In [3], efficient nonlinear broadening was observed when applying 100% forward pumping. However with this kind of pumping without any Bragg grating in the laser design outlined in Fig. 1, the lasing was disrupted by cascaded Brillouin transition lines and dual wavelength oscillations. As a consequence, only a small power region was dominated by a single-wavelength lasing. This justifies the favorable selection of 100% backward pumping in this report where the single-wavelength lasing prevailed in almost the entire operation. The main objective of this work was to study the dynamics of spectral broadening at 1584nm operating wavelength by applying tight-confinement in the fiber dimension of random laser.
Before starting any experiment, the basis behind Raman waves interactions in this advanced setup should be understood first. The simplified diagram is shown in Fig. 2 where the feedback loop is replaced by a broadband mirror. In theory, wave kinetics [9] induced by net accumulations of strong backreflected waves originating from the mirror only happens in one direction. A significant broadening implies the dominance of these backreflected waves supported by backscattered Raman photons when propagating together towards the output port (see Fig. 5 in [3]). This is followed by saturation of the output power as most energy is utilized to compensate the energy conversion required for nonlinear broadening (see Fig. 4 in [3]). In contrast, its smaller bandwidth attainment signifies the struggle of these backreflected waves (green arrows) over the opposite direction of backscattered Raman photons emerging from the backward pumping waves (orange arrows). This is depicted in Fig. 2 and as a consequence, a stronger distributed virtual mirror is introduced at port A. Therefore together with the assistance of the broadband mirror that completes the laser cavity, most of the energy is consumed to produce higher lasing power instead of assisting the spectral growth. This justifies the insight behind weak generation of wave kinetics when specifically implementing this kind of pumping-arrangement (see Figs. 2 and 3 in [3]). The “weak” terminology is just a relative definition when referring to a more efficient nonlinear response in 100% forward pumping (compare Figs. 2 and 4 in [3]). However, this is already adequate to initiate early studies on FWHM growth with further prospects of improvements in the future endeavors.
Once the principles above are understood, detail studies on various lasing modes particularly in the nonlinear broadening regimes are presented. The bandwidths were evaluated by considering the full-width at half maximum, FWHM and we only focus on the important developments since some elaborations have been covered in the past reports [3, 8, 9]. In random lasing, it was very difficult to differentiate the boundary between CW to other types of operation especially mode-locking owing to the modeless interacting waves. To resolve this issue, there were a few states that we described ourselves depending on the bandwidth values. The first was beyond 4nm band that is identified as wider spectra, “WS” and the second was spectral broadening effect “SBE”, both are discussed in setup 1. The first-state was realized by immediate spectral increase at a certain pump power level as depicted in Fig. 3. This was recognized by a sound “click” that also implies a switching power transformation as a substantial output power drop was produced. All of these occurred synchronously in the stable power region. However, no any audio was heard at the end of its performance that was represented by immediate bandwidth reduction accompanied concurrently by a significant increase in output power. For the second state of “SBE”, the bandwidth expanded slowly from 1.76 nm up to 3.53nm range as proven in Fig. 3. The flow almost satisfies the weak wave kinetics theory as outlined in Fig. 2 where no audio change to signalize its initial and end-phases. Sometimes a wider spectral regime was characterized as “UWS” in Fig. 7 where character “U” denotes its occurrence in the unstable power zone. As output power fluctuations were developed in this case, the bandwidth widened and shrunk accordingly at the same time which clarifies the hardship to complete the measurements.
Throughout experiment, the RPU pump power that was stable at all levels within ± 1mW was increased gradually up to 1 decimal point. The lasing behaviors namely output power and its stability in the corresponding pump power domain were observed carefully between 3 to 5 minutes. All data shown in Figs. 3 to 7 fell in the stable regimes that correspond to output power variations within ± 1mW. The unstable power performance, namely the “UWS” state is only presented in Fig. 8 together with its specifications in Table 3. At both laser setups (Figs. 3 and 7), stimulated Brillouin scattering (SBS) lines were sometimes observed on the Raman gain bandwidth near threshold. This led to instability of the CW regimes as manifested by unstable generated power. Despite of this, the Raman gain transition dominated over the SBS lines with the increase in power generation above threshold. This indicates the total removal of gain competition between both nonlinearities during random lasing especially when implementing full backward pumping arrangement (Fig. 1). Once this was attained, improvements to the output power stability were realized which was similar to that elaborated in [1]. Moreover the utilization of a CW, single-mode RPU in this experiment indicates no requirement for a multimode pump source to suppress the SBS generation in conventional designs [26, 27]. This shows some of the advantageous nature of random lasing mechanism. Thus to understand this physical concept, the quality of spectral broadening is elaborated.
At the outset of this assessment, the results of output power and bandwidth progressions for setup 1 are presented in Fig. 3. A power threshold, \({P_{th}}\) of 814mW was satisfied from the quadratic curve of the lasing plot (black line). Initially two output power plots at 116 and 118mW (green diamond) at the end spectral narrowing region (orange circle) deviated from the lasing trendline before jumping up to 199mW when the pump power was set to 1190mW. This might be because the Raman energy was not enough to compensate losses introduced by the DCF until certain pump power was achieved. After this level, the power almost complied with the nonlinear quadratic fit without showing any saturation sign which is in consistency to that described earlier (see Fig. 2 in [3]). Maximum output power, \({P_{out}}\) of 489mW was attained that yields 29% optical-to-optical efficiency with respect to the pump power. Simultaneously, the bandwidth development was also inspected thoroughly on the OSA. Both settings for video bandwidth, VBW. of 2.1kHz and sensitivity, Sens. of -65dBm were maintained at the same levels for the entire operation. In particular, the main topics of interests are namely on “WS” and “SBE” regions that are illustrated in Fig. 4 to 6. Below threshold, the spectral envelope consisted mainly of amplified spontaneous emission (ASE) that explained a wide bandwidth development of more than 10nm. Once lasing begun, the spectral wing envelopes were slowly dominated by stimulated Raman transitions that were responsible for the narrowing effect (orange circle). The beginning state of “SBE” was realized at 1190 mW pump power where the bandwidth increased to 1.76nm (brown circle). This signifies the weak initiation of wave kinetics (Fig. 2) as the optical bandwidth expanded slowly to 2.88nm with the continuous increment in pump power domain up to 1390mW. Both bandwidth behaviors were found to be in accordance to the predicted theories (see Fig. 3(b) in [9]).
Unpredictably beyond the physical concept outlined in Fig. 2, when the pump power \({P_p}\) was enhanced slightly from 1390 to 1417.6mW a noticeable “WS” regime was started (Fig. 3). Almost 32mW output power drop was measured within this small order of pump power increase. A sound “click” was also heard together with the concurrent commencement of 1.7 times wider spectrum. This is demonstrated by the blue profile in Fig. 4(a) where the maximum bandwidth, BW of 4.88nm was produced. The \({P_{out}}\) was 290mW that relates to 20.5% optical-to-optical-efficiency \(\eta\). The central wavelength was at 1584.2nm where the “WS” state was stable. The distinctive “WS” regime only covered a small pump power portion from 1417.6mW to 1529.8mW as represented by triangle plots in Fig. 3. At 1520mW pump power, a spectral width of 4.05nm was measured at a central wavelength of 1584.1nm [Fig. 4(b)]. The \({P_{out}}\) was 353mW that corresponds to \(\eta\) of 23.2%. We need to stress that very careful observation was performed with the gradual increment in the pump power where the “WS” signal was still in effect at \({P_p}\) = 1529.8mW although this was not recorded. Just a minor increase in the pump power to 1529.9mW resulted in nearly 1.4 times bandwidth constriction. This is depicted in Fig. 4(b) when the blue spectrum evolved into that of the red one. Synchronously, this was followed by a substantial increase of 56mW output power which signified the end of this state. As a result, the output power improved drastically to 409mW with a smaller FWHM of 2.85nm. The complete representation of Fig. 4 in dBm peak power is demonstrated in Fig. 5. The main idea of showing the latter diagram is to prove that the “WS” state was primarily induced by CW spontaneous radiation without fluctuations or Q-switching instabilities at the band edges. From both figures as no any Bragg reflector was placed inside the cavity to discriminate lasing at another peak wavelength, a secondary peak around 1594nm was observed. It was more pronounced in the “WS” modes due to the stronger broadening in comparison to that in other modes. Besides this, the transition between “SBE” to “WS” and vice versa as shown in Fig. 3 is not understood well yet owing to our lack of expertise and facilities. In addition, as no any external mode-locking modulations as explained in [15, 22, 23] were inserted to this cavity-free structure, no ultrashort pulses that become the subject of our main interest were detected similar to that assessed earlier (see Fig. 5 in [17]). The pre and post “WS” state that was classified as “SBE” almost satisfied the weak wave kinetics signature as described in Fig. 2. This is owing to the overall bandwidth development from 1.76 to 3.53nm as manifested in Fig. 3 (brown circles). Other spectral changes with the absence of “WS” signals are depicted in Fig. 6. Although minor ripples were the feature for “SBE” category, smoother beam curves were the typical attribute at a lower power segment (\({P_p} \leqslant\)1160mW). From this figure also the precise central wavelength, \({\lambda _c}\) that begins at 1583.13nm shifted 1.38nm to the right with the increase in \({P_p}\) from 843 to 1690mW. The main reason is believed due to thermal effects (see Fig. 2(b) in [29]) which explains the generation of asymmetrical spectral shapes in the previous assessment (see Fig. 2(a) in [3]).
From [3] that demonstrates a 100% forward pumping layout, the pumping enhancement up to 1585mW led to the maximum bandwidth of 0.9nm. This is almost 2.8 times wider compared to the 3dB reflection band of the FBG. In the similar pumping arrangement, the spectral broadening from 0.11 to 0.96nm was developed with the pumping increment from 14mW to 2.88W [15]. In other assessments, spectral broadening at the first Stokes wave emission was lower than 1.8nm [9–14]. The spectra were limited to below 2.6nm and 3.5nm at second-Stokes [13, 14] and third-Stokes components [13], respectively. These indicate that above 1.76nm FWHM could only be fulfilled in the CW-pumped [3, 9–13, 15] and ASE-pumped RRFLs [14] at higher-orders with multiple Watts output power generation. The exception to this bandwidth rule was observed in later advancements at 1st-Stokes order [16, 17]. A stable statistical mean value of 14nm spectral width was achieved in an open cavity RFL, but this is only a ratio of 0.875 to the linewidth of the incoherent ASE pump source [16]. Although up to a maximum of 7.31nm spectral width was attained in [17] but it is only 3.8 times wider compared to the reflection bandwidth of the FBG. Both results also necessitate high output power scaling from below 10W [16] to over 200W levels [17]. Nevertheless, the superior low power laser design of setup 1 was proven for better applications that favor nonlinear broadening at the fundamental Stokes waves emission from 1.76 up to 4.88 nm range. The latter implies exceptionally 6.3 times more expansion than the linewidth of the pump source where no any spectral selective element was employed. This report represents the best broadening proportion in any CW and ASE pumping based random fiber lasers at just within hundreds mW output power levels. In fact this claim is justified by taking into account that no pump-pulse synchronization or external mode-locking modulations were used as incorporated in past attempts [15, 22, 23].
Next, the results for setup 2 that comprised no DCF (Table 1) are illustrated in Fig. 7. The threshold power \({P_{th}}\) of 760mW was satisfied where the output power plot conformed to the linear trendline estimation (black line) without any apparent deviation. This is in contrast to the quadratic power behaviors in the prior setup. In addition, no series of “WS” state and its corresponding power drop and rise from the linear fit was initiated. For the bandwidth study, we concentrate only on the development beyond 957mW pump power as represented by the brown circles. Initially, smooth broadening transition was observed from 1.58 to 2.22nm. This was followed by an inconsistent flow from 2.28 to 3.3nm range at the middle power scheme. Another segment above 1429mW pump power also did not obey the wave kinetics owing to saturation as most of the energy was used for the conversion to higher lasing power. Nevertheless, the maximum BW in this power portion is similar to that obtained in Fig. 3. No stable “WS” state was realized but the unstable one, “UWS” was achieved in the \({P_p}\) segment that is denoted by the blue arrow in Fig. 7. It should be acknowledged again that all \({P_p}\) levels were stable within ± 1mW estimation at all time. It was the respective output power that fell in unstable regimes. Therefore, it was very difficult to observe the significant \({P_{out}}\) drop and increase from the linear fit that signify the beginning and end of this characteristic. Without the tight DCF geometry, the scheme was struggling to control its stability. The depictions on this behavior are given in Fig. 8. Prior to this, the initial BW was 3.3nm at \({P_p}\) = 1580mW before undergoing broadening that reached a maximum up to 4.25nm at \({P_p}\) of 1587.4mW. This is the widest record that we managed to acquire manually because the formation and the elimination of “UWS” state took place immediately at the same time. The FWHM was narrower than that obtained in the previous setup and this is expected by referring to the net GVD listed in Table 1. The process started from the initiation of q-switching instability as manifested by the band-edges of a few signals in Fig. 8. Our main aim was not to measure the pulses and peak power initiated by Stimulated Brillouin Scattering, SBS as this has been detailed in [30]. In addition to power fluctuations, the spectral widths also kept changing and both properties are described in Table 3. Sometimes narrow linewidth Brillouin lasing lines popped-up “ON” and “OFF” on the lasing spectra. The BW continued to decline with the continuous pumping in a similar behavior to the triangle plots in Fig. 3. In comparison to “WS” state, the categorization of “UWS” did not only depend on above 4nm bandwidth definition. This also included the dynamics of indispensable q-switching response observed on the lasing signals that thwarted further broadening. Owing to this reason, the inconstant signals below 4nm-order are still classified in “UWS” category [see Fig. 8(c to e)]. This was totally terminated when the pumping was enhanced to 1661.5mW. The FWHM was reduced to 3.31nm together with stabilization in output power to 538mW. To avoid any confusion, it should be emphasized again that the data in the stable region is presented in Fig. 3 to 7 and the unstable data denoted by the blue arrow in Fig. 7 is specified only in Fig. 8 and Table 3.
Table 3
“UWS” state properties in setup 2.
Lasing state | \({P_p}\) (mW) | \({P_{out}}\) (mW) | BW (nm) |
Stable mode | 1580 | 484 ± 1 | 3.3 |
UWS | 1587.4 | 494 ± 7 | Fluctuates from 4.25 to 3.525 |
UWS | 1590 | 502.5 ± 3.5 | Fluctuates around 3.425 |
UWS | 1625 | 518.5 ± 19.5 | Fluctuates around 3.425 |
Stable mode | 1661.5 | 538 ± 1 | 3.31 |
For concise explanations, the summaries are presented in Table 4 below. As references in [3–5, 8] verified the similar behavior between random lasing to that of a conventional one, we assume that a reasonable analogy to Kerr-lensing effects in solid-state lasers [24, 25] can be drawn between both lasing criteria. From the experimental setup (Fig. 1), the pump beam mode is restricted in the gain medium owing to the filtering scheme introduced by two WDMs (WDM1 & WDM2). The residual pump light is removed out through the corresponding port of WDM1. With the formation of virtual cavity during lasing, the laser beam mode propagates in the entire oscillator and overlaps with the pumping mode in the gain medium. The smaller core DCF provides assisted reshaping of the laser beam diameter as a function of intensity along the fiber longitudinal dimension. The purpose is to facilitate self-focusing and efficient gain guiding influences that afford an improved overlap (mode-coupling) between the laser beam and the pump mode volumes. Owing to this, the reduction of the beam spot size is realized that serves to enhance the radial and temporal-dependent intensity,\(I(r,t)\) of the propagating waves (see Fig. 3 in [25]). As a result, increases in the self-amplitude modulation (SAM) along with self-phase modulation (SPM) and self-focusing are satisfied. The SPM generates exploitable extra bandwidth between the modeless beating waves. Together with the relatively more optimized net GVD in the first setup (Table 1) justifies the attainment of better spectral width values. This agrees well to the simulation in [18] that described the favourable properties for spectral broadening with the improvement in Kerr-nonlinearity, \(\gamma\) and the minimization in fiber dispersion, \({\beta _2}\). With the strengthening of \(\gamma\) induced by the tighter geometry also results in the initiation and stabilization of the “SBE” as well as the “WS” states. For the latter, its common trait to that of “UWS” was almost nice and smooth spectral shapes without Kelly sidebands (Figs. 4, 5 and 8).
Table 4
Summary of lasing characteristics at different configurations.
Parameters | Setup 1: Fig. 3 | Setup 2: Fig. 7 |
Fiber-geometry | Tighter confinement (stronger \(\gamma\)) | All ordinary SMF fiber (weaker \(\gamma\)) |
Output power flow | Almost quadratic fit. | Almost linear fit. |
Wave kinetics response | Smoother trendline for “SBE” state from 1.76 to 3.53nm band | Overall ups and down from 1.58nm to 3.53nm as well as suffering saturation. |
“WS” & “UWS” states | Stable zone, BW range = 4.05 to 4.88nm. \(\eta\)= 21–23% | Unstable zone, max. BW ~ 4.25nm. Difficult to maintain this state. |
In solid-state lasers (SSLs), a tight crystal to mirror length tolerance is needed for the optimization of \(\gamma\) [see Fig. 3(a) in (28)]. Once this is satisfied, instabilities to CW power operation is initiated which is comparable to “UWS” state in this laser scheme. Further adjustment of the free-space optics together with a small mechanical perturbation lead to stable operation. This is similar to the realization of “WS” in this report that requires no necessity for physical or optical adjustment. Although involving only a small power segment, this was just the first step that inspires a more promising scientific breakthrough. We only introduce analogy to SSLs but do not claim of any presence of some strict stability map in random lasers. This text also does not cover the behaviours of optical rogue waves presented in [16] that might hints the underlying physics behind “WS” and “UWS” states formation. The main scopes are on the exceptional spectral growth characteristics and stabilization at low power scaling by including discussions on a fraction of commonly known nonlinear fiber optics theories. More investigations and mathematical analysis need to be done in the future to gain better understanding of the insight behind this phenomenon. It should be stressed again that the nature of the experimental design itself, comprising no FBG permits the freedom of wave kinetics assisted spectral improvement from narrower to broader widths as clarified in Fig. 2. The tighter geometry eases and boosts the creations of all stable broadening states. In fact through multiple attemps, we ourselves have proven experimentally the repeatability and reliability of the results obtained.