Gemstone Identification Through Molecular Spectroscopy and Spectral Imaging: A Non-Invasive Method for Authenticating Turquoise

Turquoise is one of the oldest known and widely used gemstones throughout history and across various cultures and civilizations. This gemstone is an opaque cryptocrystalline mineral that occurs in blue, blue-green, and green colors, with the chemical formula CuAl₆(PO₄)₄(OH)₈·4H₂O. Due to its beautiful color, high popularity, and unique significance to people, there are numerous simulants of turquoise available in the market today. The use of turquoise simulants dates back at least 6000 years in Egypt and among the Roman civilization (1). Some minerals, such as chrysocolla, microcline, lazulite, serpentine, odontolite, variscite (2), and prosopite (3) can be mistaken for turquoise due to their color and appearance.

In addition to natural and mineral imitations, other turquoise imitations, such as dyed calcite, halite, turquoise powder mixed with glue, glass, porcelain, plastics (2), and colored agate (often referred to as "turquoise agate"), which has recently been introduced to the market as a turquoise substitute (4), can also be mentioned. Natural turquoise is usually highly porous. Various treatments, such as saturation, dyeing, or the Zachery process (5), can increase its strength and reduce its porosity. However, it is not difficult to distinguish between filled turquoise with resin and untreated turquoise (6). Since the early 1970s, artificial turquoise has also entered the market extensively (1).

The variety of turquoise gemstones available and the potential for misuse by forgers have increased the importance of identifying natural types of this gemstone from others. Today, the use of techniques that cause minimal contact and damage to the gemstone is a priority for gemologists. Bernardino and associates (7) utilized Raman spectroscopy as a non-destructive method to identify and distinguish various turquoise samples with different appearances. This type of spectroscopy has been recognized as an effective method for identifying provenance and determining gem treatments in gemological laboratories since the late 1970s (5). Distinguishing between natural and treated turquoises (8), as well as tracing the provenance of archaeological turquoise (9–11), are other beneficial outcomes of Raman spectroscopy in turquoise studies.

On the other hand,ultraviolet-visible-near infrared (UV-vis-NIR) spectroscopy, in both reflection and absorption modes, is another non-destructive and non-invasive method for identifying genuine turquoise from fake, dyed, and resin-filled specimens (12,13). Fiber optic reflectance spectroscopy (FORS) is a useful alternative to other classic gemological techniques and provides beneficial results for distinguishing and identifying natural gemstones from their simulants (14–16).

Spectral imaging, another non-destructive technique for studying gemstones, has been developed over the past 60 years and has yielded effective results in the analysis of paintings, textiles, written documents, and the conservation of artworks (17–24). Among these techniques, precious and semi-precious stones are frequently analyzed using the ultraviolet-induced visible luminescence (UVL) imaging method, which relies on the observation of visible luminescence produced under UV-A and UV-C ultraviolet light (25,26). An examination of natural turquoise using ultraviolet light reveals yellow-green to light blue luminescence under long wavelengths, while short wavelengths exhibit neutral luminescence (27). This visible luminescence is not only beneficial for the initial assessment of the gem but also offers valuable insights regarding its treatment (28).

Hyperspectral imaging, which provides spectral cubes and unique information about each pixel, can play a significant role in identifying and authenticating precious and semi-precious stones (29,30). In Gurschler and associates’ report (31), hyperspectral imaging was used to distinguish natural turquoise from fake specimens. Additionally, identifying and discriminating between turquoises of different hardness is one of the advantages of hyperspectral imaging, particularly in assessing vast turquoise resources (12).

On the other hand, multi-band imaging is also recognized as another non-invasive and non-destructive spectral imaging method, offering extensive information with each image capture. However, distinguishing natural gemstones from fakes using this method is a less-studied subject. Therefore, this study focuses on the non-invasive authentication of gemstones by separating and identifying six turquoise imitations using various spectroscopic methods, such as Raman spectroscopy and fiber optic reflectance spectroscopy, as well as spectral imaging techniques, including hyperspectral and multi-band imaging.

Materials and Methods
Micro-Raman Spectroscopy
Raman scattering signals were acquired at room temperature using a Tekram micro-Raman spectrophotometer (Teksan) equipped with a 532 nm (green) laser. Raman spectra were recorded with a resolution of 6 cm⁻¹ and a 10x objective. Signals were collected using a 1200 gr/mm (750 nm) grating with 3 accumulations of 3 s. The RRUFF database (https://rruff.info) was also used to match the resulting spectra with the reference material spectra.

Fiber Optic Reflectance Spectroscopy
For fiber optic reflectance spectroscopy, a portable Ozhen spectrometer equipped with a dual tungsten-deuterium light source capable of recording spectra within the UV-vis-NIR range was used. A Y-shaped fiber optic from Thorlabs and polytetrafluoroethylene (PTFE) as a reference were utilized. The range for data collection was 400-900 nm, wherein spectra were recorded for a 1 s integration time through 5 scans. Three spectra from different points of each sample were taken, and the average of these spectra was reported.

Hyperspectral Imaging
Hyperspectral imaging was conducted using the SPECAM hyperspectral camera model Gamma, which captures information within the 400-950 nm range. Four halogen lamps with a 60-degree angle relative to the sample were used as light sources, and the exposure time was set to 120 ms. The false-color image and reflectance spectra were obtained by processing the data in MATLAB R2021b.

Multi-Band Imaging
Multi-band imaging was conducted using a modified Nikon D750 camera, which had its internal UV-IR block filter removed, and an AF Micro Nikkor 60mm 1:2.8 D lens capable of capturing images in the 350-1100 nm range. Two Youngenu NY660 xenon flashes and two Osram PAR38 IR lamps (150W) were used as light sources for the ultraviolet (UV) to infrared (IR) and IR images, respectively. These light sources were positioned at a 45-degree angle to the samples. An X-Rite color checker was also utilized as a color reference.

Images were captured in RAW format at the highest resolution (24 megapixels: 6016×4016) using the filters specified in Table I (enlarge all figures and tables by opening them in a new tab). They were then converted to 16-bit TIFF (Tagged Image File Format) format using Adobe Photoshop software. Post-processing and calibration were performed according to Kushel's method (32) and Cosentino's recommendations (33). Additionally, false-color images were generated using two methods for infrared images (IR-Raman-based vibrational spectroscopy [RG] and IR-gas chromatography [GB]) and two methods for ultraviolet images (RG-UV and GB-UV) (34).

Results and Discussion
The Raman spectra of the investigated samples are presented in Figure 1. The spectral characteristics of the six gemstones reveal three distinct types of materials. Typically, the Raman spectrum of natural turquoise displays indicative bands associated with two phosphate stretching vibrations at about 1066 and 1042 cm⁻¹, phosphate antisymmetric stretching vibrations at 1184-1106 cm⁻¹, and bending vibrations of the PO₄³⁻ tetrahedron at 410-500 cm⁻¹ and 540-650 cm⁻¹ (10,35,36). These bands are well-identified in the spectra of the S5 and S6 samples. Furthermore, a comparison of the Raman spectra of these two samples indicates a high level of agreement with the reference spectrum of turquoise (R050225), confirming the turquoise nature of these two gemstones (Figure 1a).

The Raman spectrum of sample S3 does not match the reference spectrum of turquoise, despite its similar appearance (Figure 1b). Therefore, this gem is likely one of the turquoise simulants, such as chrysocolla, amazonite, howlite, and magnesite, which have been identified as natural and mineral pseudo-turquoise (2,7). The Raman spectrum of S3 revealed the main Raman bands of microcline at about 512 and 478 cm⁻¹ (related to SiO vibrations) and 290 cm⁻¹ (related to external lattice modes) (37,38). This pattern showed a perfect correlation with microcline spectra taken from the RRUFF (R050054) reference mineral collection and with the published literature (39). Thus, this turquoise simulant was identified as microcline. It is important to note that amazonite or microcline (KAlSi₃O₈) is sometimes replaced and sold as turquoise due to its blue-green color (7,40). However, it has physical and chemical properties distinct from turquoise.

The three remaining samples, namely S1, S2, and S4, exhibited a high and similar background fluorescence in the range of 100 to 4000 cm⁻¹ when subjected to the 532 nm laser, as depicted in Figure 1c. This observation suggests the possible presence of dyes, polymers, or resin compounds within these samples. Additionally, the thermal test results further confirmed their organic nature. Hence, it can be concluded that these three fake gems are artificial imitations of turquoise with organic constituents.

Figure 2 presents the reflection spectra obtained from hyperspectral imaging and FORS. The two spectroscopic methods yield similar results; however, hyperspectral imaging provides higher detail and accuracy. Turquoise gemstones (S5 and S6) exhibit a characteristic reflectance band peaking at 480-460 nm. In contrast, microcline (S3) displays such a band at approximately 510-500 nm, and fake turquoises at about 510-530 nm. Natural turquoise does not exhibit a specific reflectance after 600 nm, whereas microcline and particularly fake turquoise exhibit significant reflection after 700 nm and in the NIR range. Furthermore, fake turquoise reflection spectra differ from other samples in that they show strong absorption around 680 nm. The discrepancy in the reflection bands of the spectra of the samples suggests the possibility of distinguishing gemstones using the spectral imaging methods discussed below.

Compared to the other samples, natural turquoise demonstrates high absorbance in the NIR region, indicating the efficacy of utilizing IR imaging as a tool for gemstone identification. The obtained image using a hyperspectral camera at a wavelength of 830 nm is displayed in Figure 3. As expected, the natural turquoise, with its high absorbance, is portrayed as black, whereas microcline appears dark gray, and fake turquoise, with high reflectance, is presented as light gray. Moreover, Figure 3 includes a false-color image generated from the spectral cube by selecting three wavelengths (480, 650, and 820 nm) with high potential for absorbance and reflectance in the reflection spectra of the samples. This image accentuates the visible contrast between the original turquoise, with its orange hue, and the other samples. Microcline is identified as a purple-brown color, whereas fake turquoise is portrayed as light purple in the hyperspectral false-color image.

Multi-band imaging has yielded highly promising and high-resolution results in the identification of the studied types of turquoise, as shown in Figure 4 depicting various multi-band images. The UV-R_-UV-365 and IR images (IR-R-720_{_IR}, IR-R-785_{_IR}, IR-R-830_{_IR}, and IR-R-950_{_IR}) provide evidence of high reflectance and absorption, respectively, for natural turquoise, indicating low absorption in the UV range and high absorption in the NIR range. As a result, it was possible to distinguish turquoise from other simulants in false-color images, with natural turquoise appearing blue in IR-RG images (IR-RG_{_IR-950}, IR-RG_{_IR-830}, IR-RG_{_IR-785}, and IR-RG_{_IR-720}), whereas other samples appeared pink. Similarly, this gem appeared bluish-white in GB-UV images, while other specimens appeared olive. The IR-GB (IR-GB_{_IR-950} and IR-GB_{_IR-720}), RG-UV, and multiband reflectance (MBR) images also helped differentiate turquoise from similar gems. The IR-L_{_VIS} image also yielded comparable outcomes to the IR reflectance images. Additionally, luminescence-based images proved to be useful indicators for microcline identification, with weaker luminescence evident in the Vis-L_{_UV-365}, Vis-L-470_{_UV-365}, and Vis-L-540_{_UV-365} images for the gem being studied, as opposed to other samples. The Vis-L_{_UV-254} image revealed clear identification of microcline due to its characteristic red luminescence, while other samples exhibited light blue luminescence. Additionally, microcline was the only sample that showed luminescence in the IR-L_{_UV-365} method, indicating the high potential of spectral imaging methods in identifying turquoise and its simulants.

Conclusion
This study has provided valuable insights into the identification and separation of various types of gemstones using non-destructive and non-invasive methods. The results of the research demonstrate the potential use of Raman spectroscopy, fiber optic reflectance spectroscopy, hyperspectral imaging, and multispectral imaging in identifying genuine turquoise, microcline, and fake turquoise. Reflectance spectroscopy has shown promise in preliminary gemstone studies. Spectral imaging has proven to be an efficient and cost-effective tool for primary gemstone classification when dealing with a large volume of gemstones. However, the expansion of spectral imaging in gemology requires the creation and development of spectral imaging databases, which have been partially covered in this series of articles. This research is significant for the gemstone industry, as it addresses concerns surrounding the increasing diversity of precious and semi-precious stones among various stakeholders. Overall, this study contributes to the advancement of non-destructive and non-invasive methods for identifying and separating gemstones, which can have practical applications in various fields such as jewelry making, art history, and antique dealing.

Statements and Declarations
Funding
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Competing Interests
The authors declare that they have no competing interests.

Data Availability
All data generated or analyzed during this study are included in this published article.

Contributions
AK contributed to the conceptualization, methodology, validation, formal analyzes, investigation, writing, original draft, and visualization of the paper. SAG contributed to the methodology, investigation, resources, writing, and original draft of the paper. BJB contributed to the investigation, writing, and visualization of the paper.

Acknowledgment
The authors would like to express their gratitude to Majid Jazayeri for his assistance in preparing the gemstones studied in this research.

References

1. Grande, L.; Augustyn, A. Gems and Gemstones: Timeless Natural Beauty of the Mineral World; University of Chicago Press, 2009.
2. Schumann, W. Gemstones of the World; Sterling Publishing Company, 2009.
3. Duma, P. J.; Fryer, C. W. Prosopite, an Effective Turquoise Substitute. J. Gemmol. 1976, 15, 205-208.
4. Jie, L. U. O.; Xiaoyue, W. A. N. G.; Suwei, Y. U. E.; Peilian, W. A. N. G.; Xunzhong, Z. O. U.; Zhensheng, L. I. U. Gemmological Characteristic of Dyed Agate Similar to Turquoise. Gems Gemol. 2021, 23 (2), 19-25. DOI: 10.15964/j.cnki.027jgg.2021.02.003
5. Fritsch, E.; Rondeau, B.; Hainschwang, T.; Karampelas, S. Raman Spectroscopy Applied to Gemmology. In Raman Spectroscopy Applied to Earth Sciences and Cultural Heritage; Dubessy, J., Caumon, M. C., Rull, F. Eds.; Vol. 12; European Mineralogical Union, 2012.
6. Liu, L.; Yang, M.; Li, Y.; Di, J.; Chen, R.; Liu, J.; He, C. Technical Evolution and Identification of Resin-Filled Turquoise. Gems Gemol. 2021, 57 (1), 22-35. DOI: 10.5741/GEMS.57.1.22
7. Bernardino, N. D. E.; Izumi, C. M. S.; de Faria, D. L. A. Fake Turquoises Investigated by Raman Microscopy. Forensic Sci. Int. 2016, 262, 196-200. DOI: 10.1016/j.forsciint.2016.03.041
8. Chen, Q. L.; Yuan, X. Q.; Chen, J. Z.; Qi, L. J. Study on the Treatment Turquoise Using Raman Spectroscopy. Spectrosc. Spect. Anal. 2010, 30 (7), 1789-1792. DOI: 10.3964/j.issn.1000-0593(2010)07-1789-04
9. She, L. Z.; Qin, Y.; Feng, M.; Mao, Z. W.; Xu, C. Y.; Huang, F. C. A Primary Raman Microscopic Study of the Turquoise and its Role in Provenance-Tracking. Spectrosc. Spect, Anal. 2008, 28 (9), 2107-2110. DOI: 10.3964/j.issn.1000-0593(2008)09-2107-04
10. Chen, Q. L.; Qi, L. J.; Chen, J. Z. Raman Spectrum Study on Turquoise. Spectrosc. Spect. Anal. 2009, 29 (2), 406-409. DOI: 10.3964/j.issn.1000-0593(2009)02-0406-04
11. Rossi, M.; Rizzi, R.; Vergara, A.; Capitelli, F.; Altomare, A.; Bellatreccia, F.; Saviano, M.; Ghiara, R. M. Compositional variation of turquoise-group minerals from the historical collection of the Real Museo Mineralogico of the University of Naples. Mineral. Mag. 2017, 81 (6), 1405-1429. DOI: 10.1180/minmag.2017.081.055
12. Qiu, J.-T.; Qi, H.; Duan, J. Reflectance Spectroscopy Characteristics of Turquoise. Minerals 2016, 7, 3. DOI: 10.3390/min7010003
13. Moe, K. S.; Moses, T. M.; Johnson, P. Polymer-Impregnated Turquoise. Gems Gemol. 2007, 43 (2), 149-151. DOI: 10.5741/GEMS.43.2.149.
14. Aceto, M.; Calà, E.; Gulino, F.; Gullo, F.; Labate, M.; Agostino, A.; Picollo, M. The Use of UV-Visible Diffuse Reflectance Spectrophotometry for a Fast, Preliminary Authentication of Gemstones. Molecules 2022, 27 (15). DOI: 10.3390/molecules27154716
15. Aceto, M.; Calà, E.; Robotti, F.; Gil, J. P.; Castronovo, S.; Gulmini, M.; Labate, M.; Angelo Agostino, A. Non-Invasive Analyses of Italian “Ostrogothic” Jewellery: The Desana Treasure. Heritage 2023, 6 (2), 1680-1697. DOI: 10.3390/heritage6020089
16. Liu, L.; Yang, M.; Li, Y. Unique Raindrop Pattern of Turquoise from Hubei, China. Gems Gemol. 2020, 56 (3), 380-400. DOI: 10.5741/GEMS.56.3.380
17. Fischer, C.; Kakoulli, I. Multispectral and Hyperspectral Imaging Technologies in Conservation: Current Research and Potential Applications. Stud. Conserv. 2006, 51 (sup1), 3-16. DOI: 10.1179/sic.2006.51.Supplement-1.3
18. Koochakzaei, A.; Jelodarian Bidgoli, B.; Naserahari, M. A Multi-Analytical Approach to Identify Colorants in the Qajar Painted Wooden Decorative False Ceilings, From Northwest Iran. Microsc. Microanal. 2024 30 (2), 401-414. DOI: 10.1093/mam/ozae029
19. Koochakzaei, A.; Nemati Babaylou, A.; Jelodarian Bidgoli, B. Identification of Coatings on Persian Lacquer Papier Mache Penboxes by Fourier Transform Infrared Spectroscopy and Luminescence Imaging. Heritage 2021, 4, 1962-1969. DOI: 10.3390/heritage4030111
20. Koochakzaei, A.; Ghaffari, T. Identification of Traditional Black Persian Inks by Spectroscopic and Spectral Imaging Techniques: Presenting a Flowchart Method. Vib. Spectrosc. 2023, 127, 103545. DOI: 10.1016/j.vibspec.2023.103545
21. Koochakzaei, A.; Jelodarian Bidgoli, B.; Safapour, S.; Nemati-Babaylou, A. Multispectral Imaging and Hyperspectral Techniques Applied to Dyed Fibers: A Classification Approach. Prog. Color Color. Coat. 2023. DOI: 10.30509/pccc.2023.167158.1229.
22. Koochakzaei, A.; Mobasher Maghsoud, E.; Jelodarian Bidgoli, B. Non-Invasive Imaging and Spectroscopy Techniques for Identifying Historical Pigments: A Case Study of Iranian Manuscripts from the Qajar Era. Herit. Sci. 2023, 11 (1), 157. DOI: 10.1186/s40494-023-01011-5
23. Koochakzaei, A.; Oudbashi, O. A Courtly Brocade Belt Belonging to Qajar Period (1789–1925), Iran. Herit. Sci. 2023, 11 (1), 29. DOI: 10.1186/s40494-023-00875-x
24. Cosentino, A. Identification of Pigments by Multispectral Imaging; A Flowchart Method. Herit. Sci. 2014, 2 (1), 8. DOI: 10.1186/2050-7445-2-8
25. Read, P. G. Gemmology; Butterworth-Heinemann, 2005.
26. Karampelas, S.; Kiefert, L.; Bersani, D.; Vandenabeele, P. Gem Analysis. In Gems and Gemmology: An Introduction for Archaeologists, Art-Historians and Conservators, Karampelas, S., Kiefert, L., Bersani, D., Vandenabeele, P. Eds.; Springer International Publishing, 2020; pp. 39-66.
27. Choudhary, G. A New Type of Composite Turquoise. Gems Gemol. 2010, 46 (2). DOI: 10.5741/GEMS.46.2.106
28. Gyarmati, J.; Maróti, B.; Kasztovszky, Z.; Döncző, B.; Szikszai, Z.; Aradi, L. E.; Mihály, J.; Koch, G.; Szilágyi, V. Hidden Behind the Mask: An Authentication Study on the Aztec Mask of the Museum of Ethnography, Budapest, Hungary. Forensic Sci. Int. 2022, 333, 111236. DOI: 10.1016/j.forsciint.2022.111236
29. Gomez, R. B.; Del Re, N. Hyperspectral Imaging: Gem Identification and Authentication. In Multispectral and Hyperspectral Remote Sensing Instruments and Applications II; SPIE 2005, 5655, 325-332.
30. Zhao, X.-F.; Wang, Z.-Q.; Qiu, J.-T.; Song, Y. Reflectance Spectral Characteristics of Minerals in the Mboukoumassi Sylvite Deposit, Kouilou Province, Congo. Minerals 2016, 6 (2), 55. DOI: 10.3390/min6020055
31. Gurschler, C.; Serafino, G.; Spöck, G.; Bianco, A.; Kraft, M.; Kulcke, A. Spectral lmaging for Classification of Natural and Artificial Turquoise Samples. In OPTO 2002, Paris, France; 2002.
32. Kushel, D. Photographic Techniques. In The AIC Guide to Digital Photography and Conservation Documentation, 3rd ed.; Frey, F. S. Ed.; American Institute for Conservation of Historic and Artistic Works, 2017.
33. Cosentino, A. Practical Notes on Ultraviolet Technical Photography for Art Examination. Conserv. Patrim. 2015, 21, 53–62. DOI: 10.14568/cp2015006
34. Moon, T.; Schilling, M. R.; Thirkettle, S. A Note on the Use of False-Color Infrared Photography in Conservation. Stud. Conserv. 1992, 37 (1), 42-52. DOI: 10.2307/1506436
35. Frost, R. L.; Reddy, B. J.; Martens, W. N.; Weier, M. The Molecular Structure of the Phosphate Mineral turquoise—A Raman Spectroscopic Study. J. Mol. Struct. 2006, 788 (1), 224-231. DOI: 10.1016/j.molstruc.2005.12.003
36. Čejka, J.; Sejkora, J.; Macek, I.; Malíková, R.; Wang, L.; Scholz, R.; Xi, Y.; Frost, R. L. Raman and Infrared Spectroscopic Study of Turquoise Minerals. Spectrochim. Acta A: Mol. Biomol. Spectrosc. 2015, 149, 173-182. DOI: 10.1016/j.saa.2015.04.029
37. Freeman, J. J.; Wang, A.; Kuebler, K. E.; Jolliff, B. L.; Haskin, L. A. Characterization of Natural Feldspars by Raman Spectroscopy for Future Planetary Exploration. Can. Mineral. 2008, 46 (6), 1477-1500. DOI: 10.3749/canmin.46.6.1477
38. Velde, B.; Boyer, H. Raman Icroprobe Spectra of Naturally Shocked Microcline Feldspars. J. Geophys. Res. Solid Earth 1985, 90 (B5), 3675-3681. DOI: 10.1029/JB090iB05p03675
39. Nigro, L.; Gallo, E.; Gharib; Mura, F.; Macrì, M.; Rinaldi, T. An Egyptian Green Schist Palette and an Amazonite Gemstone from the “Palace of the Copper Axes” at Batrawy, Jordan. Vicino Oriente 2020, 24, 1-26. DOI: 10.53131/VO2724-587X2020_1
40. Querré, G.; Vigier, E.; Cassen, S. La parure en callaïs du Néolithique européen; Archaeopress, 2019.

Alireza Koochakzaei, Samane Alizadeh Gharetapeh, and Behrooz Jelodarian Bidgoli are with the Faculty of Cultural Materials Conservation at Tabriz Islamic Art University, in Tabriz, I.R. Iran. Direct correspondence to a.koochakzaei@tabriziau.ac.ir or Alireza.k.1989@gmail.com