Journal of Crystal Growth 401 (2014) 511–513 Contents lists available at ScienceDirect Journal of Crystal Growth journal homepage: www.elsevier.com/locate/jcrysgro 4H-SiC surface structure transitions during crystal growth following bunching in a fast sublimation process Filip Krzyżewski n Institute of Physics Polish Academy of Sciences, Al. Lotników 32/46, 02-668 Warsaw, Poland art ic l e i nf o a b s t r a c t Available online 16 November 2013 Kinetic Monte Carlo wurtzite structure crystals of 4H-SiC simulations were performed. The first stage of system evolution was the sublimation process which ended when bunched structure appeared. Then the crystal growth started. Vanishing of bunched structure and its transition to the double stepped one was observed on setting the proper parameters during growth up. For high incoming fluxes during growth 2D nucleation happened at wide terraces and surface debunching was blocked. Application of low resulting flux of incoming and out-coming particles have also blocked the surface smoothing process. & 2013 Elsevier B.V. All rights reserved. Keywords: A1. Crystal growth A1. Kinetic Monte Carlo A1. Sublimation A1. Surface structure Experimental processes of crystal growth are always preceded by etching or chemical polishing procedures [1–6]. Such a surface preparation is carried out due to the removal of impurities and cleaning the surface in order to obtain the highest possible substrate quality. After such a process experimental parameters change and crystal growth starts. Here the results of kinetic Monte Carlo simulations reproducing two successive processes of etching followed by crystal growth are presented. Previous simulations performed for wurtzite structure crystals [7–10] contained modeling of growth or sublimation processes only. It was shown that step doubling or step meandering occurs during growth [7,8], however during crystal etching step bunching was observed [9,10]. It is also known [8,11,12] that during crystal growth step bunching occurs when electromigration, Ehrlich– Schwoebel effect [13] or defects are present in the system. Otherwise step movement equalizes step widths and smoothing of bunched surface phenomenon occurs. On the other hand, during annealing of crystal surface step velocity has opposite sign hence step movement which leads to the bunching process. It is interesting how vicinal crystal surface evolves when growth process takes place after sublimation. Especially worth investigation is the case when after desorption stage bunched structure is present at the surface and then the growth starts. During simulations kinetic Monte Carlo (kMC) lattice model of hexagonal crystal structure is used. Two species of atoms are taken into account in order to reproduce 4H politype of silicon carbide (SiC). 4H-SiC elementary cell consists of four diatomic layers which contain single Si and single C atom. Layers are shifted towards each other and form ABAC stack [14]. The paper concerns Si face of n Tel.: þ 48 22 116 3493. E-mail address: [email protected] 0022-0248/$ - see front matter & 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.jcrysgro.2013.11.011 silicon carbide hence on the top of every double layer Si atoms are situated. Nearest (NN) and next nearest neighbours (NNN) attractive interactions are present in the model. All the particle neighbours are chosen in such a way to reproduce geometry of bonds which characterizes 4H-SiC crystals. Hence direction of inter-atomic bonds varies for consecutive atomic layers. NN forces are present between atoms of different types (silicon and carbon) however NNN attraction occurs between atoms of the same type, two Si or C particles. Following interaction values are used F SiC ¼ 0:75 eV for nearest neighbours and F SiSi ¼ 0:35 eV or F CC ¼ 0:65 eV for silicon–silicon or carbon–carbon bonds respectively [15]. Every lattice site of the system is equivalent to the simulated crystal elementary cell. Transition rates for particle jumps from one site to the neighbouring one are given by a common Boltzmann factor: P D ¼ ν expð BD =kB TÞ; ð1Þ where BD is a jump barrier dependent on initial and final configurations of jumping atoms NN and NNN [7–10], kB is the Boltzmann constant and T is the system temperature. Prefactor ν ¼ 1 sets the system time scale up. Sublimation process occurs with probability P sub ¼ ν expð ðEi þ Bsub Þ=kB TÞ; ð2Þ where Ei is the energy of bonds with all NN and NNN of sublimating particle which corresponds to the depth of potential well where particle resides. Bsub is the additional parameter which is added to the particle energy in order to control speed of sublimation process. It can be negative or positive as well. Probability of adsorption depends on chemical potentials μSi and μC for silicon and carbon respectively, P ax ¼ ν expð μx =kB TÞ; where subscript x ¼Si for silicon and C or carbon atoms. ð3Þ 512 F. Krzyżewski / Journal of Crystal Growth 401 (2014) 511–513 Every simulation starts with N straight, parallel steps. Steps with silicon on the topmost layer alternate with the ones with carbon on the top. N is the number divisible by eight. It is due to the fact that model reproduces 4H-SiC crystal with elementary cell containing eight layers, and helical [7–10] boundary conditions, applied to the system. First stage of every simulation is the annealing process. Beginning of the system evolution is presented in Fig. 1. It shows the system state after first Monte Carlo step so steps are already rough according to the studied temperature, but they did not change their relative position yet. All simulations presented below were done for N ¼48, which is the number that corresponds to the height of six silicon carbide elementary cells. During annealing stage system temperature is T ¼1600 1C which is used during similar experimental processes [16]. Silicon and carbon chemical potentials are μSi ¼ μC ¼ 1 which in fact means no adsorption at the surface. Desorption parameters for silicon and carbon from Eq. (2) are equal and Bsub ¼ 4 eV. It means that desorption is the dominating process in the system. However the model used allows for particle diffusion it is particularly absent due to the fact that every single surface particle desorbs immediately. Moreover, desorption from the step edges is faster then the diffusion of particle which detaches from the step edge. Such sublimation conditions were chosen due to the fact that this is the simplest and the fastest way to obtain explicit step bunches in the described system. Bunch structure obtained after annealing process is shown in Fig. 2. Colours in the picture represent height i.e. the number of layers in every site. Brighter areas are higher than the darker ones. Sharp change of the colour from bright one to the dark indicates position of the bunch. In this case bunch consists of 42 steps. When bunched surface structure is present growth process starts. Temperature rises to T ¼2300 1C which is the value used during SiC growth [2]. Desorption barriers also change and Bsub ¼ 1 eV for silicon and carbon, hence sublimation process is slow. In the first case chemical potentials for silicon and carbon were μSi ¼ μC ¼ 3:3 eV and bunch obtained during annealing changed into even double steps, hence the surface smoothing was observed. Step doubling is the result of different strength of inter-atomic bonds for two species of atoms forming SiC crystals. Carbon is binded more strongly than silicon, and in conditions where adsorption and desorption barriers are equal for C and Si, the first one covers whole crystal surface. Such a double stepped, smooth structure is shown in Fig. 3. Evolution of the surface is presented in Fig. 4 where time dependent root mean square (RMS) roughness parameter is plotted, red solid line corresponds to the smoothed system. Three stages of the system evolution can be found there. First, when sublimation process takes place and RMS parameter grows slowly. The first stage finishes, when sudden jump of the RMS occurs. This is the moment when bunch appears at the crystal surface. One can notice that creation of a bunch is very sudden. If only, as a result of a fluctuation, one of the terraces becomes significantly wider then the others it expands further very fast and squeezes the rest. Such a sudden bunching is the result of very fast sublimation process. In such conditions very fast step movement causes rapid bunching. When bunch is already present growth stage takes place. One can see, the slow decrease of RMS parameter which corresponds to the surface smoothing process. During the last stage RMS parameter is constant exact to the system fluctuations. It means that debunching process came to the end, and further significant changes of the crystal surface are finished. In another case, when growth process is faster with μ ¼ μSi ¼ μC ¼ 2:7 eV, 2D nucleation of surface atoms is observed. Islands are created at the widest terrace, and the process of the surface Fig. 3. Even double steps after slow growth stage. Crystal growth was carried out for chemical potentials μ ¼ μC ¼ μSi ¼ 3:3 eV, desorption parameter Bsub ¼ 1 eV and system temperature T¼ 2300 1C. Fig. 1. Top view of the simulated surface at the beginning of the system evolution, step number N ¼ 48. Fig. 2. Top view of the bunched surface at the end of the sublimation stage and before the crystal growth process. Sublimation was carried out for desorption parameters Bsub ¼ 4 eV and temperature T ¼1600 1C. Black oval shows position of the 42-step bunch. (For interpretation of the references to colour in this figure caption, the reader is referred to the web version of this article.) Fig. 4. Time dependent RMS roughness parameter for the slowly (red solid line) and fast (green dashed line) grown system. Sublimation and growth stages are clearly visible. (For interpretation of the references to colour in this figure caption, the reader is referred to the web version of this article.) F. Krzyżewski / Journal of Crystal Growth 401 (2014) 511–513 Fig. 5. 2D nucleation at the widest terrace observed after growth in high incoming Si and C fluxes. Crystal growth was carried out for chemical potentials μC ¼ μSi ¼ 2:7 eV, desorption parameter Bsub ¼ 1 eV and system temperature T ¼ 2300 1C. 513 one where 2D nucleations occurs at the widest terraces. In such a case island catch particles diffusing at surfaces and number of atoms that attach to the steps are very low. Hence the step movement is very slow and smoothing process is blocked. In the third schema (blue asterisks in Fig. 6) resultant flux of adsorbing and desorbing particles is very low. Hence only few particles attach to the edges, steps move slowly and debunching process does not occur again. To conclude, one can state that kMC simulations have shown how properly tuned condition of 4H-SiC crystals growth can lead to the smoothing of bunched surface. It was also shown that the 2D nucleation which is present any time when the external fluxes of incoming atoms are very high blocks the debunching process. Similar situation occurs when resultant flux of incoming and outcoming particles is very low. In the last two cases step movement is blocked and surface smoothing cannot take place. Hence, the results presented in the text confirm that the debunching phenomenon is the effect of step movement which is in agreement with the previous knowledge [8,11,12]. It was also shown that low step velocity blocks the surface smoothing process. It is worth noting that the phenomenon of step debunching during crystal growth following step sublimation is weakly investigated and seems to be an interesting topic for further, theoretical and experimental research. Acknowledgments This research was carried out within the SICMAT Project financed under the European Funds for Regional Development (Contract no. UDA-POIG.01.03.01-14-155/09). Fig. 6. Phase diagram representing schemes of system evolution in β μ space. Red crosses correspond to the debunching, green -s correspond to 2D nucleation which blocks surface smoothing and blue asterisks represent area where debunching does not occur due to the very low resultant flux of adsorbing and desorbing particles. (For interpretation of the references to colour in this figure caption, the reader is referred to the web version of this article.) smoothing is blocked. Such a state is plotted in Fig. 5. Time evolution RMS parameter for that process is shown in Fig. 4 in green dashed line. One can notice that after sublimation stage which is similar to the one of the previous system, drop of RMS parameter does not occur. It means that debunching process is blocked by 2-dimensional structures that evolved during fast growth. Above and other results of simulations are gathered into the phase diagram presented in Fig. 6. In the space of β ¼ 1=kB T and μ one can observe three different phases which correspond to three schemas of system evolution during growth stage. Desorption stage was always carried out in the same conditions. At the first schema of crystal growth step movement leads to the debunching process. Results of such simulations are represented by red crosses in Fig. 6. Second scheme (green -s at the phase diagram) is the References [1] K. Buchholt, P. Eklund, J. Jensen, J. Lu, R. Ghandi, M. Domeij, C.M. Zetterling, G. Behang, H. Zhang, A. Lloyd Spetz, L. Hultamn, J. Cryst. Growth 343 (2012) 133–137. [2] N. Ohtani, M. Katsuno, T. Aigo, T. Fujimoto, H. Tsuge, H. Yashiro, M. Kanaya, J. Cryst. Growth 210 (2000) 613–622. [3] N.G. van der Berg, J.B. Malherbe, A.J. Botha, E. Friedland, Appl. Surf. Sci 258 (2012) 5561–5566. [4] S. Leone, A. Henry, E. Janzen, S. Nishizawa, J. Cryst. Growth 362 (2013) 170–173. [5] P. Lauffer, K.V. Emtsev, R. Graupner, Th. Seyller, L. Ley, S.A. Reshanov, H.B. Weber, Phys. Rev. B 77 (2008) 155426. [6] J. Hassan, J.P. Bergman, A. Henry, E. Jenzen, J. Cryst. Growth 310 (2008) 4424–4429. [7] M.A. Zaluska-Kotur, F. Krzyzewski, S. Krukowski, R. Czernecki, M. Leszczynski, Cryst. Growth Des. 13 (2013) 1006–1013. [8] M.A. Zaluska-Kotur, F. Krzyzewski, J. Appl. Phys. 111 (2012) 114311. [9] M.A. Zaluska-Kotur, F. Krzyzewski, S. Krukowski, J. Non-Cryst. Solids 356 (2010) 1935. [10] M.A. Zaluska-Kotur, F. Krzyzewski, S. Krukowski, J. Appl. Phys. 109 (2011) 023515. [11] C. Misbah, O. Pierre-Louis, Y. Saito, Rev. Mod. Phys. 82 (2010) 981. [12] M. Dufay, T. Frisch, J-M. Debierre, Phys. Rev. B 75 (2007) 241304(R). [13] R.L. Schwoebel, J. Appl. Phys. 40 (1969) 614. [14] M. Camarda, Surf. Sci. 606 (2012) 1263–1267. [15] V. Borovikov, A. Zangwill, Phys. Rev. B 79 (2009) 245413. [16] S. Tanaka, K. Morita, H. Hibino, Phys. Rev. B 81 (2010) 041416(R).
© Copyright 2026 Paperzz