Second Harmonic Generation from Artificially Stacked Transition Metal Dichalcogenide Twisted Bilayers

HSU ET AL. VOL. 8 NO. 3 2951–2958 2014 ₂₉₅₁

February 25, 2014

C 2014 American Chemical Society

Second Harmonic Generation from

Arti

ficially Stacked Transition Metal

Dichalcogenide Twisted Bilayers

Wei-Ting Hsu,†Zi-Ang Zhao,†Lain-Jong Li,‡,§Chang-Hsiao Chen,‡Ming-Hui Chiu,‡Pi-Shan Chang,† Yi-Chia Chou,†_{and Wen-Hao Chang}†,_*

†_{Department of Electrophysics, National Chiao Tung University, Hsinchu 30010, Taiwan,}‡_{Institute of Atomic and Molecular Sciences, Academia Sinica, Taipei 10617,}

Taiwan, and§Department of Medical Research, China Medical University Hospital, Taichung 40402, Taiwan

F

(M = Mo, W; X = S, Se), are semicon-ducting layered materials, exhibiting fasci-nating layer-number dependent electronic and optical properties,1which are sensitive to the underlying crystal symmetry and differ greatly from their bulk counterparts. For instance, single-layer molybdenum di-sulfide (MoS2) is known as a direct-gap

semiconductor with broken inversion sym-metry in its crystalline structure,1
3which can generate strong valley dependent po-larizations by helical lights.4
6Bilayer MoS2,

on the other hand, has vanished valley po-larization due to the restoration of crystal

inversion symmetry.5,6The presence or the absence of inversion symmetry in few-layer TMDs also manifests itself in their second-order optical nonlinearity. Intense and layer-number dependent optical second harmo-nic generation (SHG) have been observed in few-layer TMDs,7
11owing to the nonvan-ished second-order nonlinearity associated with the broken inversion symmetry in flakes with odd number of layers. However, such layer-number dependent properties are preserved only when the few-layer TMDs are Bernal stacked.7,9,10Unlike those prepared by exfoliation from nature crystals, the few-layer TMDs8 or hexagonal boron nitride (h-BN)11 grown by chemical vapor

* Address correspondence to [email protected]. Received for review January 14, 2014 and accepted February 25, 2014. Published online

10.1021/nn500228r

ABSTRACT

Optical second harmonic generation (SHG) is known as a sensitive probe to the crystalline symmetry of few-layer transition metal dichalcogenides (TMDs). Layer-number dependent and polarization resolved SHG have been observed for the special case of Bernal stacked few-layer TMDs, but it remains largely unexplored for structures deviated from this ideal stacking order. Here we report on the SHG from homo- and heterostructural TMD bilayers formed by artificial stacking with an arbitrary stacking angle. The SHG from the twisted bilayers is a coherent superposition of the SH fields from the individual layers, with a phase difference depending on the stacking angle. Such an interference effect is insensitive to the constituent layered materials and thus applicable to hetero-stacked bilayers. A proof-of-concept demonstration of using the SHG to probe the domain boundary and crystal polarity of mirror twins formed in chemically grown TMDs is also presented. We show here that the SHG is an efficient, sensitive, and nondestructive characterization for the stacking orientation, crystal polarity, and domain boundary of van der Waals heterostructures made of noncentrosymmetric layered materials.

KEYWORDS: transition metal dichalcogenide . MoS2. second harmonic generation . SHG . two-dimensional layered materials .

artificial stacking

deposition (CVD) may deviate from this ideal stacking order with interlayer rotations, displacements, or even stacking faults. The structure
property relationships in twisted bilayer graphene have been investigated in-tensively recently;12
15however, they remain largely unexplored for twisted bilayer TMDs.

Furthermore, recent advances in isolating and sub-sequent restacking of heterogeneous layered materi-als, including graphene, h-BN, and TMDs, have opened up a new research direction for exploring new proper-ties and functionaliproper-ties of artificial materials for versa-tile applications.16_{The success of these van der Waals}

hybrid materials has manifested itself inflexible photo-voltaic devices17 _{and field-effect tunneling}

transis-tors.18 Vertical stacking of TMDs homo- and hetero-structures with precise control of twist angles would create new materials with a tailored interlayer cou-pling that is strongly stacking orientation depen-dent.13,14,19,20Therefore, an efficient and noninvasive characterization tool for identifying the stacking angle is desirable. In this context, the SHG appears as a promising technique, since the second harmonic (SH) response is a sensitive probe to the underlying crystal-line symmetry and orientation. The SH microscopy has been proven to be useful for identifying the odd/even number of layers7,9,10and probing the crystallographic axis8
11 in few-layer TMDs and h-BN. However, the stacking angle dependence of the SHG from homo-and hetero-stacked TMD twisted bilayers has yet to be established.

In this work, we investigate the SH response of artificially stacked TMDs bilayers with an arbitrary stacking angle. We first start from the prototypical homo-stacked MoS2/MoS2bilayers and demonstrate

that the SHG from the twisted bilayers is a coherent superposition of the SH fields from the constituent monolayers, with a phase difference depending on the stacking angle. The resulting intensity and polarization of the SHG as a function of the stacking angle can thus be established. We further extend the study to hetero-stacked WSe2/MoS2and WSe2/WS2bilayers and show

that the same idea is applicable and insensitive to the

constituent layered materials. Finally, a proof-of-concept demonstration of using the SH microscopy to probe structural defects in CVD-grown TMDs, such as the domain boundary and crystal polarity of mirror twins, is presented.

RESULTS AND DISCUSSION

Figure 1a shows an optical microscopy image of the artificially stacked MoS2 bilayers. Monolayer MoS2

flakes were first synthesized by CVD on sapphire substrates.21The artificially stacked MoS2bilayers were

then fabricated by transferring the as-grown mono-layer flakes onto another substrate with monolayer flakes already formed thereon (see Methods). The CVD-synthesized MoS2flakes are mostly in triangular shape

with a typical side length of around∼10 μm. According to high-resolution transmission electron microscopy (HR-TEM) and optical (Photoluminescence and Raman) characterizations (see Supporting Information), these triangular MoS2flakes are monolayer thick and single

crystalline with edges along the zigzag direction (Figure 1b,c). Therefore, the crystallographic twist an-gles of the MoS2bilayers (Figure 1d) can be determined

directly by the relative orientation between the two constituent triangularflakes observed in optical micro-scopy images.

To illustrate how the stacking orientation affects the optical SHG, we select here two representative stacking orientations in nearly parallel (θ = 0
) and antiparallel (θ = 60
) configurations. Figure 2a,b shows the optical microscopy images of the two stacking configurations. The corresponding spatial mappings of SH intensity from individual monolayer MoS2flakes and the

stack-ing regions are displayed in Figure 2c,d. In this experi-ment, we measured the total SH power, which is independent of the polarization direction of the in-cident laserfield with respect to the crystallographic axis. For the case of antiparallel stacking (Figure 2c), the SH intensity from each MoS2 flake is quite uniform,

except for the stacking region, where the SH intensity is strongly suppressed. It is well-known that the SHG signal occurs (vanishes) in materials with the absence Figure 1. (a) An optical microscopy image of the artificially stacked MoS2bilayers. Monolayer (1L), bilayer (2L), and substrate

regions are indicated by arrows. (b) A low-magnification TEM image for a triangular MoS2monolayerflake. The scale bar is

5_{μm. (c) The FFT filtered HR-TEM image of the MoS}2flake shown in (b). The hexagonal lattice structure is clearly observed,

with brighter spots corresponding to the Mo sites. (d) A schematic showing the lattice structure of an artificially stacked MoS2

bilayer with a stacking angleθ.

HSU ET AL. VOL. 8 NO. 3 2951–2958 2014 2953

(presence) of inversion symmetry. Monolayer MoS2

crystal belongs to the D3h point group with broken

inversion symmetry and thus allows strong SHG. For MoS2 bilayers exfoliated from bulk crystal with the

Bernal (2H) stacking order, the crystal belongs to the D3dpoint group with inversion symmetry and hence

vanishes the SH response. The strongly suppressed SH

response in the stacking region shown in Figure 2c indicates that the underlying crystal symmetry is close to the Bernal stacked bilayer. On the contrary, as shown in Figure 2d for the case of nearly parallel stacking, an enhanced SH intensity can be observed in the stacking region. The SH intensity is about∼3.2 times stronger than that of individual monolayers. This demonstrates that the SH response is sensitive to the stacking orientation, which cannot be observed in few-layer MoS2exfoliated from bulk crystal with the ideal Bernal

stacking.

The underlying crystal symmetry and crystallo-graphic orientation of individual monolayer MoS2

flakes and the stacking regions can be further exam-ined by the polarization-resolved SHG. We select here a particular bilayer with a stacking angle of θ = 25
betweenflakes 1 and 2 (Figure 3a). The SHG experi-ment is performed in a back-reflection geometry using a pump laser normally incident on the sample (Figure 3b). The incident pump laser is linearly polar-ized along the x direction, which is initially aligned with the armchair direction of theflake 1 (Figure 3c). For the monolayer MoS2 crystal with a three-fold rotational

symmetry, the SH response with a polarization parallel (perpendicular) to the incident laser polarization is known to exhibit a six-fold rotational symmetry, with a SH intensity varies as I₎ cos23j (I_^ sin23j),8
11,22 wherej is the azimuthal angle between the incident laser polarization and the armchair direction. The SH Figure 2. (a and b) Optical microscopy images for stacked

MoS2bilayers with a stacking angle of (a)θ = 60
and (b) θ =

0
_{. (c and d) The corresponding false color-coded SH} intensity obtained by pixel-to-pixel spatial mappings of flakes in (a) and (b). The scale bars are 5 μm.

Figure 3. (a) The optical microscopy image of a particular MoS2bilayer with a stacking angle ofθ = 25
. The scale is 5 μm.

(b and c) Schematics showing (b) the measurement geometry and (c) the polarization directions of the incident laser and the measured SHG. The azimuthal anglej defines the angle between the laser polarization and the x-axis, which is aligned with the armchair direction of theflake (1). (d
f) Polar plots of the polarization-resolved SH intensity as a function of azimuthal anglej measured from (d) the monolayer region (1), (e) the monolayer region (2), and (f) the stacking region (3).

response thus shows an intensity maximum when the incident laser polarization aligns with the armchair (zigzag) direction for the parallel (perpendicular) po-larization configuration.8
10As shown in Figure 3d,e, the polarization-resolved SHG of individual flake (measured under the parallel polarization con figura-tion) shows a six-petal pattern, with petals lying along the perpendicular bisectors of each flake. This also agrees with our TEM analysis, where the perpendicular bisectors (edges) of the MoS2triangles are aligned with

the armchair (zigzag) directions. Interestingly, the SH light from the stacking region also show a six-petal polarization pattern (Figure 3f), but with petals lying along a direction between the two nearest perpendi-cular bisectors offlakes 1 and 2. This clearly indicates that the SHG from the stacking region is not contrib-uted independently from individualflakes.

The SH response in the artificially stacked MoS2

bilayers can be modeled as two electrically decoupled layers of atomically thin sheet dipole moments driven coherently by the fundamental laserfield.10The line-arly polarized fundamental laserfield generates

line-arly polarized SH fields from each layer, with a

polarization direction determined by the angle be-tween laser polarization and the respective armchair direction of each layer. The SHfields generated from the two layers then interfere coherently with a phase difference depending on the stacking angle. The re-sulting polarization of SHG from the stacking region thus retains the six-fold rotational symmetry, but with petals lying between the twoflakes. Furthermore, the coherent superposition of SH fields can be either constructive or destructive, leading to an enhanced or suppressed SH intensity in the stacking region. This simple model has been employed to explain some experimental signatures of SHG observed in CVD-synthesized few-layer MoS28and h-BN.11Kumar et al.

found that the bilayer regions in the central area of MoS2triangles have a stronger SH intensity as

com-pared with monolayer regions,8 indicating that the MoS2bilayer is not in the 2H-stacking order. Kim et al.

found that h-BN bilayers exhibit a stronger (weaker) SH intensity in the AB (AA0) stacking configuration,11and the SH intensity in the AB stacking regions is∼4 times stronger than that in monolayer regions, in good agreement with the interference model described above.

Here we provide a full description of the model for SHG in bilayers TMDs with an arbitrary stacking angle. Let us consider the stacked bilayers as two electrically decoupled layers, and first neglect the propagating effects of electromagnetic waves in these atomically thin layers. If the laser electricfield EB(ω) makes an angle ofj1(j2) with respect to the armchair direction offlake

1 (flake 2), the generated SH electric field EB1(2ω)

[EB2(2ω)] from the flake 1 (flake 2) is then 3j1(3j2)

away from the laser polarization (Figure 4a). The total

SH electric field EBs(2ω) in the stacking region can

then be obtained by a vector superposition: EBs(2ω) =

E

B1(2ω) þ EB2(2ω). The total SH intensity in the stacking

region is proportional to |EBs(2ω)|2, which can be

ex-pressed as Is(θ) ¼ I1þ I2þ 2 ffiffiffiffiffiffiffi I1I2 p cos3θ (1)

where Is, I1, and I2stand for the SH intensity in the

stacking region, the monolayerflake 1, and flake 2, respectively; andθ = j1
j2is the stacking angle,

which is defined as the angle between two nearest perpendicular bisectors of the two triangular flakes (Figure 4b). That is,θ = 0
yields a completely con-structive interference, whereasθ = 60
yields a com-pletely destructive interference of SHfields (Figure 4c,d). Although the artificial stack with θ = 60
is analogous to natural MoS2bilayer with D3d

3

symmetry, it should be noted that the artificial stack might present lateral displacements between the two layers, making the θ = 60
stacking not truly in the Bernal stacking. However, such a lateral displacement is unable to resolve by the SHG technique, since the optical wave-length is much longer than the lateral misalignment of less than one lattice constant.

We have examined the SH response of the artificially stacked bilayers with different stacking angles distrib-uted in the range of 0
e θ < 60
. Figure 5a shows the optical microscopy images of a series of MoS2bilayers

with different stacking angles. The corresponding SH intensity mappings are displayed in the same color scale in Figure 5b. It is clear that the SH intensity in the bilayer region changes dramatically with the stacking angle. According to (1), the stacking angle dependent SH intensity should vary as cos 3θ and from Is(0
) = 4I0

(fully constructive) to Is(60
) = 0 (fully destructive) if I1=

I2= I0, i.e., the SH intensities from different flakes are the

Figure 4. (a) A schematic for illustrating the vector super-position of the SH electricfields, where EB(ω) is the electric field of the fundamental light, EB1(2ω) [EB2(2ω)] is the SH

electricfield from the flake 1 (flake 2), and EBs(2ω) is the

resulting SH electricfield from the stacking region. (b
d) Schematics for stacked bilayers with (b) an arbitrary stack-ing angleθ, (c) θ = 0
, and (d) θ = 60
(or θ = 180
).

HSU ET AL. VOL. 8 NO. 3 2951–2958 2014 2955

same. However, such an angular dependence could be screened by the presence of SH intensity variation

(∼10
15%) among different flakes. Therefore, we

define here a dimensionless parameter κ(θ)  (Is

I1
I2)/2(I1I2)1/2and then examine its angular

depen-dence. In this way, the intensity variation among different flakes can be normalized and, in principle, the angular dependence of κ(θ) = cos 3θ should be recovered. It should be noted that the value of cos 3θ changes sign atθ = 30
. A positive (negative) κ(θ) value indicates a constructive (destructive) interference and, hence, an enhanced (suppressed) SH intensity in the stacking region, i.e., Is> I1þ I2(Is< I1þ I2). As shown in

Figure 5c, the extracted κ(θ) for the homo-stacked MoS2/MoS2 bilayers follows closely to cos 3θ (black

curve), in agreement with the simple model predic-tion. However, the measured interference contrast is

somewhat lower than the simple model prediction; i.e., the experimental data always lies below the cos 3θ dependence. Moreover, the extractedκ(θ) values also change sign atθ < 30
. To account for the reduced interference contrast, we have further included the propagating effects of electromagnetic waves through the TMD layers and the underlying substrate. The phase shift, absorption, and reflection of both the fundamental laserfield and the SH fields in the TMD layers and at the adjacent interfaces have been taken into account (see Supporting Information for detail). After including the propagating effects of electromag-netic waves, the calculated SH response (red curve) reproduces the measuredκ(θ) very well. In this calcula-tion, we found that the optical absorption of the SH light by the upper TMD layer plays a key role, leading to incomplete interferences. This can be verified from a Figure 5. (a) Optical microscopy images for a series of MoS2bilayers with a stacking angle distributed fromθ = 0
to 60
. (b)

The corresponding false color-coded SH intensity mappings offlakes shown in (a). The scale bar is 5 μm. (c
e) The measured angular dependence of _{κ(θ) for (c) homogeneous MoS}2/MoS2bilayers, (d) heterogeneous WSe2/WS2 bilayers, and (e)

heterogeneous WSe2/MoS2bilayers. Curves are calculatedκ(θ): black solid curves, κ(θ) = cos 3θ; red, (blue dashed) curves,

calculations considering the propagation effects of electromagnetic waves with (without) including material absorption. Symbols are data points collected from all the investigated bilayers. (f) The measured SH intensities for MoS2, WS2, and WSe2

monolayers.

calculation without including the material absorption, where only a minor change in the interference contrast caused by the multilayer reflections. The absorption coefficient at the SH wavelength used in this calcula-tion was obtained by differential reflectance measure-ments (see Methods). For monolayer MoS2, the

absor-bance is estimated to be∼12.9% at wavelength λ = 405 nm, corresponding to an absorption coefficient of R ≈ 2.1  106

cm
1(see Supporting Information). This value is comparable to the reported values in literature for bulk MoS2.23

On the basis of the understanding of MoS2/MoS2

homo-stacked bilayers, we have further extended the study to hetero-stacked bilayers. The stacking angle dependences of κ(θ) extracted from selected WSe2/

WS2 and WSe2/MoS2 hetero-stacked bilayers are

shown in Figure 5, panels d and e, respectively. In both cases, the model including the propagation effects of electromagnetic waves also reproduces the experi-mental data very well. According to the measured differential reflectance spectra of monolayer WSe2

and WS2, the absorption coefficient at the SH

wave-length is nearly the same as the monolayer MoS2(see

Supporting Information). Therefore, the model predic-tions for the homo- and hetero-stacked TMD bilayers are expected to be similar, regardless of the constitu-ent layered materials. However, we noted that the measured κ(θ) for the WSe2/MoS2 bilayers is more

scattered (Figure 5e). This might be due to the large difference between the SH efficiencies of MoS2 and

WSe2monolayers. The SH intensity of monolayer MoS2

is about 3 times weaker than that of WSe2and WS2

(Figure 5f), which could lead to incomplete interfer-ence and more scattered data. Nonetheless, our results demonstrate that the SHG is capable of probing the underlying symmetry of stacked van der Waals heterostructures made of noncentrosymmetric layered materials.

From the measuredκ(θ) for the homo- and hetero-stacked TMD bilayers, it is also clear that all the triangular TMD monolayers are of the same crystal polarity, i.e., either the transition-metal-polar (Mo, W) or the chalcogen-polar (S, Se), rather than a mixture of both polarities. As has been mentioned above, mono-layer TMD crystals exhibit a three-fold rotational sym-metry, while the polarization-resolved SH intensity has a six-fold rotational symmetry, since a phase shift ofπ in the SHfield EB(2ω) cannot be distinguished by the measurements of the SH intensity |EB(2ω)|2. In the artificially stacked TMD bilayers, the stacking angle

dependent SH intensity becomes electric-

field-resolved due to the interference effect. If the bilayers are made of two TMD monolayers with opposite polarity, the resulting stacking angle dependence will beκ(θ) = sin 3θ. The invariably observed κ(θ) = cos 3θ for all the homo- and hetero-stacked bilayers indicates that all the investigated triangular monolayers are

single-domain with the same crystal polarity. Accord-ing to the HR-TEM study, our MoS2triangularflakes are

terminated by Mo-polar zigzag edges, implying that the investigated WS2and WSe2triangularflakes also

have W-polar zigzag edges.

Since the SHG is crystal-orientation-dependent, the SH microscopy in principle is capable of probing the crystallographic axes and grain boundaries of TMD films, especially for tilted grains in polycrystalline aggregates or continuous sheets formed during CVD growths. Notably, however, the intensity-resolved SH microscopy is unable to distinguish the crystal orienta-tions of mirror twins formed by two domains with a relative in-plane rotation of 180
. By laying down a monolayer TMD with known crystal polarity on top of mirror twins, the crystal orientation becomes visua-lized due to the electric-field resolved interference effect. A proof-of-concept demonstration for the char-acterization of mirror twins is shown in Figure 6. Let us focus on the butterfly-shaped and the triangular MoS2

monolayers highlighted by the dotted lines. In the stacking region, the SH intensity shows a strong con-trast between the two wings of the butterfly-shaped monolayer, indicating that it consists of two mirror-twin domains, from which the SHfields interfere con-structively and decon-structively with that from the trian-gular MoS2monolayer. The mirror twin boundary in the

monolayer region is also visualized for thefirst time by the SH microscopy, as indicated by the arrow near the intersection between the two wings (Figure 6b). Since the SH dipole moments generated in the mirror twins are completely out of phase, the SH response should vanishes at the mirror twin boundary, resulting in a strongly suppressed SH intensity.

CONCLUSIONS

We have investigated the SHG from artificially stacked TMD bilayers with arbitrary stacking angles. The SHG from the twisted bilayers is a coherent super-position of the SHfields from the individual layers, with a phase difference depending on the stacking angle. Such an interference effect is insensitive to the con-stituent layered materials and thus applicable to hetero-stacked bilayers. The intensity and the polarization Figure 6. (a) An optical microscopy image for a triangular MoS2monolayer overlapped with a butterfly-shaped MoS2

monolayer. (b) The corresponding false color-coded SH intensity obtained by pixel-to-pixel spatial mappings of flakes in (a). The butterfly-shaped MoS2monolayer consists

of two mirror-twin domains with a twin boundary located at the intersection of the two wings (indicated by arrow).

HSU ET AL. VOL. 8 NO. 3 2951–2958 2014 2957

of SHG as a function of the stacking angle have been established. A proof-of-concept demonstration of using the SHG to probe the domain boundary and crystal polarity of mirror twins formed in chemically grown TMDs is also presented. We have demonstrated

here that the SHG is an efficient, sensitive, and non-destructive characterization for the stacking orienta-tion, crystal polarity, and domain boundary of van der Waals heterostructures made of noncentrosymmetric layered materials.

METHODS

Growth of Monolayer TMDs. High-crystal-quality monolayers MoS2, WSe2, and WS2were synthesized on sapphire substrates

by chemical vapor deposition (CVD) in a horizontal hot-wall chamber. The detailed fabrication procedures can be found elsewhere.21,24,25High purity MoO3(99%, Aldrich), WO3(99%,

Alfa), Se and S powders (99.5%, Alfa) were used as initial reactants. The MoS2 and WS2 samples were obtained by

annealing at 650 and 800
C for 5 min with a heating rate of 15
C/min and Ar flow at ambient pressure. For the synthesis of WSe2, the center heating zone with the WO3powders was

heated to 925
C, while the Se powders placed at the upper stream side were maintained at 270
C. The WSe2samples were

obtained by placing the substrate at the downstream side using an Ar/H2flowing gas at 1 Torr for 15 min.

Transfer of Monolayer TMD To Form Bilayers. The artificially stacked TMD bilayers were fabricated by transferring monolayer TMD flakes onto another as-grown monolayer TMD formed on a sapphire substrate. For MoS2transfer, a layer of

poly-(methylmethacrylate) (PMMA) (950K A4) was first coated on the monolayer MoS2flakes as a transfer supporting layer by

spin-coating (step 1, 500 rpm for 10 s; step 2:, 3000 rpm for 60 s), followed by baking at 100
C for 60 min. Then the PMMA-capped MoS2was exfoliated from the sapphire substrate in a

2 M NaOH solution at 100
C for 60 min. After dilution of etchant and residues in deionized water, the PMMA-capped monolayer MoS2 was transferred onto another as-grown MoS2 sample,

followed by the removal of PMMA film using acetone.26_Finally,

the sample was cleaned by isopropyl alcohol and deionized water. According to the atomic force microscopy (AFM) analyses (Supporting Information), the thickness of MoS2 monolayer

(bilayer) is 0.65 nm (1.3 nm). However, we noted that the PMMA cannot be completely removed by using acetone only. To avoid PMMA contamination in the interlayer region, the PMMA layer was always kept on top of the MoS2layers during the transfer

process. In this way, the PMMA residues will be mostly residing on the TMD surface. After 300
C annealing in Ar/H2gas flow for

4 h, these PMMA residues can be completely removed. We have also compared SHG from the stacked bilayers with and without thermal annealing. The measured SHG behavior is basically the same. This indicated that the surface PMMA residues have negligible effect on the interlayer coupling of SHG. All the data presented in this work are obtained from samples without thermal annealing.

Optical Measurements. The SHG measurements were per-formed in a homemade microscopy system. The fundamental laser field was provided by a mode-locked Ti:sapphire laser at a wavelength of 810 nm with a pulse width of∼150 fs and a repetition rate of 80 MHz. The laser pulse was focused to a spot size of∼1.1 μm on the sample by a 100 objective lens. The generated SH light was sent to a 0.75 m monochromator and detected by a nitrogen-cooled CCD camera. For spatial map-ping of SH intensity, the sample was mounted on a fast motorized x
y stage with a step of 0.25 μm. For polarization-resolved SHG, the sample was rotated by a motorized rotational stage with a step of 5
. The linear polarizations of the funda-mental and the SH light were selected and analyzed by indivi-dual polarizers and half-wave plates. According to the measured SH intensity, we estimated that the magnitude of the second-order susceptibilityχ(2) _{(bulk value) of our}

CVD-grown MoS2 monolayer is about χ(2) = 1.5  10
10 m/V.

Although the absoluteχ(2)value depends sensitively on the experimental conditions (such as the laser wavelength, spot size, pulse width/shape and the collected solid angle), the

deducedχ(2) value is very close to the reported value for MoS2 monolayers prepared by exfoliation from natural

crystals.9,10For differential reflectance measurements, the ap-paratus are the same as the SHG measurements, except that the light source was replaced by a fiber-coupled tungsten-halogen lamp.

Conflict of Interest: The authors declare no competing financial interest.

Acknowledgment. This work was supported in part by the National Science Council of Taiwan under Grant Nos. NSC101-2628-M-009-002-MY3 and NSC102-2119-M-001-005-MY3. W.-H.C. acknowledges the supports from the Center for Interdisciplinary Science under the MOE-ATU project for NCTU. L.-J.L. acknowl-edges the supports from the Academia Sinica (IAMS and Nano program).

Supporting Information Available:AFM analyses and optical characterizations, including Raman, PL and differential reflec-tance measurements for identifying the number of layers and material quality; determination of the absorption coefficient from the differential reflectance spectra; the effects of multi-layer reflections and absorptions on the SHG; the second-order susceptibilityχ(2)_{for bilayers; and the wavelength dependent}

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

REFERENCES AND NOTES

1. Wang, Q. H.; Kalantar-Zadeh, K.; Kis, A.; Coleman, J. N.; Strano, M. S. Electronics and Optoelectronics of Two-Dimensional Transition Metal Dichalcogenides. Nat. Nano-technol.2012, 7, 699.

2. Mak, K. F.; Lee, C.; Hone, J.; Shan, J.; Heinz, T. F. Atomically Thin MoS2: A New Direct-Gap Semiconductor. Phys. Rev.

Lett.2010, 105, 136805.

3. Xiao, D.; Liu, G.-B.; Feng, W.; Xu, X.; Yao, W. Coupled Spin and Valley Physics in Monolayers of MoS2 and Other

Group-VI Dichalcogenides. Phys. Rev. Lett. 2012, 108, 196802.

4. Cao, T.; Wang, G.; Han, W.; Ye, H.; Zhu, C.; Shi, J.; Niu, Q.; Tan, P.; Wang, E.; Liu, B.; Feng, J. Valley-Selective Circular Dichroism of Monolayer Molybdenum Disulphide. Nat. Commun.2012, 3, 887.

5. Zeng, H.; Dai, J.; Yao, W.; Xiao, D.; Cui, X. Valley Polarization in MoS2Monolayers by Optical Pumping. Nat.

Nanotech-nol.2012, 7, 490–493.

6. Mak, K. F.; He, K.; Shan, J.; Heinz, T. F. Control of Valley Polarization in Monolayer MoS2by Optical Helicity. Nat.

Nanotechnol.2012, 7, 494–498.

7. Zeng, H.; Liu, G.-B.; Dai, J.; Yan, Y.; Zhu, B.; He, R.; Xie, L.; Xu, S.; Chen, X.; Yao, W.; Cui, X. Optical Signature of Symmetry Variations and Spin-Valley Coupling in Atomically Thin Tungsten Dichalcogenides. Sci. Rep.2013, 3, 1608. 8. Kumar, N.; Najmaei, S.; Cui, Q.; Ceballos, F.; Ajayan, P. M.;

Lou, J.; Zhao, H. Second Harmonic Microscopy of Mono-layer MoS2. Phys. Rev. B2013, 87, 161403(R).

9. Malard, L. M.; Alencar, T. V.; Barboza, A. P. M.; Mak, K. F.; de Paula, A. M. Observation of Intense Second Harmonic Generation from MoS2 Atomic Crystals. Phys. Rev. B

2013, 87, 201401(R).

10. Li, Y.; Rao, Y.; Mak, K. F.; You, Y.; Wang, S.; Dean, C. R.; Heinz, T. F. Probing Symmetry Properties of Few-Layer MoS2and

h-BN by Optical Second-Harmonic Generation. Nano Lett. 2013, 13, 3329–3333.

11. Kim, C.-J.; Brown, L.; Graham, M. W.; Hovden, R.; Havener, R. W.; McEuen, P. L.; Muller, D. A.; Park, J. Stacking Order Dependent Second Harmonic Generation and Topological Defects in h-BN Bilayers. Nano Lett.2013, 13, 5660–5665. 12. Li, G.; Luican, A.; Lopes dos Santos, J. M. B.; Castro Neto, A. H.; Reina, A.; Kong, J.; Andrei, E. Y. Observation of Van Hove Singularities in Twisted Graphene Layers. Nat. Phys. 2010, 6, 109–113.

13. Havener, R. W.; Zhuang, H.; Brown, L.; Hennig, R. G.; Park, J. Angle-Resolved Raman Imaging of Interlayer Rotations and Interactions in Twisted Bilayer Graphene. Nano Lett. 2012, 12, 3162–3167.

14. Kim, K.; Coh, S.; Tan, L. Z.; Regan, W.; Yuk, J. M.; Chatterjee, E.; Crommie, M. F.; Cohen, M. L.; Louie, S. G.; Zettl, A. Raman Spectroscopy Study of Rotated Double-Layer Graphene: Misorientation-Angle Dependence of Electronic Structure. Phys. Rev. Lett.2012, 108, 246103.

15. Ponomarenko, L. A.; Gorbachev, R. V.; Yu, G. L.; Elias, D. C.; Jalil, R.; Patel, A. A.; Mishchenko, A.; Mayorov, A. S.; Woods, C. R.; Wallbank, J. R.; et al. Cloning of Dirac Fermions in Graphene Superlattices. Nature2013, 497, 594–597. 16. Geim, A. K.; Grigorieva, I. V. Van der Waals

Heterostruc-tures. Nature2013, 499, 419–425.

17. Britnell, L.; Ribeiro, R. M.; Eckmann, A.; Jalil, R.; Belle, B. D.; Mishchenko, A.; Kim, Y.-J.; Gorbachev, R. V.; Georgiou, T.; Morozov, S. V.; et al. Strong Light-Matter Interactions in Heterostructures of Atomically Thin Films. Science2013, 340, 1311–1314.

18. Britnell, L.; Gorbachev, R. V.; Jalil, R.; Belle, B. D.; Schedin, F.; Mishchenko, A.; Georgiou, T.; Katsnelson, M. I.; Eaves, L.; Morozov, S. V.; et al. Field-Effect Tunneling Transistor Based on Vertical Graphene Heterostructures. Science 2012, 335, 947–950.

19. Terrones, H.; López-Urías, F.; Terrones, M. Novel Hetero-Layered Materials with Tunable Direct Band Gaps by Sandwiching Different Metal Disulfides and Diselenides. Sci. Rep.2013, 3, 1549.

20. Kou, L.; Frauenheim, T.; Chen, C. Nanoscale Multilayer Transition-Metal Dichalcogenide Heterostructures: Band Gap Modulation by Interfacial Strain and Spontaneous Polarization. J. Phys. Chem. Lett.2013, 4, 1730–1736. 21. Lee, Y.-H.; Zhang, X.-Q.; Zhang, W.; Chang, M.-T.; Lin, C.-T.;

Chang, K.-D.; Yu, Y.-C.; Wang, J. T.-W.; Chang, C.-S.; Li, L.-J.; et al. Synthesis of Large-Area MoS2Atomic Layers with

Chemical Vapor Deposition. Adv. Mater.2012, 24, 2320– 2325.

22. Boyd, R. Nonlinear Optics; Academic Press: London, 2008. 23. Beal, A. R.; Hughes, H. P. Kramers-Krönig Analysis of the Reflectivity Spectra of 2H-MoS2, 2H-MoSe2and 2H-MoTe2.

J. Phys. C: Solid State Phys.1979, 12, 881–890.

24. Lee, Y.-H.; Yu, L.; Wang, H.; Fang, W.; Ling, X.; Shi, Y.; Lin, C.-T.; Huang, J.-K.; Chang, M.-T.; Chang, C.-S.; et al. Synthesis and Transfer of Single-Layer Transition Metal Disulfides on Diverse Surfaces. Nano Lett.2013, 13, 1852–1857. 25. Huang, J.-K.; Pu, J.; Hsu, C.-L.; Chiu, M.-H.; Juang, Z.-Y.;

Chang, Y.-H.; Chang, W.-H.; Iwasa, Y.; Takenobu, T.; Li, L.-J. Large-Area Synthesis of Highly Crystalline WSe2

Mono-layers and Device Applications. ACS Nano2014, 8, 923– 930.

26. Zhang, W.; Chuu, C.-P.; Huang, J.-K.; Chen, C.-H.; Tsai, M.-L.; Chang, Y.-H.; Liang, C.-T.; Chen, Y.-Z.; Chueh, Y.-L.; He, J.-H.; et al. Ultrahigh-Gain Photodetectors Based on Atomically Thin Graphene-MoS2Heterostructures. Sci. Rep.2014, 4,

3826.