Central Laboratory of Mineralogy and Crystallography, 1000 Sofia

Introduction As a mineral species corundum (Al₂O₃) is sufficiently common in nature, but its beautiful natural crystals (ruby, sapphire) are quite sparse. At present its unique properties find ample application in industry. In nature the growth conditions for corundum crystals are various. They come up in metamorphosed aluminium-rich sedimentary rocks, in some igneous rocks, in pegmatites, in high temperature hydrothermal veins or in placers, but never in great quantities (Hughes, 1990). This explains why different methods to grow man-made corundum crystals, nearly all of them by melt, have been developed (Latvinov, 1990). Experiments to grow corundum crystals by the hydrothermal method have been made by Laudise, 1958, Kuznetzov, 1964, 1965, 1967 and others. Investigations of the Al₂O₃-H₂O system at higher pressures and temperatures (Laudise, 1970) have shown that independently of the pressure, corundum formation can be achieved only at temperatures exceeding 400 °C. The stable phase at lower temperatures is diaspore, while at pressure lower than about 15 MPa boehmite forms. The good solubility of Al₂O₃ has been proved in various alkaline solutions (NaOH, KOH, Na₂CO₃, K₂CO₃, Na₂HCO₃, KHCO₃, Na₂B₄O₇, etc.). However, in stronger alkaline solutions (NaOH or KOH) the solubility temperature dependence is very low, while compounds formed with Na₂B₄O₇ are rather stable (Levinson et al., 1965). Therefore, carbonate and bicarbonate solutions are recommended as the best medium for hydrothermal growth of corundum crystals (Laudise, 1958, Kuznetzov, 1964, 1965, 1967). The task we have pursued with this investigation was to achieve hydrothermal growth of corundum crystals, utilizing the existing literature information. Simultaneously, some peculiarities of the growth process have been investigated in view of its optimization.
 

Experimental The runs were carried out in 150 ml unlined steel autoclaves, at crystallization temperature of about 500 °C  (Laudise, 1970) and temperature difference of 15-28 °C. The filling was changed from 65 to 70% and depending on the type and concentration of the solution used, pressures of about 140-170 MPa have been obtained. The baffle opening was 20 and 10% of the autoclave cross-section (Laudise, 1958), and some runs have been made without baffle. The growth was realized in bicarbonate solutions. In some cases a certain quantity of NaCl was added to the solvent in order to reduce the pressure. Differently oriented seed plates cut from melt-grown man-made corundum crystals were used. Crushed large crystals from melt-grown corundum, electrocorundum, corundum ceramic, and metallic aluminium were used as a nutrient.
 

Results and discussion In most of the runs (details for some of them are given in table 1), we used bicarbonate solutions, and 10-20 mm pieces of man-made corundum crystals as a nutrient. We succeeded to achieve a visible and perfect growth of the seed plates. Some of the obtained crystals are shown in plate I,1. Depending on seed orientation and size, some crystals reached 5-10 g weight within a week. The growth rate of the rhombohedron face reached 2 mm per day. On the crystals basal pinacoid c{00.1}, second-order hexagonal prism a{11.0}, positive rhombohedron r{10.1} and hexagonal bipyramid n{22.3} appeared as well. According to Hughes, 1990 the most common simple forms of natural corundum crystals are: c{00.1}, a{11.0}, n{22.3}, m{10.0}, S{02.1}, R{01.2}, r{10.1}, p{11.3}, etc. Working in conditions similar to ours (crystallization temperature about 550-600 °C, temperature difference 50-60 °C, 60% filling and pieces of man-made corundum) Kuznetzov, 1964 also observed only the above mentioned four simple forms and communicated the following growing rate (R) of crystal faces: Rr>Rn>Ra>Rc. In a series of runs we succeeded to prove this ratio in growing a seed plate cut parallel to the rhombohedral face (Fig.1). The growth rate of r{10.1} and n{22.3} reached 1.5-2 mm per day while the growth rate at c{00.1} did not exceed some tenths of one millimeter per day. In about 30 days the crystal reached its final growth form which contained only the basal pinacoid c{00.1} and second-order hexagonal prism a{11.0}. Depending on the run conditions - crystallization temperature, temperature difference, solution kind and concentration, nutrient kind and quality, as well as on a number of factors difficult to control, such as spurious nucleation, the growth ratio Rc/Rr, remaining always less than unity, varied considerably. The runs in which the growth amounts to grams per day (table 1 crystals number 1-7, and others) produce crystals with strongly curved basal face. Just over the seed this face is flat and covered by a number of small hexagonal plates (plate II, 1) and over the grown part of the crystal the growth is by dislocations producing sheets. The faces of the rhombohedron r{10.1}, when presented, are smooth and finely striated parallel to the c-axis (plate II, 3). On the prismatic faces the striation is more rough and perpendicular to the c-axis (plate.II, 2). In runs where the common growth is lower than 0.3 g/day the basal face is very flat, entirely covered by hexagonal plates similar to those on plate II, 1 and has pearly luster. Sparsely grown rhombohedron seed faces are also overgrown by the same hexagonal plates oriented parallel to the basal face. The multitude of interconnected factors determining these two growth patterns has not permitted till now to specify the causes for their appearance. The question deserves a more detailed study. As a whole, crystals grown in KHCO₃ solutions are muddier and even opaque, while crystals grown in NaHCO₃ solutions are comparatively more clear and transparent. The grown layer was transparent at slower growing rates, while at higher growing rates it was opaque. The colouring of the growing layer varied greatly depending on the run conditions. It was colourless in runs with metallic aluminium, light-greenish in runs with copper insertion and yellowish-brown in runs without insertion. In some runs intense indigo-bluish colouring was observed. In most of the cases the colouring was presumably due to the iron extracted from the walls of the autoclave which evidently is incorporated in corundum. To verify this a series of experiments was conducted, which attempted to limit the iron quantity in the crystals. For this purpose a copper or an aluminium insertion 1 mm thick was put in the autoclaves and all metallic parts into the autoclave were also of copper or aluminium. In the runs where a copper insertion was used despite the insufficient isolation of the iron, semitransparent light-green crystals with an interesting "avanturine" effect were obtained. This effect is due to small (less than 1 mm) well shaped octahedral copper crystallites caught into the corundum during the growth (plate I, 3). Simultaneously with improvement of crystal quality the growing rate lowered, probably as a result of the decreased temperature difference.

A spontaneous crystallization takes place at higher temperature differences and a crust of numerous of tiny (below 1 mm) corundum crystallites forms on the autoclave walls. Although barely discernible, this crust disturbs seriously the temperature difference, lowers the oversaturation and greatly constricts the seed growth. In order to avoid the spurious nucleation it is necessary to maintain a lower temperature difference but this leads to a decrease of the growth rate in general. The crust formed can not be removed mechanically, but is easily dissolved in Na₂B₄O₇ (Levinson et al., 1965). Our attempts to optimize this process will continue.

It was observed that even after a thorough cleaning of the autoclave from parasite crystals, if we use the same nutrient crystals in several succeeding runs, the growth rates gradually decrease. The observation of the nutrient pieces in such series of runs has shown that they are continuously covered by crystal faces. The phenomenon can be explained with the existing although minimal temperature difference in the autoclave dissolution zone itself. The crystals in the hotter (lower) part of this zone dissolve, while the crystals in its cooler (upper) part grow. Thus, although slowly, the crystal pieces in the upper part reach up to forms covered by slower dissolving crystal faces - mainly basal pinacoid c{00.1} and hexagonal prism a{11.0}. To avoid this different proceeds can be applied. Best results were achieved by crushing the nutrient crystals before each run, which leads to the increase of the surface of the faster dissolving faces. It is possible to conduct runs without baffle, which decreases the temperature difference and so affects unfavorably the growing rate. It is also possible to search for another nutrient. The latter forced us to conduct a series of experiments, in which we checked the solubility of different materials to be used for further recrystallization of corundum. The observed good solubilities (table 2) justify a further investigation. In short runs with electocorundum (size 1-2 mm) we obtained a good growth, but in more prolonged runs we again came across the problem of recrystallization and growth of the particles placed in the cooler part of dissolution zone. A crust of intergrowing fine (1-2 mm) corundum crystals was formed and this limited the access of the working solution to the main part of the intensively dissolving crystals below it. The problem can be avoided by conducting shorter experiments (up to 7 days) or by inserting of a few screens in the lower part of the autoclave.

Despite the good solubility of corundum ceramics (table 2) we could not grow crystals with it. The subject requires further investigation.

Interesting results were obtained in using metallic aluminium. As mentioned above, aluminium was initially used as an isolation of the autoclave. It appeared that at the high temperature and pressure used, aluminium oxidizes and dissolves rather quickly in the weak alkaline solutions. The processes take significant water quantities (about 1.1 g water to 1 g aluminium). Thus the initial water quantity in the experiment gradually decreases. This changes of the filling, the pressure and the density of the working solution, as well as of other not easily controlled parameters. At the end of the experiments the whole aluminium amount always turned into corundum (plate I, 3). On the other hand, only in these experiments the seed growth was colourless. Therefore the use of metal aluminium as a nutrient deserves further attention.
 

Conclusion Recrystallization as well as synthesis of corundum under hydrothermal conditions has been realized. The runs with bicarbonate solutions, rhombohedron seeds, crystallization at temperature higher than 500 °C and temperature difference lower than 10-15 °C proved to be the most successful. Hydrothermal growth of corundum crystals from melt-grown crystals, electrocorundum and metallic aluminium has been achieved. In our further studies special attention will be paid to the growth from corundum ceramics and to the synthesis of corundum from metallic aluminium. The optimization of the process will be looked for together with a more detailed investigation of the impurities influence, the recrystallization of the nutrient crystals, optimal temperature conditions and other.