Introduction

From their discovery to their conservation, the value of the cultural artefacts makes them subject to theft, trafficking, and damage. Such illegal activities prevent them from being studied to acquire shared knowledge of our past. On a global scale, it is estimated that illicit trafficking in cultural heritage is one of the four most significant forms of illegal trafficking, generating €7.8 billion in revenue in 20211 and experiencing steady growth over the years2. However, such crimes particularly suffer from a lack of data on their true extent and insufficient policing3,4.

Thus archaeological artefacts hold a particular role in illegal trafficking, which has various sources: theft from institutions or private collections, looting of excavation sites. Once stolen, the trafficking of cultural artefacts is heavily linked to other criminal activities, such as money laundering and other illegal trades5,6. It is especially concerning that terrorist groups exploit artefacts to fund their criminal activities. By pillaging historical sites when gaining territories, they generate revenues that are used to further destabilize regions by fuelling their armed guerrilla7.

The difficult distinction between legal and illegal markets8,9 is due to the lack of effective and standardized controls capable of establishing the provenance of artefacts. All these factors contribute to the large magnitude of these trades. Actions previously taken to reduce the damage to our cultural heritage have mostly concentrated on finding and returning stolen artefacts10. Great efforts have been made in these areas, which have sometimes been successful in returning objects. However, this solution is incomplete as cultural trafficking continues to grow, and even successful returns do not guarantee that the artefacts did not sustain damage11.

Such a reactive stance has so far been unable to combat the rise in heritage crime. A different approach that can reduce pillaging is needed. Several means can be used to achieve that goal. One solution is the marking of artefacts with a security tag to enact a deterrent effect12. The aim of this method is to establish secure means of identifying objects to prevent their trafficking or counterfeiting.

In the field of security markings, three categories of reading levels are generally used: overt, covert, and forensic13.

  • An overt marking is visible and legible to the naked eye and requires no tools to read. These markings include variable optical markings or holograms, visible inks, and other patterns and prints.

  • A covert marking is also visible but requires a technical tool to be read by an operator. This category includes all coded markings that can be read by algorithms or other machines. It also includes markings with properties not visible to the naked eye, such as microscopic, topographical, magnetic, and fluorescent markings.

  • Forensic marking refers to markings that can only be read by state-of-the-art equipment, usually found in laboratories. Examples of this type of marking are nanoscaled structures, products with precise elementary or molecular markers, or biochemical signatures.

Obviously, there are differences in security beyond these reading levels, particularly the difficulty of reproducing or, in other words, counterfeiting these markings. It is important to note that cultural artefacts hold particular importance compared with other goods. They are part of humanity’s heritage, and their handling is framed by rigorous codes of deontology. Any marking must also be reversible and must not alter the object.

The use of Forensic Traceable Liquid has already shown great promise in reducing illegal trafficking. The solution provided by DeterTech under the name “Smartwater” has been implemented on archaeological artefacts in several locations, such as Syria and Iraq14. It is identifiable by a bright UV dye, and each iteration contains precise identification information contained in the chemical composition of a polymer synthesized from a selection of 24 rare elements and microscopic additives. Over the years, the deterrent effects of such products on theft crimes have been repeatedly shown15,16, and this also applies to heritage crimes17. Even though the use of “Smartwater” has been deemed a cost-effective tool, it does require forensic-level characterization to identify a solution. Meanwhile, the fluorescent species in the solution only seem to serve as a permanent, simple marker.

The use of nanomaterials exhibiting unique fluorescent properties and requiring advanced chemical synthesis to achieve similar behaviour could improve the Forensic Traceable Liquid solution. If the fluorescent properties of the tag can contain some level of security, systematic forensic analysis could be avoided. Without the need to wait for analysis lab results, the fluorescent tag could be used to alert authorities that an object has been registered and that its legal provenance must be checked before being traded or transported across borders without requiring advanced characterization.

Fluorescent nanomaterials, such as carbon dots or quantum dots, emit light by fluorescence when exposed to UV and visible light. Using nanomaterials allows us to take advantage of their specific properties, such as fluorescent emission variation according to size, smaller emission peak widths, and improved photostability. In practice, the fluorescent light emitted by such materials is intense and carries precise specific properties, impossible to reproduce with standard organic dyes18. Their use in security tags can constitute a first level of security compared to the dye solely used as a marker in traceable liquid solutions. Different categories of nanomaterials have been used in this work.

Firstly, carbon dots are carbon-based 0D nanomaterials discovered in 2004 by Xu et al.19. They are composed of either graphitic carbon, amorphous carbon, or a mix of both. These nanoparticles have attracted attention as they can exhibit strong and diverse fluorescence emission while excluding any heavy metals from their composition20. While the width, quantum yield, and stability of their fluorescent emission are not on par with other quantum dots, their high biocompatibility constitutes their main advantage21.

Cadmium Selenide (CdSe) quantum dots are among the most well-known fluorescent nanomaterials. Chalcogenides nanoparticles were discovered in 198022. They are composed of elements from groups II-VI or III-V of the periodic table. By reducing such compounds to the nanometric scale, smaller than their exciton Bohr radius, their electrons undergo a quantum confinement effect and exhibit distinct optical and electronic properties. However, they show poor dispersibility in water and other highly polar solvents, and the elements used are highly toxic to humans23.

Perovskites are crystalline materials known for their remarkable physical properties and photovoltaic capabilities. Metal halide perovskites were synthesized as quantum dots in 201424, and shortly afterward, fully inorganic perovskite quantum dots (PRKs)25 were obtained by hot-injection synthesis. The nanocrystals formed using this technique reach sizes below 20 nm, thus they are also subject to quantum confinement. Nonetheless, their bonding characteristics also cause their greatest weakness: the structural and colloidal stability of metal halide perovskite QDs can be degraded by polar or ionic compounds. Light and electrical fields can also cause halide migration inside the crystal26.

Therefore, complementing other Traceable Liquid solutions, such as the “Smartwater” solution, we have conducted a study on implementing those fluorescent nanoparticles in ready-to-use marking solutions for archaeologists. Several types of fluorescent quantum dots have been synthesized and incorporated into stable conservation resins, which were applied to antique ceramic materials. The stability of the tags was evaluated by climatic and UV aging. XRF (X-Ray Fluorescence) spectroscopy measurements are a common tool to analyse the composition of archaeological materials, thus, the tagged ceramics were also analysed using two different methods to determine if the presence of the tag could undermine those analyses.

Methods

Materials

1,6-dihydroxynaphthalene, citric acid, urea, cadmium oxide, zinc acetate, oleic acid, octadecene, selenium, sulfur, sodium myristate, methanol, hexane, trioctylphosphine, trioctylphosphine oxide, oleylamine, di-2-pyridylketone, sulfuric acid, were purchased from Sigma-Aldrich (St. Louis, MO, USA).

Caesium carbonate, lead chloride, lead bromide, and lead iodide were purchased from Thermo Fisher Scientific Inc. (Waltham, MA, USA).

All reactants were of synthesis grade and used without further preparation. The water used was ultra-purified using a Veolia system.

Security tags

Since a traceable liquid solution needs to be widespread in order to induce a sufficient deterrent effect, the proposed solution must be viable and adoptable by all the parties involved: archaeologists, customs officers, and forensic scientists in particular. The literature, as well as our consultations with police teams, show that they are familiar with covert means of identification such as UV lamps and can easily adopt them27,28. Forensic teams also already have labs capable of reading forensic-level markings such as “Smartwater” solutions and identifying nanomaterials. Archaeological teams, on the other hand, carry out excavation campaigns that are limited in time and require expensive resources. It must be possible to apply a marking solution without requiring extra time and at the lowest possible cost; otherwise, archaeologists may resist adopting the proposed solution.

Consultations and field tests were performed at the Kition archaeological site in Larnaca, Cyprus. It was observed at the time that markings on the artefacts are made using commercial transparent varnish, usually based on nitrocellulose resin. The varnish penetrates the porosity of the material to prevent any ink from permanently marking the artefact, after which identification numbers are written on the varnish29,30. Although only archaeological artefacts of significance or those identifying a stratigraphic unit are marked, this still represents a large number of objects for teams of archaeologists to tag.

To increase the chances of our solution being adopted by archaeologists, our goal was to adapt as much as possible to their marking techniques. The chosen solution was to integrate our fluorescent particles directly into the varnish applied to the artefacts. This avoids any additional handling for the archaeologists and produces tags indistinguishable from the usual markings.

Resins

While adapting to the usual techniques used by the project’s archaeologists, we also sought to obtain the most stable markings possible. Natural resins such as the nitrocellulose previously used can undergo significant degradation when exposed to UV light31, and the radicals resulting from this degradation can also induce degradation in the fluorophores present in the varnish of a marking32. Archaeological artefacts are particularly prone to this as they are often stored in areas subject to high temperatures and/or high humidity. However, some resins used for the conservation of cultural artefacts show better resistance to UV exposure.

The main cause of degradation in the resins used by archaeologists is the absorption of UV light. This problem has also been encountered in the conservation of fragile cultural artefacts, mainly paintings. Their conservation over several centuries is made possible by layers of protective varnish. Historically, mastic or dammar resins have been used almost exclusively. These resins are aesthetically pleasing due to their light molecular weight, which produces a thin layer of polymer with minimal alteration of the refractive index above the painting. Nevertheless, yellowing or crumbling of the varnish has been observed over several decades due to the formation of free radicals in the resins induced by UV exposure.

As a result, more resistant resins were sought by custodians and manufacturers. Paraloid B72 is a resin composed of methyl and ethyl methacrylate, which is used extensively in archaeology today, thanks in particular to its superior stability33. While a humid environment can cause whitening of the resin during its application, it has not been shown to impact the resin stability in the long term34. However, its stability is also diminished when exposed to UV light35. Given that Paraloid B72 and other similar resins have a high molecular weight, their aesthetic appearance over paintings was not deemed suitable for conservation by some. Other resins were subsequently developed.

Cyclohexanone resins (MS2, AW2, Laropal) formed by the condensation of (methyl)cyclohexanone and formaldehyde are a low molecular weight alternative, but the presence of ketone groups results in problematic UV absorption. A reduction reaction of these ketone groups results in resins that are more chemically stable but fragile, with the layers formed subject to rapid crumbling, especially at high temperatures. To overcome this fragility, manufacturers have developed resins with a similar molecular weight, derived from the condensation of urea and aldehyde, namely Laropal A81 resin. It is more stable and has a sufficiently high glass transition temperature (54.6 °C)36 to prevent the resin from crumbling.

Finally, another type of resin has been identified for its stability. Hydrocarbon resins composed of low molecular weight saturated carbon chains avoid the UV absorption associated with the carbon-carbon double bond (C=C)37. Regalrez 1094 resin, derived from the hydrogenation of vinyl-toluene and alpha-methyl-styrene oligomers, therefore benefits from enhanced stability. It is soluble in apolar solvents but, on the other hand, its glass transition temperature is only 34 °C. Although it remains hard at room temperature, a phase change at high temperatures can lead to the incorporation of airborne particles and alter the varnish. These temperatures are commonplace in many regions of great archaeological interest (Central America, the Middle East, etc.).

Paraloid B72, Regalrez 1094, and Laropal A81 were chosen to fabricate security tags to be tested under harsh temperature, humidity, and lighting conditions. They were chosen for their conservation qualities and low potential for UV degradation. By selecting resins soluble in apolar (Regalrez), mildly polar (Laropal, Paraloid) solvents, and up to polar alcohols (Laropal), we will also be able to incorporate different kinds of nanomaterials to fabricate the tags. Integration into those resins should also result in a gain in stability. The literature shows that similar nanocomposites showed improved stability compared to resins and particles alone28,38,39,40.

The security markings were obtained by applying the nanoparticle resin solutions with a standard nylon brush on Gallo-Roman unslipped test ceramics. Three types of nanoparticles were used to produce the markings: carbon dots, chalcogenides quantum dots and perovskites quantum dots.

Carbon dots

Blue carbon dots (b-CDs) were synthesized using a well-known synthesis procedure with citric acid and urea as organic precursors in a solvothermal synthesis in dimethylformamide (DMF)41. A total of 1 g of citric acid and 2 g of urea were weighed and then dissolved in 30 mL of di-methylformamide (DMF). The solution was stirred for 10 minutes and subjected to an ultrasonic bath when necessary to completely dissolve the precursors. It was then placed in a teflon liner, subsequently placed in an autoclave reactor from Asynt capable of withstanding 200 bar, which was hermetically sealed and heated to 200 °C. After 24h of reaction time during which the processors underwent condensation, oligomerization, and carbonization, the carbon dots were recovered after cooling the reactor. The crude reaction product was then purified by dialysis using a membrane from Spectrum Labs with at least 1 kDa of Molecular Weight Cut Off (MWCO) for several days. A purple solution of b-CDs was finally obtained.

Cyan carbon dots (c-CDs) were synthesized from di-pyridylketone and sulfuric acid using a similar process as described in a previous communication42. Di-pyridylketone was heated in the presence of H2SO4 and ultrapure water for 24h at 200 °C in the same autoclave reactor. During the heating process, the reactant was dehydrated and carbonized by the sulfuric acid, the carbon functional groups reacted to form a graphitic sp2 core first, then the remaining functional groups linked to the surface of the particles. A dark red solution was obtained, which was purified using dialysis membranes with a MWCO of 500 Da from Spectrum Labs to remove any molecular precursors or impurities until the dialysate is no longer fluorescent. A slightly brown solution of c-CDs was then obtained.

To synthesize orange carbon dots (o-CDs), a total of 3 mmol of 1,6-dihydroxynaphthalene was weighed and dissolved through stirring and ultrasound in 30 mL of ethanol. Once transferred to a Teflon liner placed inside the autoclave reactor, the solution was heated to 200 °C for 7 hours. The formed carbon dots are then collected and purified using column chromatography. Verification through thin layer chromatography established a 10:90 mixture of methanol and dichloromethane as the optimal eluent. The first fraction was characterized by its yellow colour and blue fluorescence typical of PAHs. The fraction of interest was identified by its orange colour and fluorescence and is henceforth referred to as o-CDs.

Chalcogenides quantum dots

Blue quantum dots (CdZn/ZnS) and green quantum dots (CdSe/ZnS) were synthesized using a single-step method to produce gradient compositions in the nanocrystal, following the method of Bae et al.43. The blue quantum dots (b-QDs) were synthesized according to the following protocol. 1 mmol cadmium oxide (CdO), 10 mmol zinc acetate, 7 mL oleic acid was dissolved in 15 mL octadecene (ODE) in a 250 mL three-neck flask fitted with a thermocouple temperature probe and a rubber septum. The solution was placed in an inert atmosphere using argon gas and then heated to 150 °C to complex the metal ions with oleate ions. During this time, 2 mmol of sulphur was dissolved in 3 mL of ODE, and 8 mmol of sulphur was dissolved in 3 mL of trioctylphosphine (TOP). The temperature of the initial solution was then raised to 300 °C and the sulphur-ODE solution was added to the flask by syringe. After 8 minutes the temperature was raised to 310 °C and the sulphur-TOP solution was added to the medium. After a further 40 minutes the flask was cooled to room temperature.

The solution was first centrifuged at a speed equivalent to 10,000 rcf for 5 min to remove any solid residues of the reagents. The supernatant was then diluted with 30 mL of ethanol and centrifuged at 8000 rcf for 5 min to remove any reagents still present in solution. The supernatant was then discarded, and the QDs deposited are solubilized in 2 mL of toluene. This step was repeated once more, then twice, replacing the ethanol with acetone. Finally, the blue QDs are solubilized in toluene. The solution obtained should be transparent and slightly yellowish.

The green quantum dots (g-QDs) were also synthesized using the same method by hot injection. The CdSe/ZnS gradient structure was obtained by introducing 0.2 mmol CdO, 4 mmol zinc acetate, and 4 mL oleic acid into 15 mL ODE. Once the oleate compounds were formed, the temperature was raised to 305 °C and a solution containing 0.1 mmol selenium (Se) and 3.5 mmol sulphur dissolved in 2 mL TOP was added to the flask. The reaction was continued for only 10 min before carrying out the same washings as described above. The solution of g-QDs that was then obtained was completely yellow.

The orange and red quantum dots were synthesized using a protocol communicated by the team of Thomas Pons from LPEM44. CdSe cores were first synthesized according to a method by Cao et al.45. The size of this core determines the emission wavelength of the quantum dots. To form the red-emitting particles, a larger core was fabricated during this first step. A multilayer shell of CdS/CdZnS/ZnS was then formed using a successive ionic layer adsorption and reaction method46 (SILAR). During the synthesis of orange (o-QDs) and red quantum dots (r-QDs), reagents were prepared prior to the synthesis. A myristate-complexed cadmium compound was prepared from CdO and sodium myristate (NaMyr). 3.13g of NaMyr was dissolved in 250 mL of methanol for 1h, and 1.23 g of cadmium nitrate tetrahydrate was dissolved in 40 mL of methanol. The cadmium nitrate solution was then slowly added to the sodium myristate solution, and cadmium myristate then forms as a white precipitate. Cd(Myr)2 was then recovered by Buchner filtration and washed 3 times with methanol to be used as our cadmium precursor.

Solutions of cadmium oleate at 0.5mol/L, zinc oleate at 0.5mol/L, sulphur at 0.1mol/L, and Se at 0.1 mol/L in ODE were also prepared for layer growth. 6.42g of cadmium oxide was weighed and introduced with 100 mL of oleic acid into a 250 mL tri-col flask fitted with a temperature probe and rubber septum. The solution was heated to 160 °C until colourless. It was then cooled to 100 °C, at which temperature the reaction medium was degassed to remove all traces of water. The solution was then collected warm as it solidifies at room temperature. To prepare the zinc oleate, the same procedure was used, replacing the cadmium oxide with 4.07g of zinc oxide and heating to 280 °C instead of 160 °C. To prepare the sulphur-ODE solution, 0.32 g of sulphur was added to 100 mL of previously degassed ODE in a 250 mL flask. The solution was heated to 120 °C until the sulphur was completely dissolved, then the solution was cooled and stored at room temperature. Finally, 1.185 g of Se was added into 10mL of ODE and sonicated until totally dissolved. Then 140ml of previously degassed ODE was heated to 205 °C into a 250 mL three-neck flask fitted with a temperature probe and rubber septum. As the temperature rose, the Se was added into the hot ODE dropwise. After the addition the solution in the flask was kept at 205 °C for 30 minutes and was then cooled down. These precursor solutions can be stored for several months after preparation and used to prepare several batches of QDs.

Once all precursors are prepared, the QDs synthesis was performed as follows. 12mg of Se was added to 8 mL of ODE. This solution was then added to the flask after being dispersed using ultrasound. Given the low dispersibility of Se, the pillbox of the solution was then rinsed with 8 mL of ODE, which were added to the flask to ensure that all of the weighed Se was present in the flask. 174mg of cadmium myristate was also added and the flask was placed under vacuum to degas the solvent. It was then placed in an argon atmosphere and the temperature was raised to 220 °C. Once the temperature had been reached, the cadmium myristate decomposed, and the reaction forming cadmium selenide took place for 10 minutes before the flask was rapidly cooled to room temperature to stop the reaction. As the flask cooled, excess oleic acid was added to stabilise the QDs. The CdSe core solution was then topped up with 20 mL oleic acid and 80 mL ethanol before being centrifuged at 6000 rcf. The core deposit was then redispersed in hexane before being used for SILAR shell growth. To synthesize red QD cores, once the cores have formed at 220 °C, the flask was heated to 300 °C instead of being cooled. 0.6 mL of cadmium oleate solution and 3 mL of Se precursor solution are mixed, then the obtained solution was injected slowly as 300 °C was reached inside the reaction medium. Then it was left to react for 10 minutes at 300 °C before being washed and centrifuged as described above.

Next, to form the CdS/CdZnS/ZnS shells by the Successive Ion Layer Adsorption and Reaction (SILAR) method, solutions of cadmium oleate, zinc oleate and sulphur-ODE were used. This method aims to produce layers incrementally by introducing only the necessary amount of material into the medium. By calculating the radius of the CdSe core, we can calculate the amount of cadmium, zinc, or sulfur needed to produce one atomic layer at each step of the synthesis. The solution of QD cores dispersed in hexane was introduced with 10 mL of ODE and 2 mL of oleylamine into a 100 mL three-neck flask using the same set-up as above. The hexane was then evaporated from the medium by placing the flask under vacuum while maintaining its temperature at 30 °C. During this time, the oleate solutions were liquefied so that they can be used later. Once the hexane had evaporated, the flask is placed in an argon atmosphere and the equivalent of one layer of cadmium oleate and sulphur was added, then the temperature was raised to 230 °C. Subsequently, the equivalent of two layers of cadmium oleate and then sulfur were added progressively, waiting 10 minutes between each addition to the medium. Cadmium oleate and zinc oleate solutions were mixed at a 1:1 ratio to form the CdZnS layer. After the addition of cadmium and sulphur, the mixed Cd:Zn oleate solution was added in turn with the sulphur solution to produce the equivalent of 2 layers. The temperature was raised to 250 °C and the zinc oleate solution was added with the sulphur solution to produce the two final layers. Finally, the QD solution was cooled to room temperature, 10 mL of oleic acid was added to the solution, and 60 mL of ethanol was also added. The solution was centrifuged at 6000 rcf and the QDs were redispersed in 10 mL of hexane. In the case of red QDs, the injection quantities are adapted to the larger core size.

When synthesized, the blue and green quantum dots are surrounded by oleic acid and TOP ligands, whereas the orange and red quantum dots are only linked to oleic acid ligands. To make them all soluble in solvents of higher polarity, a ligand exchange reaction between oleic acid and TOP was performed. To perform the ligand exchange, we proceeded as follows. The previously synthesized QD solution was placed in a three-neck flask equipped with a thermocouple temperature sensor and a rubber septum. 10 g of triotylphosphine oxide and 5 mL of TOP were introduced in the solution, it was then degassed at 70 °C to remove hexane and any water or oxygen. After any bubbling had stopped for more than 10 minutes, the solution was put under an argon atmosphere and left to react for 1h. Once 1h had passed, the solution was cooled and introduced into 50 mL centrifugation tubes and centrifuged at 6000 rcf for 10 minutes. The QDs then precipitated and could be dispersed in the chosen tagging solution.

Synthesis of Perovskite quantum dots

To fabricate our security tags, we sought to produce PRKs that could withstand solvents of higher polarity to be able to produce security tags not only from Regalrez but also from Laropal and Paraloid resins. To achieve that goal, we synthesized caesium lead halide PRKs following the work of Wu et al.47. Firstly, a caesium oleate precursor was prepared. 0.49 mmol (159 mg) of Cs2CO3 was placed in a 50 mL three-neck flask fitted with a temperature probe and a rubber septum. 5 mL of ODE and 0.5 mL of oleic acid were added to the flask, then the medium was degassed by heating to 120 °C in order to degas the solution. Once the water had been removed from the medium, the flask was placed in an argon atmosphere, and the temperature was raised to 150 °C until the Cs2CO3 was completely dissolved. This constitutes our caesium precursor solution that can be used to produce several different perovskite nanocrystals.

On another set-up, 5 mL of ODE, 1.17 mmol of oleic acid, 1.23 mmol of oleylamine, and 0.85 mmol of TOPO were added to a 100 mL three flask in a configuration similar to the first. The solution was dried under vacuum for 30 min and 0.188 mmol of a lead halide was added to the flask. After the lead halide compound was completely dissolved, the temperature was raised to 160 °C. Then 0.4 mL of caesium precursor solution was quickly injected into the medium. After 5 seconds of reaction, it was suddenly stopped by cooling the flask using an ice bath. Finally, the obtained solutions were purified by adding 3 mL cyclohexane to the solution and centrifuging it at 7000 rcf. The supernatant was discarded while the nanocrystals were dispersed in 5 mL cyclohexane. They were centrifuged one last time at 4000 rcf to remove any aggregates or reactant leftovers. The particles obtained showed enhanced stability in polar solvents up to ethanol and thus retain their fluorescence properties. By using chlorine, bromine, and iodine, we could produce PRKs showing purple, green, and red fluorescence. Furthermore, by mixing two subsequent halides, we can fabricate nine different PRK solutions exhibiting fluorescence across the visible spectrum: purple (p-PRK), dark blue (db-PRK), blue (b-PRK), cyan (c-PRK), green (g-PRK), light green (lg-PRK), yellow (y-PRK), and red (r-PRK) perovskite quantum dots. The table below (Table 1) shows the reactants used for each perovskite and the wavelength of their emission peak.

Table 1 Summary of the composition and emission wavelengths of the perovskites nanocrystals synthesized and used for security tags
Full size table

Accelerated aging

Once we explored the possibilities of integrating fluorescent nanomaterials into conservation resins and successfully applied them to archaeological artefacts, we sought to simulate how the environment in an archaeological storage area could affect the integrity of the coating and the fluorescence of the emitters contained inside.

All the security tags mentioned above were placed in an ESPEC SH-262 temperature and humidity chamber set to produce a 50 °C and 100% relative humidity environment for 5 hours. The parameters were chosen to expose the tags to the harshest temperature and humidity conditions they might encounter during their discovery, storage, and transport. The fluorescence was recorded by exciting the tags using a 5 W LED lamp emitting at 365 nm and collecting the emitted light using a parabolic mirror mounted on an optical fibre leading to a QE65000 spectrometer from Ocean Optics Inc. A 405 nm long-pass filter was used to eliminate scattered excitation light.

The resistance of a long exposition to daylight was also tested using a Suntest XXL + from Metek. All security tags were exposed to a dose of 1.044W/m2 (340 nm) from 3 1700W xenon arc lamp in front of sunlight filters during 5 h. During the test the temperature inside the chamber rose to 42 °C.

XRF analysis

To determine the influence of marking solutions on possible future geochemical investigations of archaeological ceramics, XRF analyses were performed on all the tags containing different nanomaterials. Several different methods of XRF measurements exist and are used in archaeology, in this communication we used wavelength-dispersive X-ray fluorescence spectrometry (WDS-XRF) and portable energy-dispersive X-ray fluorescence analysis (ED-XRF) to compare two widely different methods. WDS-XRF was carried out in the Archaeology and Archaeometry Laboratory in Lyon (France), using a Bruker S8 Tiger spectrometer with a Rh excitation source according to the procedure developed by the laboratory.

Samples were cut out with a diamond-coated saw. The external surfaces, liable to chemical alteration during burial, and all traces of glazes or slips were removed. Heating the samples at 950 °C (necessary to remove water, volatiles, and organics) was followed by cooling and grinding with a tungsten-carbide ball mill. A total of 800 mg of powdered sample was then mixed with 3200 mg of flux (lithium metaborate and tetraborate), and the resultant mix was heated and fused to produce a glass disc. The measurement can then be performed on this glass disc of homogeneous composition, which corresponds to a mean chemical composition representative of the initial material. Indeed, this procedure provided the bulk chemical composition of the ceramic (matrix plus inclusions) and consequently of the material used for its manufacture48,49. For each sample, 24 components were determined based on calibration curves established with 40 international geostandards (Centre de Recherches Pétrographiques et Géochimiques (CRPG), United States Geological Survey (USGS), National Institute of Standards and Technology (NIST), British Chemical Standards, etc.). Ten samples were analysed using this method, one control sample and nine samples labelled with different solutions.

ED-XRF was carried out in the field on the Kition archaeological site in Larnaca. The Vanta M spectrometer from Olympus was utilized for this50. This Instrument has been used at the Ceramics Research Centre at Goethe University specifically for analysing ceramics since 2018. The spectrometer is finely calibrated for measurements on ceramics using around 140 standard reference samples, meaning that the results are compatible with other calibrated (laboratory) methods. Nine main elements (Si, Ti, Al, Fe, Mn, Mg, Ca, K, P) and 17 trace elements (S, V, Cr, Co, Ni, Cu, Zn, As, Se, Rb, Sr, Y, Zr, Nb, Cd, Ba, Pb) can be precisely determined with the configuration of the spectrometer and used for the analysis. The fluorescence energy of the light elements (from hydrogen to sodium), which cannot be quantified, is totalled by the spectrometer and indicated under the value LE (Light Elements).

Results

Security tags

Regarding carbon dots security tags, the obtained o-CDs solutions were dried and redispersed in 20 wt.% Paraloid ethyl acetate solutions (Fig. 1a). A solution of 20 % Laropal in the same solvent quenched most of the fluorescence of the carbon dots and was unusable. Then, since the b-CDs and c-CDs are only soluble in polar solvents, the only conservation resin used to produce tagging solutions was Laropal A81. The solutions were dried and then redispersed in 20 wt.% Laropal ethanol solutions (Fig. 1c). Polyvinyl alcohol (PVA), a polymer often used in water solutions, was also used as an alternative. To prepare tagging solutions, the particles were redispersed in 10 wt.% PVA aqueous solutions (Fig. 1b).

Fig. 1: Carbon dots security tags.
figure 1

Security tags on Gallo-Roman ceramics made with a o-CDs in 20 wt.% Paraloid B72 ethyl acetate solutions, b b-CDs in 10 wt.% PVA aqueous solution and c c-CDs in 20 wt.% Laropal ethyl acetate solutions.

Full size image

After their initial purification step, blue (b-QDs), green (g-QDs), orange (o-QDs), and red (r-QDs) chalcogenides quantum dots solutions in hexane are obtained. They can be used to make security tags by dissolving 20 wt.% of Regalrez resin into them.

Then, following the ligand exchange to replace oleic acid with TOP, the quantum dot solutions can be dried and redispersed in 20 wt.% Paraloid or Laropal ethyl acetate solutions to produce other security tags. All four chalcogenides QDs Laropal tags are displayed in Fig. 2. The o-QDs Laropal, Paraloid and Regalrez are also displayed in Fig. 3. It can be observed that under daylight, the Paraloid tag is more distinguishable compared to the Laropal and Regalrez tags, which form thinner coatings because of their low molecular weight.

Fig. 2: Quantum dots security tags.
figure 2

Security tags on Gallo-Roman ceramics made with a blue, b green, c orange and d red chalcogenides quantum dots in 20 wt.% Laropal ethyl acetate solutions.

Full size image
Fig. 3: O-QDs security tags using different conservation resins.
figure 3

o-QDs security tags made with a Laropal, b Paraloid, and c Regalrez resins.

Full size image

After purification, the solutions are dried and redispersed in 20 wt.% Paraloid and Laropal ethyl acetate solutions to produce the tagging solutions (Fig. 4). Unfortunately, we observed that yellow and red PRKs lose their fluorescent properties when redispersed in the Laropal solution. Nanocrystals containing lead iodide are the least stable, and Laropal ester groups may lead to oxidation.

Fig. 4: Perovskites quantum dots tags.
figure 4

Security tags on Gallo-Roman ceramics made with a purple, b dark blue, c blue, d cyan, e green, f light green, g yellow, and h red perovskite quantum dots in 20 wt.% Paraloid B72 ethyl acetate solutions.

Full size image

A total of 40 different combinations of resins and nanomaterials resulted in tagging solutions stable enough for application to Gallo-Roman Ceramics. The tags produced were bright enough for identification with a low-power UV lamp and their measurements by a standard spectrometer. All the combinations of resin and nanoparticles are summarized in the following table (Table 2). As specified above, the b-CDs and c-CDs were not soluble in apolar solvents and could not be integrated into the Paraloid and Regalrez resins. However, another water-soluble resin, PVA, was used to remedy that issue. On the contrary, o-CDs could not be used with Laropal, which quenched its fluorescence and could not be dissolved in an apolar solvent to be used with Regalrez. Since they are also not water soluble, PVA could not be used to produce a supplementary tag, and only Paraloid tags could be produced. While carbon dots can show water solubility, it is harder to achieve with PRK and QDs. Processes to alter their surface were not readily available, which reduced our options. To produce more tags, ligands should be used to modify the solubility of the carbon dots. Other, less-used conservation resins, like Paraloid B67, which is water soluble, could also be used to remedy that situation.

Table 2 Summary of successful combinations producing stable tagging solutions
Full size table

Then, regarding perovskites y-PRK and r-PRK could not be used in Laropal tags as they did not retain stability long enough to be tested. It appears that iodine-containing PRK showed even less colloidal stability than their counterparts. Different synthesis, or other stabilising ligands, may be used to produce such tags.

Accelerated climatic aging

After the accelerated aging process, almost all security tags retained fluorescence easily visible to the naked eye. The environmental conditions did affect all the tags, as the fluorescent intensity recorded at a constant position on the tags decreased each time. By analysing the light spectra emitted by the tags, we determined the initial brightness of each one as well as the ratio between the initial brightness and the brightness measured after aging as a means to compare their performance.

Given the low number of samples and resin combinations possible with the carbon dots synthesized, no clear trends can be observed from the results. It can still be concluded that they represent a viable option as they all retain sufficient fluorescence to be identified. While they showed the best resistance in fluorescence intensity, the integrity of the PVA tags was greatly affected by the climatic aging, making it significantly difficult to read the identification written (Fig. 5). As the polymer is soluble in water, it is clear that it is sensitive to such conditions of temperature and humidity. This alternative may only be suitable in mild climatic conditions. Laropal tags showed medium resistance by retaining about 50% of their initial signal, however, the initial intensity of the b-CDs Laropal tag was considerably weak, which indicates it is the hardest tag to identify. In comparison, the Paraloid o-CDs tag showed weak resistance (35%) but had sufficient initial intensity to be identified clearly before and after the climatic aging process. (Fig. 5)

Fig. 5: Climatic aging of carbon dots tags.
figure 5

Relative fluorescent intensities of carbon dots security tags after aging. The bars show the percentage of fluorescence signal remaining after the aging process on the left axis. Markers show the initial fluorescence intensity measured by the spectrometer on the right axis in a logarithmic scale. Pictures showing PVA coatings after aging are placed above the corresponding bars.

Full size image

In the chalcogenides security tags category, only the o-CDs Regalrez tag showed unsatisfactory results. It had the least initial fluorescence recorded and only kept 16.0% of its initial fluorescence signal. This represents a significantly less bright and stable tag than its counterparts. The blue and green tags have an overall better resistance across all resins, as their Laropal and Paraloid tags retain more than 80% of the initially measured emission, while Regalrez tags also kept 64.3% and 76.8% of intensity. The Regalrez resin does show the least resistance with QD tags. It records the lowest value in all but the r-QDs tag, for which our error margin places it similar to the Laropal tag (Fig. 6). All tags showed strong initial fluorescent signal above 1000 counts indicating that all tags would remain easily identifiable even if the concentration of QDs was lowered. The lowest values were also recorded on the Regalrez tags. Even as all QDs tag show strong resistance, it appears that Regalrez 1094 is the least adapted resin for application on archaeology tags.

Fig. 6: Climatic aging of quantum dots tags.
figure 6

Relative fluorescent intensities of chalcogenides security tags after aging. The bars show the percentage of fluorescence signal remaining after the aging process on the left axis. Markers show the initial fluorescence intensity measured by the spectrometer on the right axis in a logarithmic scale.

Full size image

As stated in the experimental section, some combinations of PRK and resins were unavailable because of their low stability in solvents of higher polarity. The results for the y-PRK and r-PRK Laropal tags are thus absent. It was observed that the majority of the PRK tags exhibit strong initial fluorescence, making them easily observable even at lower concentrations of nanoparticles. The p-PRK tags and r-PRK tags emit wavelengths at the edge of the visual spectrum and at the lowest initial intensity measured, which could make identification impossible if a lower power UV or a lower concentration is used.

When comparing their resistance to the aging process, some of the Laropal PRK tags still performed adequately, such as the c-PRK and g-PRK tags retaining more than 60% of their fluorescence. However, the other showed poor emission stability, p-PRK recorded 48%, and all the other Laropal tags less than 30% of their initial emission. The results are similar for the Regalrez tags, which exhibited medium resistance for b-PRK, c-PRK, and r-PRK (60.7%, 60.0%, and 40.4%). However, the other Regalrez tags resisted poorly and all recorded less than 20% of their initial signal. Considering their low initial brightness and resistance, the viability of the p-PRK, db-PRK, and r-PRK tags is questionable. Nevertheless, the Paraloid tags still showed the strongest resistance, for more than half of the tags, more than 60% of their previously measured fluorescence remained (Fig. 7).

Fig. 7: Climatic aging of perovskites quantum dots tags.
figure 7

Relative fluorescent intensities of perovskite security tags after aging. The bars show the percentage of fluorescence signal remaining after the aging process on the left axis. Markers show the initial fluorescence intensity measured by the spectrometer on the right axis in a logarithmic scale. Pictures show the Regalrez c-PRK and lg-PRK tags under UV lighting before and after aging as an example.

Full size image

We can observe that for the majority of the nanomaterial tags, the Regalrez resin exhibited a stronger decrease than the Laropal and Paraloid resins under the treatment at 50 °C and 100 % relative humidity for 5h. Under those conditions, this can be attributed to its low glass transition temperature and the lower thickness of the Regalrez tags. Regarding the different nanomaterials used, the carbon dots proved to be bright and resistant enough to be viable when they can be incorporated in conservation resins. The PVA alternative did not show sufficient physical resistance to the most extreme temperature and humidity conditions, but could still be considered for other applications. Both the chalcogenides quantum dots and most perovskites showed more than sufficient brightness, but the p-PRK and r-PRK markings could be harder to identify in some cases. Regarding their resistance to the accelerated aging, the chalcogenides quantum dots recorded the least diminution of their fluorescence overall, while the less stable perovskite quantum dots could fall to 10–20% of their initial recorded emission intensity. We can conclude that the ionic characteristic of their bonding leads to weaker resistance and homogeneity of the tags.

Accelerated light aging

Tags that showed only minor deterioration from climatic aging were used again for the light aging experiment. Any samples that showed a significant decrease of their fluorescent properties were cleaned and a new tag was applied. However, we were unable to perform the measurements before the Laropal and Paraloid solutions of b-PRK and g-PRK lost colloidal stability. To ensure that the results presented are comparable to the climatic aging, the results of those tags are absent.

Carbon dots showed stronger resistance to light aging than climatic aging. In all cases, their initial fluorescence was bright enough for identification. Laropal, Paraloid, and PVA tags for c-CDs and o-CDs retained more than 80% of their fluorescent intensity. The PVA tags were the most deteriorated, but also kept at least 80% of the initial fluorescence signal. However, the b-CD Laropal tag was the most sensitive to light aging and lost almost all of its initial emission (Fig. 8).

Fig. 8: Light aging of carbon dots tags.
figure 8

Relative fluorescent intensities of carbon dots security tags after light aging. The bars show the percentage of fluorescence signal remaining after the aging process on the left axis. Markers show the initial fluorescence intensity measured by the spectrometer on the right axis in a logarithmic scale.

Full size image

Chalcogenides quantum dots showed similar resistance to climatic aging. The Paraloid tags showed remarkable resistance, always retaining 95% of the fluorescence measured before the aging process. Laropal and Regalrez tags also performed well for g-QDs and r-QDs tags, with more than 80% relative fluorescent intensity, Laropal tags produced the weakest signal measured for b-QDs and o-CDs tags but still exhibited around 60% of their initial signal (Fig. 9).

Fig. 9: Light aging of quantum dots tags.
figure 9

Relative fluorescent intensities of chalcogenides security tags after light aging. The bars show the percentage of fluorescence signal remaining after the aging process on the left axis. Markers show the initial fluorescence intensity measured by the spectrometer on the right axis in a logarithmic scale.

Full size image

PRK tags were once again the least stable nanomaterials measured, although they showed different results than under the climatic aging. Only the lg-PRK tag showed strong brightness and conserved more than 50% of its initial fluorescence for all 3 resins. A majority of PRK tags decreased under the 50% threshold, and a significant portion fell below 30%. The p-PRK tags lost their signal almost entirely or, in the case of the Regalrez tag, kept only 49.5% of its initially weak fluorescent emission. db-PRK and r-PRK showed similar results, and taking into account the results of their resistance to climatic aging, they should not be deemed fit for use in the field with the current formulation. Additionally, b-PRK and g-PRK Regalrez tags also showed a strong decrease in their signal and could be unfit for use as well. When comparing all the PRK resins under light aging, no particular trend arose for PRK tags although c-PRK and lg-PRK tags showed remarkably similar behaviour for light and climatic aging (Fig. 10).

Fig. 10: Light aging of perovskites quantum dots tags.
figure 10

Relative fluorescent intensities of perovskite security tags after light aging. The bars show the percentage of fluorescence signal remaining after the aging process on the left axis. Markers show the initial fluorescence intensity measured by the spectrometer on the right axis in a logarithmic scale.

Full size image

This behaviour can be attributed to the environmental temperature in the aging process, which constantly exceeded its low glass transition point (34 °C). The Regalrez 1126 resin could be proposed as an alternative since it exhibits a higher glass transition of 65 °C.

XRF analyses

As explained above, the implementation of such a marking technique depends in part on how it is perceived by the teams of archaeologists for whom it is intended. The application of the marking solution to the artefacts was therefore been designed to adapt to the existing working methods of the HiSoMA laboratory’s team of archaeologists. However, their work also involves more in-depth characterization of certain archaeological artefacts, in particular dating using carbon-14 isotope measurements and XRF to identify materials using the concentration of elements such as Ca, Fe, Si, Zn, and more. It was therefore decided to measure the composition of ceramics marked by all of the glazes shown above in order to determine whether the presence of an inconspicuous marking could interfere with the characterization of a ceramic in order to establish its origin.

The measurements were carried out both by WD-XRF in a lab and using a portable ED-XRF device on site. WD-XRF enables analyses with much smaller quantities of material but requires highly destructive sample preparation. The portable ED-XRF, on the other hand, is non-destructive and can be used on directly on excavation sites. Since the portable device allows in-situ measurement of the tags, it is possible to measure carbon dots and perovskite nanocrystal tags, which do not withstand the temperature required for the sample preparation in WD-XRF.

In the case of portable XRF, all nanomaterials presented in this communication were analysed. Regalrez 1094 varnishes loaded with nanoparticles were applied to the surface of a modern ceramic. The measurements of the ceramic alone and of each marking can be compared to show their influence on portable XRF analysis.

The results clearly show the influence of the markings on the measurements. All the chalcogenides tags show higher levels of cadmium, zinc, and sulfur than the ceramics alone (Fig. 11). They also show the presence of selenium, which is usually not detected at all in ceramics alone, except for the b-QDs, which have a CdZn core rather than a CdSe core. Perovskite tags show increased amounts of lead compared with ceramics, but caesium and halogens are not detected. Most of the organic elements making up carbon dots are included within the ‘light elements’ (LE) category of the analysis and were detected in higher quantities, but no distinction between carbon, oxygen, and nitrogen can be made. On the other hand, sulfur was clearly identified and was found in greater quantities in samples of o-CDs and c-CDs, which are composed of organic sulfur groups due to the presence of sulfuric acid during the synthesis.

Fig. 11: ED-XRF analysis of a ceramic carrying different tags.
figure 11

Concentration in parts per million (ppm) of a Pb, b light elements, c S, d Se, e Zn, and f Cd detected by portable ED-XRF on a ceramic marked by different security tags.

Full size image

If such tags are to be used on material from archaeological excavations, the nanomaterial quantities used could be greatly reduced, which may be easily feasible for chalcogenides but less so for PRK and carbon dots. Some areas may also be designated as free of tagging products to perform undisturbed analysis.

Nevertheless, different results were obtained when analysing similar samples using WD-XRF. Using the preparation process described in the experimental methods, only chalcogenides quantum dots could be analysed without being destroyed. However, other nanomaterials susceptible to be used in various marking technologies were deposited onto ceramics to evaluate their potential pollution as well (iron oxide, gold, silver, aluminium, tin and glass nanoparticles).

All following chemical species were measured: CaO, Fe2O3, TiO2, K2O, SiO2, Al2O3, MgO, MnO, Na2O, P2O5, Zr, Sr, Rb, Zn, Cr, Ni, La, Ba, V, Ce, Y, Th, Pb, Cu. Among all the targeted compounds, only zinc showed strong variations from the ceramic covered with red CdSe/CdS/CdZnS/ZnS quantum dots (Fig. 12). While the red chalcogenides quantum dots do have a slightly thicker ZnS layer than their orange counterpart, this significant difference can be better explained by a the presence of a much larger concentration of nanomaterials on this particular sample.

Fig. 12: WD-XRF analysis of a ceramic carrying different tags.
figure 12

Concentration in parts per million (ppm) of zinc detected by WD-XRF on a Gallo-Roman ceramic marked by different security tags.

Full size image

A future study monitoring the amount of fluorescent tag needed for detection and the detection threshold of WD-XRF for chalcogenides quantum dots will be needed to guarantee unpolluted analysis of the tagged artefacts.

Discussion

In this communication, the integration of fluorescent nanomaterials into identification tags used by archaeologists was introduced as a complementary solution to forensic traceable liquid technology to reduce pillages and illicit trafficking of cultural artefacts. Such a solution could be easily spread and introduce some security elements at a covert level rather than requiring lab analysis for every object. Conservation resins developed for their long-term stability were chosen to produce security tags incorporated with fluorescent nanoparticles. Three types of fluorescent solutions were fabricated: carbon dots, chalcogenides quantum dots, and perovskite quantum dots. Every resin and particle combination was used on Gallo-Roman ceramics to make security tags similar to those used by archaeologists.

Once the tags were fabricated, their resistance to climatic and lighting conditions was evaluated through accelerated aging processes. Perovskite quantum dots showed some instability in the harsh conditions that may be reached in storage after excavation. Some were shown to lose the necessary fluorescence intensity to be identified. On the other hand, perovskites Paraloid coatings had good resistance compared to the other resins for all nanomaterials. Carbon dots were harder to integrate but are much less toxic than their quantum dot counterparts, which shows great promise if sufficient progress can be made regarding their optical properties. As mentioned, regarding the choice of the resin to be used the Paraloid tags held the most resistance over all the measurements. The thicker coating they yielded seems to protect the fluorescent nanomaterials from their environment. While the Laropal and Regalrez resins were synthesized as more stable counterparts for long-term conservation, the fluorescent properties of the nanomaterials they contained were further altered. This is especially true for Regalrez 1094.

The influence of these tags, which can sometimes be indistinguishable without UV lighting, on XRF analysis was also evaluated. Portable ED-XRF showed that chalcogenides tags contain enough zinc and sulfur to disturb the identification of the ceramic. The lead content of the perovskite and sulfur content of some carbon dots could also interfere with the analysis. However, WD-XRF showed significantly less interference and could be used to verify portable XRF results when used on tagging products. We hope that this work will spark interest in the use of nanomaterials to tag cultural artefacts and lead the way for more specific studies or new solutions. Only by reaching a widespread solution or solutions can our cultural heritage be efficiently protected.