Acknowledgements
2. Description of technique
2.1.
Examples
When crystals grow from a melt, we learn something about the relevant phase diagram. Conversely, knowledge of the relevant phase diagram will allow one to set carefully some of the conditions for crystal growth. We seldom have, needless to say, the phase diagram information necessary, and it is often much faster to optimize growth conditions without learning much in detail about the phase diagram. This is in part because factors other than those contained in the phase diagram also contribute to the process. Nevertheless, access to the tabulations of binary phase diagrams (i.e, Hansen 1958, Elliott 1965, Shunk 1969, and Moffatt 1981) is extremely useful when experimenting with possible fluxes for growth of a particular material.
The simplest way to introduce the concepts appropriate to growth of crystals from metal fluxes is with a simple binary eutectic system. Suppose we wish to produce crystals of Si. We know perfectly well that there is an entire industry built
56 Z. FISK and J.P. REMEIKA
on extremely high quality Si single crystals. Nevertheless, we might wish to p e r f o r m some experiments on small crystals which we could easily p r o d u c e ourselves. Si is chemically r a t h e r inert with respect to acids such as H N O 3 , HC1 and basic solutions such as N a O H . This would allow us to extract Si crystals from a metallic flux which is attacked by the above reagents. The simplest metals to consider as solvents are those with relatively low melting points: A1, Bi, Ga, In, Sn, and Zn. Zn has a high v a p o r pressure and is a p o o r solvent for Si, as we find out in Hansen. Bi is also a p o o r solvent. It dissolves about 5 a / o Si at 1000°C, G a about 20 a / o and Sn something in the vicinity of 5 a / o . A1 appears to be best in this respect, dissolving about 45 a / o at 1000°C, and the eutectic composition contains 11.3% Si (fig. 1). This makes the A1 flux especially attractive for growing Si single crystals: large solubility at a reasonable t e m p e r a t u r e , ease of removal and reasonable composition range b e t w e e n eutectic composition and solubility at starting temperature. It is clear that if this last range is small, little material will be available for crystallization before the entire melt solidifies.
T h e small scale laboratory p r o c e d u r e for this A1 flux growth of Si crystals then involves weighing out the appropriate amounts of the elements corresponding to the t e m p e r a t u r e where you wish growth to initiate and placing t h e m in a suitable crucible. A1 will attack SiO 2 when molten and SiO 2 cannot, t h e r e f o r e , be used.
A120 3 crucibles are readily available and will not be attacked by A1 until, in our experience, temperatures near 1600°C when some solubility of A120 3 in A1 begins and where the v a p o r pressure of A1 becomes an additional problem, partly because the thin oxide skin, which often covers molten A1, breaks up and loses its integrity, resulting in a much increased evaporation of A1. A protective atmos-
1 4 0 0
\
~ 1 1 a / o S i i
A
o v 1 2 o 0 LLI CC E:) I.- 1 0 0 0 ,<
CE LU :~ 8 0 0 l.U F--
6 o o
[ I I I
1 6 0 0
I _ l A _ _ _ I _ _
4 0 0 0 20 4 0 6 0 8 0 1 0 0
A I a / o S i S i
Fig. 1. Temperature-composition phase diagram for the A1-Si sys- tem. (After Hansen 1958.)
G R O W T H OF SINGLE CRYSTALS F R O M M O L T E N M E T A L FLUXES 57
phere must be provided for the melt, and inert gas ( H e or Ar) is preferred over vacuum because more A1 evaporation will occur in vacuum. The inert a t m o s p h e r e can either be provided by sealing the crucible in a quartz tube, simply done at a glass bench, or using a gas atmosphere in a tube furnace arranged vertically to a c c o m m o d a t e the crucibles. Horizontal tube furnaces can also be used if their bore is sufficient to accept the chosen crucible size as it stands up. In this case, the crucible(s) can be loaded in on a ceramic boat. T h e r e is an important point to r e m e m b e r when sealing a crucible into a quartz tube: the tube as it stands vertically will contact the A120 3 crucible b o t t o m at its end taper. A1203 has a much larger thermal expansion than quartz and will crack the quartz as the load is heated. This cracking can be p r e v e n t e d either by flattening the quartz tube b o t t o m so that the crucible can rest there without contacting the sides, or a short stub o f quartz can be used in the tube b o t t o m to spring the crucible away from the nesting location. We also find some A1 v a p o r attack on quartz tubes from the A1 in the crucible load. This attack is slower, obviously, when the evaporation kinetics are hindered by an inert gas a t m o s p h e r e in the quartz tube. It is reasonable to adjust the gas pressure in the tube so that it is of o r d e r 1 atm at the highest working t e m p e r a t u r e , although properly sealed quartz tubes of - 1 cm internal diameter can easily take 5 atm at t e m p e r a t u r e s up to 1200°C. While it is possible to use quartz above 1200°C, this is the range above which quartz starts to soften. Then it is important both to keep the quartz tubes scrupulously free of fingerprints and surface dirt (locations where devitrification starts) and to careful- ly adjust the gas pressure to be 1 atm at the operating t e m p e r a t u r e to p r e v e n t either the collapse or the expansion of the tube.
It is generally, but not always, the case that the larger the container, the larger the crystals. It becomes difficult, if you are not a professional glass blower, to seal off crucibles larger than about 2 cm in diameter in a quartz tube, and for crucibles above this size it is much easier to work in a tube furnace equipped with a tube through which inert gas is flowing. Mullite tubes (an A 1 2 0 3 - S i O 2 ceramic) are relatively inexpensive and can be used to at least 1500°C, a c o m m o n u p p e r t e m p e r a t u r e limit of operation for furnaces heated with silicon carbide heating elements. Mullite tubes b e c o m e somewhat porous at elevated temperatures, so that a slight over pressure of inert gas is r e c o m m e n d e d at all times. It is also important to realize that there is usually present trace amounts of unwanted gases such as H 2 0 and 0 2 in high purity H e and A r delivered from gas bottles. The gas can be first run through a commercial gas purifier, or one can be constructed by packing a quartz tube with granular Z r metal, which is heated in a small tube furnace to 800°C and through which the gas is first passed. Even at the p p m level, 02 can be gettered from the gas stream by the metal flux. It makes sense, therefore, to not flow more gas past the crucible then necessary. This can be accomplished by causing the gas flow to by-pass the tube, while maintaining a positive pressure within it, as shown in fig. 2. It is useful to know that the McDaniel C o m p a n y which sells mullite tubes also sells gas tight end-cap sealing fixtures for their tubes which require no additional preparation of the mullite tubes.
58 Z. FISK and J.P. R E M E I K A