Colorless and Green Grossularite From Tanzania
By PIETER MUIJE, Ph.D. Associate Professor of Metallurgy, Georgia Institute of Technology
CORNELIUS S. MUIJE, F.G.A., G.G. and LILIAN E. MUIJE, G.G. Lili Gemologists, lnc., Louisville, Kentucky
Introduction green variety is limited. (l thru II) When the original East African Literature references on the other finds of transparent green grossularite varieties are frequent but very limited were made in the late sixties, the in content.U ,2,4 thru 12) , excitement created by this new gem Discussions with a number of overshadowed the simultaneous finds dealers who have bought and sold large of colorless, yellow, and golden gros- quantities of Tanzania and Kenya gros- sularite. All of these varieties were sulars, and who have subjectively esti- discovered in excellent gem quality. mated the relative quantities of the Dr. H. Bank, in 1968, was the first to various varieties of grossularite, lead to describe colorless grossularite.i.U The these conclusions: green variety was promotionally intro- Yellow or golden grossularite is duced to the public by Tiffany & Co. substantially more rare than green in September, 1974, under the name grossularite, and colorless grossularite "tsavorite.,,(2) But the other varieties is the rarest of all. received no special names. Tsavorite Furthermore, according to these was named after the Tsavo National dealers, all colorless grossularite is Park, a park in Kenya located near the from finds in Tanzania. No more than discovery site in Tanzania. Even ten cut colorless grossulars in excess of though tsavorite has received wide- four carats are known to exist, none spread recognition and acceptance, in- larger than eight carats. Allor most of formation in the literature on this these stones are in the possession of
162 GEMS & GEMOLOGY private collectors. It is noteworthy have greenish tints and the diamonds that even the Smithsonian Institution of the diamond master set, yellowish by May 1978 did not have a colorless tints. During the color grading of a grossular on display. Areasonable esti- great number of stones, consistent mate is that there are about forty results were always obtained by differ- stones ranging from two to four carats ent gemologists working inde- and a few hundred between one and pendently. two carats. The total weight of the The best GIA grade found among stones below the size of one carat hundreds of small stones evaluated was should be about six-hundred to eight- G/B. And the best GIA grade found hundred carats. among the many stones over one carat We were fortunate to acquire a was I. Typically, the CIA grade for a large number of cut East African stone was K or L. We classify an N gemstones for examination. All were grade and grades of darker tones as reported to be grossulars and to have near colorless. Even with a small bril- been found in 1968 and without a liant cut, an N rated gem is noticeably doubt from the early finds in Tan- colored to the naked eye. zania. They were machine cut and in a Many dealers offer as colorless gros- variety of colors. Some were pure sulars, stones that will grade between colorless. All were transparent and of Q and ~. In fact, all Kenya "colorless" excellent gem quality. The examina- stones we have seen and graded have tions presented here were restricted to been darker than 0 with clearly ob- stones .which were pure colorless or servable green or other tints. This may green. The green stones had various be why most prominent dealers claim intensities ranging from near colorless that all colorless grossulars come from with only a touch of green, through Tanzania despite the fact that in the medium green, to deep green. literature there are references to color- Our main objective was to deter- less grossularite from Kenya, Canada, mine the characteristics of the color- California, and Mexico.(2,5,9,11) less stones, so that it could be verified that they were in fact colorless gros- sulars. Such information on the Tan- Chemical Analyses zania colorless grossulars was not in Three colorless gemstones and one the literature. Another objective was near-colorless gemstone were chemi- to better understand the cause of cally analyzed by the nondestructive green color in grossularite. technique of X-ray energy spec- trometry (XES). Their GIA grades Colorless Defined were .Q/H, !, M, and Q, respectively. The colorless and near-colorless Twelve tsavorites which had been gemstones were color graded against a selected to form a color suite with GIA-Graded Diamond Master Set. colors ranging from very light green to Such grading presented no difficulties, deep green, in small increments, were even though the gemstones tended to also chemically analyzed. All stones
SUMMER 1979 163 were brilliant cut and about 3 rom in The X-ray spectrum is induced in the diameter. sample by placing it in the electron We used a Jeol Electron Microscope microscope and bombarding the Model 100-C with a Kevex-Ray Model sample with electrons. For accuracy of 5100 Energy Spectrometer for the chemical composition, tsavorite, cal- analyses. This Kevex 5100 system cium carbonate, and some high-purity identifies and provides a quantitative oxides were used as comparative analysis of the chemical elements of a standard. It is recognized that oxygen sample by measuring in the spectrum and lighter elements cannot be de- the dispersive X-ray energy that is a tected by this method. Therefore, it unique characteristic of each element. was assumed that oxygen was present
TABLE I CHEMICAL COMPOSITION OF GROSSULARS (X-Ray Energy Spectroscopic Analyses)
Sample No. Weight Percent - Color -- CaO A 12O:J Si02 MgO Ti02 MnO FeO ----Cr203 V203 1 38.5 21.2 39.7 0.1 0.3 0.1 0.1 NO--- NO Colorless, G/H 2 38.0 20.9 40.3 0.4 0.2 0.1 0.1 NO NO Colorless, I 3 39.7 19.9 39.5 0.1 0.4 0.2 0.1 NO 0.06 Colorless, M 4 32.3 22.8 43.3 1.0 0.3 0.1 0.1 NO 0.05 Near Colorless, 0 5 35.1 21.3 41.9 0.9 0.4 0.1 0.1 0.09 0.10 VL Green 6 33.9 22.4 42.1 0.9 0.4 0.1 0.1 NO 0.09 L Green 7 34.5 21.5 42.8 0.5 0.3 0.2 0.1 NO 0.11 L Green 8 34.5 20.5 43.0 0.3 0.2 0.6 0.1 0.09 0.7 M Green 9 34.8 21.7 41.6 0.1 0.3 0.4 0.03 0.06 1.0 MO Green 10 33.6 21.6 42.2 0.4 0.4 0.6 0.1 0.09 1.0 MO Green 11 33.7 21.1 42.5 0.7 0.2 0.5 0.04 0.10 1.1 MO Green 12 35.5 20.5 41.9 0.4 0.2 0.3 NO 0.02 1.2 MO Green 13 34.5 20.5 42.5 0.3 0.3 0.5 0.1 0.12 1.2 MOGreen 14 34.6 21.0 42.0 0.3 0.2 0.4 0.03 0.14 1.4 0 Green 15 349 20.5 41.9 0.4 0.3 0.4 0.1 0.15 1.4 0 Green 16 36.3 20.6 40.3 0.2 0.2 0.2 NO 0.10 1.5 o Green
•• Ideal grossular is: 37.36% CaO, 22.63% A 1203, 40.01 % Si02.
- Grossulars are ranked according to GIA diamond color grade and depth of green color; VL = Very Light; L = Light; M = Medium; MO = Medium to Deep; D = Deep.
-.. Not Detectable
164 GEMS & GEMOLOGY in such amounts as to form perfect purities were all weight percentages metal oxides, and that elements of calculated from the chemical composi- atomic number less than oxygen were tions, assuming that there were twelve not present. oxygen atoms to every eight metal The determined chemical composi- atoms. We recognized that the oxygen tions of all gemstones, ranked accord- atoms associated with the impurity ing to color grade or intensity of green metal atoms were not impurity oxygen color, are given in Table I in terms of atoms but oxygen atoms of the basic, oxide weight percentages. Their chemi- ideal crystal. cal compositions in terms of the The X-ray energy spectrum of color- number of metal atoms per twelve less Sample No.2 is shown in Fig. 1. For oxygen atoms are given in Table II. comparison, the X-ray spectrum of Table II is a convenient way to com- tsavorite with a deep green color is pare the measured chemical composi- shown in Fig. 2. The similar nature of tions of the stones with that of color- the silicon, aluminum, and calcium less, ideal grossularite. Colorless, ideal peaks of the colorless gemstone and grossularite has a stoichiometric tsavorite may be seen by comparing formula of Ca3 Al2 (SiO 4)), indicat- Fig. lA with Fig. 2A between 0.5 and ing that there are three calcium atoms, 3.5 KeY. In Figs. lB and 2B the 2.8 to two aluminum atoms, and three silicon 7.9 KeY range is magnified to show (a) atoms for every twelve oxygen atoms. the absence of vanadium and chro- As is customary, it was accepted when mium in the colorless gemstones, (b) constructing Table II that magnesium, the vanadium and chromium peaks in manganese and iron replace calcium; the tsavorite, and (c) the lower manga- that chromium and vanadium replace nese and iron peaks in the colorless aluminum; and that titanium replaces gemstones. silicon. There is doubt that this can be The statement is often made .that applied strictly.(13,14) However, con- pure or nearly pure garnets do not sidering the low concentrations of exist in nature. However, the colorless these impurity elements, the results grossulars we examined are very close would not be significantly different. to the pure end member grossularite. The chemical analyses showed that They are probably as pure a grossu- the chemical compositions of the larite as can be found in a natural colorless or near-colorless gemstones deposit. The purest grossular pre- were similar to those of the tsavorites viously reported in the literature had a and very close to the calculated, purity of 98.9%.(3) However, the theoretical composition of colorless, color of this stone was pale yellowish ideal grossularite. Sample No. 1 and green rather than colorless. This is in Sample No.2, the two most colorless agreement with the lower purity ob- stones, were found to be calcium- served. We calculated the purity of this aluminum silicate of 99.6% and 99.5% grossular from the reported chemical purity, respectively. Colorless Sample composition and used the same No. 3 had a purity of 99.4%. These assumptions as before. The reported
SUMMER 1979 165 C'?C'?C\J <:0 "'"0) .•....1"'" r--.•....,<:0 0:>q q·z°I0:> «0000) 000 ~IC\JOO C\J ~c:io""': MoM
C'?L[)C\J010 C\J0) .•....1 C\J r--.•....,<:0 «00000) «0000) .•....0 .•.... ~,C\ioooC\i ""':00""': MoM o C'?C'?O\CD r-, <:0.•....ICD r--.•....j<:O <:0000<:0 ~oqo:> .•....0 .•.... 41 ~IC\ioooC\i ~ OO~ MoM en 0) C'?C'?.•....,CD C'?r-, .•....1 ...... C\J.•....IC'? a:: r--O 00 <:0 ~qClO:> C\J0 C\J :Jen o T"".•....T""IC'? <:0 01<:0 0) T""IO .•.... 1 ~ooO'r"" ° q>oq en ~zz~ 0 Moo oM .•.... T"" C\JOC'? 0> CO Q) «ss s « :2 ~I O~~LL «>0 iiif= 166 GEMS & GEMOLOGY A Q) c:: c:: ..c.co o Q; Ti a. .I!l Ca Ca c:: 5 Full scale = 1,000 counts. o >- ."= CIlc Q) +-' c:: B Quantum energy, KeV Figure 1. X-ray energy spectrum of colorless grossular (Sample No.2). SUMMER 1979 167 f!:J1Iscale =: ~~/000 counts, A Q) c::: c::: co .J::. ...U Q) 0- •....ct) c::: :::l 0 U •....>- 'c;; c::: •....Q) c::: B Ti Nln+ C"r V+Ti Vt\Cr Figure 2. X-ray energy spectrum of tsevorite. 168 GEMS & GEMOLOGY chemical composition had been de- which contained 0.09% chromium- termined by a technique of X-ray oxide and 0.10% vanadium-oxide, was spectrometry comparable to our own. lighter green than Samples No.6 and The results of the chemical No.7, which contained no chromium analyses, augmented by our physical and about the same amount of and optical property determinations, vanadium. Except for the chromium- further showed that the colorless and oxide content, the chemical composi- near-colorless stones formed an ex- tions of these three grossulars were panded color suite with the tsavorites. very similar. Therefore, it appeared This expanded suite consisted of the that chromium did not increase the sixteen grossulars listed.in Tables I and intensity of the green color,· but II, with tones and intensities of green rather, had a slightly negative, or ranging all the way from QI!!.colorless lightening, effect. The chromium- to deep green. This greatly facilitated oxide content was mostly not greater the evaluation of the effect of chemi- than about 0.1%. It was 0.15% as a cal composition on color. maximum. Therefore, the effect of A direct relationship was found chromium on the green color of the between the vanadium content and the grossulars of this investigation seemed intensity of the green color of the to be insignifican 1. grossulars, as may be seen in Table 1. Except for influencing the tones of No vanadium or chromium was de- the colorless and near-colorless stones, tected in the two most colorless no remaining element at the levels stones. And as the vanadium-oxide present seemed to influence color sig- content gradually increased to 1.5%, nificantly. the intensity of the green color gradu- The magnesium-oxide content ally increased to deep green. In the varied erratically between 0.1 % and colorless grossular that graded an M 1.0% in the color suite, with' no color, the vanadium-oxide content was significant effect on the green color. 0.06%, but no chromium was detected In the colorless stones, the in this stone, either. In fact, no chro- manganese-oxide content was only mium was detected in the first seven 0.1 % or 0.2%. The manganese-oxide grossulars of the color suite except for contents of the remaining stones Sample No.5, and the color of Sample varied erratically between 0.1 % and No.6 was already rated as light green. 0.6%. Within this range, the manganese Therefore, chromium could not did not noticeably modify the green account for, nor contribute to, the color of the tsavorites. However, the green colors of these grossulars. green color was always of high in- Unfortunately, there was no stone tensity at the high levels of manganese. in our color suite that contained This may have obscured any effect of chromium without vanadium. There- manganese on color. fore, we were not able to determine The titanium-oxide content was the effect of chromium without always in the range of 0.2% to 0.4%, vanadium. However, Sample No.5, and the iron-oxide content was never SUMMER 1979 169 more than 0.1%. Under these condi- UVSL-25 Combination Mineralight tions, no noticeable influence of these Lamp using an ultraviolet viewing cabi- elements on the green color of the net. tsavorites was observed. Specific Gravity: The specific gravity None of the gemstones examined of the colorless grossularite was 3.62 had a measurable nickel content. as determined by the direct-weighing method. The small colorless stones of Optical and Physical Properties our color suite were too small for The optical and physical property accurate, reliable measurement of determinations were restricted to the specific gravity. Therefore, we made colorless gemstones. We evaluated bire- three different batches of large grossu- fringence, refractive index, ultraviolet lars, each consisting of three stones, fluorescence, and specific gravity. In and determined the joint specific addition, some gemstones were micro- gravity in each case. The weight of scopically examined for inclusions. each batch was between four and five Birefringence: None of the colorless carats. And the measured specific gemstones examined exhibited bire- gravities were 3.61,3.62, and 3.62. fringence under the GEM Illuminator Characteristic Inclusions: We ex- Polariscope. All were found to be amined many of the colorless grossu- singly refractive but with medium lars under the microscope at magnifi- strain. Refractive Index: The small colorless cations up to 60x for inclusions. To gemstones, Sample No.1, Sample No. date we have not been able to cata- 2, and Sample No.3, all showed a gorize any of the inclusions we ,ob- refractive index of 1.731 to 1.732. served as "characteristic." One stone Their GIA color grades were QlH, 1. exhibited short stubby inclusions (Fig. and M, respectively. Ten colorless gem- 3). Another showed two long parallel stones which were one to two carats in needles (Fig. 4). A third had short size showed an average refractive index needles and angular inclusions (Fig. 5). of 1.732. Five of these measured And a fourth displayed a plane of 1.731, and five measured 1.732. Their small "fingerprint-like" inclusions GIA color grades ranged frorn ], to M. (Fig. 6). All refractive indices were determined Further Discussion with the GEM Duplex II Refrac- Bank reports the refractive index of tometer. colorless grossularite to be in the range Ultraviolet Fluorescence: Under both of 1.732 to 1.736.(1) Eppler reports short and long wave length ultraviolet pure synthetic grossularite as having a radiation, the colorless grossularite refractive index of 1.734.(5) And fluoresced a weak to moderate golden- Mackowsky gives 1.735 for the refrac- yellow visible light. The color of this tive index of pure grossularite.(1) light could also be called yellow with a Webster reports 1.738 to 1.744 for touch of orange. The ultraviolet trans arent green grossularite (tsavor- fluorescence was determined with the ite).(1) The GIA shows 1.736 to 1.742 170 GEMS & GEMOLOGY Figure 3. Colorless grossular with short Figure 4. Colorlessgrossular with long stubby inclusions (BOx). Arrow points to parallel needles (60x). Each arrow points to cluster of inclusions. a long needle-shapedinclusion. Figure 5. Colorless grossular with short Figure B. Colorless grossular with small needles and angular inclusions (BOx). Two fingerprint-like inclusions (60x). Arrows arrows point to short needles, and one point to plane with small inclusions in arrow points to one of numerous angular "fingerprint" pattern. inclusions. SUMMER 1979 171 for green transparent grossularite chromium is not essential. Further- (tsavorite ). more, it appears that chromium at the The colorless gemstones we ex- low levels found in tsavorite during amined showed an average refractive both investigations - 0.3% chromium index of 1.732. Therefore, our experi- oxide or less - does not have any mental results are in agreement with significant effect on green color, the data reported by Bank for color- neither in a positive nor negative sense. less grossularite, even though some of our colorless stones measured 1.731, the lowest refractive index ever re- Conclusions ported for colorless grossularite. It The chemical compositions and the should be noted that the measured optical and physical properties, de- refractive indices of our colorless gros- termined during this investigation, sulars were at the lower end of the verify the reports that the colorless 1.732 to 1.736 range reported by gemstones discovered in Tanzania in Bank. And they were lower than the the late sixties are colorless grossulars. 1.734 reported for pure synthetic gros- Such grossulars are exceedingly rare sur~rite by Eppler and lower than the and have characteristics that qualify 1.735 reported for pure grossularite by them as desirable gemstones. It Mackowsky. Since we define colorless appears that most, if not all, "color- rather tightly, it appears that many of less" grossulars from Kenya are not the stones examined by the other colorless, but "near colorless" with investigators were near colorless rather clearly observable green or other tints. than colorless. This seems especially Similar to diamonds, colorless gros- true since the GIA shows a refractive sulars can be color graded accurately index of 1.736 to 1.742 for green and with consistent results. A GIA- transparent grossularite (tsavorite). Graded Diamond Master Set is suitable Previous work by Switzer suggested for such grading. that the green color in tsavorite was The beautiful green color in related to the vanadium content and tsavorite is caused by vanadium alone. probably in part to the chromium Chromium is not essential. In the content.(3) However, both vanadium absence of chromium, a vanadium- and chromium were always present in oxide content of 0.1 % already causes a the four green grossulars Switzer ex- light green color. And a vanadium- amined, so that the effectiveness of oxide content of 1.5% causes a deep each of these elements alone to impart green color. Presumably, the tsavorites green color could not be determined. that black out have higher vanadium- Also, an interaction of vanadium and oxide content. It appears that chromium could not be ruled out. chromium at the levels found in The present work clearly showed tsavorite during this investigation has that vanadium is the effective element no significant effect on green color. that gives tsavorite its intense, beauti- The chromium-oxide contents were ful green color. The presence of about 0.2%, or less. 172 GEMS & GEMOLOGY References 8. Farn, A. F., Notes from the Labora- tory, Journal of Gemmology, Vol. XV, 1. Webster, R., GEMS: Their Sources, No.1, 1976, pp 8-10. Descriptions and Identification, Butter- 9. Arem, J. E., Color Encyclopedia of worth & Ltd., London, 3rd edi- ce., Gemstones, van Nostrand Reinhold tion, 1975, p 156. Company, N.Y., 1977, p 55. Bridges, C. R., Green Grossularite 2. 10. Mitchell, R. K., African Grossular Gar- Garnets ("Tsavorites") in East Africa, nets; Blue Topaz; Cobalt Spinel; and Gems & Gemology, VoL XIV, No. 10, Grandierite, Journal of Gemmology, Summer 1974, pp 290-295. Vol. XV, No.7, 1977, pp 354-355. 3. Switzer, G. S., Composition of Green II. Schumann, W., Gemstones of the Garnet from Tanzania and Kenya, World, Translated by Stem, E., Sterling Gems & Gemology, Vol. XIV, No. 10, Publishing Co., Inc., N.Y., 1977, p Summer 1974, pp 296-297. 106. 4. Bank, H., From the World of Gem- 12. Schlossmacher, K., Edelsteine und stones, Translated by Rutland, E. H., Perlen, E. Schweizerbart'sche Verlags Pinguin Verlag, Innsbruck, 1973, p 73. Buchhandlung, Stuttgart, 5th edition, 5. Eppler, W. F., Practische Gemmologie, 1969, p218. Ruhle-Diebener Verlag KG, Stuttgart- 13. Hartman, P., Can Ti4+ replace Si4+ in Degerloch, 1973, p 195. silicates? Mineralogical Magazine, Vol. 6. Liddicoat, Jr., R. T., Handbook of 37, No. 287, 1969, pp 366-369. Gem Identification, Gemological Insti- 14. Huggins, F. E., Virgo, D., and Hucken- tute of America, Los Angeles, 10th h olz , H. G., Titanium-containing edition, 1975, p 244. silicate garnets. I. The distribu tion of 7. Anderson, B. W., Gemstones for Every- Al Fe3'=!- and Ti4+ between octahedral man, Faber & Faber, London, 1976, and tetrahedral sites, American Min- pp 187-188. eralogist, Vol. 62, 1977, pp 475-490. SUMMER 1979 173