1. Introduction

Tunable Raman lasers are highly desirable for use in numerous applications including biomedical diagnostics, spectroscopy, and the sensing of gases, strain, and temperature [1–3]. All previous attempts to achieve stable and tunable Raman emission utilize a changing magnetic field, tunable pump diodes, and optical tuning components [4–6]. However, such systems are rather complex, which limits their size, simplicity, and cost. These limitations make it difficult to construct efficient, compact, and affordable laser systems.

A cascaded resonating Raman fiber laser scheme is a promising alternative for achieving tunability. Cascaded Raman lasing has been previously reported in silica-based fibers and microresonators [7,8], and more recently in semiconductors driven both electrically and optically with low thresholds and high efficiency [9–12]. In addition, such devices can be fabricated with sufficiently small dimensions using cavity couplers coated directly onto both ends of the fiber [13,14]. However, the monolithic architecture of cascaded Raman lasers hinders the implementation of conventional tuning elements for tailoring cavity modes. While temperature is a tuning mechanism for mode selection, it is difficult to obtain enough resonant wavelength shifts in this way due to the relatively small thermo-optical and thermal expansion coefficients (TEC) of the Raman gain medium.

In this study, we propose and demonstrate the first realization of strain-activated Raman tuning based on a Cr⁴⁺:Y₃Al₅O₁₂ (Cr⁴⁺:YAG) crystalline core fiber packaged with a Cu-Al alloy. By simply adjusting the temperature of the Cu-Al alloy, a relatively large thermal stress can be easily achieved, resulting in both a strain-induced change in the stimulated Raman shift and a strain-enhanced piezospectroscopic effect. We also show the potential of this scheme for highly sensitive temperature measurements. A piezospectroscopic coefficient of 0.273 nm/^oC was found, a value at least one order of magnitude higher than that found in existing bare silica fiber-based thermal sensing elements [15–21]. In addition, due to the enhanced piezospectroscopic effect from the Cu-Al alloy, a temperature-strain crosstalk of 530 με/^oC was verified over a large dynamic range of 4.24 × 10³ με. This crosstalk value is an order of magnitude higher than that found in polymer fiber-based strain sensors [22]. Moreover, when coupled mirrors are directly coated onto the ends of the crystal-glass hybrid fiber ends, which serves as the resonant cavity for the Raman pump and signal, the Raman threshold was significantly reduced. The presented Raman threshold of 215 mW is the lowest both in continuous wave (CW) Raman crystalline media and fiber forms. In this way, CW tunable Raman operation, an efficient cascaded process with low-threshold lasing, and improved thermal and strain sensing can be obtained.

2. Experimental

First, a 68 μm Cr⁴⁺:YAG single crystalline fiber was drawn from a 500 × 500 μm² bulk ingot using the laser-heated pedestal growth (LHPG) technique [13,14,23–25]. The fiber was then put into a fused silica hollow tube to form a double-clad structure with a 20 μm diameter core using the same LHPG system. The Cu-Al alloy was packaged at 780 ^oC in order to facilitate heat dissipation and enhance the strain-activated Raman shift of the crystalline fiber laser. This process was followed by an annealing treatment accomplished by cooling slowly at a rate of 1 ^oC/min down to room temperature. Both faces of the as-grown Cr⁴⁺:YAG double-clad crystalline core fiber were then mechanically processed before application of the dielectric TiO₂/SiO₂ coatings used to form the laser cavity. TiO₂ and SiO₂ were chosen due to their large difference in refractive index, offering excellent antireflection for the excitation and broadband high reflection near 1.4 μm. The input and output reflectance of the laser coating employed in the experiment were 99.8% and 96.0%, respectively.

Figure 1(a) shows the schematic of the Raman laser described in this work. To characterize the Raman lasing performance, a single-mode fiber (SMF-28) was used as the CW 1064-nm pump delivery fiber butt-coupled to the Cr⁴⁺:YAG fiber crystalline core through a dichroic-coated input face. A 1060/1550 coupler connected to an optical spectrum analyzer was used to detect backward lasing behavior. The Cu-Al-alloy-packaged laser cavity of a Cr⁴⁺:YAG crystalline fiber was attached to a thermoelectric cooler. The inset of Fig. 1a shows end views of a Cu-Al-packaged Cr⁴⁺:YAG crystal-core glass-clad hybrid fiber with and without a dielectric coating. After the LHPG process, the atomic-scale structure of the crystalline core was investigated by high-resolution transmission electron microscope (HRTEM). HRTEM observations were performed with a FEI Tecnai G² F20 FEG-TEM operating at 200 kV. Figure 1(b) shows an [111] HRTEM image of the core region. The image shows a hexagonal-like pattern with a close-packaged configuration indicative of the shape of a [111]-YAG crystalline core. Figure 1(c) shows the corresponding selected area electron diffraction pattern, and the sharp bright diffraction spots demonstrate that the crystalline core has a perfect YAG single-crystal structure grown by the LHPG technique, as confirmed by the measured lattice parameter a = 12.008 Å (JCPDS file 33-0040).

 

Fig. 1 (a) Schematic setup of the Raman fiber laser. Insets are photomicrographs of the fiber ends with and without a dielectric coating, showing that the periphery is in good contact with the Cu-Al-alloy packaging. (b) [111]-HRTEM image of the crystalline core. The scale bar is 5 Å. (c) The corresponding selected area electron diffraction pattern of (b).

Download Full Size | PDF

3. Results and discussion

3.1. Strain-activated SRS shifts

Figure 2(a) shows representative Raman spectra of Cr⁴⁺-doped and undoped YAG in bulk form and the Cr⁴⁺:YAG hybrid fibers using a 532-nm confocal Raman spectroscope. All samples were probed at 24°C, with the exception of the Cu-Al-packaged hybrid fiber, which was probed at 16°C. This result clearly shows that the first Raman peak of the Cu-Al-alloy-packaged Cr⁴⁺:YAG crystalline-core fiber occurs at 123.65 cm⁻¹, which is consistent with the corresponding value (approximately 123 cm⁻¹) of the cascaded anti-Stokes (AS) Raman shift of a 1414-nm Cr⁴⁺:YAG laser (Fig. 2(b)). The ordinary 1414-nm lasing excited by 1064-nm light serves as the Raman pump. The 1414-nm emission is primarily determined by the highly reflective wavelength of the broadband mirror coating. For the undoped YAG in bulk form, the corresponding peak occurs at approximately 144 cm⁻¹ [26]. Note that water absorption in this Cr⁴⁺:YAG crystal-glass hybrid fiber may lead to the first peak being favored over the stronger nearby Raman peak. This preference may arise because the inner cladding of the Cr⁴⁺:YAG crystal-glass hybrid fiber is a polycrystalline structure [27] primarily composed of amorphous SiO₂ even though the whole cavity is made of Cr⁴⁺:YAG crystal.

 

Fig. 2 (a) Strain-dependent Raman spectra of Cr⁴⁺:YAG in the bulk and fiber forms. (b) Cascaded stimulated Raman generation of the Cu-Al-packaged crystal-core glass-clad hybrid fiber, corresponding to the first Raman shift in (a).

Download Full Size | PDF

Another important conclusion drawn from Fig. 2(a) is that the transition of the first Raman peak shows strain-dependent behavior. The blueshift of the first Raman peak in the bulk and fiber forms of Cr⁴⁺:YAG relative to the peak of the undoped bulk YAG shows that the YAG matrix undergoes considerable strain [27]. This strain can be attributed to Cr⁴⁺ dopants replacing Al³⁺ ions at tetrahedral sites. Cr⁴⁺ and Al³⁺ ions have effective ionic radii of 0.041 and 0.039 nm, respectively, with a coordination number of 4 [28]. The 123.65-cm⁻¹ Raman shift of the Cu-Al-packaged crystal-glass hybrid fiber corresponds to the largest strain. This strain results from the applied stress from the large TEC difference between the crystalline YAG core and the Cu-Al alloy packaging. This difference in TEC makes a large temperature-dependent Raman shift possible, as is discussed later.

3.2. Temperature-dependent SRS tuning

Figure 3 compares the crystalline core fiber laser performance with the Cu-Al-alloy packaging to that without the packaging. Both lasers operate at 16 °C. For the bare fiber, there is no onset of a second threshold, which would normally be a strong indicator of the Raman laser threshold. The laser oscillates at a single wavelength as in a conventional laser, with a slope efficiency η_ordinary of 20.2%. For the Cu-Al-packaged fiber, the 1064-nm energy initially transfers to the 1414-nm Cr⁴⁺:YAG laser (i.e., the Raman pump) with a slope efficiency η_{Raman pump} of 5.2%. Afterwards, the intracavity Raman pump generates cascaded AS Raman emission. This intracavity resonance leads to a maximum CW Raman output power of over 25 mW, a slope efficiency η_{AS Raman} of 14.3%, and a threshold of 215 mW. It is worth emphasizing that the Raman threshold of 215 mW is the lowest both in CW Raman crystalline media and fiber forms [29,30].

 

Fig. 3 Crystalline fiber laser performance with and without the Cu-Al-alloy packaging, indicating the AS Raman laser threshold, marked by the x-intercept of the dashed line.

Download Full Size | PDF

One can reasonably obtain the effective Raman pump intensity I_p that is required to reach the Raman threshold when the gain exceeds the loss, i.e., R₁R₂exp(2g_RI_pL) ≥ 1. In this expression, R₁, R₂, g_R, and L are the input and output reflectance, the Raman gain coefficient, and the Raman crystal length, respectively. Note that this expression is valid for Stokes generation. However, we observe that the AS lasing at is accompanied by Stokes lasing, with identical threshold characteristics. After reaching the threshold pump power, both emissions grow linearly with the increasing incident pump level. Therefore, the theory employed here is effective for both Stokes and AS generation. It should be noted that an estimate of the effective stimulated Raman threshold I_p can be deduced by assuming that the resonator losses are dominated by coupler transmittance. This case is analogous to that accounting for the propagation loss α_R at the AS wavelength. In this case, the Raman crystal length L can be expressed as the effective crystal length L_eff, where L_eff = [1-exp(-α_RL_R)]/α_R [29]. By taking into account that α_R = 0.04 dB/cm (for both the Raman pump and AS signal) [31] and L = 1.65 cm, the resulting L_eff was approximately 1.638 cm. As a result, the propagation losses within the Cr⁴⁺:YAG hybrid fiber laser at the AS wavelength are negligible.

Since the Raman gain coefficient g_R of Cr⁴⁺:YAG crystal has not yet been reported, we adopted a simple model based on the macroscopic polarizability β to estimate a reasonable value of g_R. The dominant factor that determines the wavelength-dependent g_R is the frequency-dependent polarizability β. β is also associated with the refractive index in optical frequency. As a result, β accounts for the effects of the pump wavelength on the gain through the relation g_R$\propto$(n²-1)²/n² [32], where n is the refractive index of the Raman medium. By taking the n of the commonly used Raman crystal KGd(WO₄)₂ as a reference, the resulting g_R of Cr⁴⁺:YAG was calculated to be ~3 cm/GW. In our case, for a 16.5 mm Cr⁴⁺:YAG crystal-core glass-clad hybrid fiber with g_R~3 cm/GW in a resonating cavity with R₁R₂ > 0.95, the first stimulated Raman threshold was ~5 MW/cm². This value is far below the ~68 MW/cm² attained for a 215 mW pump power (corresponding to the Raman threshold in Fig. 3) within a 20 μm diameter core without resonance. This resonating scheme increases the effective Raman pump intensity (~68 MW cm²) required to reach the stimulated Raman threshold (~5 MW cm²). Therefore, this design leads to CW low-threshold lasing with high quantum efficiency.

Figure 4 shows the temperature-dependent wavelength tuning of a strain-activated Cr⁴⁺:YAG crystal-core glass-clad hybrid fiber laser under different working conditions. The first AS shift shows a redshift from 132.58 to 123.18 cm⁻¹ with a corresponding decrease in temperature from 24 °C to 16 °C, which agrees with the pressure evolution of the phonon frequencies of YAG and other members of the aluminum garnet family due to the large value of the Grüneisen parameter in the T_2g mode [33]. This shift represents a redshift of ~2.2 nm, indicating that the hybrid fiber undergoes a higher strain. This wavelength tuning implies a sensitivity Δλ/ΔT of 0.049 and 0.273 nm/^oC for the Raman pump and AS signal, respectively. Taking into account the temperature-dependent refractive index n and the TEC α of the Cr⁴⁺:YAG crystalline core, Δλ/ΔT = λ₀[α + (dn/dT)/n], where λ₀ = 1414 nm, α = 7 × 10⁻⁶ K⁻¹, dn/dT = 9 × 10⁻⁶ K⁻¹, and n = 1.82 [34]. The extracted value of Δλ/ΔT for the Raman pump and AS signal wavelength are ~0.017 nm/^oC, which are much smaller than the experimental values of 0.049 and 0.273 nm/^oC, respectively. This comparison demonstrates that the large wavelength sensitivity originates from different optical mechanisms.

 

Fig. 4 Temperature-dependent stimulated Raman tuning of (a) AS Raman and (b) Raman pump.

Download Full Size | PDF

In Fig. 5, the AS Raman spectra at high power show central dips and finite linewidths for all temperatures, which is a strong indication of four-wave mixing [35]. Multiple copies of the Raman gain peak are observed at both the two Stokes and two AS wavelengths due to four-wave mixing. This result also clearly reveals that the Cr⁴⁺:YAG crystal-glass hybrid fiber is highly dispersive. Another noticeable feature that can be observed at high powers is that only the AS Raman spectrum survives, corresponding to a full depletion of Raman pump due to the use of broadband couplers [36]. This pump depletion has commonly been observed in existing Raman lasers [36–38]. This nonlinearity is additional evidence that the onset of the blueshift emission can be attributed to AS Raman laser action.

 

Fig. 5 (a) Nonlinear characteristics of AS Raman spectra at high powers. (b) Close-up view of (a) showing the four-wave mixing characteristics.

Download Full Size | PDF

There are two fundamental contributors to the shift of the Raman peak with temperature. One contribution arises from the changes in the vibrational amplitude of the atoms, that is, the change in the occupation of the phonon states. The other contribution is a consequence of the change in the interatomic distances with a change in temperature. As the unit cell volume changes with the thermal expansion, the Raman peaks shift due to the piezospectroscopic effect (i.e., the peak position depends on the applied strain). Typically, frequency shifts from changes in the interatomic distances (or phonons) are small with such small temperature changes. In the present case, where there are significant thermal expansion effects, the Cu-Al-alloy packaging enhanced the heat dissipation but also led to additional strain in the crystalline fiber, as evident from the 123.65 cm⁻¹ Raman shift (Fig. 2(a)). The large TEC of the embedded Cu-Al alloy of up to 1.266 × 10^{-4 o}C⁻¹ [39] makes large temperature-dependent Raman shifts possible even for small temperature changes.

Taking into account the TEC, the overall thermally-induced volume change for the Cu-Al alloy for the observed temperature change (24 ^oC to 16 ^oC) is ~0.1%. Furthermore, for a the thermally-induced Raman shift difference of 9.4 cm⁻¹ (132.58 to 123.18 cm⁻¹), one can estimate a corresponding stress change of ~0.37 GPa via the pressure-induced shift of the representative R line of Cr³⁺ ions at 25.4 cm⁻¹/GPa [40]. Note that about 95% of the Cr ions in the YAG core have an oxidation state of 3 + . Considering the volume changed ratio of the Cu-Al alloy [41], this ~0.37 GPa pressure change results in a ~0.4% volume difference, in agreement with the TEC-derived value of ~0.1%. In this way, we can explain the large temperature-dependent Raman shift as arising from the piezospectroscopic effect of the Cu-Al-alloy package even for such small temperature changes.

3.3. Crystal-fiber-based sensing with superior sensitivity

The strain-activated AS Raman shifts in Figs. 6(a) and 6(b) demonstrates a potential application of the Cr⁴⁺:YAG crystalline core fiber to highly sensitive thermal and strain sensing. The resonant wavelength shifts linearly with temperature with a slope of 0.273 nm/^oC, a value that is superior to that of other bare fiber Bragg gratings (~0.007–0.012 nm/^oC) [15–18], photonic crystal fibers (~0.01 nm/^oC) [19,20], and SMF-28-based systems(~0.01 nm/^oC) [19,21]. This shift corresponds to a strain sensitivity of ~0.5 pm/με, similar to the ~1 pm/με of a passive fiber Bragg grating sensing scheme [19,20]. In addition, by taking into account the piezospectroscopic effect, a temperature-strain crosstalk of around 530 με/^oC was found over a large dynamic range of 4.24 × 10³ με, which is an order of magnitude higher than that of polymer fiber-based strain sensors [22].

 

Fig. 6 (a) AS Raman shifts as a function of temperature. A high thermal sensitivity of 0.273 nm/°C and 1.134 cm⁻¹/°C was obtained due to the enhanced piezospectroscopic effect of the fiber laser packaged by the Cu-Al alloy. (b) Estimates of the stress and the corresponding volume change ratio of the Cu-Al alloy with the temperature change. The temperature-strain crosstalk calculated from the linear fit of the volume change ratio is 530 με/°C. These sensitivities are at least one order of magnitude higher than the sensitivities achieved by conventional silica fiber-based thermal sensors and polymer fiber-based stress sensors.

Download Full Size | PDF

4. Conclusion

In conclusion, we demonstrated the stain-activated tuning of a low-threshold CW Raman crystal-glass hybrid fiber laser using a cascaded resonating scheme. A high thermal sensitivity of 0.273 nm/°C (i.e., 1.134 cm⁻¹/°C) was obtained from the strain-enhanced stimulated Raman shift of the Cu-Al-alloy packaging; considering the pressure-induced Raman shift of 25.4 cm⁻¹/GPa, this shift corresponds to a pressure change of 0.37 GPa. This pressure change results in an approximately 0.4% volume difference, which agrees well with the TEC-derived value of approximately 0.1%. Thus, the compact and efficient device fabricated in this study is a promising candidate for crystal-fiber-based temperature- and stress-sensing applications.