Bubbles size (diameters range from 10 lm to 45lm) in Tidoped
sapphire crystals (Fig. 9) are comparatively larger than in
sapphire crystals (diameters range from 2 lm to 5 lm) grown by
CZ technique. Fig. 9a shows some individual spherical bubbles
and Fig. 9b shows some ‘‘bubbles’’ present various geometry and
size. As analyzed in the section of undoped sapphire crystals, the
‘‘bubbles’’ in various shapes are formed by several adjacent spherical
bubbles. Fig. 10a represents magnified image of individual
bubbles, Fig. 10b and c illustrate connected bubbles arrange periodically
in line while Fig. 10d indicates adjacent bubbles arranged
curvilinearly. For Ti-sapphire crystals, when the growth rate was
lower or equal to 1.5 mm/h and rotation rate 8 rpm, the whole
crystal is free of bubbles or any other macro defects. When the
pulling rate was increased to 2.5 mm/h, the crystal tended to
contain micro-bubbles in the core of beginning and end parts,
bubbles of beginning part are quantitatively more numerous
than bubbles of end part, while the middle part is always
exempted of bubbles. When the growth parameters was still
4 mm/h and 8 rpm, but the started titanium concentration was
enhanced from 0.28 atom% to 0.5 atom%, serious titanium segregation
took place and redundant titanium was finally solidified in the
end part, thus more bubbles were captured in the end part accompanied
the rejection of titanium (Fig. 11). The difference of bubbles
distribution between sapphire and Ti-doped sapphire crystals is
due to that, for pure alumina the destabilization of solid–liquid
interface results from the gas rejection while for doped materials
from the rejection of the impurities, which is accordant with
[19]. Besides, a qualitative measurement of the Ti3+ spatial distribution
inside the Ti-sapphire crystals inspected by microluminescence
analysis validated this phenomenon further. The spectra
were recorded on the wafers of 30 mm in diameter and 5 mm in
thickness at different part of the crystals. For the measurement,
the wafer was placed under a microscope (lens 15) and was
directly excited with a focused beam of 532 nm frequency-doubled
and Q-switched Nd:YAG laser. The luminescence was recorded
through the microscope by an optical fiber and transmitted to a
monochromator that dispersed the emitted light onto a CCD detector
cooled to 20 C. The measurement was performed in different
transversal sections of the wafer exempt from bubbles or containing
bubbles in the core. The curves in Fig. 12a indicate relativelyhigher luminescence intensity and higher Ti3+ ion concentration at
the periphery than the core of the wafers exempt bubbles. The
radial distribution of Ti3+ in Fig. 12a express as luminescence at
730 nm is illustrated in Fig. 12b, we assume that the variation of
luminescence intensity is linear with the concentration of Ti3+ in
the crystal. On the contrary, Fig. 13a shows relatively higher luminescence
intensity and higher Ti3+ ion concentration in the core
region contains bubbles than the periphery region without bubbles,
the radial distribution of Ti3+ in Fig. 13a express as luminescence
at 730 nm is illustrated in Fig. 13b. The explicit contrast of
Figs. 12 and 13 manifests that for Ti-doped alumina the destabilization
of solid–liquid interface results from the titanium rejection,
which lead to more bubbles were captured in the core of crystals.