Sources of threading edge dislocations

 

 

The primary source of threading edge dislocations in silicon carbide boules grown along [0001] directions are seeds used in the process.  The PVT grown crystals are cylindrical in shape with seed about the same diameter as the boule itself.  It is, therefore, to be expected that all of the dislocations with line direction along c-axis will propagate into the overgrowth.  In addition, there are several mechanisms which increase the concentration of threading edge dislocations in SiC crystals.

 

One of these mechanisms is plastic deformation of boules during growth and/or post-growth cooling.  The evidence of this process is discussed below.  Fig. 1 is an optical micrograph of an etched Si(0001) face of a 4H SiC wafer.  The wafer edge is visible in the lower right corner of the figure and a part of a polytypic inclusion or a misoriented grain at the edge of the wafer is shown in the lower center.  The most apparent features in this figure are the two parallel bands of small etch pits extending from the bottom to the top of the figure.  Their full lengths were about 4 and 10 mm, respectively (the wider of the two bands is marked with an arrow in Fig. 1).  It is also noticeable that there are several additional less well defined bands of pits rotated 60° away from the dominant bands.  The widths of the bands were between 30 and 200 mm with the etch pit densities in the bands in the range of 0.5~2.2 ´106 cm-2.  Laue x-ray diffraction was used to determine the orientation of the bands.  All of them extended along the <11-20> directions.

 

 

Fig. 1  Optical micrograph of the etch pit bands on KOH etched Si(0001) face of a 4H SiC wafer with etch pit bands along <11-20> directions.

 

In the figure, most small etch pits show regular circular shapes on the surface, which implies that these are due to threading dislocations approximately normal to the basal plane and wafer surface.  In particular, they are different from the shell etch pits assigned to basal plane dislocations by Takahashi et al. (J. Takahashi, M. Kanaya, Y. Fujiwara, J. Crystal Growth 135, 61 (1994)).  Conventional and high resolution TEM was used to determine the character (line direction and Burgers vector) of the dislocations in the bands.

 

 

Fig. 2  Plan view bright field conventional TEM micrograph showing a part of a dislocation array shown in Fig. 1.

 

Fig. 2 is a plan view conventional TEM image showing a part of a band shown in Fig. 1.  Five dislocations are visible in the figure forming a straight array along a [11-20] direction.  This image was taken in a two beam diffraction condition tilted from the c-axis by less than 5°.  The fact that the lattice distortion contrast features due to the dislocations are nearly point-shaped implies that the dislocation lines are almost parallel to the c-axis.  Subsequently, high resolution TEM was applied to determine the line directions and Burgers vectors of the individual dislocations more precisely.  Fig. 3 is a plan view high resolution lattice image around a dislocation in the array shown in Fig. 2.  The image was taken by choosing the c-axis as the zone axis.  The dislocation is of pure edge type and threading along the c-axis without tilt.  Its core is at the intersection of the two extra half planes marked with two rows of dots.  The corresponding Burgers vector, determined by drawing a Burgers circuit around the core, is a/3[11-20] with a direction marked with an arrow.  The dislocation array and the Burgers vectors of the individual dislocations in it were determined to be parallel to each other.

 

 

Fig. 3  C-axis plan view lattice image around a dislocation in the array of Fig. 2.  The two extra half planes are marked with two arrays of dots and the corresponding Burgers vector direction with an arrow.

 

The type of dislocations making up the array, the relationship between their Burgers vector and the array direction, and the array morphology are consistent with a slip band generated by the prismatic slip: <11-20>{-1100}.  At initial stages of single crystal deformation, relatively few dislocation sources are operative.  New dislocations will be multiplied at the sources and glide in the plane determined by their Burgers vector and line direction.  Repeated multiplication would generate an array parallel to the Burgers vector as observed in the above experiments.  It should be pointed out, that such arrangement of dislocations corresponds to a high energy configuration because the strain fields of individual dislocations add up and create long range stress fields.  It is highly unlikely that such array could be grown-in.  Also, the grown-in dislocations are typically between misoriented sub-grains nucleated and grown independently, and form low angle grain boundaries.  The <11-20> arrays discussed here are not misorientation boundaries like the low angle boundaries.  Another characteristic feature of the <11-20> arrays is their width (30~200 mm in Fig. 1).  Glide dislocations typically do not lie in one glide plane but in a set of parallel planes with widening caused by multiple cross glide (D. Hull, D. J. Bacon, Introduction to dislocations, 3rd ed., 179 (Butterworth-Heinemann, Oxford, 1984).  All of the above characteristics lead us to interpret the origin of the arrays as the prismatic slip.  It is well known that the primary mechanism of SiC deformation is the basal plane slip: <11-20>(0001). While the activation of the secondary slip is more difficult, it has been observed experimentally by Maeda et al. [K. Maeda, K. Suzuki, S. Fujita, M. Ichihara, S. Hyodo, Phil. Mag. A 57, 573 (1988)].  TEM images of samples indented at room temperature show dislocation loops with <11-20> and [0001] segments.  This indicates that while the Peierls energy in SiC is significant, the activation of secondary slip systems is possible.  Considering the fact that the Peierls stress decreases with increasing temperature, it should be much easier to activate the secondary slip while cooling from the growth temperature (~2300 °C) than at room temperature.  Similar dislocation etch pit arrays (the arrays were along the <11-20> directions on etched basal plane surfaces) were reported by Amelincx et al. [S. Amelinckx, G. Strumane, W. W. Webb, J. Appl. Phys. 31, 1359 (1960)] in SiC crystals grown by the Lely method.  They interpreted them as glide traces in a pyramidal slip system <11-20>{3-301}forming dislocation pile-ups against a screw dislocation or screw dislocation groups.

 

As illustrated in Fig. 1, the apparent origins of most slip bands are misoriented grains or polytypic inclusions at the periphery of the wafers.  The grains or inclusions are frequently observed to nucleate on the growth crucible walls.  The following argument points toward a likely mechanism of the slip band formation.  It is plausible to assume that the crystal around the misoriented grains is stress free at the growth temperature.  During post-growth cooling, the thermal expansion anisotropy between the matrix boule and the misoriented grains leads to stress build-up at the interface as well as inside the grains and the boule.  Similarly, any polytypic inclusions would result in stresses due to differences in thermal expansion.  As the temperature decreases, the stresses will increase until they exceed the critical resolved shear stress and activate the dislocation glide in the corresponding slip system.

 

In summary, the results presented above provide evidence of threading edge dislocations produced in SiC boules by plastic deformation.  The apparent source of stresses are misoriented grains and/or polytypic inclusions.  It is not clear at this time if prismatic slip could be activated by macroscopic stresses induced by temperature gradients in the crystal during growth.  It is also likely that there are other mechanisms that lead to creation of edge dislocations with a/3<11-20> Burgers vector and line direction of [0001].

 

Return to:

 

[ CMU SiC Page ] [ MSkowronski Home Page ]