Gemstone Inclusion Study by 3D Raman-Mapping and High-Resolution X-ray Computed Tomography: The Case of Trapiche Emerald from Swat, Pakistan

Abstract

Gemstones usually contain inclusions as the natural hallmark of their formation. Studies on inclusions may contribute to our understanding of the gem and better adapt to the gems and jewelry industry. In this study, we applied advanced 3D Raman mapping and high-resolution X-ray computed tomography (HRXCT) and conducted a thorough experiment on one emerald with a special trapiche pattern containing a colorless core, solid minerals, and fluids. Hematite and magnetite were identified as metallic minerals by the Raman spectrum. The hexagonal core is beryl, and the voids are primarily filled with CO₂ and N₂. HRXCT demonstrated a visualized distribution of these inclusions within the gem host by reconstructing a 3D illustration. Further calculation of the volume of inclusions regarding the host showed that the minerals take up 0.07%, void 0.03%, and the hexagonal core 8.25%. The combined application of Raman-Mapping and X-ray Micro Computed Tomography proved to be a very promising technique for tracing the gemstones by characterizing the unique inclusions (identification and morphology) within the gem host.

1. Introduction

Gemstones are usually highly valued and prominent in the trade owing to their beauty and rarity. The gem materials usually contain inclusions trapped during their formation. As the inclusions contain the message of the ancient earth, well-studied research may unveil nature’s secret.

Furthermore, the inclusions, such as trapped mineral crystals, rock relics, and fluids, are unique to the gem host. Based on the phase, they are classified as solid, liquid, and gaseous. Minerals are the most commonly encountered, while the liquid and gaseous usually come together as two-phase inclusions, such as the negative crystal containing water and CO₂ in an emerald. Solid inclusions can be categorized into protogenetic, syngenetic, and epigenetic types inclusions based on the order in which they formed concerning their hosts [1]. A protogenetic inclusion refers to those forms before the gem host, i.e., older than the gem, such as calcite found in Burmese ruby [2] and mica found in Zambian emerald [3,4]. A syngenetic inclusion represents inclusions forming simultaneously with the host, meaning that the gem and inclusions are of the same age. Garnet and apatite were found as syngenetic inclusions in carbonado [5]. An epigenetic inclusion means inclusion forms after the host, meaning it is younger than the host. Gems usually contain open fissures owing to the unideal growth environment. Hence, the substance around the gem may go inside the cracks and stay within. Limonite is commonly encountered in Sri Lanka sapphire and partially heals the fracture of the host [6].

The Raman effect was first reported in 1928 [7]. Afterward, Raman spectroscopy was developed and applied to many fields, including gemology [8]. Micro-Raman spectroscopy has been applied to sapphires from various localities, and the mineral inclusions were successfully identified, such as anatase, rutile, and calcite [9]. Additionally, less common minerals can also be detected by Raman, such as palygorskite fibers in pink opals from Mexico and Peru [10] and diopside needles in demantoids from Russia [11]. More inclusion identification cases and related references can be detailed in previous literature [12,13].

The 3D mapping takes the Raman application to gemstones to a higher level, demonstrating the tiny difference in a specific selected area and presenting a much more visualized scene of the inclusion within the gem host. Eaton-Magaña and the co-authors have conducted the application of Raman mapping on gem materials such as diamonds [13]. Their experiment mainly focuses on the surface and the near-surface part, and the results are exciting and promising. In this research, we couple the Micro-Raman spectrometer with a 3D moving sample stage, enabling us to set the accurate position of the confocal laser beam and perform the mapping on the inclusions of selected samples.

Secondly, the confocal micro-Raman spectroscopy could identify the inclusion based on their unique spectral pattern. 3D mapping further scans the selected area of the inclusions and helps us understand the overall morphology of the inclusions and their relationship with the host in a three-dimensional way.

X-ray image techniques have been a promising newcomer since the 1990s [14] and have been widely applied to gemology, mainly in pearls and diamonds. Sunagawa et al. [15] used X-ray diffraction with cathodoluminescence to determine whether two faceted diamonds were cut from the same rough. Later research also demonstrated the application of an X-ray fingerprinting routine for cutting diamonds [16]. The X-ray has also been proven helpful in deciding whether pearls are natural or cultured and grown in saltwater or freshwater based on the details of their internal features [17,18].

HRXCT is a non-destructive method that combines the X-ray source and micro-objective coupled detector. It uses the comparison of X-ray attenuation on a series of continuous X-ray plates with different viewing angles to reconstruct the 3D distribution of regions of different densities in many materials. A review by Cnudde and Boone in 2013 demonstrated the high-resolution X-ray computed tomography applied to geoscience materials [19]. The in situ nature of gold particles and associated minerals can be characterized using high-resolution X-ray computed tomography and PVB analysis, which can reveal particle sizes, volumes, shapes, orientations, and textural relationships. This was demonstrated in a recent study on orogenic gold ores from Brazil [20].

The first brief report on the X-ray 3D applied on a gemstone was by Giuliani et al., who performed an X-ray 3D scanning on a demantoid and presented a clear result, revealing the internal morphology of the fluid and vapor within the cavity [21]. Furthermore, Richard et al. [22] performed high-resolution X-ray computed tomography (HRXCT) on the fluids and melted inclusions of several gemstone varieties. Their results showed that HRXCT is a promising tool for the non-destructive characterization of fluid and melt inclusions. Heyn et al. performed the X-ray CT on several gemstone varieties, including corundum, topaz, and quartz [23].

Usually, the standard microscopic observation under a gemological microscope equipped with various lighting conditions could reveal the inclusion scene and provide valuable information on the gemstone’s variety, treatment, and origin. However, such a method is prohibited mainly if the stone is not polished or a faceted stone is set into metal as jewelry. The colored stones are collected as rough from underground and fashioned and polished into faceted or cabochon gemstones. The interior features are barely seen under a microscope if the stone is in a rough state with a raw surface, since the raw surface will diffuse the light, prohibiting it from being captured by the microscope. By comparison, X-ray micro-computed tomography can penetrate the bulk of the sample and avoid interference from the surface, enabling us to see the internal scene of the gemstone at a more accurate level.

The new application of advanced technology to the inclusion of gemstones can contribute to elevating the research level and advance our understanding of the formation and growth mechanism of the gemstone and the adaption of the inclusion in the gems and jewelry industry.

In this study, we demonstrate the capabilities of HRXCT by scanning a trapiche emerald with a distinctive growth pattern in three dimensions, displaying the form of fluid and mineral inclusions and recreating the overall volume and volumes of each phase using post-process software. Raman mapping and HRXCT recorded the experiment results performed on the trapiche emerald sample. The combined technique of 3D Raman mapping and HRXCT has proven to be quite valuable in furthering our understanding of the inclusions within a gemstone.

2. Materials and Methods

2.1. Materials

In this study, we have selected one trapiche-like emerald from Swat Valley, Pakistan, to demonstrate the combination of 3D Raman mapping and high-resolution X-ray computed tomography (HRXCT). Trapiche-like emeralds contain multiple inclusions, including minerals and fluids, which could serve as a good example to exhibit the high power of Raman and HRXCT. The chosen sample represents a trapiche-like growth pattern within emeralds, which can be seen in emeralds from Swat Valley, as shown in Figure 1. This chosen sample weighs 0.535 ct, with a size of 6.32 × 5.65 × 1.56 mm³, and its trapiche pattern consists of minute black inclusions surrounding a colorless core, as shown in Figure 2.

2.2. Methods

In this study, we have applied three methods to study the inclusions of a gemstone, including standard gemological microscopic observation, 3D Raman mapping, and high-resolution X-ray computed tomography (HRXCT).

2.2.1. Standard Gemology

Standard gemology combined with gemological microscopic observation allows for visual inspection of the inclusion under a well-equipped microscope and presents a broad scene of inclusion distribution. Inclusions were observed with an 80× magnification gemological microscope with S6D Leica optics equipped with different lighting sources, including transmitting light, reflective illumination, and oblique fibers.

2.2.2. Raman

The Horiba Xplora Plus Raman system (Horiba France SAS, Palaiseau, France) performed Raman micro-spectroscopy on the inclusions in this study. The machine was equipped with three lasers of 473 nm, 532 nm, and 783 nm. All the lasers were applied to the sample in this study to find the most prominent and effective one. As a result, we applied a 473 nm laser to the emerald sample. The spectra were recorded in the 400–2000 cm^–1 range with a spectral resolution of 4 cm^–1 and 8 scans at 7.5 kHz scanning speed.

Raman spectra were primarily collected in the range of 2000–100 cm⁻¹. The 3D Ram mapping was performed on a selected sample area coupled with an XYZ stage, which the software can automatically control. The site’s position was chosen as a cube, with specific x, y, and z position values constrained within a particular range, with increments between 1 and 2 μm.

We randomly chose three plates of hematite and magnetite for Raman mapping, which could show the overall form and size in three dimensions. The mapping area is strictly measured by the X, Y, and Z-axis, determined by the XYZ sample stage, with accuracy as low as 0.1 μm. The XYZ parameters include the starting point, ending point, increment step, and the number of points along the axis, which can be seen in detail in

Table 1. The 3D Raman mapping parameters on the magnetite and hematite inclusions.

Parameters		Mag-02	Hem-02
X	Range/μm	(−15, 15.9)	(−17.5, 17.5)
	Step/μm	0.8	3.9
	Size	40	10
Y	Range/μm	(−9.7, 7)	(−23.1, 19.9)
	Step/μm	0.8	3.9
	Size	22	12
Z	Range/μm	(−5, 7)	(−100, 100)
	Step/μm	0.5	4
	Size	25	51
Total Volume/μm3		6212.4	301,000
Total Time		36 h 30 min	14 h 21 min
Total Spectrum		22,000	6120

. The 3D Raman mapping is pretty time-consuming, usually taking hours to acquire a decent result and half to one day to get a good one, depending on the quality of the finish spectrum and reconstructed images.

2.2.3. HRXCT

We applied nanoVoxel-3000 X-ray three-dimensional microscopy computerized tomography to analyze a 0.55 ct trapiche-like emerald from Swat Valley, Pakistan. We collected 1170 images by scanning them, layer by layer of the sample, within 90 min using the VoxelStudio Scan software V1.0 in Tianjin, China. The testing parameters include 60 kV voltage, 50 μA current, and 3.24 μm resolution. The post-processes are reconstructed based on the finished pictures by the VoxelStudio Recon software.

3. Results and Discussion

Trapiche emeralds from Colombia have been well-known in the gem trade. They were first described by the French mineralogist Emile Bertrand [24]. Pogue’s study [25] showed the first mention of inclusions of carbonaceous substance in the emeralds “arranged in a six-rayed centering about a tapering hexagonal core”. A recent study [26] showed an advanced understanding of the texture and growth mechanism of Colombian trapiche emeralds. Colombia was the only one producing trapiche emeralds until it was reported that the trapiche-pattern was also observed in emeralds from Swat Valley, Pakistan [27].

Unlike the Colombian counterpart, the Pakistan material was composed of four parts from rim to core: a green edge, a light green area, six black arms, and a colorless center, as illustrated in Figure 3. The edges were a highly saturated green, and most were relatively clean except for several fractures and tiny fluid inclusions. The edges ranged from approximately 1 to 2 mm wide. Although the boundary between the green border and the light area was not very sharp, a hexagonal boundary was visible. Six black arms spread in a hexagonal symmetric pattern, with each arm perpendicular to the hexagonal side. The colorless core usually had a hexagonal shape.

As demonstrated in Figure 4, the image record from Micro-CT shows a general structure of the sample viewing from three angles along with a body view. This sample can be divided into four parts based on density: the emerald host, metallic minerals, voids, and the core. Figure 5 shows a clear boundary between the core and other parts, indicating a distinct difference in density. The metallic mineral shows a bright luster, and the voids exhibit black dots, which are indicated by red and blue circles, respectively, in Figure 5. The core exhibits a clear and well-formed hexagonal outline showing darker color.

3.1. The Metallic Inclusions

Microscopic observation revealed metallic minerals, and the magnetite (Mag) concentrated in small size with a gray color in a systematic way. At the same time, the hematite (Hem) shows subhedral hexagonal growth habits showing bright interference color, as shown in Figure 6. Raman tests identify two mineral varieties as magnetite and hematite. An X-ray CT result showed that the metallic inclusions are scattered along with the gem host, and they take up about 0.07% by volume in total. We will demonstrate the results and discuss them separately.

3.1.1. Magnetite

Magnetite is a commonly seen iron oxide mineral with a chemical formula of Fe₃O₄, belonging to the isotropic crystal system. In this study, the magnetite minerals within the emerald host appear black when observed by the naked eye. The well-distributed approach of magnetite gives rise to the trapiche pattern of the emerald host. Further observation under high magnification of the microscope showed that the magnetite is thin platy with a hexagonal outline, while some may exhibit a skeleton appearance. The typical reflective grey color is characteristic of magnetite, as shown in Figure 7.

Raman test reveals feature peaks at 661 and 534 cm⁻¹, which agrees with magnetite from the RRUFF online database, as shown in Figure 8. The prominent peak at 686 cm⁻¹ and a weak peak at 399 cm⁻¹ originate from the emerald host. No distinct magnetism was observed when we applied a magnet close to the emerald sample. 3D Raman mapping was conducted on one selected magnetite to study the morphology of the magnetite further. Figure 9 contains seven Raman spectra recorded on this chosen magnetite with a constant z-axis value ranging from −5 μm to 3 μm with fixed x and y, as indicated by the legend. The intensity of the feature peak of magnetite at 661 cm⁻¹ starts to increase as the Z value starts from −5 μm, and it became most distinct when z = −1 μm and decreased to nearly none at the z value moved to 3 μm.

The recorded 3D mapping image results are illustrated in Figure 10, where the legend bar indicates the intensity of the selected prominent Raman peak with specific colors. For example, orange designates a weak signal of 661 cm⁻¹, while other colors are in between, and red denotes a strong signal. As a result, the color change caused by the magnetite’s primary peak intensity can be used to illustrate the mineral’s morphology. As shown in Figure 10, the images explicate the outline shape of the platy magnetite. More importantly, the fluctuation in the magnetite’s primary peak may be a valuable tool for deciphering the magnetite’s growth process within the gem host. Based on the images and z-axis testing range, this magnetite inclusion thickness can be estimated as less than 8 μm.

Furthermore, a 3D image of magnetite was generated based on the 661 cm⁻¹ peaks, as shown in Figure 11, within a restricted cuboid area. Moreover, the length, width, and height are 31 μm, 15.7 μm, and 12 μm, respectively, forming a three-dimensional volume of 5840.4 μm³. This image demonstrates the distribution and the overall morphology of the magnetite inclusion. The animation allows the viewer to see the inclusions from different angles, usually prohibited by microscopic observation.

3.1.2. Hematite

Hematite is another iron oxide mineral with a chemical formula of Fe₂O₃. In this study, the naked eye barely sees the hematite inclusions. Unlike the concentrated and evenly dispersed magnetite, hematite does not contribute to the trapiche-pattern of the emeralds host. However, many brownish-red colored platy metallic inclusions are also visible under a microscope. They generally showed inhomogeneous size, with their width ranging from 20 to 200 μm, as indicated in Figure 12.

Further Raman tests identified them as hematite. As seen in Figure 13, Raman spectrum analysis identifies the large, brownish-red platy inclusion as hematite. It has two feature peaks at 225 and 290 cm⁻¹, weak peaks at 407, 495, 607, and 658 cm⁻¹, and a prominent broad, centered at 1310 cm⁻¹, all of which agree with the main peaks of hematite from the RRUFF online database.

The 3D Raman mapping was conducted on one selected plate to further study the hematite’s morphology. Figure 14 contains seven Raman spectra recorded on this chosen hematite with a constant z-axis value ranging from −48 μm to 48 μm with fixed x and y, as indicated by the legend. The intensity of the feature peak of hematite at 1310 cm⁻¹ starts to increase as the Z value starts from −48 μm, and it became most distinct when z = 0 μm, and decreased to nearly none at the z value moved to 48 μm.

The 3D Raman mapping images in Figure 15 demonstrated the hematite plate’s hexagonal shape based on the variation in the strength of the 1310 cm⁻¹ peaks of the chosen hematite in Figure 14, with fixed x and y, while the Z value ranged from −48 μm to 48 μm. The thickness of this hematite plate is much thicker than the magnetite, nearly 96 μm. A 3D model of the selected hematite was reconstructed based on the 3D Raman mapping, selecting 1310 cm⁻¹ as a criterion, as shown in Figure 16.

HXRCT imaging reveals the overall distribution of all the metallic inclusions, as seen in Figure 17 At the same time, the Raman spectrum identifies the metallic inclusions as magnetite and hematite, creating a 3D image of the carefully specified platies. Post-processing software calculation reveals a 0.07% volume proportion regarding the host. Although they seem to be generally uniformly distributed, the layer-by-layer scanning result shows a fluctuating proportion ranging from 0.02% to 0.15%, with an average value of 0.07%, as shown in Figure 17 and Figure 18.

3.2. The Voids

Among the host, numerous tiny voids were observed, filled with multiple phases, as shown in Figure 19. Most of the voids showed a flattened shape parallel to the (0001) basal face with an irregular or elongated outline (Figure 20). Raman tests identify the phases as a mixture of CO₂ and N₂, with two prominent peaks at 1280 and 1386 cm⁻¹ assigned to CO₂ and one strong peak at 2328 cm⁻¹ attributed to N₂ (Figure 21). The movability of these fluids prohibits further 3D Raman mapping.

The HRXCT test showed a continuous change in the volume proportion change of the voids along the direction of sample thickness, as shown in Figure 22. While a distinct fluctuation occurs between the range of 251–301 tiers, indicating large internal fractures in this area, most of the voids vary between 0.01% and 0.05%, taking up 0.03% of the gem host by volume, on average (Figure 23). These HRXCT results proved useful in forecasting internal fracture, usually unseen by microscopic observation.

3.3. The Hexagonal Core

The X-ray image shows the core with a nearly regular hexagonal shape (Figure 24). The size measurement shows that the side length of the hexagon is approximately 0.671 mm, and the distance between parallel side lengths is 1.140 mm. The internal angle is 121.298 degrees, very close to the standard inner angle of the hexagonal crystal system mineral. The deviation in the inner angle value can be attributed to the position of the sample wafer, which is precisely cut and polished to position the C-axis of the emerald host perpendicular to the wafer planes. Based on the value of the side length and the inner angle, the area of the hexagonal cross-section is approximately 3.49 mm². Further calculation based on the value acquired provides the volume of the core. As the scanning continues along the direction of sample thickness, the volume proportion varies slightly, but no distinct fluctuation is observed.

The volume proportion change in the core is relatively stable compared with the voids and metal inclusion (Figure 25), suggesting that the center is a well-formed hexagonal crystal. Furthermore, the final result showed that the core takes up approximately 8.25% of the host by volume, on average (Figure 26). Again, the volume proportion could be even more stable and less deviated if the wafer is cut perpendicular to the c-axis.

The color variations in Figure 26 indicate that the core and the emerald host possess different densities, since the CT image shows different colors. It is worth pointing out that no distinct gemological feature variations were observed in the current state, such as the luster and R.I.

Further Raman testing was performed on the core, and the recorded spectrum shows peaks at 685 cm⁻¹ and 399 cm⁻¹, which agrees with beryl (Figure 27). Further line mapping with 10 selected points from the host to the center was carried out. The Raman spectrum of the core agrees with the emerald host very well, and no distinct spectrum features’ variations were found between the host and the core, as shown in Figure 27.

The core may be the starting crystal seed, which crystallized previous to the emerald host, with insufficient trace elements to produce a prominent green color. Further study will focus on the crystalline relationship between the emerald host and metallic minerals.

When considering the distribution of fluid, minerals, and the core within the emerald host, a whole structure was reconstructed three-dimensionally (Figure 28).

4. Conclusions

The inclusion study has been a significant subject in gemology, which helps identify the gem, detect the treatment, and provide geographic information. Standard gemological microscopic observation enables us to see the inclusion. Further, Raman microscopy proves to be an efficient technique to identify the inclusion without damaging the stone. These three techniques cover the visual, spectral, and structural perspectives. Their combination could serve as a comprehensive solution to study the inclusion of gemstones.

A seasoned cutter usually performs the cutting process to maximize the rough gem value in color, clarity, and weight. Any large fractures or undesired inclusion may cause damage to the gem, thus, decreasing the value or even eliminating the whole stone. Hence, the scanning by high-resolution X-ray CT can provide much more information on the distribution of inclusions and fractures within the rough, which is barely known to a cutter.

Lastly, high-quality and large-sized colored stones, such as ruby, blue sapphire, and emerald, are usually highly valued. This combined technique may further provide unique information from the notable inclusion within the stone. Such information is very critical for the appraiser and auction house. Moreover, even when the stones are mounted or recut, the inclusion information generated by Raman mapping and CT is still valid as the interior features remain untouched.