1. Introduction

The femtosecond laser direct-write technique was first demonstrated in 1996 [1]. Since that time a dynamic field of research has grown around this technique because of its unique qualities compared with conventional fabrication techniques such as ion-exchange and photolithography. In particular, it enables rapid prototyping when coupled with high precision translation stages and permits the fabrication of complex 3D structures.

The femtosecond laser direct-write technique has already been successfully utilized to create waveguide lasers based on crystalline and glass materials. In crystals, femtosecond laser irradiation can cause an expansion of the lattice in the focal volume and is generally accompanied with a decrease in refractive index [2]. Waveguides can be created on the sides of the modified region where the refractive index is increased through a stress-induced effect. Typically, the inscribed waveguides usually support only one particular polarization and the direction of the polarized mode is strongly dependent on the properties of the crystal itself [3, 4]. In contrast to crystals, glass has a random network which can support either unpolarised waveguide lasers (or polarized with the addition of stress formers) albeit with typically lower emission cross sections and hence lower gains. The induced refractive index change can be positive or negative depending on the glass properties and fabrication parameters [5, 6]. To date, ultrafast laser written waveguide lasers have been reported in a range of different rare-earth doped glass hosts [7–13], such as Er:Yb-doped phosphate glass [9], Yb-doped bismuthate glass [7], Er:Yb-doped oxyfluoride silicate glass [11], Ho-doped fluorozirconate glass [12] and Yb-doped silicate glass [13]. The emission of these lasers covers a broad range of wavelength spectrum ranging from 1 µm to the mid-infrared. In addition, various types of structures have also been utilized to realize special functions. For example, Ams et al. demonstrated a dual-wavelength waveguide laser based on sampled gratings written in Yb-doped phosphate glass [14]. The body of literature on femtosecond laser written waveguide lasers in glass is predominantly focused on time independent laser operating characteristics such as output power, laser wavelength, slope efficiency and beam quality. To date, the time dependent performance of this type of laser has yet to be studied in detail.

In this paper we examine the time dependent performance (over 30 hours) of femtosecond laser written Yb-doped DFB waveguide lasers fabricated in two of the most commonly investigated glass hosts, namely phosphate and silicate glass. Self-annealing during laser operation is shown to result in diminished performance for the waveguide laser written in the Yb doped phosphate glass whereas the laser threshold and output power for the Yb-doped silicate glass laser is shown to improve over time. The mechanisms underpinning these changes are also discussed.

2. Experiment

In this study the two glass samples used were a 9 wt. % Yb-doped phosphate glass (“QX” Kigre Inc.) and an 8 wt. % Yb-doped silicate glass (“IOG-10” Schott). In each glass sample, devices were fabricated using a regeneratively amplified Ti:sapphire laser which had a 1 kHz repetition rate, 120 femtosecond pulse duration, and operated at 800 nm. The laser beam was circularly polarized in order to induce the maximum refractive index change [15]. The beam was focused into the glass by either a 20 × (NA 0.46) or 40 × (NA 0.6) microscope objective at a depth of 170 µm below the surface of the glass sample. A slit was also placed in the beam path in order to produce waveguides with a circular cross section [16].

After fabrication, the glass samples were polished with a Logitech (PM5) lapping and polishing machine to expose the ends of the devices. The morphology and guided mode field profile of the laser written devices were investigated with a differential interference contrast (DIC) microscope (Olympus IX81) and CCD camera imaging system with a light source at 976 nm respectively.

A series of waveguides were first written in the glass samples in order to characterize the laser induced modifications. Following this, DFB lasers were fabricated by writing waveguides composed of uniform first order Bragg gratings. The performance of these DFB lasers in the two kinds of substrates was then compared. Finally, to understand the improved performance of the silicate host over the phosphate host, we characterized the grating strength at 1550 nm as well as the transmission loss at 1310 nm before and after pumping the gratings at 976 nm. 1550 nm and 1310 nm were chosen as they fall away from the pump/laser absorption lines near 1 µm.

3. Results and discussion

3.1. Physical characteristics of femtosecond laser modified Yb-doped silicate and phosphate glass

Table 1 lists the pulse energy, physical diameter and mode field diameter (MFD) of waveguides inscribed in the phosphate and silicate samples using a 20 × and 40 × microscope objective. MFDs were taken using the ISO standard D4σ or second moment width. For doped phosphate glass, the threshold for inducing a positive refractive index change was found to be 0.9 µJ and 0.35 µJ when using the 20 × and 40 × microscope objectives respectively. Waveguides written with the 20 × microscope objective had a physical diameter ~4.7 µm with a mode field diameter ~15 µm. However, for waveguides written using the 40 × microscope objective, the size of the waveguides were too small to effectively guide light near 976 nm. The minimum pulse energy for inducing a positive refractive index modification in the doped silicate sample with the 20 × and 40 × microscope objective was 2.6 µJ and 0.25 µJ respectively. The physical diameter of waveguides inscribed using the 20 × objective was as large as 10.5 µm, due to the high pulse energy required to reach the modification threshold, leading to multi-mode guiding near 976 nm. However, waveguides written using the 40 × microscope objective had physical diameters of ~6.5 µm guiding a single mode with MFD ~10 µm. This is in contrast to the work reported by Palmer et al. who in the thermal regime required a depressed cladding arrangement to construct a single mode waveguide in the same substrate [13].

Table 1. Characteristics of waveguides fabricated in Yb-doped samples using different writing parameters.

View Table

3.2 DFB waveguide laser characterization and comparison

To create a DFB WGL, waveguide Bragg gratings (WBGs) were fabricated. To achieve this, the writing laser was 100% intensity modulated with a 50:50 mark-space ratio while the glass sample was being translated. In silicate glass, first order WBGs were fabricated with the 40 × microscope objective and pulse energy of 0.55 µJ while for phosphate glass, the 20 × microscope objective and a pulse energy of 1 µJ was used. While writing parameters were optimized to minimize MFD and hence lasing thresholds, no effort was made to optimize grating characteristics and hence laser output powers.

The WBG was pumped by a laser diode (central wavelength of 976 nm) through a wavelength division multiplexer (WDM) as shown in Fig. 1.The laser output through the WDMs was recorded with an optical spectrum analyzer (OSA) and a power meter. The OSA has a resolution bandwidth of 10 pm. Index matching oil, with a refractive index that closely approximated that of the glass, was used between the butt-coupled fiber and the glass sample in order to reduce the Fresnel reflection. Small sections of graded index fiber were spliced to the ends of the coupling fibers, increasing the pump mode to match the single guided mode of the WBG near 976 nm.

 

Fig. 1 Schematic diagram of the setup used to characterize the DFB waveguide lasers. WDM – Wavelength division multiplexer, WGL – waveguide laser, OSA – Optical spectrum analyzer. (The second pump shown in the dotted box was only used for grating depth studies which required a bidirectional pumping scheme).

Download Full Size | PDF

The DFB lasers were continuously pumped for 30 hours at a pump power of 300 mW in order to monitor their time dependent performance. Prior to each measurement being taken, small adjustments between the alignment of the pump fibers and WBGs were made to maximize the light throughput. The output power and the emission peak, plotted as a function of time, are shown in Fig. 2and Fig. 3, respectively.It can be seen in Fig. 2 that for the phosphate laser, the output power decreased rapidly from 12 mW to less than 6 mW in the first three hours followed by a lower decay rate. After 30 hours of operation, the output power of the phosphate laser was close to zero. In contrast, the output power for the silicate laser increased slightly over the duration of the study. In addition, the threshold of the laser was also observed to decrease. Both lasers experienced a blue-shift of the emission peak (see Fig. 3), typically an indication of a reduction in the WBG index contrast. However, the range of the wavelength shift of the silicate laser was smaller than that of the phosphate laser.

 

Fig. 2 Variation of DFB laser output power as a function of time

Download Full Size | PDF

 

Fig. 3 Variation of DFB laser emission peak as a function of time

Download Full Size | PDF

The mechanism by which refractive index changes are produced in glass substrates exposed to femtosecond lasers is very much dependent on the irradiation conditions. Two different mechanisms have been proposed. In the thermal regime, associated with high repetition rate lasers, a local densification or rarefaction coupled with species migration has been found to be the main contribution to the refractive index change [17, 18]. However, in the athermal regime, associated with low repetition rate lasers, the induced refractive index modification has been attributed to both densification and the conversion of bridging oxygen atoms to non-bridging oxygen hole centers which result in the formation of color centers [18, 19]. In Yb-doped phosphate glass, color centers in this form are not photo-stable and can be removed when exposed to UV/visible light. Photo-annealing experiments conducted by Dekker et al. [20] have shown when pumping at 976 nm (with the associated cooperative luminescence near 500 nm) that the refractive index contrast of a WBG in Yb-doped phosphate decays rapidly in the first 3 hours and then decreases slowly, consistent with that of the laser output shown here in Fig. 2. The blue shift of the grating and decrease in laser output is also due to the decay of the effective grating index through the same mechanisms. In contrast to the degradation of performance of DFB lasers based on Kigre QX (phosphate), the output of the DFB lasers based on Schott IOG-10 (silicate) improves over time.

In order to further study the influence of pump light absorption on the refractive index of IOG-10 waveguides and the associated effect on DFB waveguide laser performance, first-order WBGs at 1545 nm were fabricated and bi-directionally pumped at 976 nm for 15 hours. The maximum pump power was approximately 300 mW from each end resulting in complete saturation of the pump power over the entire 11 mm long device. After each exposure time the sample was left to cool for more than 20 minutes before measuring the grating depth and Bragg wavelength using a swept wavelength system. The grating depth as a function of accumulated pump exposure time is shown in Fig. 4.Clearly the grating depth increases almost immediately after 976 nm excitation had commenced and then relaxes over many hours to a point close to its original value. The Bragg wavelength (not shown) varies by approximately 10 pm during the course of the measurement without any clear pattern, a drift that is attributed to both the resolution of the OSA and variation in the room temperature of around 1°C. These results indicate that the improved laser performance of the silicate host over time is not simply attributed to improved grating reflectivity. Therefore, it is postulated that the improved performance of the silicate laser over time including increased output power, reduced threshold and the accompanied blue shift are not due to the refractive index contrast variation of the grating but are instead related to a reduction in propagation loss.

 

Fig. 4 Grating depth of a 1545 nm waveguide Bragg grating in silicate glass after 976 nm pump exposure as a function of time.

Download Full Size | PDF

To prove this we fabricated WBGs near 1020 nm and measured the transmission of off-resonance 1310 nm light through the WBG before and after pumping at 976 nm. The WBG was pumped from both ends and lased during these experiments. After each pump duration the 1310 nm probe laser was connected and the transmitted power measured. The results are shown in Fig. 5. From this data it can be seen that there is a large increase in transmission immediately following exposure to the pump. The spread of the experimental data seen in Fig. 5 is a result of variation caused by adjustments in pump-laser alignment after reconnecting the 1310 nm probe over multiple repetitions of the experiment. To rule out the effect of a change to the MFD of the WBG after exposure to the net increase in WBG transmission, we measured the MFD both before and after 976 nm exposure. We found the MFD reduced slightly, resulting in less than 1% change to the coupling. These measurements we believe affirm our understanding that the improved laser performance of the silicate DFB waveguide laser over time are due to reduced scattering of the WBG after 976 nm light exposure. This data also explains why there is a slight blue shift in laser wavelength while maintaining the same or higher refractive index contrast (associated with the grating strength). There was no phase change in the WBG so the laser wavelength was free to vary along the edge of the grating resonance dependent on the net gain which typically blue shifts with reducing losses as is typical in 3 level laser systems.

 

Fig. 5 Off-resonance throughput measured at 1310 nm for silicate glass waveguide Bragg gratings after pump exposure at 976 nm over time. The Bragg wavelength for the grating was 1020 nm.

Download Full Size | PDF

4. Conclusion

DFB waveguide lasers were fabricated in Yb-doped phosphate and silicate glass hosts using the femtosecond laser direct-write technique. The time dependent performance of these waveguide lasers was compared over 30 hours. The phosphate based laser degraded during the course of experiments while the silicate based laser improved in performance. In the low repetition rate regime (1 kHz), the refractive index change of these glasses is largely due to densification and the formation of color centers. In a phosphate host these color centers fade away during lasing operation. In contrast, the silicate host showed improved laser and optical properties after pumping, with a constant grating refractive index contrast suggesting a photo-stable color center concentration. The reduced threshold and increased output power over time are attributed to a reduction in propagation loss.