Domain Architectures and Grain Boundaries in Chemical Vapor

Aug 4, 2016 - Recent studies have shown that vapor phase synthesis of structurally isotropic two-dimensional (2D) MoS2 and WS2 produces well-defined d...
1 downloads 6 Views 1MB Size
Download PDF
Subscriber access provided by Northern Illinois University

Communication

Domain Architectures and Grain Boundaries in Chemical Vapor Deposited Highly Anisotropic ReS2 Monolayer Films Kedi Wu, Bin Chen, Sijie Yang, Gang Wang, Wilson Kong, Hui Cai, Toshihiro Aoki, Emmanuel Soignard, Xavier Marie, Aliya Yano, Aslihan Suslu, Bernhard Urbaszek, and Sefaattin Tongay Nano Lett., Just Accepted Manuscript • DOI: 10.1021/acs.nanolett.6b02766 • Publication Date (Web): 04 Aug 2016 Downloaded from http://pubs.acs.org on August 4, 2016

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Nano Letters is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 12

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Nano Letters

Domain Architectures and Grain Boundaries in Chemical Vapor Deposited Highly Anisotropic ReS2 Monolayer Films Kedi Wu1, Bin Chen1, Sijie Yang1, Gang Wang2, Wilson Kong1, Hui Cai1, Toshihiro Aoki1, Emmanuel Soignard1, Xavier Marie2, Aliya Yano1, Aslihan Suslu1, Bernhard Urbaszek2, and Sefaattin Tongay1* 1

School for Engineering of Matter, Transport and Energy, Arizona State University, Tempe, AZ 85287, USA

2

Universit´e de Toulouse, INSA-CNRS-UPS, LPCNO, 135 Av. Rangueil, 31077 Toulouse, France

Abstract Recent studies have shown that vapor phase synthesis of structurally isotropic two-dimensional (2D) MoS2 and WS2 produces well-defined domains with clean grain boundaries (GBs). This is anticipated to be vastly different for 2D anisotropic materials like ReS2 mainly due to large anisotropy in interfacial energy imposed by its distorted 1T crystal structure and formation of signature Re chains along [010] b-axis direction. Here, we provide first insight on domain architecture on chemical vapor deposited (CVD) ReS2 domains using high-resolution scanning transmission electron microscopy (HRSTEM), angle-resolved nano-Raman spectroscopy (ANRS), reflectivity, and atomic force microscopy (AFM) measurements. Results provide ways to achieve crystalline anisotropy in CVD ReS2, establish domain architecture of high symmetry ReS2 flakes, and determine Re-chain orientation within sub-domains. Results also provide a first atomic resolution look at ReS2 GBs, and surprisingly we find that cluster and vacancy defects, formed by collusion of Re-chains at the GBs, dramatically impact the crystal structure by changing the Re0 chain direction and rotating Re-chains 180 along their b-axis. Overall results not only shed first light on domain architecture and structure of anisotropic 2D systems, but also allow to attain much desired crystalline anisotropy in CVD grown ReS2 for the first time for tangible applications in photonics and optoelectronics where direction dependent dichroic and linearly polarized material properties are required. Keywords: 2D materials, ReS2, anisotropic materials, Chemical vapor deposition, Grain boundary, synthesis

Introduction Rhenium disulfide (ReS2) is a member of semiconducting transition metal dichalcogenides (TMDCs) with an optical bandgap around 1.4 eV (bulk) and 1.6 eV (monolayer)1. In comparison to group VITMDCs such as MoS2 and WSe2, the rhenium atom in ReS2 has an extra electron that is shared between two neighboring rhenium atoms1. Earlier literature and recent work have shown that highly oriented ReRe chains (Re-chains) are formed along b-axis lattice direction due to strong interaction and dimerization between the adjacent Re atoms (Figure 1a). The presence of in-plane anisotropy distinguishes ReS2 and ReSe2 from the other members of TMDCs family structurally1-4, electronically5-9, and optically2, 10. More recently, the influence of structural anisotropy on the physical properties of ReS2 and ReSe2 has been studied on exfoliated layers by various teams, and findings suggest that group-VII TMDC monolayers may offer unique applications in polarized detectors, sensors, and photonic devices where dichroism and light polarization capabilities are essential5, 8, 11 . Currently, these studies are based on mechanically exfoliated samples which is ideal for understanding their fundamental properties and exploring their potential in variety of applications. To unravel their true potential, however, it is essential to develop feasible and industrially compatible growth techniques to synthesize these layers at large scales. Thus far, vertically oriented ReS2 nanosheets have been synthesized using chemical vapor deposition (CVD) for energy applications12, 13. More recently, synthesis of 2D planar ReS2 flakes have been demonstrated by Keyshar14 using lowtemperature CVD process onto thermal SiO2 which has enabled ReS2 synthesis in 2D form. However, 1 ACS Paragon Plus Environment

Nano Letters

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

owing to amorphous nature of thermal oxide (SiO2) substrate the growth conditions cannot be controlled to achieve high enough crystallinity required to yield high structural anisotropy, i.e., well oriented Re chains. Following this promising growth route14, a number of important questions arise: 1) Can highly crystalline ReS2 monolayers with noteworthy structural anisotropy be synthesized? This is of particular concern as anisotropy in interfacial energy (due to crystalline anisotropy) stabilizes dendritic growth mode, and prevents well-define crystalline orientation. Herein results show that synthesis of structurally anisotropic ReS2 CVD domain growth is possible. 2) How does domain arrangement / architecture differ in ReS2 compared to structurally isotropic TMDCs such as MoS2 and WS2? And how is the b-axis (Re chain direction) oriented within each ReS2 domain? Current state of knowledge on domain structure of CVD MoS2 is relatively well-established but is completely unknown for CVD anisotropic materials like ReS2. Our results present the first data on this by demonstrating how sub-domains arrange themselves to build larger scale flakes, and demonstrate in which direction Re-chains are oriented within each sub-domain. 3) Grain boundaries (GBs) of 2D materials (like graphene and MoS2) are known to extend out linearly in one particular direction, vacancy defects are energetically most stable when arranged in 4|6, 4|8, 5|7, and 6|8 rings. How is this different for anisotropic materials? From geometrical considerations alone, formation of GBs and atomic arrangement of 0D defects located at the GBs are anticipated to be vastly different from isotropic group-VI TMDCs. Our results show that cluster defects and a variety of other vacancy defects are formed when Re-chains coalesce into each other at different approaching angle. Interestingly, we find that these vacancy defects alter the direction of Re-chains and even cause unusual Re-chain rotation around b-axis. As summarized above, our results establish the domain architecture and grain boundaries of ReS2 for the first time using nm resolution angle-resolved nano Raman spectroscopy (ANRS), high resolution scanning transmission electron microscopy (HRSTEM), atomic force microscopy (AFM), cryogenic temperature reflectivity spectroscopy, and kelvin probe force microscopy (KPFM surface potential). Established domain architecture and the attained crystalline anisotropy in CVD grown ReS2 observed here for the first time is the cornerstone for developing direction dependent dichroic and linearly polarized material properties for photonics and optoelectronics. Understanding growth dynamics. Since many attractive properties of ReS2 heavily rely on their structural anisotropy2, 5, 8, 9, it is essential to develop techniques to achieve highly crystalline ReS2 monolayers with well-defined domain structure where a- and b-axis lattice directions can be confidently distinguished. In this work, monolayer ReS2 was grown using by transporting ammonium perrhenate (NH4ReO4 ≥99% Sigma Aldrich) and sulfur gas precursors which have been recently shown to be effective14. Unlike prior work on amorphous SiO2 substrates14, we have chosen double polished c-cut sapphire wafers owing to lower energy barrier for nucleation at atomically sharp sapphire step edges, and better control over surface cleanliness by chemical methods. Based on measurements and analysis performed on more than 100 CVD trials and samples, we find that use of sapphire substrate is essential to achieve control over thickness, crystallinity, and structural anisotropy. During growth, precursors were transported to sapphire wafers at different temperatures (490 0C – 520 0C), flow-rates (10-50 sccm), and NH4ReO4 pre-cursor amounts (10-50 mg) (see methods and supplementary information S1 for details). Here, NH4ReO4 precursors provide highly volatile Re2O7 and ReO2 rhenium sub-oxides after their thermal decomposition at 300 °C - 500 °C, and at a suitable 2 ACS Paragon Plus Environment

Page 2 of 12

Page 3 of 12

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Nano Letters

temperature, reaction time, and sulfur amount, reaction proceeds to carry out oxide sulfurization reaction to yield ReS215. Based on this principle, it is postulated that after formation of intermediate rhenium oxide complexes and physi-sorption on the surface, they diffuse on the substrate till they lose their kinetic energy at the terrace, kink, or defects. Once sulfurized, ReS2 is nucleated and growth is initiated in one of the common 2D growth modes (layer by layer, dendritic, or screw-dislocation driven).

Figure 1 Synthesis of monolayer ReS2 on c-cut sapphire substrates and characterization. a. Schematic depiction of monolayer ReS2 identifying b-axis [010] Re-chain direction and a-axis [100] across Re-chains. Optical images taken on b. triangle-like (truncated) c. a group of preferentially oriented hexagons in two distinct directions (red and black dashed lines) d. preferentially aligned and nearly coalesced hexagons, and e. full merged ReS2 monolayers. f. Comparison of Raman spectra for bulk (blue), exfoliated monolayer (orange), and CVD grown ReS2 (black) g. electron energy loss spectroscopy (EELS) and micro-absorption spectroscopy spectrum acquired on CVD grown monolayer ReS2. AFM measurements on h. hexagonal and i. triangle-like flakes showing uneven and dendritic edges.

After typical chemical vapor transport (CVD) process, we have observed a variety of highly crystalline different kind of ReS2 monolayer domain shapes, including separated truncated ReS2 triangles (Figure 1b), relatively large area separated or nearly merged hexagonal ReS2 domains (Figure 1c-d), and fully merged ReS2 monolayers (Figure 1e). We note that hexagonal features were observed at lower temperatures (490 0C) and low flow-rates (12-15 sccm), whereas truncated triangle shapes were achieved at much higher temperatures (500-520 0C) in relatively broad flow-rates (12-50 sccm) with reasonable reproducibility (Figure S1 and discussions). It is noteworthy to mention that perfect triangles (3-fold symmetry) were not observed, and all triangle-like features appeared to be truncated as shown 3 ACS Paragon Plus Environment

Nano Letters

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

in Figure 1i and Figure 2g possibly due to reduced crystalline symmetry of ReS2 in comparison to MoS2 with perfect 3-fold symmetry. A small nanodot like formation in the middle of each hexagonal ReS2 domain in Figure 1c-d is identified as a nucleation site, suggesting that ReS2 is nucleated at random locations with specific nucleation density (Figure S1) on the bare substrate. During the growth, the nucleation sites continue to grow and form grain boundaries when two or more domains met, resulting in a partially continuous film (Figure 1c-d). These large-area monolayer ReS2 samples measure nearly a centimeter in size. Single point Raman spectroscopy measurements and comparison to exfoliated ReS2 confirmed the presence of ReS2 (Figure 1f), and unintentional defect density was low enough to achieve FWHM values in the 4-6 cm-1 range. Nano-electron energy electron loss (nano-EELS) and microscale absorption spectroscopy data further confirmed the presence and optical properties of monolayer ReS2 as shown in Figure 1g. It is noteworthy to highlight that ReS2 domains preferentially orient themselves in mostly two distinct directions on sapphire surface (see red and black hexagons in Figure 1c-d) which is in large contrast to ReS2 growth on amorphous (thermal) SiO2 known to yield thick ReS2 layers completely randomly oriented across the wafer12, 14. Previously, theoretical and experimental studies have shown a significant reduction in the energy barrier of 2D material (MoS2 or graphene) nucleation close to the step edges as compared to flat substrates16, 17. We propose that similar catalytic processes are also involved in nucleation and diffusion of ReS2 monolayer growth at the sapphire terraces, which results in preferentially oriented hexagonal domains (Figure 1b-c). Through further studies, we determined that temperature, flow rate, and pre-cursor amount are the most crucial growth parameters that directly control the domain structure, hence the anisotropy of ReS2 monolayers. These parameters and their effects on the nucleation density, monolayer coverage, domain size and shape are discussed in the supplementary information (Figure S1). Despite largely different growth conditions, we note that the outer edges of these flakes always display dendritic like growth suggesting that the monolayer flake itself may be grown in dendritic form without any coherent / well-defined structural anisotropy (Figure 1 h-i). Representative AFM images on ReS2 triangle and hexagonal domains show that unlike MoS2 the edges do not appear to be atomically sharp, and dendritic features can be easily identified. Due to random nature of dendritic growth, this may imply that synthesized flakes may lack a structural anisotropy within each domain due to randomly oriented dendritic features, and may compromise its overall direction dependent properties. Surprisingly, however, we find that structural anisotropy is attained within each domain as discussed below. Internal domain structures of ReS2 monolayers. To develop deeper understanding of growth and internal domain structure of ReS2, we have performed angle resolved nano-Raman spectroscopy measurements on ReS2 triangular-like and hexagonal-like structures (see methods). As shown in Figure 1f, a series of Raman modes appear in the range of 100–400 cm-1, and their peak positions match well with reported values in the literature1. These modes are mostly associated with in-plane (E2g) and outof-plane (Ag) like vibrations, and due to lower symmetry (1T’) of ReS2, the total number of peaks is much larger compared to other 2D systems such as MoS24. Here, one particular Raman peak (located at 214 cm-1) is noteworthy to discuss: Recent studies have shown that intensity of 214 cm-1 mode reaches its maximum when polarization vector is nearly parallel to the b-axis lattice direction2-4 (see Figure S2 discussions for more details). Measuring Raman intensity of 214 cm-1 peak at different angles yields 2lobed shapes that can be described as a cos2(x) function, and the direction of the 2-lobe allows one to 4 ACS Paragon Plus Environment

Page 4 of 12

Page 5 of 12

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Nano Letters

identify the absolute orientation of a ReS2 domain with respect to the laboratory axis. In line with these studies2-4, our angle-resolved Raman measurements on bulk and exfoliated ReS2 confirmed similar trends as shown in Figure S2.

Figure 2 Angle-resolved nano-Raman and reflectivity spectroscopy a. Angle-resolved nano-Raman spectroscopy (ANRS) mapping data at 214 cm-1 peak at different polarization angles. Angle refers to the angle between the b-axis lattice chain direction and the polarization vector b. Construction of grain boundaries (black dashed line) and Re-chain direction in each sub-domain (see colored straight lines within a hexagonal flake). c. ANRS data taken from the middle section of the hexagon yielding isotropic-like response due to averaging over all domains with different Re-chain orientation. Angle dependent polar plots for d. domain A (red), B (green), and C (blue) e. D (orange), E (magenta), and F (grey) in a hexagonal flake. f. Angle dependent reflectivity data acquired at T= 4 K in normalized intensity g. Construction of domain structure of ReS2 truncated triangle flakes. Red and blue lines refer to grain boundaries and Re-chain direction respectively (red, green, and blue arrows). Angle dependent polar plots for h. different domains (red, green, and blue dots) and i. grain boundaries (orange, magenta, and olive dots in 2g). Fitted cos2(x) functions reach a maximum value along b-axis direction and allow for determining Re-chain direction.

To understand the domain structure of ReS2 monolayers and determine Re-chain direction, we have first focused on hexagonal ReS2 flakes and performed Raman intensity mapping at 214 cm-1 peak at 5 ACS Paragon Plus Environment

Nano Letters

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

different polarization angles from 00 (Polarization P ∥ b-axis) at 600 steps. In Figure 2a, we show one of our main findings: Successive Raman mapping images taken at different polarization angles demonstrate that different segments of the hexagons yield at least an order of magnitude more signal compared to other regions. More interestingly, these individual segments (sub-domains that are ~10 microns in size) appear to be triangular in shape, and six different domains can easily be identified. Despite the presence of dendritic edges observed in AFM images (Figure 1h-i), this finding convincingly demonstrates that hexagonal flakes are not made of randomly oriented Re-Re chains (b-axis), but instead are made of subdomains. These domains appear to have grain boundaries roughly running from 300 to 2100, 900 to 2700, and 1500 to 3300 as highlighted by red lines in Figure 2b. A closer look at Figure 2a shows that equally bright triangles are also positioned 180 degrees from each other. This suggests that Re-chain direction is closely aligned for sub-domains located across from each other. We note that polarization dependent response within triangular sub-domains (constituting the hexagon itself) is in stark contrast to group-VI TMDCs domains with no polarization dependence in Raman intensity mapping due to their highly isotropic crystal structure. Here, one natural question arises: What is the domain structure of hexagonal ReS2 flakes and what is the b-axis orientation of each individual sub-domain? Raman mapping data already suggest that subdomains across from each other form pairs with closely matched b-axis direction (see bright triangles across from each other in Figure 2a). To determine the b-axis direction within each domain, we have performed angle resolved measurements on various spots within each triangular domain (A, B, C, D, E, and F, see notation in Figure 2b), and plotted 214 cm-1 peak intensity as a function of angle. Here, the Raman laser was tightly focused onto each triangular domain to collect structural information, and special care was given to collect data only from one single domain. Angle resolved measurements on each triangular sub-domains, however, yield two-lobed polar plot and the orientation of b-axis with respect to laboratory axes can be identified by fitting to y=A+B*cos2(α+α0). Results show that 214 cm-1 Raman intensity in domain A, B, C, D, E, and F reaches its maximum at 1360, 500, 3480, 3030, 2140, and 1600 degrees (Figure 2d-e). Considering ~100 degree offset between these values and the b-axis (see our calibration work in Figure S2), the following can be concluded from the angle-resolved Raman spectroscopy data: i) Opposite triangular domains (A-D, B-E, and C-F) have the nearly similar b-axis orientation (180 degree rotation) which explains why domains across from each appear bright at a given polarization angle (Figure 2a). ii) b-axis of domain A, B, D, and E appear nearly at the grain boundary whereas b- lattice direction runs perpendicular to the edge in domain C and F. Based on these, the proposed domain structure is depicted in Figure 2b (each arrow indicated anticipated Re-chain direction). While individual domains yield strongly anisotropic response, middle of the hexagon or probing the whole hexagonal ReS2 (using lower magnification larger area probing spot) yields isotropic response in Figure 2c. This suggests that each hexagonal flake behaves as if isotropic MoS2 at macroscale but each sub-domain has the well-defined structural anisotropy. In addition to ANRS measurements, we have performed micro-reflectivity measurements at different polarization directions to confirm proposed Re-chain directions by an independent technique. It is noteworthy to emphasize that reflectivity measurements were preferred over photoluminescence measurements as reflectivity probes the intrinsic properties of ReS2 and is less sensitive to defect density and polarization direction of excitons which are not necessarily along high symmetry crystal direction. Moreover, PL emission line from ReS2 has a strong overlap with broad sapphire-defect peaks at 1.6 eV which prevents one to distinguish one another. In Figure 2f, we show reflectivity data acquired 6 ACS Paragon Plus Environment

Page 6 of 12

Page 7 of 12

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Nano Letters

from individual domains within the hexagonal flake at low (4K) temperatures. Comparison between Figure 2d and 2f reveals that angle resolved reflectivity response from domains A, B, and C matches closely with angle-resolved Raman spectrum from the same domains (thus, proposed Re-chain directions). Similar to angle resolved Raman signals, reflectivity reaches to its maximum value when polarization direction is parallel to b-axis, and offers an additional way to confirm Re-chain direction in synthesized ReS2 layers (Figure S3 for methods). It is noteworthy to mention that when samples are cooled rapidly after growth or under other nonideal circumstances, such as thermal variations and flow-rate changes during synthesis, hexagonal domain architecture display slight changes from those presented in in Figure 2. In such case, Re-chain direction may deviate within +/- 10 degrees, but retains its overall architecture and triangular subdomain structure as shown in Figure S4. In contrast to hexagonal flakes, truncated triangular ReS2 can reconstruct in one of the two possible domain arrangements depicted in Figure 2g (red and blue dashed lines). In order to assess domain architecture of these triangular regions, we have performed similar angle-resolved Raman measurements as shown in Figure 2h-I (see also mapping in Figure S5). Polar Raman plots of 214 cm-1 peak on red, green, and blue dots display good anisotropy and fitting to cos2(x) function reveal that Rechain directions are oriented towards the corners (90, 330, and 210 degrees, respectively). In contrast, measurements on orange, magenta, and dark green spots are nearly isotropic, implying that Re-chains are randomly oriented in close proximity to grain boundaries either because crystalline anisotropy is partly cancelled due to different b-axis orientation in the adjacent domains, or (as discussed in greater detail in the next section) defects at the boundaries strongly impact Re-chain direction. Overall findings suggest that grain boundaries can be described as red lines and Re-chains within each sub-domain run parallel to blue dashed lines as shown in Figure 2g. Overall, b-axis within each sub-domain orients perpendicular to the truncated edges as depicted in Figure 2g. Based on similar measurements on a variety of samples (>10 samples) this appears to be a common phenomenon for truncated triangles. However, a few exceptions were observed especially when these flakes were grown under non-ideal extremely high temperature- conditions as discussed for hexagonal flakes in Figure S4. This is perhaps due to high number of defects introduced during non-ideal growth condition which causes anisotropy to be reduced as discussed in Figure 3. Grain Boundary and TEM studies. From growth and manufacturability perspective, discussed data suggests that more advanced growth techniques must be invented to achieve anisotropic (structural, optical, and electronic) response at chip scales. To achieve a controlled anisotropic growth one needs to know how and why grains are formed and arranged. Previous studies have shown that grain boundaries in graphene can be described as arrays of dislocations formed by mainly 5-7 carbon rings and variety of different kinds of carbon arrangements mediated by carbon vacancy (VC). It has been also well documented that grain boundaries of isotropic binary 2D materials, such as h-BN and MoS2, contain similar pristine (vacancy) or sulfur substituted 4|6, 4|8, 5|7, and 6|8 B-N and Mo-S rings16, 18, 19. However, grain boundary structure of ReS2 is anticipated to be vastly different from other 2D systems due to anisotropy in interfacial energy and vast number of different atomic arrangements around defect sites11. However, little to none is known about the detailed grain boundary structures in anisotropic 2D systems such as ReS2 and ReSe2.

7 ACS Paragon Plus Environment

Nano Letters

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Other structural imperfections also exist away from the boundary region: For example, green dashed circles in Figure 3 highlight commonly observed S-Re-S pair vacancies - triangular black missing patches(Figure 3f). These defects change atomic level local interactions sufficiently enough that it can lead to 180o in-plane chain rotation by Re slip motion as shown in Figure 3f. Interestingly, atomic structure around cluster defects (red dashed circles in Figure 3d) is heavily influenced by the strain imposed through these imperfections: Closer look at the highlighted false-color red strips (Re chains) demonstrate that these chains do not extend along infinitely but instead change their directions in multiples of 60o around these cluster defects (Figure 3c and 3g). Since a large number of defects are present at the boundary, similar b-axis rotation action is apparent in close proximity to grain boundary (blue strips near boundary Figure 3c). We note that these observed defect behaviors are largely different from other isotropic 2D systems, such as graphene and MoS2, and provides valuable insight into structure of ReS2 grain boundary.

Figure 3 High-resolution scanning transmission electron microscopy (HRSTEM) characterization at the grain boundaries. a. HRSTEM images taken from ReS2 monolayers transferred onto TEM grids showing the quasi-1D nature of synthesized monolayers and b. schematic depiction of ReS2 monolayers and b- and a-axis lattice directions. c. HRSTEM image taken at ReS2 grain boundaries and specific zoom in images from d-e. grain boundary region, f-g away from the grain boundaries. h. HRSTEM image and SAED diffraction pattern collected from ReS2 bilayers, and i. constructed images of each individual layer showing ~14 degree rotation angle. j. Schematic depiction of vertically stacked ReS2 bilayers with ~140 degree rotation angle between adjacent layers.

Even though CVD process predominantly results in monolayer ReS2, when grains continue to grow on top of each other, growth also yield bilayer CVD ReS2 local regions. High resolution STEM images taken from bilayer ReS2 displays moiré fringes (Figure 3h), and inspection of the diffraction pattern (Figure 3h inset) reveals that each ReS2 reciprocal lattices are rotated by ϕ=14° with respect to each other as Figure 3h inset. The fast Fourier transform analysis in Figure 3i shows the two ReS2 monolayers stacked onto 8 ACS Paragon Plus Environment

Page 8 of 12

Page 9 of 12

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Nano Letters

each other and their superposition of the two lattices produces moiré fringes in Figure 3h. Here, our studies suggest that ReS2 layers may also stack onto each other rotationally instead of vertical or other more complex stacking of Re-chain. Discussion. Above arguments and findings suggest that structural anisotropy can be attained within sub-domains constituting the high symmetry (hexagonal or truncated triangle) flakes as shown in Figure 2. However, the degree of micro-scale structural anisotropy of ReS2 sub-domains is likely to be reduced by alterations in Re-chain direction (Figure 3c-g) imposed by variety of imperfections such as cluster defects and vacancies. Therefore, uncontrolled anisotropy (caused by imperfections) at nanoscale prevents polarization selection and reduces the polarization dependent material properties. This is particularly the case in close proximity to defect rich grain boundaries in which Re-chain direction is strongly altered by the presence of variety of vacancy defects. The structural anisotropy within ReS2 sub-domains is quickly recovered away from grains boundaries in relatively defect free-regions, and these regions display rather large anisotropy. Indeed, angle-resolved Raman mapping and angle resolved reflectivity measurements presented in Figure 2 support the presence of these sub-domains with well-defined b-axis direction and associated polarization dependent Raman/reflectivity response. Based on our findings above, we predict that structural anisotropy and polarization dependent properties will be negligible for layers with high density of defects due to nearly randomly oriented Rechains caused by imperfections. It is also anticipated that achieving macroscale (chip-scale) ReX2 (X=S and Se) with large anisotropy will be a large challenge, and either more sophisticated molecular beam epitaxy (MBE) technique needs to be utilized to achieve large area single crystalline ReS2 layers, or selective epitaxy routes need to be developed for fabrication at a targeted area with well-defined crystalline anisotropy. Conclusion. State-of-art angle resolved nano-Raman spectroscopy, high resolution scanning transmission electron microscopy, 4K angle resolved reflectivity spectroscopy, atomic force microscopy and kelvin-probe force microscopy measurements allowed us to explore domain architecture, grain boundaries, and crystalline anisotropy of CVD grown anisotropic ReS2 2D materials for the first time. Improved CVD synthesis of ReS2 monolayers onto [0001] (c-cut) sapphire substrates allowed production of highly crystalline ReS2 domains with well-defined structural anisotropy for the first time. Angle resolved nano-Raman spectroscopy mapping measurements convincingly demonstrate the presence of structurally anisotropic ReS2 sub-domains. This is particularly surprising since anisotropic interfacial energy stabilizes (completely random) dendritic growth as evidenced by dendrite like features at the edges of the CVD flakes. Angle resolved Raman and reflectivity measurements allowed us to determine the direction of Rechains (b-axis) within each sub-domain (of hexagonal and truncated triangular flakes), and establish the domain architecture of quasi-1D anisotropic 2D material systems for the first time. Our HRSTEM measurements provided the very first insight into atomic restructuring at ReS2 grain boundaries (GBs): We find that variety of cluster and vacancy defects at the GBs are formed by collusion of two Re-chains approaching at different angles, and these vacancy defects are much complicated in comparison to 4|6, 4|8, 5|7, and 6|8 rings commonly found in other 2D systems. Surprisingly, we find that defects near and away from the GBs cause Re-chains to change direction and even rotate chains in 180 degrees around their own (b-) axis, and reduce the anisotropic response of ReS2 layers. Methods 9 ACS Paragon Plus Environment

Nano Letters

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 12

Growth: Prior to growth, sapphire substrates were treated with oxygen plasma for 3 minutes to remove any contaminants on the surface, and placed on the top of alumina boat containing ammonium perrhenate (NH4ReO4) precursor. This ceramic boat was placed at the center of the heating zone of the furnace, and sulfur (≥ 99.88% Sigma Aldrich) precursor was placed approximately 18.5 cm upstream from the NH4ReO4 precursor. The monolayer growth was performed at atmospheric pressure (ATM), and precursors were transported to the substrate under o high purity argon flow. Samples were first heated to 300 C (100 sccm Ar flow), and flow was gradually decreased o 0 0 to 50 scmm at the growth temperature (490-520 C). Sulfur precursors were sublimated at ~350 C - 420 C 0 temperature range and carried to the surface by Ar flow. Growth was carried out at 490-520 C for 25 minutes o -1 0 0 -1 followed by controlled cooling (5 C min ) to 420 C and fast cooling (>10 C min ) to room temperature. STEM and EELS measurements: The CVD grown ReS2 monolayers were transferred from sapphire substrates to copper grids with holey carbon film. Valence electron energy loss spectroscopy was performed using Nion HighResolution Monochromated EELS STEM (HERMES) system consisting of a Nion STEM 100 scanning transmission electron microscope (STEM) equipped with a Nion high energy resolution monochromator and a modified Gatan Enfinium EEL spectrometer. The energy resolution, measured from the half width of the zero-loss peak was set to be 60 meV. To improve S/N, zero loss peak (ZPL) was cut off from 0.5 eV. The accelerating voltage was 40 kV, the probe size was about 0.3 nm, and the probe current was approximately 12 pA . Optical characterization: Raman measurements for the as grown samples were performed in a Renishaw InVia spectroscopy system with a 100× objective lens using a laser source with 488 nm wavelength. The laser was focused onto the sample with a spot diameter of 0.5 μm and a power of 0.5 mW. Angle-dependent data was collected by mounting samples on a rotatory stage and Raman was taken every 15 degree of rotation. The microabsorption spectrum were measured under a UV-Enhanced Aluminum coating 15x reflective objective lens using the halogen lamp only of the ocean optics DH2000 as radiation source with a power of 20 W. The data were collected using an Acton 300i spectrograph and a back thinned Princeton Instruments liquid nitrogen cooled CCD detector. AFM characterization: CVD grown ReS2 samples were scanned by Dimension Multimode 8 under tapping mode. The scanning rate was set to 0.5 Hz at the resolution of 512 x 512.

Associated Content Supporting information on growth details and parameters, angle resolved Raman spectroscopy and absorption spectroscopy for determining crystalline axis is available free of charge via the Internet at http://pubs.acs.org

Author information Corresponding author email: [email protected]

Acknowledgments S.T. acknowledges funding from National Science Foundation (DMR-1552220) and National Science Foundation (CMMI-1561839). G.W, X.M, and B.U acknowledge funding from ERC Grant No. 306719, ANR MoS2ValleyControl. Note: The authors declare no competing financial interest."

References 1. Tongay, S.; Sahin, H.; Ko, C.; Luce, A.; Fan, W.; Liu, K.; Zhou, J.; Huang, Y. S.; Ho, C. H.; Yan, J.; Ogletree, D. F.; Aloni, S.; Ji, J.; Li, S.; Li, J.; Peeters, F. M.; Wu, J. Nat Commun 2014, 5, 3252. 2. Chenet, D. A.; Aslan, O. B.; Huang, P. Y.; Fan, C.; van der Zande, A. M.; Heinz, T. F.; Hone, J. C. Nano Letters 2015, 15, (9), 5667-5672. 3. Hart, L.; Dale, S.; Hoye, S.; Webb, J. L.; Wolverson, D. Nano Letters 2016, 16, (2), 1381-1386. 10 ACS Paragon Plus Environment

Page 11 of 12

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Nano Letters

4. He, R.; Yan, J.-A.; Yin, Z.; Ye, Z.; Ye, G.; Cheng, J.; Li, J.; Lui, C. H. Nano Letters 2016, 16, (2), 14041409. 5. Zhang, E.; Jin, Y.; Yuan, X.; Wang, W.; Zhang, C.; Tang, L.; Liu, S.; Zhou, P.; Hu, W.; Xiu, F. Advanced Functional Materials 2015, 25, (26), 4076-4082. 6. Yang, S.; Tongay, S.; Li, Y.; Yue, Q.; Xia, J.-B.; Li, S.-S.; Li, J.; Wei, S.-H. Nanoscale 2014, 6, (13), 7226-7231. 7. Yang, S.; Tongay, S.; Yue, Q.; Li, Y.; Li, B.; Lu, F. Scientific Reports 2014, 4, 5442. 8. Liu, F.; Zheng, S.; He, X.; Chaturvedi, A.; He, J.; Chow, W. L.; Mion, T. R.; Wang, X.; Zhou, J.; Fu, Q.; Fan, H. J.; Tay, B. K.; Song, L.; He, R.-H.; Kloc, C.; Ajayan, P. M.; Liu, Z. Advanced Functional Materials 2016, 26, (8), 1169-1177. 9. Shim, J.; Oh, A.; Kang, D.-H.; Oh, S.; Jang, S. K.; Jeon, J.; Jeon, M. H.; Kim, M.; Choi, C.; Lee, J.; Lee, S.; Yeom, G. Y.; Song, Y. J.; Park, J.-H. Advanced Materials 2016, 201601002. 10. Yang, S.; Wang, C.; Sahin, H.; Chen, H.; Li, Y.; Li, S. S.; Suslu, A.; Peeters, F. M.; Liu, Q.; Li, J.; Tongay, S. Nano Lett 2015, 15, (3), 1660-6. 11. Horzum, S.; Cakir, D.; Suh, J.; Tongay, S.; Huang, Y. S.; Ho, C. H.; Wu, J.; Sahin, H.; Peeters, F. M. Physical Review B 2014, 89, (15) 155433. 12. Gao, J.; Li, L.; Tan, J.; Sun, H.; Li, B.; Idrobo, J. C.; Singh, C. V.; Lu, T.-M.; Koratkar, N. Nano Letters 2016, 16, (6), 3780-3787. 13. Zhang, Q.; Tan, S.; Mendes, R. G.; Sun, Z.; Chen, Y.; Kong, X.; Xue, Y.; Rümmeli, M. H.; Wu, X.; Chen, S.; Fu, L. Advanced Materials 2016, 28, (13), 2616-2623. 14. Keyshar, K.; Gong, Y.; Ye, G.; Brunetto, G.; Zhou, W.; Cole, D. P.; Hackenberg, K.; He, Y.; Machado, L.; Kabbani, M.; Hart, A. H. C.; Li, B.; Galvao, D. S.; George, A.; Vajtai, R.; Tiwary, C. S.; Ajayan, P. M. Advanced Materials 2015, 27, (31), 4640-4648. 15. Glemser, O. Academic Press, New York. 1963, 1, 1476-85. 16. Najmaei, S.; Liu, Z.; Zhou, W.; Zou, X.; Shi, G.; Lei, S.; Yakobson, B. I.; Idrobo, J.-C.; Ajayan, P. M.; Lou, J. Nat Mater 2013, 12, (8), 754-759. 17. Gao, J.; Yip, J.; Zhao, J.; Yakobson, B. I.; Ding, F. Journal of the American Chemical Society 2011, 133, (13), 5009-5015. 18. Liu, Y.; Zou, X.; Yakobson, B. I. ACS Nano 2012, 6, (8), 7053-7058. 19. Gibb, A. L.; Alem, N.; Chen, J.-H.; Erickson, K. J.; Ciston, J.; Gautam, A.; Linck, M.; Zettl, A. Journal of the American Chemical Society 2013, 135, (18), 6758-6761.

11 ACS Paragon Plus Environment

Nano Letters

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42

0o

60o

120o B

F

Page 12 of 12

A

C E

D B

F

A

C D

E

180o

240o ACS Paragon Plus Environment

300o