Applied Physics Express     

LETTERS

A self-limiting layer-by-layer etching technique for2H-MoS2

To cite this article: Choong Hee Lee et al 2017 Appl. Phys. Express 10 035201

 

View the article online for updates and enhancements.

Related contentOn-Stack Two-Dimensional Conversion ofMoS2 into MoO3Taeg Yeoung Ko, Areum Jeong, WontaekKim et al.

-

Effects of plasma treatment on surfaceproperties of ultrathin layered MoS2Suhhyun Kim, Min Sup Choi, Deshun Quet al.

-

In situ thermal oxidation kinetics in fewlayer MoS2Rahul Rao, Ahmad E Islam, Philip MCampbell et al.

-

Recent citationsRemote N2 plasma treatment to depositultrathin high-k dielectric as tunnelingcontact layer for single-layer MoS2MOSFETQingkai Qian et al

-

This content was downloaded from IP address 128.146.189.69 on 12/02/2019 at 21:32

A self-limiting layer-by-layer etching technique for 2H-MoS2

Choong Hee Lee1*, Edwin W. Lee, II1, William McCulloch2, Zane Jamal-Eddine1, Sriram Krishnamoorthy1,Michael J Newburger4, Roland K. Kawakami3,4, Yiying Wu2, and Siddharth Rajan1*

1Department of Electrical and Computer Engineering, The Ohio State University, Columbus, OH 43210, U.S.A.2Department of Chemistry and Biochemistry, The Ohio State University, Columbus, OH 43210, U.S.A.3Program of Materials Science and Engineering, University of California, Riverside, CA 92521, U.S.A.4Department of Physics, The Ohio State University, Columbus, OH 43210, U.S.A.

*E-mail: lee.5107@osu.edu; rajan.21@osu.edu

Received November 3, 2016; accepted January 23, 2017; published online February 7, 2017

We report controlled layer-by-layer removal of large-area, sulfurized, single-crystal molybdenum disulfide (MoS2) films using a digital etchingtechnique, which utilizes oxidation and removal of the oxidized layer. We demonstrate a self-limiting oxidation process where Mo oxide covered thesurface of MoS2. A constant etching rate of one monolayer/cycle and the uniformity of the etching process were also verified. We show that theetching of an integer number of MoS2 layers can be precisely controlled. No noticeable film quality degradation was observed after multiple cyclesof digital etching, as confirmed by Raman mapping of the ratio of the E1

2g and A1g peak intensities. © 2017 The Japan Society of Applied Physics

The layered structure of semiconducting transition-metal dichalcogenides (TMDs) has been studiedextensively.1–4) Modulating the thickness of TMDs

enables tuning of the bandgap of the layered material owingto quantum confinement effects.5,6) For instance, bulk MoS2exhibits a 1.2 eV indirect bandgap, while monolayer MoS2has a 1.8 eV direct bandgap.7) The ability to control thenumber of layers is especially useful in optoelectronics dueto the transition from the indirect to direct bandgap.8,9)

Furthermore, it is advantageous in designing heterojunctiondevices as it allows one to tune the bandgap, affording greaterflexibility in device design.

Recently, various other controllable and repeatable etchingprocesses have been investigated. Castellanos-Gomez et al.10)

used a laser to thin down few-layer MoS2 flakes to a mono-layer by decomposing the exposed area via laser irradiation.Lu et al.11) air-annealed MoS2 flakes at 650 °C for severalhours and showed sustained crystalline quality and surfaceroughening. Other groups have reported dry etching process-es which used Ar, CF4, and XeF2 plasmas.12–14) Such dryetching processes are not self-limiting. These factors enhancethe possibility of etching rate fluctuation, which makes theprecise layer-by-layer control of etching challenging.

Here, we report layer-by-layer digital etching of sulfurized,large-area, single-crystal MoS2 films. The detailed synthesisprocess and the structural quality of the films are reported inour previous work.15) This digital etching technique wasadapted from existing etching techniques used for III–Vcompound semiconductors via a cycle of surface layer oxi-dation and removal of the oxidized layer by wet etching.16,17)

We show a stable “etching window”, which removes a singlemonolayer in one cycle. The oxidation conditions we useonly oxidize the first monolayer of the bulk MoS2 surface.Next, this oxide layer is removed by hydrochloric acid (HCl).Due to its chemical and physical properties, MoS2 has strongchemical resistance to common acids.14) Owing to suchproperties, the MoS2 itself functions as an excellent etchingstop layer. Therefore, the Mo oxide layer formed on thesurface of the sample can be selectively wet etched, makingthis etching technique a self-limiting process.

MoS2 samples were prepared by sulfurizing a Mo-coatedsapphire wafer in a quartz tube with MoS2 powder as thesulfur precursor. The tubes are loaded in a furnace and heatedto 1100 °C for 4.5 h. An array of a 6 × 55 µm2 rectangle

pattern was defined by photolithography to measure theetching depth. After patterning the photoresist, the samplewas placed in a plasma asher (Diener Pico) and oxidizedusing an oxygen plasma with a plasma power of 40W and aprocess pressure of 1.1 Torr. The plasma exposure time wasset to 5, 10, 20, and 30min to investigate the etching rate as afunction of oxidation time. The sample was then soaked inHCl for 5min to remove the oxidized layer followed by adeionized (DI) water rinse. Multiple oxidation=oxidized layerremoval cycles were repeated to perform layer-by-layeretching of the MoS2 films. Photoresist was then removedbefore measuring the etching depth and surface morphologyusing atomic force microscopy (AFM; Bruker INOVAmicroscope). The absolute film thickness was measured bymeasuring the step height between the MoS2 layer and ascratched area where MoS2 was removed. The number ofMoS2 layers after etching was also confirmed by measuringthe separation of the E1

2g and A1g peaks using Raman spec-troscopy (Renishaw Raman FTIR microprobe) with a 514 nmwavelength laser (1mW). To avoid sample-to-sample varia-tions, one MoS2 sample (area of 1 × 0.8 cm2) was used formeasuring the etching depth. The surface chemical bondingof the oxygen-plasma-treated MoS2 and the post-wet-etchingoxidized MoS2 film were analyzed using X-ray photoelectronspectroscopy (XPS; Thermo VG Multilab 2000).

Digital etching of MoS2 consists of a two-step process asillustrated in Fig. 1(a). 1) Oxidize a monolayer of MoS2 fromthe surface through oxygen plasma treatment. 2) Wet etch theoxide using HCl followed by a DI rinse. Steps 1 and 2 arerepeated to etch the film layer-by-layer down to the desiredthickness. The XPS spectra in the Mo 3d, S 2p, and O 1sregions for oxidized and wet-etched MoS2 samples are shownin Figs. 1(b)–1(d), respectively. All peaks were fitted using aGaussian function, and each fitted peak was plotted as theconvolution of the original XPS data. For both samples,S2− 2p3=2 (blue), S2− 2p1=2 (violet), S 2s (orange), Mo4+ 3d5=2(yellow), and Mo4+ 3d3=2 (green) peaks were observed atbinding energies of 163, 164, 227, 229.9, and 233.1 eV,respectively. After oxygen plasma treatment, the O 1s peakintensity was enhanced. At the same time, the S 2p andMo 3d peak intensities dropped markedly. Furthermore, twonew peaks appeared at 168.8 and 236.5 eV correspondingto S6+ and Mo6+ ligands, respectively. Here, the transitioninvolving moving to a higher oxidation state, from Mo4+ to

Applied Physics Express 10, 035201 (2017)

https://doi.org/10.7567/APEX.10.035201

035201-1 © 2017 The Japan Society of Applied Physics

Mo6+, clearly indicates the formation of MoO3 on thesurface.18) This process of oxidation of the MoS2 surfacelayer follows the chemical equation of 2MoS2 + 7O2 →2MoO3 + 4SO2, where SO2 is volatile, while MoO3 remainson the surface. The weakened peaks related to Mo–S bonds inthe MoS2 film are detected from the MoS2 layers underneaththe oxidized MoO3 surface layer due to the long penetrationdepth (several mm) of the illuminating X-ray. The appearanceof the S6+ peak indicates that phases other than MoO3 werecreated during oxidation of the MoS2 surface.19) After aHCl wet etching, the Mo6+ and S6+ species almost completelyvanished.

AFM images of one and two cycles of the digital etchingof MoS2 are shown in Figs. 2(a) and 2(b), respectively.10min of oxidation was performed and then a 30 × 30 µm2

area was scanned. The dark and bright regions indicateetched and non-etched areas, respectively, and the white linedepicts the path along which the height profile was taken inthe AFM image. Though the etching depth was less than1 nm, the color contrast between etched and non-etched areasgives us clear evidence of uniform etching. The etchingdepths from different oxidation times and numbers of cycleswere measured and are compiled in Fig. 2(c). Error bars weredetermined based on data measured from several differentareas on one sample. Based on XPS results, we assumedthat 5min of HCl exposure completely etched the oxidizedlayer in every cycle.

The constant etching depths of 0.7 nm (1 cycle) and 1.4 nm(2 cycles) clearly indicate that the etching rate is 1 monolayer(ML)=cycle. The etching depth remains constant despiteincreasing the oxidation time to as much as 30min. It canalso be inferred that even a short oxidation time of 5mincompletely oxidizes the surface layer of the bulk MoS2.Based on the oxidation process of MoS2, the oxide islands

nucleate randomly on the surface with a preferential directionand coalesce.20) This oxidation process is self-limiting due tothe rapid reduction of the diffusion rate of oxygen atoms,resulting in a constant etching depth for different oxidationtimes. Unlike digital etching of III–V semiconductors,16) thedigital etching rate of MoS2 does not follow a logarithmiccurve. Instead it shows a constant etching rate of 1ML=cyclewith negligible variation. We ascribe this to the structural

(a)

(b) (c) (d)

Fig. 1. (a) Schematic diagram of the digital etching process for layer-by-layer removal. XPS spectra of core levels of (b) Mo 3d and S 2s, (c) S 2p, and(d) O 1s measured from oxygen-plasma-treated and wet-etched MoS2 film.

(a) (b)

(c)

Fig. 2. 30 × 30 µm2 AFM scans of selectively etched MoS2 film with(a) 1 cycle and (b) 2 cycles of digital etching. (c) Etching depth for differentnumbers of cycles of digital etching as a function of oxidation time under O2

plasma.

Appl. Phys. Express 10, 035201 (2017) C. H. Lee et al.

035201-2 © 2017 The Japan Society of Applied Physics

difference between three-dimensional (3D) and two-dimen-sional (2D) semiconductors and the absence of out-of-planebonds in 2D layered materials.

We performed Raman mapping of the peak shifts andAFM measurements as the etched region inside the exposedpattern was thinned down from the bulk to a few monolayersbetween each digital etching cycle. Since etching down to amonolayer often formed islands instead of a coalesced filmdue to the inherited surface roughness of the MoS2 film, allfilms were etched down to 2ML. Figure 3(a) shows theRaman spectra of the MoS2 film for thicknesses ranging frombulk to 2ML. As the number of etching cycles increased andthe film thickness was reduced to a few monolayers, the peakintensities dropped and the frequency difference between thecharacteristic E1

2g and A1g peaks was reduced from 25 to22.4 cm−1. This reduction in peak spacing is due to the blueand red shift of the E1

2g and A1g modes, respectively, and canbe utilized in identifying the number of MoS2 layers.21)

Figure 3(b) shows both the frequency difference of the twopeaks and the corresponding MoS2 film thickness as afunction of the number of layers. A linear trend of the overallfilm thicknesses was confirmed by measuring the AFM stepheight on the scratched area of the same sample. The errorbars could be from the film roughness, which has an averageRMS value of 0.6 nm, close to the thickness of 1ML MoS2(shown later).

In Fig. 3(c), selectively etched MoS2 with patterning isshown in the optical microscope image, where the etchedarea (dark region) can be identified by the color contrast.Raman line scans and mapping were measured from thedashed line (20 µm) and area (25 × 35 µm2), respectively.Figure 3(d) shows Raman spectra along the vertical directionin Fig. 3(c). The non-etched region exhibits more than anorder of magnitude stronger intensity of the E1

2g and A1g

peaks as compared to the etched area. It can be seen that the

peak intensities change abruptly as the laser spot moves fromthe non-etched regions to the etched regions and vice versa.

Uniformity of the etched area was confirmed by perform-ing spatial Raman mapping inside the dashed square inFig. 3(c). Spatial Raman maps of both the frequency dif-ference and peak intensity ratio (E1

2g : A1g) are presented inFigs. 3(e) and 3(f), respectively. The peak frequency differ-ence ranges from 21 to 23 cm−1 in the etched area (orange),whereas it varies from 25 to 26 cm−1 in the bulk region (lightyellow), which verifies that the selective area is uniformlyetched. The variations of color contrast in the etched area aredue to the initial film thickness fluctuations originating fromthe CVD growth process. Further characterization of the filmquality after etching was achieved by mapping the intensityratio of the two dominant peaks. The intensity ratio in the as-grown sample shows a range of 1.0–1.3, which was reportedin our earlier article.4) The etched area shows an intensityratio in the range of 1.0–1.2, while the bulk region shows ahigher intensity ratio of 1.1–1.5. However, the peak intensityratio in both the etched and bulk regions falls within in thereference range of 1.0–1.3, indicating that the film qualityis sustained after the digital etching process. The photo-luminescence data shows a peak at ∼1.85 eV (A1 exciton),which corresponds to the direct bandgap at the K symmetrypoint in MoS2 (see the online supplementary data at http://stacks.iop.org/APEX/10/035201/mmedia). Based on the PLmeasurement of an exfoliated MoS2 film, as reported by Maket al.,22) it can be concluded that our etched MoS2 film is∼3ML thick, and this is in agreement with the AFM data.

Figure 4(a) shows AFM images (5 × 5 µm2) of etchedMoS2 films with different numbers of layers. The morphol-ogy of bulk MoS2 is also added as a reference for compar-ison. All images are presented with the same Z-scale of 5 nm.Similar surface morphologies and a similar RMS roughnessvalue of ∼0.6 nm were observed, even though multiple

(a) (b) (c)

(d) (e) (f)

Fig. 3. (a) Raman spectra of MoS2 film with thickness ranging from bulk (13ML) to 2ML after multiple cycles of digital etching. (b) Frequency differenceof E1

2g and A1g (red) and corresponding thickness (blue) as a function of number of layers. (c) Optical microscope image of selectively etched MoS2 film. Thedark areas denote selectively etched regions in MoS2. (d) Line scan Raman spectra of the MoS2 film along the dashed line in (c). (e) Contour map of frequencydifference and (f) intensity ratio of the E1

2g and A1g peaks taken from dashed area in (c).

Appl. Phys. Express 10, 035201 (2017) C. H. Lee et al.

035201-3 © 2017 The Japan Society of Applied Physics

etching cycles were applied. The RMS surface roughnessranges from 0.55 to 0.65 nm, marked by the diagonal lines inFig. 4(b), and it shows no clear correlation with the numberof digital etching cycles, indicating that the surface mor-phology is not affected by the etching process. Since theRMS roughness is close to the thickness of 1ML of MoS2,there can be a 1–2ML thickness variation across the sample,which was shown earlier in the Raman mapping images. Anadditional etching experiment was also conducted usingexfoliated WS2 to investigate whether the technique demon-strated in this work can be applied to other 2D materials.Multiple cycles of oxidation and wet etching were used toetch 4 nm exfoliated WS2 (see the online supplementary dataat http://stacks.iop.org/APEX/10/035201/mmedia).

In conclusion, we have introduced a digital etchingtechnique to thin large-area, multilayered MoS2 down tothe desired thickness by removing a single monolayer percycle. A constant etching rate of one monolayer=cycleand uniform etching were confirmed by AFM and Ramanmeasurements. The controllable layer-by-layer etching ofselective areas in MoS2 films can be applied to control thelayer thickness in epitaxially grown films, and can enablelateral engineering of structures such as bulk-monolayerheterojunctions.

Acknowledgment We acknowledge the support from Air Force Officeof Scientific Research (Dr. Kenneth Goretta) under Contract No. FA9550-15-1-0294.

1) B. Radisavljevic, A. Radenovic, J. Brivio, V. Giacometti, and A. Kis, Nat.Nanotechnol. 6, 147 (2011).

2) B. Radisavljevic, M. B. Whitwick, and A. Kis, Appl. Phys. Lett. 101,043103 (2012).

3) E. W. Lee, II, C. H. Lee, P. K. Paul, L. Ma, W. D. McCulloch, S.Krishnamoorthy, Y. Wu, A. R. Arehart, and S. Rajan, Appl. Phys. Lett. 107,103505 (2015).

4) C. H. Lee, W. McCulloch, E. W. Lee, II, L. Ma, S. Krishnamoorthy, J.Hwang, Y. Wu, and S. Rajan, Appl. Phys. Lett. 107, 193503 (2015).

5) A. Kuc, N. Zibouche, and T. Heine, Phys. Rev. B 83, 245213 (2011).6) R. Frindt, Phys. Rev. 140, A536 (1965).7) A. Splendiani, L. Sun, Y. Zhang, T. Li, J. Kim, C.-Y. Chim, G. Galli, and F.

Wang, Nano Lett. 10, 1271 (2010).8) W. Ho, J. C. Yu, J. Lin, J. Yu, and P. Li, Langmuir 20, 5865 (2004).9) E. Gourmelon, O. Lignier, H. Hadouda, G. Couturier, J. Bernede, J. Tedd, J.

Pouzet, and J. Salardenne, Sol. Energy Mater. Sol. Cells 46, 115 (1997).10) A. Castellanos-Gomez, M. Barkelid, A. Goossens, V. E. Calado, H. S.

van der Zant, and G. A. Steele, Nano Lett. 12, 3187 (2012).11) X. Lu, M. I. B. Utama, J. Zhang, Y. Zhao, and Q. Xiong, Nanoscale 5, 8904

(2013).12) Y. Liu, H. Nan, X. Wu, W. Pan, W. Wang, J. Bai, W. Zhao, L. Sun, X.

Wang, and Z. Ni, ACS Nano 7, 4202 (2013).13) M. H. Jeon, C. Ahn, H. Kim, K. N. Kim, T. Z. Lin, H. Qin, Y. Kim, S. Lee,

T. Kim, and G. Y. Yeom, Nanotechnology 26, 355706 (2015).14) Y. Huang, J. Wu, X. Xu, Y. Ho, G. Ni, Q. Zou, G. K. W. Koon, W. Zhao,

A. C. Neto, and G. Eda, Nano Res. 6, 200 (2013).15) L. Ma, D. N. Nath, E. W. Lee, II, C. H. Lee, M. Yu, A. Arehart, S. Rajan,

and Y. Wu, Appl. Phys. Lett. 105, 072105 (2014).16) J. Lin, X. Zhao, D. A. Antoniadis, and J. A. Del Alamo, IEEE Electron

Device Lett. 35, 440 (2014).17) T. Meguro, M. Hamagaki, S. Modaressi, T. Hara, Y. Aoyagi, M. Ishii, and

Y. Yamamoto, Appl. Phys. Lett. 56, 1552 (1990).18) J. Yang, S. Kim, W. Choi, S. H. Park, Y. Jung, M.-H. Cho, and H. Kim,

ACS Appl. Mater. Interfaces 5, 4739 (2013).19) N. M. Brown, N. Cui, and A. McKinley, Appl. Surf. Sci. 134, 11 (1998).20) M. Yamamoto, T. L. Einstein, M. S. Fuhrer, and W. G. Cullen, J. Phys.

Chem. C 117, 25643 (2013).21) L. Liang and V. Meunier, Nanoscale 6, 5394 (2014).22) K. F. Mak, C. Lee, J. Hone, J. Shan, and T. F. Heinz, Phys. Rev. Lett. 105,

136805 (2010).

(a)

(b)

Fig. 4. (a) AFM images of surface morphologies of MoS2 film etcheddown from bulk to 2ML. (b) RMS roughness as a function of number ofMoS2 layers. Orange diagonal lines indicate the range of the RMS values.

Appl. Phys. Express 10, 035201 (2017) C. H. Lee et al.

035201-4 © 2017 The Japan Society of Applied Physics