This content has been downloaded from IOPscience. Please scroll down to see the full text.

Download details:

IP Address: 148.88.189.226

This content was downloaded on 21/02/2015 at 16:07

Please note that terms and conditions apply.

Structural, optical and electrostatic properties of single and few-layers MoS2: effect of

substrate

View the table of contents for this issue, or go to the journal homepage for more

2015 2D Mater. 2 015005

(http://iopscience.iop.org/2053-1583/2/1/015005)

Home Search Collections Journals About Contact us My IOPscience

2DMater. 2 (2015) 015005 doi:10.1088/2053-1583/2/1/015005

PAPER

Structural, optical and electrostatic properties of single and few-layers MoS2: effect of substrate

Benjamin J Robinson1, Cristina EGiusca2, YuremaTeijeiroGonzalez2, NicholasDKay1, OlgaKazakova2 andOlegVKolosov1

1 Department of Physics, Lancaster University, Lancaster, LA1 4YB,UK2 National Physical Laboratory,HamptonRoad, Teddington, TW11 0LW,UK

E-mail: olga.kazakova@npl.co.uk and o.kolosov@lancaster.ac.uk

Keywords: two dimensionalmaterials,MoS2, nanomechanics, substrate interaction, Kelvin probe, ultrasonic forcemicroscopy, Raman

AbstractWehave decoupled the intrinsic electrostatic effects arising inmonolayer and few-layerMoS2 fromthose influenced by the flake-substrate interaction. Using ultrasonic forcemicroscopy nanomechani-calmapping, we identify the change from supported to suspended flake regions on a trenched sub-strate. These regions are correlatedwith the surface potential asmeasured by scanning Kelvin probemicroscopy. Relative to the supported region, we observe an increase in surface potential contrast dueto suppressed charge transfer for the suspendedmonolayer. Using Raman spectroscopywe observe ared shift of the E12gmode formonolayerMoS2 deposited on Si, consistent with amore strainedMoS2on the Si substrate compared to the Au substrate.

Introduction

Layered transition metal dichalcogenides haveattracted significant attention due to their potentialapplications in electronic and optical devices [1].Molybdenum disulphide (MoS2) is one of the moststable layered materials of this class. In the bulk formthis semiconductor material has an indirect band gapof ∼1.29 eV and is used in a broad range of diverseapplications, e.g. as a photocatalyst and dry lubricant,as well as for photovoltaic power generation [2] andphoto-electrochemical hydrogen and Li ion batteriesproduction [3]. Monolayer MoS2 has a ∼1.75 eVdirect band gap and prominent electro- and photo-luminescent properties, making it a likely candidatefor applications in photodetectors and light-emittingdevices operating in the visible range [4]. Additionally,monolayer MoS2-based field-effect transistorsdemonstrated very promising electronic characteris-tics, such as a large current on/off ratio and sub-threshold swing [5].

With a rapidly increasing interest in the develop-ment of ultrathin MoS2-based devices, measurementmethods allowing for multifunctional characteriza-tion of physical properties, easy identification of MoS2layer number and interaction of the flakes with a sub-strate are in high demand. As electronic and optical

properties of MoS2 are strongly thickness and layer–substrate interaction dependent [6], it is essential toprecisely ascribe the measured parameters to indivi-dual layers. Here, we have used Raman spectroscopy,scanning Kelvin probe microscopy (SKPM) andatomic and ultrasonic force microscopy (AFM/UFM)for the mapping of mechanically exfoliated MoS2flakes with domains of the different thickness with theaim to precisely correlate their optical, nanomechani-cal and electrostatic properties on the nanoscale as wellas to explore the effect of interaction of MoS2 flakeswith a substrate.

Raman spectroscopy has been widely used todetermine the number of graphene [7] and MoS2 lay-ers, strain in both materials [8, 9] and, in a bulk form,identification of various crystalline forms of the mate-rial [10, 11]. For off-resonant laser lines (i.e. 532 nm),there are four well-defined first-order Raman activemodes, which can be typically observed in bulk MoS2.Themost prominent of them are the E12g mode (oppo-site in-plane vibration of two S atoms with respect tothe Mo atom) and A1g mode (out-of-plane vibrationof S atoms in opposite directions) [12]. It has beenshown that for single- and few-layers MoS2 both ofthese modes are thickness-dependent and shift awayfrom each other in frequency with increasing thick-ness [13], hence providing a convenient and reliable

OPEN ACCESS

RECEIVED

22 September 2014

REVISED

13 January 2015

ACCEPTED FOR PUBLICATION

28 January 2015

PUBLISHED

20 February 2015

Content from this workmay be used under theterms of theCreativeCommonsAttribution 3.0licence.

Any further distribution ofthis workmustmaintainattribution to theauthor(s) and the title ofthework, journal citationandDOI.

© 2015 IOPPublishing Ltd

means for determining layer thickness with atomic-level precision [14–17].

Similarly, the work function or surface potential oflayered materials is also strongly dependent on thenumber of layers [18]. While the Raman studies ofMoS2 are relatively common, direct measurements ofthe screening length are comparatively rare and therole of the electrostatic coupling in layered materials islargely neglected, with an exception of graphene [19–21]. In previous works, thickness-dependent inter-layer screening effects in MoS2 were measured bymeans of electrostatic force microscopy [22] andSKPM [23] techniques. For example, it has beendemonstrated that the surface potential of pristineMoS2 flakes decreased with increasing thickness (intip-biased studies) and increasing with thickness (insample-biased ones). Substrate related effects (i.e.trapped charges) and contamination caused by mate-rial processing can significantly affect the work func-tion value [24, 25]. The estimated characteristicscreening effects vary considerably in different studies,being within the length of ∼5 nm (8 layers) [23] or∼30–50 nm [22]. In both cases the obtained screeninglength is significantly larger than in graphene, where astrong-coupling regime is achieved already at∼1–2 nm, albeit strongly depending on the initialcharge density [18, 19]. However, in a recent experi-ment the screening length of ∼2.96 nm was reported[24], making it similar to the one in graphene despitetheir significant differences in the conductivity and theelectronic structure. Thus, the observed electrostaticproperties of layered MoS2 were attributed to a weak-coupling regime leading to the reduced screeningproperties.

UFM has been shown to be highly effective atdetermining the nanomechanical properties of MoS2,arising from both intrinsic structure and defects aswell as from the sample–substrate interface [4]. Theinterlayer coupling in MoS2 is largely dependent onlayer stacking where properties such as folds candecrease the coupling strength by up to a factor of 5[26]. This decrease in coupling strength between thelayers will manifest itself as a decrease in the mechan-ical strength of the flake. Themechanical properties ofMoS2 can also be greatly affected by the substrateproperties and morphology. It has been shown thatMoS2 deposited through mechanical exfoliation ontoa layer of thermally oxidized SiO2 does not follow itsnanoscale rough surface but instead adheres to thehigh points on the substrate [4]. This surface rough-ness affects the coupling of the flake to the surface andas such may cause a local variance in the flake’smechanical and electrical properties [4, 25, 27].Experimentally, variations in the Young’s modulusand pre-tension in MoS2 flakes have been observedand attributed to changing defect densities and adhe-sion to the substrate [27]. Using nanomechanicalmapping by UFM [28, 29] one can observe the localmechanical properties, stresses [30] and adhesion of

the multi-layer solid state structures [9, 31, 32] and,therefore probe the level of substrate-flake interactionand layer–layer interaction. Additionally, by studyingsuspended MoS2 films it is possible to eliminate thesubstrate flake interaction altogether and observepurely the effect of interlayer coupling of the flake onitsmechanical properties.

In this work we study the local optical, nano-mechanical and electrostatic properties of single andfew-layers MoS2 as measured by a combination offunctional scanning probemicroscopy techniques andRaman spectroscopymapping.

Materials andmethods

Sample preparationThe MoS2 layers were produced by mechanicalexfoliation from bulk crystals using the well-estab-lished ‘scotch tape’ method [33], with the finalexfoliation step performed using cross-linked adhesivepolymer film (Gel-pak® 4x adhesion) that does notleave surface residues. Single-, double- and few-layerMoS2 flakes were transferred onto uncoated and Au-coated (5 nm Cr/40 nm Au) Si/SiO2 substrates with300 nm thick thermal silicon oxide on doped Sisubstrate. The substrates had narrow (150–200 nmwide) trenches through all depth of SiO2 producedusing optical edge lithography described elsewhere[34]. Prior to use, substrates were cleaned in piranhasolution (3:1 concentrated H2SO4 to 30% H2O2) andimmediately before transfer by 98% Ar/2% oxygenplasma (PlasmaPrep2, Gala Instruments) for threeminutes to remove any remaining organic materialand facilitate attachment of theMoS2 to the surface.

ScanningKelvin probemicroscopy (SKPM)The surface potential (VCPD)measurements have beenperformed by SKPM, which also provided informa-tion on sample morphology, as well as a quantitativedetermination of the local thickness of MoS2. Thethickness of the flake was defined using the histogrammethod, allowing the highest accuracy of measure-ments of thin layers [35]. SKPM measurements wereconducted in ambient conditions, on a Bruker IconAFM, using Bruker highly doped Si probes (PFQNE-AL) with a force constant ∼0.9 Nm−1. Frequency-modulated SKPM (FM-SKPM) technique operated ina single pass mode has been used in all measurements.FM-SKPM operates by detecting the force gradient(dF/dz), which results in changes to the resonancefrequency of the cantilever. In this technique, an acvoltage with a lower frequency (fmod = 3 kHz) thanthat of the resonant frequency (f0 = 300 kHz) of thecantilever is applied to the probe, inducing a frequencyshift. The feedback loop of FM-KPFM monitors theside bands, f0 ± fmod, of cantilever vibration andminimizes themby applying an offset dc voltage whichis recorded to obtain a surface potential map [36].

2

2DMater. 2 (2015) 015005 B J Robinson et al

Sample-biased setup has been used in the pre-sent work.

Raman spectroscopyRaman intensity maps were obtained using a HoribaJobin-Yvon HR800 System. A 532 nm wavelengthlaser (2.33 eV excitation energy) was focused onto thesample through a 100× objective with Raman mapsdata taken with a spectral resolution of(3.1 ± 0.4) cm−1 and lateral spatial resolution of(0.4 ± 0.1) μm. The lateral spatial resolution wasdetermined using Si calibration gratings with a periodof 3 μm from the first derivative of Si 520 cm−1 bandscan lines extracted from the Raman map. The widthof the first derivative peak obtained by fitting with aGaussian is the lateral resolution.

Laser power was kept to aminimum to avoid heat-induced damage to the sample [37, 38], which can alsolead to broadening and shifting of the Raman peaks.Raman intensitymaps for E12g and A1gmodes were col-lected simultaneously by taking individual spectra onpreselected areas across each flake. The Raman spectrawere processed afterwards using Matlab software to fitthe E12g and A1g peaks using single Lorentzians and theresulting frequency shifts plotted to yield frequencyshiftmaps.

Ultrasonic forcemicroscopy (UFM)UFM has previously shown to be highly sensitive toboth the surface, subsurface and layer-interface struc-ture of MoS2 [4], graphene [32] and other 2Dmaterials [39]. UFM is a modification of standardcontact mode AFM where the sample is oscillated atlow amplitude (5–10 Å) and high frequency(4–5MHz in ourmeasurements). At these frequenciesthe cantilever (Contact-G, Budget Sensors) probesbecomes dynamically extremely rigid resulting inperiodic indentation of and separation from thesample. Resulting detection of the ultrasonic vibrationvia non-linear interaction forces between the probeand sample during an oscillation cycle is then stronglydependent upon the local mechanical structure of thesample [40] allowing effective non-destructive map-ping of the nanomechanical properties with a lateralresolution of 2–3 nm.

Results and discussion

Functional scanning probemicroscopyTopography image of the exfoliated and transferredMoS2 flake on a gold substrate, containing regions ofdifferent thicknesses, is shown in figure 1(a). Thick-ness of individual regions has been defined usingtapping mode AFM, where the first layer on thesubstrate has a thickness of ∼1 nm. Consequent layersthicknesses have been estimated from a topographicline profile (figure 1(d)) considering an interlayerseparation of 0.7 nm, yielding, correspondingly, 1, 5, 8

layers thickness and bulk (∼40 layers). The verticaltrench in the substrate is highlighted by an arrow.

Surface potential mapping was obtained simulta-neously with topography imaging and is presented infigure 1(b). For the entire MoS2 flake, VCPD value issignificantly lower than that of the gold substrate,whereVCPD value of a single layer is notably the lowest.The VCPD value increases with the layer thickness,consistent with literature [24], though the bulk value(as well as a part of the flake) is compromised by dec-oration of the surface by environmental adsorbates(also seen as bright clusters in figure 1(a)), leading to adecrease in surface potential of the MoS2 layers [24].With this exception, distribution of the surface poten-tial within each layer is relatively homogeneous withinthe 20 mV accuracy. Figure 1(e) demonstrates the dis-tribution of the surface potential across an area of dif-ferent thicknesses indicated by the frame in figure 1(b)and a histogram analysis of the acquired SKPM dataover the enclosed area is presented next. The averagevalues of themeasured surface potentials on the differ-ent thickness domains are obtained by peak fitting, asindicated by the green lines in figure 1(e), showing theindividual resulting components. The results are bestdescribed by the fitting of three Lorentzian peaks, cor-responding to the areas of 1, 5 and 8 MoS2 layers withtheir absolute VCPD values being −106, −62 and−33 mV, respectively. The assignment of individualdomains thicknesses to the deconvoluted peaks in thehistogram has been made in correlation with the levelof contrast in the associated SKPM image, with themost intense peak corresponding to the brightest con-trast in the surface potential image.

Using the absolute values for the contact potentialdifference and a work function of 4.5 eV for the scan-ning tip [41], the work function can be estimatedaccording to Φsample =Φtip +VCPD, resulting in4.39 eV, 4.44 eV and 4.47 eV for 1, 5 and 8 layers,respectively, consistent with the values reported byOchedowski et al [24]. However, a significantly largervalue (up to 5.25 eV) for bulk MoS2 was used by oth-ers, see e.g.[23]. It should also be noted that althoughan opposite thickness-dependent trend of the surfacepotential was experimentally observed in [22, 23], thisdiscrepancy arises from the use of different type ofbiasing (tip or sample biasing) in the SKPMsetup.

UFM-derived nanomechanical mapping(figure 1(c)) shows stiffness variations arising fromboth the MoS2 flake thickness, the local sample–sub-strate interface and the interlayer coupling; these sour-ces may be interlinked, for example significantly morevariation are observed in the monolayer region com-pared to the thicker areas of material (figure 1(c)),where brighter contrast corresponds to moremechanically stiff areas. UFM has previously beenused for quantitative determination of 2D materialproperties [32, 42], such as elastic modulus; however,here the key advantage of UFM is the identification ofthe trench which can be clearly seen as the black line

3

2DMater. 2 (2015) 015005 B J Robinson et al

(zero signal) running across the flake under its surface;zero UFM signal indicates complete local decouplingof the sample from the substrate and allows us to pre-cisely mark the transition from supported to sus-pended MoS2. The position of the trench remainsclearly distinguishable even under the 8 LMoS2 wheretopographical effects are minimal, while regions ofdifferent thickness are easily identified and generallyuniformUFM response within each region suggests anexcellent MoS2–substrate contact. Comparison ofUFM and SKPM responses for the different MoS2layer thicknesses (1 L, 5 L and 8 L areas) over the sup-ported/suspended transition region (figures 2(a)–(c)),shows a clear peak (enhanced surface potential con-trast) in the surface potential (blue dashed lines) overthe UFM identified trench (black solid line) for the 1 Lregions, smaller changes are observed for the thicker5 L and 8 L regions.

Surface potential data arising from comparison ofthe supported and suspended regions are shown intable 1. It is noteworthy that the area of unsupportedsingle layerMoS2 above the channel is characterized byan enhanced surface potential contrast, i.e. ∼100 mVincrease as compared to the main part of a single flake.This is most likely due to suppressed charge transferfor the suspended monolayer compared to the

Topography SKPM – Surface Potential UFM - Nanomechanics 60

40

20

-20

0

200

-200

-400

0

3

2

1

0

6

5

4

4

3

3

2

2

1

1

0

0

-1

Position (μm)

Hei

ght (

nm)

(d) (e) Surface Potential (v)

-0.20 -0.15 -0.10 -0.05 0.00 0.05

Dis

tribu

tion

15

10

5

0

1 layer:-106 mV

-62 mV

-33 mV

5 layers:

8 layers:N=8

N=5

N=1

N=0

Figure 1. (a) Topography (z range: 96.82 nm), (b) SKPMsurface potential (potential range: 785.1 mV) and (c)UFMnanomechanicalimages (non-linear response range 3.0 nm2) of theMoS2flake onAu substrate, the number of layers is indicated in (a). (d)Topography profile along the line indicated in (a). (e)Distribution of surface potential across regions of different thickness ofMoS2 asindicated by the frame in (b). The green lines are result of thefitting for individual domains, the red line shows the overall fit.

1.5

1.5-150

-100

1.0

1.0

0.5

0.50.0

1.5

1.0

0.5

0.00.0 1.51.00.50.0

sreyaL 8 sreyaL 5 reyaL 11.5

1.0

0.5

0.00.0 0.5 1.0 1.5

-150

-100

-50

0

Position (μm) Position (μm) Position (μm)

Non

linea

r Res

pons

e (n

m2 )

Non

linea

r Res

pons

e (n

m2 )

Non

linea

r Res

pons

e (n

m2 )

Sur

face

Pot

entia

l (m

V)

Sur

face

Pot

entia

l (m

V)

Sur

face

Pot

entia

l (m

V)

Surface PotentialUFM Response

-50

0

-150

-100

-50

0

(a) (b) (c)

Figure 2.Comparison ofUFMnanomechanical (black solid lines) and SKPM surface potential (blue dashed lines) responses in the 1layer (a); 5 layer (b) and 8 layer (c) regions. The position of the trench is represented by the grey shaded area.

Table 1.Average enhanced surface potential contrastsbetween the supported and suspended regions oftheMoS2.

Increase in surface potential

contrast

MoS2 thickness Absolute change % change

1 layer 94 ± 7 mV (67± 5%)

5 layers 39 ± 17 mV (53± 23%)

8 layers 41 ± 4 mV (98± 10%)

4

2DMater. 2 (2015) 015005 B J Robinson et al

supported one [22, 24]. The effect of the trench on theabsolute value of the enhanced surface potential is sig-nificantly less pronounced in the case of thicker (5 Land 8 L) flakes.

The width of the trench as observed by UFM istypically ∼150 nm for both the 1 and 5 layer regionsand ∼190 for the 8 layer region, in good correlationwith the SKPM surface potential peaks. Apparentbroadening of the trench in thicker regions is wellknown due to increased mechanical rigidity of thethicker flake.

Small dark points observed by UFM (figure 1(c))

on the monolayer region are assigned to surface con-

tamination of the flake and not delaminated MoS2regions due to their non-zero non-linear response.

These points give rise to the variation in the observed

non-linear response (figures 2(a)–(c)) of the sup-

ported flake and corresponding surface potential,

which is in turn a significant component of the errors

calculated in table 1. The contamination is believed to

arise from absorbed water and other airborne mole-

cules, as MoS2 is mildly hydrophilic and sensitive to

atmospheric water vapour. Annealing of the flake led

only to partial disappearance and redistribution of this

contamination. Due to MoS2 polarity and hydro-

philicity, the effect of atmospheric species can be

rather strong, e.g. through the formation of hydrogen

bond with water molecules. Such adsorbates act as

charge trappers, affecting the surface potential and

charge distribution at ambient atmosphere. A cleaning

procedure using contact AFM and soft cantilevers,

similar to the one described in Goosens et al [43], has

been employed, however no significant improvement

in theflake appearancewas noticed after a few cleaningcycles.

Ramanmapping: thickness dependenceRaman maps of E12g and A1g modes were obtainedusing 532 nm laser line for a flake containing domainsof 1 L, 2 L, 3 L and 7 L thickness on Si (figure 3).Corresponding maps of the frequency shift andintensity are shown in figures 3(a) and (b) and 2(c)and (d), respectively. Themaps demonstrate that boththe frequency shift and intensity maps are thickness-dependent, allowing one to use such maps for a clearidentification of the layer number.

Spatial maps of the Raman frequency shift for theE12g and A1g modes (figures 3(a) and (b)) demonstratethat the E12g vibration softens (red shift), while the A1g

vibration stiffens (blue shift) with increasing samplethickness. For example, in figure 3(a) the E12gmode haslowest contrast for 1 L thickness and highest for thethicker material, indicating the red shift for the latter.Conversely, the opposite trend is observed for the A1g

modemap (figure 3(b)), where the 1 L domain has thebrightest and the bulk area—the darkest contrast,indicative of the blue shift for the thickermaterial.

In order to further quantify Raman maps, weextract individual Raman spectra obtained on flakes ofdifferent thickness. Typical Raman spectra high-lighting the region of E12g and A1g modes are shown infigure 3(e). The dependence of peak position on thenumber of layers is plotted in figure 3(f). The Ramanshift between E12g and A1g modes, which becomes gen-erally larger with layer number, is in good agreementwith literature [13–17]. Intensity maps of E12g and A1g

modes are shown in figures 3(c) and (d), respectively.

3.0

2.5

Nor

mal

ised

inte

nsity

(a.u

)

2.0

1.5

1.0

0.5

0.0370 380 390 400 410 420

Raman shift (cm-1)R

aman

shi

ft (c

m-1

)

410

408

406

404

384

382

0 1 2 3 4 5 6 7 8 bulkNumber of Layer (N)

0 1 2 3 4 5 6 7 8 bulkNumber of Layer (N)

5.5

5.0

4.5

4.0

3.5

3.0

2.5MoS2/Si MoS2/Au

MoS2/Si MoS2/Au

FWH

M (c

m-1

)

E1

E1

2g

2g

E12g

FWHM

FWHMA1g

A1g

A1g

Peak PositionPeak Position

7L

3L

2L

1L

(e) (f) (g)

spam ytisnetni namaR spam tfihs ycneuqerf namaR

Figure 3.Relative Raman frequency shift (a),(b) and intensity (c),(d)maps of E12g (a),(c) andA1g (b),(d) peaks for theMoS2/Siflakewith 1, 2, 3 and 7 layers domains. (e) Representative Raman spectra for different thicknesses of theMoS2flake. (f) Thickness-dependent Raman shifts of E12g andA1gmodes forMoS2flakes onAu and Si substrates. (g) Linewidths of E

12g andA1gmodes forMoS2

onAu and Si substrates as a function of the number of layers.

5

2DMater. 2 (2015) 015005 B J Robinson et al

For both types of vibrations, the intensity increaseswith thickness between 1 L through to 7 L. The widthsof the peaks are generally independent on the layerthickness for small number of layers, with peaksbecoming narrower for bulk MoS2, also in agreementwith previous results [14].

Further, we study the effect of the substrate onvibrational modes of MoS2. Comparison of the influ-ence of a metallic substrate (Au), relative to the semi-conducting one, on the vibrational properties of theMoS2 is summarized in figures 3(f) and (g), with cor-responding relative Raman intensity and frequencyshift maps presented in figure 4. The differencebetween the E12g and A1g peak positions for 1 L is com-parable for both substrates: ΔωAu = 19.7 cm−1 andΔωSi = 19.3 cm−1. However, a noticeable red shift isobserved for both E12g and A1g modes when MoS2 isdeposited on Si in comparison with the Au substrate.This observation is consistent with a more strainedMoS2 on the Si substrate compared to the Au one, as itwas shown that tensile strain is associated with a redshift of the Raman E12gmode formonolayerMoS2 [24].At the same time, for single layer MoS2 on Au, the A1g

peak becomes sharper and E12g broader than for singlelayer MoS2 on Si. It is in line with the UFM examina-tion in figure 1 where the absence of delaminationssuggests that this flake is either at neutral or tensilestrain [9]. It would be interesting to examine how thesurface potential is affected by strain of the flake, how-ever due to the semi-insulating nature of the Si sub-strate giving rise to charging of the flake duringscanning, SKPM data has only been acquired on theAu substrate, so a direct comparison is not possiblebased on the current data.

Corresponding images for Raman intensity(figures 4(a) and (b) and frequency shift (figures 4(c)and (d) maps of the MoS2 flake supported on the Ausubstrate (the same flake as shown in figure 1) demon-strate the same thickness-dependent tendencies as dis-cussed above for MoS2 on Si. However, the trend is notobserved for the E12g mode in the bulk domain(figure 4(c)), which might be explained by contamina-tion of the representative area. A ca. 20% increase inintensity is observed for suspended regions with respectto the supported ones for both E12g, A1gmodes. It shouldbe pointed out that the channel in the substrate of150–190 nm width is narrower than the lateral spatialresolution of the Raman system, and so, the resultantspectra show a mixed contribution from the materialsuspended over the channel and adjacent regions sup-ported on the Au substrate not allowing for the spectro-scopic assessment of the peak shift position.At the sametime, assuming axially symmetrical Lorentzian shapedprobe of 400 nm width (see Materials and Methods),the trench area would contribute from 21 to 33%of thetotal signal indicating, togetherwith observed local 20%increase in total signal, that the per area Raman inten-sity from the suspended region is 50–100%higher com-pared to the supported region. Raman intensities areknown to be sensitive to the orientation of single crys-tals in relation to the scattering geometry [44]. Sinceourmeasurements are averaged over an area larger thanthe width of the channel, a misalignment of the crystalplane for the suspended MoS2 in respect to the area ofthe flake supported by the Au substrate is unlikely to besolely accounted for the increase in intensity weobserve. Another possible reason could be due to inter-ference effects caused bymultiple reflections of the laserbeam at the edges of the channel, as it is well recognized

Figure 4.Raman intensity (a),(b) and frequency shift (c),(d)maps of E12g (a),(c) andA1g (b),(d) peaks for theMoS2/Au flakewith 1, 5,8 layers and bulk domains.

6

2DMater. 2 (2015) 015005 B J Robinson et al

that optical interference effects have a strong impact onthe intensity of Raman spectra, as shown e.g. for gra-phene [45]. Contrary to our observations, for mono-layer MoS2, calculations of the enhancement factor ofRaman peak intensity due to optical interference effectsshow a stronger Raman response for supported config-urations than for suspended ones [46]. This could pointto a different mechanism giving rise to the increase inintensity we observe, possibly based on emission andabsorption effects at the edges of the channel and themetallic nature of the substrate as compared to thedielectric ones usually employed for graphene andother 2D materials. Furthermore, we also observe anincreased Raman signal at the channel edge in the baresubstrate not associated with the flake, which could bean indication of SERS-like enhancement, as manynanoscale-textured metallic surfaces have been foundcapable of producing strong electromagnetic fields thatgive rise to SERS enhancements. The enhancementcould be caused by the polarization of the incident laserlight possibly resonating with the narrowAu channel togive an enhancement of the absorption and scatteringprocesses cross sections [47, 48].

Raman maps also demonstrate that the signal forboth vibrational modes, while being generally homo-geneous within each individual supported flake, showsa change of the contrast at the flake border. This ismanifested as an increase of intensity of both E12g andA1g modes. The effect is evident for thicker flakes,however is less pronounced for 1 L MoS2. For Ramanshifts the behaviour is more complex, i.e. the E12gmodeis characterized by a darker contrast (i.e. experiences ablue shift), whereas the A1g mode demonstrates abrighter one (i.e. experiences a red shift) at the edges ofindividual flakes. This behaviour is opposite to thethickness-dependent trends as observed above andlikely to reflect the defective nature and possible inho-mogeneity of the chemical composition of the flakeboundaries due to adsorption of adatoms and creationof vacancy defects [49, 50].

Conclusions

In conclusion, we have performed a comprehensivestudy of the optical and electrostatic properties ofMoS2 in dependence on the layer thickness andsample–substrate interaction. Using subsurface sensi-tive nanomechanics sensitive UFM mapping we haveidentified the change in the MoS2 flake indicating thetransition between supported and suspended flakeregions of different thicknesses (1, 5 and 8 layers) andcorrelated these regions with SKPM derived surfacepotential and Raman mapping. We observe anincrease in the surface potential contrast for suspendedregions of all thicknesses relative to the supportedareas, with the monolayer region demonstrating a∼100 mV (∼67%) increase, which is believed to bedue to suppressed charge transfer for the suspended

monolayer compared to the supported one. Further-more, a corresponding increase in Raman intensity forthe E12g and A1g modes is observed for the monolayerregion but not for thicker regions of the flake.Additionally, we demonstrate a noticeable red shift forboth E12g and A1g modes whenMoS2 is deposited on Siin comparison with the Au substrate. This observationis consistent with a more strained state of the MoS2flake on the Si substrate.

These results provide a detailed understanding ofthe layer properties, which are essential for potentialoptoelectronic applications by decoupling the electro-static properties of MoS2 from substrate-inducedeffects.

Acknowledgments

Authors acknowledge support of EC grants GrapheneFlagship (No. CNECT-ICT-604391), FUNPROB,QUANTIHEAT and EMRP under project Gra-phOhm, EPSRC grants EP/G015570 and EP/K023373/1, NMS under the IRD Graphene Projectand NowNANO Doctoral Training Centre. Authorsthank Dr Spyros Yannopoulos and Dr Alina Zoladek-Lemanczyk for useful discussions on Raman spectro-scopy and DrMark Rosamond and Dr Dagou Zeze forsupplying the trenched substrates.

References

[1] GeimAKandGrigorieva I V 2013Van derwaals hetero-structuresNature 499 419–25

[2] FontanaM,Deppe T, BoydAK, RinzanM, LiuAY,ParanjapeMandBarbara P 2013 Electron–hole transport andphotovoltaic effect in gatedMoS2 schottky junctions Sci. Rep.3 1634

[3] Stephenson T, Li Z,Olsen B andMitlinD 2014 Lithium ionbattery applications ofmolybdenumdisulfide (MoS2) nano-composites Energy Environ. Sci. 7 209–31

[4] SercombeD, Schwarz S, Del Pozo-ZamudioO, Liu F,Robinson B J, Chekhovich EA, Tartakovskii I I, KolosovO andTartakovskii A I 2013Optical investigation of the naturalelectron doping in thinMoS2films deposited on dielectricsubstrates Sci. Rep. 3 3489

[5] Radisavljevic B, Radenovic A, Brivio J, Giacometti V andKis A2011 Single-layerMoS2 transistorsNat. Nanotechnology 6147–50

[6] SercombeD, Schwarz S, Del Pozo-ZamudioO, Liu F,Robinson B J, Chekhovich EA, Tartakovskii I I, KolosovO andTartakovskii A I 2013Optical investigation of the naturalelectron doping in thinMoS2films deposited on dielectricsubstrates Sci. Rep. 3 3489

[7] Ferrari AC et al 2006Raman spectrumof graphene andgraphene layers Phys. Rev. Lett. 97 187401

[8] Ni ZH,WangYY, YuT and ShenZX 2008Ramanspectroscopy and imaging of grapheneNanoRes. 1 273–91

[9] Trabelsi A BG, Kusmartsev FV, Robinson B J, Ouerghi A,KusmartsevaOE, KolosovOV,Mazzocco R,Marat BG andOueslatiM2014Charged nano-domes and bubbles inepitaxial grapheneNanotechnology 25 165704

[10] Verble J L andWieting T J 1970 Latticemode degeneracyinMoS2 and other layer compounds Phys. Rev. Lett. 25362–5

[11] Chen JMandWangC S 1974 Second order Raman spectrumofMoS2 Solid State Commun. 14 857–60

7

2DMater. 2 (2015) 015005 B J Robinson et al

[12] Bertrand PA 1991 Surface-phonon dispersion ofMoS2Phys.Rev.B 44 5745–9

[13] Lee C, YanH, Brus L E,Heinz T F,Hone J andRyu S 2010Anomalous lattice vibrations of single- and few-layerMoS2ACSNano 4 2695–700

[14] LiH, ZhangQ, YapCCR, Tay BK, EdwinTHT,Olivier A andBaillargeat D 2012 Frombulk tomonolayerMoS2: evolution ofRaman scatteringAdv. Funct.Mater. 22 1385–90

[15] Zhao Y et al 2013 Interlayer breathing and shearmodes in few-trilayerMoS2 andWSe2Nano Lett. 13 1007–15

[16] LiH, Yin Z,HeQ, LiH,HuangX, LuG, FamDWH,TokA I Y, ZhangQ andZhangH2012 Fabrication of single-andmulti-layerMoS2film-basedfield-effect transistors forsensingNOat room temperature Small 8 63–7

[17] Rice C, YoungR J, ZanR, Bangert U,WolversonD,Georgiou T, Jalil R andNovoselov K S 2013Raman-scatteringmeasurements and first-principles calculations of strain-induced phonon shifts inmonolayerMoS2Phys. Rev.B 87081307

[18] Datta S S, StrachanDR,Mele E J and JohnsonATC2008Surface potentials and layer charge distributions in few-layergraphene filmsNano Lett. 9 7–11

[19] KurodaMA, Tersoff J andMartynaG J 2011Nonlinearscreening inmultilayer graphene systems Phys. Rev. Lett. 106116804

[20] Sanchez-Yamagishi J D, Taychatanapat T,WatanabeK,Taniguchi T, Yacoby A and Jarillo-Herrero P 2012Quantumhall effect, screening and layer-polarized insulating states intwisted bilayer graphene Phys. Rev. Lett. 108 076601

[21] Panchal V, GiuscaC, Lartsev A, Yakimova R andKazakovaO2014 Local electricfield screening in bi-layer graphene devicesFront. Phys. 2 3

[22] Castellanos-Gomez A,Cappelluti E, RoldánR, Agraït N,Guinea F andRubio-Bollinger G 2013 Electric-field screeningin atomically thin layers ofMoS2: the role of interlayercouplingAdv.Mater. 25 899–903

[23] Li Y, XuC-Y andZhen L 2013 Surface potential and interlayerscreening effects of few-layerMoS2 nanoflakesAppl. Phys. Lett.102 143110

[24] Ochedowski O,MarinovK, ScheuschnerN, PoloczekA,BussmannBK,Maultzsch J and SchlebergerM2014 Effect ofcontaminations and surface preparation on thework functionof single layerMoS2Beilstein J. Nanotechnol. 5 291–7

[25] KayND, Robinson B J, Fal’koV I,NovoselovK S andKolosovOV2014 Electromechanical sensing of substratecharge hidden under atomic 2D crystalsNano Lett. 14 3400

[26] Castellanos-Gomez A, van der ZantH J and Steele G 2014FoldedMoS2 layers with reduced interlayer couplingNanoRes. 7 1–7

[27] Castellanos-Gomez A, PootM, Steele GA, van der ZantH S J,Agrait N andRubio-Bollinger G 2012 Elastic properties offreely suspendedMoS2 nanosheetsAdv.Mater. 24 772–5

[28] Dinelli F, AssenderHE, KirovK andKolosovOV2000 Surfacemorphology and crystallinity of biaxially stretched PET filmson the nanoscale Polymer 41 4285–9

[29] KolosovO andYamanakaK1993Nonlinear detection ofultrasonic vibrations in an atomic-forcemicroscope Japan. J.Appl. Phys. 2 32 L1095–8

[30] KolosovOV,CastellMR,MarshCD, BriggsGAD,Kamins T I andWilliams R S 1998 Imaging the elasticnanostructure ofGe islands by ultrasonic forcemicroscopyPhys. Rev. Lett. 81 1046–9

[31] McGuigan AP,Huey BD, BriggsGAD,KolosovOV,Tsukahara Y andYanakaM2002Measurement of debonding

in cracked nanocomposite films by ultrasonic forcemicro-scopyAppl. Phys. Lett. 80 1180–2

[32] Robinson B J, Rabot C,Mazzocco R,DelamoreanuA,Zenasni A andKolosovOV2014Nanomechanicalmapping ofgraphene layers and interfaces in suspended graphene nanos-tructures grown via carbon diffusionThin Solid Films 550472–9

[33] GeimAK andKimP 2008Carbonwonderland Sci. Am. 29890–7

[34] RosamondMC,Gallant A J, PettyMC, KolosovO andZezeDA2011A versatile nanopatterning techniquebased on controlled undercutting and liftoffAdv.Mater. 235039–44

[35] Burnett T L, Yakimova R andKazakovaO2012 Identificationof epitaxial graphene domains and adsorbed species inambient conditions using quantified topographymeasure-ments J. Appl. Phys. 112 054308

[36] Panchal V, Pearce R, Yakimova R, TzalenchukA andKazakovaO2013 Standardization of surface potentialmea-surements of graphene domains Sci. Rep. 3 2597

[37] Sahoo S andArora AK 2010 Laser-power-inducedmulti-phonon resonant raman scattering in laser-heatedCdS nano-crystal J. Phys. Chem.B 114 4199–203

[38] Gupta R, XiongQ, AduCK, KimU J and Eklund PC 2003Laser-induced fano resonance scattering in silicon nanowiresNano Lett. 3 627–31

[39] Robinson B J, KayND andKolosovOV2013Nanoscaleinterfacial interactions of graphenewith polar and nonpolarliquids Langmuir 29 7735–42

[40] Robinson B J andKolosovOV2014 Probing nanoscalegraphene–liquid interfacial interactions via ultrasonic forcespectroscopyNanoscale 6 10806–16

[41] Panchal V, Pearce R, Yakimova R, TzalenchukA andKazakovaO2013 Standardization of surface potentialmea-surements of graphene domains Sci. Rep. 3 2597

[42] Robinson B J andKolosovOV2014 Probing nanoscalegraphene-liquid interfacial interactions via ultrasonic forcespectroscopyNanoscale 6 10806–16

[43] Goossens AM,CaladoVE, Barreiro A,WatanabeK,Taniguchi T andVandersypen LMK2012Mechanicalcleaning of grapheneAppl. Phys. Lett. 100 073110

[44] Turrell G 1972 Infrared AndRaman SpectraOf Crystals(London: Academic)

[45] Wang YY,Ni ZH, ShenZX,WangHMandWuYH2008Interference enhancement of Raman signal of grapheneAppl.Phys. Lett. 92 043121

[46] Li S-L,MiyazakiH, SongH,KuramochiH,Nakaharai S andTsukagoshi K 2012Quantitative Raman spectrum and reliablethickness identification for atomic layers on insulating sub-stratesACSNano 6 7381–8

[47] Sow I, Grand J, Lévi G, Aubard J, Félidj N, Tinguely J C,HohenauA andKrenn J R 2013Revisiting surface-enhancedRaman scattering on realistic lithographic gold nanostripesJ. Phys. Chem.C 117 25650–8

[48] NatelsonD, Li Y andHerzog J B 2013Nanogap structures:combining enhanced Raman spectroscopy and electronictransport Phys. Chem. Chem. Phys. 15 5262–75

[49] ZhouW,ZouX,Najmaei S, Liu Z, Shi Y, Kong J, Lou J,Ajayan PM, YakobsonB I and Idrobo J-C 2013 Intrinsicstructural defects inmonolayermolybdenumdisulfideNanoLett. 13 2615–22

[50] AtacaC, S ahinH, Akturk E andCiraci S 2011Mechanical andelectronic properties ofMoS2 nanoribbons and their defectsJ. Phys. Chem.C 115 3934–41

8

2DMater. 2 (2015) 015005 B J Robinson et al