Letter pubs.acs.org/NanoLett
In Situ Observation on Dislocation-Controlled Sublimation of Mg Nanoparticles Qian Yu,*,† Min-Min Mao,† Qing-Jie Li,‡ Xiao-Qian Fu,† He Tian,† Ji-Xue Li,† Scott X. Mao,§,† and Ze Zhang† †
Center of Electron Microscopy and State Key Laboratory of Silicon Materials, Department of Materials Science and Engineering, Zhejiang University, Hangzhou, China, 310027 ‡ Department of Materials Science and Engineering, Johns Hopkins University, Baltimore, Maryland 21218, United States § Department of Mechanical Engineering and Materials Science, University of Pittsburgh, Pittsburgh, Pennsylvania 15261,United States S Supporting Information *
ABSTRACT: Sublimation is an important endothermic phase transition in which the atoms break away from their neighbors in the crystal lattice and are removed into the gas phase. Such debonding process may be significantly influenced by dislocations, the crystal defect that changes the bonding environment of local atoms. By performing systematic defects characterization and in situ transmission electron microscopy (TEM) tests on a core−shell MgO−Mg system, which enables us to “modulate” the internal dislocation density, we investigated the role of dislocations on materials’ sublimation with particular focus on the sublimation kinetics and mechanism. It was observed that the sublimation rate increases significantly with dislocation density. As the density of screw dislocations is high, the intersection of screw dislocation spirals creates a large number of monatomic ledges, resulting in a “liquid-like” motion of solid−gas interface, which significantly deviates from the theoretically predicted sublimation plane. Our calculation based on density functional theory demonstrated that the remarkable change of sublimation rate with dislocation density is due to the dramatic reduction in binding energy of the monatomic ledges. This study provides direct observation to improve our understanding on this fundamental phase transition as well as to shed light on tuning materials’ sublimation by “engineering” dislocation density in applications. KEYWORDS: Dislocation density, sublimation rate, “liquid-like” motion, monatomic ledge, Mg nanoparticles
D
information on the kinetics of vaporization in real materials where defects significantly matter is limited due to the difficulties in experimental methods.15−17 Here, by performing systematic defects characterization and in situ transmission electron microscopy (TEM) tests on a core−shell MgO−Mg system, which enables us to “modulate” the internal dislocation density, we investigated the influence of dislocations on the sublimation of materials in terms of sublimation temperature, sublimation rate, and kinetic process. Mg nanoparticles were produced by a direct current (dc) arc plasma method.18 All the nanoparticles were single crystalline. The size of the particles is ranging from hundreds of nanometers to several microns in diameter. As shown in Figure 1a, the smaller nanoparticles displayed a threedimensional (3D) faceted structure with 6-fold-symmetry. The morphology of the nanoparticles was analyzed in previous study; the surface planes were {0001}, {1010̅ }, and {1011̅ },
efects have great influence on the mechanical and functional properties of materials.1−3 As one of the most common types of defect, dislocations not only help to accommodate the shape changes of crystals by tuning the debonding and rebonding process but also influence electron or photon transportation, resulting in significant change of electrical or photonic property of materials.4,5 Dislocations also play an important role in endothermic phase transition of materials, which thereby affects the physical stability of material itself.6 For instance, the melting process of solid has been reported to be highly related to the dislocation structure in the materials.7,8 Sublimation is another important endothermic phase transition in which the atoms break away from their neighbors in the crystal lattice and are removed into the gas phase. Understanding the sublimation mechanism may have great influence not only from the scientific side of view but also in many technological applications including purification,9−11 thin film deposition12,13 and patterning of MEMS device.14 However, most of the previous studies on sublimation focused on the equilibrium thermodynamic parameters (vapor pressure, enthalpy, free energy) in the vaporization reaction, the © 2016 American Chemical Society
Received: November 1, 2015 Revised: December 18, 2015 Published: January 22, 2016 1156
DOI: 10.1021/acs.nanolett.5b04439 Nano Lett. 2016, 16, 1156−1160
Letter
Nano Letters
Figure 1. (a) The SEM images of a 200 and a 400 nm Mg particle. (b) EELS mapping of an Mg nanoparticle. The thickness of the MgO shell is about 2 nm. (c) The HAADF-STEM image of the Mg−MgO interface (upper) and its corresponding inverse-FFT image (lower). Interface and the interface dislocations are marked. (d) A bright-field TEM image of a Mg nanoparticle (upper). g = [21̅1̅0]. Screw dislocations are visible. The lower image is captured from its corresponding tomogram, showing the 3D dislocation structure in this nanoparticle.
thermodynamically favored once the temperature goes beyond this critical value during in situ TEM heating experiments. For repeatability, in situ TEM heating experiments were carried out in several microscopes (Hitach 9500, JEOL 3010, FEI F20) with Protochips and Gatan heating holders, respectively. By using the MgO−Mg core−shell structure, we designed the following three series of experiments to investigate the impact of dislocation density on sublimation mechanism: (1) direct rapid heating with heating rate at 10−20 °C/second; (2) slow heating with heating rate at 0.5 °C/second; and (3) annealing at a medium temperature (∼300 °C) for 2 h followed by a slow heating with heating rate at 0.5 °C/second. The particles with similar sizes were used for comparison. In the first case, the significant difference in thermal expansion of Mg and MgO at elevated temperature would result in huge stress at the interface (the finite element calculation shows that the stress can be as high as several GPa; detailed description is shown in Supporting Information), generating an ultrahigh density of dislocations, through dislocation multiplication and nucleation, which unfortunately have no time to move and escape. In contrast, with lower heating rate the second case offers time for dislocations to move around and possibly annihilate at interface or surface, reducing the internal stress. Thereby, dislocation multiplication and nucleation should be less significant and a medium value of internal dislocation density is expected. Finally, the annealing process in the third case is designed to significantly reduce the dislocation density to an extremely low level. The results show that the direct rapid heating in the first case resulted in an instantaneous sublimation in which the Mg core evaporates within extremely short time (less than a second). It is noticed that such rapid sublimation occurred at about 320− 350 °C in almost all the particles regardless of their particle sizes. One typical in situ TEM movie of case 1 was shown in Supporting Information as Movie 1. The appearance and movement of strong strain contours in the TEM images shown in Figure 2 captured from Movie 1 were observed right before the instantaneous sublimation, indicating the generation of ultrahigh density of dislocations (the particle largely remained crystalline before sublimation, see Supporting Information for the analysis). The TEM analysis demonstrated that only an
respectively. Differently, the larger nanoparticles tend to be spherical, as shown in the lower SEM image in Figure 1a. Because of the high reactivity of magnesium with oxygen, a thin layer of Mg oxide forms on the surface of the single crystal Mg nanoparticles once the particles are exposed to air, building a core−shell structure. The chemical analysis and the atomic structure investigation of the core−shell interface were performed in an aberration-corrected Titan microscopy and the double aberration-corrected TEAM1 microscopy, respectively. Electron energy loss spectroscopy (EELS) mapping displays the oxygen enriched core with the average thickness at about 2 nm (shown in Figure 1b). Figure 1c shows the highangle annular dark-field scanning transmission electron microscope (HAADF-STEM) image and the inverse fast Fourier transform (IFFT) image in which the noise and high frequencies were filtered out because only the first order diffractions were used to form the image. It is demonstrated that (1) the MgO shell is epitaxial and is in the form of monocrystalline and (2) the epitaxial MgO-Mg interface is decorated with array of dislocations, indicating the possibility of having ultrahigh internal pressure upon heating because of the interfacial incoherency. It is also noticed that the Mg core contains quite high density of pre-existing dislocations with an estimated value at about 1014−1015/m2. The upper TEM image in Figure 1d shows the typical dislocation structure in the nanoparticles viewed by using g = [21̅1̅0]. High density of screw dislocations was observed. The three-dimensional dislocation structure in this nanoparticle was further studied by performing dislocation tomography in a JEOL 3010 microscopy. According to the tomogram, which is attached in Supporting Information, it is observed that the screw dislocations were ⟨a⟩-type lying on basal planes; they were almost parallel to each other, as shown in the lower image in Figure 1d. One typical tomogram was shown in Supporting Information. It has been calculated that the increase of vapor pressure of Mg at elevated temperature is significant. On the basis of the evaluation from the equation log Ptorr = 8.6047−7560.3/T (223−385 °C) (1 Torr = 133 Pa),19 the vapor pressure of Mg would exceed the pressure in TEM column (10−6−10−9 Torr) at about 245 °C; therefore, the sublimation of Mg would be 1157
DOI: 10.1021/acs.nanolett.5b04439 Nano Lett. 2016, 16, 1156−1160
Letter
Nano Letters
though it was confirmed by diffraction analysis that the materials remained in the middle of the sublimation process was still crystalline Mg, the dynamic sublimation process showed a “liquid-like” activity of the solid−gas interface, that is, the sublimation surface does not reside on a specific welldefined crystal plane but constantly undulates with some degrees of freedom, as shown in Movie 2. This is obviously different from the prediction of theory and recent experiments, which illustrates that sublimation usually happens along a welldefined crystal plane.20−22 When we further reduce the dislocation density by annealing, observable sublimation appeared at about 480−520 °C in different particles. The sublimation rate was lower comparing with that in the second case; a sublimation process was completed in several to 10 minutes. In strong contrast to the “liquid-like dancing” of the solid−gas interface in case 2, sublimation in this case proceeded along well-defined crystal planes as predicted by the theory. A typical movie is shown in Supporting Information as Movie 3. Figure 4a shows the sublimation of a particle in the third case. The particle was viewed along [011̅0] beam direction and has a size similar to the one in Figure 3a. The diffraction contrast seen as black lines here indicates the existence of straight screw dislocations, which was marked by the green dashed lines. It was observed that the two-dimensional projection of the sublimation surface is always flat and is aligned with the basal plane. As shown in the series of TEM images in Figure 4a, there were steps on the solid−gas interface moving in a “spiral-like” form around the dislocation lines, which was indicated by the red dash lines. The moving direction of the interface was along [0001] c-axis. Figure 4b shows the dynamic sublimation process of a larger particle viewed from [211̅ 0̅ ] direction with 500 nm in diameter. Consistent with that in smaller particles, the solid−gas interfaces were also faceted with sharp angles. The interface moved in tiny steps. The moving direction of the interface was parallel to c-axis until the very end where the sublimation happened along multiple facets. We also performed rapid heating in these annealed particles. The results show that the sublimation process and temperature is similar comparing with the case above. Figure S3 in Supporting Information shows one TEM image of a clean particle with flat sublimation surface
Figure 2. Series of TEM images captured from a movie for case 1. The rapid heating here results in instantaneous sublimation. At t = 1.54 and 1.59 s, significant strain contrast was generated presumably due to the buildup of ultrahigh dislocation density.
empty MgO shell was left behind after sublimation, which is in a nanocrystalline structure showing a ring pattern in diffraction. However, the sublimation kinetics can be slowed down by tuning down the dislocation density. In case 2 where the value of dislocation density is estimated to be medium, the sublimation event proceeded gradually thereby the details of the dynamic sublimation process were obtained. In general, sublimation completed in 20−30 s in all the particles. Figure 3a,b shows the details of the sublimation process in the nanoparticles with 200 and 500 nm in diameter, respectively. The smaller particle was viewed along [011̅0] beam direction and the larger one was viewed along [211̅ 0̅ ] beam direction. The corresponding movies are attached in Supporting Information. According to the in situ TEM observation, observable sublimation started at about 380−400 °C. Even
Figure 3. (a) The series of TEM images from a movie of case 2, showing the sublimation process of a 200 nm sized Mg nanoparticle. The diffraction pattern at right was taken during sublimation, indicating that the Mg remains crystallized in this process. g = [011̅0]. (b) The series of TEM images from a movie of case 2, showing the sublimation process of a 400 nm sized Mg nanoparticle. The diffraction pattern at right was taken during sublimation, indicating that the Mg remains crystallized in this process. g = [21̅1̅0]. The sublimation process shows a “liquid-like motion” of gas− solid interface in both particles. 1158
DOI: 10.1021/acs.nanolett.5b04439 Nano Lett. 2016, 16, 1156−1160
Letter
Nano Letters
Figure 4. (a) The series of TEM images from a movie of case 3, showing the sublimation process of a 200 nm sized Mg nanoparticle. The diffraction pattern at right was taken during sublimation, indicating that the Mg remains crystallized in this process. g = [011̅0]. b) The series of TEM images from a movie of case 3, showing the sublimation process of a 400 nm sized Mg nanoparticle. The diffraction pattern at right was taken during sublimation, indicating that the Mg remains crystallized in this process. g = [21̅1̅0]. Sublimation was preceded along c-axis in both. Gas−solid interface is along basal plane.
observed after rapid heating. The heating rate independence observed in these annealed particles with few or no pre-existing screw dislocations indicates that the sublimation is less relevant to the interfacial dislocations generated during heating but is indeed controlled by the density of screw dislocations. The main results of the above three heating experiments are summarized in Table S2 of Supporting Information. By investigating the change of sublimation phenomena with dislocation densities, it is clearly shown that dislocations, specifically screw type, play the very important role in sublimation of material. As an inverse process, it has been previously reported that the screw dislocation core could strongly affect the deposition of atoms; thereby they could act as pole/tunnel for crystal growth.23−25 The impact of screw dislocations on sublimation of metals has been also previously discussed by Hirth et al.26 It has been proposed that the α Ρe sublimation rate J is determined from J = 1/2 , where Pe,
Figure 5. (a) Schematic of the dislocation-controlled sublimation mechanism. The intersection of screw dislocation spirals produces monatomic ledges for sublimation. (b) Configurations of smooth surface and surface with a ledge. Left: Top view and side view of a (0001) smooth surface. The per atom binding energy of such smooth surface is −1.51 eV. Right: Top view and side view of a (0001) surface with a ledge. The per atom binding energy of this monatomic ledge is −0.64 eV. Atoms are colored according to their coordination numbers.
(2πmkT )
T, m, k, and α are the vapor pressure of the metal, temperature, atomic mass, Boltzmann’s constant, and the vaporization coefficient that may have values from 10−6 to 1. Generally speaking, the value of α reaches the maximum at the monatomic ledges. The sources of monatomic ledges include crystal edges or grain boundaries, pores, cracks, screw dislocations and macroscopic ledges. However, it is worth noting that because high α is only limited to a small area at the center of the spiral, screw dislocations have to be present in a high density in order to act as effective source of monatomic ledges and result in a globally high α. As illustrated in Figure 5a, when dislocation density is high, the spiral areas of neighboring dislocations may overlap, producing numbers of ready sources for monatomic ledges in sublimation. In order to confirm the above analysis and estimate the influence of surface monatomic ledges produced by the screw dislocations on the kinetics of sublimation, we calculated the binding energies of a smooth (0001) surface and a monatomic surface ledge based on density functional theory (DFT). As shown in Figure 5b, our calculations were performed using a slab configuration consisting of 14 layers with and without a monatomic surface ledge (details of calculations can be found in Supporting Information). The binding energy per atom of a smooth surface is defined as
BEsurf =
1 (En − En − 1 − mEatom) m
where m is the number of atoms on a smooth surface, En is the total energy of slab consisting of n layers, En−1 is the total energy of slab with n − 1 layers and Eatom is the energy of a free atom. BEsurf can be considered as the energy per atom required to simultaneously remove a single surface layer of atoms into free atoms. When n → ∞, BEsurf → Ecoh, the cohesive energy of bulk Mg. Thus, we calculated the cohesive energy as the binding energy per atom of a smooth surface. Similarly, the binding energy per atom of a monatomic surface ledge is defined as BE ledge =
1 L (En − En − lEatom) l
where l is the number of atoms on a monatomic surface ledge, ELn is the total energy of a n-layer slab with a monatomic surface ledge. Similarly, BEledge can be viewed as the energy per atom required to simultaneously remove a ledge of atoms into free atoms. According to the above definitions, the binding energies were calculated to be BEsurf = −1.51 eV and BEledge = −0.64 eV, respectively. As can be seen, BEledge is 2.4 times smaller than 1159
DOI: 10.1021/acs.nanolett.5b04439 Nano Lett. 2016, 16, 1156−1160
Letter
Nano Letters
(4) Nakamura, S. Science 1998, 281 (5379), 956−961. (5) Szot, K.; Speier, W.; Bihlmayer, G.; Waser, R. Nat. Mater. 2006, 5 (4), 312−320. (6) Driver, J. H.; Papazian, J. M. Mater. Sci. Eng. 1985, 76 (0), 51−56. (7) Kierfeld, J.; Vinokur, V. Phys. Rev. B: Condens. Matter Mater. Phys. 2000, 61 (22), R14928−R14931. (8) Nelson, D. R.; Halperin, B. I. Phys. Rev. B: Condens. Matter Mater. Phys. 1979, 19 (5), 2457−2484. (9) Montag, B. W.; Reichenberger, M. A.; Edwards, N.; Ugorowski, P. B.; Sunder, M.; Weeks, J.; McGregor, D. S. J. Cryst. Growth 2015, 419 (0), 133−137. (10) Montag, B. W.; Reichenberger, M. A.; Sunder, M.; Ugorowski, P. B.; Nelson, K. A.; McGregor, D. S. J. Cryst. Growth 2015, 419 (0), 143−148. (11) Fitzner, R.; Reinold, E.; Mishra, A.; Mena-Osteritz, E.; Ziehlke, H.; Korner, C.; Leo, K.; Riede, M.; Weil, M.; Tsaryova, O.; Weiss, A.; Uhrich, C.; Pfeiffer, M.; Bauerle, P. Adv. Funct. Mater. 2011, 21 (5), 897−910. (12) Oliva, A. I.; Castro-Rodriguez, R.; Solis-Canto, O.; Sosa, V.; Quintana, P.; Pena, J. L. Appl. Surf. Sci. 2003, 205 (1−4), 56−64. (13) Pauleau, Y.; Fasasi, A. Y. Chem. Mater. 1991, 3 (1), 45−50. (14) Knieling, T.; Lang, W.; Benecke, W. Sens. Actuators, B 2007, 126 (1), 13−17. (15) Couchman, P. R.; Jesser, W. A. Nature 1977, 269 (5628), 481− 483. (16) Pitchimani, R.; Burnham, A. K.; Weeks, B. L. J. Phys. Chem. B 2007, 111 (31), 9182−9185. (17) Schultz, R. D.; Dekker, A. O. J. Chem. Phys. 1955, 23 (11), 2133−2138. (18) Chen, J. H.; Lu, G. H.; Zhu, L. Y.; Flagan, R. C. J. Nanopart. Res. 2007, 9 (2), 203−213. (19) Gilbreath, W. P. The Vapor Pressure of Magnesium between 223 and 385 °C; NASA Technical Note NASA TN D-2723; NASA: Washington, DC, 1965. (20) Somorjai, G. A. Science 1968, 162 (3855), 755−760. (21) Hellebusch, D. J.; Manthiram, K.; Beberwyck, B. J.; Alivisatos, A. P. J. Phys. Chem. Lett. 2015, 6 (4), 605−611. (22) van Huis, M. A.; Young, N. P.; Pandraud, G.; Creemer, J. F.; Vanmaekelbergh, D.; Kirkland, A. I.; Zandbergen, H. W. Adv. Mater. 2009, 21 (48), 4992−4995. (23) Li, Y. G.; Wu, Y. Y. Chem. Mater. 2010, 22 (19), 5537−5542. (24) Morin, S. A.; Bierman, M. J.; Tong, J.; Jin, S. Science 2010, 328 (5977), 476−480. (25) Morin, S. A.; Forticaux, A.; Bierman, M. J.; Jin, S. Nano Lett. 2011, 11 (10), 4449−4455. (26) Hirth, J. P.; Pound, G. M. J. Chem. Phys. 1957, 26 (5), 1216− 1224.
BEsurf, consistent with the observation that rapid heating requires the lowest sublimation temperature as fast thermal expansion created a high density of dislocations, thus more surface ledges. It should be noted that if the stress field of the screw dislocations is considered, the binding energy for the monatomic surface ledge might be even lower. Comparing to that in crystal growth, the role of screw dislocation in sublimation is actually quite different. In crystal growth, screw dislocation spiral directly offers tunnel for atoms’ deposition. Here it is suggested that the monatomic ledges generated by the intersections of screw dislocation spirals contribute to the sublimation, thereby the kinetics of sublimation (e.g., sublimation temperature, rate etc.) is a function of dislocation density. On the basis of the dislocationcontrolled sublimation mechanism, one may be able to tune the sublimation rate by engineering the material with certain dislocation density. This in situ study provides direct observation on sublimation process and systematically stated the change of sublimation behavior with varying dislocation densities, which to our knowledge has not been achieved before, shedding light on our understanding on this fundamental phase transition and its related applications.
■
ASSOCIATED CONTENT
S Supporting Information *
The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.nanolett.5b04439. Analysis on the interfacial stress, DFT calculation method, and table for the types of experiments. (PDF) Tomogram of a nanopartical with high screw dislocation density; sublimation of a 200 nm sized particle with direct rapid heating. (AVI) Tomogram of a nanopartical with high screw dislocation density; sublimation of a 200 nm sized particle with slow heating.(AVI) Tomogram of a nanopartical with high screw dislocation density; sublimation of a 200 nm sized particle with annealing and slow heating.(AVI)
■
AUTHOR INFORMATION
Corresponding Author
*E-mail:
[email protected]. Notes
za
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS This work was supported in China by grants from the Chinese 1000-Youth-Talent Plan and the 973 Program of China No. 2015CB65930. We also thank Penghan Lu from CAMPnano, Xi’an Jiaotong University, China for assistance with the first experiment. Ze Zhang’s work was supported by grant Nos. 11234011 and 11327901.
■
REFERENCES
(1) Banhart, F.; Kotakoski, J.; Krasheninnikov, A. V. ACS Nano 2011, 5 (1), 26−41. (2) Meyers, M. A.; Mishra, A.; Benson, D. J. Prog. Mater. Sci. 2006, 51 (4), 427−556. (3) Uberuaga, B. P.; Andersson, D. A.; Stanek, C. R. Curr. Opin. Solid State Mater. Sci. 2013, 17 (6), 249−256. 1160
DOI: 10.1021/acs.nanolett.5b04439 Nano Lett. 2016, 16, 1156−1160
