1. Introduction

The development of crystalline Raman lasers has been a subject of great interest over the last decade [1,2]. Raman conversion of Nd-based lasers gives access to wavelengths ranging from 1.1μm to 1.5μm. Recently, diamond has been introduced as a Raman laser medium [3–7]. Its properties – a large Raman gain coefficient (75cm.GW⁻¹ at λ = 532nm) [4,8,9], large Raman shift (1332cm⁻¹) [10], incomparable thermal conductivity (2000W/m/K) [11] and transparency between λ = 230nm and λ = 2500nm [12,13] – make diamond an attractive Raman laser material. The high thermal conductivity – two to three orders of magnitude higher than conventional Raman laser materials – should enable significant power scaling of Raman lasers; combined with the large Raman gain coefficient, it will also permit the use of shorter samples – a few millimeters rather than the few centimeters typical of conventional materials – leading to more compact devices.

To our knowledge, the only report to date of an intra-cavity pumped diamond Raman laser is from Demidovich et al. [3]. In this work, pulse energies of 450nJ were obtained at the Raman laser wavelength within a microchip laser although full published details are lacking. (We have previously reported a CW intra-cavity-pumped diamond Raman laser, but the output powers achieved (a few milliwatts) were trivial [6]).

The use of natural diamond is, however, limited by the restricted availability of large diamond pieces and by the variability of the optical properties [14]. The availability of high optical quality diamond has greatly improved recently with the development of synthetic single-crystal diamond [15,16]. Mildren et al. presented the use of chemical vapour deposition (CVD) grown synthetic single-crystal diamond in an external-cavity Raman laser configuration resulting in a conversion efficiency of 63.5% when pumped with mJ pulses at 532nm [5], competitive with conventional Raman laser materials.

In this paper, we report the development of a diamond Raman laser in an intra-cavity configuration (i.e. where both the fundamental and Raman lasers share the same cavity) utilising high quality CVD-grown single-crystal material. This approach is the most applicable to CW Raman lasers – because the threshold is considerably reduced compared to an external cavity Raman laser – and the demonstration of a pulsed diamond Raman laser described in this paper presents the initial steps towards the development of an efficient CW diamond Raman lasers. However, it is an approach that requires high optical quality diamond: both low birefringence and low optical losses are needed. The recent development of low birefringence (Δn<10⁻⁶) synthetic diamond opens up the potential for intra-cavity applications in lasers [17].

In the next section, the diamond sample used in this experiment will be described. Then, a description of the diamond Raman laser cavity is provided in section 3. Finally, a discussion of the experimental results is given in section 4.

2. Synthetic single-crystal diamond for Raman laser applications

The Raman crystal used in this experiment was a synthetic single-crystal diamond with dimensions of a = 3mm x b = 2mm x c = 3.3mm from Element 6 Ltd. The faces of the CVD-grown cuboid are all {100} planes. The front and back faces (ab) were dielectrically coated ensuring minimal losses at both λ = 1064nm and λ = 1240nm (R~0.1%). The crystal was grown so that the birefringence along the optical path was minimized (Δn<5.10⁻⁷), albeit still about 5 times higher than the birefringence of the synthetic diamond used in [5]. In order to measure the depolarization losses due to the diamond, a fused silica plate (thickness = 5mm) was inserted at Brewster’s angle in a 3-mirror laser cavity built around an end-pumped 5mm long Nd:YVO₄ rod. From this, the round-trip depolarization losses due to the diamond were estimated to be 0.12%. To measure the losses in the diamond sample, the optical losses of the Nd:YVO₄ laser cavity, now without the fused silica window, were measured using a Caird analysis [18] with and without the diamond. In this way, the round-trip losses attributable to the diamond were estimated to be 2% at the laser wavelength of 1064nm. During this experiment, the reflection from the front and back surfaces were negligible and no scattered light was observed suggesting that absorption was mainly responsible for the loss. Therefore, an absorption coefficient of 0.03cm⁻¹ was estimated. This value is almost three times larger than that for the synthetic diamond used in [5]. The absorption value measured here is also at the higher end of results reported for synthetic single-crystal diamond in Ref [15]. where the loss coefficients range from 0.003 to 0.07cm⁻¹. Although the material reported in [15] had relatively high birefringence, as the growth technology develops, it should soon be possible to source low birefringence material with losses towards the 0.003cm⁻¹ level.

3. An intra-cavity Raman laser resonator

The Raman laser cavity, shown in Fig. 1 was built around an end-pumped 5mm long a-cut Nd:YVO₄ rod (doping concentration = 1 atm.%). The laser rod faces were anti-reflection (AR) coated at λ = 1064nm and 1240nm (R~0.1%). The rod was oriented so that the fundamental laser output was horizontally polarized. The pump light was delivered using a 35W fibre-coupled (core diameter = 100μm, NA = 0.22) diode laser emitting at 808nm. The pump beam radius at the laser rod was measured to be 150μm. An acousto-optic Q-switch was used to obtain short pulses and its faces were AR coated for both 1064 and 1240nm (R<0.2%). The diamond was placed at an intra-cavity beam waist where the calculated fundamental cavity mode radius was ~45μm (Fig. 1). All cavity mirrors bar the output coupler were coated to be highly-reflective (HR) for both wavelengths (R>99.9%). The coating of one of the curved mirrors had a transmission of about 5% for λ = 1.18μm. This introduced sufficient loss at this wavelength to prevent self-Raman oscillation in the short Nd:YVO₄ crystal. The transmission of the output coupler at λ = 1240nm was 5% (R>99.9% at λ = 1064nm).

 

Fig. 1 Schematic of the intra-cavity diamond Raman laser.

Download Full Size | PDF

4. Results

In the first instance, the diamond was oriented so that the polarization of the fundamental beam was parallel to a <100>direction in the diamond. As expected from the third order susceptibility tensor for the crystal class of diamond [19], the polarization of the first-order Raman laser output was perpendicular to the fundamental laser polarization. Oscillation of the 1st and 2nd Stokes components (corresponding to λ = 1240nm and 1485nm, respectively) was observed (Fig. 2 ). The maximum average output power at λ = 1240nm was measured to be 155mW for 9W of diode pump power. The resultant energy of the 1240nm pulses was 25μJ. About 25mW of residual output power at 1064nm and 30mW of second Stokes component at λ = 1485nm were also observed. The pulse durations of the 1240nm and 1064nm components were 80ns and 100ns, respectively (Fig. 3 ). The beam propagation factors were M² _x = 4.1, M² _y = 3.4 for the 1st Stokes component (for the horizontal and vertical planes, respectively) and M² _x = 1.6, M² _y = 1.5 for the fundamental. The thresholds for the fundamental laser and the first and second Stokes components were 0.5W, 1.2W and 2W of absorbed diode laser pump power, respectively. Using an optical spectrum analyzer of 10pm spectral resolution, the full-widths at half maximum of the spectra were measured to be 0.18nm, 0.22nm and 0.34nm for the λ = 1064nm, λ = 1240nm and λ = 1485nm components, respectively.

 

Fig. 2 Optical spectrum of the diamond Raman laser output showing the fundamental, first and second Stokes components.

Download Full Size | PDF

 

Fig. 3 Pulse profiles of the (a) fundamental laser at λ = 1064nm and (b) Raman laser at λ = 1240nm recorded using a 100MHz bandwidth oscilloscope.

Download Full Size | PDF

The diamond was then rotated around the cavity axis so that the fundamental polarization was parallel to a <110> direction in the crystal. For this orientation, the polarization of the first Stokes component was parallel to the fundamental (i.e. horizontally polarized) as expected. The maximum average output power at λ = 1240nm was measured to be 230mW for 8W of absorbed pump power (Fig. 4 ). The pulse durations for the fundamental and the first Stokes components were 60ns and 50ns, respectively at a repetition rate of 6.3kHz. The resultant pulse energy at λ = 1240nm was therefore 37μJ. The beam propagating factors for the fundamental and the first Stokes were measured to be M² _x = 1.2, M² _y = 1.6 and M² _x = 3.3, M² _y = 1.8, respectively. Thermally-induced distortions within the Nd:YVO₄ rod, resulting in the degradation of the transverse mode, limited the output power above 9W of absorbed pump power. The conversion efficiency from the absorbed diode laser to the 1st Stokes component output was about 3%.

 

Fig. 4 Average output power of the Raman laser as a function of the absorbed diode laser pump power.

Download Full Size | PDF

The laser cavity was then modified in order to provide more output coupling at λ = 1240nm. Given that output couplers with larger transmission were not available, two output couplers were used. A T = 5% output coupler was used to fold the resonator and a T = 1% output coupler was used as an end-mirror (see Fig. 5 ). The total coupling for the laser cavity was therefore T = 11%. The total average output power at λ = 1240nm was 375mW for 9.5W of absorbed pump power giving an absorbed diode laser to the Raman laser conversion efficiency of 4%. The pulse durations at λ = 1064nm and λ = 1240nm were 80ns and 45ns, respectively, while the beam propagating factors were M² _x = 1.6 M² _y = 1.8 and M² _x = 2.6, M² _y = 2.1 for the fundamental and 1st Stokes component, respectively. The pulse energy at λ = 1240nm was 60μJ. Again, thermal lensing within the Nd:YVO₄ rod limited the power that could be obtained.

 

Fig. 5 Schematic of the intra-cavity diamond Raman laser featuring a combination of output couplers.

Download Full Size | PDF

Intra-cavity Raman lasers are particularly sensitive to intra-cavity loss [1]. The intra-cavity optical losses at both resonant wavelengths (approximately 4.5% per round-trip with 2% attributable to the diamond was measured at 1064nm) seriously hampered the performance of the laser reported here. Future effort to improve the efficiency of these lasers will focus on reducing these losses. It is possible to fabricate synthetic single-crystal diamond with significantly lower losses [15] and, as the growth technology matures, it should be possible to source material combining low birefringence and low loss.

In Raman lasers, so-called ‘beam clean-up’ usually leads to an improvement in the beam quality of the Stokes output over the fundamental [1,20,21]. In the experiments reported here, the M² of the Stokes output was significantly larger than that of the fundamental. The reasons for this are not fully understood but may relate to the small (sub-optimal) output coupling and the tight focusing in the Raman laser material. This would lead to a relatively low threshold for Raman laser oscillation on high order transverse modes, exaggerating the effects of any mismatch between the Raman and the fundamental laser modes. This is consistent with the fact that the value of the M² dropped with output coupling. The larger Raman laser fundamental mode within the laser gain medium (10% greater than that for the 1064nm laser) may also have sampled greater aberration from the thermal lens in the Nd:YVO₄. Future efforts to improve the beam quality will focus on improving the mode overlap in the diamond and reducing the thermal lens strength in the Nd:YVO₄.

5. Conclusion

For the first time to our knowledge, low-birefringence single-crystal CVD diamond has been used as a Raman laser medium in an intra-cavity configuration. The Raman laser output power and conversion efficiency were mainly limited by the thermal aberrations taking place within the conventional laser gain medium and the relatively high absorption within the diamond (1% per pass). Using a laser with an 11% of output coupling, average powers of 375mW were observed resulting in an absorbed diode-laser pump power to the Raman laser output power conversion efficiency of 4%. This demonstration is a first step towards the development of an efficient CW diamond Raman lasers which will, in turn, open up the potential for considerable power scaling of CW Raman lasers. The remaining challenge is the reduction of the bulk losses within the diamond. Further investigations are under way to identify synthetic diamond material combining the low loss that has been reported elsewhere for such material [15] with the low birefringence properties required for Raman lasers.