In Situ Resonant Raman Spectroscopy to Monitor the Surface

Subscriber access provided by UNIVERSITY OF TOLEDO LIBRARIES

Article

In Situ Resonant Raman Spectroscopy to Monitor the Surface Functionalization of MoS and WSe for High-k Integration: A First-Principles Study 2

2

Qingkai Qian, Zhaofu Zhang, and Kevin J. Chen Langmuir, Just Accepted Manuscript • DOI: 10.1021/acs.langmuir.7b03840 • Publication Date (Web): 05 Feb 2018 Downloaded from http://pubs.acs.org on February 6, 2018

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Langmuir is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 30 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Langmuir

In Situ Resonant Raman Spectroscopy to Monitor the Surface Functionalization of MoS2 and WSe2 for High-k Integration: A First-Principles Study Qingkai Qian,† Zhaofu Zhang,† and Kevin J. Chen*,† †

Department of Electronic and Computer Engineering, Hong Kong University of Science and Technology, Clear Water Bay, Hong Kong SAR, China

KEYWORDS: MoS2, WSe2, surface functionalization, Raman scattering, first-principles calculations.

ABSTRACT

Surface functionalization of the dangling-bond free MoS2, WSe2 and other TMDs (transition metal dichalcogenides) is of large practical importance, for example, in providing nucleation sites for the subsequent high-k dielectric integration. Of the surface functionalization methods, the reversible O or N atom adsorption on top of the chalcogen atoms is most promising. However, hazards such as severe oxidation or nitridation persist when the adsorption coverage is high. An in situ characterization technique, which can be integrated with the surface functionalization and dielectric deposition chamber, becomes valuable to enable the real-time monitoring of surface adsorption conditions. Raman spectroscopy, as a non-destructive characterization method without vacuum requirement, is a strong candidate. By utilizing first-principles calculations, Raman spectra of single-layer MoS2 and WSe2 with

ACS Paragon Plus Environment

1

Langmuir 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 30

various O/N adsorption coverages are studied. The calculations suggest that the low-coverage O/N adsorbates will act as perturbations to the periodic lattice and activate the acoustic-phonon Raman scatterings. While high-coverage adsorptions will further activate and intensify the optical-phonon Raman scatterings of previously silent and modes, due to the breaking of reflection symmetry in the z direction. New phonon modes associated with the adatom oscillations are also introduced. All these evidences together with the peak shifts of previously active and modes, suggest that in situ

resonant Raman spectroscopy is capable of providing important information to quantify

the O/N

adsorption coverage, and can be used as a valuable real-time characterization technique to monitor and control the surface functionalization conditions of MoS2 and WSe2.

INTRODUCTION Transition metal dichalcogenides (TMDs) are two-dimensional materials with varied properties that can be tailored by the chemical compositions,1,2 crystalline phases1,3 and layer thicknesses.4-6 Molybdenum disulfide (MoS2) and tungsten diselenide (WSe2), owing to their good thermal stability, high mobility and intriguing optoelectronics applications,7-12 are two widely studied semiconductors from the TMD family. MoS2 and WSe2 have a dangling-bondfree surface, which could be their advantage to reduce the surface scattering and benefit the carrier mobility even with only atomic-scale thickness. However, the dangling-bond free surface also makes it difficult to integrate high-quality high-k dielectric on MoS2 and WSe2 as gate dielectric or passivation layer.6,8,13-16 Additional surface functionalization is needed to provide nucleation sites for subsequent atomic layer deposition (ALD) to improve the film’s uniformity. Of the surface functionalization methods, remote O2 or N2 plasma treatment,14,15 or similar UVO3 treatment6,13,16 could be most promising. Because these methods modify the MoS2/WSe2


2

Page 3 of 30 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Langmuir

surfaces via reversible O or N adsorptions on top of S/Se. The adatoms could then serve as nucleation sites to anchor the precursors during ALD.13,15,16 If the kinetic energy tail of O or N ions is controlled below the knock-on energy threshold to generate a vacancy,17 no irreversible damages will be introduced during the short-time treatment.13-16

However, hazards still persist when the sample is overexposed during the plasma treatment or UV-O3 exposure.6,14-16 As experimentally verified, high-coverage O adsorptions will trigger the area oxidation and the eventual layer-by-layer oxidation of MoS2.14,15,18 Long-time exposure to N2 plasma will also cause N substitutions and other defects.15,19 To realize surface functionalization of MoS2 and WSe2 with robust controllability and yield, an in situ real-time monitoring technique would be highly valuable. Raman spectroscopy has been widely used to characterize the crystal defects, strain and layer thicknesses of MoS2 and WSe2.4,5,10,20,21 It also has the advantages of fast, non-contact, damage-free characterization without vacuum requirement. By simply adding one glass window to allow laser in and out, Raman spectroscopy can be easily integrated with the treatment chamber and the in situ ALD system. Similar to the Raman D-peak of graphene, the acoustic-phonon Raman scattering of MoS2 has already been shown capable of sensitively revealing the crystal defects.21 Since the adsorbates could also be regarded as "defects" or surface perturbations to the periodic crystal lattices, Raman spectroscopy should have the potential to detect the surface adsorption conditions with good sensitivity.

In this paper, Raman spectra of MoS2 and WSe2 with varied O and N adsorptions are studied by first-principles calculations, aiming to provide guidelines for Raman spectroscopy as a technique to monitor the surface functionalization conditions. The calculations show that low-


3

Langmuir 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 4 of 30

coverage adsorbates will act as perturbations to the periodic crystal lattice and activate the acoustic-phonon Raman scatterings. Moreover, high-coverage adsorbates will further activate and intensify the previously silent optical-phonon Raman scatterings, due to the breaking of the reflection symmetry in the z direction. New Raman modes associated with the adatom oscillations are also introduced. With gradually increased adsorption coverage, the previously peaks shift continuously. These Raman spectrum changes can be used as active and

important signatures to quantify the surface adsorption coverage, and to realize robust surface functionalization of MoS2 and WSe2.

RESULTS AND DISCUSSION For ideal MoS2 and WSe2, only phonon modes at Γ point could be Raman-active in singlephonon Raman scattering process, as a result of momentum conservation.21 Low-coverage adsorbates on MoS2 and WSe2 can be regarded as "defects" to the periodic lattices, which will break the translational symmetry and make it possible for phonon modes at Brillouin zone edge to be Raman-active. In real situations, the adsorbates on MoS2 and WSe2 are random, which are difficult to be simulated in first-principles calculations. In practice, the varied adsorbate influences can be studied in a relatively large supercell. The supercell will be periodically repeated during the calculation, thus again only phonon modes at the Γ point of supercell will be Raman-active. However, because the large supercell has smaller 1st Brillouin zone, previous phonon modes at Brillouin zone edge of unitcell are actually now folded to the new Γ point. In this way, the breaking of translational symmetry and activation of new Raman modes can be studied.


4

Page 5 of 30 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Langmuir

Phonon dispersions of MoS2 and WSe2 are calculated by VASP and Phonopy codes. PBE functional and PAW potentials are adopted, with the van der Waals interaction corrected by the DFT-D3. Figure 1a and Figure 1b show the atomic structure of the ideal 1×1 MoS2 and the corresponding phonon dispersion. Excluding the three acoustic-phonon branches, there are six optical-phonon modes at the Γ point. Since the crystal structure of single-layer MoS2 belongs to point group D3h, the corresponding phonon modes have irreducible representation labels of A1', E' and E" according to Mulliken notation.22 However, by convention, labels for bulk MoS2 are still used to name the phonon modes at the Γ point in this paper, as marked in Figure 1b.2,4,23,24 and phonons are singly degenerate, while and are doubly degenerate. In

backscattering geometry, only and are Raman-active and are widely used to measure the

layer thicknesses and crystal qualities,4,5,10,15,21 while and are Raman-silent due to the reflection symmetry along z direction.


5

Langmuir 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 6 of 30

Figure 1. Activation of MoS2 acoustic-phonon Raman scatterings due to O and N adsorptions. (a) Atomic structure of 1×1 MoS2 unitcell and (b) the corresponding phonon dispersion. (c) Atomic structure of 3×3 MoS2 supercell and the folding of 1st Brillouin zone from that of 1×1 MoS2. (d) Phonon dispersion of 3×3 MoS2 supercell. The acoustic phonons of 1×1 MoS2 near 1st Brillouin zone edge are now folded to the Γ point. Calculated single-phonon Raman spectra of (e) ideal 3×3 MoS2 supercell, (f) MoS2 supercell with single O atom adsorption and (g) MoS2 supercell with single N atom adsorption, assuming excited by 514 nm laser. The Raman intensities are decomposed by the phonon modes with A1 and E symmetries.

When adsorbates are present, the supercell no longer has periodicity for the included unitcells. The translational symmetry is broken, thus the phonon modes at the 1st Brillouin zone edge of the unitcell can become Raman-active. A 3×3 MoS2 supercell is used in our study, and the atomic structure is shown in Figure 1c. The 3×3 MoS2 supercell without any adsorption already has 27 ions and yields 78 non-zero phonon modes at the Γ point. To calculate the corresponding Raman tensors using finite-difference method,25,26 DFT calculations need to be done 156 times for the supercell with high accuracy to obtain the frequency-dependent dielectric tensors. The 3×3 MoS2 supercell is chosen based on the considerations of computational time and resources. Compared with the 1×1 unitcell, the supercell has smaller 1st Brillouin zone, thus the previous phonon modes at the Brillouin zone edge are now folded into the smaller Brillouin zone, as schematically drawn in Figure 1c. Figure 1d shows the phonon dispersion of the 3×3 MoS2 supercell. The typical acoustic-phonon modes of 137 cm-1 and 210 cm-1 at the Γ point are highlighted, which are consistent with the predicted results in Figure 1b. Because the 3×3


6

Page 7 of 30 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Langmuir

supercell is still relatively small, the folding is not completely even over the whole Brillouin zone, making the distribution of phonon modes at the Γ point in the supercell deviated from the VDOS (vibrational density of states, see supporting information Figure S1). Specifically, the typical 137 cm-1 and 210 cm-1 modes for acoustic phonon in Figure 1d are different from the 169 cm-1 and 226 cm-1 peaks of VDOS, which will result in slight errors in predicting the peak positions of acoustic-phonon Raman spectrum.

Finite-difference method is used to calculate the Raman tensor.25,26 Firstly, the ion displacements for all phonon modes at the Γ point are calculated. The ion displacements are normalized according to ∑ , , , = , where is the mass of ion , and , is displacement of ion in Cartesian direction i for mode . Within independent particle

approximation and excluding the local field effect, the frequency-dependent dielectric tensor

can be calculated using VASP codes. By differentiating the dielectric tensor with respect to the ion displacements, Stokes Raman tensor ℜ of phonon mode with ion oscillation amplitude

, can be obtained. Specifically,

ℜ , = ∑ , (/ , ) ,

(1)

Furthermore, based on the Raman tensor ℜ , the Raman scattering cross section of supercell for mode can be calculated by26

ℏ! 01233 * ! !# $ 1 , . ℜ . , + + +, 7," 9 ℏ −1

= (4&'"

1A

(2)


7

Langmuir 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 30

where B and B- are incident and scattering light frequencies, , and ,- are the polarization

directions of incident and scattering light, C);DD is the supercell volume, B , , = B is the

energy of phonon mode with amplitude , , and 1/(,

ℏ

− 1) is

the Bose-Einstein factor.

The calculated Raman spectra are shown in Figure 1e, f and g for ideal 3×3 MoS2 supercell, and MoS2 supercells with single O and N adsorptions, assuming excited by 514 nm laser in backscattering geometry, with both the incident and scattering light non-polarized. The most stable single O and N atom adsorptions are on top of sulfur.13,16 The corresponding atomic structures are shown in insets of Figure 1f and Figure 1g. The ideal 3×3 MoS2 supercell still belongs to point group D3h. However, after O/N adsorption, the 3×3 MoS2 belongs to point group C3v, and the Raman active modes can be classified into A1 and E by symmetry.22 A1 mode is singly degenerate, while E mode is doubly degenerate. In fact, and modes of ideal MoS2

with higher symmetries can also be regarded as a special case of A1 and E respectively.22 To visualize the Raman spectrum, Raman peaks are broadened by Lorentz profile. For MoS2 Raman peaks, the FWHM (full width at half maximum) data from previous publication are adopted.21 Namely, modes with symmetry near peak are broadened by 7 cm-1, and modes with

symmetry near peak are broadened by 4 cm-1, while all other modes are roughly broadened

by 15 cm-1. Consistent with the experimental results, for ideal MoS2, only and modes are Raman-

active in the single-phonon scattering, as shown in Figure 1e.21,23 All the newly folded phonon modes are Raman-silent for ideal 3×3 MoS2, which is reasonable, since the ideal 3×3 MoS2


8

Page 9 of 30 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Langmuir

supercell is physically equivalent to the 1×1 MoS2 unitcell. With the introductions of O and N adsorbates, not only do they change the electronic structure (see supporting information Figure S2), they also activate and introduce new phonon modes. The activated acoustic-phonon Raman peaks are labeled. The peak positions are consistent with the folding results in Figure 1d.

peak is observed to shift right after O adsorption. Compared with O adsorption, N adsorption reduces the and Raman intensities more significantly. Both the O and N adsorptions

introduce new Raman-active modes at 1107 cm-1 and 1082 cm-1 respectively, as shown in the insets of Figure 1f and g. The new modes are noted by (E) and (F), because they are

related to the local oscillations of O and N atoms with symmetry. (E) and (F) modes have much higher vibrational frequencies even than the two-phonon Raman scatterings of MoS2,23 making them useful as special signatures to identify the specific atom adsorptions.

Figure 2. Activation mechanism of acoustic-phonon Raman due to O or N adsorption. (a) 210 cm-1 acoustic-phonon mode of ideal 3×3 MoS2 and (b) the corresponding charge oscillation. (c) The charge oscillation is averaged in 1×1 MoS2 unitcell, which completely vanishes for ideal


9

Langmuir 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 30

3×3 MoS2. Same averaged charge oscillations for acoustic phonon modes of (d) 3×3 MoS2 supercell with single O adsorption and (e) 3×3 MoS2 supercell with single N adsorption. The charge oscillations (or intuitively electron wavefunction density oscillations) no longer cancel each other in different unitcells due to adsorptions.

Phonon modes at the Brillouin zone edge are Raman-silent for the ideal 1×1 MoS2 unitcell, due to momentum conservation. Even though these phonon modes can be folded to the Γ point for ideal 3×3 MoS2 supercell, they are still Raman-silent. The explanation is no longer momentum conservation, but because the oscillations in different 3×3 unitcells have different phases, which will completely cancel each other in the radiations of Raman scattering, resulting in zero contributions to the Raman tensor. Figure 2a shows the oscillation pattern of one folded acousticphonon mode at the Γ point of ideal 3×3 MoS2 supercell. This phonon has frequency of 210 cm-1 and is six times degenerate due to the Brillouin zone folding in Figure 1c. These degenerate modes have been renormalized during the phonon eigenvector calculations in the new supercell, thus the oscillation in Figure 2a shows no repeated patterns in different unitcells. The corresponding charge-density oscillation is shown in Figure 2b. The orange and green colors correspond to the increasing and decreasing of electron densities respectively. The chargedensity oscillations in the nine unitcells have different characteristics. However, after averaging, they completely cancel each other and have no net charge oscillations at all, as shown in Figure 2c. Using the same method, the average charge oscillations of similar acoustic phonons for 3×3 MoS2 supercells with single O and N adsorptions are shown in Figure 2d and Figure 2e respectively. Clearly, owing to the perturbations of O and N adsorbates, the charge oscillations in different unitcells no longer completely cancel each other, indicating possible net contributions


10

Page 11 of 30 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Langmuir

to the Raman tensors and thus accounting for the activations of the acoustic-phonon Raman scatterings.

Defects can be generated during the surface functionalization process, since the high-energy tail of the active ions could surpass the knock-off energy of chalcogen atoms,17 or chemical reaction like oxidation or nitridation might happen when the adsorption coverage is too high.14,15,18 Figure 3a shows the calculated Raman spectra of 3×3 MoS2 with one sulfur vacancy. The sulfur vacancy not only activates the acoustic-phonon Raman and mode (295 cm-1), it also severely decreases the and Raman intensities and broadens the and peak

widths. Experimentally the irreversible sulfur vacancy can be intentionally introduced by remote H2 plasma treatment. Figure 3b plots the measured Raman spectra of MoS2 before and after introducing S vacancies. The theoretical calculation shows good agreement with the experimental results, such as the intensity dropping and width broadening of and peaks,

and the activation of acoustic-phonon modes and mode. Due to previously mentioned finite supercell size and thus sparse folding of Brillouin zone, the acoustic-phonon Raman peaks show some deviations from the experimental result. and intensity ratio in the theoretical

calculation are slightly different from the experimental results, which can be related to the ignored excitonic effect, or the notorious band gap underestimation of DFT and the Raman intensity dependence on the exciting laser energy.25,27 After all, the above results have already indicated that first-principle calculations can provide powerful and reliable predictions about the MoS2 Raman spectrum.


11

Langmuir 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 30

Figure 3. (a) Theoretical Raman spectrum of 3×3 MoS2 supercell with single sulfur vacancy. (b) Experimental Raman spectrum of MoS2 with sulfur vacancies intentionally introduced by remote H2 plasma treatment.

When the sample is overexposed during the surface functionalization process, the adsorption coverage can become high, defects like severe oxidation or nitridation might happen.13-15 Raman spectra of MoS2 with single point substitution of S by O and N atoms are calculated and shown in supporting information Figure S3. Based on the experimental results, the long-time exposure of O2 plasma more likely tends to cause area oxidation of MoS2 instead of point defects.15,28 The area oxidation process is very fast once MoS2 is overexposed, which will finally result in layerby-layer oxidation of MoS2.14,18 The oxidation of MoS2 can be identified by the overall dropping of Raman intensity without other significant modifications of the Raman profile.14,15,18 Singlelayer MoS2 is more stable in remote N2 plasma, but still defects can be introduced after long-time exposure, which will activate the acoustic-phonon modes.15 In summary, the overexposure of MoS2 and WSe2 to remote O2/N2 plasma should be avoided.16 Raman spectroscopy as a


12

Page 13 of 30 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Langmuir

technique to realize in situ and real-time monitoring of surface adsorption conditions becomes meaningful.

Figure 4. Raman spectrum evolutions of MoS2 with increasing O/N adsorption coverages. Raman spectra of MoS2 with increasing (a) O surface coverage and (d) N surface coverage. Raman intensities are decomposed into A1 (black) and E (green) modes by the phonon symmetries. Phonon modes of MoS2 with fully covered (b) O and (e) N atoms. The labeled percentage shows the portion of phonon energy associated with the oscillations of adatoms. The extracted peak shifts due to increasing (c) O and (f) N atom adsorptions.

Raman spectra of MoS2 with varied O/N adsorption coverages are calculated in Figure 4a and d. The corresponding supercells are shown in supporting information Figure S4. As shown


13

Langmuir 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 14 of 30

before, the low-coverage adsorption activates the acoustic-phonon Raman scattering, which can be used as a sensitive indicator of the initial adsorptions. With higher coverage adsorption, the low-frequency Raman scattering shows no obvious characteristics. For MoS2 with 100% O coverage, the acoustic-phonon mode will become completely silent, since the crystal has now obtained translational symmetry again. The phonon modes of MoS2 with 100% O adsorption are shown in Figure 4b, which are named based on their similarities to phonons of ideal MoS2 (see supporting information Figure S5). The labeled percentages are the energy ratios of O adatom oscillations. New phonon modes associated with the strong oscillations of adatoms are named according to the phonon symmetries and the specific adatoms, i.e. A1(O) means that the phonon has A1 symmetry and is associated with local oscillations of O adatoms.

To more clearly identify the peak shifts of different modes in Figure 4a, we decompose the Raman intensity according to the phonon symmetries. The black and green lines correspond to Raman intensities of phonon modes with A1 and E symmetries respectively. The peak shift trend is marked by the dotted line to guide the eye, which would otherwise be difficult to be identified without the above decompositions. With increasing O coverage, the peak position

gradually shifts left and finally crosses . The peak shift trends are more clearly plotted in

Figure 4c. A continuously strengthened A1(O), softened and modes are observed. Similar Raman spectrum calculations, intensity decompositions and peak shift extractions are

also carried out for MoS2 with increasing N adsorptions, as shown in Figure 4d, e and f. Experimentally, the decompositions of Raman intensities by symmetry can be achieved in Raman measurements with different laser light polarizations. Because based on the Raman tensors of phonons with different symmetries,22


14

Page 15 of 30 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Langmuir

G ℜ( ) = 90 0

0 G 0

0 0 0A, ℜ(J ) = 9 −K I −L

−K 0 0

K 0 −L 0 A and ℜM = 90 −K 0 L 0

0 LA 0

(3)

+,M . ℜ , . ,J + in equation (2) only becomes non-zero for J mode, but +,J . ℜ , . ,J +

can be non-zero for both M and modes. Thus, only J modes are observable in N(OP)N̅ scattering, i.e. backscattering geometry with incident and scattering light cross-polarized. But both and M mode are observable in N(OO)N̅ geometry. The calculated Raman spectra of MoS2 in different polarization geometries are plotted in Figure S6 in supporting information. Because J and M are degenerate, E mode has the same Raman intensities in both N(OO)N̅ and N(OP)N̅

geometries. By subtracting the Raman intensity of N(OP)N̅ geometry from that of N(OO)N̅

geometry, Raman intensity with A1 symmetry can be obtained, thus the peak shift trends due to adsorptions can be experimentally identified.

Figure 5. Activation of A2u and E1g in backscattering geometry due to breaking of the reflection symmetry along z direction. Side and top views of (a) singly degenerate A2u mode, doubly


15

Langmuir 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 16 of 30

degenerate (b) E1g(x) and (c) E1g(y) modes for MoS2 without and with O adsorptions. The symmetry operations and the corresponding parities are labeled. (d) Raman tensors for the above phonon modes. The Raman tensor equations are based on σS operation, which establish or not depending on the reflection symmetries of phonon modes along z direction.

Noticeably, with high-coverage O/N adsorptions, the previously silent and modes in

backscattering geometry are now active, as shown in Figure 4a and d. For ideal MoS2, both

and phonon modes are antisymmetric with respect to the mirror reflection in z direction (σS ), i.e. they have odd parity with respect to σS operation, as schematically drawn in Figure 5a, b and

c. Moreover, the singly degenerate is invariant by C3(z) rotations, while the doubly

degenerate modes are either symmetric or antisymmetric with respect to the mirror reflection

in x-axis (σJ ). The C3(z) and σJ symmetries of and maintain the same after O/N adsorptions. However, the σS symmetries are broken, which will determine whether the Raman

tensor equations in Figure 5d establish or not. The Raman tensors for MoS2 phonons without/with adsorptions can be derived and are listed in Figure 5d. and of ideal MoS2 are Raman-silent in backscattering geometry, since the xx, yy and xy components of their Raman tensors are zero. While for MoS2 with surface adsorptions, these components can be non-zero, thus and modes are activated in backscattering geometry now, due to the breaking of σS symmetry.


16

Page 17 of 30 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Langmuir

Figure 6. Raman spectra of WSe2 with increasing (a) O and (c) N coverage. Acoustic-phonon Raman scattering is activated at low adsorption coverage, which can be used to detect the initial adsorptions. and modes become obvious for high-coverage adsorptions. Raman peak shifts with increasing (b) O and (d) N coverage percentage. The peak shifts could be used as indicators to monitor the high-coverage O/N adsorptions.

First-principles calculations are also carried out for WSe2 to study the influences of O/N adsorptions over the Raman spectra. The electronic band structures of WSe2 with O/N adsorptions are provided in supporting information Figure S7. The Raman spectra of WSe2 with increasing O and N coverage are shown in Figure 6a and c. The Raman intensities are decomposed into A1 (black lines) and E (green lines) modes, and the extracted peak shifts are


17

Langmuir 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 18 of 30

peaks are broadened by 3 cm-1,27 while all other shown in Figure 6b and d. Both and

modes are broadened roughly by 10 cm-1. For ideal WSe2, and have very close energy,

but has much lower Raman intensity, which are consistent with the experimental results.27

The and modes are Raman-silent for ideal WSe2, due to the mirror reflection symmetry in z direction. Similar to the Raman of MoS2, both low-coverage O and N adsorptions activate

the acoustic-phonon Raman scatterings of WSe2. High-coverage adsorptions activate and intensify the and modes, even though mode is relatively weak for the case of N adsorptions and is strongly mixed with the N atom oscillations. The O/N adsorptions also introduce a new phonon mode with A1 symmetry associated with the local adatom oscillations with high frequency, noted by A1(O) and A1(N) for O and N adsorptions respectively. With high coverage O/N adsorptions, the intensity of mode increases, which finally becomes

comparable to . These Raman peak shift trends and activations of new Raman modes could be used as signatures to monitor and control the surface adsorptions on WSe2.

CONCLUSION Resonant Raman spectra of single-layer MoS2 and WSe2 with increasing O and N adsorptions are studied in supercell models by first-principles calculations for the first time. The activations of both acoustic- and optical-phonon Raman scatterings are analyzed based on the breaking of translational and reflection symmetries. Specifically, the low-coverage O/N adsorption will act as perturbation to the periodic crystal lattice, breaking the translational symmetry and activating the acoustic-phonon modes. The acousticphonon Raman scattering could be used as a sensitive indicator to measure the low-coverage adsorptions. High-coverage adsorptions will break the reflection symmetry in z direction, activating and further intensifying the previously silent A2u and E1g modes. New phonon modes associated with local adatom oscillations with A1 and E symmetries are also introduced. To better identify the A1g and peak shifts,


18

Page 19 of 30 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Langmuir

Raman intensities are decomposed by phonon symmetries, which can be experimentally realized by adopting both N(OO)N̅ and N(OP)N̅ scattering geometries. With increasing O and N adsorptions, A1g mode

is monotonously softened. The A1g and peak shifts, together with the above activations of acoustic-

and optical-phonon Raman scatterings, can be used as scalars to quantify the adsorption coverages and to realize robust surface functionalization.

ASSOCIATED CONTENT Supporting Information Phonon dispersions and vibrational density of states (VDOS) of MoS2 and WSe2; electronic density of

states (DOS), electronic band structures of MoS2 and WSe2 with single O atom and N atom adsorptions; Raman spectra of MoS2 and WSe2 with one sulfur substituted by O atom and N atom; Raman spectra of MoS2 and WSe2 with various O/N adsorptions in different laser polarization geometries. This

material is available free of charge via the Internet at http://pubs.acs.org.

AUTHOR INFORMATION Corresponding Author *Email: [email protected] Notes The authors declare no competing financial interest.

ACKNOWLEDGMENT The computational resources for first-principles calculations were provide by the Tianhe-2 system of the National Supercomputer Center in Guangzhou (NSCC-GZ).


19

Langmuir 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 20 of 30

References 1.

Xu, M.; Liang, T.; Shi, M.; Chen, H., Graphene-Like Two-Dimensional Materials. Chem.

Rev. 2013, 113, 3766-3798. 2.

Zhang, M.; Wu, J.; Zhu, Y.; Dumcenco, D. O.; Hong, J.; Mao, N.; Deng, S.; Chen, Y.;

Yang, Y.; Jin, C.et al., Two-Dimensional Molybdenum Tungsten Diselenide Alloys: Photoluminescence, Raman Scattering, and Electrical Transport. ACS Nano 2014, 8, 7130-7137. 3.

Kappera, R.; Voiry, D.; Yalcin, S. E.; Branch, B.; Gupta, G.; Mohite, A. D.; Chhowalla,

M., Phase-Engineered Low-Resistance Contacts for Ultrathin MoS2 Transistors. Nat. Mater. 2014, 13, 1128-1134. 4.

Lee, C.; Yan, H.; Brus, L. E.; Heinz, T. F.; Hone, J.; Ryu, S., Anomalous Lattice

Vibrations of Single- and Few-Layer MoS2. ACS Nano 2010, 4, 2695-2700. 5.

Sahin, H.; Tongay, S.; Horzum, S.; Fan, W.; Zhou, J.; Li, J.; Wu, J.; Peeters, F. M.,

Anomalous Raman Spectra and Thickness-Dependent Electronic Properties of WSe2. Phys. Rev. B 2013, 87, 165409. 6.

Wang, J.; Li, S.; Zou, X.; Ho, J.; Liao, L.; Xiao, X.; Jiang, C.; Hu, W.; Wang, J.; Li, J.,

Integration of High-k Oxide on MoS2 by Using Ozone Pretreatment for High-Performance MoS2 Top-Gated Transistor with Thickness-Dependent Carrier Scattering Investigation. Small 2015, 11, 5932-5938.


20

Page 21 of 30 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Langmuir

7.

Chiu, M.; Zhang, C.; Shiu, H.; Chuu, C.; Chen, C.; Chang, C. S.; Chen, C.; Chou, M.;

Shih, C.; Li, L., Determination of Band Alignment in the Single-Layer MoS2/WSe2 Heterojunction. Nat. Commun. 2015, 6, 7666. 8.

Sachid, A. B.; Tosun, M.; Desai, S. B.; Hsu, C.; Lien, D.; Madhvapathy, S. R.; Chen, Y.;

Hettick, M.; Kang, J. S.; Zeng, Y.et al., Monolithic 3D CMOS Using Layered Semiconductors. Adv. Mater. 2016, 28, 2547. 9.

Jo, S.; Kang, D.; Shim, J.; Jeon, J.; Jeon, M. H.; Yoo, G.; Kim, J.; Lee, J.; Yeom, G. Y.;

Lee, S.et al., A High-Performance WSe2/h-BN Photodetector using a Triphenylphosphine (PPh3)-Based n-Doping Technique. Adv. Mater. 2016, 28, 4824-4831. 10. Lim, Y. R.; Song, W.; Han, J. K.; Lee, Y. B.; Kim, S. J.; Myung, S.; Lee, S. S.; An, K.; Choi, C.; Lim, J., Wafer-Scale, Homogeneous MoS2 Layers on Plastic Substrates for Flexible Visible-Light Photodetectors. Adv. Mater. 2016, 28, 5025-5030. 11. Radisavljevic, B.; Radenovic, A.; Brivio, J.; Giacometti, V.; Kis, A., Single-Layer MoS2 Transistors. Nat. Nanotechnol. 2011, 6, 147-150. 12. Yu, Z.; Ong, Z.; Pan, Y.; Cui, Y.; Xin, R.; Shi, Y.; Wang, B.; Wu, Y.; Chen, T.; Zhang, Y.et al., Realization of Room-Temperature Phonon-Limited Carrier Transport in Monolayer MoS2 by Dielectric and Carrier Screening. Adv. Mater. 2016, 28, 547-552. 13. Azcatl, A.; McDonnell, S.; K. C., S.; Peng, X.; Dong, H.; Qin, X.; Addou, R.; Mordi, G. I.; Lu, N.; Kim, J.et al., MoS2 Functionalization for Ultra-Thin Atomic Layer Deposited Dielectrics. Appl. Phys. Lett. 2014, 104, 111601.


21

Langmuir 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 22 of 30

14. Zhu, H.; Qin, X.; Cheng, L.; Azcatl, A.; Kim, J.; Wallace, R. M., Remote Plasma Oxidation and Atomic Layer Etching of MoS2. ACS Appl. Mater. Interfaces 2016, 8, 1911919126. 15. Qian, Q.; Zhang, Z.; Hua, M.; Tang, G.; Lei, J.; Lan, F.; Xu, Y.; Yan, R.; Chen, K. J., Enhanced Dielectric Deposition on Single-Layer MoS2 with Low Damage using Remote N2 Plasma Treatment. Nanotechnology 2017, 28, 175202. 16. Azcatl, A.; Santosh, K. C.; Peng, X.; Lu, N.; McDonnell, S.; Qin, X.; de Dios, F.; Addou, R.; Kim, J.; Kim, M. J.et al., HfO2 on UV-O3 Exposed Transition Metal Dichalcogenides: Interfacial Reactions Study. 2D Materials 2015, 2, 14004. 17. Zan, R.; Ramasse, Q. M.; Jalil, R.; Georgiou, T.; Bangert, U.; Novoselov, K. S., Control of Radiation Damage in MoS2 by Graphene Encapsulation. ACS Nano 2013, 7, 10167-10174. 18. Ko, T. Y.; Jeong, A.; Kim, W.; Lee, J.; Kim, Y.; Lee, J. E.; Ryu, G. H.; Park, K.; Kim, D.; Lee, Z.et al., On-Stack Two-Dimensional Conversion of MoS2 into MoO3. 2D Materials 2017, 4, 14003. 19. Azcatl, A.; Qin, X.; Prakash, A.; Zhang, C.; Cheng, L.; Wang, Q.; Lu, N.; Kim, M. J.; Kim, J.; Cho, K.et al., Covalent Nitrogen Doping and Compressive Strain in MoS2 by Remote N2 Plasma Exposure. Nano Lett. 2016, 16, 5437-5443. 20. Zhu, C. R.; Wang, G.; Liu, B. L.; Marie, X.; Qiao, X. F.; Zhang, X.; Wu, X. X.; Fan, H.; Tan, P. H.; Amand, T.et al., Strain Tuning of Optical Emission Energy and Polarization in Monolayer and Bilayer MoS2. Phys. Rev. B 2013, 88, 121301.


22

Page 23 of 30 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Langmuir

21. Mignuzzi, S.; Pollard, A. J.; Bonini, N.; Brennan, B.; Gilmore, I. S.; Pimenta, M. A.; Richards, D.; Roy, D., Effect of Disorder on Raman Scattering of Single-Layer MoS2. Phys. Rev. B 2015, 91, 195411. 22. Loudon, R., The Raman Effect in Crystals. Adv. Phys. 1964, 13, 423-482. 23. Frey, G. L.; Tenne, R.; Matthews, M. J.; Dresselhaus, M. S.; Dresselhaus, G., Raman and Resonance Raman Investigation of MoS2 Nanoparticles. Phys. Rev. B 1999, 60, 2883-2892. 24. Chakraborty, B.; Bera, A.; Muthu, D. V. S.; Bhowmick, S.; Waghmare, U. V.; Sood, A. K., Symmetry-Dependent Phonon Renormalization in Monolayer MoS2 Transistor. Phys. Rev. B 2012, 85, 161403. 25. Miranda, H. P. C.; Reichardt, S.; Froehlicher, G.; Molina-Sánchez, A.; Berciaud, S.; Wirtz, L., Quantum Interference Effects in Resonant Raman Spectroscopy of Single- and TripleLayer MoTe2 from First-Principles. Nano Lett. 2017, 17, 2381-2388. 26. Liang, L.; Meunier, V., First-principles Raman Spectra of MoS2, WS2 and Their Heterostructures. Nanoscale 2014, 6, 5394. 27. Zhao, W.; Ghorannevis, Z.; Amara, K. K.; Pang, J. R.; Toh, M.; Zhang, X.; Kloc, C.; Tan, P. H.; Eda, G., Lattice Dynamics in Mono- and Few-Layer Sheets of WS2 and WSe2. Nanoscale 2013, 5, 9677-9683. 28. Islam, M. R.; Kang, N.; Bhanu, U.; Paudel, H. P.; Erementchouk, M.; Tetard, L.; Leuenberger, M. N.; Khondaker, S. I., Tuning the Electrical Property via Defect Engineering of Single Layer MoS2 by Oxygen Plasma. Nanoscale 2014, 6, 10033-10039.


23

Langmuir 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 24 of 30

Table of Contents


24

Page 25 of 30 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Langmuir

170x100mm (300 x 300 DPI)


Langmuir 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

84x74mm (300 x 300 DPI)


Page 26 of 30

Page 27 of 30 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Langmuir

170x74mm (300 x 300 DPI)


Langmuir 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

170x107mm (300 x 300 DPI)


Page 28 of 30

Page 29 of 30 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Langmuir

170x96mm (300 x 300 DPI)


Langmuir 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

85x72mm (300 x 300 DPI)


Page 30 of 30

In Situ Resonant Raman Spectroscopy to Monitor the Surface

Recommend Documents