CRYSTAL GROWTH FROM MELTS IN 0-G ENVIRONMENT
L. D. Fullmer and R. M. Housley
NORTH AMERICAN ROCKWELL CORPORATION
THOUSAND OAKS, CALIFORNIA
The value of highly perfect single crystals to
technology is discussed. Many crystals used sell
in the range $10,000 - $100,000 per po\ind. Many
potentially useful crystals which cannot be
grown on earth because the melt cannot be con-
tained could be grown from containerless melts
in 0-g. Very desirable increases in perfection
of other crystals currently being used could
probably be achieved in 0-g. The factors limiting
perfection and their relation to gravity are
discussed. Different techniques of growth in 0-g
are discussed and a very versatile crystal pulling
apparatus which combines the advantages of float
zone refining and Czochralski growth is schematically
Of the many possibilities which have been suggested for space manu-
facturing it seems to us that crystal growing offers the most promise
of rapid economic success. Many crystals now used in technology for
such purposes as lasers, laser modulators, semiconducting devices,
transducers, substrates for large scale integrated circuits, etc., sell
for prices in the range of $10,000 - $100,000 per pound. The fact that
these prices are paid gives a measure of the technological value of the
Many potentially useful crystals have not realized their promise hecause
it has not proven possihle so far to produce them of suitable size and
perfection. In a number of cases this seems to he solely due to the
unavailability of a crucible material that will hold a melt from which
the crystal can be grown. Noteworthy examples of difficult to contain
melts are transition metal oxides including rare earth iron garnets and
compounds with high alkali and alkali earth content.
We believe crystals of this class can be grown relatively easily from
containerless melts in space and that the economic rewards from such
an accomplishment would clearly be great .
In many other cases the crystals currently used in technology are
limited in perfection by fluctuations in composition and/or the
accompanying strains. Semiconductors, for example, are very sensitive
to dopant levels (carrier concentrations) and even small fluctuations
can render the device useless. Ferroelectric crystals used for laser
modulation, second harmonic generation, and other nonlinear effects are
extremely sensitive to fluctuations in stoichiometry which induce
domain structure and/or result in birefringence variations. This
effect causes detrimental optical losses in the device.
There are at least three well recognized causes for these composition
fluctuations in melt grown crystals l) Faceting results from the
irregular nucleation of new growth along low index planes in a low
temperature gradient. 2) Constitutional supercooling results from the
instability of the growth front to a perturbation in the composition of
the melt under conditions of high temperature gradient. In extreme
cases this leads to the formation of trails of precipitated solute.
3) Temperature fluctuations leading to compositional fluctuations
result from turbulent or oscillatory^ convection in the melt.
These convective thermal oscillations during growth have long been ,
known to lead to oscillations in the carrier density of semiconductors.
Growth striations due to this cause have also recently been positively
identified in-'-'^ rare earth doped CaF2, Nd-CaWOl)., Ba2NaNb50i2,
Nd-Y3Al50i2, Cr-Al203, and Cr-MgAl20l^ .
Convection and the associated oscillations would of course be absent
in 0-g. Absence of convection and the constraints imposed by a melt
container in 0-g also offer the possibility of arranging the heater and
insulation geometries in such a way that the freezing isotherm is not
near parallel to any low index crystal planes and at the same time the
temperature gradients are small. This would minimize problems due to
both faceting and constitutional supercooling. Therefore, there is
real promise that better quality crystals of these materials can be
grown in o-g.
Modest improvements in crystal quality resulting in devices with
improved characteristics or higher yields of acceptable devices could
again have a tremendous economic importance.
Conceptually the simplest way of growing crystals in 0-g is from a
seeded free floating melt. This possibility was mentioned by several
speakers last year. Consideration of surface energies indicates that
the seed and growing crystal would always stay inside the spherical melt
until the maximum dimension of the crystal corresponded to the diameter
of the melt. Perfection of crystals obtained by this technique would
be limited by faceting and constitutional supercooling as are other
melt grown crystals.
The cooling time necessary to grow high quality crystals would certainly
be measured in hours. Clearly some method of keeping the melt centered
in the furnace for this length of time would be needed. Magnetic
induction and gas jets have been suggested as means for positioning such
a melt. Both would probably lead to temperature fluctuations in the melt
and hence imperfections in the crystal.
In addition the constant attention of an operator would be necessary to
keep the melt centered. Alternatively, a servo system could possibly
be devised but this would involve an additional development of sensors
and control circuitry. It might be difficult to design sensors and
positioning devices compatible with a furnace which will provide the
necessary temperatures and temperature gradients.
All of this is not to say that good crystals could not be grown from
a free floating melt. It is only to convince you that the modified
Czochralski method which we propose is actually simpler as well as far
A schematic drawing of o\ar design is shown on slide 1. It can also be
looked at as a modification of the floating zone principle to incor-
porate the advantages of the Czochralski method. These are the use of
a well crystallized seed and the ability to neck down the crystal to
prevent the propagation of twin boundaries, dislocations, etc.
The basic equipment would be two opposing crystal pullers with reversible
and varying pull rates and a suitable heating arrangement with controls
and programmer. The insulation would be designed and fabricated from
suitable materials for the specific crystal pulled. A versatile puller
would have interchangeable containers for a variety of atmospheres, i.e.,
reducing or oxidizing. This design would not require positioning of
the melt .
Slide 2 shows the operation of the puller. The end of a large, pre-
pared ingot or crystal would be positioned and melted in the furnace.
The ingot could be prepared by pressing and/or sintering a combination
of materials yielding the right chemical composition. After the melt
is obtained and thermally stabilized, the opposite puller would intro-
duce a seed into the melt. The melt and seed could be simultaneously
moved within the furnace to establish the proper gradients, etc. The
seed piiller would then "pull" the crystal as in the Czochralski tech-
nique. As the material is removed from the melt, the feed rod puller
could replenish the supply at a steady rate. We believe a crystal
puller of the type described here could be engineered, built, and tested
within a year, and hence could easily be ready to go on a 1973 flight.
Space Manufacturing Process Chamber #2 appears to have enough room for
equipment designed along fairly conventional lines.
One final subject must be discussed. That is the source of heat which
will be used. This has two aspects, the type of heating element in the
fiornace and the ultimate power soxirce. Each useful crystal produced by
melt growth will require from several kilowatt-hours to several tens of
kilowatt-hours of energy. The only ultimate soiirces of power which
appear practical are solar power and nuclear power. If nuclear power
is used, the heating elements could be of any conventional type, i.e.,
resistance, electron beam, glow discharge, induction, etc. This has
obvious advantages .
If solar power is to be used, it seems that high priority should be
given to design of such a furnace and a Space Manufactin-ing Process
Chamber compatible with it. A polar orbit would probably be required
for the successful use of solar power. Solar power might have advan-
tages for certain, refractory insulating materials, but glow discharge
heating should also be considered.
Since in the 1973 tests only battery pack power will be available,
experimental rions should be made with a relatively low melting point
material to limit power consumption. The material should also have
been thoroughly studied on earth so that the results can be interpreted
unambiguously. Many materials qualify, for example, KCl.
Production of say 100 crystals with a value of several million dollars
would require of the order of 1000 kilowatt-hours of energy. Assuming
this power is available, quantity production could begin in the 1975
A real evaluation of the cost effectiveness of crystal growth from melts
in space depends on a realistic assessment of the power cost.
1. B, Cockayne, J. Crystal Growth 3_, ^, 60 (1968), and references
2. W. R. Wilcox and L. D. Fullmer, J. Appl, Phys. 36, 2201 (I965).
3. D. T. J. Hurle, J. Gillman, and E. J. Harp, Phil. Mag. lk_, 205
k. D. T. J. Hurle, Phil. Mag. 13, 305 (1966), and references therein.
Fig. I Pull for 0-6ravity Environment
Fig. 2 Procedure for Crystal Growth in 0-Gravity Puller