. To determine the relative intensity of the diffraction spot, the structure is defined as where is the reciprocal vector corresponding to a diffraction spot, and the coordinates can be represented by. By substituting the coordinates of each atom and the reciprocal vector, the structure factor can be further simplified into It can be easily found that the reciprocal vectors marked with light blue dots always have structure vectors in the form of and its conjugates. On the contrary, the reciprocal vectors marked with dark blue dot have structure factor of, which leads to stronger diffraction intensity. This explains the extinction and enhancement phenomenon after the Ti treatment. NATURE NANOTECHNOLOGY 5

6 Simulated band diagram of pristine InSe and InSe-Ti complex: Figure S5. Simulated band diagram of pristine InSe and InSe-Ti complex with different Ti coverage rate (6.25%, 25%, 100%). The valence band maxima are chosen as energy zero in all the cases. DFT calculations show that both systems have a direct band gap at point. By comparison, it can be observed that the Ti 4+ ion does not induce trap states in the band gap regardless the coverage rate. Only some distortion is found in the band structures, especially deep in the conduction band. However, the first conduction band valley and the first two valence band peaks at point are nearly not affected. These results show that the electronic properties of InSe are well preserved by the Ti treatment doping, which is more advanced than substitution doping (that degrades the crystal quality). In the band structures of these lower Ti concentrations, we do not observe any localized states either. (Note that the curvatures of the bands change because of the change of Brillouin zone size). Each system has been relaxed to a maximum force of 0.02 ev/å, and eventually Ti atoms prefer to stay in the centre of the InSe hexagons rather than cluster, even at lower coverage rate. 6 NATURE NANOTECHNOLOGY

7 SUPPLEMENTARY INFORMATION XPS study on Ti and Cl in InSe-Ti complex Figure S6. a) shows the XPS data ranging from 454 ev to 470 ev. Before TiCl 4 treatment (red) no XPS peak was detected in this range. After Ti treatment (black), the peak associated with Ti 4+ appeared in InSe sample, indicating the existence of Ti ions. b) shows the XPS data ranging from 190 ev to 210 ev. Before TiCl 4 treatment (red), a very weak Cl peak can be barely distinguished from the noise background, which may result from contamination. After the treatment (black), obvious Cl peaks appeared. NATURE NANOTECHNOLOGY 7

8 Resonant Raman Process in InSe and InSe-Ti: Figure S7. The resonant Raman process in InSe and InSe-Ti samples. As shown in (a), a resonant Raman scattering is an electron-associated photon-phonon scattering in which electrons are excited to an actual energy level and emit or absorb phonons during the relax process. In InSe case, the electron is excited from Se p xy orbit to the bottom of conduction band (Note that it is not an excitation from valence band top to conduction band bottom. Refer to reference 11 in the main text.) Because the resonant Raman involves a special ground state and an actual excited state, the excitation wavelength should be precisely fixed to induce the electron transition between these two energy levels. As a result, the resonant Raman is very sensitive to the changes on energy gap between these two energy levels, the occupation of the ground 8 NATURE NANOTECHNOLOGY

9 SUPPLEMENTARY INFORMATION state, and etc. In our experiment, the resonant peak in InSe-Ti is strongly attenuated comparing with pristine InSe. This is because after the formation of InSe-Ti complex, the electron density in InSe (b) is lowered by Ti in InSe-Ti (c), so is the occupation of electrons in the Raman ground state (Se p xy -orbital). Since fewer electrons can take part in the resonant Raman process, the resonant peak is attenuated. Figure S8. Phonon spectrum of strained InSe (1%) compared to intrinsic InSe calculated by DFT. Table S1. Wave numbers of the Raman peaks calculated by DFT. Raman Peak (unit: cm -1 ) InSe InSe (strained) A E A A It can be found by applying a 1% strain over InSe, A1 ane E peaks do not change obviously, but A2 experiences an obvious red shifting, which agrees well with the experimental observation. This indicates that the lattice strain due to the introduction of Ti 4+ leads to this effect. NATURE NANOTECHNOLOGY 9

10 Atomic force microscopy study on devices under test: Figure S9. AFM measurements on devices. (a) AFM mapping of device shown in Figure 3a inset. (b) AFM profile along the yellow line in (a) indicates a thickness of ~ 7 nm. (c) AFM mapping of device 10 NATURE NANOTECHNOLOGY

11 SUPPLEMENTARY INFORMATION shown in Figure 3d inset. (d) AFM profile along the yellow line in (c) indicates a thickness of ~ 10 nm. (e) AFM mapping of device shown in Figure 4b inset. (f) AFM profile along the yellow line in (e) indicates a thickness of ~ 7 nm. IV character of the p-n junction test structure before Ti treatment. a b mw/cm 2 Dark Current (pa) Photocurrent (pa) Voltage (V) Voltage (mv) Figure S10. The dark current (a) and photocurrent (b) of the p-n junction test structure shown in Figure 3d inset before Ti treatment. In the dark, the device does not show obvious rectification behaviour, although the IV curve is not very symmetry. This asymmetry feature is led by PMMA. More importantly, the test structure showed no photovoltaic effect under illumination (the IV curve passes through the frame origin), suggesting no p-n junction was established before Ti treatment. Short circuit current (SCC) and incident photon current efficiency (IPCE) of InSe-Ti/InSe p-n junction SCC (pa) SCC IPCE IPCE (%) Illumination Intensity (mw/cm 2 ) NATURE NANOTECHNOLOGY 11

12 Figure S11. The SCC (black) increased nearly linearly with the incident light power, because the photo charge carrier generation rate is proportional to the number of absorbed photons. The IPCE (red) fluctuated between 4.8%~5.2%. Considering the photon absorption rate of few layered InSe (~10%), the internal quantum efficiency can be higher than 52 %. Structure of N719, Control experiment of N719 sensitization without Ti-bridge: Figure S12. (a) The structure of N719 dye ([RuL 2 (NCS) 2 ]:2 TBA (L = 2,2'-bipyridyl-4,4'-dicarboxylic acid; TBA = tetra-n-butylammonium), the positive ionic group is not shown, since it does not take part in reaction.) (b) and (c) shows the dark current and photocurrent (under 120 mw/cm nm illumination) before and after N719 treatment, respectively, without Ti-coordination. As discussed in the main text, Lewis acid plays a very important role in surface modification of 2D materials which mainly are Lewis base. The reactive anchor points of N719 are the COO - groups, in which oxygen atoms have lone pair electrons, i.e., N719 still acts as Lewis base. In this case, N719 cannot be bonded to the surface of 2D materials without the help of Lewis acid bridge. This is proved by our control experiment as shown in Figure S6 (b) and (c). It can be found that without Ti-coordination, there is not photoresponse enhancement, in significant contrast with the result shown in Figure NATURE NANOTECHNOLOGY