1. Introduction

Ground-based Extremely Large Telescopes (ELTs) [1] are expected to revolutionize the field of astronomy, as they will be used to conduct detailed studies of planets orbiting other stars, super-massive black holes, dark matter, and dark energy. These telescopes will rely on state of the art adaptive optics systems to correct for atmospheric distortions. In order to provide broad sky coverage, an artificial guide star will be an integral part of such systems. Experimental results so far have clearly demonstrated the benefits of using 589 nm lasers to generate artificial guide stars [2]. The 589 nm light excites sodium atoms in the mesosphere and a portion of the resonantly scattered light is directed to a wavefront sensor that measures the distortion caused by the atmosphere. This distortion is then corrected via a deformable mirror.

First generation sodium beacons relied on dye materials that lased directly at 589 nm [3]. Since then, several research groups have developed more powerful, yet compact and ruggedized, 589 nm sources. One approach that has recently attracted attention is the construction of a single-frequency fiber amplifier operating at 1178 nm. The light can be subsequently frequency doubled to 589 nm using a resonant second harmonic generation (SHG) cavity. Since Yb-doped fibers have very low gain at 1178 nm, amplifiers utilizing conventional fibers would suffer from amplified spontaneous emission (ASE). Double clad Yb-doped photonic bandgap fiber (PBGF) pumped at 976 nm and specifically designed to suppress lasing in the 1-1.1 µm region have shown promise [4]. However, Yb-doped PBGFs have relatively high cost and are, as of yet, not been demonstrated for monolithic all-fiber systems. Alternatively, core pumped Raman fiber amplifiers (RFAs) utilizing single-mode passive fiber can be used to generate 1178 nm light [5,6]. An important general point to consider with Raman amplification in silica fibers is that the gain is both broad and tunable and thus RFAs can offer access to a range of wavelengths that are not traditionally accessible through laser host materials. Hence, in addition to guide star applications, they can be potentially utilized in RGB sources for digital projection displays [7].

In a single-frequency RFA, the backward travelling Stokes light is amplified through both Brillouin gain as well as Raman gain. The signal light drives the former while the pump light drives the latter. As shown in [8], both of these effects are significant and should be taken into account when modeling the system. It was also shown in that work, that when the fiber length is optimized based on available pump power, the single-frequency Raman output at stimulated Brillouin scattering (SBS) threshold scales linearly with pump power. Currently, commercially available 1120 nm sources are limited to approximately 100 W. Thus, a single-stage RFA seeded with a diode source with an output power of ~10 mW would require a fiber length >50 m to generate a multi-watt signal. It would appear that at such fiber lengths SBS would be a limiting factor preventing scaling beyond 1 W. However, two considerations are in order: 1) the nonlinear effective length in an RFA is reduced compared to the actual fiber length due to the very rapid rise of the signal at the output end and 2) there are some fibers commercially available, which have an appreciably smaller Brillouin gain coefficient,$g_B$, than the typical value for optical fibers of 2.5 × 10⁻¹¹ m/W.

The highest output power reported in the literature from single-stage single-frequency RFAs is 18 W for a polarization-maintaining (PM) fiber [9] and 39 W for a non-PM fiber [10]. In the former work, an acoustically tailored fiber design was used to suppress SBS. The latter work does not provide information on the suppression of SBS. Recently, Zhang et al. reported on a 44 W two-stage PM RFA whereby SBS suppression was achieved by applying a longitudinally varied strain [11].

In this work, we investigate experimentally and theoretically power scaling in PM single-frequency RFAs operating at 1178 nm. As a first step, we used a commercial-off-the-shelf (COTS) PM fiber to generate 10 W of 1178 nm single-frequency output in a counter-pumped single-stage configuration. We present pump-probe measurements showing $g_B$to be ~1.2 × 10⁻¹¹ m/W for this type of fiber. In order to obtain 10 W of output power, SBS suppression in the RFA was achieved by creating a two-step temperature profile (i.e. three temperature regions). A study was also conducted to investigate the signal output at SBS threshold for optimal fiber length. This study revealed a linear dependence with pump power, which is in agreement with the theoretical prediction of [8]. Additionally, we conducted a study of single-frequency output power as a function of seed power by constructing a counter-pumped two-stage RFA that utilizes the acoustically tailored fiber described in [9]. To the best of our knowledge, our experimental studies of the scalability of single-frequency RFAs as a function of both pump and seed powers are the most detailed studies reported so far in the literature in this area. For the counter-pumped two-stage RFA, up to 22.2 W of power of single-frequency output was obtained. In contrast, when we utilized a co-pumped configuration, the linewidth of the output signal broadened considerably.

2. Experimental results: COTS fiber

2.1 Pump-probe experiment

A key consideration in constructing a relatively high power RFA from COTS fiber is to identify fibers with relatively low$g_B$. Dopants contained in the fiber core can radically affect the Brillouin gain coefficient. Mermelstein compared the SBS process in aluminum-doped and germanium-doped fibers [12]. For the former, he measured$g_B$to be ~1.0 × 10⁻¹¹ m/W. This relatively low value was attributed to the acoustic anti-guiding properties of the fiber core due to the presence of aluminum.

The COTS PM980-XP from Nufern contains high aluminum content [13]. This single-mode passive fiber possesses core and cladding diameters of 6 µm and 125 µm, respectively. We examined the Brillouin gain spectrum (BGS) of this type of fiber using the well-established pump-probe technique. The experimental set-up is shown in Fig. 1. Two non-planar ring oscillators (NPRO) sources operating at approximately 1064 nm and with nominal linewidths on the order of KHz were used in the setup as the pump and probe (Stokes) laser sources. Frequency tuning of the probe NPRO was achieved by slowly modulating the temperature of the Nd:YAG crystal. The pump signal was amplified by utilizing a single-mode Yb-doped fiber amplifier and propagated through a polarizing beam splitter (PBS) in order to separate the Stokes light. The polarizations of the input beams were oriented along the slow axis of the PM980-XP fiber using half-wave plates. The fiber length used was 10 m. Fused fiber tap coupler/splitters (TAP 1 and TAP 2) were used to separate 1% of the signal to be later combined with a 50/50 coupler. The photodiode (PD1) was used with the RF spectrum analyzer (RFSA) to measure the beat note of the two signals separated by ~16 GHz; which is approximately equal to the Brillouin shift frequency in an optical fiber at a wavelength of 1064 nm.

 

Fig. 1 Experimental setup for the pump-probe technique to measure the Brillouin gain bandwidth. ISO 1 and ISO 2 are optical isolators.

Download Full Size | PDF

We scanned a frequency span of approximately 1 GHz with a resolution on the order of several MHz. The intensity of the output probe was monitored using a photodiode (PD 2). A plot of the BGS at a pump power of 400 mW and a probe power of 10 mW is displayed in Fig. 2. As shown, the peak output was obtained at approximately 15.9 GHz corresponding to the center Brillouin shift. The FWHM is approximately 78 MHz. At this pump power, the single pass Brillouin gain for 10 m of this type of fiber is on the order of 1, and therefore gain narrowing is negligible [14]; thus 78 MHz is approximately equal to the spontaneous Brillouin gain bandwidth. The 78 MHz value is relatively large for a silica fiber and is indicative of a lower Brillouin gain, as the acoustic phonon lifetime is proportional to the reciprocal of this value [14].

 

Fig. 2 Experimental data of the Brillouin gain spectrum in the PM980-XP fiber obtained by conducting a pump-probe experiment. The peak gain occurs at a Brillouin shift of approximately 15.9 GHz and the bandwidth is 78 MHz.

Download Full Size | PDF

In order to obtain an estimate of $g_B$we conducted a set of experiments whereby the input probe power was kept at 1 mW throughout the measurements. We varied the pump power and recorded the output Stokes at peak gain (PD2) to generate the data in Fig. 3. The peak gain for each pump power was determined by scanning through the frequency range. This data was fitted to the numerical solutions obtained by solving the coupled system of equations describing the evolution of the signal and Stokes. This fit yielded a value for $g_B$of $1.2\times {10}^{-11}$ m/W. This value is approximately equal to that reported by Mermelstein for the aluminum-doped fiber and in line with the value used in the simulations of [8].

 

Fig. 3 Experimental data and the numerical fit corresponding to the pump-probe study of the Brillouin gain in the PM980-XP fiber. The fit yielded a value for the Brillouin gain coefficient of 1.2 × 10⁻¹¹ m/W.

Download Full Size | PDF

2.2 Experimental set-up

Both co- and counter-pumped single-stage RFA configurations were investigated. To the best of our knowledge, very little experimental data has been provided in the literature on comparing the two configurations for single-frequency RFAs. We found that the former configuration led to considerable broadening of the signal light linewidth; thus rendering the output unsuitable for frequency doubling in a resonant cavity. More details of the spectral broadening in a co-pumped RFA can be found in section 3. As we will show below, no such broadening was observed for the counter-pumped configuration. The experimental set-up for this case is shown in Fig. 4. A 30 mW Toptica DFB diode laser operating at 1178 nm provided the seed light. This fiber coupled diode laser has a nominal linewidth of 2 MHz and an isolator to prevent damage from the backward traveling light. The pump source was an IPG water-cooled broadband Raman fiber laser centered at approximately 1120 nm with a maximum output of 100 W. Both lasers were aligned to the slow axis of the fiber and a system of WDMs was used to combine/separate the 1178 and 1120 nm light. The amplifier output was angle polished.

 

Fig. 4 Experimental setup of the counter-pumped single-stage RFA. The WDMs were used to combine/separate the different wavelengths. The TAP was used to monitor the forward and backward traveling light and the amplifier output was angle polished.

Download Full Size | PDF

As shown in Fig. 4, a 3% TAP allowed us to monitor both the forward and backward traveling light at the input end of the RFA. WDM 1 was used to couple the 1178 nm light into the RFA while sending the counter propagating unabsorbed 1120 nm light out of the system and into a pump dump. The maximum 1178 nm power available to seed the RFA was measured past WDM 1 to be 15 mW. Two WDMs (WDM 2, WDM 3) were used at the output end of the RFA. The WDMs allow the light originating from the IPG to pump the RFA while the signal is sent down a different port for characterization. The maximum pump power after WDM 2 was measured to be approximately 75 W. It is worthwhile to point out here that since we were constructing a PM RFA, the WDM components were required to be PM. The power handling capability of commercially available PM WDMs is typically < 10 W. Gooch and Housego manufactured a specialty high-power WDM rated for 50 W for this work. In order to drive the component to powers, in some cases exceeding the manufacturer’s recommendations, we mounted and thermally cooled the WDMs to prevent damage. There was some leakage (~3%) of 1178 nm light into the 1120 nm port and vice versa.

We used a high-resolution optical spectrum analyzer (OSA) to examine the spectral content of the 1120 nm IPG laser. This laser is a linearly polarized oscillator comprised of a passive fiber pumped with an Yb-doped fiber laser operating at approximately 1064 nm. The laser light at 1120 nm is generated through a first-order Raman shift. A second-order Raman shift would then introduce light at 1178 nm into the predominantly 1120 nm output. The spectral content (resolution of 0.1 nm) at 50% and 90% of the maximum output power is shown in Fig. 5. The primary peak occurs near 1120 nm. Furthermore, a broadband secondary peak is present near 1178 nm. As expected from a second-order Stokes effect, the relative spectral content near 1178 nm rises with increased output power. An important consideration was the amount of noise in the 1178 nm region introduced into the RFA. This noise can compete with the seed for Raman gain as well as act as a seed for the growth of SBS. To be certain, WDMs 2 and 3 filtered out a significant amount of this noise. However, even then, we could clearly observe it on the OSA as we pumped the RFA without having the seed light on.

 

Fig. 5 Spectral content of the IPG 1120 nm output at 50% and 90% of total output power indicating relative rise in 1178 nm light as the output power is increased. The 1178 nm light is due to the second-order Stokes shift within the oscillator of the pump laser.

Download Full Size | PDF

2.3 Single-stage RFA

The maximum power that can be obtained from a single-frequency RFA in the vicinity of the SBS threshold depends on several factors. These factors include the Brillouin and Raman gain coefficients of the fiber, the available pump and seed powers, the length of fiber, and mitigation techniques used to suppress SBS. One technique to suppress SBS is to create different temperature zones along the fiber as the Brillouin resonance frequency is temperature dependent [15]. A similar effect can be obtained by varying the stress [16].

We subjected three longitudinal segments of a 63 m long fiber to three different temperatures (i.e. a two-step temperature profile). The temperature difference between adjacent segments was ~40 ˚C. This allowed for sufficient separation among the Stokes frequencies so that there was little overlap among the Brillouin gain bandwidths. Since the signal rises rapidly at the output end of an RFA, the lengths were chosen such that the longest segment was at the input end followed by the segment adjacent to it. The process of optimizing the lengths of the fiber segments for maximum SBS suppression could have been tedious. However, it was made easier by following the procedure described in [8]. Consider $N$ segments with one end of each segment located at$L_1$,$L_2$,$L_3$,…$L_N$, where $L_N=L$. The optimal lengths are estimated by solving the following set of equations:

.

.

.

Figure 6 provides plots of the 1178 nm Raman power and the backward power as a function of launched pump power for the two-step temperature case. For comparison, the corresponding plots when a uniform temperature was applied throughout the fiber length are shown in the figure. The SBS threshold definition is somewhat arbitrary and in the literature there are multiple definitions [17]. Nevertheless, the SBS threshold is characterized by a rapid increase in the backward power. In the comparison of thresholds presented herein, we define the SBS threshold as occurring at the point corresponding to minimal increase in the forward power (<5%) due to any further increase in the pump power. This point corresponded in our experiments to a reflectivity of ~5%. For the two-step temperature case, 10.1 W of 1178 nm was obtained at SBS threshold when pumped with approximately 70 W of 1120 nm light. For the uniform temperature case, the pump power was approximately 58 W, which provided an 1178 nm output of ~3.8 W at SBS threshold. The spectral linewidth of the Stokes light for this case was captured by a Toptica FPI-100 Fabry-Perot interferometer (FPI). The etalon is a piezoelectrically scanned confocal Fabry-Perot interferometer, with finesse > 500 and a free spectral range of 1 GHz. The 2 MHz resolution of the FPI is sufficient to characterize the bandwidth of the Brillouin gain spectrum. Due to SBS gain narrowing [14], the measured linewidth was considerably smaller than the spontaneous Brillouin bandwidth measured using the pump-probe experiment. Furthermore, as shown in Fig. 7, the measured bandwidth of the Stokes light decreased from 30.8 MHz to 26.2 MHz as the output power increased from ~3 W to ~3.8 W.

 

Fig. 6 1178 nm signal and backward power vs. 1120 nm pump power for the Nufern PM980-XP fiber for the cases of a two-step and uniform temperature profiles. The application of thermal gradients led to 2.6x the output power of the uniform temperature case.

Download Full Size | PDF

 

Fig. 7 Stokes light spectrum as captured by a Fabry-Perot interferometer for two different reflectivities. Due to gain narrowing, the bandwidth is much smaller than the spontaneous Brillouin gain bandwidth. The plot in green is the captured spectrum below SBS threshold at an output power of ~3 W, while that in blue was obtained at ~3.8 W.

Download Full Size | PDF

The application of a two-step temperature profile provided approximately 2.6 times more power, which is reasonably close to our calculated theoretical limit of 3 from our model [8]. Still, additional power can be attained through further optimization of the temperature separation and more precise selection of the lengths of the fiber segments. Moreover, further power scaling can be achieved by applying additional temperature steps. However, we point out that from a practical point of a view a maximum of four temperature regions is recommended as the temperature of the fiber used in our experiments should not exceed 120 ˚C for long term reliability.

We also examined the forward power using the FPI. The spectral linewidth of the 1178 nm signal was monitored from initial low power to the highest power at 10.1 W. Unlike the co-pumped configuration, there were no indications of any spectral broadening for the counter-pumped RFA. The spectrum of the signal is displayed in Fig. 8.

 

Fig. 8 Spectral linewidth of the 1178 nm light at 10 W output showing it to be within the resolution limit of the interferometer. No spectral broadening was observed for a counter-pumped RFA.

Download Full Size | PDF

We also conducted a set of experiments to study the dependence of the output power on the pump power. Both the Brillouin and Raman processes are nonlinear in nature and thus the scalability of single-frequency RFAs in relation to pump power warrants some consideration. It was shown theoretically in [8] that when the fiber length is optimized, the amplifier output scales linearly with available pump power. Optimization of fiber length in this context refers to selecting the fiber length, $L_{optm}$, such that all available pump power is utilized at SBS threshold. Thus, for fiber lengths <$L_{optm}$, the output power will be pump limited and consequently the signal power will be less than that obtained at$L_{optm}$. For fibers lengths >$L_{optm}$, the output power is in this case SBS limited and is also less than that obtained by using a fiber length of $L_{optm}$.

We conducted the study by starting with a fiber of length 80 m. The pump power was then increased until the SBS threshold was encountered. Both pump and signal powers were recorded at SBS threshold. We then performed a cutback experiment whereby the same procedure was repeated for fiber lengths of 75 m, 70 m, 65 m, 60 m, and 55 m. The entire study was conducted without the benefit of a thermal gradient. The simulations and theoretical analysis presented in [8] assumed a seed power that is much smaller than the pump power, which is similar to our experiments as we were seeding with 15 mW throughout the study. The results are shown in Fig. 9. In the plot, the output signal (1178 nm) power is normalized to the output signal power at a length of 55 m while the pump (1120 nm) power is normalized to the corresponding pump power. Also, shown in the figure is the linear fit with a coefficient of determination,$R^2$, of ~0.997. It can therefore be inferred that our experimental results are in good agreement with the theoretical prediction.

 

Fig. 9 Normalized signal power vs. normalized pump power at SBS threshold for counter-pumped RFA. The fiber lengths used in the studied were varied from a length of 55 m to 80 m in increments of 5 m. The data indicates a linear dependence.

Download Full Size | PDF

3. Two-stage Raman fiber amplifier

3.1 Numerical simulations of a two-stage amplifier

The numerical simulations in [8] indicate that for a given fiber length, the output power at SBS threshold actually decreases with increased seed power. On the other hand, the efficiency defined as the ratio of the signal power to the pump power increases. The work in [8] did not address the power scaling if the fiber length was optimized for available seed power. We conducted a numerical study to examine the scalability of single-frequency RFAs in relation to seed power. The system of equations used in the study is described in detail in [8] and entails solving the evolution of the pump, signal, and Stokes light as functions of position in the fiber with the SBS process being initiated from thermal noise.

We proceeded by choosing different seed powers and optimizing the fiber length such as all available pump power is utilized at SBS threshold. The results from a counter-pumped RFA without the benefit of a thermal gradient are shown in Fig. 10. Not considering spectral broadening, one would expect the results for a co-pumped RFA to behave similarly [8]. The fiber simulated is based on the Raman gain enhanced acoustically tailored fiber described in [9]. This type of fiber was utilized in our two-stage RFA experimental work presented below. This fiber allowed us to conduct the study of dependence of output power on seed power more readily while allowing for power scaling to >20 W without the added complexity of utilizing different temperature regions.

 

Fig. 10 Simulation results of output power at SBS threshold vs. seed power for the acoustically tailored fiber described in [8]. Also, shown is the corresponding optimal fiber length.

Download Full Size | PDF

The maximum available pump power in the simulations is taken to be 80 W to match with the pump power used in the two-stage experiments. For this fiber, $g_R$was estimated to be 8 × 10⁻¹⁴ m/W [9]. We estimated $g_B$to be 3.4 × 10⁻¹² m/W based on the acoustically tailored RFA results presented in [9]. As a comparison, a similar estimate for the COTS fiber provided $g_B$~1 × 10⁻¹¹ m/W. This value is lower than the 1.2 × 10⁻¹¹ m/W estimate obtained from the pump-probe experiment that was performed at 1064 nm (see Section 2). As shown in the figure, the output power at SBS threshold increases with seed power. A relatively sharp increase in the output power is obtained as the seed power is varied from 5 mW to 500 mW. Beyond 1 W, the output scales approximately linearly with seed power. The figure also provides the corresponding optimal fiber length. Therefore, in order to achieve further power scaling there is a need to build a two-stage RFA. The first stage is used to generate 1178 nm optical power at a level >500 mW in order to seed the second stage. The experimental effort to achieve this is described in the next two sections.

3.2 Counter-pumped two-stage RFA

The experimental set-up for the counter-pumped two-stage RFA is shown in Fig. 11. A 50 W 1120 nm laser from IPG was used to counter- pump the first stage. This laser is similar to the 100 W 1120 nm laser described above. The 100 W laser was used to pump the second stage. For the first stage, 80 m of the Raman gain enhanced acoustically tailored fiber was used for amplification [9]. Both lasers were aligned to the slow axis of the fiber. The nominal core and cladding diameters of the fiber are 6 µm and 125 µm, respectively. The first stage RFA allowed us to generate >4 W. The same type of fiber was utilized in the second stage. A fiber-coupled isolator at 1178 nm was inserted between the two stages. This isolator was rated for a total power handling capability of 3 W and consequently we kept the maximum seed power originating from the first stage at ~2.7 W. We estimated the insertion loss of the isolator to be 1.4 dB. Two WDMs (WDM 4, WDM 5) were inserted between the second stage fiber and the isolator in order to manage the unabsorbed pump and to ensure that the isolator is not damaged. Our measurements indicated that ~1.8 W of 1178 nm was available to be coupled into the second stage after passing through the isolator and WDMs. The WDMs used to couple the pump light into the second stage were an improved version of the WDMs used in section 2. As a result, they possessed lower insertion loss leading to a maximum of ~82 W of 1120 nm power coupled into the RFA.

 

Fig. 11 Experimental setup of two-stage counter-pumped RFA. The first stage and second stage are comprised of acoustically tailored fiber. A 3 W isolator (ISO) is inserted between the amplifier stages to protect against backward travelling light.

Download Full Size | PDF

According to the simulations presented above, the fiber length at seeding powers of >500 mW is <30 m. Consequently, we chose a fiber length of approximately 25 m. For all seed powers, there were no signs of spectral broadening for the two-stage counter-pumped RFA. At a seed power of 100 mW, we were pump-limited with an output signal power of 7 W. As expected, when the seed power was increased, the output power increased. At 500 mW of seed power, approximately 18 W of 1178 nm power was obtained. Still, in this case, we were operating below the SBS threshold (i.e. the output power was pump limited).

We found that the output power was near maximum when the second stage amplifier was seeded with 900 mW; which was slightly higher seed power than what the simulations indicate for a fiber of length 25 m. At this seed power, we were operating at SBS threshold while utilizing a little less than the maximum available pump power. The measured 1178 nm output power was 22.2 W. This output power was also slightly higher than what was obtained from the simulations, but well within margins of experimental error and estimates of the Brillouin and Raman gain coefficients. Increasing the seed power beyond this point actually led to a decrease in the output power as the SBS threshold was encountered at lower pump powers. At a seed power of 1200 mW, the output power was 18.5 W and was SBS limited. The plots for signal power vs. pump power for the different seed levels discussed above are shown in Fig. 12.

 

Fig. 12 1178 nm output power vs. 1120 nm pump power for several seed powers. The length of the RFA was ~25 m. For seed powers of 100 mW and 500 mW, the output is pump limited; however, the output for 900 mW and 1200 mW was SBS limited.

Download Full Size | PDF

Further power scaling can be achieved by reducing the length of the second-stage amplifier to an optimal value. This optimal length would provide for utilization of the maximum available seed power (~1.8 W) at the maximum available pump power. The simulations presented in Fig. 10 indicate that would provide a small enhancement of ~2 W over the 22.2 W obtained in the experiment. The 22.2 W output power represents approximately a 2x improvement over the 11.2 W power obtained from a single-stage acoustically tailored RFA. The latter result was reported in [9]. Both results were obtained without the benefit of the thermal gradient. As shown in [9], the power for the latter was increased to 18.3 W by application of a one-step thermal gradient. The application of multi-step thermal gradients to our two-stage acoustically tailored RFA for further power scaling is currently under investigation.

3.3 Co-pumped two-stage RFA

The experimental set-up for the co-pumped two-stage RFA is shown in Fig. 13. The first stage is identical to the counter-pumped configuration. However, the second stage is co-pumped with the 100 W 1120 nm laser. Two WDMs (WDM 4, WDM 5) were inserted after the pump to reduce the 1178 nm noise introduced by the pump as shown in Fig. 5. Similar to the counter-pumped setup, both lasers were aligned to the slow axis of the fiber, and the second stage utilized 25 m of the Raman enhanced acoustically tailored fiber with a WDM 6 at the output to separate the unabsorbed pump from the Raman amplified signal.

 

Fig. 13 Experimental setup of co-pumped second stage RFA. It is seeded through a counter-pumped RFA. Both stages are comprised of acoustically tailored fiber. A 3 W isolator (ISO) is inserted between the amplifier stages to protect against backward travelling light.

Download Full Size | PDF

Our experiments utilizing co-propagating single-stage RFAs pumped with an 1120 nm Raman fiber lasers have revealed considerable broadening of the 1178 nm signal. In those experiments, a maximum seed power of 15 mW was available to us. The broadening was observed at fairly low pump powers (< 20 W). As the pump power was increased, further broadening occurred. At 500 mW of output power, the spectral FWHM of the 1178 nm signal has broadened to 0.1 nm as observed on a high resolution OSA. It is not definitive to us the exact mechanism responsible for this effect. One possible source for this effect is the broadband 1178 nm noise introduced by the 1120 nm pump. Our characterization of the spectral content of the pump, which is presented in Section 2.2, revealed parasitic lasing in the proximity of 1178 nm due to second-order Stokes process (see Fig. 5). This broadband noise can potentially interact through four-wave mixing (FWM) with the amplified 1178 nm seed signal; leading to further spectral broadening. Even in a counter-pumped configuration, this parasitic noise is undesirable as it can potentially seed the SBS process. To find out if higher seed power and shorter fiber length can suppress the spectral broadening, we investigated the spectral content of the signal in the co-pumped RFA.

A shorter fiber may lead to a reduction in FWM. However, our studies revealed that spectral broadening still occurred. Figure 14 shows the results for a case in which the second stage RFA was seeded with 360 mW. The various spectra correspond to different output powers at 1178 nm and clearly indicate spectral broadening with increased pump power. Based on the secondary spectral peaks, one may infer a FWM process is occurring in the RFA. Further increase in our seed power had a marginal effect on mitigating the spectral broadening; in all cases the FWHM approached 0.1 nm when the signal output was of the order of a few watts. Therefore, at least for our experiments, this rendered co-pumping as unsuitable for generating 1178 nm for a guide star application.

 

Fig. 14 Spectral content near 1178 nm as captured on a high-resolution optical spectrum analyzer indicating spectral broadening as the pump power was increased in a co-pumped second stage RFA.

Download Full Size | PDF

To conclude this section, we believe, further investigations of the spectral broadening in a co-pumped RFA are warranted. While FWM has been investigated extensively in Raman fiber lasers [18], to the best of our knowledge scant information has been reported in the literature in this area when considering RFAs seeded with single-frequency sources. Finally, we point out that there are several techniques that can potentially alleviate the problem of introducing into the system unwanted 1178 nm light from the pump end. These techniques include the use of various fiber designs to filter out the unwanted light or the use of Yb-doped 1120 nm fiber lasers as pump sources.

4. Summary

In summary, we have conducted investigations of core-pumped single-stage and two-stage PM RFAs. Both co-pumped and counter-pumped configurations were considered. For the former, spectral broadening was observed; rendering the output unsuitable for frequency doubling in a resonant SHG cavity. No spectral broadening was observed for the latter configuration. For a counter-pumped single-stage RFA, COTS fiber was used to generate 10.1 W through the use of externally applied temperature steps to suppress SBS. In addition, a fiber cutback experiment revealed a linear dependence at SBS threshold on pump power in accordance with the theoretical predictions. Further power scaling to 22.2 W was achieved by using acoustically tailored fiber in a two-stage RFA system.