1. Introduction

Femtosecond (fs) lasers are now being widely used in the internal modification of transparent materials. Because of a multiphoton absorption process, the interaction between the fs laser and the material only occurs in the vicinity of the focal spot, leaving the surface of the material intact. One of the most important applications of this ability is to modify the refractive index of bulk glasses such as pure and boron-doped fused silica glasses [1,2], soda-lime glass [3], and phosphate glass [4]. A variety of photonic devices such as waveguides [1], couplers [2,5], and gratings [3,6] have been fabricated inside the glasses with two-dimensional (2D) or three-dimensional (3D) configuration. In general, the refractive index change increases with laser intensity [6] or exposure time [5], until the refractive index change reaches a saturation value. However, high laser intensity or long exposure time usually makes the modified area large, resulting in a low spatial resolution of modification. Moreover, high intensity laser irradiation causes high roughness of the internal structure, and thereby induces the scattering of light. The conflict in simultaneously obtaining the high spatial resolution and the high refractive index change has been an important issue for the fabrication of photonic devices with fine structures, such as photonic crystals [7] and optical gratings.

Although the mechanism is still under investigation, some authors have suggested that a densification from local melting and the subsequent rapid quenching after the optical breakdown is responsible for the refractive index change [1,8]. In this sense, the direct induction of the refractive index change in glasses by fs laser irradiation is a heat-induced process. Therefore, the resolution of fabrication is not only determined by the area where the multiphoton absorption occurs, but is also dependent on the area of heat diffusion. In many cases, the latter factor becomes the bottleneck for obtaining high spatial resolution even for extremely short pulse duration of the fs laser.

To overcome the above difficulty, a new strategy is to use the fs laser merely for initiation of a secondary technological process. In this work, we attempt to modify the optical properties of glass, in particular, a photosensitive glass Foturan, by the photochemical reaction of a fs laser with small pulse energy which sows the seeds of the modification for postbaking.

Foturan glass has been successfully used for microstructuring using UV light [9,10]. It is composed of lithium aluminosilicate glass doped with trace amounts of silver and cerium. By exposure to UV light, the cerium ions (Ce³⁺) release electrons to the silver ions (Ag⁺) to induce silver atoms. In the following postbaking, first the silver atoms diffuse and agglomerate to clusters at around 500°C, and then the crystalline phase of lithium metasilicate grows into an amorphous glass matrix using the silver clusters as nuclei at around 600°C. Since the crystalline phase of lithium metasilicate is much more soluble in a dilute solution of hydrofluoric (HF) acid than the glass matrix, it can be preferentially etched away. Recently, it was found that the photosensitivity of Foturan glass can also be activated with infrared fs laser pulses by a multiphoton absorption process [11], and 3D microstructures such as microfluidic structures and microoptical structures have been embedded inside the glass [11–13].

Since it has been pointed out that a dielectric medium with embedded metallic nanoparticles changes its permittivity as well as the refractive index [14,15], modification of optical properties of Foturan glass becomes possible if the postbaking of the sample is conducted at a temperature below that for crystallization. The required intensity of the fs laser is only the threshold of the multiphoton absorption which is much smaller than that for direct refractive index change by the fs laser [11]. In this case, the refractive index change is actually realized by the postbaking, and the modified region is confined only in the latent image region written by fs laser irradiation. In this manner, a high spatial resolution of modification, controllable refractive index change, and smooth internal structure could be achieved.

2. Experiment

The experiment was performed at a commercial fs laser workstation at Hoya Photonics Corporation, which was described in detail elsewhere [16]. The laser wavelength, pulse width and repetition rate were 775nm, 140±5fs and 1kHz, respectively. The focusing lens was a 50X microscope objective with a numerical aperture (NA) of 0.80. The sample under fabrication was translated by a PC-controlled XYZ stage at a resolution of ±0.5µm. The fabrication process was displayed on the PC monitor by a charge-coupled device (CCD).

In order to form the internal grating structure, we first focused the fs laser beam on the surface of the glass on the stage, and then lowered the objective lens by 20µm. By taking into account the refractive index of Foturan glass which is ~1.5 (the accurate refractive index of Foturan glass is 1.515 at 546.1nm, 25°C [9]), the focal spot was determined to be ~30µm below the surface. The laser pulse energy was chosen as 70nJ/pulse, and the scanning speed was set at 200µm/s. Under these irradiation conditions, no visible modifications of the glass such as color change or ablation could be observed on the PC monitor, for either case of focusing on the surface or focusing into the glass. Latent grating structures with a pitch of 5µm were inscribed into the glass by laser direct writing. The size of all gratings was 1.5mm×1.5mm to accommodate the beam size of a He-Ne laser used for examination of diffraction. After exposure to the fs laser, the samples were subjected to postbaking. The temperature was first ramped from room temperature to 520°C at 5°C/min, and then held at 520°C, which was sufficiently high for silver nanoparticle formation but lower than the crystallization temperature, for various durations from 3h to 18h. Finally, the samples were naturally cooled down to room temperature. We then inspected the grating structures using an optical microscope, and examined their diffraction efficiencies using a He-Ne laser beam.

3. Results

We first checked the samples immediately after exposure to the fs laser irradiation. No visible change of the glass could be found under the optical microscope, and also no diffraction effect could be observed by irradiating the samples with the He-Ne laser beam. So the refractive index change of the glass is negligible. However, after the sample underwent postbaking, a significant change in color could be observed under the optical microscope, which is a signature of the precipitation of metallic silver nanoparticles. Figure 1(a) shows a grating baked for 3h. The first-order and second-order diffraction patterns could be clearly observed when the sample was irradiated using the He-Ne laser beam, as shown in Fig. 1(b).

 

Fig. 1. (a) Optical micrograph of a grating structure embedded in the photosensitive glass. The sample was baked for 3h. (b) Diffraction pattern of the grating with a He-Ne laser beam.

Download Full Size | PDF

 

Fig. 2. (a) Optical micrograph of a grating structure embedded in the photosensitive glass. The sample was baked for 18h. (b) Diffraction pattern of the grating with a He-Ne laser beam.

Download Full Size | PDF

The color of the grating became darker with increasing postbaking duration, implying that more silver atoms agglomerate to metallic silver nanoparticles. Figure 2(a) shows a micrograph of another grating structure which was baked for 18h. The diffraction pattern from the grating is presented in Fig. 2(b), showing a diffraction efficiency much higher than that of the grating in Fig. 1(a). It is interesting to note that the width of each line inscribed by the fs laser was almost unchanged in Figs. 1(a) and 2(a), indicating that the resolution of modification is merely determined by the parameters of the fs laser irradiation. The width of each line is measured to be approximately 1.25µm, which is smaller than the diameter of the focal spot of the fs laser inside the glass (~1.5µm), since the multiphoton absorption took place only in a region smaller than the focal spot. Higher resolution is possible by more carefully controlling the pulse energy according to the critical dose of the multiphoton absorption of Foturan glass [11], or by choosing a larger NA for the objective lens.

As has been pointed out in Refs. 14 and 15, the formation of metallic silver nanoparticles within a dielectric medium results in a refractive index change which is a function of the volume fraction of metallic silver. The refractive index change of the Foturan glass can be calculated by the diffraction efficiency of the grating structure, and thereby the volume fraction of metallic silver can be determined. In order to clarify the influence of postbaking duration on the refractive index change, we baked the samples for 3h, 6h, 9h, 12h, 15h, and 18h, and then checked the first-order diffraction efficiencies for these samples. Here we define the first-order diffraction efficiency as the ratio of the averaged power of the ±1- order diffraction to the power of the incident light. To exclude the influence of the Fresnel reflections at the front surface and the back surface of the glass, we measured the power of the incident light by propagating the He-Ne laser beam through an unexposed area of the sample. The refractive index changes were calculated by the simple assumption that the fabricated internal structure is a thin grating with a pure sinusoidal phase modulation, in which the maximum phase change is Δϕ=2π·Δn·d/λ, where Δn is the refractive index change at the center of the grating line, d is the thickness of the grating structure, and λ=633nm is the wavelength of the He-Ne laser. To obtain the value of d, we cleaved one grating structure and directly measured d as ~12µm using the optical microscope. It should be noted that a thin grating implies that the grating thickness parameter, which is defined as Q=2π·λ·d/(n·Λ²), satisfies Q<1, whereas a thick grating implies that the grating thickness parameter Q>10 [17]. For our grating structures, the refractive index is n≈1.5, and the pitch is Λ=5µm. Thus the grating thickness parameter Q is about 1.27, making it more reasonable to use the diffraction theory for a thin grating.

Figure 3 shows the average efficiency of the first-order diffraction as a function of postbaking duration. It appears that the increase in diffraction efficiency with postbaking duration can be roughly divided into two stages, namely, postbaking durations below and above 12h. The diffraction efficiency shows an accelerated increase with postbaking duration up to 12h; however, it shows a saturation tendency above 12h.

Figure 4 shows both the refractive index change and the volume fraction of metallic silver in the glass as a function of postbaking duration. The volume fractions of metallic silver were calculated using Eqs. (2–5) in Ref. [15] based on the calculated refractive indices. The curves in Fig. 4 again show different increase rates for the duration of postbaking below and above 12h. The almost linear increase in refractive index up to 12h may be ascribed to an increase in size of silver nanoparticles with baking time which should be related to the volume fraction. We suspect that the slowdown or saturation of the increases in refractive index and volume fraction of metallic silver for postbaking duration over 12h is due to the depletion of silver atoms within the glass. In fact, the volume fraction of metallic silver reached 1.17×10^-3 for the sample baked for 18h. Considering that the density of metallic silver is 10.49g/cm³ and the density of Foturan glass is 2.37g/cm³ [9], the weight ratio of metallic silver is therefore ~0.52%. The weight ratio of metallic silver in the sample baked for 18h is fairly consistent with the concentration of Ag₂O (<1%) in commercial Foturan glass [18]. At the postbaking duration of 18h, the refractive index increase is estimated to be as high as ~4×10^-3.

 

Fig. 3. The averaged first-order diffraction efficiency as a function of postbaking duration.

Download Full Size | PDF

 

Fig. 4. The refractive index change (square) and the volume fraction of metallic silver (circle) as functions of postbaking duration.

Download Full Size | PDF

4. Discussion

Controllable refractive index change in the photosensitive glass has been achieved by fs laser initiated photochemical process followed by programmed postbaking. The connection between the volume fraction of metallic silver and the refraction index change suggests that a higher refractive index change is achievable if one can raise the doping concentration of silver ions in the photosensitive glass. For instance, if the volume fraction of metallic silver in the glass reaches 1%, the refractive index change will be 3.3×10^-2. A high volume fraction of metallic silver in the glass also enables fabrication of novel optical materials, such as zero-permittivity materials with band gaps in the visible range [14].

Gaussian beam of the fs laser with pulse energy controlled close to the multiphoton absorption threshold can achieve the spatial resolution beyond the diffraction limit. However, the direct refractive index change using fs laser reported by many groups so far would be heat-induced process, resulting in degradation of the spatial resolution due to heat diffusion. On the other hand, although the refractive index change presented in our work also relies on the diffusion of silver atoms and the subsequently formation of silver clusters, the obtained grating lines would be still confined in the laser-irradiated area even after the postbaking duration of 18h, remaining the high spatial resolution of the microstructure. To discuss this phenomenon, we must go back to the mechanism of the formation of silver nanoparticles inside the Foturan glass. One mechanism proposed is that after the photosensitive glass is exposed to UV light, the cerium ion (Ce³⁺) gives up an electron to become Ce⁴⁺ ion. These electrons could reduce the silver ions (Ag⁺) to form silver atoms (Ag⁰). In the subsequent heat treatment, the silver atoms diffuse and agglomerate to form silver clusters at about 500°C. If this mechanism is true, after the long-time postbaking, the silver atoms are likely to diffuse out of the laser irradiated area. However, this is not the situation occurred in our experiment. In fact, the width of modified regions was unchanged for different durations of the heat treatment. Thus the mechanism containing only reduction of silver ions to atoms appears not sufficient to explain the formation of silver nanoparticles.

A more reasonable explanation suggests that after exposure to the UV light, the free electrons created from Ce³⁺ would not be trapped by the silver ions immediately. On the contrary, each free electron would stay near its Ce⁴⁺ parent ion to form an electron-ion pair. In the subsequent heat treatment, the silver ions would move to these free electrons due to a Coulomb force and then agglomerate to silver clusters [19]. Applying this mechanism to explain our experimental results just needs a little modification, because in fs laser irradiation, Ce³⁺ seems to be unnecessary from our current experiment. In this case, free electrons seem to be generated by oxygen bond breaking, resulting in oxygen deficient centers. It may be possible that the free electrons stay near the oxygen deficient centers after the exposure step, becoming the nucleation agents for the formation of silver nanoparticles. This mechanism can justify the high spatial resolution achieved in our experiment, because in this case the diffusion of silver atoms is actually not random but an organized movement toward only to the region exposed to the fs laser irradiation. If the mechanism is true, the theoretical limit of the spatial resolution achievable in the fs laser microstructuring of Foturan glass based on the multiphoton process will ultimately be determined by the size of the silver nanoparticles, which is about tens to a few hundreds of nanometers, depending on the doping concentration of the silver ions in the glass, the exposure conditions, and the postbaking temperature and duration. Currently, the mechanism of silver agglomeration in the photosensitive glass is still under investigation, and we are now designing new experiment relevant to this issue.

Finally, we would like to point out that, before this work, the selective precipitation of silver nanoparticles inside a glass with fs laser pulses had been reported by Qiu et al. [20]. However, distinct experimental conditions were applied in our work and Qiu’s. In Qiu’s work, high intensity laser pulses which induce optical breakdown in the glass were chosen, consequently, the color change could be directly generated by the fs laser irradiation even without postbaking. However, in our experiment, we chose low intensity laser pulses to avoid the optical breakdown for achieving a high spatial resolution. Moreover, we paid more attention to the refractive index change than to the color change, pursuing applications of the technique in the fabrication of photonic materials and photonic devices.

5. Conclusion

In conclusion, we demonstrate that the refractive index change can be realized and controlled in Foturan glass by fs laser irradiation followed by postbaking. The advantages of using a low pulse energy in the fabrication include (1) enhancement of spatial resolution, (2) reduction of roughness, (3) enlargement of exposure area in the case that the internal grating structure is produced by interference of multiple fs laser beams [3,21], and (4) modification of deep regions without the use of a chirped pulse [21] since the multiphoton absorption occurs only at the modified region. The high spatial resolution is of great use to the fabrication of 3D photonic devices with fine structures, such as photonic crystals or multiplayer volume gratings. To achieve higher refractive index changes with this technique, the doping concentration of silver ions should be raised in the glass; however, the refractive index increase of ~4×10^-3 attained in this study is sufficiently high for practical applications.