Optoelectronic crystal of artificial atoms in strain-textured molybdenum disulphide

Abstract

The isolation of the two-dimensional semiconductor molybdenum disulphide introduced a new optically active material possessing a band gap that can be facilely tuned via elastic strain. As an atomically thin membrane with exceptional strength, monolayer molybdenum disulphide subjected to biaxial strain can embed wide band gap variations overlapping the visible light spectrum, with calculations showing the modified electronic potential emanating from point-induced tensile strain perturbations mimics the Coulomb potential in a mesoscopic atom. Here we realize and confirm this 'artificial atom' concept via capillary-pressure-induced nanoindentation of monolayer molybdenum disulphide from a tailored nanopattern, and demonstrate that a synthetic superlattice of these building blocks forms an optoelectronic crystal capable of broadband light absorption and efficient funnelling of photogenerated excitons to points of maximum strain at the artificial-atom nuclei. Such two-dimensional semiconductors with spatially textured band gaps represent a new class of materials, which may find applications in next-generation optoelectronics or photovoltaics.

ARTICLE

Received 25 Feb 2015 | Accepted 1 May 2015 | Published 19 Jun 2015

Hong Li1, Alex W. Contryman2, Xiaofeng Qian3, Sina Moeini Ardakani4, Yongji Gong5, Xingli Wang5, Jeffery M. Weisse1, Chi Hwan Lee1, Jiheng Zhao1, Pulickel M. Ajayan5, Ju Li6, Hari C. Manoharan7& Xiaolin Zheng1

The isolation of the two-dimensional semiconductor molybdenum disulphide introduced a new optically active material possessing a band gap that can be facilely tuned via elastic strain. As an atomically thin membrane with exceptional strength, monolayer molybdenum disulphide subjected to biaxial strain can embed wide band gap variations overlapping the visible light spectrum, with calculations showing the modied electronic potential emanating from point-induced tensile strain perturbations mimics the Coulomb potential in a mesoscopic atom. Here we realize and conrm this articial atom concept via capillary-pressure-induced nanoindentation of monolayer molybdenum disulphide from a tailored nanopattern, and demonstrate that a synthetic superlattice of these building blocks forms an optoelectronic crystal capable of broadband light absorption and efcient funnelling of photo-generated excitons to points of maximum strain at the articial-atom nuclei. Such two-dimensional semiconductors with spatially textured band gaps represent a new class of materials, which may nd applications in next-generation optoelectronics or photovoltaics.

1 Department of Mechanical Engineering, Stanford University, Stanford, California 94305, USA. 2 Department of Applied Physics, Stanford University, Stanford, California 94305, USA. 3 Department of Materials Science and Engineering, Dwight Look College of Engineering, Texas A&M University, College Station, Texas 77843, USA. 4 Department of Civil and Environmental Engineering, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, USA. 5 Department of Materials Science & NanoEngineering, Rice University, Houston, Texas 77251, USA. 6 Department of Nuclear Science and Engineering and Department of Materials Science and Engineering, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, USA.

7 Department of Physics, Stanford University, Stanford, California 94305, USA. Correspondence and requests for materials should be addressed to X.Z. (email: mailto:[email protected]

Web End [email protected] ).

NATURE COMMUNICATIONS | 6:7381 | DOI: 10.1038/ncomms8381 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 1

& 2015 Macmillan Publishers Limited. All rights reserved.

DOI: 10.1038/ncomms8381 OPEN

Optoelectronic crystal of articial atoms in strain-textured molybdenum disulphide

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms8381

Straining two-dimensional (2D) materials18 with a spatially varying designer strain can lead to new articial materials with exotic properties. Assembling such tuned articial

materials in condensed matter is an emerging eld and has employed tools such as atomic manipulation9 and lithographic techniques10. Here we focus on monolayer molybdenum disulphide (MoS2), a direct band gap semiconductor that shows promise for applications in photonics and optoelectronics due to its extraordinary physical properties1118. Elastic strain offers a novel and exciting opportunity to tune the band gap of monolayer MoS2 (refs 14) since it can sustain very high elastic strain before rupturing compared with its bulk counterpart19,20.

For instance, uniaxial strain has been shown to modulate the electronic structure57 and reduce the optical band gap (OBG) of monolayer MoS2 up to 100 meV6.

Besides pushing the magnitude of the strain, the ability to apply a spatially controllable strain is even more crucial because it enables the realization of a graded band gap semiconductor eagerly sought for wider photonic, optoelectronic and photovoltaic applications21,22. It has been calculated that indenting monolayer MoS2 creates a tensile strain eld that reduces the local quasiparticle band gap (QBG). The resulting modied electronic potential falls off inversely with distance from the indentation, playing the role of an effective electronic potential centred on a mesoscopic articial atom8. This electronic potential funnels photogenerated excitons from larger band gap regions into smaller band gap regions, resulting in potential broader spectrum light absorption and more efcient photocarrier concentration. The exciton funnel concept has been successfully employed to interpret the OBG of wrinkled few-layer MoS2. However, the indirect band gap of few-layer MoS2 prevents the distinctive photoluminescence (PL) intensity enhancement expected from the exciton funnelling process from being directly observed23. Assembly of the aforementioned articial atoms would result in a large-area functional articial crystal. Currently there exists no fundamental proof-of-principle of this idea, neither at the single articial atom level nor at a more practical macroscopic scale.

In the following, we apply spatially modulated biaxial tensile strain in monolayer MoS2 using a patterned nanocone substrate to realize the optoelectronic crystal consisting of articial atoms. MoS2 on top of the nanocones experiences high strain, which gradually decreases from the tip to the perimeter of the nanocones (Fig. 1a). Since the band gap decreases with increasing tensile strain, the nanocone apex marks the nucleus of the articial atom. The strain prole has been optimized for maximum exciton solar funnel collection by controlling the geometry and dimensions of the nanocones.

ResultsCreation of the articial-atom crystal in strain-textured MoS2.

The as-grown MoS2 sheet (Supplementary Fig. 1) is transferred onto the SiO2 nanocones (Fig. 1b) assembled using nanosphere lithography24,25 (Supplementary Fig. 2). The as-transferred MoS2 is soaked in ethylene glycol to remove the trapped air bubbles and optimize the strain (Supplementary Fig. 3b). The evaporation of ethylene glycol that lls the gap between the MoS2 and nanocones generates capillary force that pulls down the MoS2 sheet, causing it to conform to the topography of the nanocones and accomplishing the nanoindentation. The clearly visible wrinkles between nanocones (Fig. 1c) indicate the presence of deformation and strain in the MoS2 sheet, which is also veried by atomic force microscopy (AFM; Fig. 1d). A scanning tunnelling microscopy (STM) topograph (Fig. 1e) shows the details of a single MoS2 articial-atom element. We note that a similar attempt for templating strain has been carried out in graphene with limited strain magnitude due to a not-yet optimized process25,26.

Scanning Raman spectroscopy of the articial-atom crystal. The spatially varying strain distribution is veried by micro-Raman spectroscopy. Figure 2a shows the typical Raman spectra of the most strained-MoS2 (on tip of nanocones), less strained-

MoS2 (between nanocones) and unstrained-MoS2 (on at SiO2

a

b e

Strain crystal Less strained

Monolayer MoS2

Most strained Nanocone

Quasiparticle energy gap

Distance

Artificial atom

Single MoS2 artificial atom

z (nm)

Nanocone

SiO2

p++ Si

20%

c d

2D MoS2 strain crystal 2D MoS2 strain crystal

0 80

Figure 1 | Assembly of an articial-atom crystal via spatially patterned biaxial strain within a monolayer of MoS2. (a) Schematic of strained MoS2 indented by SiO2 nanocones, where the regions on the tips of nanocones exhibit highest tensile strain while the areas between nanocones are less strained. The MoS2 energy band gap inversely tracks the strain prole and becomes spatially modulated, forming articial atoms at the points of peak strain where the strain-induced potential mimics the Coulomb potential around ions in a crystal. (b) Cross-sectional scanning electron microscopy (SEM) image of the nanocone substrate. Dotted lines label the SiO2/p Si and SiO2/vacuum boundaries. Solid curve delineates the nanocone shape. Scale bar is 500 nm.

(c) Tilted false-colour SEM image of the 2D strained MoS2 crystal dened by the nanocone array. Scale bar is 500 nm. (d) AFM topography of the 2D MoS2 strain crystal. Scale bar is 1 mm. (e) STM topography of a single articial atom building block within the crystal. Scale bar is 100 nm.

2 NATURE COMMUNICATIONS | 6:7381 | DOI: 10.1038/ncomms8381 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

& 2015 Macmillan Publishers Limited. All rights reserved.

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms8381 ARTICLE

a b c

Mo S

360 380 400 420

A E peak frequency A peak frequency

cm1 cm1385.9

381.1

Intensity (a.u.)

405.2

403.8

MS-MoS2

LS-MoS2

US-MoS2

E

Raman shift (cm1)

Figure 2 | Scanning Raman spectroscopy of the MoS2 strain crystal. (a) Raman spectra of most strained-MoS2, less strained-MoS2 and unstrained-MoS2. Symbols are measurement data; curves are tting data. Inset: schematic atomic displacement of the in-plane E12g and out-of-plane A1g modes. (b,c)

Scanning Raman spectroscopic maps plotting (b) E12g peak frequency and (c) A1g peak frequency of strain-textured MoS2 on the nanocone substrate. Scale bars are 1 mm.

A exciton MS-MoS2

LS-MoS2

US-MoS2 B exciton

PL energy (eV)

a b

surface) [Supplementary Fig. 4]. The unstrained-MoS2 has the typical Raman spectrum of monolayer MoS2 with two dominant peaks at 385.6 (E12g) and 404.9 cm 1 (A1g)27,28. The strained

MoS2 samples show signicant redshifts for both E12g and A1g peaks: 0.8 and 0.3 cm 1 for less strained-MoS2, and 2.4 and0.6 cm 1 for most strained-MoS2. From an overall t to the redshift magnitudes and the unstrained values, we estimate that(0.2300.035)% and (0.5650.025)% biaxial tensile strains are sampled on average by the 450-nm diameter laser beam in less strained-MoS2 and most strained-MoS2 according to our theoretical calculation discussed later. The Raman mappings of the E12g (Fig. 2b) and A1g (Fig. 2c) peak frequencies show that the periodically varying strain is consistent over the entire scanned articial crystal (25 mm2). The darker (lower frequency) and brighter (higher frequency) colours (Fig. 2b,c) indicate that tensile strain is highest on the tips of nanoconesat articial atom nucleiand gradually decreases to a minimum towards the perimeters of the nanocones.

Scanning PL spectroscopy of the articial-atom crystal. The effect of strain on the OBG of MoS2 strain crystal is examined by micro-PL spectroscopy. Figure 3a compares the representative PL spectra of the most strained-MoS2, less strained-MoS2, and unstrained-MoS2, where the strong PL peaks arise from the direct band gap emissions in monolayer MoS2. First, unstrained-MoS2 shows a typical PL spectrum of monolayer MoS2 with a principal peak at 1.83 eV (A exciton) and a minor peak at 1.95 eV (B exciton)11. Second, the strained MoS2 clearly exhibits redshift of the A exciton energy: 18 and 50 meV for the less strained-MoS2 and most strained-MoS2, respectively, indicating a strain-induced OBG reduction6,7. From the magnitude of the redshift and our theoretical calculation, we estimate approximately 0.2 and 0.5% biaxial tensile strains are optically sampled in less strained-MoS2 and most strained-MoS2, respectively. These values agree well with the strains derived from Raman peak shifts above. Third, the mapping of the PL wavelength (Fig. 3b) shows that spatially varying OBG reduction is consistently observed over scanned areas. Finally and signicantly, the A exciton peak intensity is more than doubled in the most strained-MoS2 (Fig. 3a), and the intensity increase is highly reproducible over the scanned region (Fig. 3c). There are a few factors that may affect the PL intensity including strain-modulated oscillator strength, varying local emission geometry, variation in underlying SiO2 thickness and strain-induced exciton funnelling29. First, according to our exciton calculations (see Discussion below), the change of the oscillator strength on elastic strain is relatively small (B6%

increase under 1% biaxial strain), which cannot explain the observed 4100% PL intensity enhancement. Second, the geometry-dependent PL intensity is expected to be enhanced on

a at substrate as the PL intensity of monolayer MoS2 peaks at normal collection angle30,31, which is opposite to our observation. Third, a recent study of PL emission from MoS2 shows a at dependence on underlying SiO2 thickness in the range of 180 270 nm, corresponding to oxide thicknesses variations in our samples from etching the nanocones32. In fact, the PL emission intensity of homogeneously strained monolayer MoS2 was experimentally found to decrease when strain increases due to the increased probability of non-radiative relaxations6, again opposite our observations. Finally, we note that enhanced PL intensity due to exciton drifting and concentrating has been observed in strained semiconductor nanowires29,33,34. Therefore, we rule out alternate factors discussed above and attribute the anomalous enhanced PL intensity to the novel 2D exciton funnel effect8. The Raman mappings (Fig. 2b,c) show that the tensile strain increases from the perimeters to the tips of the nanocones, so the band gap of MoS2 decreases from the perimeters to the tips of the nanocones. Consequently, a built-in electric eld pointing

Wavelength

nm

700

674

Intensity (a.u.)

1.7 1.8 1.9 2.0

c d

Intensity CCD

counts

6,400

3,100

I. Exciton

generation

II. Exciton drifting

III. Exciton

recombination

Figure 3 | Scanning PL spectroscopy of the MoS2 strain crystal.(a) PL spectra of most strained-MoS2, less strained-MoS2 and unstrained-

MoS2. Symbols are measurement data; curves are tting data. Scanning PL maps with (b) wavelength and (c) peak intensity of the same sample and area depicted in Fig. 2. Scale bars are 1 mm. (d) Schematic of the funnel effect that consists of three processes: (I) excitons are induced in MoS2 upon illumination; (II) photogenerated excitons drift in the articial atom potential towards the atom center formed by the nanocone tip; and(III) concentrated excitons give emission with longer wavelength.

NATURE COMMUNICATIONS | 6:7381 | DOI: 10.1038/ncomms8381 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 3

& 2015 Macmillan Publishers Limited. All rights reserved.

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms8381

from the tip towards the perimeter of the nanocone is created (Fig. 3d) and the photogenerated excitons drift towards the funnel center, as can be modelled analytically and numerically29. Since the photogenerated excitons in monolayer MoS2 can drift up to 660 nm before recombination8, the majority of photogenerated excitons are able to drift from the valleys to the tips of nanocones ( 245 nm from pitch radius) without signicant recombination.

As a result, excitons are concentrated at the tips of nanocones upon illumination and they eventually recombine, giving rise to the enhanced PL emission localized at the nanocone tips (Fig. 3c,d) and supporting the articial atom and energy funnel concept.

STM and STS measurements of the articial-atom crystal. The local electronic structure modulation of MoS2 caused by strain is directly veried by STM. Figure 4a shows an STM topograph (V 2.4 V, I 18 pA) in the vicinity of three nanocones with

height variations of 80 nm. Scanning tunnelling spectroscopy (STS) is performed in different regions of the area shown, and representative dI/dV tunnel spectra for unstrained-MoS2, less strained-MoS2, and most strained-MoS2 are shown in Fig. 4b. It is worth noting that these spectra show the Fermi level near mid-gap position due to the electron depletion caused by the gold

substrate35. The QBG sizes extracted from the dI/dV spectra (Supplementary Fig. 5) are labelled at each strain level. Unstrained-MoS2 shows a QBG of 2.29 eV, which is consistent with the OBG of 1.83 eV obtained by PL measurement considering the exciton binding energy B0.5 eV8. The extracted local QBG varies from2.29 eV down to 1.83 eV from STS measurements acquired from unstrained-MoS2 (topographic valley), through less strained-MoS2 (intermediate regions) and to most strained-MoS2 (topographic peak) locations. This corresponds to a measured local strain approaching B3% for most strained-MoS2. These strains are much higher than the maximum of B0.6% measured by Raman spectroscopy, which can be explained by the inherently atomic scale measurement of STM/STS compared with the optical measurements, which average over the nite laser beam size (B450 nm).

To quantify this argument and verify the complete strain prole of the crystal, we simulate the strain induced in a monolayer MoS2 sheet by applying a constant pressure on the sheet, and LennardJones interatomic interactions between the sheet and substrate (Fig. 4c, Supplementary Fig. 6). To calibrate the mapping between the local tunnelling measurements and the optically averaged Raman/PL measurements, we plot in Fig. 4d the cone-to-cone calculated strain proles (Fig. 4c) before and after a convolution step. The raw atomic strain ebi is convolved

a b c e

100

75

50

25

Optically averaged strain [afii9830]bi (%)

Local

MoS2

QBG

0.0

0.2

0.4

0.6 0.8 1.0

390

405

404

403

A 1gfrequency (cm

[afii9830]bi (%)

3

3

1 )

1 )

LS-MoS 2MS-MoS 2

US-MoS 2

388

386

384

382

3802.32.22.12.01.91.81.71.6

E1 2gfrequency (cm

1.83 V

2

1

600

dl/dV(pA per V)

0 0 200 200

200

400 y (nm)

E12g phonon

x (nm)

400 600

0

d

80 MS

z (nm) LS

0 US

Optically averaged strain [afii9830] bi(%)

Quasiparticle energy gap (eV)

1.841.821.801.781.761.741.721.70

Exciton energy (eV)

A exciton

2.01 V

0.5

0.4

2

Direct

Indirect

0 1 2 3 4 5

0.3

0.2

1

Local gap

A1g phonon

2.29 V

US LS MS

0.1

0.0

0 2 1 0 1 2

00

400

600

800

Sample bias (V)

Local atomic strain [afii9830] bi(%)

Distance (nm)

0.6

Local atomic strain [afii9830]bi (%)

Figure 4 | STM/STS of the MoS2 strain crystal and experimenttheory correlation. (a) STM topography in the vicinity of three nanocones. Height colour bar indicates unstrained (topographic valleys), less strained (intermediate strain) and most strained (topographic peaks) to denote the amount of local strain expected at each height z (dashed arrow indicates direction of increasing strain). Scale bar is 100 nm. (b) Representative local dI/dV measurements acquired from the region shown in (a) from unstrained (grey), less strained (blue) and most strained (red) locations. Each strain shows three individual dI/dV measurements (open symbols), which are acquired in tight groups o10 nm in extent over each location. The solid lines are the averages of each group of three curves. Band gaps extracted from the curves are enumerated for each strain amount with the trend indicated by dashed arrows.

(c) Calculated local biaxial strain distribution ebi over several unit cells of the crystal. (d) Cross-section of the strain prole showing local atomic strain ebi (red), measured by STS and ranging from 0 to 2.85% tensile strain, and the optically averaged strainbi (green) sampled by scanning Raman/PL measurements, ranging from 0.23 to 0.56% tensile strain. The broadened distribution is calculated from theoretical local strain using the 450-nm optical

Gaussian beam diameter. (e) Top panel shows theoretical (open symbols) and experimental (lled circles) Raman frequencies for the indicated phonon modes as a function of strain. Dashed lines are linear ts to the theory ( 1.02 cm-1 per % for A1g phonon and 4.48 cm-1 per % for E12g phonon) and are

used for the optically averaged strain assignment (top axis). Bottom panel shows theoretical (open symbols) and experimental (closed symbols) data for the OBG (magenta, right axis) and the local QBG (cyan, left axis). Axes have been scaled using the results of (d) as discussed in main text. Dashed lines match t of 0.11 eV per % (direct OBG) and dotted line shows an increased 0.24 eV per % (indirect OBG) falloff of the OBG beyond 2% biaxial strain.

Star symbols denote the three specic measurements shown in (b). Error bars are determined by s.d. of each spectroscopic measurement.

4 NATURE COMMUNICATIONS | 6:7381 | DOI: 10.1038/ncomms8381 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

& 2015 Macmillan Publishers Limited. All rights reserved.

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms8381 ARTICLE

with a 2D Gaussian of 450-nm 1/e2 width, corresponding to the 450-nm diameter Gaussian beam used in our optical experiments. With no tting parameters, this yields a predicted optically averaged strain sampled by scanning Raman/PL; this average strainbi ranges from 0.233 to 0.562% and is in excellent agreement with the Raman/PL results from less strained-MoS2 and most strained-MoS2 regions presented above. This result provides conrmation that the local atomic strain ebi is approaching the unprecedentedly high value of 3%.

DiscussionWe use this calibration to unify all experimental (closed symbols) and theoretical (open symbols) results (Fig. 4e). We calculated Raman spectra underbi from 0 to 1% (red and blue triangles in

Fig. 4e; Supplementary Fig. 7a). Asbi increases from 0 to 1%, the E12g and A1g peaks shift towards lower frequencies. Such Raman peak shifts vary linearly with biaxial strain, and the slopes are

4.48 and 1.02 cm 1 per 1% of biaxial strain for E12g and A1g

modes, respectively (dashed red and blue lines in Fig. 4e). The t to all Raman data in unstrained, less strained and most strained regions with respect to these two lines yieldsbi values quoted earlier, (0.2300.035)% (less strained) and (0.5650.025)% (most strained) that are shaded in Fig. 4e. The calculated A exciton energy at variousbi ranging from 0 to 1% is shown in

Fig. 4e (lower panel, open purple triangles; Supplementary Fig. 7b). The A exciton peak exhibits a linear redshift rate of

0.11 eV per 1% of biaxial tensile strain (dashed purple line in Fig. 4e, also in good agreement with other calculations1,4). Our recorded PL energies when assigned to the Raman-extracted strain values above fall along this redshift line within experimental error.

The above discussion correlates OBG measurements tobi. To make the nal quantitative link to the intrinsic QBG and the local atomic strain ebi within the articial crystal, we use the results of the convolution analysis elaborated above. Accordingly, we show in Fig. 4e a local atomic strain axis ebi, which is linked to the optically averaged strain axisbi by the ratio of maxebi:maxbi 2.85:0.562 B 5:1. On this axis we rst plot

the calculated theoretical QBG from 0 to 5% local strain for direct (open cyan triangles) and indirect (open cyan circles) transitions8. The t to the direct gap calculations (Fig. 4e, dashed cyan line) also diminishes at 0.11 eV per % biaxial strain and the gap axis

there is scaled by the same 5:1 ratio to show this match. The calculations8 furthermore reveal a direct-to-indirect gap transition at B2% biaxial tensile strain, resulting in an increased drop of B 0.24 eV per % of the (now indirect) gap with strain

(Fig. 4e, dotted cyan line). We add other acquired STS data to this plot. The closed cyan star symbols correspond to the curves shown in Fig. 4b and provide the anchor for the strain measurement: the highest QBG of 2.29 eV corresponds to zero local strain, and the lowest QBG of 1.83 eV corresponds to the highest strain at the articial-atom nanocone tips, and must correspond to the largest optically averaged strains deduced. The remaining STS measurements have known gap values, and strain values that can be inferred by assigning ebi to each value along the dotted line: gaps falling at 0.11 eV per % from the unstrained-

MoS2 value and rising at 0.24 eV per % from the most strained-MoS2 value, meeting at the 2% direct-to-indirect transition. We note that the strain magnitude of 3% is calculated based on biaxial strain. This is the lower bound as larger strain magnitude is necessary to achieve the observed band gap modulation if the strain is uniaxial, which could exist in local areas due to the imperfect geometry of the nanocones.

We note that this summary shows that our local crystal strain ebi is sufciently high to access the indirect band gap of

monolayer MoS2. However, because the Raman/PL measurements are optically averaged, they are dominated by the majority of the points within the laser beam that have direct transition as evidenced by the high PL efciency. The measurements also conrm a remarkably large exciton binding energy of B0.5 eV, only recently veried to be a general feature of dichalcogenides36. From the STS measurements, a modulation of over 20% in the QBG is directly observed. From the mapping to the optically averaged strain (Fig. 4) and the deduced exciton binding energy, this represents a huge modulation of the OBG, exceeding 25%.

This engineered OBG enables a broadband light absorption by increasing the absorption bandwidth from 677 nm (unstrained-MoS2) to 905 nm (most strained-MoS2), which covers the entire visible wavelength and most intensive wavelengths of the solar spectrum. While already without precedent, since these strains are not yet even rupture limited, we anticipate even larger band gap variations and electronic elds can be embedded in such articial crystals in the future.

Methods

Transfer and strain of MoS2 monolayer. MoS2 monolayers grown by chemical vapour deposition were transferred onto the SiO2 nanocone substrate using the PMMA-assisted wet transfer process. The transferred sample was then baked at 100 C for 30 min and the PMMA was removed by sequential soaking in acetone at 60 C for 10 min followed by chloroform at 60 C for 1 h. Afterwards, the SiO2 nanocone with transferred MoS2 was immersed in ethylene glycol in vacuum for 1 h to ensure that both sides of the MoS2 sheet were wetted by ethylene glycol. Lastly, the sample was dried in ambient air to completely evaporate ethylene glycol.

Raman and photoluminescence characterizations. The Raman and PL measurements were performed with the excitation laser line of 532 nm using a WITEC alpha500 Confocal Raman system in ambient air environment. The power of the excitation laser line was kept below 1 mW to avoid damage of MoS2. The

Raman scattering was collected by an Olympus 100 objective (N.A. 0.9) and

dispersed by 1,800 (for Raman measurements) and 600 (for PL measurements) lines per mm gratings.

STM/STS measurements. STM/STS measurements were performed at 77 K in ultrahigh vacuum. A bilayer of titanium (10 nm) and gold (80 nm) lms were deposited onto the silicon oxide nanocone surface, then monolayer MoS2 akes were transferred and strained on the metalized nanocone surface. A control sample was prepared alongside the STM sample and characterized using SEM and AFM to ensure conformal coating of MoS2 on the metalized nanocones. The STM sample was annealed in the STM ultrahigh vacuum chamber at 200 oC for 1 h to clean the MoS2 surface for a reliable STM measurement.

Modelling of strain distribution of MoS2 on nanocones. The MoS2 sheet was modelled using honeycomb lattice with Tersoff potential and the substrate was modelled as a fcc lattice. A LennardJones potential was employed to include the Van der Waals interaction between substrate and the MoS2 sheet. A constant force was placed on each atom downward to emulate the pressure difference across the MoS2 sheet (see Supplementary Fig. 6 for more details).

Theoretical Raman spectra of monolayer MoS2. The theoretical biaxial strain-dependent Raman spectra were calculated using rst-principles density-functional perturbation theory implemented in the Quantum-ESPRESSO package with a plane-wave cutoff of 120 Rydberg, a MonkhorstPack k-point sampling of12 12 1, and an exchange correlation functional of the PerdewZunger form

within the local density approximation. The spin-orbit coupling was not included. In addition, norm-conserving HartwigsenGoedeckerHutter pseudopotentials were used to take into account the core electrons and reduce the computational efforts. All biaxially strained conguration were fully relaxed with a convergence criteria of 0.0001 a.u. for the maximal residual force. The calculation was carried out in a periodic supercell with a vacuum spacing of 20 along the z (plane normal) direction to reduce the spurious interaction between the neighbouring unit cells.

Theoretical optical absorption spectra of monolayer MoS2. The theoretical biaxial strain-dependent optical absorption spectra were calculated by solving the BetheSalpeter equation within the TammDancoff approximation. The key parameters used in the BetheSalpeter equation including quasiparticle energies and screened-Coulomb interactions calculated were obtained from many-body perturbation theory with the Hedins GW approximation. All the calculations were

NATURE COMMUNICATIONS | 6:7381 | DOI: 10.1038/ncomms8381 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 5

& 2015 Macmillan Publishers Limited. All rights reserved.

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms8381

performed using the Vienna Ab initio Simulation Package with plane-wave basis and the projector-augmented wave method. A plane-wave cutoff of 350 eV, a MonkhorstPack k-point sampling of 18 18 1, and an exchange correlation

functional of the PerdewBerkeErnzerhof form within the generalized gradient approximation were used. All biaxially strained congurations were fully relaxed with the maximal residual force of r0.0001 eV 1 using density-functional theory calculations.

References

1. Johari, P. & Shenoy, V. B. Tuning the Electronic Properties of Semiconducting Transition Metal Dichalcogenides by Applying Mechanical Strains. ACS Nano 6, 54495456 (2012).

2. Scalise, E., Houssa, M., Pourtois, G., Afanasev, V. & Stesmans, A. Strain-induced semiconductor to metal transition in the two-dimensional honeycomb structure of MoS2. Nano Res. 5, 4348 (2012).

3. Shi, H., Pan, H., Zhang, Y.-W. & Yakobson, B. I. Quasiparticle band structures and optical properties of strained monolayer MoS2 and WS2. Phys. Rev. B 87, 155304 (2013).

4. Wang, L., Kutana, A. & Yakobson, B. I. Many-body and spin-orbit effects on direct-indirect band gap transition of strained monolayer MoS2 and WS2. Ann.

Phys. 526, L7L12 (2014).5. Rice, C. et al. Raman-scattering measurements and rst-principles calculations of strain-induced phonon shifts in monolayer MoS2. Phys. Rev. B 87, 081307 (2013).

6. Conley, H. J. et al. Bandgap Engineering of Strained Monolayer and Bilayer MoS2. Nano Lett. 13, 36263630 (2013).

7. He, K., Poole, C., Mak, K. F. & Shan, J. Experimental Demonstration of Continuous Electronic Structure Tuning via Strain in Atomically Thin MoS2.

Nano Lett. 13, 29312936 (2013).8. Feng, J., Qian, X., Huang, C.-W. & Li, J. Strain-engineered articial atom as a broad-spectrum solar energy funnel. Nat. Photon. 6, 866872 (2012).

9. Gomes, K. K., Mar, W., Ko, W., Guinea, F. & Manoharan, H. C. Designer Dirac fermions and topological phases in molecular graphene. Nature 483, 306310 (2012).

10. Polini, M., Guinea, F., Lewenstein, M., Manoharan, H. C. & Pellegrini, V. Articial honeycomb lattices for electrons, atoms and photons. Nat. Nanotechnol. 8, 625633 (2013).

11. Mak, K. F., Lee, C., Hone, J., Shan, J. & Heinz, T. F. Atomically thin MoS2: a new direct-gap semiconductor. Phys. Rev. Lett. 105, 136805 (2010).

12. Splendiani, A. et al. Emerging Photoluminescence in Monolayer MoS2. Nano Lett. 10, 12711275 (2010).

13. Bernardi, M., Palummo, M. & Grossman, J. C. Extraordinary sunlight absorption and one nanometer thick photovoltaics using two-dimensional monolayer materials. Nano Lett. 13, 36643670 (2013).

14. Sundaram, R. S. et al. Electroluminescence in Single Layer MoS2. Nano Lett. 13, 14161421 (2013).

15. Radisavljevic, B., Radenovic, A., Brivio, J., Giacometti, V. & Kis, A. Single-layer MoS2 transistors. Nat. Nanotechnol. 6, 147150 (2011).

16. Butler, S. Z. et al. Progress, challenges, and opportunities in two-dimensional materials beyond graphene. ACS Nano 7, 28982926 (2013).

17. Wang, Q. H., Kalantar-Zadeh, K., Kis, A., Coleman, J. N. & Strano, M. S. Electronics and optoelectronics of two-dimensional transition metal dichalcogenides. Nat. Nanotechnol. 7, 699712 (2012).

18. Mak, K. F., McGill, K. L., Park, J. & McEuen, P. L. The valley Hall effect in MoS2 transistors. Science 344, 14891492 (2014).

19. Bertolazzi, S., Brivio, J. & Kis, A. Stretching and breaking of ultrathin MoS2.

ACS Nano 5, 97039709 (2011).20. Castellanos-Gomez, A. et al. Elastic properties of freely suspended MoS2 nanosheets. Adv. Mater. 24, 772775 (2012).

21. Kuykendall, T., Ulrich, P., Aloni, S. & Yang, P. Complete composition tunability of InGaN nanowires using a combinatorial approach. Nat. Mater. 6, 951956 (2007).

22. Kim, C.-J. et al. On-nanowire band-graded Si:Ge photodetectors. Adv. Mater. 23, 10251029 (2011).

23. Castellanos-Gomez, A. et al. Local strain engineering in atomically thin MoS2.

Nano Lett. 13, 53615366 (2013).

24. Hulteen, J. C. & Van Duyne, R. P. Nanosphere lithography: a materials general fabrication process for periodic particle array surfaces. J. Vac. Sci. Technol. A 13, 15531558 (1995).

25. Tomori, H. et al. Introducing nonuniform strain to graphene using dielectric nanopillars. Appl. Phys. Express 4, 075102 (2011).

26. Reserbat-Plantey, A. et al. Strain superlattices and macroscale suspension of graphene induced by corrugated substrates. Nano Lett. 14, 50445051 (2014).

27. Lee, C. et al. Anomalous Lattice Vibrations of Single- and Few-Layer MoS2.

ACS Nano 4, 26952700 (2010).28. Li, H. et al. From Bulk to Monolayer MoS2: Evolution of Raman Scattering. Adv. Func. Mater. 22, 13851390 (2012).

29. Fu, X. et al. Tailoring exciton dynamics by elastic strain-gradient in semiconductors. Adv. Mater. 26, 25722579 (2014).

30. Schuller, J. A. et al. Orientation of luminescent excitons in layered nanomaterials. Nat. Nanotechnol. 8, 271276 (2013).

31. Liu, X. et al. Strong lightmatter coupling in two-dimensional atomic crystals. Nat. Photon. 9, 3034 (2015).

32. Lien, D.-H. et al. Engineering light outcoupling in 2D materials. Nano Lett. 15, 13561361 (2015).

33. Nam, D. et al. Strain-induced pseudoheterostructure nanowires conning carriers at room temperature with nanoscale-tunable band proles. Nano Lett. 13, 31183123 (2013).

34. Fu, X. et al. Exciton drift in semiconductors under uniform strain gradients: application to bent ZnO microwires. ACS Nano 8, 34123420 (2014).

35. Walia, S. et al. Characterization of metal contacts for two-dimensional MoS2 nanoakes. Appl. Phys. Lett. 103, 232105 (2013).

36. Ye, Z. et al. Probing excitonic dark states in single-layer tungsten disulphide. Nature 513, 214248 (2014).

Acknowledgements

This work was nancially supported by the 2013 Global Research Outreach Program (Award #: IC2012-1318) of the Samsung Advanced Institute of Technology (SAIT) and Samsung R&D Center America, Silicon Valley (SRA-SV) under the supervision ofDr. Debasis Bera and Anthony Radspieler, Jr. STM/STS experiments (A.W.C. and H.C.M) were supported by the US National Science Foundation (Grant DMR-1206916). J.L. acknowledges support by NSF CBET-1240696 and DMR-1120901. Computational time on the Extreme Science and Engineering Discovery Environment under the grant number TG-DMR130038 is gratefully acknowledged.

Author contributions

H.L. and X.L.Z. designed the experiments. H.L. performed device fabrication and optical measurements. A.W.C. and H.C.M conducted the STM/STS measurements. X.F.Q, S.M.A. and J.L. carried out the calculation and modelling. Y.J.G, X.L.W and P.M.A. carried out the CVD MoS2 growth. H.L., X.L.Z., H.C.M. and A.W.C. wrote the

manuscript, and all authors discussed the results and commented on the manuscript.

Additional information

Supplementary Information accompanies this paper at http://www.nature.com/naturecommunications

Web End =http://www.nature.com/ http://www.nature.com/naturecommunications

Web End =naturecommunications

Competing nancial interests: The authors declare no competing nancial interests.

Reprints and permission information is available online at http://npg.nature.com/reprintsandpermissions/

Web End =http://npg.nature.com/ http://npg.nature.com/reprintsandpermissions/

Web End =reprintsandpermissions/

How to cite this article: Hong, L. et al. Optoelectronic crystal of articial atoms in strain-textured molybdenum disulphide. Nat. Commun. 6:7381 doi: 10.1038/ncomms8381 (2015).

This work is licensed under a Creative Commons Attribution 4.0 International License. The images or other third party material in this article are included in the articles Creative Commons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/

Web End =http://creativecommons.org/licenses/by/4.0/

6 NATURE COMMUNICATIONS | 6:7381 | DOI: 10.1038/ncomms8381 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

& 2015 Macmillan Publishers Limited. All rights reserved.