The growth of semiconductor nanowires has broadened the range of viable geometries available for defect-free heterostructures. This is primarily because the volumes are so small and their total strain energy is smaller than the cost of inserting misfit dislocations. Thus, systems with large lattice mismatch form axial and core-shell nanowire heterostructures below critical radii without danger of forming dislocations. Testing whether equilibrium models reliably predict the critical geometries in nanowires has been of great interest, extending similar studies for planar interfaces.This talk will show examples of strain relaxation in various semiconductor core-shell nanowire heterostructures where strain relaxation has occurred via dislocations. The types of dislocations observed and the residual strains are evaluated directly using transmission electron microscopy (TEM) as a function of nanowire composition, size, and geometry. Their geometries are compared to theoretical predictions giving insights about likely relaxation mechanisms. Dislocations must nucleate and grow at nanowire interfaces without the help of existing line defects that might exist in a planar substrate. If they form at the surface of the wire they must diffuse or glide to the interfaces. Most dislocations found in core-shell nanowires that have grown along {111} or {0001} directions are edge type and formed likely by diffusing the short distance from the surface to the interfaces. This also explains their highly uniform spacing along the wires. A glide process from the top or bottom of the wire along the interfaces is not as likely since the typical sidewall interface is not a low energy glide plane.Figure 1 shows an example of a bright-field TEM image of a wurzite InAs-GaAs (0001) core-shell nanowire that has relaxed the axial mismatch strain1. In this case there is a clear indication both from the image and the diffraction pattern that there are two lattices present. The fringes in the image and the spacing of spots in the diffraction pattern show that a mismatch of 7% has relaxed along the axial direction equal to the expected lattice mismatch strain of this system. The multiple spots are due to double electron diffraction events occurring in the GaAs shell and InAs core material. Misfit dislocations along the core-shell interfaces are directly visible from the strain fields in the image and from lattice fringes in higher magnification images.Figure 2 shows another TEM example of a relaxed GaAs-GaSb (111) core-shell interface with a set of dislocations visible2. The extra plane of an edge component is indicated by the white lines on the image. The diffraction pattern has double spots with 95% relaxation both axially and radially within this plane. The shell thickness and core radius were 9 nm and 18 nm, respectively, larger than predicted critical geometries for this radius (2 nm shell)3. Wires grown with less then 2 nm shell thicknesses were coherently strained.In the Ge-Si (111) core-shell system the critical thicknesses are 3 nm shells for a 15 nm radius core compared to 1 nm in a bulk planar growth4. Relaxation shows evidence that the mechanism is via glide along {111} planes oblique to the {110} sidewall facets. There remain questions about whether there is a complete dislocation loop on all sidewall interfaces. In some cases, for example GaAs-GaP core-shell, the GaP sidewall thickness is not uniform developing a triangular facet and therefore misfit dislocations are unnecessary on parts of the nanowire. The differences between the group IV and the III-V systems will be discussed.Acknowledgements:We thank our many collaborators including Watkins, Salehzadeh (SFU), Ruda, Savelyev, Blumin (U. Toronto), Dayeh, Picraux (Los Alamos), and Swadener (Aston U.) and are grateful for funding support from NSERC and CFI/BCKDF.
Read full abstract
Dec 15, 2014
Irene Palacio + 12
Open Access
