Fundamental mechanisms of strain relaxation in lattice-mismatched heteroepitaxy/ FRG



Main Page
Curriculum Vitae
Research Group
Current Projets
Other MSE Links










contact persons: Jennifer Gray (jlg9v@virginia.edu); Chi-Chin Wu (cw9r@virginia.edu)

1. Stress relaxation in SiGe/Si epitaxial films:  Tailoring surface morphology and mapping relaxation regimes           

Heteroepitaxial growth of semiconductor structures is important for the fabrication of many types of devices.  However, due to the lattice mismatch, large amounts of strain develop in the epitaxial layer.  This strain may be reduced through the formation of islands on the surface of the film1 when there is sufficient adatom mobility, and also through the injection of misfit dislocations at a critical thickness2.  Strain-relieving islands can self-assemble into ordered two-dimensional arrays on the film surface.  There has been recent widespread interest in using island morphologies produced by such strain relieving mechanisms as a possible way to form quantum structures for new devices.  In these experiments we measure relaxation in a growth regime where both of these mechanisms compete and interact as a function of the deposition variables. By mapping out relaxation regimes as a function of growth parameters and combining with detailed microscopy analysis, we can gain insight into how to control the surface and defect structure of thin films in general.

The surface morphology that develops in SixGe1-x films grown on Si is known to be strongly dependent on the composition and growth temperature.  Growth at temperatures above about 700°C follows the widely observed morphological evolution in the Ge/Si system of the formation of discrete hut clusters followed by a transition to dome shaped islands at greater thicknesses 3,4.  Growth at lower temperatures generally produces a rippled surface morphology due to limitations on adatom diffusion.5,6  However, we have found that for relatively low growth temperatures (550°C, Si0.7Ge0.3), it is possible to form ordered quadruplet arrays of SiGe nanoclusters, with possible applications in quantum cellular automata architectures.7

In these experiments,  Si0.7Ge0.3 films are grown using molecular-beam epitaxy (MBE) on (001) Si substrates.  The MBE system is equipped with reflection high-energy electron diffraction (RHEED), for monitoring the surface of the film during growth, and a multi-beam optical stress sensor (MOSS) system8 that measures wafer curvature which is related to stress through Stoney’s equation .  The in-situ wafer curvature measurements from films grown at 550°C and a rate of 0.9 Å/s indicate that the stress in these films remains relatively constant until approximately 20 nm of growth (Region I in Fig. 1).  Around this thickness, relaxation of 10% or less of the initial stress in the film occurs, as is determined by the change in slope of the stress-thickness curve at this point.  Further growth beyond the islanding transition leads to a second major transition in the stress-thickness slope at approximately 50 nm.  This rapid relaxation regime corresponds to the introduction of misfit dislocations in the film as we have confirmed by plan-view transmission electron microscopy.

Atomic force microscopy (AFM) shows the initial development of small, shallow pyramidal surface pits (Fig. 2a) at film thicknesses of 5-15 nm, below the islanding transition.  With further growth past the islanding transition, nucleation of islands surrounding the edges of the pits along the <100> directions (Fig. 2b) occurs, forming a structure that we term “quantum fortresses”.  This resembles the cooperative nucleation process as described by Jesson, et al9.  The sidewall angle of the pits is close to that of the typical hut cluster island <501> facet. After the introduction of dislocations the “quantum fortress” islands begin to be broken up by a cross-hatch morphology caused by the dislocations along the <110> directions.  A similar island geometry has been observed previously in the SiGe/Si system, using deliberate incorporation of C.10   However, the quantum fortress morphology in these experiments appears to be intrinsically associated with the growth conditions. 

When the growth rate is decreased to 0.15 Å/s, the in-situ curvature measurements are similar, however, the resultant surface morphology consists of initial small and densely packed square islands that elongate into island ridges along the <100> directions as growth is continued (Fig. 2c).  For higher growth temperatures (650, 750 ºC) we do not observe the quantum fortress morphology, only islands for these growth rates.  It thus appears that the role of growth rate upon surface morphology is far more significant at lower growth temperatures in this system.  If we anneal films grown under the “quantum fortress” growth conditions, we can get yet another morphology.  By stopping growth just after pit formation and annealing, we find that the pits rapidly elongate in one <100> direction to become slits (Fig. 2d).  It may be possible to tailor the surface morphology further by repeated sequences of growth, followed by anneals. 

These results show that under kinetically limited conditions, growth rate has a large impact on the resulting surface morphology and can be used to form complex ordered nano-architectures.  More experiments are currently being done to determine the temperature, growth rate and composition ranges over which “quantum fortress” type structures can be formed.  The in-situ wafer curvature data will allow us to map out when these relaxation transitions occur as well providing the quantitative measurements for relaxation amounts.  This combined with the microscopy data will provide better understanding of the relaxation behavior of strained heteroepitaxial systems and the control of surface morphology and defect microstructure.



Text Box: FIG.1. In-situ wafer curvature data for Si0.7Ge0.3 grown at 550°C and 0.9Å/s.   Region I: flat coherent growth.  
Region II: coherent islanding.  Region III: dislocations 
Text Box: FIG.2. AFM image of (a) 15nm film and (b) 30nm film grown at 550°C and 0.9Å/s.   (c) 30nm film grown at 550°C and 0.15Å/s.  (d) 5nm film grown at  550°C and 0.9Å/s, then annealed for 1 hr.  









1.  R.J. Asaro and W.A. Tiller, Metall. Trans. 3, 1789 (1972).

2.  J.W. Matthews and A.E. Blakeslee, J. Cryst. Growth 29, 273 (1975).

3.  Y.-W. Mo, D.E. Savage, B.S. Swartzentruber, and M.G. Lagally, Phys. Rev. Lett. 65, 1020 (1990).

4.  J.A. Floro, E. Chason, R.D. Twesten, R.Q. Hwang, and L.B. Freund, Phys. Rev. Lett. 79, 3946 (1997).

5.  A.G. Cullis, D.J. Robbins, A.J. Pidduck and P.W. Smith, J. Crystal Growth 123, 333 (1992).

6.  J.A. Floro, E. Chason, L.B. Freund, R.D. Twesten, R.Q. Hwang, G.A. Lucadamo, Phys. Rev.B 59, 1990 (1999).

7.  G. Bernstein, C. Bazan, M. Chen, C.S. Lent, J.L. Merz, A.O. Orlov, W. Porod, G.L. Snider, and P.D. Tougaw, Superlattices and Microstructures 20, 447 (1996).

8.  J.A. Floro and E. Chason, Mater. Res. Soc. Symp. Proc. 406, 491 (1996).

9.  D.E. Jesson, K.M. Chen, S.J. Pennycook, T. Thundat and R.J. Warmack, Phys. Rev. Lett. 77, 1330 (1996).

10. X. Deng and M. Krishnamurthy, Phys. Rev. Lett. 81, 1473 (1998). 


2. Investigation of stress relaxation and dislocations for periodic sinusoidal Si1-xGex /Si

                Due to the need to release the great strains and stresses induced by interface misfit, the evolution of the undulation instabilities lead to a sinusoidal-like morphology. Any corrugation of the SiGe surface is very detrimental in the fabrication of high-speed electronic devices that require abrupt interfaces and flat layers. On the other hand, this periodic sinusoidal morphology can be used as self-patterned templates for selective growth of self-assembled nanostructures. Before achieving that, there is an urgent need to perform extensive studies both in modeling/ simulations with calculation techniques and observe them experimentally to understand the possible correlations among competing factors of strain relaxation, morphology and dislocations.

            In order to gain fundamental understandings about the strained layer epitaxial systems for the study of dislocations and strain relaxation kitnetics for semiconductor devices. The research is conducted through experimental observations from state-of-the art transmission electron microscopy to in-situ observe the dislocation movements as well as undulation morphology and correlate this information with the stress distributions and dislocation interactions from simulation and modeling with finite element analysis techniques. This work involves extensive collaborations with researchers at the IBM Watson Research Center in Yorktown Heights, New York State. The research system mainly focuses on Silicon-Germanium alloy thin film with various thickness and undulation dimensions grown on single crystalline silicon substrate. The main tasks of this research are to investigate the effect of epitaxial layer morphology (such as sinusoidal curve, various types of islands or any other networks) on the strain relaxation mechanism and dislocation interactions.

                The correlations between the simulations of dislocation interactions with complicated self-interactive programs, stress modeling with finite element methods, and in-situ observations make this research unique. With these tools and advanced research resources, it is possible to analyze and better understand the strain relaxation mechanism which is a crucial information for successful thin film growth in semiconductor industry. With the IBM code, this research is able to simulate where a specific existing dislocation moves at any interested time instant and how it interacts with other dislocations inside the system. With in-situ TEM experiments, we are able to observe dislocation movements and correlate them with the stress relaxations in local regions. With models done by the finite element analysis method, the preliminary results have demonstrated that the strains and stresses can be relaxed by inducing surface undulations in such a way that the stress is released in the top surface regions and compressed in the trough regions, as shown in the figure below. The same technique can also study the composition segregation and precisely correlate such information with stress relaxations, which is otherwise difficult to carry out in TEM experiments in satisfactory accuracy.