*3.1. Optical and SEM Observations*

The optical and SEM images are shown in Figure 2 together with the EDX analysis. A close inspection of the patterns revealed that in addition to the familiar eye and dark and light stripes, there were etched rings, as is apparent from Figure 1. Observation under the optical microscope did not **3. Results and Discussion** 

*3.1. Optical and SEM Observations* 

reveal any clear boundaries between the white ring and the dark background. Figure 2a displays the photographs of the one-eye and two-eye side of the bead together with the optical microscopy, and Figure 2b displays the SEM images showing a representative etched ring. These rings were all over the surface. In addition, from the slice in Figure S2, one could see the white region extended considerably into the bead; this confirms the report of Ebbinghouse and Winsten on the crafting of the white patterns that the white region extends significantly below the surface [2]. reveal any clear boundaries between the white ring and the dark background. Figure 2a displays the photographs of the one-eye and two-eye side of the bead together with the optical microscopy, and Figure 2b displays the SEM images showing a representative etched ring. These rings were all over the surface. In addition, from the slice in Figure S2, one could see the white region extended considerably into the bead; this confirms the report of Ebbinghouse and Winsten on the crafting of the white patterns that the white region extends significantly below the surface [2].

were etched rings, as is apparent from Figure 1. Observation under the optical microscope did not

*Heritage* **2020**, *3* FOR PEER REVIEW 5

**Figure 2.** (**a**) The approximate boundary between the light and dark areas of the bead's pattern (depicted with a white dashed line), as observed using an optical microscope. (**b**) Optical (top) and scanning electron microscopy (SEM) (bottom) image of the etched rings that are all over the surface of the bead in both the light and dark regions and across the boundary. (**c**) EDX from two specified areas denoted 1 and 2 in the SEM (Electron High Tension (EHT): 30 kV). Dominant components are Si and O, as expected for agate. **Figure 2.** (**a**) The approximate boundary between the light and dark areas of the bead's pattern (depicted with a white dashed line), as observed using an optical microscope. (**b**) Optical (top) and scanning electron microscopy (SEM) (bottom) image of the etched rings that are all over the surface of the bead in both the light and dark regions and across the boundary. (**c**) EDX from two specified areas denoted 1 and 2 in the SEM (Electron High Tension (EHT): 30 kV). Dominant components are Si and O, as expected for agate.

One can also see from Figure 2b that the etched rings were about 1 mm in diameter and clearly exhibited cracks and crevices, as can be seen in both the optical and SEM images. EDX recorded in the selected area marked with blue and red rectangles identified Si and O as the dominant elements, with a tiny amount of Al in the flat region in Figure 2c. This finding immediately confirms the elemental composition of the Dzi bead was consistent with that of agate (quartz). One can also see from Figure 2b that the etched rings were about 1 mm in diameter and clearly exhibited cracks and crevices, as can be seen in both the optical and SEM images. EDX recorded in the selected area marked with blue and red rectangles identified Si and O as the dominant elements, with a tiny amount of Al in the flat region in Figure 2c. This finding immediately confirms the elemental composition of the Dzi bead was consistent with that of agate (quartz).

### *3.2. XRD 3.2. XRD*

Figure 3 shows a comparison of the lab and synchrotron XRD recorded with (a) a powder sample of the Dzi bead and (b) the edge of a slice of the Dzi bead, respectively. The XRD using VESPERS was collected with a microbeam and an area-sensitive detector, yielding diffraction rings that were then converted to the 2θ display (Figure 1b and Figure S3) with the same scale as Co Kα for ease of comparison. From Figure 3a, one clearly sees that the Dzi bead powder exhibited a pattern characteristic of α-quartz (vertical lines), supporting the EDX results that the materials making up the Dzi bead were SiO2 quartz. There was, however, a small but noticeable peak marked by the arrow. This conspicuous peak was not seen in the synchrotron data (Figure 3b), where a comparison was also made with the XRD of graphite and CuO. The latter had no match, e.g., CuO had strong peaks at 2θ between 30° and 40° (not shown), strongly indicating that this peak likely came from graphite. It should be noted that both graphite and CuO appear black. Figure 3 shows a comparison of the lab and synchrotron XRD recorded with (a) a powder sample of the Dzi bead and (b) the edge of a slice of the Dzi bead, respectively. The XRD using VESPERS was collected with a microbeam and an area-sensitive detector, yielding diffraction rings that were then converted to the 2θ display (Figure 1b and Figure S3) with the same scale as Co Kα for ease of comparison. From Figure 3a, one clearly sees that the Dzi bead powder exhibited a pattern characteristic of α-quartz (vertical lines), supporting the EDX results that the materials making up the Dzi bead were SiO<sup>2</sup> quartz. There was, however, a small but noticeable peak marked by the arrow. This conspicuous peak was not seen in the synchrotron data (Figure 3b), where a comparison was also made with the XRD of graphite and CuO. The latter had no match, e.g., CuO had strong peaks at 2θ between 30◦ and 40◦ (not shown), strongly indicating that this peak likely came from graphite. It should be noted that both graphite and CuO appear black.

From the above analysis, the Dzi bead was made of SiO<sup>2</sup> (quartz) and appeared to be genuine. A graphite signature was detected in the powder sample but not from microdiffraction at the edge of a slice; presumably, the latter probed a much smaller area than the powder sample examined with a large beam and a relatively macroscopic sample, although possible carbon contamination during the grinding process cannot be ruled out.

grinding process cannot be ruled out.

*Heritage* **2020**, *3* FOR PEER REVIEW 6

From the above analysis, the Dzi bead was made of SiO2 (quartz) and appeared to be genuine. A graphite signature was detected in the powder sample but not from microdiffraction at the edge of a slice; presumably, the latter probed a much smaller area than the powder sample examined with a

**Figure 3.** (**a**) Representative X-ray diffraction (XRD) of Dzi powder ground from a broken piece recorded using lab X-rays from a Co anode (λ = 1.8 Å). The corresponding XRD pattern of quartz (PDF 46-1045) is also shown in purple vertical lines. A conspicuous peak not belonging to quartz is marked with an arrow, which is consistent with graphite (004) diffraction. (**b**) XRD from the edge of a slice of Dzi bead obtained with 8 keV X-rays using the VESPERS beamline converted to the same 2θ scale as Co Kα for ease of comparison. The insert shows the diffraction rings from the Pilatus detector, which has been converted to the intensity-2θ pattern as shown. The XRD pattern of graphite (PDF 41-1487) is also shown as vertical red lines. The conspicuous peak seen in the powder sample is absent. *3.3. XRF, Imaging and Micro-XANES Analysis*  **Figure 3.** (**a**) Representative X-ray diffraction (XRD) of Dzi powder ground from a broken piece recorded using lab X-rays from a Co anode (λ = 1.8 Å). The corresponding XRD pattern of quartz (PDF 46-1045) is also shown in purple vertical lines. A conspicuous peak not belonging to quartz is marked with an arrow, which is consistent with graphite (004) diffraction. (**b**) XRD from the edge of a slice of Dzi bead obtained with 8 keV X-rays using the VESPERS beamline converted to the same 2θ scale as Co Kα for ease of comparison. The insert shows the diffraction rings from the Pilatus detector, which has been converted to the intensity-2θ pattern as shown. The XRD pattern of graphite (PDF 41-1487) is also shown as vertical red lines. The conspicuous peak seen in the powder sample is absent.

### Figure 4 shows the XRF map of elements in a region of interest covering both dark and light *3.3. XRF, Imaging and Micro-XANES Analysis*

areas. XRF tracks the specific X-ray emissions from elements, e.g., Cu Kα X-ray emissions arise from the radiative de-excitation upon the removal of a 1s electron from a Cu atom (2p electron filling a 1s core hole, emitting an X-ray photon). While XRF can be excited by any high-energy radiation—most commonly electrons, as in SEM, EDX and X-rays—a synchrotron X-ray has the advantage over EDX using SEM in that, in addition to being many orders of magnitude brighter than laboratory X-rays, one can tune the excitation energy to just above an absorption edge of interest to enhance absorption and, hence, the intensity of the fluorescence or, under favorable conditions, to avoid an unwanted fluorescence signal by tuning the X-ray to below the edge of an unwanted element. Let us first discuss the XRF results from VESPERS, a hard X-ray beamline where we used 12 keV X-rays, which will be able to excite Ca and all first-row transition elements, yielding characteristic K<sup>α</sup> Figure 4 shows the XRF map of elements in a region of interest covering both dark and light areas. XRF tracks the specific X-ray emissions from elements, e.g., Cu Kα X-ray emissions arise from the radiative de-excitation upon the removal of a 1s electron from a Cu atom (2p electron filling a 1s core hole, emitting an X-ray photon). While XRF can be excited by any high-energy radiation—most commonly electrons, as in SEM, EDX and X-rays—a synchrotron X-ray has the advantage over EDX using SEM in that, in addition to being many orders of magnitude brighter than laboratory X-rays, one can tune the excitation energy to just above an absorption edge of interest to enhance absorption and, hence, the intensity of the fluorescence or, under favorable conditions, to avoid an unwanted fluorescence signal by tuning the X-ray to below the edge of an unwanted element.

and Kβ X-rays; the Kα was used to track elements of interest. Let us first discuss the XRF results from VESPERS, a hard X-ray beamline where we used 12 keV X-rays, which will be able to excite Ca and all first-row transition elements, yielding characteristic K<sup>α</sup> and K<sup>β</sup> X-rays; the K<sup>α</sup> was used to track elements of interest.

From Figure 4a,b, we can see the etched rings covering the surface regardless of the dark and light regions. The scan covered an area of ~1.1 × 1.1 mm, as marked by the red square. It began in a dark region and ended in a light region, covering ~50–50 of both the light and dark region. From Figure 4c, it is apparent that the most interesting feature was the hot spots seen in the element maps of Ca, Fe and Cu, especially Cu. The location of these rings had no correlation with the dark and light regions of the surface. Let us now concentrate on the Cu map. The spots were ~20 µm in diameter and aligned regularly along the circumference of the ring, with a spacing of ~100 µm. The maps of Ca, Fe and Cu could be directly overlaid on each other. To determine the chemical composition of the Cu, we conducted Cu K-edge XANES measurements with a microbeam at the hot spots. Figure 5a,b shows a magnified image of this region, which reveals that the etched ring spanned both the light and dark regions. Figure 5c displays the micro-XANES of a hot spot (5 × 5 µm) recorded in the Cu K<sup>α</sup>

fluorescence yield compared with the XANES of the reference compounds, Cu metal, Cu2O and CuO, representing the 0, +1 and +2 oxidation states of Cu. It should be noted that despite low count rates in micro-XANES, the signal we obtained (eight scans) required about an hour, and it already clearly confirms that the hot spot Cu was from CuO. The discovery of CuO led us to compare the XRD data with those of CuO XRD patterns, and there was no sign of any CuO peaks seen in the bead's data, as noted in the XRD section above. This indicates that the amount of CuO was either amorphous or very small and not detectable in the XRD. *Heritage* **2020**, *3* FOR PEER REVIEW 7

**Figure 4.** (**a**,**b**) show the starting and ending point of the scan (crosshairs), respectively, of the region marked with a square. (**c**) X-ray fluorescence (XRF) maps of several elements, Ca, Fe and Cu, are shown (XRF data are stored in a Multichannel Analyzer (MCA); counts within an energy window characteristic of the element are tracked and color-coded). These maps show hot spots on the circumference of the etched ring, most apparent in the case of Cu. The size of the hot spot is ~20 µm, with ~100 µm spacing along the circumference. **Figure 4.** (**a**,**b**) show the starting and ending point of the scan (crosshairs), respectively, of the region marked with a square. (**c**) X-ray fluorescence (XRF) maps of several elements, Ca, Fe and Cu, are shown (XRF data are stored in a Multichannel Analyzer (MCA); counts within an energy window characteristic of the element are tracked and color-coded). These maps show hot spots on the circumference of the etched ring, most apparent in the case of Cu. The size of the hot spot is ~20 µm, with ~100 µm spacing along the circumference.

From Figure 4a,b, we can see the etched rings covering the surface regardless of the dark and light regions. The scan covered an area of ~1.1 × 1.1 mm, as marked by the red square. It began in a dark region and ended in a light region, covering ~50–50 of both the light and dark region. From Figure 4c, it is apparent that the most interesting feature was the hot spots seen in the element maps of Ca, Fe and Cu, especially Cu. The location of these rings had no correlation with the dark and light While the experiment using VESPERS provided fruitful results, its energy was not suitable to track SiO2, the major component of agate based on which Dzi beads are made and crafted. We next conducted measurements with the SXRMB beamline using its tender X-ray microprobe endstation. SXRMB is designed for tender X-ray (1.7–5 keV) investigations, with accessible photon energy extending to 10 keV [16,17].

regions of the surface. Let us now concentrate on the Cu map. The spots were ~20 µm in diameter and aligned regularly along the circumference of the ring, with a spacing of ~100 µm. The maps of Ca, Fe and Cu could be directly overlaid on each other. To determine the chemical composition of the Cu, we conducted Cu K-edge XANES measurements with a microbeam at the hot spots. Figure 5a,b shows a magnified image of this region, which reveals that the etched ring spanned both the light and dark regions. Figure 5c displays the micro-XANES of a hot spot (5 × 5 µm) recorded in the Cu K<sup>α</sup> fluorescence yield compared with the XANES of the reference compounds, Cu metal, Cu2O and CuO, representing the 0, +1 and +2 oxidation states of Cu. It should be noted that despite low count rates in micro-XANES, the signal we obtained (eight scans) required about an hour, and it already clearly confirms that the hot spot Cu was from CuO. The discovery of CuO led us to compare the XRD data with those of CuO XRD patterns, and there was no sign of any CuO peaks seen in the bead's data, as noted in the XRD section above. This indicates that the amount of CuO was either amorphous or very small and not detectable in the XRD. The microprobe provides a beam size of ~11 × 11 µm across this energy range with KB mirror focusing [17]. To investigate the Si K-edge, we used InSb(111) crystals in the double-crystal monochromator. XRF, imaging and micro-XANES were all conducted in a vacuum chamber; a Si drift detector was used to track the fluorescent X-rays. Figure 6a shows the optical image of the area of interest. Figure 6b displays the XRF maps excited at a photon energy of 7130 eV, just above the Fe K-edge, and selected regions of interest (ROI), marked with color-coded boxes. Figure 6c–e is the XRF maps of Si, K, Ca and Fe of selected ROI, respectively. While the etched rings were apparent, instead of looking more like holes in the Cu map, the Si in Figure 6a shows the optical image from the video camera; it again shows that the etched rings were all over the surface. The boxes define the area where XRF maps of relevant elements were collected, as shown in Figure 6b, from Si, K, Ca and Fe Kα<sup>1</sup> emissions. It can be clearly seen that the rings contained extruding Si materials, while other elements appeared to be less defined or depleted in the case of K, which tracked Si well in the ring circumference. It was desirable to see if the Si map was correlated with the Cu image obtained using VESPERS (Figure 5). While Cu mapping using SXRMB was difficult due to low flux at the Cu K-edge, we were

and CuO.

irregularities.

able to track the Cu distribution while tracking Si and other element maps at 7130 eV by observing the Cu K<sup>α</sup> produced by second-order radiation (14260 eV). The results are shown in Figure S4 for ROI displayed in Figure 6d; one can see that despite the weak signal, there was a good correlation between Cu and Si. *Heritage* **2020**, *3* FOR PEER REVIEW 8

**Figure 5.** (**a**) Optical image of the magnified region of the etched ring inside the region of the XRF map. (**b**) XRF map from Cu showing the hot spots and the circled hot spot where micro-X-ray absorption near edge structure (XANES) has been obtained. (**c**) Cu K-edge micro-XANES from a hot spot circled in (**b**), obtained with a microbeam (5 × 5 µm) together with the references, Cu metal, Cu2O **Figure 5.** (**a**) Optical image of the magnified region of the etched ring inside the region of the XRF map. (**b**) XRF map from Cu showing the hot spots and the circled hot spot where micro-X-ray absorption near edge structure (XANES) has been obtained. (**c**) Cu K-edge micro-XANES from a hot spot circled in (**b**), obtained with a microbeam (5 × 5 µm) together with the references, Cu metal, Cu2O and CuO. *Heritage* **2020**, *3* FOR PEER REVIEW 9

we were able to track the Cu distribution while tracking Si and other element maps at 7130 eV by observing the Cu Kα produced by second-order radiation (14260 eV). The results are shown in Figure S4 for ROI displayed in Figure 6d; one can see that despite the weak signal, there was a good correlation between Cu and Si. **Figure 6.** (**a**) Optical image of the Dzi bead surface; the rectangle encloses the region of the XRF maps shown in (**b**). (**b**) XRF map from Ca, K, Si and Fe as noted; the color-coded boxes in the Si map define the regions of interest (ROI) displayed in (**c**–**e**) with the same color code, where all four elements are tracked; the color code is also shown. **Figure 6.** (**a**) Optical image of the Dzi bead surface; the rectangle encloses the region of the XRF maps shown in (**b**). (**b**) XRF map from Ca, K, Si and Fe as noted; the color-coded boxes in the Si map define the regions of interest (ROI) displayed in (**c**–**e**) with the same color code, where all four elements are tracked; the color code is also shown.

Figure 7 displays the Si K-edge micro-XANES of four spots from hot to cold. We can see that the XANES for all the spots was nearly identical except for spot 2, which was at the edge of the crack and Figure 7 displays the Si K-edge micro-XANES of four spots from hot to cold. We can see that the XANES for all the spots was nearly identical except for spot 2, which was at the edge of the crack and

XANES in different ROIs depicted in Figure 6c,d and found similar results (not shown) with no

**Figure 7.** (**a**) Si map of a ROI of the Dzi bead sample, as shown in Figure 6e, where four spots (single pixel of 11 × 11 µm) were selected for micro-XANES analysis as marked. Spot 1 is the hot spot, spot 4 is the cold spot, spot 2 is at the edge of a crack and spot 3 is with moderate intensity. (**b**) Si K-edge XANES of selected spots. The spectral features are unmistakably characteristic of quartz SiO2 (inset).

tracked; the color code is also shown.

showed a slight broadening of the resonances, indicating some disorder. They were characteristic of the XANES of quartz (inset), with the white line (spike at the edge jump) at ~1848 eV, arising from the Si 1s to t<sup>2</sup> transition in the tetrahedral surrounding of Si in quartz. We have also conducted micro-XANES in different ROIs depicted in Figure 6c,d and found similar results (not shown) with no irregularities. the XANES of quartz (inset), with the white line (spike at the edge jump) at ~1848 eV, arising from the Si 1s to t2 transition in the tetrahedral surrounding of Si in quartz. We have also conducted micro-XANES in different ROIs depicted in Figure 6c,d and found similar results (not shown) with no irregularities.

showed a slight broadening of the resonances, indicating some disorder. They were characteristic of

**Figure 6.** (**a**) Optical image of the Dzi bead surface; the rectangle encloses the region of the XRF maps shown in (**b**). (**b**) XRF map from Ca, K, Si and Fe as noted; the color-coded boxes in the Si map define the regions of interest (ROI) displayed in (**c**–**e**) with the same color code, where all four elements are

Figure 7 displays the Si K-edge micro-XANES of four spots from hot to cold. We can see that the

*Heritage* **2020**, *3* FOR PEER REVIEW 9

**Figure 7.** (**a**) Si map of a ROI of the Dzi bead sample, as shown in Figure 6e, where four spots (single pixel of 11 × 11 µm) were selected for micro-XANES analysis as marked. Spot 1 is the hot spot, spot 4 is the cold spot, spot 2 is at the edge of a crack and spot 3 is with moderate intensity. (**b**) Si K-edge XANES of selected spots. The spectral features are unmistakably characteristic of quartz SiO2 (inset). **Figure 7.** (**a**) Si map of a ROI of the Dzi bead sample, as shown in Figure 6e, where four spots (single pixel of 11 × 11 µm) were selected for micro-XANES analysis as marked. Spot 1 is the hot spot, spot 4 is the cold spot, spot 2 is at the edge of a crack and spot 3 is with moderate intensity. (**b**) Si K-edge XANES of selected spots. The spectral features are unmistakably characteristic of quartz SiO<sup>2</sup> (inset).
