Scheduled System Maintenance on May 29th, 2015:
IEEE Xplore will be upgraded between 11:00 AM and 10:00 PM EDT. During this time there may be intermittent impact on performance. We apologize for any inconvenience.
By Topic

• Abstract

SECTION 1

## INTRODUCTION

The main criteria that determine laser properties in conventional lasers are the appropriate choice of the gain medium, as well as the resonator design [1]. The gain medium is required to provide amplification via stimulated emission, and the corresponding modes structure is produced after one cavity round trip. In addition, the spatial confinement of circulating light in the laser structure is realized with the inclusion of high reflectivity mirrors at both cavity ends. Once the integral balance between gain and intracavity losses is induced, lasing oscillation is established. It has a specific direction of emission which is dictated by the existence of a visible optical cavity.

However, random lasers are the simplest sources of stimulated emission without the requirement for the standard cavity arrangement. This is because their optical feedback is created in two methods. The first is based on multiple scattering effects within an active volume and the second depends on light reflection from sample surfaces. In 1960s, Letokhov et al. had pioneered the invention of this new type of laser by exploiting negative resonant absorption [2] and nonresonant feedback [3]. Since then, this research has progressed well with several advancements in the laser design [4], [5], [6], which can be promising attempts in competing with the conventional lasers. In fact, the concept of “random lasing” refers to the emitted radiation that is stochastically distributed over the whole solid angle. It has several advantages, particularly its simple technology that eliminates the formation of a precise microcavity in diode lasers construction. Additionally, the optimum temperature to fabricate a powder-based substrate is only at 500 °C approximately, in comparison with the fabrication of an ordinary laser crystal that requires elevated temperatures above 700 °C.

The same principle is also utilized in a Raman laser that involves a considerable stimulated Raman scattering (SRS) transition. The fiber waveguides provide a tight optical confinement in a very small core diameter of less than 10 $\mu$ m that favors a high-intensity intracavity light propagation. This is further supported by the low loss coefficient $\alpha\sim 0.2\ \hbox{dB/km}$ at 1550 nm. The induction of multiple Rayleigh backscattering (RBS) effects owing to microinhomogeneities in the fiber composition leads to the formation of a randomly distributed reflector. This has been proven in preliminary research reports [7], [8] that incorporated tens of km fiber length as a gain medium. The realization of this extended cavity laser could prove extremely useful for remote sensing applications. The sensing variations could be provided by the changes to the fiber Bragg grating (FBG) or the RBS feedback. Although the influence of RBS in the fiber core is extremely negligible, it is crucial over a very long distance. Recently, the operational characteristics of an 83-km Raman fiber laser in an open cavity was successfully elucidated [9]. In open cavity, substantial length of fiber was needed to produce sufficient RBS to induce lasing, which achieve laser slope efficiency of only about 30% due to the random weak distributed feedback [9]. In comparison, the slope efficiency of conventional FBG-based cavity was 72% with decreasing output power at longer fiber length due to higher cavity loss [10]. Deployment of a single FBG reflector with RBS feedback at the other end has also been performed, which achieved modeless laser with narrower bandwidth [8]. The fundamental clarification of scattering processes within this waveguide contributes further to the development of this particular area. Therefore, in this paper, we demonstrate a 51-km Raman laser that incorporated various pump coupling distributions along the fiber structure. In our previous study, we have already looked into the broadening effect associated with random lasers [11]. In this laser scheme that consisted of a FBG at one cavity end, its detailed performances in other aspects especially threshold levels and output powers are discussed instead with respect to the changes in pumping percentages.

SECTION 2

## EXPERIMENTAL SETUP

In this assessment, the ultralong Raman fiber laser (ULRFL) was pumped by a Raman pump unit (RPU) that operated at 1455 nm wavelength. The RPU that has a maximum pump power of 1585 mW was spliced to the fiber through two 1480/1550-nm wavelength selective couplers (WSC1 & WSC2) according to its pumping arrangement. In the middle of these components, a set of couplers was inserted individually as illustrated in Fig. 1 to divide the pump power radiation along the fiber in forward and backward directions. In this case, the term “forward” and “backward” relate to the pumping direction with respect to the FBG position. The couplers' properties were 10/90%, 20/80%, 30/70%, 40/60%, and 50/50% where with these inclusions the power percentage was enhanced toward backward pumping from 10% up to 90% with 10% increments. However, for 100% pumping, the coupler was excluded, and the RPU was connected directly to either WSC1 or WSC2 for maximum pumping in the forward or backward direction, respectively.

Fig. 1. Schematic diagram of an ULRFL in the forward Pf and backward, Pb pumping arrangements. The pump directions are represented by the red arrow (forward) and the yellow arrow (backward). A series of coupler with ratios of 10/90, 20/80, 30/70, 40/60, and 50/50 % was connected to the WSC's with the increased percentage towards WSC2.

To simplify this, the term coupling ratio $(CR)$ was introduced, which can be expressed as TeX Source $$CR = {P_{b} \over P_{T}}\eqno{\hbox{(1)}}$$ where the percentage for backward pumping is denoted by $P_{b}$, where $P_{T} = 1$ is the total absolute value. $P_{T}$ is composed of the addition between $P_{f}$ and $P_{b}$, where $P_{f}$ signifies the forward pumping percentage. From (1), the increase of $CR$ from 0 to 1 with a 0.1 increment represents the continuous addition of $P_{b}$ from 0% to 100% with 10% steps, respectively. Meanwhile, the expansion in $P_{b}$ lead to the corresponding decrease of $P_{f}$ from 100% to 0% with similar intervals.

In this asymmetric laser scheme, the FBG that has a high reflectivity $HR \sim 98\%$ at 1553.3 nm acted as the first mirror. It was placed next to WSC1 and has 3- and 20-dB reflection bandwidths of 0.32 nm and 0.454 nm, respectively. The cavity was completed by providing another feedback initiated by multiple scattering reflectors. This was done by utilizing a 51-km single-mode fiber (SMF) as a gain medium. The extremely long SMF implies inherent RBS effects that behaved as a distributed random mirror to support laser operation. Two isolators were utilized together with an angle-cleaved fiber end to eliminate Fresnel reflection or backreflected light to the optical elements [8]. These ensure the propagation of Raman and remaining pump photons in only a single direction.

Once the Raman gain has exceeded the threshold required to overcome the intracavity losses, lasing oscillation was started. For the entire assessment, an optical power meter (OPM) or an optical spectrum analyzer (OSA) was positioned at port (A) in Fig. 1 to measure the corresponding laser parameters. For a better spectral evaluation, the OSA was set at a resolution around 0.02 nm. The resulting first Stokes wavelength was close to 1555 nm, which agrees well with the 13-THz SRS shift in the silica fiber.

SECTION 3

## RESULTS AND DISCUSSIONS

At the outset of this evaluation, variations in threshold powers were investigated when varying the pumping distribution along the two fiber end-facets, as illustrated in Fig. 1. Similar to regular lasers, the threshold condition in this laser scheme is also characterized by suppression of a relatively broad spontaneous emission spectrum to a single line that corresponds to the FBG wavelength. As a result, the output power generated increased by several orders of magnitude where the threshold level can be estimated from the x-intercept of the graph linear fit. The results are demonstrated in Fig. 2 where an increment was observed at the first eight pump CRs. With the implementation of 100% forward pumping scheme $(CR = 0)$, a low threshold operation of 788 mW was achieved. After this, the threshold continued to expand slowly until $CR = 0.7$ where from that point onwards, the threshold was maintained around 1200 mW, as demonstrated in Fig. 1.

Fig. 2. Threshold power as a function of coupling ratio CR. The red line indicates the curve fit of this graph.

Theoretically, a threshold begins once the balance of gain and loss in the systems is met. In Raman lasers, the total amount of gain is mainly supplied by the vibrational energy of a molecule. This is proportional to the absorption of pump photons that results in subsequent emission of Raman photons. The higher ratio of the remaining Raman photons diffused in the fiber results in higher inelastic scattering. The main feedback in this laser structure was mainly contributed by 1553.3-nm amplification provided by the FBG. It was then enclosed virtually at the other end by multiple scattering effects inside the 51-km SMF. The loss is caused by the light transmission through the sample surface and the spatial distribution of gain is governed by the spreading of pump photon in the laser structure. Thus, the variation of CR from the 0 to 1 at 10% increase implies the reduction of pump photons at the start of the forward pumping input. As a consequence, gradual decreasing in amplification of 1553.3 nm waves at the FBG was introduced, which explains the increment of threshold powers, as manifested in Fig. 2. In addition, the propagation distance at which light is randomized also influences the attainment of low threshold levels. As the distance required for the photons to travel to the FBG in the forward pumping configuration was shorter than the backward pumping, this also contributes to its lower threshold characteristic.

Next, the output powers at all pumping distributions were analyzed thoroughly. These parameters grew continuously from the initial value of 67 mW to achieve a maximum of 220 mW. From Fig. 3(a), this phenomenon was observed for the expansion in backward pumping from 0% to 60%. The spectra presented in Fig. 3(b) illustrate the growing peak power as the CR was varied. Owing to this fact, the gradual improvement of RS effects at the final fiber length $L_{f}$ illustrated in Fig. 4 provided a more influential distributed reflector. Together with the FBG that initiated main amplification at 1553.3 nm, an optimized resonator was realized, which justifies the maximization of lasing power at this specific design $(CR = 0.6)$. After this condition, the power maintained around 176 mW for the next three pumping stages before increasing back to 194 mW at 100% backward pumping. The trend of output power growth for $CR = 0.6$ to 1 was confirmed to be accurate after repeating this evaluation for a few times. This is believed to occur due to the interactions between the backscattered waves in the modeless regimes that at the moment has yet to be understood. In fact, the enhancement of backward pumping from 70% to 100% that implies a stronger RS reflector at $L_{f}$ was compensated by a decline in forward pumping portions from 30% to 0% at $L_{o}$. This weakened the amplification of the first Stokes Raman photons at the FBG within this range, which describes their lower power generation. As the fiber length implemented in this assessment might also be another factor that influences this effect, further study is suggested by implementing a set of SMF with lengths up to 200 km.

Fig. 3. (a) Comprehensive power development at all pumping ratios and (b) measured optical spectra for different ratios at 0.4, 0.5, and 0.6.
Fig. 4. Miniaturized propagation of Raman photons in the gain structure. L0 and Lf represent the fiber length at initial and final positions, respectively. The thickness of backscattered Raman photons within the 51-km fiber closed-loop signify their corresponding intensities.

From Fig. 5(a), which indicates the highest power scaling generation when utilizing $CR = 0.6$, a threshold was satisfied at 1100 mW. This agrees well with the predicted value suggested in a preliminary report [8]. The maximum output power was 220 mW, which corresponds to an optical-to-optical efficiency of 14%. The slope-efficiency estimated from the linear fit of the graph was in the vicinity of 44% as evident in Fig. 5(b). In general, these overall laser characteristics are reasonable, which shows the potential of this novel type of laser to challenge conventional lasers [10]. In a full analogy to the power progression in Fig. 3, the slope-efficiency also shows improvement with a quadratic increment for the first seven pumping arrangements, as depicted in Fig. 5(b). Then, at the last four pumping stages, the differential efficiency changed to a linear fit before reaching the top value at 51%.

Fig. 5. (a) Continuous-wave lasing performance at CR = 0.6 and (b) the slope-efficiency improvement against pump coupling ratio. The red lines represent linear and quadratic fit of the graph.

For CR larger than 50%, the performances of this laser configuration are better than [9], which has higher lasing threshold of $\sim$1600 mW and slope efficiency of only 30%. This can be attributed to the use of FBG which has higher reflectivity and leads to lower cavity loss. Moreover, substantial length of fiber (84 km) with high pump power was needed in order to produce sufficient RBS feedback for lasing [9]. The employment of FBG at one end compensated for the lower RBS feedback at the other end and allowed lasing to occur in shorter cavity at lower pump power. Additionally, the narrow reflectance bandwidth of the FBG limited the spectral development of the laser [8].

SECTION 4

## CONCLUSION

In this research assessment, the exploitation of variable pumping distribution along two fiber-end-facets with respect to the FBG position resulted in numerous laser properties. These were threshold levels, output powers, and slope-efficiencies, which can be adjusted to satisfy special requirements from low threshold to high power operation. A low threshold is dictated by a large fraction of available Raman photons that reaches the FBG to initiate significant laser amplification. This is also supported by the reduction of the photon travel distance $L_{d}$ in the microscopic laser structure. In addition, the best power scaling generation is influenced by an efficient formation of multiple scattering reflectors at the end of the fiber length $L_{f}$, together with the amplification provided by the FBG. For this specific fiber length, a correct combination between 40% forward and 60% backward pumping ratio is required to increase the amplitude feedback. In the future, to provide an insight to this physical issue, a more advanced design that incorporates longer gain medium and larger pump power can be implemented. Careful consideration has to be given to the fiber length to balance the need for larger RBS feedback while minimizing cavity loss.

## Footnotes

This work was partly supported by the Ministry of Higher Education, Malaysia, under High Impact Research Grant #A000007-50001 and the Universiti Putra Malaysia under Research University Grant Scheme 05-01-10-0894RU, postdoctoral research fellowship, and graduate research fellowship schemes. Corresponding author: M. A. Mahdi (e-mail: mdadzir@eng.upm.edu.my).

## References

No Data Available

## Cited By

No Data Available

None

## Multimedia

No Data Available
This paper appears in:
No Data Available
Issue Date:
No Data Available
On page(s):
No Data Available
ISSN:
None
INSPEC Accession Number:
None
Digital Object Identifier:
None
Date of Current Version:
No Data Available
Date of Original Publication:
No Data Available