1. Introduction

Waveguides in optical media are an essential part of a wide variety of important optical devices. In recent years, pulsed laser processing has become an established method for writing waveguides in optical materials; successful examples using femtosecond (fs) laser pulses exist for glasses and single crystals [1–6]. In recent years transparent polycrystalline ceramics have been receiving significantly increased attention for optical applications such as laser source materials [7], solid state lighting [8] and light manipulation [9]. Ceramics offer high temperature and chemical stability and relatively efficient fabrication compared to glasses and single crystals thus promising to increase the application space for optical materials. These advantages have prompted investigations and various successful demonstrations of waveguide written in ceramics [10,11].

A potential drawback to the widespread application of waveguide structures in ceramic-based devices is the relatively high power that is necessary to induce permanent optical changes in ceramics i.e. for waveguide writing. Here we present a method for writing waveguides in transparent polycrystalline ceramics using fs laser pulses with remarkably low energy of 5nJ. This laser energy is orders of magnitude lower than previously reported for waveguide writing in ceramics. Such low energy required for writing waveguides should make these optical ceramics more amenable to industrial application leading to their integration in devices.

2. Experimental procedure

2.1 Ceramic fabrication

Commercial (Tosoh Corporation, Tokyo, Japan) nanocrystalline 8YSZ powder with a reported grain size of 50 nm was densified using current activated pressure assisted densification (CAPAD). Each of the samples was prepared in a graphite die with 19 mm inner diameter. Temperature was measured using a grounded N- type thermocouple placed in a hole drilled halfway through the thickness of the die. The CAPAD processing was done in a custom built apparatus that has been described before [12]. 1.5 g of 8YSZ powder was loaded into the die for each sample. The pressure in the system was raised to 106 MPa before the current was applied.

When this pressure was reached, the current heated the sample to 1200 °C using a heating rate of 200 °C/min. Samples were held for 10 min at final pressure and temperature. The samples were mechanically polished for microstructural and optical characterization. The densified samples were fractured and examined using scanning electron microscopy (SEM) (Phillips FEI). Figure 1 shows an SEM micrograph and optical photograph of a typical sample. The sample was annealed for 8 hours in air at 750 °C, in order to thermally etch the surface.

 

Fig. 1 (a) SEM micrograph of the thermally etched surface of an 8YSZ sample produced via CAPAD at 1200⁰C with a 10 min hold time at temperature, 106 MPa (b) Photograph of the same 8YSZ on top of backlit text showing transparency of the ceramic.

Download Full Size | PDF

2.2 Ceramic annealing

It was previously established that the optical properties such as absorption coefficient are highly dependent on oxygen stoichiometry in 8YSZ [13]. Oxygen stoichiometry in YSZ can be readily controlled by exposure to either reducing or oxidizing atmospheres. The as-processed ceramics in this study are oxygenic deficient do to the reducing nature of CAPAD processing conditions [14] Annealing the YSZ samples in air have the effect of diffusing oxygen back into the sample and removing the oxygen vacancies that were created at high temperatures [13]. Thus, as the transmission of light through a sample is related to the average concentration of oxygen vacancies, there is an increase in transmission when the samples are annealed for various times. In order to evaluate the effects of annealing on waveguide writing, the samples were annealed in air at 750 °C for various times (10 min to 8hrs).

2.3 Femtosecond laser processing

When a femtosecond laser pulse is focused inside a transparent material the irradiance reached in the focal volume may induce nonlinear absorption through a combination of multiphoton absorption, avalanche ionization and tunneling ionization [15]. If enough energy is deposited in the material through this nonlinear absorption, permanent structural changes such as refractive index changes can be induced. Since the nonlinear absorption can be only reached for high irradiances, the changes are induced only in the focal volume. Therefore, by simply translating the sample with respect to the focus and using a continuous train of pulses it is possible to induce permanent changes in the material in a reproducible and controlled manner. In this work we used this technique to write waveguide-like structures in the YSZ ceramic.

For the laser processing of the ceramics a homemade Ti:sapphire oscillator with a central wavelength of 800 nm, a repetition rate of 70 MHz, a maximum on-target per pulse energy of 5 nJ and a pulse duration (FWHM) of about 66 fs was used. The laser beam was focused on the surface sample through a lens of 0.6 N.A. and a focal distance f₁ = 4 mm. To find and visualize the beam waist onto the sample (ω₀ = 1.9 µm, z_R = 7.8 µm) we used a lens of focal distance f₂ = 500 mm and a CCD camera (PL-A774, Pixelink). This system was set in a configuration similar to an image relay arrangement. The samples were translated with a constant speed of 530 µm/s and perpendicular to the incident beam as shown in Fig. 2 . To determine the refractive index change threshold, a series of about 2 mm long lines were written varying the number of laser scans and the per pulse delivered energy by using an attenuator made of a combination a half- wave plate and a polarizer.

 

Fig. 2 Schematic of laser processing set-up.

Download Full Size | PDF

3. Results and discussion

3.1 Waveguide-like structure fabrication

A top view optical micrograph of a series of waveguide-like structures on as processed YSZ is shown in Fig. 3 . These figures show the effects of varying the energy pulse per [(a)3.6, (b) 4.6 and (c) 5 nJ] and the number of passes at each energy. The micrographs reveal that with each of the laser energies in this study, pulsed laser processing produced written structures that are brighter compared to the surrounding material, indicating a permanent change in the optical properties of the irradiated region. This color change can be caused by a change in absorption coefficient, refractive index or both as will be discussed below.

 

Fig. 3 Optical micrograph (in transmission) of the waveguide-like structures written in the YSZ ceramic using varying energies a) 3.6, b) 4.6 and c) 5 nJ per pulse. The inset number over the structures indicates the number of scans along the same track.

Download Full Size | PDF

It is worth emphasizing the extremely low per pulse energy used to write the structures. The energies used here (3.6 nJ to 5 nJ) are at least three or four orders of magnitude lower than previously reported in the literature for laser-induced index trimming in transparent crystals and ceramics [5,6,10,11]. The lowest (threshold energy) for writing is 3.6 nJ (I = 4.4 x 10¹¹ W/cm², F = 31 mJ/cm²). In order to verify that we worked below the ablation threshold, we analyzed the irradiated zones by SEM. We did not find any evidence of ablation.

We also found that the number of scans plays an important role in waveguide writing at all energies. For example at 3.6 nJ (Fig. 3(a)) at a low number of scans (100) no visible change was observed while increasing the number of scans to 150 produced a weak visible change and at 200 scans, the wave-guide structure is clearly visible. The effect of scans was similar at higher energies (Figs. 3(b) and 3(c)); Increasing scan numbers produce more clearly demarcated structures. The dependence on the total number indicates that the total incident pulses per unit area on the sample is important.

Changes in the optical properties during fs processing are often attributed to laser-induced structural changes due to the heat accumulation and field or irradiance related effects [16,17] over the irradiated zone. We do not expect such phase changes in our samples since is the cubic structure of fully stabilized YSZ is known to be very stable [18]. Instead we believe that the changes we observe are related to point defects.

As mentioned earlier, in previous studies [13], we were able to induce color change in YSZ by changing the oxygen stoichiometry in the samples by annealing in the presence of oxygen. The primary absorption centers in YSZ are oxygen vacancies, _{$V_{\ddot{O}}$}with trapped electrons, $e^{\prime }$producing oxygen vacancies with a single positive charge written in Kroger-Vink notation as:

Equation (1) suggest that there are two possibilities for controlling _{$[V_{\dot{O}}]$}: 1) The concentration of$V_{\ddot{O}}$ can vary or 2) The trapped electron, _{$e^{\prime }$}can become de-coupled from the $V_{\ddot{O}}$. Both mechanisms are feasible with pulsed laser processing.

Mechanism 1) requires the thermally driven diffusion of oxygen vacancies. fs laser irradiation could increase the temperature of the samples to temperatures where oxygen vacancies have sufficient mobility to cause significant diffusion. This is feasible because of the time between successive delivered pulses (14 ns, i.e. 70 MHz repetition rate), which for this material, could be much shorter than the characteristic time for thermal diffusion out of the focal volume. As a result, the delivered train of laser pulses deposits energy faster than the time required for heat diffusion to occur, leading to a high temperature rise of the material over the focal region [19,20]. For long enough laser exposures, the heat deposited by the successive pulses of the oscillator diffuses towards the surrounding material inducing changes beyond the focal volume. This can be seen in Fig. 3(c) where the width of the structures is larger than the diameter of the laser spot (≈4 µm). Eaton et al. [21,22] noticed that using a high-repetition rate femtosecond laser it is possible to reach high temperatures by heat accumulation processes.

It is also possible that high electric field caused by the fs laser interaction with YSZ causes decoupling of electrons trapped in vacancies (Mechanism 2). Two types of oxygen vacancies have been identified in reduced cubic YSZ: T- and C-Type oxygen vacancies. The T-type vacancies occur in weakly reducing conditions, while the C-Type occurs in strongly reducing atmospheres. Merino and Orera showed that that the energy gaps between the T- and C-Type oxygen vacancies and the conduction band in reduced YSZ are ~3.3eV and ~2.6eV, respectively [23]. Since we are irradiating the samples with λ = 800nm, corresponding to a photon energy of E_ph = 1.55 eV it is apparently not possible to induce changes in the YSZ via Mechanism 2 with linear photonic absorption. However, since we are employing fs-laser pulses, the intensities are very high indicating the possibility of non-linear effects, in particular 2-photon absorption. The doublet of E_ph is 3.10 eV which is enough energy to decouple trapped electrons from the C-type oxygen vacancies via Mechanism 2.

At this time it is not possible to say with certainty which mechanism is responsible of written structures in this work. Investigations are underway to elucidate the governing mechanism.

In order to determine whether the induced color change in the irradiated zones also led to a refractive index change, we analyzed the post-laser processed ceramics using a phase contrast microscope (Olympus, model BX41). Figure 4 shows a phase contrast micrograph of a waveguide-like structure fabricated at 5 nJ per pulse energy (F = 44 mJ/cm², I = 6.7 x 10¹¹ W/cm²) and 200 scans. The micrograph reveals a high contrast between the resulting structure and the surrounding material. This clearly indicates a refractive index change over the irradiated track. The bright spots visible in sample are likely dust particles and/or other imperfections caused by polishing.

 

Fig. 4 Phase contrast micrograph of a waveguide-like structure written at 5 nJ and 200 scans.

Download Full Size | PDF

While the color change in YSZ has been associated with variance in the oxygen vacancy concentration, we are not aware of experiments showing that the refractive index is also coupled to the degree of oxygen reduction. However measurements of the permittivity have been conducted that can help explain why the refractive index is affected. Henn et al. clearly show an inverse relation between the oxygen vacancy concentration and the relative permittivity in YSZ [24]. Their measurements were done at significantly lower than optical frequencies (10³ Hz), but presumably similar changes could occur over a wide frequency range. Since the refractive index is related to the permittivity by:

3.2 Annealing time influence

Table 1 shows the irradiation results in samples with different annealing time (i.e. different optical properties). All the results were obtained for a per pulse energy of 5 nJ (F = 44 mJ/cm², I = 6.7 x 10¹¹ W/cm²) and 200 scans. We found that the waveguide-like structures writing is easier for samples with a lower annealing time that have lower optical transmittance (higher linear absorption coefficient). This is further evidence that the waveguide writing process is an analogous effect to annealing, i.e is dependent on the concentration of oxygen vacancies in the sample$[V_{\dot{O}}]$.

Table 1. Laser Induced Changes in the YSZ Ceramic as a Function of the Annealing Time

View Table

3.3 Light confinement in the pulsed laser written structures

The ability of the written structures to behave as an optical waveguide was proven by coupling a He-Ne laser trough a single mode optical fiber. The output of the waveguide was collected by a 10x microscope objective coupled to a CCD camera. Figure 5 shows the waveguide-fiber coupling into a not-annealed sample. It can be seen that there is indeed light confinement in the written structures. These results also indicate that the change induced over the irradiated zone corresponds to a refractive index increment (positive Δn). We found that the structures written at higher energies present a better light confinement than those written at lower energies (not shown here). We also observed a better light confinement in structures written at higher number of scans. This confirms a higher refractive index increment over the irradiated zone for structures written at higher energies and higher number of scans.

 

Fig. 5 a) Light confinement in a waveguide-like structure written at 5 nJ and 200 scans. b) Coupling light in a zone out of the waveguide-like structure.

Download Full Size | PDF

Figure 6 shows the intensity profile at the output face of a waveguide written using 5 nJ (F = 44 mJ/cm², I = 6.7 x 10¹¹ W/cm²) and 200 scans. As it can be seen, the waveguide does not produce a single mode intensity distribution at the coupling wavelength (632 nm). This is shown in the inset picture of Fig. 6, which shows the mode intensity distribution of the waveguide. It presents two bright spots which could be two propagation modes of the waveguide. Another important point to note is that the transmittance of the YSZ ceramics increases with increasing wavelengths (see [13]), suggesting that the transmittance of the waveguide structures should be higher at higher wavelengths. This opens the possibility for the use of this material in telecommunications applications (e.g. 1310 and 1550 nm); investigations are underway in this direction as well.

 

Fig. 6 Intensity profile at the output face of the waveguide-like structure for a transmitted beam for a wavelength of 632 nm.

Download Full Size | PDF

4. Conclusion

In conclusion, by using ultralow energy (5 nJ) and ultrafast (66 fs) laser pulses, we were able to write waveguide-like structures in the polycrystalline ceramic YSZ. The energy we used for writing these waveguide-like structures is at least three or four orders of magnitude lower than earlier reported for ceramics. To the best of our knowledge, these are the lowest energies ever reported for the successful writing of waveguides in a ceramic material. We found that the refractive index change increases with the pulse energy and the number of scans. We also found that the writing of these structures is easier for samples with low annealing time (high linear absorption coefficient).