Beryllium Diffusion Coloration of Sapphire:
A summary of ongoing experiments
Introduction
Dr. John Emmett is a member of the AGTA-GTC Board of Governors and is a preeminent researcher into the physics and chemistry of corundum.
The following is a summary of the data collected by Dr Emmett and his co-worker Troy Douthit concerning the recent bulk diffusion practices that were started in Thailand. The availability of this data and the ongoing experiments that will help in the future understanding of the bulk diffusion treatment of corundum has been made possible in a large part by an infusion of funds from the AGTA-GTC Research Fund. The AGTA-GTC Research Fund is split into various project headings and all funds donated for a particular project are spent on that project alone.
AGTA through it's President, Richard Greenwood and the tireless work of its Laboratory Committee Chairman, Jeff Bilgore, continue to support fully these very important research initiatives.
As more experimental data becomes available, this will also be published at www.agta.org. For further background information concerning the bulk diffusion of corundum, see: http://www.agta.org/consumer/gtclab/treatedsapps01.htm
Beryllium Diffusion Coloration of Sapphire:
A summary of ongoing experiments
Sept. 4, 2002 – In early January 2002, AGTA Gemological Testing Center issued an alert warning traders that orange sapphires enhanced by a new process in Thailand appeared to be diffusion treated. (Surface diffusion is the common gemological term for this process but we will use the more scientifically correct term - bulk diffusion.) Evidence of this was a layer of orange color concentrated at the surface and just below the surface of stones, with that layer exactly conforming to the shape of the cut gemstone.
This sent gemologists scrambling to find the precise cause of color. Initial reports suggested a number of possibilities, but further studies have revealed that such stones are colored by bulk diffusion, with beryllium (Be) thought to be the primary causative agent.
Early observations of the surface conformal color layers in the pink-orange Madagascar sapphires indicated to us that the likely cause was the bulk diffusion into the gemstone of light elements such as beryllium, magnesium, or calcium (or perhaps lithium, sodium or potassium) and we so advised both GIA and the AGTA-GTC staff. These light elements substituting for aluminum in the sapphire lattice often create what is known in the scientific literature of corundum as "trapped-hole color centers". The trapped-hole color center in corundum causes a yellow coloration. This yellow coloration superimposed on a gem with a pink body color appears as orange. In colorless stones it appears yellow, but a very different color of yellow than is created by iron impurities.
At the February 2002 GILC meeting in Tucson, we discussed the coloration caused by trapped-hole color centers and the observations that some light element had obviously been diffused into the stones. At that meeting, Shane McClure of GIA presented data that indicated enhanced beryllium concentration in the colored layer. This does not occur naturally - it is an induced result.
Following the Tucson Show, we initiated a set of experiments by diffusing beryllium into a wide variety of sapphire types. A discussion of these first experiments entitled Understanding the New Treated Pink-Orange Sapphires can be found at www.palagems.com/treated_sapphire_emmett.htm. We have continued to conduct additional experiments and summarize the combined results here.
Our initial experiments used natural chrysoberyl from Madagascar as a source of beryllium, as it was our surmise that this new Thai process was discovered with the accidental inclusion of chrysoberyl in a parcel of sapphire being heat treated. Chrysoberyl is commonly found in parcels of sapphire from Madagascar. In the initial experiment we conducted two types of diffusion. The first was diffusion from a molten flux, as we judged this would most likely approximate Thai practice. Beryllium was added to both borate and phosphate fluxes by the addition of 2–4% chrysoberyl powder by weight. The stones were coated with the flux and then heated for 25 hours in an oxygen atmosphere at 1800°C. For the second type of diffusion, 2–4% chrysoberyl was mixed with high purity reagent grade aluminum oxide (sapphire) powder, and the stones imbedded in the powder. In this case the stones were heated in an oxygen atmosphere for 100 hours at 1780°C. The sapphires used for these experiments were pink and pale yellow from Madagascar, Songea sapphire, the "colorless" Sri Lanka sapphire that results from heat treating some types of geuda material, and high purity synthetic colorless sapphire.
These experiments reproduced both the complete range of colors and diffusion phenomenology that are observed in gemstones which are in the marketplace, plus a few more colors. We observed little differences among fluxes. The flux-processed stones show well-defined surface conformal color layers, while the powder-diffused stones are colored nearly completely through. This is, of course, because of the longer diffusion times in the powder experiments. As a result of these experiments, we have obtained a rough estimate of the chemical diffusion coefficient as being 100 times that of titanium or magnesium, or about 1/100 that of hydrogen for these conditions.
While these experiments were fairly definitive in the sense of showing that beryllium diffusion into sapphire was a key element in the new Thai heat treatment process, several questions remained. To determine if some of the impurities in the natural chrysoberyl or in the flux were important contributors to the final result, a new series of experiments was undertaken. In these experiments the chrysoberyl was replaced with high purity BeO (beryllium oxide) powder. Thus 0.8% BeO was added to high purity sapphire powder and the stones imbedded in the mixture. The stones were then heat treated for 33 hours in an oxygen atmosphere at 1780°C. The results of these experiments were the same as the earlier experiments using chrysoberyl, that is, all of the padparadscha, orange, gold, and yellow colors were produced. We then conducted the null experiment in which a similar group of stones were imbedded in pure sapphire powder without any beryllium compound. The stones were heated in the same way at 1780°C for 33 hours with the result that no color changes were produced. Thus it is quite clear that the beryllium diffused into these stones in an oxygen atmosphere is the single causative agent in the color changes.
The colors that are produced by the diffusion of beryllium into sapphire are primarily caused by a broad and strong absorption band in the blue region of the spectrum which produces a strong yellow coloration. This type of coloration is known in the scientific literature as a trapped-hole color center and similar coloration could, in principle, be caused by other divalent elements such as magnesium diffused into the gemstone. To determine if diffusion of magnesium into the gemstone could be a factor, we added 2% high purity MgO (magnesium oxide) into the high purity sapphire powder in which we imbedded a similar group of sapphires as in the experiments described above. The stones were then heated at 1800°C for 100 hours. There was no induced coloration in the stones. This is consistent with what we would expect, since the diffusion rate of magnesium into sapphire is extremely low. However, magnesium does produce yellow coloration where it occurs naturally in sapphire, and synthetic sapphire grown with magnesium doping also becomes yellow when processed at high temperature in an oxygen atmosphere.
We have also been experimenting with the diffusion of beryllium into other types of natural sapphire. The sapphire of Dry Cottonwood Creek, Montana is routinely processed at high temperature in an oxygen atmosphere to develop yellow and orange stones. This coloration results from the presence of naturally occurring magnesium. Typically, about 10% of a mine run sample develops strong coloration in this process. We chose a group of stones that when processed, did not develop such coloration and thus remained a very pale greenish color. These stones, when diffused with beryllium, all became a strong yellow or gold color. We next tested a natural unprocessed sample of Rock Creek, Montana sapphire with the result that all stones developed yellow to orange coloration. Sapphire from Madagascar that was a very pale yellow also became yellow or golden with beryllium diffusion, as did the bluish and greenish stones from Songea. The "colorless" Sri Lanka sapphire which results from heat treating certain types of geuda also becomes yellow, gold, or orange with beryllium diffusion. Finally, we tested some of the greenish-yellow Australian stones from the Subera deposit with the result that all of the greenish overtones were removed and the stones became good yellows and golds.
It is quite clear that it is easy to change very low value sapphires into much higher value yellow, gold, and orange sapphires. It is therefore very likely that many of these sapphires have been introduced into the market in the last one to two years have been diffused with beryllium.
We are also examining beryllium diffusion into high purity, colorless, synthetic sapphire. Two types of synthetic sapphire are being used in these experiments. The first of these is Czochralski (CZ) grown colorless sapphire produced by Union Carbide corporation in 1992. The second material is very high purity sapphire grown this year by the heat exchanger method (HEM) by Crystal Systems Inc. The latter material is regarded as among the highest purity synthetic sapphire available in the world today. Both of these materials were imbedded in the same mixture of high purity BeO and sapphire powders as described above, and heated for 33 hours at 1780°C. Both sapphire samples were highly colored. Both samples exhibit broad strong absorption in the blue region of the spectrum which is similar to what we observe in the natural sapphire. However, in addition there is a weak absorption band in the red at 700 nm which shifts the perceived color from yellow to brown. Thus both samples are brown in color. In the case of the CZ grown sample, the brown coloration penetrated approximately 1.7 mm into the sample and exhibited a very sharp color boundary. The HEM grown high purity was uniformly colored brown indicating that the diffusion penetrated the entire sample which was 7 mm thick. The difference in diffusion rates between the two identically processed samples indicates that the inward diffusing beryllium is chemically reacting with some impurity in the CZ sample (and, probably, in all natural sapphire). It is likely that the impurity is tetravalent and thus titanium, silicon, and zirconium are possibilities. Of these, silicon appears to us to be the most likely. It is clear from these experiments that beryllium alone, unlike hydrogen, can strongly color sapphire.
Our experiments thus far have only served to provide an outline to the phenomenology of beryllium diffusion in sapphire. Much remains to be done to understand what is actually occurring at a microscopic level. To make further progress the following tasks must be undertaken:
- Test many other types of sapphire for their reaction to beryllium diffusion.
- Characterize the trace element composition for a wide variety of natural and synthetic sapphire types and compare their reactions to beryllium diffusion.
- Conduct careful optical spectroscopy on the sapphires before and after beryllium diffusion to characterize the additional absorption bands induced by the beryllium, and correlate the parameters of the absorption bands with trace element chemistry.
- Finally, we must develop a low cost easily conducted test for beryllium diffusion, for if we cannot do this, the market for fancy colored sapphires will be adversely affected by the Thai sapphire treaters.
We would like to thank the AGTA-GTC Research Fund for providing a large portion of the much needed funding for these and future experiments, Terry Coldham, Mark Smith, Joe Belmont, Tom Cushman, Rudi Wobito, Hans-Georg Wild, Markus Wild, Dick Hughes, Bill Larson, Roland Naftule, Dave Witter and Garth Billings for graciously supplying stones and other support for these experiments. We would also like to thank Ken Scarratt, Tom Moses, Shane McClure, and Wuyi Wang for the instant sharing of data they developed. A number of stimulating conversations with George Rossman are also greatly appreciated, as is information and data provided by Tobias Häger.
