1. Introduction

Simultaneous stimulated Raman scattering (SRS) in a Raman-active crystal and sum-frequency generation (SFG) in a nonlinear crystal within the optical cavity of a Nd³⁺:crystalline laser has proved to be a very flexible and effective means of realizing efficient all-solid-state CW green-yellow laser sources [1–5]. For example, CW powers >5 W in the lime green (559 nm) and yellow (586 nm) have been demonstrated with good beam quality and diode-visible optical conversion efficiencies around 20% for Nd:GdVO₄ self-Raman lasers with intracavity SFG in LBO [6,7]. Multi-Watt visible CW sources have also been demonstrated using different intracavity combinations of separate Nd³⁺ laser, Raman-active (e.g. BaWO₄, SrMoO₄, KGd(WO₄)₂) and nonlinear crystals [8–10].

We have recently expanded studies of CW crystalline Raman lasers with intracavity SFG to the low-power, miniature regime, where diode pump powers of only 1-2 W might be used to generate CW green-yellow outputs ~100 mW [11,12]. Simple, efficient, miniature wavelength-versatile sources of this type are expected to find application in areas like biomedical diagnostics and various forms of laser writing/display. However achieving high conversion efficiencies for such low diode pump powers is very challenging due to the high circulating optical intensities required to reach threshold for simultaneous SRS and SFG, and the potentially high insertion losses for multiple crystals within the laser resonator. Despite this, we have achieved CW output powers >400 mW at 559 nm for SFG of the fundamental (1064 nm) and first-Stokes (1176 nm) wavelengths of a Nd:YVO₄ self-Raman laser, and 200 mW for Second Harmonic Generation (SHG) of the first-Stokes using intracavity BBO, for diode pump power only 3.8 W. Corresponding diode-visible conversion efficiencies are 14% and 7% respectively [12].

Self-Raman laser crystals, in which a single crystal such as Nd:YVO₄ or Nd:KGW performs the dual functions of generating the fundamental laser field and the Stokes field, provide for low insertion loss and short resonators (which offsets the effects of thermal lensing in the crystals) but due to the limited number of practical self-Raman laser crystal materials, only a limited number of visible output wavelengths can be generated. In principle, using different combinations of separate intracavity laser crystals and Raman crystals with intracavity SFG/SHG allows a much wider spectrum of output wavelengths to be generated. Separating the laser crystal from the Raman crystal also has the advantage that waste heating in the laser crystal is separated from Raman heating, enabling improved thermal management. However in the low-power miniature regime, the additional insertion losses and longer resonators that result from using separate laser and Raman crystals can lead to significantly higher laser thresholds and lower overall output powers. In this paper we report an efficient miniature CW visible source using 3 separate intracavity crystals: a Nd:YVO₄ crystal as the laser medium, a BaWO₄ crystal as the Raman medium, and an LBO crystal for sum-frequency or second-harmonic generation. The high Raman gain (8.5 cm/GW [13]) of BaWO₄ offsets the higher insertion loss of the additional crystal.

Although these miniature Raman lasers involve comparatively low powers, the intracavity intensities are high, and the effects of competition between SRS and SFG/SHG within the resonator can significantly reduce output power at the desired wavelength [12, 14]. Another effect of the intracavity SFG/SHG is to enhance competition between different Stokes lines, which may arise from the two Raman-active materials (YVO₄ and BaWO₄) in the resonator and/or from the two significant Raman transitions in BaWO₄ itself. Such competition effects have been observed previously in both low and high power systems, but not studied in detail. For example, competition between Raman transitions in Nd:GdVO₄ and SrMoO₄ was noted in [9], while competition between the Raman transitions in BaWO₄ was noted in [8]. Competition between Raman transitions in Nd:GdVO₄ have also been observed in [15]. It is clear that competition of this sort has a detrimental effect on laser performance, and understanding how to mitigate competition will be of importance to a wide range of Raman laser designs. Here we consider how and why this type of competition arises and present a parameter analysis of the key factors determining the phenomenon. We demonstrate how to avoid Stokes competition by controlling the crystal dimensions and orientation to obtain stable CW yellow output at 590 nm (SHG of the 1180 nm first-Stokes line of BaWO₄) with powers ~200 mW at diode-visible conversion efficiency of 5% (diode pump-power 3.8 W).

Intracavity Raman lasers with intracavity frequency conversion are particularly complex systems because of the interplay of the intracavity fields through both the ${\chi }^{(2)}$and ${\chi }^{(3)}$non-linearities. Without a ${\chi }^{(2)}$(SFG) interaction, an ideal Raman laser operating at the first Stokes wavelength, once above threshold will hold the fundamental field at its intensity at that threshold, analogous to a conventional laser holding the laser gain at its threshold value. It follows that once the Stokes field with highest gain reaches laser threshold, any weaker Stokes lines cannot then reach their thresholds.

In reality, the situation is typically more complex. While some CW Raman lasers show definite clamping behaviour [16, 17], others do not [18–20], and substantial spectral broadening of the fundamental field is one mechanism that has been invoked to explain this [19, 20]. Other plausible mechanisms include the presence of multiple transverse modes and power-dependent mode sizes due to strong thermal lensing.

However in the case of a Raman laser with intracavity SHG, clamping is not predicted: rather the ${\chi }^{(2)}$ interaction provides a non-linear output coupling for the desired Stokes field that increases as the laser is pumped harder, and the fundamental field strength rises to maintain steady state. This increase in the fundamental field strength then allows weaker Stokes lines at wavelengths that experience less or no ${\chi }^{(2)}$loss to progressively reach threshold at higher pump powers. For this reason, competition between Stokes lines arising from different Raman transitions is much more problematic in Raman lasers which include intracavity SFG/SHG. Once the weaker Stokes lines reach threshold, they then may clamp the fundamental field strength, so that higher pump power does not lead to higher output power at the target visible wavelength, but increases the power in the unwanted Stokes lines.

2. Experimental setup

Figure 1 shows a schematic of the pump geometry and cavity layout of our experimental laser system. We used 1at.%, a-cut Nd:YVO₄ crystals with dimensions of either 4 × 4 × 3 mm or 4 × 4 × 1 mm as the laser medium. The high emission cross-section of Nd:YVO₄ (14.4 × 10⁻¹⁹ cm²) [21] enables low pump-power threshold and generally high efficiency for laser operation on the 1064 nm fundamental wavelength at modest pump powers. The laser crystals, anti-reflection (AR) coated for 808 nm and 1064 - 1200 nm (Castech Inc), were mounted in a water-cooled copper heat-sink held at a temperature of 20°C.

 

Fig. 1 Laser setup.

Download Full Size | PDF

A 4 × 4 × 3 mm a-cut BaWO₄ (BW) crystal, grown by the Institute of Crystal Materials, Shandong University, was used as the Raman gain medium for the experiments. By means of its strongest Raman transition at 925 cm⁻¹, BaWO₄ can shift the 1064 nm fundamental wavelength to the first Stokes wavelength of 1180 nm. The BaWO₄ crystals were also AR coated for 808 nm and 1064 - 1200 nm, and were maintained at a temperature of 20°C.

A non-critically phase matched (NCPM, θ = 90° φ = 0°) LBO crystal with dimensions of 4 × 4 × 5 mm was used for intracavity frequency-mixing. The crystal was AR coated (R = 0.45% @ 1064 nm, R = 0.1% @ 1176 nm) and housed in a copper block incorporating a resistive heater; the temperature of this mount was monitored using a PT100 sensor. For the present experiments the LBO crystal was maintained at 40 °C to phase-match for SHG of 1180 nm to 590 nm.

The pump source used in this work was a high brightness, free space diode laser (Unique Mode UM4200-M20-CB-TEC) giving maximum output power of 3.8 W at 808 nm. The direction of polarization of this pump light was parallel to the c-axis of the Nd:YVO₄ crystal to take advantage of higher 808 nm absorption (48.4 cm⁻¹) compared to the a-axis (14.2 cm⁻¹) [21]. We used a telescope arrangement for expanding and collimating the diode output and then focused it into the laser crystal through a 50 mm focal-length lens. The diameter of pump beam waist in the laser crystal was estimated to be 160 µm. Most of the pump light was absorbed in the laser crystal: 98.7% for the 3 mm long crystal and 91.3% for the 1 mm long crystal.

We employed a compact, linear resonator configuration formed by mirror M1 and M2, as depicted in Fig. 1. Input mirror M1 was a flat mirror coated reflectivity R>99.994% for 1064 nm and 1176-1180 nm, transmissivity T = 98% for 808 nm. Output coupler M2 was a concave mirror with 50 mm radius of curvature, coated as R >99.994% for 1064 nm and 1176-1180 nm, T = 98.9% for 588-590 nm. The resonator length was around 18 mm.

3. Preliminary experiments using 3 mm-long Nd:YVO₄ and 3 mm-long BaWO₄

Initial experiments used a combination of 3 mm-long Nd:YVO₄ and 3 mm-long BaWO₄ oriented with the c-axis parallel to the plane of polarisation of the fundamental (1064 nm) laser field to take advantage of the highest Raman gain (see middle plot of Fig. 2(a) ). This laser exhibited highly unstable yellow output, with large fluctuations in output power as shown in Fig. 2(b). Large fluctuations in output power were consistently observed, and the yellow output, measured using a fibre coupled optical spectrum analyser with 0.05 nm resolution (Ocean Optics USB HR4000) and shown in Fig. 2(c), was found to not be spectrally pure.

 

Fig. 2 (a) Spontaneous Raman spectra for the cases of laser excitation polarised along the a- and c-axes of BaWO₄ and the c-axis of YVO₄. (b) Laser output powers; and (c) spectral characteristics of visible output and (d) infrared fields (3.8 W incident pump) of the laser formed using 3 mm long Nd:YVO₄ and 3 mm long BaWO₄ crystals.

Download Full Size | PDF

The source of these problems was quickly traced to unwanted Stokes lines at 1176 nm (corresponding to the 890 cm⁻¹ shift in Nd:YVO₄) and 1103 nm (corresponding to the 332 cm⁻¹ shift in BaWO₄), which competed with the desired Stokes line at 1180 nm; all of these shifts are indicated in Fig. 2(a). As indicated in Fig. 2(b), the 1176 nm line was observed for pump powers above 1.6 W, while the 1103 nm line was observed for higher powers, above 2.4 W. Figure 2(d) shows a snapshot of these Stokes lines, observed through the output coupler using an infrared fibre coupled optical spectrum analyser with 0.05 nm resolution. The visible output spectrum shown in Fig. 2(c) was observed for a pump power of 3.8 W. As well as strong 590 nm output arising from SHG of the 1180 nm first-Stokes line of BaWO₄, both SHG of the first-Stokes of YVO₄ at 1176 nm to 588 nm and SFG of the two Stokes lines to 589 nm are also clearly evident. The acceptance bandwidth of the 5 mm long LBO crystal was 44 nm at 1180 nm [22] and was sufficient to phase-match these interactions simultaneously. Note the spectra shown in Fig. 2(c) and 2(d) are indicative only, since the distribution of energy between the various fields was continuously changing.

4. Analysis of Raman-gain competition and strategies for overcoming it

4.1 Competition between Nd:YVO₄ and BaWO₄

A useful parameter for considering competition is ${\kappa }_R=g_R{l}_R/A_R$, which characterises the strength of the Raman coupling for a given Stokes line [23], $g_R$ is the Raman gain for the line, $A_R$is the area of the mode in the associated Raman crystal of length $l_R$. We can use these coupling parameters to find the ratio of the strengths of the Raman interaction from the strongest shifts in the BaWO₄ and Nd:YVO₄ crystals as follows.

We investigated laser performance using this combination of 1 mm-long Nd:YVO₄ and 3 mm-long BaWO₄, and verified experimentally that no infrared signal at 1176 nm was observed when using this combination of crystal lengths. Figure 3(a) shows the performance of this laser. We obtained stable Stokes operation at 1180 nm without 1176 nm competition and pure output at 590 nm over the full pump range. Up to 146 mW yellow output was observed for 3.8 W pump power with a low threshold pump power of 0.4 W. Note that although a yellow beam was generated in both directions, only the light propagating through the output coupler was measured in this work; the beam propagating into the laser crystal was mostly absorbed, but could in principle be extracted using an intracavity mirror. We estimated the circulating intensity for the 1064 nm fundamental wavelength to be the order of 6.77 MW/cm², and for the 1180 nm Stokes to be of order 0.4 MW/cm^2, for diode-pump powers of 3.8 W, using the measured power of 1064 nm and 1180 nm leaked through output mirror. Because of the small mode sizes in this laser the circulating intensities are comparable to those of multi-Watt Raman lasers despite the much lower pump powers used here.

 

Fig. 3 (a) Yellow laser performance with crystal combination of 1 mm-long Nd:YVO₄ and 3 mm-long BaWO₄ (oriented so the fundamental laser is polarized along its c-axis); and (b) residual fundamental-field and Stokes power collected through the output coupler.

Download Full Size | PDF

The BaWO₄ crystal was oriented in the same way as the experiment with the two 3 mm long crystals, i.e. with the fundamental laser field polarized along its c-axis, to access the highest Raman gain coefficient for the 925 cm⁻¹ Raman shift. However, while the 1176 nm line from the YVO₄ crystal was completely suppressed, emission at 1103 nm, corresponding to Stokes-shifting of the fundamental wavelength by the BaWO₄ 332 cm⁻¹ Raman peak was observed for diode pump powers above 2.4 W. Because of competition between this line and the desired 1180 nm Stokes line, there was relatively little increase in 590 nm output power above 2.4 W pump power. Figure 3(b) shows how the intracavity fundamental field, monitored from leakage through mirror M2, increases with pump power. As expected, owing to the increasing output coupling (through SHG) presented to the 1180 nm field as it intensifies, the fundamental field intensity eventually becomes high enough to enable the 1103 nm line to oscillate. Because the 1103 nm line is not phase-matched in the LBO crystal to generate visible output, the intensity of the fundamental field clamps, and the strength of the 1180 nm line and its second harmonic no longer increases, the power instead being fed into the 1103 nm line.

4.2 Competition between Raman shifts in BaWO₄

By examining the spontaneous Raman spectra shown in Fig. 2(a) and scaling to the known Raman gain for the 925 cm⁻¹ shift, we can determine the Raman gain coefficients for the 332 cm⁻¹ shift to be 3.1 cm/GW for E║c and 0.92 cm/GW for E║a. In this case, the mode-sizes and crystal lengths are the same, and so the ratio of the Raman coupling strengths for the 925 cm⁻¹ and the 332 cm⁻¹ shifts is simply determined by their respective gains. Accordingly we can calculate Raman coupling ratios ${\kappa }_{R-BW}^{925}/{\kappa }_{R-BW}^{332}$ to be 2.7 in the case where the fundamental-field is polarised parallel to the c-axis of BaWO₄, and 7.8 in the case where the fundamental-field is polarised parallel to the a-axis of the BaWO₄. We can therefore predict that competition between the 332 cm⁻¹ and 925 cm⁻¹ shifts is far less likely to occur for the latter case.

We note that comparing the peak values of the spontaneous scattering lines compares their steady-state Raman gain coefficients, as used throughout this paper. This is most appropriate for determining which Raman transition will reach threshold first since the linewidth of the fundamental field close to Raman threshold is somewhat narrower than the width of all the Raman transitions under consideration. For higher pump powers, the linewidth of the fundamental field does increase to be of the same order as the Raman linewidths; this will favour Raman transitions with broader linewidths (shorter dephasing times) such as the 332 cm⁻¹ line, effectively altering the kappa ratios. This can further promote the onset of competition at higher pump powers [17, 18].

In our experiments, we re-oriented the BaWO₄ crystal such that the crystals a-axis was parallel to the polarisation of the fundamental field, and found there was indeed no emission at 1103 nm. Figure 4 shows the performance of the laser at 590 nm with all competition eliminated. The maximum output power at 590 nm was 194 mW, achieved for 3.8 W pump power, and the corresponding optical (diode to yellow) conversion efficiency was 5.1%. The yellow output was spectrally pure, and as seen in the inset, the only Stokes field present was at 1180 nm. For comparison, the power performance for the laser with polarisation along the c-axis of BaWO₄ is also reproduced in Fig. 4, where it can be seen that for diode pump powers up to 2.5 W, slightly higher powers were obtained when the fundamental-field polarisation was aligned along the c-axis of BaWO₄, this being consistent with the slightly higher Raman coupling for this orientation. Above this power level however, it is clearly preferable to have the fundamental field polarisation along the a-axis of BaWO₄ in order to avoid competition.

 

Fig. 4 Yellow laser performance with crystal combination of 1 mm-long Nd:YVO₄ and 3 mm-long BaWO₄, oriented with the fundamental field polarized along its a-axis (circles), and its c-axis (umbrellas). Shown inset is a spectrum of the IR wavelengths leaking from the output coupler at maximum pump power.

Download Full Size | PDF

5. Conclusion

In conclusion, we have presented a study of competition between different Stokes lines in different crystals (Nd:YVO₄ and BaWO₄), and between different lines in the same crystal (BaWO₄). While we initially observed poor output power at the desired visible wavelength, we determined that this was due to the onset of undesirable Stokes lines within the Nd:YVO₄ host and BaWO₄ Raman crystals. By analysing the ratio of Raman coupling parameters (κ) we could estimate the likely competition between lines, and we suppressed the competition between 1176 nm and 1180 nm by selecting the lengths of the Nd:YVO₄ and BaWO₄ crystals appropriately.

We also presented an analysis of the generation of 1103 nm and demonstrated its successful suppression by orienting the polarization of the fundamental field along the a-axis of BaWO₄, despite it having weaker Raman gain at the desired line. These optimisations led to the stable operation of an efficient, miniature yellow Raman laser with a diode pump threshold of just 0.4 W. Output power of 194 mW at 590 nm was achieved from an 18 mm long resonator pumped using 3.8 W. The results demonstrate that with judicious choice of crystal length and orientation to overcome SRS competition effects, CW Raman lasers making use of separate gain and Raman crystals and pumped with low power are capable of producing visible output powers approaching that of self-Raman configurations. This is significant as the added wavelength versatility enabled by using two crystals opens avenues for compact, high efficiency Raman lasers in the field of biomedicine.