1. INTRODUCTION

Pearl, a nacreous organic gemstone formed within a mollusk, has been treasured by humans since ancient times. It is composed of aragonite and conchiolin, secreted by a mollusk’s mantle tissue, similar to the way its shell is formed. The main component of aragonite is calcium carbonate crystal, where conchiolin is made of complex organic macromolecules, mainly proteins and polysaccharides, which glue aragonites to construct pearl’s nacre layers. Pearl can form naturally or by human intervention using various types of culturing methods. Natural pearls once were the major source of pearl until 1900s, when cultured pearls started to dominate the market both due to the decline of natural pearl resources and fast developments in pearl culturing techniques. It is a multibillion dollar industry and still growing [1].

A pearl’s value depends on its many quality factors and rarity. Color is one of the most important value factors in pearls. Pearls can naturally occur in a broad range of hues of various saturations and tones from light to dark, mostly due to the natural pigments found in their host mollusks. Usually after harvest, pearls are routinely processed before hitting the market, and these processes include general cleaning and mild polishing (tumbling). Further color modification by dyeing or irradiation, which will significantly alter the appearance of the pearls, is also a common practice to treat pearls of less appealing colors into more attractive coloration and subsequently increase their commercial values. Certain types of white colored pearls also normally undergo additional treatments such as bleaching (mainly using hydrogen peroxide), maeshori (a type of luster enhancement involving alcohols), and sometimes even optical brightening (using fluorescence whiteners) to enhance or stabilize visual appearance [2]. Another value-related factor is pearl’s species. There are currently four major types of cultured pearl products in the market based on different mollusk species: akoya, South Sea, Tahitian, and freshwater cultured pearls. Cultured pearls grown in different mollusk species exhibit different intrinsic levels of quality factors, yet sometimes it is difficult to distinguish pearls from Pinctada fucata (akoya) and Pinctada maxima (South Sea) mollusks with similar diameter and nacre layer thickness. Both treatment information and species information are required to be disclosed to the public during the trade [3]. Unfortunately, in many cases, both factors are unable to be identified by visual examination.

Pearl identification is usually performed by experienced gemologists, combining visual observation with the aid of microscope and gem lamps in gemological laboratories [4]. In addition, many advanced testing techniques are necessary for treatment or species identification, including energy dispersive X-ray fluorescence (EDXRF), ultraviolet to visible (UV-Vis) reflectance spectroscopy, and Raman/photoluminescence (PL) spectroscopy [5–10]. EDXRF is a trace elemental analysis technique that can distinguish saltwater pearls from freshwater pearls, as well as detect certain artificial dyes such as silver nitrate [11]. UV-Vis reflectance spectroscopy can be used to separate naturally colored Tahitian and golden South Sea cultured pearls from treated ones [5,6,12]. Raman and PL spectroscopies are useful techniques in detecting natural pigments from naturally colored freshwater pearls, and they also can differentiate dark colored pearls from Pinctada and Pteria mollusk species [8–10,13]. These spectroscopic techniques are required to be used in tandem to positively identify pearl’s color treatment since each technique is only sensitive to specific type of treatment. Nacre thickness is also an important quality factor for bead cultured pearls, and it can be used to identify certain pearl types. Since pearl’s nacre thickness detection requires at least 5 mm axial detection range from the surface, it is usually measured by techniques that have deep axial detection range, such as real-time microradiography or optical coherence tomography (OCT) rather than spectroscopy measurements, which usually do not penetrate inside pearls [14–16]. These advanced techniques require expensive instruments, routine instrument maintenance, and well-trained technician for daily operation. In addition, the data acquisition is time consuming; for example, each sample requires several minutes under each instrument, including time for sample preparation, alignment, and measurement. As a result, the identification process is not necessarily cost- and time-effective. There is strong demand for simple and inexpensive instruments designed for general pearl treatment screening and identification, which is sensitive to most commonly encountered treatments in the trade.

Conventional pearl identification does not really involve fluorescence spectroscopy methods, only visual fluorescence observation under UV excitation by gemologists. It is prone to inconsistency and difficult to differentiate the different intensities and hues in some cases. This study presents results for pearl’s treatment detection and species classification based on the spectroscopic analysis of pearl fluorescence excited by a UV light-emitting diode (LED). Pearl’s nacre fluorescence has been targeted and measured by a photoluminescence excitation (PLE) spectroscopy to determine the effective excitation wavelength and the sensing range for optimizing the experimental prototype. To enhance the flexibility, the prototype uses a reflection fiber probe to enable loose pearls, mounted pearl jewelry, and pearl strands testing in a wide size and shape range, and can be operated at ambient lighting and room temperature conditions. This fluorescence spectroscopy technique is a quick and consistent analytical tool that can be used to further help with the identification of pearls. Unlike other methods, this fluorescence spectroscopy targets pearl’s nacre fluorescence, which is sensitive to all major pearl treatments, including dyeing, UV irradiation, and bleaching. In addition, the high usability and sensitivity of this technique also shortens the required preparation and measurement time from minutes to seconds, which can significantly reduce the time requirement for treatment detection. Finally, the system does not require expensive light source or detector as traditional spectroscopy techniques, dramatically reducing the instrument cost. Combined with spectral analysis protocol, the primary purpose of the device is to rapidly detect potential treatments in pearls, including color treatments in colored pearls and bleaching in white pearls. Additional analysis has been designed to extend the application to classify white pearls between South Sea and akoya species. The instrument’s rapid testing speed, economic cost, and high usability and sensitivity to common pearl treatments meet the needs as a pearl screening device.

2. BACKGROUND

Conventionally, mercury lamp’s 254 nm and 365 nm emission lines have been used as the short-wave and long-wave UV light sources to study the fluorescence visually in a dark environment as a routine process during pearl identification [5,9,10]. Unfortunately, visual observation of pearl fluorescence is limited by reproducibility, sensitivity, accuracy, and time requirement. Since human vision is superior in differentiating minor differences between samples but poor in quantifying the color and brightness, it is very difficult to accurately record the fluorescence features. In addition, visual observation only detects visible wavelengths, and variations in the UV or near infrared region cannot be identified by human eyes. Another limitation comes from the light source. For example, the leaking of visible components in gemological UV lamps usually interferes with the observation, especially when the fluorescence response is weak [17]. After observation, to localize and label tiny samples in dark environment could also be particularly challenging. Gemologists have to repeat the process to confidently localize all the suspicious samples in a strand. These limitations restrict the applications of using the fluorescence features in pearl identification. Therefore, fluorescence features is rarely being considered as conclusive evidences in pearl identification.

Compared to visual observation, using fluorescence spectroscopy significantly improves the reliability of using fluorescence in identification. Fluorescence spectroscopy is designed to quantify the spectral distribution, which can distinguish subtle differences in emission. Additionally, it is usually more sensitive than absorption/reflectance measurements [18], commonly used in modern day’s pearl identification. Affordable fluorescence spectroscopy has been widely accepted as a basic tool in gemstone identification, such as natural diamond screening, treated diamond identification, and determination of various mineral types of colored gemstones [19,20]. These applications demonstrate the advantages of rapid detection and nondestructive analysis in valuable gemstones and jewelry pieces. Fluorescence spectroscopy has been applied in several pearl studies. Laser-induced fluorescence spectroscopy was applied along or combined with OCT to distinguish some mother oysters [21–24]. Unfortunately, the spectral variation in fluorescence between popular pearls species is usually minor and highly related to the pearl’s body color [23,24]. For example, porphyrin’s fluorescence features at 620, 650, and 680 nm can be detected both in Pteria and Tahitian (Pinctada margaritifera) pearls, especially in dark colors [10,25,26]. As a result, pearl species identification based on characteristic fluorescence spectra is not widely accepted by the pearl industry. In addition, based on the best of our knowledge, these applications are unable to detect potential pearl treatments.

It is known that color treatments not only modify the visual appearances of the pearl but also impact the pearl’s nacre layer [27]. Nacre layer also contributes to several spectroscopy features, such as an absorption band approximately at 280 nm that can be detected in most of the naturally colored pearls [28]. Although color treatment may diminish the 280 nm absorption band in some samples, the existence of the 280 nm absorption is not reliable enough to be used for treatment detection [10,28]. The origin of this absorption feature has been reported previously as the amino acid tryptophan (Trp or W) in conchiolin, which is also the major UV fluorophore in pearl’s nacre [29]. Although the other two aromatic amino acids phenylalanine (Phe or F) and tyrosine (Tyr or Y) also generate a fluorescence feature, under mid-wave UV excitation, tryptophan becomes the dominant fluorophore due to its high emission coefficient and absorption range [18,30]. In addition, the DNA sequencing technique also proved that tryptophan is detectable in pearl’s nacre layer [31,32].

Tryptophan fluorescence is rarely observable through visual inspection or detected by a color camera since the emission band is centered approximately at 340 nm and extends to ${\sim}450\,\,\rm nm$. Only weak blue fluorescence is within the visible region. As a result, fluorescence spectroscopy is an ideal tool for its detection. A useful feature of tryptophan fluorescence is its high sensitivity to chemical and physical modifications to its surrounding environments [18]. The intensity of tryptophan fluorescence has been applied in evaluation of hair damage under UV exposure, gamma ray irradiation, bleaching, and color dyeing [33,34]. Since the pearl industry adapts similar chemical and physical methods to modify pearl’s visual appearance, tryptophan fluorescence intensity from pearl’s nacre can be used as an indicator of pearl treatments [35,36]. Another extensible application is to separate different types of pearls or mollusk species. One known difference between akoya and South Sea pearls is that the former type is routinely processed after the harvest while the later one is not [37]. Since treatment and processes result in nacre fluorescence deduction, the intensity of nacre fluorescence could be used to separate white colored akoya and South Sea pearls.

Since the defect producing absorption in pearl’s nacre also generates fluorescence, using a PLE spectroscopy can effectively determine the shape of the absorption spectrum and characterize the fluorescence. A PLE spectroscopy was used to measure an untreated pearl sample to evaluate pearl’s fluorescence from nacre [36]. A monochromatic tunable light source scanned through the absorption spectrum of the feature of interest, then a spectrometer was set to pass the fluorescence band. In most of the cases, the absorption feature is not interacting with other defects. Only light that is absorbed will produce fluorescence, so the excitation spectrum in a PLE system will effectively represent the absorption spectrum of the nacre. Due to opaque natural of the pearl sample, a reflection fiber probe was used to excite the sample and to collect the emission spectrum. In order to isolated the excitation and the emission spectra under reflection mode, a 300 nm shortpass and a 300 nm longpass filter were applied to the spectroscopy. These two filters limit the upper range of the excitation and the lower range of the emission spectra as 300 nm. Figure 1 shows pearl’s nacre has an absorption band at 250 to 300 nm with local maximum approximate at 290 nm in the excitation spectrum. The emission spectrum excited at 290 nm shows one broad fluorescence band from 310 to 450 nm with local maximum at 340 nm. No fine spectral feature was detected in this experiment. Based on the result, an excitation wavelength between 270 to 290 nm with a sensing range starting from 310 nm is recommended to analyze pearl’s nacre fluorescence.

 

Fig. 1. Excitation and emission spectra of pearl’s nacre fluorescence. The emission spectrum is centered at 340 nm with approximately 70 nm bandwidth. The excitation spectrum shows that 290 nm has the best efficiency to generate fluorescence signal.

Download Full Size | PDF

3. EXPERIMENTAL DETAILS

The ideal excitation wavelength for fluorescence spectroscopy should effectively excite the targeted spectral features from the tested sample. To optimize the detection efficiency of pearl’s nacre fluorescence, we have selected a 275 nm UV LED as the light source, coinciding with tryptophan’s excitation spectrum, as presented in Fig. 1. Other excitation wavelengths between 265 and 300 nm would also be usable; however, other LED sources in this wavelength range have lower irradiance than 275 nm [38]. Based on the best of our knowledge, the 275 nm LED is the most effective off the shelf LED light source to excite pearl’s nacre fluorescence. Mid-wave UV laser sources are too expensive and usually belong to class 3 or 4 lasers. To meet the radiation safety requirement, additional personal protection equipment and operating restrictions are required to operate class 3 or 4 laser systems, increasing the cost and operational limitations on sample handling. The use of a UV LED for this prototype provides a suitable inexpensive excitation source, as long as the user is wearing appropriate safety eye goggles, or to reduce the LED output power to meet the standard as a class 1 LED product [39].

A systematic layout of the fluorescence spectrometer is presented in Fig. 2. A 275 nm LED generating mid-wave UV is used to excite fluorescence from pearls. A reflection fiber probe guides the UV light to the sample and collects the fluorescence signal for analysis. Finally, a spectrometer records and sends the spectra to a computer for analysis. Optical longpass and shortpass filters are used to isolate pearl’s fluorescence signal from excitation light, as presented in Fig. 2(a). Since the optical filter requires small incident angles to achieve optimal cutoff efficiency, two lenses are used to collimate the LED light before the shortpass filter and to refocus the filtered excitation light to the reflection fiber probe. A pre-aligned in-line multimode fiber optic filter mount is used to collimate the fluorescence signal before the longpass filter and to focus the filtered light into a fiber spectrometer. Finally, the recorded florescence spectrum is analyzed by a custom algorithm and generating a screening or identification result.

 

Fig. 2. Optical design of the fluorescence spectrometer. (a) Schematic layout of the fluorescence spectrometer. (b) Reflection fiber probe design: the light source fiber has six $400\,\,\unicode{x00B5}{\rm m}$ core fibers, while a single $400\,\,\unicode{x00B5}{\rm m}$ fiber is used for detection. These seven fibers merge at the probe tip, with the detector fiber located at the center. (c) Fiber probe head layout: an LED switch is attached on the fiber probe head, and a Teflon spacer is mounted on the fiber probe tip to maintain the optimal sampling distance approximately 4 mm.

Download Full Size | PDF

The reflection fiber probe aligns all the light source and signal fiber cores for spectral measurement. As presented in Fig. 2(b), six fibers relay the UV light to excite the sample, and a single signal fiber transmits the fluorescence signal to the detector. In order to improve the usability, as presented in Fig. 2(c), a relay power switch was attached to the fiber probe to enable the user to easily control the on/off status of the UV LED and to position the probe to pearl sample simultaneously. A Teflon spacer is mounted on the top of the reflection fiber probe to maintain the optimal working distance, approximately 4 mm, between the fiber probe head and the sample. The spacer has a 5 mm diameter aperture, suitable for measurement from pearls with diameter greater than 5 mm. Since the samples’ fluorescence intensities can vary significantly by both the probe incidence angle and the distance between the probe and sample, using the Teflon spacer can minimize the intensity variation and improve the consistency of the measurements. In addition, the spacer reduces ambient light collected by the fiber probe, which interferes with the fluorescence measurement, and prevents any scattered UV light leaking to the environment. Finally, this Teflon spacer also prevents pearl samples from being scratched by the metal fiber probe head. We note that adding a focusing lens will improve the efficiency of excitation and light collection. It can also lower the required LED power and shorten the spectroscopy integration time. However, adding a focusing lens behind the optical fiber probe will also create additional irradiation safety concern. Extra irradiation safety standards need to be applied to the system with focusing optics. As a result, our current prototype does not include a focusing lens.

The experimental prototype shown in Fig. 3 was built to test the concept of using reflection fiber probe fluorescence spectroscopy with a UV LED light source for the pearl treatment detection and species separation. The light source for this setup was a 275 nm LED (M275L4, Thorlabs) powered by an LED driver (LEDD1B, Thorlabs). Two 35 mm focal length fused silica lenses (LA4052-UV, Thorlabs) were installed before and after the 300 nm shortpass filter (FF01-300/SP-25, Semrock) to collimate, filter, and then couple the selected UV source into the reflection fiber probe (FCR-7UVIR400-1, Avantes). The fibers were connected through a SMA connector (SM05SMA, Thorlabs). All of the light source optics from the UV LED to the SMA connector were mounted and aligned by a 30 mm cage system (Thorlabs). An in-line multimode fiber optic filter mount (FOFMS-UV) was used to collimate the fluorescence signal into a 300 nm longpass filter (FF01-300/LP-25, Semrock), and to couple the filtered fluorescence signal into a 0.22NA, $400\,\,\unicode{x00B5}{\rm m}$ multimode signal fiber (M113L01, Thorlabs). A spectrometer was used for signal detection to distinguish targeted fluorescence from other optical background signals that may interfere the result. The spectrometer (AvaSpec-Mini 2048CL, Avantes) had an effective detection wavelength range from 300 to 1100 nm with 4.8 nm spectral resolution. The entire setup was mounted on a 6 in. by 12 in. (1 in. = 2.54 cm) aluminum optical breadboard (MB612F, Thorlabs) for portable applications.

 

Fig. 3. Experimental prototype of the fluorescence spectrometer.

Download Full Size | PDF

In order to reduce unnecessary UV irradiation, a limited UV LED output power and spectrometer’s integration time were defined. The output power of the LED was set to $300\,\,\unicode{x00B5}\rm W$, which was measured by an optical power sensor (S120VC, Thorlabs) at the probe head. The spectrometer collected fluorescence spectra using a fixed, 200 ms integration time. Corresponding dark and room light spectra were subtracted from the measurements. A spectral analysis protocol was applied to screen the samples based on the intensity of characteristic fluorescence band, and the relative intensity ratio between targeted and reference fluorescence bands. The relative intensity ratio analysis is designed to compensate the impact caused by fluorescence intensity fluctuation. The details of spectral analysis criteria will be discussed in Section 4.

 

Fig. 4. Selected naturally colored pearls and treated color pearls. (a) Natural color golden. (b) Natural color pinkish orange. (c) Natural color strong purplish pink. (d) Natural color strong pinkish purple. (e) Natural color brown. (f) Natural color dark gray. (g) Natural color pistachio. (h) Treated color golden. (i) Treated color orangy yellow. (j) Treated color pinkish purple. (k) Treated color pinkish purple. (l) Treated color dark bluish gray. (m) Treated color dark greenish gray. (n) Treated color chocolate. (o) Treated color pistachio.

Download Full Size | PDF

To validate the performance of the prototype spectrometer in treatment detection and pearl species separation, we measured 478 pearl samples including 126 naturally colored pearls, and 137 treated color pearls; 27 white colored pearls samples, which were treated by different standard pearl processes; 23 white South Sea cultured pearls; and 165 white akoya cultured pearls. The tested samples were from GIA’s research sets, all previously conclusively identified by experienced gemologists using various gemological and advanced techniques, including visual observation, observation of fluorescence under long-wave UV excitation, real-time microradiography, UV-Vis reflectance spectroscopy, or Raman spectroscopy.

We note that this fluorescence spectroscopy is targeting the commonly encountered pearl treatments, which mainly impact pearl’s surface. The actual penetration depth of specific UV light is still under investigation. Due to the weak penetration capability, we believe our fluorescence spectroscopy does not have the capability to detect deeper layer information of pearl.

4. RESULTS AND DISCUSSION

A. Treated Colored Pearl Screening

To prove the concept of color treatment screening, a set of 15 pearl samples was selected, including 7 naturally colored and 8 color-treated pearls with similar hues, as presented in Fig. 4. These samples included both freshwater and saltwater cultured pearls of colors frequently encountered in the market. All pearl samples were identified by experienced gemologists to confirm their origin of color, using both gemological and advanced instrumental methods. Those color-treated pearls were modified by dyes except the chocolate and pistachio colors, which were reportedly not dyed but were treated with a series of undisclosed chemical treatments. Figures 5 and 6 show the fluorescence spectra of selected naturally colored and color-treated pearls, respectively. The horizontal axis indicates the wavelength, and the vertical axis is the detector counts normalized to spectrometer’s integration time per millisecond. Naturally colored pearls showed strong fluorescence bands between 310 and 370 nm, with the maximum signal level above 45 counts per millisecond. Visible fluorescence sidebands at 410 and 480 nm were also detectable in some samples. Conversely, most of the color-treated pearls show similar UV fluorescence bands, but the intensity was at least 80% lower than their naturally colored counterparts, and all below 20 counts per millisecond. Treated chocolate and pistachio colored pearls were almost inert to excitation. In addition, the 410 and 480 nm fluorescence bands were more obvious in color-treated pearls, compared to naturally colored pearls.

 

Fig. 5. Fluorescence spectra of naturally colored pearls, including hues of golden, pink-orange, purplish pink, pinkish purple, brown, dark gray, and pistachio [Figs. 4(a)–4(g)]. The signal level was normalized by dividing the detector counts by per millisecond integration time.

Download Full Size | PDF

 

Fig. 6. Fluorescence spectra of color-treated pearls, including hues of golden, orangy yellow, pinkish purple, bluish gray, brown, dark greenish gray, chocolate, and pistachio [Figs. 4(h) to (o)]. The signal level was normalized by dividing the detector counts by per millisecond integration time.

Download Full Size | PDF

We noticed the spectral distribution between 330 and 350 nm in our fluorescence spectrometer is slightly different than in our PLE setup. This is because the quantum efficiency of the spectrometer (AvaSpec-Mini 2048CL) in our prototype has a saddle point around 340 nm. The spectrum of a deuterium-tungsten halogen lamp (DH-2000-BAL, Ocean Optics) was used to confirm the difference in quantum efficiency around 340 nm, which was approximately 10% lower than 330 nm and 27% lower than 350 nm.

Based on the characteristic spectral features outlined above, a color treatment screening protocol was developed. It used the intensity of UV fluorescence and the ratio between UV and visible fluorescence to detect potentially treated colored pearls and separate them from naturally colored pearls. The protocol used a pre-defined intensity threshold of 45 counts per millisecond to evaluate the fluorescence band ${\sim}340\,\,\rm nm$. Close to boundary signals, the protocol evaluated the signal ratio between UV fluorescence and a reference fluorescence band ${\sim}480\,\,\rm nm$. Pearls emitted fluorescence between 30 to 45 counts per millisecond, but ratios between UV and visible fluorescence above 20 were also considered as untreated pearls. Pearls emitting fluorescence above either one of these two conditions were positively identified by the protocol as naturally color pearls while those that did not were referred for further testing. Table 1 summarizes the results of such treatment screening for total of 263 pearls. The protocol detected strong UV fluorescence in 97.6% of naturally colored pearls. On the other hand, none of 137 color-treated pearls had strong UV fluorescence detected, which means 100.0% of color-treated pearls were referred as potentially treated pearls. The test samples include with variety of colors and species, including akoya, South Sea, fresh water, and Tahitian pearls. Most of the popular treatment methods were also included in our treated color samples, such as color dyeing, silver nitrate darkening, and undisclosed chemical processing. However, further studies with a higher volume of samples and diversity of pearl species, colors, and color treatments are preferred to evaluate the usability of this method in general pearl color treatment detection.

Table 1. Data Collection for Treated Color Pearls Screening

View Table | View all tables in this article

We noted that this system does not identify the type of color treatment. Any sample referred by the treated color pearl screening is required to be further analyzed by other methods to disclose the origin of color treatment. However, screening out potentially treated samples can significantly reduce the time requirement for conclusively treatment disclosure, especially for high quantity evaluation.

B. Detection of Routine Pearl Enhancement

Figure 7 shows fluorescence spectra for white colored pearls under different types of routine processing, including 10 tumbled only, 8 maeshori only, 4 bleached only, and 5 bleached plus tumbled samples. The samples were collected from and processed by two akoya pearl farms located in Ehime and Nagasaki in Japan. Compared with untreated pearls, tumbled samples did not have any clear fluorescence reduction since tumbling does not include any chemical or optical processes, only physically polishing the pearl’s surface. Maeshori samples showed slightly lower fluorescence than the tumbled samples since alcohol and mild heat treatment were involved. Finally, bleached and bleached plus tumbled samples showed significantly reduced fluorescence, indicating that the strong oxidation effect caused by the bleaching solution contributed to severe fluorescence reduction. In addition, visible fluorescence detected from bleached samples was stronger compare with other type of treatments. Based on the literature, the visible fluorescence could come from the metabolic and oxidation products of tryptophan [34]. Based on the results, fluorescence spectrum can be used to detect several types of routine treatments, especially bleaching. However, additional studies with pearls under different amounts of bleaching times and conditions are necessary to achieve a comprehensive evaluation of pearl’s bleaching detection.

 

Fig. 7. Fluorescence spectra for pearls after different type of pearl processes. The processes including tumbling, maeshori, bleaching, and bleaching plus tumbling.

Download Full Size | PDF

C. Akoya and South Sea Pearls Separation

Typical fluorescence spectra of white colored South Sea and akoya pearls are shown in Fig. 8. Both types of pearl showed similar spectral distribution but with distinguishable differences in their fluorescence intensity. Due to the routine processes (bleaching and maeshori) applied to akoya pearls, their nacre fluorescence was 3 to 4 times weaker than South Sea pearls. The pearl species separation protocol considered the UV fluorescence intensity to classify white colored South Sea and akoya pearls. The threshold fluorescence level was selected at 120 counts per millisecond under our experimental setup. All samples above the florescence threshold were classified as South Sea, and the others were grouped as akoya.

 

Fig. 8. Fluorescence spectra of white colored South Sea pearls and akoya pearls.

Download Full Size | PDF

To prove the concept of pearl species separation, 188 white pearl samples, including 23 South Sea pearls and 165 akoya pearls, were investigated under our prototype and summarized in Table 2. The species of all the tested samples were disclosed by manufacturers and confirmed by gemologists based on their appearance, including diameter and nacre thickness. The protocol detected strong UV fluorescence in 100.0% of South Sea pearls. On the other hand, all 165 akoya pearls were below the pre-defined fluorescence threshold and were concluded correctly.

Table 2. Data Collection for Pearl Species Separation

View Table | View all tables in this article

 

Fig. 9. Evaluation of the impact of continuous UV irradiation to pearl’s nacre fluorescence. (a) Fluorescence spectra after 0 (initial), 1, 5, 10, 20, and 30 min of UV irradiation. The sample was irradiated by 275 nm UV light with $300\,\,\unicode{x00B5}\rm W$ of average power for 30 min. (b) Normalized fluorescence signal decay between 330 and 350 nm under 5 min of UV irradiation. The red line shows the average decay, and the gray shaded area represents the standard deviation.

Download Full Size | PDF

We note that, besides nacre fluorescence, other features such as blue color fluorescence from optical brightening agents (OBAs) can also be used to separate akoya and South Sea pearls since pearls of the latter type are generally not being optically brightened after harvest [40]. However, 275 nm LED is less efficient to excite OBA fluorescence compared to 385 nm LED [36].

D. Impact of UV Irradiation to Pearl’s Fluorescence Intensity

As we discussed in the previous sections, nacre fluorescence may be modified by UV irradiation. Since a mid-wave UV LED was used in our experimental prototype, any potential impacts caused by the UV irradiation should also be evaluated. Figure 9 shows the result of sequential pearl’s nacre fluorescence measurements under continuous mid-wave UV irradiation. A natural American freshwater pearl was selected as the sample since it was collected directly from a freshwater mussel to confirm no treatment was applied to the sample. The experimental prototype presented in Fig. 3 was used to irradiate the sample while collecting the fluorescence. A series of fluorescence spectra was collected in 30 min with a time interval of 10 s. Figure 9(a) shows the fluorescence spectra after different durations of UV irradiation. Similar to color treatments and bleaching, UV irradiation also reduced the nacre fluorescence intensity. In addition, it also slightly increased the visible fluorescence intensity between 410 and 480 nm, potentially due to the effect of photo-oxidation of tryptophan [34]. The fluorescence decay was collected at 16 different surface areas from the same sample to estimate the decay speed. Figure 9(b) presents the average and the standard deviation of fluorescence intensity decay between 330 to 350 nm during 5 min of UV irradiation with a sampling interval of 5 s. A double exponential decay function,

5. CONCLUSION

We have reported a rapid pearl treatment screening and species separation device based on fluorescence spectroscopy and experimentally demonstrated its functionality and accuracy. To optimize efficiency of excitation and detection of corresponding fluorescence features, a PLE spectroscopy was used to measure pearl’s nacre fluorescence to locate proper excitation wavelength and detection range. The system used a 275 nm LED as the light source, a pair of longpass and shortpass filters to isolate emission from the interference of excitation, and a miniature spectrometer to collect the fluorescence response from pearls. A revised reflection fiber probe with a power switch and spacer was added to effectively transmit both the excitation and sample emitted signals between the device and test samples. This probe expanded the flexibility in measuring all varieties of pearl samples, including loose pearls, pearl strands, and mounted pearls in jewelry. The power switch and the spacer on the probe tip improved the suitability of operating fluorescence measurement under standard office lighting conditions, reducing the hazard of unintentional UV exposure and interference from ambient light. The system operated with the spectral analysis algorithm and demonstrated the capability of identifying naturally colored pearls and referred treated colored pearls. A 97.6% naturally colored pearl positive identification rate and 0% false positive rate was achieved for color pearls frequently encountered in the market. Another application for white pearl enhancement detection was also tested for common white pearl treatments such as maeshori and bleaching. The result was extended to be used to separate white color pearls between South Sea and akoya species. Notably, none of the pearls were incorrectly identified as other species. We note that this device can be used not only on cultured pearls but also on natural pearls’ color treatment detection since similar color treatment techniques have been applied to valuable natural pearls. The technical advances in pearl culturing and post-harvest treatments are continuously evolving, which increaes the complexity of identification. Future studies will target detection of new treatment techniques and fluorescence features related to pearl identification.