Journal Archive

Platinum Metals Rev., 1973, 17, (2), 46

The Platinum Metals in Crystal Pulling

Crucibles and Other Apparatus

  • J. C. Brice
  • P. A. C. Whiffin
  • Mullard Research Laboratories, Redhill, Surrey, England

Article Synopsis

The platinum metals do not react with most molten oxides and halides and therefore they have many applications in crystal pulling. They can be used not only for crucibles but also for heat shields and baffles to control both the spatial and temporal temperature distributions. The construction and use of these components is described in this article.

Single crystals of oxides and halides have an ever increasing number of uses. For example, gramophone styli, watch bearings, lasers, piezo-electric filters, and electro-optic devices are often made from oxide crystals and some optical components are manufactured from halide crystals. In many cases the most satisfactory crystals can be pulled from the melt by the method first used by Czochralski (1). One of the major difficulties with this technique is finding materials to make the necessary parts of the growth system. For oxides and halides the platinum group metals are often a good choice since they do not react with the molten charge or the atmosphere which must be maintained over it.

Figure 1 shows a crystal being pulled from the melt. The caption explains the process and Table I enumerates conditions which have been found suitable for the growth of a number of crystals. There is an alternative process due to Kyropoulos (3) in which the crystal is not raised during growth but is allowed to grow into the melt as the crucible temperature is lowered. This process is widely used for halide crystals and requires essentially the same apparatus.

Table I

Conditions for the Growth of Some Crystals

Crystal Melting Point °C Crucible Material Atmosphere Pulling Rate mm/h Rotation Rate rev/min
Y3Al5O12 1950 Iridium Nitrogen 2 60
LiTaO3 1650 Iridium Nitrogen 10 30
CaWO4 1650 Rhodium Air 10 100
MnFe2O4 1500 Iridium Nitrogen 5 10
NaBa2Nb5O15 1450 Platinum Oxygen 5 30
Sr0·5Ba0·5Nb2O6 1406 Platinum Oxygen 5 10
LiNbO3 1250 Platinum Oxygen 5 30
ZnWO4 1200 Platinum Air 10 100
BaCIF 1008 Platinum Nitrogen/Hydrogen 8 85
Bi12GeO20 930 Platinum Oxygen 10 50
NaNO2 271 Platinum Vacuum * 70

The crystals of Y3Al5O12 were for use in lasers. LiTaO3, NaBa2Nb5O15, Sr0·5Ba0·5Nb2O6, LiNbO3, and Bi12GeO20 were for electro-optic devices. CaWO4 and ZnWO4 were for maser studies. MnFe2O4 was for tape recorder heads. BaCIF was for a quantum counting device. NaNO2 crystals were for pyroelectric devices

Kyropoulos method crucible cooled at 4 deg C/h

Fig. 1

A simple crystal pulling system. The melt is contained in a platinum crucible heated inductively. A seed crystal is lowered to touch the melt surface, is held there for a few minutes and then is raised slowly. The diameter of the growing crystal can be changed by varying the temperature of the crucible. The growing crystal is rotated to eliminateradial asymmetry. Both the lifting and rotary movements must be smooth if homogeneous crystals are to be grown; in thiscase both movements were provided by a hydraulic system (2)

Both these processes need a crucible which does not react with the melt. For the crystals with lower melting points platinum is often satisfactory but, as Table I indicates, higher melting point materials necessitate the use of rhodium or iridium. Platinum has the advantages that it is the least reactive of these materials and is relatively easily worked. It has the disadvantage that in the pure state it is very soft at high temperatures so that it distorts in use, particularly if the melt expands on freezing. (It is unusual to remove all the melt during growth.) This problem can be overcome by alloying but some alloying elements can be leached from the crucible. ThusWhiffin and Orton (4) found that it was necessary to use high purity platinum forthe growth of zinc tungstate if contamination by rhodium was to be avoided. (See Table II.) Melts containing alkali metal or bismuth ions also frequently leach impurities from platinum.

Table II

Rhodium in Zinc Tungstate

Crucible Rhodium in crystals mol.%
Rh 0.075*
Pt+10% Rh up to 0.03
Pt+0.1% Rh 0.009

This is sufficient to turn the normally colourless crystal black.

Elimination of Distortion

Problems with the distortion of pure platinum crucibles can be overcome in a number of ways. With some materials it is possible to supercool the melt by say 30°C before freezing is initiated and then rapidly raise the crucible temperature to a few degrees above the melting point before continuing to cool slowly. This results in the solid growing from the centre of the crucible rather than forming a solid skin locked to the crucible walls and trapping some unfrozen melt below it. An obvious but expensive alternative is to use a thicker crucible. However in practice it is only necessary to reinforce the rim as shown in Figure 2. A thick rim is in any case desirable in all crucibles heated inductively because small distortions can cause local hot spots producing local melting which spreads progressively. (Iridium crucibles are particularly prone to this type of failure.) Another method for preventing distortion is to hold the crucible in a rigid support; a tightly packed powder as shown in Fig. 3 is usually adequate or a casing of a castable ceramic can be used. Van Uitert (5) has described the use of an iridium crucible sealed between two platinum ones to achieve the rigidity of iridium combined with the corrosion resistance of platinum.

Fig. 2

Some platinum components used in crystal growth. At the top of the photograph is a seed holder, in the centre is a silica ring from which is suspended a platinum baffle for suppressing thermal oscillations, and at the bottom is a crucible with a reinforced rim

Fig. 3

Various heat shields to decrease the temperature gradient in growing crystals. The beehive shape is fabricated from iridium and the others from platinum. These shields couple to the radio frequency field and act as additional heat sources above the melt

Clearly, in a crystal pulling system it is necessary to hold the seed crystal. Figure 2 shows one method and some alternatives are given in Fig. 4.

Fig. 4

Seed holders constructed from strip and wire. Their materials are selected to suit atmosphere and temperature. That on the left was made from platinum and was used for BaxSr1−xNb206, that in the centre was for melts giving off corrosive vapours and that on the right in 40 per cent rhodium-platinum was for Y3Al5012

High Quality Crystals

So far, only the basic process has been discussed. However, the growth of high quality crystals often demands considerable refinement. The first improvement is to control (usually decrease*) the temperature gradient in the growing crystal (7). This can be done by placing a furnace over the crucible. However, a simpler and more reliable method is to use heat shields; Figure 3 gives some examples. If the most uniform possible crystals are required, it is necessary to eliminate the unsteady convection which can occur in melts heated from the side or from below. This convection produces a modulation of the growth rate which results in the presence of “growth striae” in the crystal. The effect can be eliminated by placing baffles in the melt (8, 9). In an unbaffled melt the temperature at a fixed point can oscillate several degrees with a period of a few seconds. A correctly located baffle can eliminate all measurable oscillations and so allow the production of extremely homogeneous crystals. Figure 2 shows the type of baffle used. This is lowered into the crucible after the charge has been melted. The baffle and two of the support wires are made of platinum, the third wire is 13 per cent rhodium-platinum. This wire with one of the others forms a thermocouple. The correct position for the baffle is the one which gives no variation in the output of this couple.

Precious metal thermocouples have other uses in crystal growth. With oxide and halide melts, for example, they may be used unsheathed to determine the temperature distribution in the melt, which is vital in the control of perfection (7). They can also be used to monitor the crucible temperature to ensure that the temperature control system is working, or even to activate it. Such couples should be welded to the crucible. Platinum: rhodium-platinum couples can conveniently be welded to platinum crucibles by heating both the couple and the crucible and then tapping the couple on to the crucible with a polished steel hammer. (This technique can also be used to fasten wires to other platinum components.) In inductively heated systems thermocouples can act as pick-up loops. This problem can be overcome by using a filter of the type shown in Fig. 5.

Fig. 5

An electrical filter for use on thermocouple circuits. The inductors L are radio frequency chokes with low self capacity and an inductance of either 13 or 23 mH. The capacitors are not critical but exceed about 2 μF. (Electrolytic capacitors may be used but have a life of only a few months)

It is sometimes necessary to increase the temperature gradient in order to maintain a constant diameter (6).

Maintenance of Components

Obviously, great care must be taken to ensure that the apparatus used in crystal growth is clean. Most melts are very reactive and will readily dissolve any contamination on surfaces which they touch. Some melts, e.g. ones containing bismuth oxide, are readily reduced by organic matter and a reduced melt will usually attack the platinum metals, so that care must be taken to remove lint and grease. The authors’ colleagues soak new components in hydrochloric acid and remove traces of old melts with a fusion mixture composed of two parts of sodium carbonate and one part of disodium tetraborate. After washing away the fusion mixture with water a further soak is given usually in hydrochloric acid but for some melts (e.g. ones containing lead ions) nitric acid is used. Large amounts of old melts are more easily removed with a trepanning drill before the use of the fusion mixture.

Apart from cleaning, precious metal components require little maintenance. Platinum parts which distort can be annealed and bent or hammered back into shape, although too much working is not desirable, and if much work must be done the material should be annealed several times. Annealing and welding are conveniently done with a gas torch but care must be taken to maintain an oxidising rather than a reducing flame.

This article discusses the main uses of the platinum group metals in crystal pulling. The ideas behind the use of the various components are given in the references so far cited and a comprehensive review is given in reference 10.


  1. 1
    J. Czochralski, Z. Phys. Chem., 1918, 92, 219
  2. 2
    J. C. Brice,, G. W. Lelievre and P. A. C. Whiffin, J. Phys. E, 1969, 2, 1063
  3. 3
    S. Kyropoulos, Z. Anorg. Chem., 1926, 154, 308
  4. 4
    P. A. C. Whiffin and J. W. Orton, Brit. J. Appl. Phys., 1965, 16, 567
  5. 5
    L. G. van Uitert, Platinum Metals Rev., 1970, 14, 118
  6. 6
    J. C. Brice,, O. F. Hill and P. A. C. Whiffin, J. Crystal Growth, 1970, 6, 297
  7. 7
    J. C. Brice, J. Crystal Growth, 1968, 2, 395
  8. 8
    P. A. C. Whiffin and J. C. Brice, J. Crystal Growth, 1971, 10, 91
  9. 9
    J. C. Brice,, O. F. Hill,, P. A. C. Whiffin and J. A. Wilkinson, J. Crystal Growth, 1971, 10, 133
  10. 10
    J. C. Brice, The Growth of Crystals from Liquids (North Holland, Amsterdam), in press .

Find an article