1. INTRODUCTION

Direct laser sources in the 2 μm range have seen recent growth in utility. 2 μm laser sources share many commonalities with organic absorption lines and are widely considered “eye-safe” due to their high absorption coefficient in intraocular fluids and tissues. In this work, we demonstrate the use of Ho:YAG ultrafast laser inscription (ULI) waveguides to create a compact laser source, which could be suitable for medical [1] and dental [2] applications and as scientific laser sources [3,4]. Typically, holmium ions in a garnet host operate in two regimes: either by in-band pumping of the holmium ions’ [5] ${}^5{\mathrm{I}}_7-{{}^5\mathrm{I}}_8$ transition or by energy transfer from other rare earth ions, in a co-doped crystal [6]. For both regimes, the laser operates on the holmium ${}^5{\mathrm{I}}_7-{{}^5\mathrm{I}}_8$ transition producing 2.09 μm radiation in a YAG crystal host at room temperature [7].

Guided wave operation of laser sources offers several advantages to bulk lasers, such as increased interaction length, compact size, and increased reliability. One approach to guided-wave operation of Ho:YAG lasers is through doping of crystalline YAG fibers [8]. However, the relative immaturity of this technique has slowed the adaptation of this technology as a way of making laser sources. A more popularized approach is holmium doping or co-doping of holmium and thulium of ZBLAN fibers [9,10]. A maximum output of 6.6 W has been experimentally shown for a Ho:ZBLAN fiber [11]. Tm, Ho, and co-doped Tm Ho silica fiber lasers have been created with output powers well in excess of 10 W [12] and are commercially available. Finally, techniques such as adhesive free bonded waveguides have shown promise for producing efficient laser sources [13,14].

Direct creation of waveguides in bulk material using ULI is an approach that was successfully used in YAG [15] and has produced waveguide lasers in Tm:YAG [16] and Tm:ZBLAN [17]. However, a ULI waveguide laser has not been reported in Ho:YAG, to the best of our knowledge. In this paper, we present demonstration of both single-mode and multimode waveguide lasers in Ho:YAG. Waveguide operation of this material allows for several advantages over bulk operation, including improved longitudinal overlap between pump and laser modes, reduced system complexity, and decreased size and weight.

2. FABRICATION

ULI has been shown to be a viable technique for creating waveguides inside of a variety of materials, including glass [18], chalcogenides [19], and various crystalline hosts [20]. For this study, multimode and single-mode waveguides were inscribed in bulk Ho:YAG crystals. This was achieved using a chirped pulse, amplified Yb-fiber laser (IMRA μJewel D1000 and IMRA μJewel D400) operating at approximately 1045 nm. The Ho:YAG samples were doped with 0.5 at. % Ho and had dimensions of $5\mathrm{mm}\times 5\mathrm{mm}\times 14\mathrm{mm}$. The waveguides were inscribed at two different repetition frequencies: 100 kHz for the multimode waveguides and 500 kHz for the single-mode waveguides. The inscription setups had pulse widths of 600 and 359 fs, respectively. Pulse energies of 1.2 μJ were used to inscribe the multimode waveguides, while pulse energies of 250 nJ were used for the single-mode waveguides. In addition, the multimode waveguides were inscribed using a 0.68 NA lens with a focal length of 4.1 mm producing a focal spot of 1 μm. The single-mode waveguides were inscribed with a 0.4 NA lens with a focal length of 6.2 mm, producing a focal spot of 2 μm. The sample was translated using an XYZ translation stage (Aerotech Model A3200) moving with a maximum velocity of 10 mm/s. The high repetition rate laser in conjunction with the fast translation speed allows single devices to be fabricated in minutes and entire arrays of waveguides to be produced in less than an hour. The resultant inscribed cladding structures can be seen in Fig. 1. The waveguides produced were uniform in the direction of propagation and were slightly nonuniform in the transverse directions. The nonuniformities arise from slight changes in the focusing of the beam due to the depth inside of the sample and change in the overlap with surrounding modified regions. It should be noted that the dark circle created by the femtosecond laser is actually a damage region, and the guiding occurs inside of the damage ring. Similar structures have been demonstrated by other authors [19,21]. The ULI device was designed to have a number of waveguides with different core diameters to demonstrate a wide variety of operating parameters. The core sizes varied between 30 and 120 μm. As core diameter increased, the number of inscription elements also increased. For example, the 50 μm was composed of 60 individual inscription elements, and the 80 μm waveguide was composed of 80 elements. In addition, two different recipes were used for creation of single-mode and multimode waveguides. The difference in the two recipes arose from creating a larger index difference between the core and the cladding. Similar results could be obtained by making the single-mode waveguides larger or by making the multimode waveguides smaller. While the optical propagation losses for these samples can vary from waveguide to waveguide, a maximum loss of 1.5 dB/cm was measured using the technique outlined by Okamura et al. [22].

 

Fig. 1. Waveguides inscribed in a 0.5 at. % Ho:YAG sample. The picture was taken looking into the direction of propagation of the waveguide.

Download Full Size | PDF

3. EXPERIMENTAL SETUP

To test the performance of the ULI waveguide devices, the waveguide sample was placed on a water cooled mount attached to a five-axis adjustable stage. The cooling block was kept at 15°C. Figure 2 shows the cavity configurations [Figs. 2A and 2B] used for testing of the lasers. The configuration in Fig. 2A was used for the multimode waveguides and the configuration in Fig. 2B was used for the single-mode waveguides. Mirror M1 was AR coated for 1.7–2 μm and HR for 2.05–2.2 μm. Mirror M2 was AR for 1.7–2 μm in addition to acting as the outcoupler for 2.05–2.2 μm, which varied from 30% to 97% reflective. For the multimode setup in Fig. 2A, lenses L1 and L2 were broadband AR coated (2–3 μm) with a focal length of 5 cm. For the single-mode setup in Fig. 2B, lenses L1 and L2 were broadband AR coated (2–3 μm) with focal lengths of 4 and 5 cm, respectively. The pump source for the Ho:YAG crystal was a CW thulium fiber laser (IPG model TLR-1908-LP), which produced a maximum output power of 12 W. Lens L1 was used to focus the pump beam to a spot size of $\sim 50\mathrm{\mu m}$. Furthermore, both entrance and exit facets of the sample were uncoated. While various waveguide sizes were manufactured and evaluated, only the best performing multimode and single-mode waveguide results are presented here.

 

Fig. 2. Cavity configuration used for testing of (A) the multimode waveguides and (B) the single-mode waveguides.

Download Full Size | PDF

4. MULTIMODE OPERATION

Under in-band pumping conditions at 1.9 μm, the sample shown in Fig. 1 demonstrated lasing on the ${}^5{\mathrm{I}}_7-{{}^5\mathrm{I}}_8$ transition. Figure 3 shows the laser performance of an 80 μm diameter multimode waveguide. From measuring the diverging light from the waveguide, we estimate that an index difference of $\mathrm{\Delta }n=0.01$ was obtained. This guide produced the highest output power given the pump conditions and setup shown in Fig. 2A. The slope efficiency was obtained for several output coupler reflectivities ranging from 30% to 90%. It should be noted that, for this measurement, the pump power is recorded as the power incident on the crystal. Using a 50% reflective outcoupler, we obtained a maximum output power of 1.88 W with a slope efficiency of 29%. As the output coupler reflectivities deviated from the 50%/70% level, output power and sloped efficiency decreased. Furthermore, the threshold powers for the 90%, 70%, 50%, and 30% outcouplers were 2.1, 2.5, 3.2, and 3.6 W, respectively. Table 1 reports the slope efficiency and output power obtained from each outcoupler reflectivity. At incident pump powers greater than 10 W, thermal rollover in the output power was observed. While not investigated here, thermal rollover could be mitigated by reducing the duty cycle of the pump or by decreasing the temperature of the cooling block.

Table 1. Slope Efficiencies and Maximum Output Powers of Fig. 3

View Table | View all tables in this article

 

Fig. 3. Laser performance for various output coupler reflectivities for the 80 μm multimode Ho:YAG waveguide.

Download Full Size | PDF

The spectral characteristics of the waveguide lasers were characterized using a Yokogawa optical spectrum analyzer (OSA) with a resolution of 0.02 nm (AQ6375). Figure 4 shows the spectral output of the multimode waveguides. It can be seen that the output wavelength depended on the reflectivity of the outcoupler. For the 30%, 50%, and 70% reflective outcouplers, the output wavelength was centered at 2090 nm. For the 90% outcoupler, the wavelength shifted to 2120 nm. This shift is due to the increased round-trip losses associated with the lower reflectivity outcoupler and increased reabsorption losses due to overlap of the emission and absorption cross sections [23]. The lasing wavelength shifts to the wavelength where the threshold power is the lowest, which, for higher round-trip losses, forces the laser to run at the peak of the emission cross section at $\approx 2090\mathrm{nm}$.

 

Fig. 4. Spectral output of the 80 μm waveguide dependent on output coupler reflectivity.

Download Full Size | PDF

5. SINGLE-MODE OPERATION

Single-mode waveguides were tested under conditions similar to those described above. The 50 μm diameter waveguide was pumped using a 1.9 μm thulium fiber laser. From measuring the diverging light from the waveguide, we estimate that an index difference of $\mathrm{\Delta }n=0.0003$ was obtained. Figure 5 shows the laser performance of the single-mode waveguides for various output coupler reflectivities. It can be seen that the 70% reflective outcoupler produced the best performance with a maximum output power of 1.78 W and a slope efficiency of 16%. Table 2 reports the results for the single-mode waveguides. Unlike the multimode waveguides where the threshold was found to be 2–4 W, the single-mode waveguide exhibited an extremely low threshold of approximately 100 mW. Previous authors have indicated that losses as low as 1.5 dB/cm could be produced in Ho:YAG waveguides [15]; however, no laser operation has been demonstrated. The low threshold here is an indication that the losses for the single-mode waveguides are significantly decreased compared to previously reported values.

Table 2. Slope Efficiencies and Output Powers of Fig. 5

View Table | View all tables in this article

 

Fig. 5. Slope efficiency for the single-mode waveguides.

Download Full Size | PDF

In addition to the performance, the spectral output of the single-mode waveguide laser was characterized with a ThorLabs OSA (OSA205) with a resolution of 0.01 nm. These results can be seen in Fig. 6. Similar to the multimode waveguides, the lower reflectivity outcouplers (60% to 80% reflective) had operating wavelengths centered at 2090 nm. The 97% reflective outcoupler redshifted the operating wavelength to 2097 nm.

 

Fig. 6. Output spectra of the single-mode 50 μm waveguide as a function of outcoupling reflectivity.

Download Full Size | PDF

For comparison, single-mode and multimode waveguides in Tm:YAG [16] present similar behavior to the results discussed above. In general, the multimode waveguides for both Tm and Ho:YAG exhibit higher slope efficiencies compared to the single-mode waveguides. The multimode waveguides show promise for power scaling of these devices. However, this is at the cost of decreased mode quality.

6. MODE PROFILES

The transverse mode profile of the 80 μm multimode waveguide can be seen in the subset picture in Fig. 7. To characterize the beam quality of the laser, the $M^2$ parameter was measured using a 10 cm lens to focus the output beam. The $1/{\mathrm{e}}^2$ beam diameter was measured at several positions and the $M^2$, was found by a least squares fit to the experimental data using

 

Fig. 7. $M^2$ measurement for the multimode waveguide. The $1/e^2$ width was plotted as a function of distance when focused using a 10 cm focal length lens. Subset picture shows the far-field output mode profile in a $250\mathrm{\mu m}\times 250\mathrm{\mu m}$ image.

Download Full Size | PDF

The $M^2$ measurement was then repeated for the single-mode waveguide. The measured beam widths and subsequent fit can be seen in Fig. 8. From the fit, the single mode waveguides produced a beam with an $M^2$ value of 1.41.

 

Fig. 8. $M^2$ measurement for the single-mode waveguide. The $1/e^2$ width was plotted as a function of distance when focused using a 10 cm focal length lens. Subset picture shows the far-field output mode profile in a $80\mathrm{\mu m}\times 80\mathrm{\mu m}$ image.

Download Full Size | PDF

7. CO-DOPED WAVEGUIDES

As it is our goal to create efficient Ho-based Mid-IR waveguide lasers, the next step is to leverage existing fiber coupled diode sources to pump the waveguide lasers. Unfortunately, 1908 nm diodes are not a mature technology, forcing the use of either a cascaded pumping scheme or an additional dopant for energy transfer. This can be realized by pumping the ${\mathrm{Yb}}^{3+}$ ion with a 940 nm diode laser, which then non-resonantly transfers to the ${\mathrm{Ho}}^{3+}$ ion [24,25]. The basic energy level diagram for this pumping configuration can be seen in Fig. 9. The ${\mathrm{Yb}}^{3+}$ acts as a sensitizer for the ${\mathrm{Ho}}^{3+}$ ions by absorbing the 940 nm radiation on the ${}^2{\mathrm{F}}_{7/2}\to {}^2{\mathrm{F}}_{5/2}$ transition. The energy then transfers to the ${\mathrm{Ho}}^{3+}$ ion via non-resonant energy transfer. Finally, through spontaneous decay from the ${}^5{\mathrm{I}}_6\to {}^5{\mathrm{I}}_7$, the upper laser level of the 2 μm transition is pumped.

 

Fig. 9. ${\mathrm{Yb}}^{3+}/{\mathrm{Ho}}^{3+}$ energy level diagram illustrating the pumping scheme of the co-doped material.

Download Full Size | PDF

To this end, a sample of Yb:Ho:YAG was doped with 2 at. % Ho and 15 at. % Yb and inscribed with a variety of waveguides for testing. The parameters used for inscribing waveguides into this sample were identical to those used for the multimode waveguides. The repetition rate of the laser was set to 100 kHz in an effort to decrease the thermal load on the sample. However, due to the presence of ${\mathrm{Yb}}^{3+}$ ions, the sample was prone to cracking during inscription due to linear absorption of the 1045 nm inscription laser. Figure 10 represents the best waveguides inscribed in the co-doped sample. The crack formation propagates out from the laser induced damage region, thus allowing light propagation in the waveguide core. Many of the waveguides inscribed in the co-doped sample formed cracks propagating inward from the region modified by the laser. The presence of cracking in the core of the waveguide significantly increased losses and prevented guiding in most of the waveguides that were tested.

 

Fig. 10. Outward radial cracking of waveguides inscribed in the Yb:Ho:YAG sample.

Download Full Size | PDF

To initially test the viability of the co-doped waveguides, we first in-band pumped the sample with a Tm fiber laser operating at 1908 nm using the setup shown in Fig. 2A. The optimum performance obtained from the 80 μm waveguide can be seen in Fig. 11. The waveguide had a maximum output power of 325 mW with a slope efficiency of 9% using a 70% reflective outcoupler. The spectral output of the waveguide was also measured and closely followed that of the multimode waveguides. The 50% and 70% reflective outcouplers had an operating wavelength of 2090 nm. The 90% reflective outcoupler shifted the operating wavelength to 2120 nm.

 

Fig. 11. Output power for in-band pumping of the Yb:Ho:YAG waveguide.

Download Full Size | PDF

Transitioning to diode pumping of the Yb:Ho:YAG sample required changing many of the cavity optics described in Section 3. However, the basic overall geometry remained the same as seen in Fig. 12. In Fig. 12, M1 was a 45 deg incoupler that was AR coated for 940 nm and HR for 2 μm, M2 was a variable percentage outcoupler, M3 was an HR mirror for 2 μm, and L1 and L2 were uncoated ${\mathrm{CaF}}_2$ lenses with a focal length of 5 cm. The co-doped sample was mounted on a water cooled heatsink and kept at a temperature of 15°C.

 

Fig. 12. Cavity configuration used for 940 nm pumping of the Yb:Ho:YAG waveguide sample.

Download Full Size | PDF

Utilizing a DILAS 40 W, 940 nm pump diode and a 200 μm diameter, 0.22 NA multimode fiber, several attempts were made to produce output from the co-doped sample. However, no lasing action was observed. While the sample exhibited strong 2 μm fluorescence and exhibited excellent guiding properties, the crystal cracked under the stress of the thermal load, destroying all available waveguides. Future research will focus on optimization of the ULI parameters to minimize the stress inside of the co-doped material.

8. CONCLUSION

In conclusion, we have demonstrated single-mode and multimode waveguides in Ho:YAG using two different inscription recipes. The single-mode waveguides were created in a 0.5 at. % doped sample with a core diameter of 50 μm. Single-mode operation of the waveguide laser produced a maximum output of 1.77 W with a slope efficiency of 16%. The multimode waveguides were also inscribed in a 0.5 at. % doped sample, but had a larger core diameter of 80 μm. The multimode waveguides produced a maximum output power of 1.89 W with a slope efficiency of 29%. Additionally, a co-doped sample of Ho:Yb:YAG was inscribed with waveguides. The waveguides produced lasing action at 2   μm using in-band pumping, albeit with decreased efficiency compared to the single-mode and multimode waveguides. Direct diode pumping of the sample at 940 nm showed strong guiding and fluorescence at 2 μm, but no lasing was observed.