Evidence for Active Atomic Defects in Monolayer Hexagonal Boron

Jan 27, 2014 - ABSTRACT: We report the formation and motion of 4|8 (square- octagon) defects in monolayer hexagonal boron nitride (h-BN). The 4|8 defe...
0 downloads 0 Views 3MB Size
Letter pubs.acs.org/NanoLett

Evidence for Active Atomic Defects in Monolayer Hexagonal Boron Nitride: A New Mechanism of Plasticity in Two-Dimensional Materials Ovidiu Cretu,* Yung-Chang Lin, and Kazutomo Suenaga Nanotube Research Center, National Institute of Advanced Industrial Science and Technology (AIST), Central 5, 1-1-1 Higashi, Tsukuba, Ibaraki 305-8565, Japan S Supporting Information *

ABSTRACT: We report the formation and motion of 4|8 (squareoctagon) defects in monolayer hexagonal boron nitride (h-BN). The 4|8 defects, involving less-favorable B−B and N−N bonds, are mobile within the monolayer at high sample temperature (∼1000 K) under electron beam irradiation. Gliding of one or two atomic rows along the armchair direction is suggested to be the origin of the defect motion. This represents a completely new mechanism of plasticity in two-dimensional materials. KEYWORDS: Transmission electron microscopy, in situ, boron nitride, defect, dislocation, plasticity

L

graphene layer, in which the Stone-Wales transformation is responsible for the plastic deformation, the deformation of hBN seems to rely on atomic row gliding along the armchair direction, which does not necessarily increase the number of B−B and N−N bonds. The h-BN single crystals were mechanically exfoliated using Scotch tape and transferred to silicon substrates with a 300 nm layer of thermal oxide. Single and few-layer flakes were then transferred to Quantifoil Mo TEM grids.9 Electron irradiation and high-resolution TEM imaging were performed in a modified Jeol JEM-2100F microscope, fitted with a cold field-emission gun and dodecapole-based aberration correctors.10 The microscope was operated at 60 kV, while the samples were held at 1000 K throughout the experiments by using a heating specimen holder. Few-layer (1−3) h-BN regions were initially irradiated with a beam current density of 2 × 107 e/nm2s (corresponding to the focused electron beam) for time intervals of the order of a minute. This led to the sputtering of thicker parts of the sample, as well as to the formation of 4|8 defects in monolayer regions. The current density was subsequently reduced to 106 e/nm2s for imaging. Under these conditions, the 4|8 defects were stable for several minutes. HR-TEM images have been Fourier-filtered using a procedure which is detailed in Supporting Information Figure S1. This procedure has been used for all images shown in this paper, unless explicitly mentioned otherwise. Molecular modeling for the various structures was done within the Accelrys Materials Studio suite. Geometry optimizations were performed using the DMol3 package.11,12 Multislice TEM image simulations were computed based on the optimized models using the QSTEM software.13

ow-dimensional materials are recently attracting a lot of attention from wide research fields because of their properties, which are different from those of the bulk. Mechanical behavior is an important issue in low-dimensional materials both from a fundamental physics point of view as well as in view of their future applications. For instance, large deformations in carbon nanotubes were observed by transmission electron microscopy (TEM),1 even though plastic deformation has hardly ever been seen in bulk graphite. In situ atomic resolution TEM has revealed that topological defects are indeed mobile during the deformation process of carbon nanotubes and are responsible for their plastic deformation.2 These activated defects are thought to originate from the 90° rotation of a C−C bond, called a Stone-Wales transformation. There has been little work done on atomic-scale defects in hBN. Simple vacancies3,4 and more recently 5|7-membered boundaries5 have been reported experimentally, while structures containing 4|8-membered rings involving only heteroelemental B−N bonds have been predicted by theory.6 No experimental evidence has been so far reported for activated defect structures in h-BN,7 which could be responsible for the plastic deformation of the h-BN layer. The lack of evidence for plasticity is mainly due to the difference in bond nature between h-BN and graphene. The ionic character of the B−N bond will not allow bond switching as easily as in the case of C−C bonds. The Stone-Wales deformation in h-BN, although predicted in the case of nanotubes,8 is believed to be highly unlikely because it inevitably induces a pair of B−B and N−N bonds, which are energetically unfavorable. Another deformation mechanism is therefore needed in order to explain the plasticity of h-BN. Here we show the first experimental evidence for activated atomic defects in single-layer h-BN at elevated temperatures under e-beam irradiation. Surprisingly, the 4|8 defects are stabilized with homoelemental B−B and N−N bonds and are indeed mobile in the h-BN lattice. Unlike the case of a © 2014 American Chemical Society

Received: December 20, 2013 Revised: January 13, 2014 Published: January 27, 2014 1064

dx.doi.org/10.1021/nl404735w | Nano Lett. 2014, 14, 1064−1068

Nano Letters

Letter

Figure 1. HR-TEM images of 4|8 line defects. (a) A defect along the armchair direction in a single layered region. (b) An enlarged version of the highlighted area in (a). (c) Atomic model of the “symmetric” case, as indicated by two yellow triangles (left) and corresponding simulated image (right). (See Supporting Information Figure S2 for the “asymmetric” case.) (d) Another HR-TEM image of a 4|8 defect showing its end. (e) Structural model for (e), containing a pentagon. Boron and nitrogen atoms are colored red and blue, respectively. Scale bar is 1 nm.

Figure 2. Experimental evidence of 4|8 defect motion. Panels (a) and (b) show series of frames highlighting trajectories for different 4|8 defects. The lower panels show models for each diffusion step. Boron and nitrogen atoms are colored red and blue, respectively. See text for details on each of the cases. Scale bar is 1 nm. See also Figure 3 for more details.

The less common case of a 4|8 defect whose end section is visible in the center of the image is shown in Figure 1d. Figure 1e shows the corresponding structural model, containing a pentagon termination. Graphene analogues for this termination have been observed in refs 14 and 15. The asymmetric case would require a series of homoelemental bonds at the intersection between the two rotated domains in the part that is not reconstructed as a 4|8 structure. This represents further proof that the defect forms in the same h-BN domain, as outlined by the three yellow triangles in Figure 1e. The atomic-scale formation process of the 4|8 defect is experimentally unclear, as it shows up instantaneously in our TEM images. No intermediate states of its formation process could be observed. We infer three possible mechanisms to form these defects. The first mechanism is the coalescence of several point defects. On the basis of previous reports for graphene,15 the most probable path to create line-defects is through the formation and diffusion of several smaller reconstructed vacancies. However, this is a process that takes place primarily through bond-rotations that are unlikely to occur for B−N bonds. These smaller defects were not observed during our experiments, although one cannot rule out their rapid coalescence because of the high sample temperature. The second possible mechanism is the preferential sputtering of

The 4|8 defects were always observed along the armchair direction of single-layered regions of h-BN during TEM observation at 700 °C (Figure 1a). An enlarged HR-TEM image of the highlighted area is shown in Figure 1b, where one can see alternating smaller and larger rings, corresponding to the four and eight membered polygons respectively. There are two possibilities for the atomic arrangement of boron and nitrogen in the 4|8 defects, namely “symmetric” and “asymmetric”. One cannot identify the two different atomic structures directly from the HR-TEM image, which in this case cannot discriminate between B and N atoms. Previous theoretical work6 has predicted the asymmetric 4|8 defects, which can only exist between two domains with 60° rotation (or mirror symmetry). In this asymmetric case, all the squares and octagons contain exclusively B−N bonds, as shown in Supporting Information Figure S2. The 4|8 structure found in our experiments can only be explained by the symmetric configuration (Figure 1c), involving B−B and N−N homoelemental bonding for the squares and octagons. As heteroelemental B−N bonds are energetically favored, the asymmetric case may sound more intuitive.6 This asymmetric structure is however unlikely, because it can only exist between two different h-BN domains with a 60° rotation between them and thus cannot form in a single h-BN domain. 1065

dx.doi.org/10.1021/nl404735w | Nano Lett. 2014, 14, 1064−1068

Nano Letters

Letter

Figure 3. The 4|8 defect motion through atomic-row glides. The three types of motion observed are presented in the three columns. Panels (a−c) show in each case the initial structure, on which the atoms which will move are indicated by a green arrow. Panels (d−f) show the optimized structures following a movement in each of the cases. The left parts of the initial and final structures are aligned. Dimers highlighted in yellow and purple serve as a guide for the eye. Boron and nitrogen atoms are colored red and blue, respectively. Scale bar is 0.3 nm.

Its first octagon (arrowed) initially moves up (perpendicularly) and joins the fixed right-side defect, after which it disappears through reconstruction into smaller rings. A dynamic version of this process is shown in Movie2. Interpreting the transformations outlined here is relatively straightforward due to the availability of step-by-step images of the 4|8 defects. In a case where the 4|8 defect has already undergone several transformations or where fewer images are available, pinpointing the exact processes would be significantly more difficult. The diffusion of the 4|8 defects can be explained through the glide of a section of the defect along the armchair direction. A schematic of this process is given in Figure 3, where the three columns highlight the three different types of motion that we observe. Panels a−c in Figure 3 show in each case the initial structure. The atoms that will move are indicated by green arrows which show the direction of motion. Panels d−f in Figure 3 show the optimized structure after each of these three movements. In order to make viewing easier, the left parts of the initial and final structures are aligned. The same two dimers are highlighted in yellow and purple in order to help visualizing the transformation undergone by the right side of the structure. Gliding implies a collective translation of one or two atomic rows (highlighted for each case) along the direction indicated by the arrows. In reason of the symmetry of the system, this vector is equal to the unit vector along the armchair direction, aAC = 1.5d (where “d” is the B−N distance). Because of the way in which two of these translations bring the two resulting 4|8 defects closer together or farther apart, they are called “inward” and “outward”, respectively. The third one is called “perpendicular”, because of its orientation with respect to the 4|8 defect axis. Macroscopically, these motions are equivalent to gliding part of the h-BN lattice along the three nonequivalent zigzag directions, which are indicated in the lower part of Figure 3a by black arrows. This comparison is given for illustrative purposes only, as the mechanism that we describe is much more energetically favorable. Finally, in the case of the more complicated perpendicular motion, an alternative mechanism could be a series of 120° bond rotations of each B−N pair between the two rows of highlighted atoms in Figure 3c. Because the number of atoms that move is the same, we expect the energy costs between the two mechanisms (collective bond rotation and atomic glide) to be similar. As

atoms along the armchair direction, followed by reconstruction to stabilize the 4|8 defect. This however contradicts the threefold symmetry of the h-BN lattice, which leads to triangularshaped defects under electron-irradiation, as has been reported by TEM observations.3,4 The third mechanism is to form the 4| 8 linear defect by gliding one part of the crystalline domain along the armchair direction. This process has the advantage of explaining the instant appearance of the structures along the armchair direction, which is precisely what we observe. However, it is at the same time the most energetically expensive. Additionally, this process should be reversible, causing the disappearance of the 4|8 defect through another glide step. Furthermore, this process does not explain why we do not see several parallel defects at the same time, or why we sometimes see the defect ending in a pentagon (Figure 1d). Experimental evidence for the motion of these structures is presented in Figure 2, which shows groups of TEM images extracted from image series acquired of the 4|8 defects. Panel (a) in Figure 2 shows a process in which an initially straight defect undergoes a transformation. The defect is separated into two sections that are vertically offset, but appear horizontally closer together. This results in a structure that shows a small vertical step between its two parts but is otherwise still continuous. In rare cases (right-hand side frame in Figure 2a), a second step is seen to appear, corresponding to an additional transformation. Movie1 shows an accelerated sequence of frames corresponding to this process. Supporting Information Figure S3a shows a similarly transformed 4|8 defect regaining its initial configuration, showing that a continuous structure is in this case energetically favorable. Panel (b) in Figure 2 shows a 4|8 defect that undergoes a two-step process. From left to right, the initially continuous 4|8 defect separates into sections that are in this case offset in a direction perpendicular to the defect axis. In a following step, the two parts become further separated following a second transformation, which increases the distance between them both horizontally and vertically. Supporting Information Figure S3b shows a more complicated process involving the same defect. The initial configuration is that of a 4|8 defect that has been split in a perpendicular direction. In the following frames, the left-side defect undergoes a two-step motion, toward the lower-right and upper-right corners of the image respectively. 1066

dx.doi.org/10.1021/nl404735w | Nano Lett. 2014, 14, 1064−1068

Nano Letters

Letter

Figure 4. Structure and motion of two intersecting 4|8 defects. An image of the new structure created by the intersection is shown in (a). The TEM simulation and corresponding optimized model are shown in (b,c), respectively. A sequence of frames showing the motion of these 4|8 defects is displayed in (d1−d6). The directions of the two defects are indicated by yellow dotted lines. Black arrows indicate portions of the defect that reconstruct along the horizontal direction. Scale bars are 0.5 nm.

can identify the polarity of the defect and assign the doublebonded atom at the center corner of the heart defect as B. While this configuration has the disadvantage of an open bond, it produces at the same time a minimum amount of distortion to the surrounding lattice. The motion of this system of 4|8 defects is simplified due to the lower number of degrees of freedom compared to the previous cases. A sequence of frames showing the movement of this configuration is shown in Figure 4d1−6. Movie3 shows this transformation in more detail. The most important observation is that the motion is predominantly horizontal, with no vertical steps for the heart defect itself. This can be explained by considering that both individual 4|8 defects move along welldefined directions. Supporting Information Figure S4 details this mechanism; initially, the glide of a row of atoms belonging to the left-side defect causes it to shift upward (Supporting Information Figure S4a). This produces an intermediate step where the heart defect is enlarged and the two sides appear farther apart (Supporting Information Figure S4b). An equivalent glide from of a row of atoms belonging to the right-side defect restores the initial heart shape, which has been translated by one zigzag unit vector (√3d). The process does not require additional lattice reconstruction, which makes it energetically favorable and explains why it is seen several times throughout the imaging process. Finally, panels d3−4 show a case where a section of the rightside defect (indicated by a black arrow) reconstructs along the horizontal direction. This transformation produces polygons which appear to have a different structure than the 4|8 configuration. This is supported by the fact that it would be impossible for another 4|8 structure to form along this zigzag direction. Although the resolution is insufficient in order to directly identify the reconstruction, a possible candidate along with its formation mechanism is described in Supporting Information Figure S5. Here, two armchair row glides produce

such, it is impossible to discriminate between the two under the present experimental conditions. All of the transformations that have been observed can be explained by a combination of these three fundamental steps. The h-BN lattice accommodates these glides through the formation of pentagon and (less frequently) heptagon rings. While these provide a unique picture for each configuration, they can be difficult to resolve in the TEM images. In order to make identification easier, we classify each structure by introducing two parameters: the vertical distance between the two parts of the separated defect (dV) and the horizontal distance between the first octagons of each section (dH), as illustrated in Figure 3e. The values for these parameters, as derived from the optimized structures in Figure 3, are listed in Supporting Information Table S1. Measuring these parameters and comparing them with the calculated values allows for an unambiguous identification of the glide process. There is an obvious connection between these two values and the unit vector along the zigzag direction. In a first approximation, they can be calculated by aZZ sin(θ) and aZZ cos(θ), respectively, where aZZ = √3d and θ is the angle between this vector and the direction of the 4|8 defect. However, this does not take lattice relaxation into account and produces results that have large errors when measuring close to the reconstructed area. The final part of the results shows the case of two 4|8 defects that intersect each other and form a 120° angle between them. A detailed view of the intersection is shown in Figure 4a. Because of the symmetry of this system, the final octagons of each 4|8 defect merge together and form a decagon termination. A simulated TEM image based on the relaxed model is shown for comparison in Figure 4b, while the model itself is displayed in Figure 4c. We call this structure, which has not been previously predicted or observed, a heart defect. Because of the presence of the step-edge nearby as well as of a larger triangular hole visible in the full image (not shown), we 1067

dx.doi.org/10.1021/nl404735w | Nano Lett. 2014, 14, 1064−1068

Nano Letters

Letter

(11) Delley, B. J. Chem. Phys. 1990, 92, 508−517. (12) Delley, B. J. Chem. Phys. 2000, 113, 7756−7764. (13) Koch, C. T. Ph.D. Thesis, Arizona State University, Phoenix, Arizona, 2002 (14) Kotakoski, J.; Krasheninnikov, A. V.; Kaiser, U.; Meyer, J. C. Phys. Rev. Lett. 2011, 106, 105505. (15) Warner, J. H.; Lee, G.-D.; He, K.; Robertson, A. W.; Yoon, E.; Kirkland, A. I.. ACS Nano 2013, 7, 9860−9866.

a line of three heptagons that are similar to the experimental structure. Their short lifetime and subsequent reconstruction back to the 4|8 structure in the following frames show that in the current case they represent a less favorable configuration. In conclusion, the structure and dynamics of linear 4|8 defects in h-BN have been observed experimentally for the first time using HR-TEM. Because of their in situ formation from the pristine h-BN lattice, the structure of these 4|8 defects must contain B−B and N−N bonds and is therefore different from the previous cases predicted by theory. Their motion has been explained in terms of atomic row glides along the armchair direction, induced by the continuous electron irradiation and high temperature. The intersection of two defects produces a newly identified heart-shaped defect structure that stabilizes the system with minimal lattice deformation. This represents the first observation of plasticity in this material and opens the way for further studies involving it, as well as other two-dimensional analogues.



ASSOCIATED CONTENT

S Supporting Information *

Additional content includes movies showing the motion of the defects, experimental and simulated TEM images, values for glide parameters, and models related to the case of the two intersecting defects. This material is available free of charge via the Internet at http://pubs.acs.org.



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Author Contributions

The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS The authors would like to thank Dr. Masanori Koshino for his advice regarding the geometric optimization of the models. The work is supported by a JST Research Acceleration program.



REFERENCES

(1) Huang, J. Y.; Chen, S.; Wang, Z. Q.; Kempa, K.; Wang, Y. M.; Jo, S. H.; Chen, G.; Dresselhaus, M. S.; Ren, Z. F. Nature 2006, 439, 281. (2) Suenaga, K.; Wakabayashi, H.; Koshino, M.; Sato, Y.; Urita, K.; Iijima, S. Nat. Nanotechnol. 2007, 2, 358−360. (3) Jin, C.; Lin, F.; Suenaga, K.; Iijima, S. Phys. Rev. Lett. 2009, 102, 195505. (4) Meyer, J. C.; Chuvilin, A.; Algara-Siller, G.; Biskupek, J.; Kaiser, U. Nano Lett. 2009, 9, 2683−2689. (5) Gibb, A. L.; Alem, N.; Chen, J.-H.; Erickson, K. J.; Ciston, J.; Gautam, A.; Linck, M.; Zettl, A. J. Am. Chem. Soc. 2013, 135, 6758− 6761. (6) Liu, Y.; Zou, X.; Yakobson, B. I. ACS Nano 2012, 6, 7053−7058. (7) Zobelli, A.; Ewels, C. P.; Gloter, A.; Seifert, G. Phys. Rev. B 2007, 75, 094104. (8) Li, Y.; Zhou, Z.; Golberg, D.; Bando, Y.; von Ragué Schleyer, P.; Chen, Z. J. Phys. Chem. C 2008, 112, 1365−1370. (9) Pacilé, D.; Meyer, J. C.; Girit, Ç . Ö .; Zettl, A. Appl. Phys. Lett. 2008, 92, 133107. (10) Sasaki, T.; Sawada, H.; Hosokawa, F.; Kohno, Y.; Tomita, T.; Kaneyama, T.; Kondo, Y.; Kimoto, K.; Sato, Y.; Suenaga, K. J. Electron. Microsc. 2010, 59 (Suppl.), S7−S13. 1068

dx.doi.org/10.1021/nl404735w | Nano Lett. 2014, 14, 1064−1068