Resonant Raman imaging of MoS2-substrate interaction
RResonant Raman imaging of MoS -substrate interaction Hongyuan Li
1, 2 and Dmitri V. Voronine
1, 31
Institute for Quantum Science and Engineering,Texas A & M University, College Station, TX, 77843 USA Department of Applied Physics, School of Science,Xi’an Jiaotong University, Shaanxi 710049, P. R. China Baylor University, Waco, TX, 76198 USA
Abstract
We report a study of long-range MoS -substrate interaction using resonant Raman imaging. We observeda strong thickness-dependent peak shift of a Raman-forbidden mode that can be used as a new method ofdetermining the thickness of multilayered MoS flakes. In addition, dependence of the Raman scatteringintensity on thickness is explained by the interference enhancement theory. Finally, the resonant Ramanspectra on different substrates are analysed. Keywords:
MoS , resonant Raman, AFM, thickness-dependence, forbidden mode a r X i v : . [ c ond - m a t . m t r l - s c i ] D ec . INTRODUCTION Two-dimensional (2D) transition metal dichalcogenide (TMD) materials have recently attractedwide attention due to their potential applications in optoelectronic devices. For example, mono-layer MoS has a large direct band gap [1], and can be used in field-effect-transistors[2]. As apowerful tool, Raman spectroscopy has been widely used for studying the various properies ofMoS . Frey et . at . studied the resonant Raman (RR) spectra of MoS nanoparticles [3]. Li andChakraborty have investigated the thickness-dependent effects for the Raman scattering of MoS [4, 5]. Also, the influence of the substrate-MoS interactions on the Raman spectra has been in-vestigated [6]. In this paper, we performed RR imaging of multilayered MoS flakes using 660 nmlaser excitation which corresponds to the A exciton of MoS [7, 8]. We obtained simultaneouslythe topographic information and correlated it with Raman imaging. The correlated AFM-Ramanimaging reveals the relation between the thickness and optical properties. We also analysed theinfluence of substrates with different dielectric properties on the RR spectra of MoS . II. RESULTS
The multilayered MoS flakes were exfoliated on SiO and gold substrate using the scotch-tapemethod. Th AFM image of a typical MoS flake is shown in Fig.1 ( a ). The thinnest part of theflake has the height of ∼ [4]. Note the colorbar inFig.1 ( a ) corresponds to the number of layers .A typical RR spectrum of the MoS flake is shown in Fig.1 ( b ). The main spectral features in-clude: strong out-of-plane A g mode near 410cm − , in-plane E g mode near 385cm − , IR-activeA u mode near 466cm − and several second-order modes, which include E g +LA at 528cm − ,2E g at 574cm − , E g +LA at 600cm − and A g +LA at 644cm − [3, 4].Resonant Raman imaging of the multilayered MoS flake on the gold substrate is shown inFig.2. The imaging step size was 100nm. The intensity maps for A g and A u modes are shownin Fig.2 ( a ) and ( b ). The intensity ratio I(A u )/I(A g ) is shown in Fig.2 ( c ). Both Fig.2 ( a ) and( b ) show that the area with the strongest intensity corresponds to the number of layers of about30, while the very thin (eg: the middle) and the very thick (eg: the left bottom) parts of the flakeshow a relatively low intensity. I(A g ) vs the number of layers is shown in Fig.4 ( b ) where themaximum intensity correspond to n=30. The intensity ratios in Fig.4 ( a ) and Fig.5 indicate that2or the thin part, the A u peak has a relatively stronger intensity compared to other modes. ( a ) Raman shift(cm -1 )300 350 400 450 500 550 600 650 700 I n t en s i t y ( a . u . ) × A E E +LA 2E E +LA A +LA2LA ( b )FIG. 1. AFM image ( a ) of MoS flake, with the colorbar representing the number of layers and a typicalresonant Raman spectrum ( b ) of a multilayered MoS flake on gold substrate excited by 660 nm laser. Peak positions also show shifts with the change of the number of layers . The peak positionmaps of the A g and A u modes are shown in Fig.2 ( d ) and ( e ). Combined with the height distribu-tion in Fig.1 ( a ), the RR spectra show a blue shift for A g and a red shift for A u for a multilayeredflake compared to bulk MoS . Fig.3 ( a ) and ( b ) show the A g and A u peak position versus thenumber of layers . The A g peak position increases slightly( ∼ − ) when the number of lay-ers increases from 7 to 140, while the A u peak has a blue shift of about ∼ − . The shift ofthe A u peak can seen more clearly in Fig.5 ( a ) and ( b ). III. DISCUSSION
We consider two main effects: (1) substrate dependence, (2) thickness-dependence, includingpeak shifts and intensity variations. It has been widely reported that, on common insulating sub-strates such as silicone and SiO , the increase of the number of layers has an influence on thetwo characteristic Raman peaks E g and A g [4, 9]. The A g mode has a blue shift due to the in-crease of the force constant which is induced by the increased interlayer van der Waals interaction.Our RR measurements of the A g peak position variation are consistent with the previous reports.However, here we show that, under the resonant condition, the Raman-inactive A u mode shows astrong thickness-dependent softening with the increase of the number of layers . This is attributed3 a ) ( b ) ( c )( d ) ( e ) ( f )FIG. 2. Resonant Raman imaging of multilayered MoS on gold substrate: intensity of A g ( a ) and A u ( b )modes, peak intensity ratio I(A u )/I(A g ) ( c ), peak portion maps for A g ( d ) and A u ( e ) modes, and peakposition difference A u -A g ( f ). to the decrease of the long-range Coulombic interaction between the Mo atoms with increasinglayer number [10]. The results on SiO substrate are consistent with previous reports. For thesmall number of layers (n < g and A u modeshow red shifts with the increase of the number of layers [4]. When the number of layers exceeds ∼
10, the change of the peak position is small and difficult to measure.However on the gold substrate, we observed a previously unreported blue shift for the A u mode. Fig.3 ( a ) shows that for a small number of layers ( ∼ u peak has a red shiftof about 2cm − compared to the bulk MoS . With the increase of the thickness, this red shiftgradually disappears with a turning point near n =
30, as shown in Fig.3 ( a ). This indicates that4
20 40 60 80 100 120 140464.5465465.5466466.5467 A P ea k po s i t i on ( c m − ) Number of layers P ea k po s i t i on ( c m − ) ( a ) −A Number of layers P ea k d i ff e r en c e ( c m − ) ( b ) P ea k po s i t i on ( c m − ) ( c )FIG. 3. Thickness-dependence of peak position. ( a ) Peak position of A u and A g modes. ( b ) Peak positiondifference A u -A g . ( c ) Peak position of 2LA. Blue and red dots represent the measurements on gold andSiO substrates, respectively. Data from several flaskes are shown as groups of points linked by dashedlines with all connected points belonging to the same flake the thickness-dependent effects can be observed within a large range of the number of layers . Thisis different from the SiO substrate where the peak shift can only be observed for a small numberof layers (n < u mode ofthe multilayered MoS flake on the gold substrate: (1) laser-induced heating of the gold substratemay have thermal effects on MoS leading to the peak shift [11]; (2) charge transfer from MoS
20 40 60 80 100 120 1400.20.250.30.350.40.450.50.550.60.650.7 Number of layers I n t en s i t y r a t i o I(A )/I(A ) ( a ) I n t en s i t y ( no r m a li s ed ) I(A ) ( b )FIG. 4. ( a ).The intensity ratio I(A u )/I(A g ) decreases rapidly with the increase of the thickness and stablizesafter a turning point of ∼ b ) A g intensity as a function of thickness showes a maximum at ∼
350 400 450 500 550 600 650 700Raman shift(cm −1 )A N=137N=30N=15N=10N=6A ( a )
440 445 450 455 460 465 470 475 480 485Raman shift(cm −1 )A N=10N=15N=30N=137N=6 ( b )FIG. 5. ( a ) Resonant Raman spectra for different number of layers N(normalised to A g and offset forconvenience). ( b ) Spectra of the A u mode, corresponding to the dashed spectral region in ( a ). to the gold substrate modifying the doping, and the electron-phonon interactions [6, 12]. Wediscard the first possibility by performing laser power dependent measurements. Different laserpower can lead to different surface temperatures. However, we find no observable changes inthe peak position vs the number of layers for different laser powers (Fig.3). The second possibleexplanation may be supported by previous literature reports which considered the non-resonantsituation. Several studies reported p-doping of MoS using various methods, including depositionon the gold substrate [6, 13], decoration with gold nano-particles (NP) [14], and using monolayer6oS transistor to adjust the doping level directly [15] resulting in stiffening of the A g mode.Here our results show the corresponding blue shift of the A g mode on the gold substrate for thenumber of layers N<10, which is consistent with the previous studies. Here, we, for the first time,observed that under the resonant conditions, the Raman-inactive A u mode of MoS on the goldsubstrate shows a strong blue shift compared with that on the SiO substrate. Fig.3 ( a ) shows ashift more than 1cm − when the number of layers is less than 10 layers.Fig.4 shows the nearly linear increase of the A g peak intensity until the number of layersreaches ∼
30 with the maximum point n=30, followed by a decrease until n=50. Note that theRaman signal intensity of bulk MoS is weaker than that with the number of layers of ∼
30. Thismay be attributed to the inference enhancement[16, 17] due to multiple reflections of the incidentlaser and emitted Raman signals in MoS flakes. Using the model of Wang et . al . , the intensity ofthe Raman signal can be expressed as I = (cid:90) d | t γ | dy , (1)where d is the thickness of the MoS flake, t is the amplitude of the electric field at the depth y ,and γ is the enhancement factor due to the multiple-reflection. The simulation is shown as a reddashed line in Fig.3 ( b ). The original model assumes no coherence between the field scatteredfrom adjacent layers. Considering coherence, Eq. (1) can be written as I = (cid:12)(cid:12)(cid:12)(cid:12) (cid:90) d t γ dy (cid:12)(cid:12)(cid:12)(cid:12) . (2)The coherent simulation is different, shown as a green dashed line in Fig.3 ( b ). Zeng et . al . pro-vided a qualititative explanation by using the accumulated phase shift for the n th reflected field anddestructive interference with the first reflected field from the flake surface [18]. ACKNOWLEDGEMENTS
We thank Profs. Marlan Scully, Alexei Sokolov and Zhenrong Zhang for helpful discussions.We also thank Prof. Marlan Scully for the access to Raman facilities and we thank Zhenrong Zhangfor help with sample preparation. D.V. acknowledges the support of NSF grant CHE-1609608. [1] A. Splendiani, L. Sun, Y. B. Zhang, T. S. Li, J. Kim, C. Y. Chim, G. Galli, and F. Wang, Nano Lett. , 1271 (2010).
2] J. Lin, H. Li, H. Zhang, and W. Chen, Appl. Phys. Lett. , 203109 (2013).[3] G. L. Frey, R. Tenne, M. J. Matthews, M. Dresselhaus, and G. Dresselhaus, Physical Review B ,2883 (1999).[4] B. Chakraborty, H. S. S. R. Matte, A. K. Sood, and C. N. R. Rao, J. Raman Spectrosc. , 92 (2013).[5] H. Li, Q. Zhang, C. C. R. Yap, B. K. Tay, T. H. T. Edwin, A. Olivier, and D. Baillargeat, AdvancedFunctional Materials , 1385 (2012).[6] M. Buscema, G. A. Steele, H. S. J. van der Zant, and A. Castellanos-Gomez, Nano Res. , 561 (2014).[7] R. Coehoorn, C. Hass, and R. A. de Groot, Phys. Rev. B , 6023 (1987).[8] J. V. Acrivos, W. Y. Liang, J. A. Wilson, and A. D. Yoffe, J. Phy. C , L18 (1971).[9] H. Li, Q. Zhang, C. C. R. Yap, B. K. Tay, T. H. T. Edwin, A. Olivier, and D. Baillargeat, Adv. Funct.Mater. , 1385 (2012).[10] A. M. Sanchez and L. Wirtz, Phys. Rev. B , 155413 (2011).[11] S. Najmaei, A. Mlayah, A. Arbouet, C. Girard, J. Leotin, and J. Lou, NANO , 12682 (2014).[12] U. Bhanu, M. R. Islam, L. Tetard, and S. I. Khondaker, Sci. Rep. , 5574 (2014).[13] B. J. Robinson, C. E. Giusca, Y. T. Gonzalez, N. D. Kay, O. Kazakova, and O. V. Kolosov, 2DMaterials , 015005 (2015).[14] Y.Shi, J. Huang, L. Jin3, Y. Hsu, S. Yu, L. Li, and H. Yang, Scientific Reports , 1839 (2013).[15] B. Chakraborty, A. Bera, D. Muthu, S. Bhowmick, U. V. Waghmare, and A. Sood, Phys. Rev. B ,161403 (2012).[16] Y. Y. Wang, Z. H. Ni, Z. X. Shen, H. M. Wang, , and Y. H. Wu, Appl. Phys. Lett. , 043121 (2008).[17] Z. Ni, Y. Wang, T. Yu, and Z. Shen, Nano Res , 273 (2008).[18] J.Zeng, J. Li, H. Li, Q. Dai, S. Tie, and S. Lan, Optics Express , 31817 (2015)., 31817 (2015).