1. Introduction

Solid-state Raman lasers have been established as efficient coherent radiation sources in spectral regions inaccessible to inversion lasers, from ultraviolet to mid-infrared regions [1–3]. Owing to the relatively low nonlinear gain of stimulated Raman scattering (SRS) process, the thresholds of external cavity continuous-wave (cw) Raman lasers are usually at least several watts even with very low output coupling, therefore making the laser quite inefficient for low- and moderate-power applications. The intracavity pump scheme is often adopted to circumvent this problem. By locating the Raman crystal within the high-Q fundamental laser cavity with minimized losses, the circulating fundamental power could achieve cw SRS threshold (usually hundreds of watts) with primary diode pump power as low as several hundred milliwatts [4,5]. Several watts of cw Stokes and frequency-doubled yellow have been demonstrated with compact and robust intracavity Raman lasers (including the self-Raman lasers), with good optical efficiencies of up to 20% and the 17% [2,6].

Intracavity Raman lasers exhibit some complex dynamics, and it has been shown in recent years that their efficiency can be impacted by broadening of the fundamental linewidth, as the SRS presents a spectrally-varying loss to the fundamental laser [6–9]. The broadened fundamental laser linewidth results in wider Stokes spectrum which could be inconvenient for some practical applications, and, more importantly, it was shown in [9] that spectral broadening of the fundamental can reduce the effective Raman gain coefficient by over 70% of its narrow linewidth steady state value. The Stokes power and conversion efficiency of intracavity Raman lasers can be enhanced by narrowing the fundamental linewidth by the use of one or multi etalon(s) [9–11]. However there has been a less emphasis given to the potential for wavelength tuning the output of crystalline Raman lasers.

Much of the work on intracavity Raman lasers has been focused on demonstrating output at yellow wavelengths suitable for ophthalmology, and around the 1.5 μm eye-safe region. However there are other applications, such as laser guide-star and for spectroscopy, which require sources having narrow-linewidth laser and are capable of being tuned to match transition lines accurately [12,13]. The spectral coverage of crystalline Raman laser wavelengths is usually described as a set of discrete lines, determined by the choice of laser and Raman crystals. Despite there being numerous crystals to choose between, and the fact that some crystals offer multiple viable laser or Raman transitions, the spectral coverage of these sources is far from complete, and the output would not usually be considered to be tunable. Hence, when targeting specific laser wavelengths for certain applications, researchers have to choose a fixed combination of laser and Raman gain mediums to achieve the emission line needed. For example, for the 589.16 nm guide star wavelength, the 1062 nm Nd:GGG fundamental transition and the BaWO₄ Raman shift of 925 cm⁻¹ [14], or 1064 nm Nd:YVO₄ fundamental transition and the CaWO₄ Raman shift of 910 cm⁻¹ may be used [12]. Having the choice of laser and Raman crystals dictated in this way, may however, compromise the laser performance in regards to other aspects such as thermal load management. In the case of this wavelength, power has been limited to the watt level and only pulsed operation.

The desire to realize wavelength tuning of Raman lasers has been one driver for the research group at Strathclyde University, who have harnessed optically-pumped semiconductors as a broadband gain media in their tunable, multi-Watt cw intracavity diamond Raman laser. Their work, published in [8] reported tunable output from 1217 nm to 1244 nm with the semiconductor disk fundamental laser being tuned by a birefringent filter [15]. Later the same group extended the wavelength tuning range to the red and the 1.4 μm by frequency-doubling the Stokes and using different semiconductor [8,11]. However, the optical efficiencies of optically pumped semiconductor fundamental lasers are usually not as good as those utilizing crystalline gain media, which may limit the efficiency of the Raman laser. Moreover, the availability of semiconductor disk gain media is also limited. Lin et al. also demonstrated watt-level tunable output in lime-yellow-orange based on semiconductor disk fundamental and KGW Raman shift [16].

Past attempts to tune crystalline Raman lasers have been limited. In 2014, Krainak et al. demonstrated an actively Q-switched c-cut Nd:YVO₄ self-Raman laser at 589 nm and discussed the possibility of tuning the yellow output to the guide star wavelength by means of injection seeding while utilizing the relatively wide emission linewidth of the c-cut vanadate [17]. Researchers have also tried some other efforts of wavelength tuning. Liu et al. tuned the Stokes wavelength of a Nd:GGG-BaWO₄ intracavity Raman laser and realized single-longitudinal-mode operation by using multiple etalons in the Stokes cavity. However, the tuning range is limited to less than 0.1 nm by the narrow Raman gain linewidth, and the Stokes power is only tens of milliwatts due to the insertion losses of etalons [18]. Stokes wavelength tuning range of ~0.5 nm has also been demonstrated by changing the temperature of YVO₄ Raman crystal within 140 °C, but the temperature-dependent Raman gain decreased the power performance seriously when heating the Raman crystal [19].

In this paper, we investigate the tunability of crystalline Raman lasers by means of solid etalons, finding that careful selection of etalons and careful resonator design can give rise to all of the following: increased output power and efficiency, substantial narrowing of the Stokes output and its second harmonic, and tunability over several nm, as required to target particular spectroscopic transitions. Using the mature a-cut Nd:GdVO₄ as fundamental laser crystal and a-cut BaWO₄ as Raman medium, we demonstrated a Stokes tuning range from 1176.46 nm to 1181.63 nm (corresponding to fundamental wavelengths from 1061.00 nm to 1065.20 nm). The Raman laser was capable of operating efficiently even tuned well into the wings of the fundamental emission band. Tunable yellow output, obtained by frequency-doubling the Stokes was achieved between 588.23 nm and 590.81 nm, which covers the guide star wavelength of 589.16 nm. The results are presented in a way that enables the reader to understand the factors that affect the choice of etalon, the obtainable tuning range, and the output power and efficiency of the laser.

2. Experimental arrangements

The design of the tunable intracavity Raman laser is shown in Fig. 1. The pump source was a fiber-coupled LD (nLight element e06) emitting at 878.6 nm with a narrow linewidth of ~0.3 nm (FWHM). The core diameter and the NA of the fiber were 200 µm and 0.22, respectively. A 1:2 multi-lens coupler was used to refocus the pump light into a a-cut Nd:GdVO₄ laser crystal which has anti-reflective (AR) coating at the pump (R<0.1%) and laser wavelengths (R<0.1%) on both facets. The 0.3-at.%-doped, 10 mm long crystal was measured to absorb ~75% of the incident pump under non-lasing condition. A 16 mm long a-cut BaWO₄ crystal which was AR coated for both 1.06 μm fundamental and 1.18 μm Stokes was used as Raman crystal. It was oriented so that the fundamental polarization was parallel to its a-axis. Both the laser and Raman crystals were wrapped in indium foil and mounted in water-cooled copper holders maintained at 20 °C. Three mirrors, M1, M2 and M3, coated for AR at 880 nm (R<0.2%) and highly reflective (HR) at the fundamental (R>99.97%) and Stokes (R>99.99%) wavelengths defined the fundamental cavity. M1 was a flat mirror, while the radii of curvature (ROC) of M2 and M3 were 100 and 500 mm, respectively. The Stokes cavity was defined by mirrors M3 and M2, together with a flat dichroic mirror M4 coated to be HR at the Stokes (R>99.9%) and highly transmissive (T>99.5%) at fundamental, and an output coupler M5. With this coupled-cavity design, we could turn on/off the SRS conveniently by blocking the Stokes cavity to observe the influence of SRS process on the fundamental laser spectrum, and also control the fundamental linewidth by inserting etalons into the fundamental cavity without disturbing the Stokes cavity. The output coupler M5 used to obtain Stokes output had a ROC of 100 mm and T = 0.45% at 1.18 μm. To achieve yellow output, M5 was removed, and replaced by a frequency doubling crystal and mirror M6 (these modifications are highlighted by the dashed boxes in Fig. 1). M6 was HR at Stokes (R>99.99%) and AR at 588-590 nm (R<1.2%) with a ROC of 150 mm. The crystal used for frequency doubling was a LBO crystal cut at θ_pm = 90°, φ _pm = 0° (dimensions 4 × 4 × 8.5 mm³ with both surfaces AR coated for 1063 – 1180 nm). It was mounted in a copper holder heated to ~41 °C to achieve non-critical phase matching (NCPM).

 

Fig. 1 Scheme of the experimental setup of the intracavity Raman laser. Fundamental cavity length was 163 mm (M1-M2-M3: 98 + 65 mm), and the Stokes cavity length was 158 mm (M3-M2-M4-M5: 65 + 50 + 43 mm) for Stokes output and 225 mm (M3-M2-M4-M5: 65 + 50 + 110 mm) for yellow output.

Download Full Size | PDF

The distances between elements are shown in Fig. 1. The 163 mm long fundamental cavity length resulted in a 195 μm cold-cavity TEM₀₀ mode size (radius) in the laser crystal which matched the 200 μm pump radius well, whilst the 120 μm fundamental spot radius in the BaWO₄ crystal was close to the 130 μm Stokes radius determined by the 158 mm long Stokes cavity. The thermal focal length in the Nd:GdVO₄ crystal was estimated to be 50-60 mm under the maximum incident pump of 18.7 W, while the BaWO₄ crystal was estimated to have a highly astigmatic negative thermal focal length of ~(-)60 mm in c-axis direction and ~(-)500 mm in a-axis based on the measured Stokes output power of around 2 W at this pump level [20]. Hence, the fundamental laser mode spot in the laser and Raman crystals became 120 × 160 μm and 170 × 180 μm ellipses and the Stokes beam size in the Raman crystal was 110 × 120 μm. For the yellow experiment, the 225-mm Stokes cavity with the mirror M5 replaced by a 150 mm ROC one resulted in beam sizes of 110 × 120 μm in the Stokes crystal and 160 × 180 μm in the LBO crystal.

For the purpose of tuning the wavelength as well as controlling the linewidth, three fused silica etalons with different thicknesses of 50, 100 and 300 μm were evaluated individually. Each etalon had same coating reflectivity of R = 30% at 1063 nm. The associated free spectral range (FSR), full width corresponding to an etalon transmission of 98% (FW@T = 98%), and the etalon tilt angle (θ_tilt) required to realize peak transmission at the wavelength of peak gain for Nd:GdVO₄ (found to be 1063.2 nm), are shown in the first 3 rows of Table 1. The laser spectrum was recorded by a Spectrum Analyzer (Bristol 771A) which measured wavelengths to an accuracy of 0.2 pm and displayed the laser spectrum to a resolution of 2 GHz.

Table 1. Parameters and experimental tuning range of each etalon used.

View Table

3. Experimental results

3.1. Fundamental spectrum control leading to power enhancement and linewidth reduction

Laser performance is first presented for the free-running laser, i.e. the case without any etalons being used. The laser characteristic is shown in Fig. 2(c). Threshold for SRS occurred for a diode pump power of 0.8 W, and 1.73 W Stokes output at 1179 nm was obtained for the maximum incident pump power of 18.7 W with corresponding optical efficiency being 9.3%. In the absence of SRS (Stokes cavity blocked), the free-running 1063 nm Nd:GdVO₄ laser spectrum had a full width at half maximum (FWHM) of ~0.15 nm at the maximum pump power of 18.7 W, as shown in Fig. 2(a). The two peaks evident in the spectra were separated by 0.12-0.13 nm (32-35 GHz) and were found to be due to the weak etalon effect between mirror M1 and the front facet of the Nd:GdVO₄ crystal which were ~4 mm apart. We note that it is relatively unusual to observe spectral output of Raman lasers with this resolution, and such structure is not readily apparent when observed using more-common spectrometers (such as our HR4000 spectrometer which has a resolution of around 0.1 nm). When SRS was enabled, by unblocking the Stokes resonator, the fundamental spectrum became significantly broader, to over 0.7 nm. Again, the spectrum contained several peaks consistent with the etalon effect mentioned above, and the observed spectral broadening is attributed to the frequency-dependent loss for the fundamental field induced by SRS, as investigated and explained in our former work and in works by other researchers [8,9]. The 1.18 μm Stokes linewidth was over 0.6 nm (130 GHz) wide – much broader than the BaWO₄ crystal’s Raman gain linewidth of 48 GHz [21] and consisted of several peaks spaced 0.15-0.16 nm (32-35 GHz) apart. The broadened fundamental spectrum, as well as the wide Stokes, would decrease the effective Raman gain seriously according to the theoretical calculation and former experimental observation [5,9]. Using the collected fundamental and Stokes spectra and the 1.6 cm⁻¹ Lorentzian Raman line shape of BaWO₄ crystal, the calculated effective Raman gain coefficient based on Eq. (3).31) in [22] was only ~20% of the steady state Raman gain coefficient g₀.

 

Fig. 2 (a) Fundamental spectrum broadening induced by SRS and corresponding Stokes spectrum under 18.7 W incident pump power. (b) Fundamental spectra with different etalons under 18.7 W incident pump power. (c) Power transfer with different etalons. Inset: typical Stokes spectra with fundamental linewidth control (300 μm etalon, 9.8 W and 18.7 W pump).

Download Full Size | PDF

Laser performance was then investigated using each of the three etalons. When each etalon was inserted into the fundamental cavity, enhanced Stokes power was obtained as theoretically expected, although the SRS threshold increased to around 1.7 W in each case because of the insertion losses [23]. The insertion losses include etalon-specific losses such as those associated with the surface figure and non-parallelism, and walk-off losses which depend on the angle at which the etalon is used. Figure 2(c) shows power transfers for the Raman laser, where the cavity has been aligned and etalons tilted for best Stokes output power. Using the 50 μm, 100 μm, and 300 μm etalons, the maximum Stokes output under the fixed pump power increased substantially from the free-running level of 1.73 W to 2.10 W, 2.78 W, and 2.80 W, respectively, with corresponding optical efficiency increases from 9.3% to 11.2%, 14.9% and 15.0% respectively. The different power enhancements observed using each etalon is attributed to the correspondingly different fundamental spectra and resultant different effective Raman gains [5]. Figure 2(b) shows typical fundamental spectra collected at the maximum Stokes output power for each etalon. It can be seen that the 100 μm and 300 μm etalons were most effective in that the fundamental comprised only one peak with a linewidth of ~0.05 nm, while the broader transmission band of the 50 μm etalon allowed 2 peaks to oscillate. The fundamental spectrum with the 50 μm etalon was narrowed compared with the free-running laser, but was significantly broader than for the thicker etalons; therefore the effective Raman gain coefficient was intermediate compared to the free-running and thicker etalon cases.

The narrowing of the fundamental was expected to give rise to narrowing of the Stokes spectra, and this was verified experimentally. For incident pump powers below 14.4 W, the Stokes spectra was found to consist of a single peak whose FWHM was instrument limited to ~2 GHz (9.2 pm @1180 nm). This is shown in the lower spectrum inset in Fig. 2(c), and considering the 0.9 GHz (4.2 pm @1180 nm) FSR of the Stokes cavity, we speculate there should be no more than 2 longitudinal modes oscillating. Above 14.4 W, the Stokes spectrum broadened and more longitudinal modes started oscillating as the pump power was increased, as shown in the upper spectrum inset in Fig. 2(c). It also can be seen that the Stokes wavelength exhibited a small red shift with increasing pump power, which is attributed to the fundamental wavelength shift due to etalon heating by the high intracavity fundamental laser power. In the case of the 50 μm etalon which exhibited the two-peak fundamental spectrum, the Stokes spectrum contained only one single peak. We speculate that the Raman gain linewidth for BaWO₄ of 48 GHz (FWHM) is larger than the interval of the two fundamental peaks of 35 GHz, so the energy in both of the two fundamental peaks can be extracted by the single Stokes peak efficiently. If only one of the two fundamental peaks with similar intensity was converted to the Stokes, the output power would drop a lot and be quite unstable due to the competition. This was not the case, as the power obtained was only ~25% lower compared to the other two etalons. The calculated effective Raman gain coefficient with the narrowed one-peak fundamental (100 μm and 300 μm etalons) and Stokes spectra was ~70% of the steady-state Raman gain coefficient g₀, and became ~50% of g₀ with the two-peak fundamental obtained using the 50 μm etalon.

3.2. Wavelength tuning

Wavelength tuning was realized by varying the tilt angle, ie the external angle at which the etalons were inclined relative to the resonator axis. Typically the tilt angle varied from near-zero to around 10 degrees. When the etalons were aligned perpendicular to the resonator axis, the non-stabilized coupled cavity effects would lead to power fluctuation and unstable spectra. In other cases however, the laser could operate with reproducible stability. Due to the layout of the laser components and associated mounts, it was difficult to determine absolute tilt angles. It was more straightforward to measure relative changes to tilt angle, by measuring tuning knob rotations.

The FWHM of the π-polarization emission spectrum of the a-cut Nd-vanadate crystal is ~1 nm, however, the tuning range achieved experimentally was much wider than this value and the tuning range reported in [24]. The widest tuning range was obtained using the 50 μm etalon, which had the largest FSR of the three etalons investigated. The Stokes wavelength (λ_s) tuning range extended from 1176.46 nm to 1181.63 nm (corresponding to a fundamental wavelength (λ_f) tuning range from 1061.00 nm to 1065.20 nm). Figure 3(a) presents information about how the fundamental and Stokes wavelengths, and the Stokes powers varied in response to tilting the etalon. At near-normal incidence, the fundamental wavelength was ~1062.60 nm, and the Stokes wavelength was ~1178.40 nm. When starting to tilt the etalon away from near-normal incidence, the fundamental and corresponding Stokes wavelength became shorter, and the output power decreased until the Stokes stopped oscillating when the fundamental wavelength was shorter than 1061.00 nm (θ = 4.7°, λ_s = 1176.46 nm). Upon further tilting of the etalon to an angle of 8.2°, stable and continuously-tunable Stokes output came back when the fundamental wavelength (determined by the neighboring etalon transmission peak) reached 1065.2 nm (θ = 8.2°, λ_s = 1181.63 nm). The Stokes and fundamental wavelengths then became shorter and the Stokes power increased until the tilt angle reached 9.7°, at which point the fundamental switched to the laser emission peak of 1063.2 nm. Further increasing the tilt angle to 10.1°, the fundamental went back to 1062.60 nm, the wavelength that had earlier corresponded to θ~0°. Thus, Stokes tuning was achieved from 1176.46 nm to 1181.63 nm, for fundamental tuning from 1061.00 nm to 1065.20 nm. In terms of “useful” tuning, Stokes powers of over 1 W were achieved for Stokes wavelengths across the range 1177.54 nm to 1181.38 nm.

 

Fig. 3 (a) Tuning curve of the Raman laser with the 50-μm etalon under incident pump power of 18.7 W and (b) fluorescence spectrum of the a-cut Nd:GdVO₄ crystal and tuning ranges with each etalons. The dotted line is a guide to the eye.

Download Full Size | PDF

In addition to the wavelength-tuning and output powers detailed above, several interesting observations were made. We noticed that the visible fluorescence in the vanadate crystal became brighter when the fundamental wavelength was tuned to ~1062 nm and shorter, which was consistent with the excited state absorption (ESA) line centered at 1058.9 nm that was reported in [25]. Such ESA, coupled with the small stimulated emission cross section in the wings of the fundamental line, may have prevented the intracavity fundamental power at wavelengths below 1061.00 nm from reaching the level required for SRS threshold. With the tilt angle θ of ~6.6°, we also got up to watt level Stokes output with the λ_f ~1066.8 nm (, λ_s ~1183.6 nm), which is not shown in Fig. 3. With θ between 6.6° and 8.2°, the fundamental was tuned continuously from 1066.8 nm to 1065.2 nm but no Stokes was generated. This is because the 1071 nm laser line of Nd:GdVO₄, which corresponded to another successive etalon transmission peak, started oscillating together with the 1.06 μm fundamental laser, thus dividing the fundamental power and pulling the power below the level required for SRS. For tilt angle a little smaller than 6.6° (λ_f > 1066.8 nm), it was the 1083 nm emission line of Nd:GdVO₄, which is 2 FSRs away, started lasing and limited the fundamental intensity to below the SRS threshold . The fundamental laser remained π-polarized during the whole process.

The tuning behaviors with the 300 μm and 100 μm etalons were much simpler, and the wavelength tunability ranges were limited by the FSR of each etalon. The FSR-limited fundamental tuning ranges were 1062.74-1063.88 nm and 1062.20-1064.80 nm, with Stokes tuning range being 1178.60-1180.00 nm and 1177.94-1181.13 nm, respectively. The laser wavelength became shorter with increasing tilt angle till it reached the shorter edge of the tuning range. Further tilting the etalon would make the fundamental laser operate in dual-wavelength which correspond to the two neighboring etalon transmission peaks, and the Stokes power fluctuated and decreased seriously, or even stopped oscillating. When the longer of the two fundamental wavelengths totally suppressed the other one with the increasing tilt angle, the Stokes output became stable again and defined the longer-edge of tuning range. The parameters and experimental tuning range of each etalon used are given in Table 1.

Figure 3(b) shows the measured π-polarization fluorescence spectrum of our a-cut Nd:GdVO₄ crystal using a spectrometer (Ocean Optics HR4000) and the fundamental tuning range over which stable Stokes output can be obtained. The vertical markers indicate the fundamental wavelengths achieved with tilt angle θ~0°. It can be seen that the fluorescence intensities on the edges of the tuning ranges were quite similar when using the 100 μm and 300 μm etalons. However, for the 50 μm etalon, the fluorescence intensity at the longer edge of the tuning range of 1065.20 nm, was still over 2 times higher than that at the shorter edge of 1061.00 nm, and did not suffer the ESA loss. Therefore we speculate that the Stokes tuning range may be further extended if the 1071 nm and 1083 nm emission line could be suppressed, eg by incorporating a mirror with transmission at these wavelengths, or by using an additional etalon or birefringence filter.

3.3. Yellow output

With the LBO crystal inserted into the Stokes cavity and the output coupler M5 replaced by M6, tunable yellow output was obtained. Figure 4(a) gives the power transfers of the yellow output with the etalons tilted for maximum output power and for the free-running case. It can be seen that the yellow output power was also improved significantly by the fundamental linewidth narrowing. The maximum one-way yellow power under incident pump of 18.7 W was 0.69 W without linewidth control of the fundamental, and this increased to 0.90 W and 1.14 W, respectively, when the 50 μm and 100 μm etalons were used. Considering the backward propagating yellow, which is reasonable to assume that the power is similar to that of the forward propagating, was not collected, it should be possible to increase the yellow output power by a factor of ~2 by using a proper yellow reflector in the cavity.

 

Fig. 4 (a) Power transfer (λ = 589.58 nm under maximum power) and (b) wavelength tuning of the yellow output with and without spectrum control. Inset: typical yellow spectrum with fundamental linewidth control (100 μm etalon, 9.8 W pump). The lines are a guide to the eye.

Download Full Size | PDF

The free-running yellow spectrum had a linewidth around 110 GHz (~0.13 nm at 590 nm). The spectrum comprised several peaks spaced at roughly half of the intervals on the Stokes spectra, which is due to the sum-frequency mixing between the Stokes peaks. With the inclusion of the etalons, the yellow linewidth remained instrument-limited ie <2 GHz (2.3 pm at 590 nm) till the pump power reached 9.8 W, and reached 4 GHz (4.6 pm at 590 nm) at the maximum pump of 18.7 W. Similar linewidths were achieved using both 50 μm and 100 μm etalons, and a typical spectrum is shown in the inset of Fig. 4(a).

The yellow power enhancement brought about by using etalons is largely a consequence of the enhanced Stokes power, rather than the accompanying spectral narrowing of the Stokes, given the wide acceptance bandwidth of 25 nm for NCPM in the 8.5 mm long LBO crystal [26]. However a secondary emission line at 1227 nm, which corresponds to the 332 cm⁻¹ Raman shift from the 1178 nm Stokes and did not occur in Stokes output experiment, was observed in the yellow output experiment with incident pump power over 12.2 W, since the Stokes cavity for yellow output was high-Q for both 1178 and 1227 nm. This limited the intracavity Stokes power and yellow output power to some extent and we anticipate higher yellow powers could be obtained by using Stokes mirrors that are tranmissive at 1227 nm.

The wavelength tuning ranges with the 50 μm and 100 μm etalons were 588.23 nm to 590.81 nm and 588.97 nm to 590.57 nm, respectively, which covered the guide star wavelength of 589.16 nm (fundamental at 1062.50 nm), as shown in Fig. 4(b). With the 300-μm etalon we got a maximum yellow power of 1.10 W which was similar to that with the 100-μm etalon, but the tuning range was limited to 589.30 nm to 590.00 nm by the smaller FSR, hence it is not plotted in the Fig. 4(a). When using the 50-μm etalon, the one-way yellow power at the guide star wavelength of 589.16 nm was 0.87 W, almost the same with the maximum output power of 0.90 W got at 589.58 nm which correspond to the fundamental emission peak of 1063.2 nm. As was the case with the laser producing Stokes output, the intracavity etalon not only narrowed the fundamental linewidth and thus improved the power and spectrum characteristics of the Raman laser, it also enabled tuning of the yellow output over a useful range, making it easier to match a crystalline Raman laser to a particular spectroscopic application.

4. Discussion

4.1. Crystal emission spectrum and wavelength tuning range

C-cut Nd vanadate crystals have been proposed as better choices for tunable output than a-cut crystals, because of their wider and smoother emission spectra [16,27,28]. However, since the a-cut crystals have stimulated emission cross sections 4-5 times larger than those of the c-cut crystals, the absolute values of their stimulated emission cross section far off the peak, i. e., at the edges of the 1061.00 nm to 1065.20 nm fundamental tuning range here, are still comparable with those of the c-cut crystals and hence enable a similar tuning range, though the FWHM of their emission spectra are much narrower. Therefore, the more commonly used a-cut Nd-vanadate crystals, which have the important benefit of polarized emission, are also promising gain mediums for laser output tunable within multi nanometers, thus make it easier to achieve the specific wavelength needed.

According to our calculations, ~250 W intracavity fundamental power (two-way) was needed to reach the SRS threshold in this cavity arrangement, while the saturation power of the 0.3-at.%-doped a-cut Nd:GdVO₄ crystal is only 2.5 W using the peak stimulated emission cross section σ_e0 of 10.3 × 10⁻¹⁹ cm², the fluorescence lifetime τ of 92 μs [29] and the 200 μm beam radius in the vanadate crystal. This means, even with the fundamental laser operating at 1061.0 nm - the shorter edge of the tuning range where the stimulated emission cross section σ_e was 7% of its peak value σ_e0 - the intracavity power at the Raman threshold was still ~7 times the saturation power so that the population inversion could be extracted to build the fundamental laser field efficiently to be further converted to the Stokes by SRS. As long as there is a gap large enough between the fundamental saturation power and the SRS threshold, the smaller stimulated emission cross section would only increase the thresholds of lasing and SRS, but not the Stokes power slope efficiency [5].

4.2. The influence of etalon losses

Etalons would not usually present substantial losses within a conventional laser having reasonable output coupling. However in the context of a cw Raman laser where it is critical to maintain low losses in order for the fundamental intensity to become sufficiently high to reach threshold for SRS, etalon losses are important to consider. For example, despite having used very high quality etalons, we found that insertion losses increased the threshold for SRS from 0.8 W to 1.7 W for our Stokes laser. Etalon losses fall into two categories: fixed and angle-dependent. The fixed losses arise from non-parallel surfaces and imperfect surface figure and flatness. The angle-dependent loss is the walk off loss, which scales with the etalon’s thickness, reflectivity R and tilt angle θ and has the potential to impact the output power and tuning range of our lasers. Here we have calculated the walk-off loss for our etalons, according to the methods described in [23]. Figure 5 shows the theoretical walk-off loss L as function of θ for the three etalons used in the experiments, calculated with the theoretical TEM₀₀ mode size of ~150 μm and the measured fundamental beam quality factor M² of ~2 at the maximum power (with SRS on). The walk-off loss with the 300 μm etalon can be as large as over 2% when tuning the laser wavelength for one FSR. This is in accordance with the experimental tuning curve- the Stokes power 1179.41 nm (fundamental 1063.40 nm) with the 300 μm etalon tilted for ~4° was below 1 W, much lower than that of 2.8 W at 1179.25 nm (fundamental 1063.27 nm) with a tiny tilt angle of less than 0.2°, though the emission cross sections at these two wavelengths were almost the same. In contrast, the tuning curve with the 100 and 50 μm etalons, which have much smaller walk-off losses, were much smoother.

 

Fig. 5 Theoretical walk-off losses with different etalon parameters. The arrows mark the tilt angle needed for tuning the fundamental wavelength over one FSR.

Download Full Size | PDF

5. Conclusions

The wavelength tuning, linewidth narrowing and power enhancement of an intracavity Nd:GdVO₄-BaWO₄ Raman laser by means of solid fused silica etalons has been investigated. By incorporating an etalon in the fundamental cavity, the fundamental laser linewidth was reduced substantially, leading to higher effective Raman gain and enhanced output powers. The spectrally-narrower fundamental resulted in narrow-linewidth Stokes output. Tilting the 50 μm etalon led to tuning of the fundamental from 1061.00 nm to 1065.20 nm, a Stokes tuning range from 1176.46 nm to 1181.63 nm, and a frequency-doubled narrow-linewidth yellow output tunable from 588.23 nm to 590.81 nm, which covered the guide-star wavelength of 589.16 nm. The results show that the commonly used a-cut Nd-vanadate crystals are capable of lasing efficiently in a tuning range of multiple nanometers, though their emission linewidth expressed using FWHM are much narrower, thus enable the intracavity Raman lasers to conveniently achieve specific wavelength requirements.