Study on the effect of heat treatment on amethyst color and the cause of coloration

The effect of heat treatment on amethyst color was studied from a new perspective of chromaticity of gemstones and the cause of amethyst coloration was discussed based on the results of X-ray diffraction, ultraviolet–visible spectroscopy. The results show that the amethyst color has no significant relationship with cell parameters but the crystallinity index decreases as temperature rises. The absorption band at 545 nm in the UV–visible spectrum can be related to a charge-transfer transition of Fe3+ and O2−, which has a significant relationship with amethyst lightness and chroma. The color at different temperatures can be divided into three stages: The amethyst stage with temperature below 420 °C, the prasiolite stage with temperature between 420 and 440 °C where the color center is the most unstable, the citrine stage with temperature above 440 °C. The color change degree of heated amethyst is related to its initial color. When the initial color is darker, the color difference of heated amethyst is larger, and the easier it is to change the color after heat treatment. A more appropriate heating temperature to obtain citrine by heating amethyst is about 560 °C.

The CIE DE2000 (1:1:1) formula (formula 3) 32 is used to calculate the color difference (ΔE 00 ) of amethyst under different heating temperatures. ΔL*, ΔC* and ΔH*, represent the difference of lightness, chroma and hue angle respectively. R T is a conversion function to reduce the interaction between chroma and hue in the blue area. S L , S C , and S H are functions to calibrate the absence of visual uniformity of the CIE LAB formula. K L , K C , K H are correction parameters of the experimental environment and K L = 1, K C = 1, K H = 1.
We used 20 pieces of natural amethyst from South Africa (Fig. 1), whose colors display continuously from pink violet to blue violet. To get the color parameters effectively, all of them are cut into a polished round plane with a diameter of 7 mm and a thickness of 4 mm from different crystallographic orientations. There is no obvious inclusion in the inner part of the samples observed by naked eyes and some samples have straight or angular color bands. In this paper, the cause of amethyst coloration is discussed based on the results of the X-ray diffraction (XRD) and UV-Vis spectroscopy measurement of amethyst at different temperatures. Based on CIE1976 L*a*b* uniform color space system, the effect of heat treatment on amethyst color is studied and a more appropriate temperature range is defined to obtain the best color of citrine by heating amethyst.

Results
X-ray diffraction measurement. When the temperature arrives 573 °C, the amethyst will change from α-quartz to β-quartz and the color of amethyst cannot be recovered by γ ray irradiation 33 . It may indicate that when the heating temperature is higher than the phase transition temperature, the color center of amethyst may be permanently destroyed 34 . In order to find out whether the color of amethyst at different heating temperatures is related to phase transition and crystal structure, four samples were cut from the same raw stone. One of them was not heat treated, and the other three were heated to 400 °C, 500 °C, and 600 °C. After being heated, they became light violet, yellow and milky white respectively. Then we tested the samples by XRD measurement and the pattern is as shown in the Fig. 2.
Four samples are all quartz, which matched the standard card PDF Card-46-1,045. It can be concluded that the crystal plane index, d-spacing (Å) and the unit cell parameters of samples heated at different temperatures are not significantly changed. As shown in Table 1, the cell volumes of the samples are decreasing to 111.61 (Å 3 ) as the temperature rises to 500 °C. When it reaches 600 °C, the cell volume begins to increase to 115.81 (Å 3 ) and the increase of cell volume could be related to the crystallization of nuclei of ferrous silicate 34 . www.nature.com/scientificreports/ The crystallinity index (CI) of samples slightly decreases from 10 to 9.55 with the increase of heat treatment temperature. The crystallinity index CI 35 can be calculated by a and b values of the (2132) peak as follows (F is the scaling factor of 1.12): When the temperature is above 600 °C (higher than the phase transition temperature of amethyst), the amethyst becomes milky and turns lighter. It can be related to the destruction of the S 1 centers ([Fe 3+ O 4 /M i + ]) and I centers (Fe 3+ in an interstitial site) 34 and the aggregation of water to form "water bubbles". The diameter of the "water bubbles" is from 20 to 100 nm, which can cause Rayleigh scattering of light 36 .  www.nature.com/scientificreports/ UV-Vis spectrum analysis. The absorption bands at different wavelengths of amethyst UV-Vis spectra correspond to the relevant color centers 12,16,37 . Therefore, the samples heated at different temperatures were tested by UV-Vis spectroscopy. The UV-Vis spectra of amethysts had a broad absorption bands at 545 nm and 345 nm. As the temperature rises, the violet color gradually became lighter and the absorption bands at 545 nm and 345 nm were gradually weakened. The color of amethyst became almost colorless and the absorption bands at 545 nm almost disappeared at 420 °C (Fig. 3b). Lehmann 9 believed that amethyst color is related to the intensity of the optical absorption near 545 nm and ascribed the color to a charge-transfer transition of Fe 3+ and O 2− . Rossman 11 reported that the violet coloration is observed only in samples where specific sites are occupied by Fe 3+ ions. Czaja 7 conducted Mössbauer spectroscopy measurements on quartz with radiation and heat treatment and the results showed that only Fe 3+ was found in the prasiolite and amethyst. According to the result of the XANES evidences and DFT structural features, Di Benedetto 38 also suggest that numerous local distortions are occurring as a consequence of the presence of Fe 3+ variably compensated by H + , when Fe replaces Si in its tetrahedral site in amethyst. They provided the evidence for the existence of Fe 3+ -H + complex structure in amethyst by calculation and proposed the structural model of the Fe substitutional sites (Fig. 3a).
According to Halliburton, the exposure of quartz to ionizing radiation may also lead to the creation of a hole (H + ) in a non-bonding p orbital of an oxygen atom adjacent to substitutional aluminum 39 . Since H + is a small and light ion, it can easily diffuses through the quartz lattice and may occupy an interstitial sites previously occupied by alkali ions ( M + i ) 40 . The color center formation of could be related to the following mechanism 41 : www.nature.com/scientificreports/ A similar centers process of Amethyst have been proposed when iron is substituting for silicon in the quartz lattice as reaction (6) 12,42 . The alkali ion released in reaction may eventually react with an electron, leading to the creation of an interstitial neutral atom as reaction (7) 14,41 : Citrine can be obtained by heating amethyst at a temperature of about 500 °C 10,40 . As for the cause of the color of citrine, During the heat treatment, the transformation of amethyst to citrine is related to the increase in the concentration of interstitial Fe 3+ related defects and the precipitation of iron particles in the quartz lattice 17,34 . According to Stock and Lehmann, the size of these iron particles is about 100 nm 43 . The brown color is partly due to these inclusions and partly due to a shift of the charge-transfer band into the visible as S 1 centers are converted to less constricted I centers and the charge-transfer is as following (Fe i represents an Fe ion at an interstitial site) 34 : In order to find out whether the absorption band at 545 nm is related to the color of amethyst, we analyzed the color parameters and the peak area (X) of the absorption band at 545 nm which is obtained by heating the samples below 400 °C. To get the average peak area ( X ) of each category, the peak area (X) is divided into 7 categories according to their initial lightness (L*) range from 30 to 70 (Table 2).
According to Fig. 3c, as the average peak area ( X ) at 545 nm increases, the average lightness ( L * ) decreases and the average chroma ( C * ) increases simultaneously. The correlation (R 2 ) between them is 0.944 and 0.955 respectively, which means the absorption band at 545 nm has a significant relationship with the color of amethyst. The larger of the peak area at 545 nm is, the darker the color of amethyst. However, the hue angle (h°) has no significant relationship with the peak area (X) at 545 nm, which is because the hue angle of amethyst has no obvious change when temperature is below 400 °C. The relationship between L * , C * and X as follows: The effect of heat treatment on amethyst color. The temperature of heat treatment has a great effect on amethyst color. Instead of observing by naked eye, we quantified the amethyst color based on CIE1976 L*a*b* uniform color space system. The distribution of the color parameters of the amethyst at different temperature in CIE1976 L*a*b* uniform color space is shown in the Fig. 4a. After heat treatment, the color of amethyst is stable and will not change obviously in dark conditions. The average color parameters and the range of a* (R a ), b* (R b ) and color difference (ΔE 00 ) of the samples between current temperatures and the previous adjacent temperature is shown in Table 3.
When the temperature is lower than 360 °C, the amethyst color does not change significantly. As the temperature rises to 380 °C, the violet color begins to fade out and gradually becomes colorless. Then it changes to light green at 420-440 °C. Therefore, prasiolite can be obtained by heating amethyst from 420 to 440 °C, such as the N-05 sample at 440 °C (Fig. 4b). As the temperature continues to rise, it slowly changes to yellow and gradually L * = 108.601 ± 6.094SEM − 1.106 ± 0.110SEMX (R 2 = 0.944) (10) C * = −11.470 ± 3.057SEM + 0.806 ± 0.071SEMX (R 2 = 0.955) www.nature.com/scientificreports/ deepens. Since some samples has color bands, we have noticed that they have different degrees of fading during the heating process at different areas of the color bands. For example, when the samples are heated to 420 °C, some areas become colorless and other areas are still light violet. It seems that the degree of color change of amethyst after heating may be related to its' initial color 44 . However, when the temperature arrives 420 °C, the hue gradually shifts from green towards yellow as the temperature rises. The reason for the hue change can be explained by the changes of a* and b* (Fig. 5a). The positive and negative of a* axis represent red and green respectively. a* shows a trend of decreasing from positive to negative and then slowly rising to 0 with increasing temperature, which causes the hue to change from red to green and then return to red. The positive and negative of the b* axis represent yellow and blue respectively, b*  www.nature.com/scientificreports/ keeps rising from negative to positive which causes the hue to change from blue to yellow. When the temperature reaches 420 °C, R a , R b and ΔE 00 shown in Table 3 respectively reach the maximum value of 26.92, 28.53, 18.22 (Fig. 5b). It seems that the color center of amethyst is the most unstable at 420 °C. Besides, we noticed that the color difference at 580 °C is 2.65 and it's different from that of 560 °C (0.98) and 600 °C (0.63). We believe that The relationship between R a , R b , ΔE 00 and temperature. R a , R b and ΔE 00 reach the maximum value at 420 °C. (c) The relationship between the degree of color change of amethyst and the initial color. ΔE 00 represents the color difference of each categories before and after heat treatment. When the initial color is darker, the amethyst color is easier to change after heat treatment. www.nature.com/scientificreports/ the suddenly increase of color difference at 580 °C can be related to the irreversible destruction of color center when the temperature is above the phase transition temperature (573 °C) of amethyst 34 .
In order to investigate whether the degree of color change of amethyst after heat treatment is related to its initial color, the amethyst color is respectively divided into three categories according to their initial lightness and chroma: L 1 * (20-35), L 1 * (35-45), L 1 * (45-55) and C 1 * (30-40), C 2 * (40-45), C 3 * (45-55). ΔL* and ΔC* respectively represent the difference between the values at current temperature of each categories and the corresponding initial values. The relationship between ΔC*, ΔL* and temperature are shown in the Fig. 5c. It can be concluded that the higher the initial chroma is, the greater the difference in chroma and the lower the initial lightness is, the greater the difference in lightness after heat treatment. Taking the change of color difference into account, the ΔE 00 of each category before and after heat treatment are calculated (Fig. 5c.). It can be known that the when the initial color is darker, the color difference of samples after heat treatment is larger. Therefore, the initial color of amethyst has an effect on its color change after heat treatment. The darker the initial color is, the easier it is to change the color after heat treatment, which explains why different areas of the same sample have different degrees of fading after heating.
The b* and C* of samples mean value at different temperatures are plotted in Fig. 6. C* first decreases and then rises with b* and the change of C* is divided into two parts: when the temperature is lower than 420 °C, C* is negatively correlated with b*; When the temperature is higher than 440 °C, C* is positively correlated with b*, and the relationship between C* and b* is as follows: In summary, the color change of amethyst can be divides into three stages with temperature: the amethyst stage with temperature below 420 °C, violet gradually weakens and the color center gradually destructs with the temperature; the prasiolite stage with temperature between 420 and 440 °C, which is the stage of transition from amethyst to citrine and the color center is the most unstable at 420 °C; the citrine stage with temperature above 440 °C, yellow gradually deepens with the temperature.
Color grading of heat-treated amethyst. In order to get a more appropriate temperature range of amethyst heat treatment, the color parameters in the citrine stage (when temperature is above 440 °C) were classified. The standard of a more appropriate temperature range is the largest number of samples in the best color category after classification. Based on CIE1976 L*a*b* uniform color space system, the classification uses two mathematical methods: K-means clustering analysis and Fisher discriminant analysis. These two methods have been verified in peridot 24,25 , tourmaline 26,27 , jadeite [29][30][31] and other colored minerals. K-means Cluster analysis is a statistical analysis technology that divides the research objects into relatively homogeneous groups. Fisher discriminant analysis is one of the important methods of multivariate statistical analysis 24 .
In this paper, three independent color parameters L*, a* and b* are selected for K-means cluster analysis. The clustering effect is the best (sig < 0.001), when the number of categories is 5 (Table 4), which shows that the classification effect is obvious and the accuracy of color data updating is 98.4%. Therefore, the scheme of dividing citrine color into five categories is effective and feasible. Fisher discriminant function is used to test the www.nature.com/scientificreports/ clustering effect, and the discriminant function corresponding to the color of five categories of citrine color is obtained as follows (Table 5): For citrine color appreciation, L* should be considered as the primary factor because it can correspond visual effect typically and directly, and then take hue and chroma into account together comprehensively 28 . According to the results of K-means cluster analysis and Fisher discriminant analysis, by imitating GIA's colored diamond grading system 15 , the color in the citrine stage were divided into five grades: Fancy Intense, Fancy Deep, Fancy Dark , Fancy and Fancy Light, when hue angle ranges from is 93.3° to 121.4° (Fig. 7). The best grade "fancy intense" has the lightness ranges from 66 to 90 and the chroma ranges from 20 to 30; the second grade "fancy" has the lightness ranges from 70 to 90 and the chroma ranges from 6 to 20; the third grade "fancy deep" has the lightness ranges from 60 to 70 and the chroma ranges from 12 to 30; the fourth grade "fancy dark" has the lightness ranges from 60 to 70 and the chroma ranges from 6 to 12; the fifth grade "fancy light" has the lightness ranges from 80 to 90 and the chroma ranges from 6 to 12. The temperatures with the largest number of samples in the Fancy intense category are 560 °C and 600 °C. Because 600 °C is higher than the phase transition temperature of α-quartz and milky white color may appear after heat treatment, it is suggested that a more appropriate temperature range to obtain the best color of citrine by heating amethyst is about 560 °C.

Discussion
The color at different heated temperatures has no obvious relationship with cell parameters, but the crystallinity index CI of amethyst decreases as the temperature rises. When the temperature is above 600 °C (higher than the phase transition temperature of amethyst), the amethyst becomes milky and turns lighter, which can be related to the destruction of the color centers and Rayleigh scattering of light due to the aggregation of water.
In the UV-Vis spectra, the absorption band at 545 nm (which is related to a charge-transfer transition of Fe 3+ and O 2− ) has a significant relationship with the color of amethyst. The larger the band area at 545 nm is, the lower the lightness and the higher the chroma, which means the amethyst color will be darker.  Table 4. ANOVA of L*, a*, and b* of citrine color stage. www.nature.com/scientificreports/ The color of amethyst under heat treatment can be divided into three stages with temperature: the amethyst stage with temperature below 420 °C, violet gradually weakens with the temperature; the prasiolite stage with temperature between 420 °C and 440 °C, the color center of amethyst is the most unstable at 420 °C; the citrine stage with temperature above 440 °C, yellow gradually deepens with the temperature. When the temperature is over 600 °C, it may appear milky white. The increase of color difference at 580 °C can be related to the irreversible destruction of color center when the temperature is above the phase transition temperature (573 °C) of amethyst.
The color change degree of amethyst after heat treatment is related to the initial color. The darker the initial color, the larger the color difference of samples after heat treatment, and the easier it is to change the color. The color of citrine obtained from heating amethyst can be divided into five categories and a more appropriate heating temperature to obtain citrine is about 560 °C.

Materials and methods
Heating method. The heating instrument is the KSL-1100X-S miniature box furnace of Hefei Kejing whose heating element is resistance wire. The temperature control system adopts intelligent programmable control, and the accuracy is ± 1 °C with the heating power keeping 1200 W. In order to heat these samples evenly, they were buried in pure quartz powder and placed in an alumina crucible. The samples were heated 16 times in the oxidation atmosphere, and the temperature ranges from 300 to 600 °C and the temperature is increased by 20 °C for each heating time. After holding for 10 min, the samples were naturally cooled down to room temperature in the furnace and then taken out to test and the cooling rate was 1.5 °C per minute. X-ray diffraction. Amethyst was powdered into 200 orders in an agate mortar. After drying, it was tested by BRUKER D8 Advance from Germany in State Key Laboratory of Earthquake Dynamics, Institute of Geology, China Earthquake Administration. The test conditions are: Cu K-α radiation (wavelength = 1.54 Å); scan step, 0.02; 40 kV tube voltage; 40 mA tube current; test angle of 3°-90°; test time of 434 s; scan mode: continuous PSD fast.
UV-Vis spectroscopy. The UV-Vis spectra were tested by using UV-3600 UV-VIS spectrophotometer.
The test conditions are as follows: range, 300-800 nm; single scanning mode; high scanning speed, sampling interval, 0.5 s; beam mode, double beam; slit width, 2.0 nm; slit program, normal; grating conversion wavelength, 720.00 nm; detector conversion wavelength, 900.00 nm; S/R conversion, standard; automatic lock the detector; step correction.
Colorimetric analysis. The color parameters were measured by X-Rite SP6 spectrophotometer in a standard illumination box with a fluorescent lamp (CCT 6504K, PHILIPS MASTERTL-D 90 De Luxe 18W/965, Holland) and the background was N9 gray level of Munsell neutral color chips. In order to obtain accurate color parameters, the average value of each sample was taken after three tests. The X-Rite SP6 spectrophotometer collects the surface reflection signal of the sample through the integrating sphere and then converts the signal into the color parameters. The test conditions are: reflection method, not including the specular reflection, D65 www.nature.com/scientificreports/