1. Introduction

Tremendous numbers of colors are generated spontaneously in nature, which enliven our surroundings. Colors, either produced from structures or colorizing agents, can be described from a physical point of view as light-matter interaction. For example, the color of glass has been known to depend on small proportions of embedded metals (now known as metallic nanoparticles (NPs)). The stained windows in churches, as well as the Lycurgus Cup, are historical evidences of advanced expertise for the manufacture of colored glass [1–4]. Beyond decorative purposes, colors are also carriers of information, which not only contribute to intraspecific communication, but also secure optical document.

To date, a variety of color rendering substrates have been fabricated as a consequence of complicated interactions between light and the structures (structure colors). As anisotropic color rendering can be readily achieved via morphology control, the colored substrates has provided a viable scheme for multidimensional optical multiplexing, information encryption, and anticounterfeiting [5,6]. In contrast, colored glass based on metallic NPs always exhibits isotropic color rendering, hence developmental studies concerning information encryption based on colored glass are quite limited even up to now.

With the development of ultrashort pulsed laser, the extremely high-power density of ultrashort laser pulses allows for strong nonlinear interactions between laser and the transparent materials, which can generate color centers in various glass and transparent crystals [7–10]. A color center is a point lattice defect consisting of a vacant negative ion site and an electron bound to the site. The color centers generated in glass can selectively absorb light and make glass colored. It has been reported that color centers can be generated in silicate glass, soda-lime glasses, rare-earth ions doped germanate glass as well as in calcium fluoride (CaF₂) and lithium fluoride (LiF), etc [11–16]. It was demonstrated that diffraction gratings can be rapidly and easily produced in glass according to the color center patterns. Although the coloration mechanism of color centers is well-established, color centers, such as F, F₂, F₂⁺, F₃⁺ and H₃⁺, were proved to be sensitive to environmental factors such as heat, ultraviolet (UV) radiation, interaction with chemicals, etc. Hence, the competition between coloration induced by the laser irradiation and the subsequent transmission recovery limits the degree of coloration. Therefore, further application of color centers for information encryption is also rather limited.

On the other hand, when interacting with glass materials, ultrashort laser pulses inject energy within an ultrashort amount of time, which leads to strong thermal accumulation and thereby increases the local pressure and temperature. This process can provide sufficient heat for nucleation and even direct crystallization of nanocrystals. So far, great efforts have been devoted to the in situ precipitation of metal halide perovskite nanocrystals, silver nanoclusters or nanoparticle in glass [17–20]. And, the demonstrated potential applications such as high-capacity optical data storage, information encryption and 3D artwork are mainly based on their extraordinary photoelectronic properties. Rare attention has been paid to their colors. Furthermore, in addition to metallic NPs, some metallic chalcogenides semiconductor nanocrystals can also be used for colorizing glass, which has been widely used for cut-off filters [21,22]. Unfortunately, effective precipitation of metallic chalcogenides semiconductor nanocrystals in glass has rarely been reported, let alone considered their colors for information encryption.

Herein, a new kind of silicate glass system co-doped with CdO and ZnTe was developed, which featuring controllable generation of intrinsic color centers of brown and green via ultrashort laser pulses. On the other hand, laser-initiated instantaneous temperature rise enhanced the diffusion of Cd²⁺ and Te^2-, and promoted the nuclei formation of CdTe QDs. The controlled thermally growth of CdTe QDs leaded to a permanent color of orange in the glass, while the high-temperature accelerated the aging of color centers. Thus, a strategy of color encryption and heat decryption was achieved. This bulk color-tunable glass with robustness can be fabricated in a cost-effective, environmentally friendly and scalable way. Importantly, user-defined multiscale interior coloration and an unprecedented multidimensional color-encryption scheme with a high-security level were demonstrated based on color multiplexing.

2. Experimental section

2.1. Sample preparation

The precursor glass with a stoichiometric composition (in mol%) of 50SiO₂-35Na₂CO₃-10ZnO-5Al₂O₃-CdO-ZnTe was fabricated via a melt-quenching route, where CdO and ZnTe acted as doping reagents. For comparison, single-doped glass with CdO or ZnTe was also fabricated, which has been specifically mentioned in the corresponding section. Analytical grade powders were mixed thoroughly in corundum crucibles and melted at 1350 °C for 30 min in ambient atmosphere. The melted glass was then removed from the furnace, cast on a preheated (350 °C) graphite plate and splash-cooled by pressing with a stainless-steel stamp. The solidified glass samples were subsequently heated to 400 °C at a muffle furnace and kept for 4 h to release the strain, thereby yielding transparent glass. All glasses were optically polished for further studies.

2.2. Laser treatment

For the interior coloring of CdTe-doped silicate glasses, a compact high repetition rate femtosecond laser system (Pharos-SP-10W) with the wavelength of 1030 nm was used, where repetition rate was tunable from 1 kHz-200 kHz and frequency doubling crystal was used to produce frequency-doubled laser (515 nm). The precursor glass was placed on a three-dimensional translational stage (Prior Stage II), and linearly polarized laser beam was guided into a microscope and focused into the sample at a depth of 600 um beneath the sample surface by a 10× objective lens (NA = 0.3). If not specifically mentioned, all the samples subjected to the optical measurements were irradiated at a scanning speed of 10000 µm/s by implementing a parallel line raster pattern process. A CCD camera (Mshot MC50) equipped on a microscope was used for real-time monitoring.

2.3. Sample characterization

An UV-VIS-NIR spectrophotometer (Jasco V570) was used to record the absorbance of the irradiated glass sample. Glass transition temperature was measured by a differential scanning calorimeter (Netzsch Sta 409) in ambient atmosphere with a heating rate of 10 K/min. And the crystal morphology and size distribution of CdTe was analyzed using a field emission transmission electron microscope (TEM) (Tecnai G2 F20 S-TWIN, FEI). Besides, defects of electrons and holes were measured and analyzed by electron paramagnetic resonance (EPR) spectrometer (Bruker EPR-300). A silicon detector (Coherent OP-2 VIS) records the incident laser power.

3. Results and discussion

3.1. Color encryption by direct laser writing

The concept of information encryption and decryption in glass matrix is illustrated in Fig. 1. A kind of silicate glass system with two dopants of CdO and ZnTe was specially designed and home-made. An objective lens was adopted to focusing the ultrashort laser pulses, and a high-resolution scanning system was chosen to facilitate the precise manipulation of DLW in the glass matrix. Upon illumination of laser pulses (184 fs at 1030 nm), the instantaneous energy goes sufficiently high enough to initiate nonlinear photoionization and finally color center forms (color of brown) (Fig. 1(a)). More importantly, when higher pulse’s energy was deposited, color centers was visually deepened accompanied by the simultaneous formation of tiny crystal nuclei of CdTe QDs. Hence, a ciphertext of “5” was deliberately designed with a combination of color centers and crystal nuclei, while the other portions of “8” with only color centers. The first heating process, accelerated the aging of color centers, and the second heating process enhanced the controllable growth of CdTe QDs. Consequently, “5” was decrypted from “8”. When laser was frequency doubled (184 fs at 515 nm), similar strategy of color encryption and heat decryption could also be applied (Fig. 1(b)). The difference was that a new color center of green was more heat resistant.

 

Fig. 1. Schematic illustration of information encryption and decryption in glass matrix enabled by DLW.

Download Full Size | PDF

3.2. Tunable color generation: brown and green

With laser wavelength fixed at 1030 nm, the optical properties of the glass samples were illustrated in Fig. 2. As can be seen in Fig. 2(a), the precursor glass appears purely transparent with faint yellow (see Fig. S1 in Supporting Information). Upon laser irradiation, the color of brown emerged, and gradually deepened when laser repetition rates increased from 1-200 kHz (pulse energy at 10 uJ). As expected, the optical transmission exhibited a decreasing trend with increasing repetition rates (Fig. 2(c)). The difference transmission spectra between the precursor and the laser-treated glass clearly demonstrated that laser-induced absorption was mainly located in the short-wave band (around 380 nm). Interestingly, with laser repetition rates increased to more than 100 kHz, there appeared another absorption band in the long-wave band (around 650 nm) (inset of Fig. 2(c)). For comparison, with higher laser pulse energy at 50 uJ (repetition rate at 50 kHz), the color of brown was greatly enhanced, and subsequent heat treatment revealed that under 200 °C, there was a rapid attenuation of brown, and at the same time, the color of green appeared (Fig. 2(b)). From the spectral analysis, one can found that with laser repetition rate fixed at 50 kHz, there was no visible absorption band around 650 nm when pulse energy was 10 uJ. Thus, laser-irradiated zone was in the color of brown. However, with pulse energy increased to 50 uJ, the absorption band around 650 nm was also observed. By prolonging the heat treatment time, the absorption band around 380 nm was greatly brought down, whereas the absorption band around 650 nm exhibited only a slight decrease (Fig. 2(d)). As a result, the color of green became dominant (Fig. 2(b)).

 

Fig. 2. The evolution of optical images with (a) repetition rates and (b) annealing time. Scale bar is 4 mm; (c) The transmission spectra and (d) the difference transmission spectra corresponding to (a) and (b), respectively. (e) Arrhenius plot and (f) experimental verification of the decay rate for the color of brown. Inset of (c): The difference transmission spectra. (Other laser parameters: 1030 nm, 184 fs).

Download Full Size | PDF

Therefore, based on the above analysis, three primary facts can be reached: (1) At relatively low pulse energy and repetition rate, only brown could be generated, and brown was mainly attributed to short-wave absorption band; (2) In addition to the short-wave absorption band, higher pulse energy or repetition rate would induce another long-wave absorption band (around 650 nm), which promotes the generation of green; (3) Green was more resistant to heat than brown, and could only be revealed when the short-wave band was greatly blenched (by annealing). The three primary facts demonstrate two scenarios behind the color of brown. One scenario was that when the short and long-wave absorption bands were simultaneously induced, green became a hidden color covered by brown. The other scenario was that when only the short absorption band was induced, no green was covered by brown. The two scenarios jointly constitute a strategy of color encryption. In such color encryption, once the brown was presented, one could not distinguish whether there existed a hidden color of green by naked eye. Hence, the ciphertext could be encrypted in the color of green, and heat annealing became the decryption key.

Based on this scheme, the thermal stability of both the brown and green became the crucial factors. Herein, accelerated aging measurements at different temperatures were adopted to calculate the lifetime of a certain color, which was estimated by the Arrhenius law [23]:

As Fig. 2(e) revealed, the lifetime of brown was calculated to be approximately 79 days at room temperature (293 K), where decay rate was evaluated by detecting the relative absorption intensity versus the annealing time at variant temperature (inset of Fig. 2(e)). On a similar note, the decay of brown at room temperature was experimentally verified to be around 68 days, which was in consistent with that estimated by Arrhenius law (Fig. 2(f)).

From above discussion, one can see that the color of brown could be easily blenched by heat, which was in favor for the decryption of green. As for green, a systematic investigation was conducted (Fig. 3(a)). It was obvious that with the increase of pulse energy, there was a simultaneous increase of both the short and long-wave absorption band. For each pulse energy, the short-wave absorption band could be generated even at low repetition rate, while the long-wave absorption band required higher repetition rate. For the notable generation of the long-wave absorption band, the higher the pulse energy was applied, the lower the repletion rate was needed. For instance, at pulse energy of 30 uJ, the visible long-wave absorption band was detected at repetition rate at 10 kHz, which was much lower than 100 kHz and 200 kHz for pulse energy of 20 uJ and 10 uJ, respectively. Another interesting phenomenon should also be noticed that the short-wave absorption band didn’t show a monotonous increasing trend with repetition rates. With repetition rate increased to 200 kHz, the absorption intensity of the short-wave absorption band decreased, and this trend goes the same for each pulse energy (see Fig. S2 in Supporting information). Coincidentally, the decrease of the short-wave absorption band was accompanied by a distinct increase of the long-wave absorption band. Hence, one can estimate that there exists a competition between the short and long-wave absorption band. Once a threshold was reached, more laser energy was preferred to be absorbed for the generation of green. In addition, pulse width was also an influential factor that matters. It was found that when pulse width was adjusted to more than 5 ps, no absorption band around 650 nm could be measured, and at the same time, the short-wave absorption band also decreased a lot (Fig. 3(b)). Therefore, for each pulse energy and repetition rate, there exist a proper range of pulse width for the tunable generation of the color of brown and green.

 

Fig. 3. (a) Statistical data of the dependence of both the short and long-wave absorption band on pulse energy and repetition rates (1030 nm, 184 fs). The dependence of the difference transmission spectra on (b) pulse width (1030 nm, 100 kHz, 30 uJ) and (c) wavelength (184 fs, 1 kHz, pulse energy of 50 uJ and 2 uJ for 1030 nm and 515 nm, respectively). Scale bar is 4 mm. (d) Arrhenius plot of the decay rate for the color of green.

Download Full Size | PDF

The above investigations were carried out based on the same wavelength (1030 nm). When the infrared laser was frequency doubled, then both 1030 nm and 515 nm was applied for comparison. It was evident that femtosecond laser with 515 nm could induce the long-wave absorption band with repetition rate as low as 1 kHz and with pulse energy as low as 2 uJ. As a result, the color of green could be directly induced without the help of heat annealing. By resorting to the same method of Arrhenius law, the decay time of green was approximately 37 billion years at room temperature (293 K). Even at higher temperature of 250 °C (523 K), the color of green could last for as long as 38 days. In contrast, the color of brown could only exist for 16 min at the temperature of 250 °C (calculated on the basis of the Arrhenius plot of brown in Fig. 2(e)). Therefore, as long as the time of heat annealing was longer than 16 min at 250 °C, the decryption of green will succeed. Hence, heat annealing at 250 °C for 8 h could be recognized as an effective decryption key.

Above discussion concerning Fig. 2 and Fig. 3 reveal that compared to the long-lasting decay lifetime of green, the lifetime of brown was so short that even at room temperature, it could only last for about 70 days. So, if the color encryption between the brown and green was applied, ciphertext has to be decrypted within a short period, thus such color encryption was much in favor for instant delivery of encrypted message. Fortunately, due to the capability of direct precipitation of green, ultrashort laser pulses of 515 nm make the encryption methods more diversified. That is, previous color encryption strategy of “brown → blank (no color) or green” was upgraded to “brown + green → blank or green + green”.

To further clarify the origin of color generation, single-doped glass with CdO or ZnTe, and co-doped glass with CdO and ZnTe were fabricated for comparison. All glass samples were generally transparent (see Fig. S3 in Supporting Information). The glass doped with CdO was in no visible color before laser irradiation, and became brown after laser irradiation. Whereas the single-doped glass with ZnTe was intrinsically dark yellow, and turned into green after laser irradiation, respectively (inset of Fig. 4(a)). Therefore, the green color can be attributed to the element of tellurium [24]. It was conceivable that the co-doped glass was in a mixed color of brown and green, which has been experimentally confirmed (Fig. 1(b)). From the electron paramagnetic resonance (EPR) spectra, there was only one distinct EPR signal for single-doped glass with CdO, which was characterized by the g value of g₁ ≈ 2.0102. The two EPR signals were superimposed for single-doped glass with ZnTe and co-doped glass (Fig. 4(a)), which was characterized by the g value of g₁ ≈ 2.0102 and g₂ ≈ 1.9746, respectively (see Supplementary Note 1 in Supporting Information). Hence, spectroscopic splitting factor of g₁ and g₂ may respectively contribute to the generation of brown and green, which indicated that both brown and green were laser-induced defects (color center) in nature. In addition to the direct observation of dangling bond type defects, EPR measurements, by calibrating the signal intensity, could also allow for the comparison of the relative defect concentration. The laser-irradiated samples used in Fig. 3(c) were also subject to EPR measurements (Fig. 4(b)). The signal intensity for green (g₂) induced by femtosecond laser with 515 nm was much more intense than that induced by 1030 nm-femtosecond laser, and the opposite results was found for the signal of brown (g₁), which was in consistent with the absorption spectra in Fig. 3(c). It is known that the g value of free electron is g_e= 2.00232. In the glass, hole or electron trapping defect is generally defined with g value larger or smaller than g_e [25]. More specifically, brown was originated from the hole trapping defect, while the green from the electron trapping defect. Glass is mostly transparent to the incident light. However, ultrashort pulsed laser enables high laser intensities, which is sufficient to cause multiphoton absorption. Thus, laser energy absorbed by glass excites electrons in the valence to the conduction band. At the same time, holes are produced in the valence band. Electrons and hole are easily captured by various defects in glass. In this way, electron or hole trapping defect is successfully generated.

 

Fig. 4. The dependence of electron paramagnetic resonance spectra on (a) doping elements and (b) laser parameters (glass co-doped with CdO and ZnTe). Scale bar is 4 mm.

Download Full Size | PDF

3.3. Two-step precipitation of CdTe QDs

Differential scanning calorimeter (DSC) analysis of precursor indicated one exothermic peak, which denoted that the glass transition temperature (T_g) was at 462.78 °C (see Fig. S4 in Supporting Information). When a precursor was firstly laser irradiated (1030 nm, 1 kHz, 50 uJ), and then heat annealed at 450 °C for 8 h (the lifetime of green at 450 °C was 4 min based on the Arrhenius plot), green was totally replaced by another color of orange (inset of Fig. 5(a)). From the following Raman measurements, no significant structure was found for the precursor glass. However, a distinct peak located at 160 cm^-1 was observed in orange region, which was ascribed to the longitudinal optical phonon of CdTe QDs [26]. Transmission electron microscopy (TEM) images in Fig. 5(b) also confirmed the homogeneous precipitation of CdTe QDs with radii around 4-6 nm among the glass matrix. The lattice fringes could be observed in a higher resolution image (Fig. 5(d)), and the distance between neighboring crystal lattice fringes was ∼ 0.229 nm, and it corresponded to the (200) crystal facet of CdTe. Selected area electron diffraction (SEAD) showed that the QDs in the glass matrix with multilayer concentric diffraction rings correspond to the (220) and (311) facets of CdTe, which were in perfect accordance with the crystalline phase of CdTe (JCPDS No.65-1082) (Fig. 5(c)).

 

Fig. 5. (a) Raman spectra of the precursor and laser-heat-treatment glass (1030 nm, 1 kHz, 50 uJ; 450 °C for 4 h). Scale bar is 4 mm. (b) TEM, (d) HRTEM images and (c) the SAED pattern of the CdTe QDs in glass matrix. Inset of (a): photograph of CdTe QDs-precipitated glass.

Download Full Size | PDF

4. Application

4.1. Interior coloration

In section 3, we have discussed that three colors could be generated in a controllable manner. On the one hand, one of the three colors could be individually fabricated in the glass matrix (Fig. 6(a)). One the other hand, the three colors could also be effectively integrated in one piece of glass matrix (Fig. 6(b))(see Table S1 in Supporting Information). For practical and scalable fabrication of interior coloration, some tips should be noted: (1) As brown itself could not be long-protected at room temperature, deep brown (brown with a hidden color of green) was advocated. During the fading process, a deep brown will eventually turn into green, thus preventing information loss; (2) To simplify the coloration procedure, femtosecond laser with 515 nm is recommended for direct coloration of green; (3) If multicolor integration is applied, orange has to be firstly generated due to the high-temperature annealing. In addition, by using a computer-controlled 3D XYZ stage, laser spatially selective machining with two or three-dimensional patterns can be achieved. Hence, interior coloration in a glass matrix is also an excellent way to permanently mark company logos and product/part information, which could be produced as a security tagging attached onto a device to ensure traceability and also for anti-counterfeiting. Furthermore, the focused light allows for a submicron voxel featuring the color of green or orange, which can facilitate a high-capacity data storage.

 

Fig. 6. (a) Individual and (b) integrated interior coloring in glass matrix. Scale bar is 3 mm.

Download Full Size | PDF

Compared with traditional coloring methods, interior coloration enabled by laser-induced defects and QDs owns it special advantages: (1) noncontact and operating without tooling requirement; (2) high processing speed, high flexibility in automation, and low operating cost (without using consumables); (3) ability to create precise features and interior coloration in a small volume; (4) outstanding design flexibility in patterns and images because no molds or stencils are required, and (5) environmental friendliness due to the chemical free process.

4.2. Multilevel encryption

As a proof-of-concept experiment, a multilevel color encryption strategy was illustrated (Fig. 7). For interior coloration, colors of brown, green and orange were regarded as three independent parameters which could be freely integrated. Unlike interior coloration, in this color encryption scheme, one color of brown/green was artificially divided into two colors as light brown/green and deep brown/green. Consequently, colors of light brown (L-Brown), light green (L-Green), deep brown (D-Grown) and deep green (D-Green) were adopted as four freely switchable variables. Each piece of glass matrix acted as a chip written with a digital 8. For a single digital 8, there are seven separate parts, each of which could be independently colored by one of the four colors. It is worth noting that although the visual difference between the light brown/green and deep brown/green was not quite obvious, the four colors were intrinsically different in physical properties: (1) L-Brown and L-Green were purely due to color centers, which could be totally bleached at 250 °C and 350 °C for 8 h, respectively; (2) D-brown was a combination of L-Brown, L-Green and tiny nuclei of CdTe QDs, and D-Green was a mixture of L-Green and tiny nuclei of CdTe QDs; (3) Tiny nuclei of CdTe QDs was very small crystals, which serve as seeds for the growth of large crystals during annealing. After annealing at 450 °C for 8 h, the invisible small crystals grow into CdTe QDs, rendering the color of orange. It has been discussed in section 3 that if color encryption was applied only between the brown and green, ciphertext has to be decrypted within a short period (brown was unstable at room temperature). In order to achieve a long-time color encryption, the precipitation of CdTe QDs was chosen as the final decryption process. Hence, ciphertext could be written by D-brown and D-Green.

 

Fig. 7. (a) Schematic and (b) a proof-of-concept of multilevel color encryption. Scale bar is 4 mm.

Download Full Size | PDF

In order to perform more color division multiplexing encryptions, four representative color encryption cases were introduced. For each case, Step 1 was the original encryption step, in which ciphertext was encrypted by precisely controlling the laser parameters (see Table S2 in Supporting Information); Step 4 was the decryption process. Only the decryption key (450 °C for 8 h) were matched, secret ciphertext was decrypted; Step 2 and Step 3 was an intermediate state, in which brown was firstly bleached in step 2, and then green disappeared in step 3. Based on this encryption strategy, we can find that when “88” were written, it could be decrypted into nothing (case I), 88 (case II) and 52 (case III). However, for case IV, a multilevel encryption was designed. A combination writing of “54” and “26” was firstly decrypted into “69” and finally “13”, which ensured a higher level of encryption security.

5. Conclusion

In conclusion, spatially selective generation of intrinsic color centers of brown and green, and tiny nuclei of CdTe were achieved in a highly controlled fashion by DLW in silicate glass co-doped with CdO and ZnTe. The controlled heating steps facilitated the growth of CdTe QDs while accelerated the aging of color centers. Hence, permanent color of orange (CdTe QDs) appeared at the cost of the disappearance of brown and green (color centers). Based on the color formation scheme, brown, green and orange were adopted as three independent colors which could be ready for scalable production of 3D volumetric interior coloration on the order of millimeters and even larger areas. Furthermore, by encoding ciphertext into the tiny nuclei of CdTe, a strategy of color encryption and heat decryption is established, which brings about the merits of expanded storage capacity and improved information security. This work provides a viable coloration scheme for multidimensional optical multiplexing, information encryption, and anticounterfeiting.