1. Introduction

Photoluminescence (PL) is commonly excited in gemstone materials when illuminated by appropriate radiation wavelengths. Atomic-scale features within, such as optical centers and impurities, absorb the incident energy and emit characteristic light that can be detected. In its simplest definition, PL can be divided into fluorescence and phosphorescence, where the former occurs shortly following photoexcitation (< 50 ns), whereas the latter luminescence is longer-lived. Traditionally, the study of a mineral’s fluorescence utilizes cathode, x-ray, or ultra-violet (UV) light as excitation [1]. Mercury vapor lamp excitation has been popular since the 1930s due to its powerful UV component coupled with its affordability, accessibility, and established safety standards [1–3]. Visual or spectroscopic evaluation of fluorescence under the lamp’s filtered longwave (LW, 365 nm) and shortwave (SW, 254 nm) UV features became a key part of routine gemstone testing protocols following the development of the mercury tube light [4]. However, the emission spectra of gem testing UV lamps usually include a mixture of all the mercury emission lines, with differing longwave and shortwave component ratios [5]. Lamp filter selection and degradation further changes their emission spectra, in turn affecting the fluorescence of the illuminated sample. This casts doubt on the reliability of fluorescence observations and their use as conclusive evidence in modern gemstone identification [5–7].

PL spectroscopy is based upon the illumination of a sample with a suitable light source, with a spectrometer recording and analyzing the spectral density of the luminescence. Compared to visual evaluation, PL spectroscopy offers several improvements to the reliability and reproducibility of visual fluorescence analysis. The spectral distribution of the luminescence signal can be quantified, and subtle differences in emission can be distinguished. PL has become an important identification technique for screening natural gemstones from laboratory-grown counterparts, mineral identification, and treatment detection due to its rapid, non-destructive capabilities [7–29]. Microscope-based PL systems that use high resolution spectrophotometers are popular in major gemological laboratories, employing excitation wavelengths ranging from 325 nm to 830 nm. Monochromatic light sources such as lasers are preferred as they are the only sources that can be used for Raman spectroscopy, often simultaneously collected with the same instrument. The strength of narrow-band light emitting diodes (LED) can be suitable for PL analysis, however, the emission lineshape is broad, in turn broadening features in the PL spectra. Broadband light sources that are coupled to a monochromator or spectrally filtered are comparatively weak or suffer from leakage that introduces artifacts in the luminescence spectra, complicating analysis [12]. For advanced PL systems equipped with multiple lasers, changes to the excitation wavelength – especially when switching between UV and visible ranges – requires the replacement of corresponding optics, detrimentally affecting the cost, speed and ease of use of the system.

The limited options for lasers, filters, and optical components, mean that UV-excited PL spectroscopy systems are relatively uncommon, with few published studies on gemstone fluorescence [5,12,26,30,31]. Considering the traditional use of SWUV for visual fluorescence characterization, it is expected that complementary SWUV excited fluorescence spectroscopy will support more rigorous and detailed gemstone analysis. Additionally, the 405 nm laser has proven to be popular for the analysis of diamond and other gemstones as it can excite several visible features [12,17,25,32].

This study presents a simple and accessible multi-excitation PL spectroscopy system using 261 and 405 nm lasers, designed to reveal characteristic PL features emitting between 270 to 1000 nm to achieve accurate, rapid, and simple gemstone identification. Applications relating to popular gemstones including diamond, corundum (ruby and sapphire), emerald, and pearl, have been studied and the corresponding identification features will be presented. Sufficient instrument design detail is included for reproduction by interested parties. The instrument’s ease of use and affordability is expected to be attractive to gemstone industry members, incorporating advanced spectroscopy techniques to enhance reliable fluorescence analysis.

2. Background

The luminescence behavior of gemstones when exposed to the filtered output of a mercury vapor lamp are commonly considered for gemstone identification, and may in some cases help separate natural and laboratory-grown gemstones or detect artificial treatments used to improve a specimen’s appearance. Luminescence properties that are studied include emission color, brightness, distribution, and luminescence decay time (fluorescence vs. phosphorescence). For example, when considering colored stones (non-diamond materials), it may be possible to separate laboratory-grown and natural rubies based on the former material’s strong red fluorescence associated with $\text {Cr}^{\text {3+}}$. This difference originates from the relatively high purity of the manmade material compared to natural corundum, which contains trace elements such as iron or titanium that effectively quench the fluorescence. However, rubies originating from marble deposits may also display high levels of fluorescence due to their low iron growth environments, making fluorescence color and intensity-based screening for natural and lab-grown rubies unreliable [33]. Laboratory-grown and thermally enhanced natural sapphires may exhibit “chalky” fluorescence under SWUV, potentially allowing them to be separated from natural samples [34–37]. Among diamonds, natural pink color samples typically fluoresce blue under LWUV, whereas the majority of pink treated natural and laboratory-grown diamonds show orange-red fluorescence due to the presence of nitrogen-vacancy (NV, where N = nitrogen and V = vacancy) centers [24,38–40]. Only about 0.6% of natural pink diamonds, known as “Golconda pink” diamonds in the gem trade, are colored by NV defects, often characterized by pale color saturations [40,41]. There are also applications for pearls, where LWUV can be used to excite blue fluorescence from brightening agents used to enhance the color of white pearls [42]. Other pearl treatments can also be identified. For example, the tryptophan in a naturally colored pearl’s nacre shows greenish blue fluorescence to SWUV, yet treated colored pearls are usually inert [21,43]. Sometimes it is not the fluorescence color that is key. Clarity enhanced emeralds, where surface reaching fractures have been filled with material to minimize their visibility, produce recognizable fluorescence patterns [44,45]. Finally, the decay time of luminescence can also be relevant. High pressure, high temperature (HPHT) grown diamonds exhibit strong, long-lived greenish blue phosphorescence following SWUV illumination, a feature that is exceedingly rare for natural diamonds [24,46–48].

While a mercury lamp can be a helpful tool for gemologists, it is known that a conclusive identification of gemstones typically requires combining fluorescence observations with other gemological tests. Visual observation of luminescence has limitations, specifically in terms of reproducibility and sensitivity. Firstly, human eyes are only sensitive to emissions in the visible range. Although human vision is adept at differentiating minor differences in color, color and brightness interpretations are qualitative, making it difficult to record fluorescence features with precision. Additionally, distinct fluorescing centers may produce similar colors, or the contributions of multiple differently colored emissions may be combined (e.g. blue + yellow = green). Variations in the excitation source spectra also introduce further complications to achieving consistent and reproducible observations. The combination gem testing LWUV lamps (primary emission at 365 nm) that are commonly used in gemology often emit at 404 and 435 nm, as well as a broad band that extends from the UV to the visible region [5]. Similarly, the SWUV lamp component emits at 254, 315 and 365 nm. These secondary features cannot be completely removed by the cobalt and nickel sulfate or colored glass filters used in gem testing lamps [5,6]. Improved filtration can be achieved using advanced filter design and coatings, but this typically results in a detrimental reduction in output intensity of the desired wavelengths, as well as increasing the material costs. Filter degradation can further exacerbate light leakage. Alternatives, such as UV LED sources, also produce a sideband extending to the visible region. These light leakage features could either stimulate additional fluorescence emissions, affecting the combined fluorescence color, or interfere with visual observations [5].

Lasers emitting in the shortwave UV region between 190 and 280 nm can be used to generate sample fluorescence, benefiting from strong narrow band emissions. Traditionally, shortwave UV lasers are generated by frequency quadrupling NIR lasers emitting at ~1000 nm. Drawbacks include their expense, large size, and temperature-dependent output powers, limiting their use for gemstone analysis. Alternatively, excimer laser at 193 nm can also be used in PL spectroscopy, but the requirement of using argon fluoride gas limits its application [49]. Recently, continuous-wave (CW) shortwave UV lasers emitting at 261 nm, based on frequency-doubling praseodymium lasers pumped by a blue laser diode, became commercially available [50]. These compact, low-cost, and simple UV lasers enable PL analysis with excitation wavelengths close to those traditionally used for gemological applications.

Laser emission in the longwave UV region between 315 to 400 nm can be generated by frequency-tripling a NIR laser or using a diode-based CW UV laser. Compared to visible wavelength lasers, these lasers are more costly and suffer from lower output powers. Consequently, longwave UV lasers are rarely used for PL spectroscopy. 405 nm lasers, which straddle the boundary between UV and visible light, are used for gemstone PL and Raman analysis as they can effectively excite luminescence features in the visible to near infrared (Vis-NIR) region for popular minerals [12,17,32,51]. This laser wavelength is able to excite the key defects that are considered when using a LWUV lamp for gemological applications, and is especially useful coupled with a spectrometer [16,20]. For example, diamond’s N3 ($\text {N}_\text {3}\text {V}^\text {0}$) fluorescence at 415 nm is used for screening natural and laboratory-grown colorless and near-colorless diamonds and simulants [16,20,23,24,52]. The 405 nm laser can also be used to detect differences in the chromium-related emission spectra of corundum, beryl, spinel, and tanzanite [20]. Finally, the blue fluorescence from optical brightening agents can be used to detect whitening treatments in pearls [42,43].

The traditional 365 and 254 nm LW and SW UV emissions from the mercury vapor lamp were somewhat arbitrarily selected as the convention for gemstone fluorescence observations based on their availability and cost [1–3]. Since 405 and 261 nm illumination can effectively excite fluorescence features from gemstones, they should also be suitable for gemstone analysis. New sets of criteria need to be developed to account for the different detailed luminescence response that they induce compared to those excited by LW and SWUV sources.

When collecting a typical PL measurement, it is possible to obtain both Raman and luminescence features within the same spectrum, providing information on the host crystal and impurities or defects. Raman scattering occurs due to the inelastic scattering of photons with a loss or gain of energy to the intrinsic crystal lattice vibrations. If the spectrum is plotted on a Raman shift scale ($\text {cm}^\text {-1}$), the Raman feature positions are independent of the excitation wavelength. However, if spectra are recorded on an absolute wavelength scale (as is typical for luminescence spectra), the Raman feature positions will depend on the excitation wavelength. The PL emission spectrum of a defect, which relates to the energy separation between the defect states and their vibronic structure, is unambiguously characteristic of the specific species of optical center. When plotted on an absolute wavelength scale, PL feature positions are independent of the excitation wavelength. The different wavelength dependence of Raman and PL signals means it is possible to distinguish their contributions to the recorded spectra. Although Raman and PL emissions can be simultaneously observed, their requirements for optimal detection differ. A high spectral resolution is necessary to resolve the Raman signals produced by minerals, over a limited spectral range of 200 to 2000 $\text {cm}^\text {-1}$ [25,53]. Meanwhile, PL emission spectra can span several hundred nanometers, yet the spectral resolution requirement is not as strict as room temperature emission features are rarely narrower than 1 nm. Gemological laboratories primarily employ PL spectroscopy due to its nondestructive nature and high sensitivity to atomic defect emissions, offering potential for treatment detection and screening of natural gemstones. Conversely, while Raman spectroscopy is suited for mineral identification, it generally cannot be used to distinguish natural vs. treated or laboratory-grown gemstones as their primary crystal structure is the same.

3. Experimental details

The schematic layout of the multi-excitation PL spectroscopy system is presented in Fig. 1. A 10 mW, 0.1 nm linewidth, 261 nm laser (Model261, UVC Photonics) mounted on a temperature-controlled breadboard (PTC-1, Thorlabs) and a fiber-coupled 50 mW, 0.1 nm linewidth, 405 nm laser (FC-D-405-50mW, CNI) are used to excite the PL features. Fiber collimators (RC08SMA-F01 or RC08 SMA-P01, Thorlabs) are used in this design to collimate laser to the system and to collect emission from the sample to send to the spectrometer. The 261 nm PL sub-system utilizes four optical filters to isolate the emission from the excitation. Two shortpass filters (XUV0400, Asahi and FF01-300/SP-25, Semrock) are used to block the leaking emission lines at 443, 522, and 722 nm from the 261 nm laser. A dichroic beam splitter (Di01-R266-25x36, Semrock) reflects the laser light to the sample and transmits the emission signal to the collection optics. A longpass filter (LP02-266RU-25, Semrock) blocks the excitation laser from entering the spectrometer. The 405 nm PL sub-system includes a bandpass laser clean-up filter (LL01-405, Semrock) to transmit only the precisely defined laser line, a dichroic beam splitter (FF409-Di03, Semrock) to reflect the laser to the sample while transmitting the sample’s PL signal, and a longpass filter (LP02-407RU-25) to further isolated the emission from the excitation. To simplify the setup, a custom 405 nm spectroscopy probe (SPC-R405 Spectra Solution) without the focusing lens barrel is used to replace the 405 nm PL sub-system. A flip mirror (FlipMirror II, Baader Planetarium) is used to select the 261 or 405 nm excitation options. A 20X/0.33NA reflective objective (39-141, Edmund Optics) is used as the objective optics. The collected PL signal is analyzed by a TE-cooled modular spectrometer (QE-pro, Ocean Insight) or a compact spectrometer (AvaSpec-Mini, Avantes) with 1.9 or 1.2 nm spectral resolution, respectively. Based on the specification of the optical filters used in the system, the sensing range for the 261 nm and the 405 nm PL sub-systems are 270 to 750 nm and 408 to 1000 nm, respectively.

 

Fig. 1. Schematic of the multi-excitation PL spectroscopy. TC: temperature-controlled breadboard, FC1-5: Fiber collimators, SP: shortpass filter, DBS1-2: dichroic beam splitters, LP1-2: longpass filter, BP: bandpass filter, FM: flipping mirror, and OBJ: reflective objective.

Download Full Size | PDF

The experimental prototype in Fig. 2 was built to validate the concept of multi-excitation PL spectroscopy for gemstone analysis. The relay between laser, optics and spectrometer are based on 600 and 100 $\mu$m core diameter fibers for the 261 and 405 nm PL sub-systems, respectively. Finally, the optics, including the fiber couplers, filters, dichroic beam splitters, flipping mirror, and the reflective objective, are mounted on a manual x-y-z translation stage for proper sample positioning. Faceted gemstone samples are held below the objective, presenting the table facet (top flat surface on a faceted stone) to the beam path, as illustrated in Fig. 1. The entire setup is mounted on a $12^{\prime\prime}\times12^{\prime\prime}$ aluminum optical breadboard (MB12, Thorlabs) for portable applications. Due to the loss from the reflective optics used in the system, the laser power at the sample position are ~800 $\mu$W and 45 mW for the 261 and 405 nm lasers, respectively. One of the primary advantages of this multi-excitation PL spectroscopy setup is that an area of interest on the sample could be investigated using both 261 and 405 nm lasers, detecting luminescence across the UV-Vis-NIR range due to the selected deep UV-enhanced reflective objective. Additionally, its long working distance (18.88 mm) is compatible with commercially available cryogenic sample stages for temperature dependent PL analysis. The total cost of the prototype spectrometer is approximately $\$$50,000 U.S. dollars, a significant discount compared to commercially available PL microscopes, that start at $\$$150,000 U.S. dollars.

 

Fig. 2. Experimental prototype of the multi-excitation photoluminescence spectroscopy.

Download Full Size | PDF

4. Results and discussion

To validate the performance of this PL spectroscopy system, measurements were collected on a range of commercially important gemstone samples, including natural and lab-grown diamonds, colored gemstones, and naturally colored and treated pearls. For each type of species at least 2 or more samples were used to collect reference spectra. The tested samples included gemstones from GIA’s research sets and Dr. Edward J. Gübelin Gem Collection, all previously conclusively identified by experienced gemologists using gemological observations and advanced techniques including Fourier transform infrared (FTIR) and UV-Vis-NIR absorption, reflectance, PL, and Raman spectroscopies, and real-time microradiography. All measurements for faceted samples were collected from the table facet. Spectrometer integration times ranged from 500 ms – 6 s and 10 ms – 1 s for 261 and 405 nm excited PL experiments, respectively. All measurements were conducted at room temperature. The spectrometers’ spectral responses were adjusted using an integrating sphere based halogen calibration lamp (AVASPHERE-50-LS-HAL-CAL-12, Avantes), accounting for differences in transmission. Finally, the spectral baseline was corrected by subtracting dark readings from analysis locations with the response blocked by optical filters. Characteristic PL features of each sample group will be discussed.

4.1 Diamonds

Diamonds are conventionally classified according to the detection of nitrogen- or boron-related features using FTIR spectra [23,54–58]. Briefly, diamonds that contain detectable nitrogen are classed as type Ia (aggregated) or Ib (isolated). Those with no detectable nitrogen nor boron are type IIa, whereas those with boron are type IIb. When considering D-to-Z color (colorless – faintly colored) diamonds, the vast majority of natural diamonds are type Ia (~98 – 99%) whereas laboratory-grown diamonds produced by HPHT or chemical vapor deposition (CVD) methods for gem applications are type IIa or IIb [14,59]. Weak type Ib CVD-grown and type Ia HPHT-grown diamonds (produced by higher temperature growth or annealing) are rarely encountered as they suffer from poorer colors with brown or yellow hues that degrade their value [60–63]. For further discussion on nitrogen in laboratory-grown diamonds see the reviews by Ashfold, Green and D’Haenens-Johansson et al. [23,24,57]. Examples of characteristic PL spectra for natural type Ia and IIa, HPHT- and CVD-grown, and annealed CVD-grown diamonds excited by 261 and 405 nm lasers are presented in Fig. 3.

 

Fig. 3. 261 and 405 nm PL spectra for (a) natural type Ia, (b) natural type IIa, (c) HPHT-grown, (d) CVD-as grown, and (e) annealed CVD-grown diamonds; and the phosphorescence spectrum for HPHT -grown diamond generated by 261 nm laser excitation is also presented in (c). The spectra were either normalized to diamond’s Raman peak or the dominant PL feature.

Download Full Size | PDF

Figure 3(a) shows PL spectra of a typical natural type Ia diamond that is dominated by the N3 defect fluorescence with zero phonon line (ZPL) at 415 nm and a vibronic sideband extending to longer wavelengths. The intensity of the diamond Raman peak (1332 $\text {cm}^\text {-1}$) at 428 nm under 405 nm excitation is usually much weaker than the N3 fluorescence and may be overwhelmed by the N3 defect’s vibronic sideband. When using the 261 nm excitation, the diamond Raman peak is detected at 270 nm, with a second order Raman peak (2664 $\text {cm}^\text {-1}$) at 279 nm. Type Ia samples with aggregated nitrogen in the form or nitrogen pairs, known as A-aggregates, have strong absorption below 300 nm. Sometimes referred to as the “secondary absorption edge” it is thought to arise from overlapping vibronic bands associated with defects that correlate with A-aggregates, as well as photoconductivity from A-aggregates [56]. Consequently, N3-containing type Ia diamonds generally show weaker Raman peaks with 261 nm excitation compared to other diamond types. This laser excitation produces N3 fluorescence, but in this case the intensity level is similar to that of the Raman peak, with the excitation efficiency being noticeably poorer compared to using a 405 nm laser. A broad luminescence band centered at 520 nm is detected, dominating the spectrum. Based on the emission spectra, the fluorescence color shifts from blue to greenish blue when changing from 405 to 261 nm excitation.

The natural type IIa diamond produces few defect-related emissions under similar conditions (Fig. 3(b)), with the diamond Raman peak dominating the spectra. Besides the Raman peak, neutral nitrogen vacancy ($\text {NV}^\text {0}$) centers with ZPL at 575 nm can also be excited by both excitation wavelengths. Additionally, the 261 nm laser excites several broad band emission features between 400 and 700 nm, creating whitish color emission. Due to the overall low nitrogen concentration, N3 fluorescence may not be detectable in some natural type IIa diamonds.

D-to-Z colored HPHT-grown diamonds are typically type IIa or weak type IIb due to boron contamination during growth, primarily from the source of carbon [24,64–66]. Characteristic spectra are shown in Fig. 3(c), where the Raman diamond peak dominates. The 261 nm laser is also able to induce boron-related phosphorescence centered at $\sim$500 nm [46,48]. The long-lived phosphorescent nature of this band can be confirmed by recording the luminescence spectrum immediately following the extinction of the incident 261 nm laser. Although not exclusive to HPHT-diamonds, this greenish-blue phosphorescence is rare enough among natural diamonds to be useful as a screening indicator [47,52,65,66].

Figure 3(d) shows that different fluorescence features are excited by the 261 and 405 nm PL in an “as-grown” (untreated) CVD-grown diamond. Both lasers excited the $\text {NV}^\text {0}$ center, whereas the nitrogen-related 389 and 468 nm defects were only excited by the 261 and 405 nm lasers, respectively. These latter two defects are common in as-grown CVD diamonds and can be used for screening purposes [24,60,67].

The color of CVD diamonds is frequently improved by HPHT annealing [14,24,60], resulting in changes to the PL spectra. Figure 3(e) shows the PL features of an annealed CVD-grown diamond. The 261 nm laser excites a broad band at $\sim$320 nm, which has so far not been observed in as-grown CVD diamonds, suggesting it could be useful for annealing treatment identification. Though shifted to a lower wavelength, the feature appears similar to the broad band centered around 340 nm detected in HPHT-annealed CVD diamonds (also produced by Gemesis) by Wassell et al. following excitation with deep-UV light (190 – 227 nm, 50 – 100 $\mu$s delay) [31]. The 261 nm laser also excited an asymmetric broad band $\sim$510 nm. Unlike deep-UV excited room temperature spectra by Wassell et al. and McGuiness et al., neither 499 nor 503 nm zero-phonon lines were detected, suggesting this band differs despite being in a similar region to their associated vibronic bands [27,31]. The 405 nm laser excited an asymmetric ramp-like feature peaking around 460 nm, comparable to a rise in luminescence that is resolved into an unidentified series of peaks at 77 K using 325 nm excitation for Gemesis HPHT annealed CVD-grown diamonds [68]. The 405 nm laser was also efficient at exciting the negatively charged silicon-vacancy ($\text {SiV}^\text {-}$) defect at 737 nm, a feature that is common in CVD-grown diamonds (both as-grown and treated) [24,60,69]. Although annealed CVD samples often exhibit turquoise phosphorescence under shortwave UV excitation, this features is too weak and short-lived to be recorded by this PL system [31,68]. Careful synchronization between laser and the spectrometer is required to record fast decaying luminescence signals.

4.2 Colored gemstones

Examples of characteristic PL spectra for colored gemstones are presented in Figs. 4–6. Heat treatment is commonly applied to corundum (ruby and sapphire) to improve the visual appearance, such as body color and transparency. Unheated and heat-treated blue sapphire samples can be separated based on the combination of FTIR absorption spectroscopy, UV-Vis-NIR absorption spectroscopy, microscopy, and SWUV fluorescence by gemologists [34,35,70,71]. The relative intensity of FTIR and UV-Vis-NIR absorption spectroscopy peaks and bands can be optically sensitive to heat treatment [70,72]. Observation of the sample’s inclusions under a microscope can also reveal evidence of heat treatment. For example, a partially healed fracture is a useful indicator of heat treatment in corundum [73]. Traditionally, SWUV lamps have been used to aid in heat treatment identification. Heat-treated sapphires may show “chalky” color fluorescence under SWUV excitation, which is thought to be related to the $\text {Ti}^\text {4+}$ ions and Ti-Al vacancies in corundum created by rutile ($\text {TiO}_\text {2}$) being dissolved into the corundum lattice during heat treatment [36,37,70]. As a result, the intensity and distribution of this “chalky” fluorescence is positively correlated with the heating temperature. We note that heating the sample up to 1500$^{\circ }$C may start to annihilate the fluorescence bands [70]. The iron concentrations can also quench this fluorescence by preferentially compensating the $\text {Ti}^\text {4+}$ over the Al-Ti vacancy. Such limitations lower the usability of SWUV lamps in heat treatment identification.

 

Fig. 4. (a) 261 and 405 nm PL spectra for a heat-treated blue sapphire. (b) Normalized 261 nm PL spectra for heat-treated and laboratory-grown blue sapphires and rubies, with a typical spectrum for an unheated sapphire for comparison. The spectra have been offset in the vertical axis for clarity. The heat-treated and lab-grown ruby spectra were normalized within the displayed range; and the other samples were normalized to the $\text {Cr}^{\text {3+}}$ emission peak (not shown in this figure).

Download Full Size | PDF

In a preliminary study, six samples were examined: a flame-fusion lab-grown blue sapphire, a flame-fusion lab-grown ruby, a heat-treated metamorphic Madagascan blue sapphire, a heat-treated marble-hosted Burmese ruby and two unheated metamorphic Sri Lankan blue sapphires. Figure 4(a) shows a heat-treated metamorphic Madagascan blue sapphire’s PL under 405 and 261 nm excitations. The major PL features are the strong and sharp chromium emission peaks at 693 and 694 nm, which can be detected by both excitations. Due to the limited spectral resolution of our system, the chromium emission doublet presents as a single band. A broad emission band between 400 and 550 nm was excited by the 261 nm laser, as presented in Fig. 4(b).This broadband emission captures the “chalky” color observed under SWUV illumination during conventional visual evaluation. The band was also detected in the heat-treated ruby and the laboratory-grown sapphire and ruby. While one of the unheated samples did not display detectable fluorescence in this region, the remaining unheated sapphire displayed a similar emission band, however, slightly shifted to the green region compared to the heat-treated samples, as presented in Fig. 4(b). While not studied here, anecdotal observations of unheated basalt-related sapphires may also display similar fluorescence, thought to be attributed to their relatively higher heat growth environment. While PL spectroscopy has the exciting potential to help identify heat treatment in corundum, future studies with higher sample sizes and a broader range of materials are required to better explore the correlation between “chalky” fluorescence and heat-treatment in corundum. The higher sensitivity and reproducibility provided by PL spectroscopy enhances the reliability of gemstone fluorescence evaluation.

Multi-wavelength PL spectroscopy can also be used for emerald screening. Emerald is one of the most important colored gemstones, and particularly popular in the United States and parts of Asia. Emerald is a variety of the mineral beryl, and is characterized by a beautiful green to bluish green body color caused by trace amounts of chromium or vanadium, with often both elements present. One identification challenge for emeralds is that there are many other colored gemstones that look similar, such as green colored garnet, topaz, and tourmaline. The majority of green gemstones can be screened out from emerald based on their characteristic fluoresce features [20,28]; however, uvite, the green color magnesium/iron-rich member of the tourmaline group, shares a similar spectral signature. Uvite’s separation from emerald can become a challenge if the spectral resolution of the system is insufficient to distinguish the minor difference in these minerals’ $\text {Cr}^\text {3+}$ emission peak positions [28]. Figure 5 shows typical PL spectra for uvite and emerald under 405 and 261 nm excitations. For the former excitation, their respective $\text {Cr}^\text {3+}$ peaks are located at 682 and 684 nm, along with a broad fluorescence band around 720 nm. While the 405 nm excited PL spectra for the two minerals appear quite similar, the 261 nm excited spectra reveal a major difference that can be leveraged for their separation. The 261 nm excited PL spectral features for emerald remain unchanged compared to those for the 405 nm excitation, yet uvite’s is drastically different, being dominated by a broad emission band between 400 to 750 nm without clear $\text {Cr}^\text {3+}$ emission. The difference in luminescence response enables easy separation between emerald and uvite based on PL spectroscopy.

 

Fig. 5. PL spectra for green color gemstones. (a) 405 nm PL spectra for uvite and emerald. (b) 261 nm PL spectra for uvite and emerald. The spectra were normalized to the dominant PL feature.

Download Full Size | PDF

Figure 6 shows typical PL spectra of a range of other colored gemstones under 405 and 261 nm laser excitations. Under traditional UV lamp evaluation, these samples show different colors or intensities in their fluorescence response under LWUV and SWUV lamps. Although Raman scattering is typically observed near the excitation source under 405 nm laser illumination, it may not be detected if the scattering signal is absorbed by a strong absorption band in the material or overwhelmed by a fluorescence background. To avoid confusion, we note that many photoluminescence features at longer wavelengths are associated with chromium [28]. We noted that some leaking laser signal at 404.5 nm could also be detected in our 405 nm PL spectra. Although the origin of SWUV-induced luminescence features are often unknown, their existence or absence may still serve as an important evidence for gemstone identification or treatment detection, as illustrated by the following examples. The PL spectra of kunzite, a variety of the mineral spodumene, exhibits major differences under 405 and 261 nm excitations, as presented in Fig. 6(a), resulting in distinct perceived colors. A strong orange emission band centered at 605 nm was detected under 405 nm excitation, while emission bands at 298 nm and 425 nm were excited by the 261 nm laser. The gemstone amethyst, a purple color variety of quartz, was weakly fluorescing under 405 nm excitation, but a strong, deep red color luminescence band centered at 740 nm was detected under 261 nm excitation, as shown in Fig. 6(b). Chromium emission at 680 nm was the major fluorescence feature under both 405 and 261 nm excitations for alexandrite, a variety of chrysoberyl, as presented in Fig. 6(c); however, a broad emission band at 535 nm only appeared under 261 nm excitation. Similarly, chromium emissions at 692 and 709 nm were the major PL features in zoisite under 405 nm excitation (Fig. 6(d)), yet under 261 nm excitation the only emission detected was a band at 318 nm. Figure 6(e) shows the PL spectra for pink liddicoatite tourmaline. Several peaks ranging between 560 to 650 nm were present under 405 nm excitation, yet the major feature under 261 nm excitation was an emission band located at 311 nm. Finally, the PL spectra of sunstone labradorite are presented in Fig. 6(f). Two emission bands at 560 and 697 nm detected under 405 nm excitation where absent under 261 nm excitation, which was instead dominated by emission bands located at 384 and 713 nm. These examples illustrate differences in the multi-excitation PL spectra for colored stones which could be exploited for their identification and separation.

 

Fig. 6. Typical 261 and 405 nm PL spectra of colored gemstones with different fluorescence color under short wave UV and long wave UV mineral lamps, including (a) spodumene(kunzite), (b) quartz(amethyst), (c) chrysoberyl(alexandrite), (d) zoisite, (e) liddicoatite, and (f) labradorite(sunstone).

Download Full Size | PDF

4.3 Pearls

PL spectroscopy can also serve as an important indicator to separate naturally colored and treated color pearls, and to detect optically brightened and bleached white colored pearls. Pearls’ color can naturally occur in a broad range of hues from light to dark, mainly due to the natural pigments found in their host mollusks. Further color modification after harvest, such as dyeing and irradiation, can significantly alter pearl’s appearance [74–77]. Such treatment can convert those less appealing colors into more attractive coloration to increase their commercial value. Certain types of white colored pearls may also undergo additional process such as bleaching and optical brightening to enhance or stabilize visual appearance [42]. The key luminescence features considered include the amino acid tryptophan (Trp or W) originating in pearl’s nacre layer, emitting at 340 nm, and the artificially applied optical brightening agent that emits at 430 nm [21,42]. Naturally colored pearls typically show strong tryptophan fluorescence, but color treatments and processing damage the nacre layer, reducing the intensity of the 340 nm fluorescence band. The optical brightening agent applied to some white colored pearls for color and luster enhancement usually produces additional blue fluorescence.

Figure 7 compares the difference of tryptophan fluorescence between naturally colored and treated colored pearls under 261 nm laser excitation. Based on the result, the fluorescence intensity of naturally colored pearls were noticeably higher than color treated pearls. Figure 8 shows the PL spectra of naturally colored, optical brightened, and bleached white color pearls under 405 and 261 nm laser excitation. Under 405 nm excitation, naturally colored and bleached white color pearls showed emission band centered around 480 nm and 460 nm, respectively. The brightened white color pearl showed noticeable blue fluorescence band around 420 nm, which comes from the remaining optical brightening agent and fluorescence maxima at around 440 nm.

 

Fig. 7. 261 nm PL spectra for naturally colored golden, grey, and pink color pearls, and treated color golden, dark grey, and dark blue pearls.

Download Full Size | PDF

 

Fig. 8. PL spectra for white pearls. (a) Normalized 405 nm PL spectra for naturally colored, brightened, and bleached white color pearls. (b) 261 nm PL for naturally colored, brightened, and bleached white color pearls.

Download Full Size | PDF

Figure 8(b) compared the tryptophan fluorescence level between the 3 different white color pearls. Naturally colored white pearl had the strongest fluorescence level, while optical brightened white pearl showed slightly lower fluorescence intensity along with detectable fluorescence band around 430 nm, possibly from the remaining optical brightening agent. The bleached white color pearl presented noticeable lower fluorescence intensity compared with the naturally colored white sample, along with emission band around 455 nm. Using a focused laser as excitation can significantly reduce the required integration time down to 1% but can still achieve similar spectral quality compared with the previously reported UV LED based system [21]. The illuminated spot size can also be reduced from 2 mm down to several um in diameter by using the reflective objective. Although the UV irradiation does not affect the visual appearance of the sample, it may still reduce the durability of pearl’s nacre layer [78] .

5. Conclusions

This study proposed a multi-excitation PL spectroscopy to extend the capability of using luminescence features in gemstone identification. The proposed system combines PL sub-systems using 261 and 405 nm lasers, with a reflective objective and a flipping mirror for wavelength selection. Compact fiber spectrometers were used as the sensors to analyze gemstone luminescence features between 270 and 1000 nm, covering the UV to NIR region. Both lasers were aligned together to illuminate the same spot of the sample, allowing the sample to be analyzed by both wavelengths without the need to switch optical components. Commercially important gemstones samples, including diamond, corundum, emerald and green colored gemstones, pearls, and some other colored gemstones were selected as examples to demonstrate its capabilities for gemstone analysis. Luminescence features from diamonds, such as fluorescence in the UV region from CVD-grown diamonds and the long lasting phosphorescence from HPHT-grown diamonds, can be used to conclusively identify lab-grown diamonds. The greenish blue broadband emission has the potential to be used in identifying heat-treated sapphires and separating laboratory-grown sapphires from natural. The difference in emission spectra created by 261 and 405 nm laser can be used to separate emeralds and green colored gemstones, and potentially in other colored gemstones, such as spodumene, quartz, chrysoberyl, zoisite, liddicoatite, and labradorite. Finally, the intensity of the tryptophan fluorescence at 340 nm can be used to identify colored treatment in pearls. These results suggest this spectroscopy-based approach could be a complementary technique or potential replacement to traditional gemstone fluorescence evaluations. The difference in excitation wavelengths between this proposed PL system and the mercury vapor lamp could introduce variations in the luminescence behavior from some gemstones. Thus, a statistically meaningful spectral database for commonly encountered natural, lab-grown, and treated gemstone samples should be built and disclosed to the public. For each species, multiple samples with different colors or appearances should be included in a reference spectra collection. The technical advances in gemstone synthesis and treatments are continuously evolving, increasing the complexity of gemstone identification. Future studies will target detection of luminescence features related to new treatment techniques and lab-grown gemstones.