Wide-Spectrum Tuning and Narrowing of 780 nm Broad-Area Diode Laser with Littrow-Type Transmission Gratings

Abstract

Spectrum-narrowed and -locked broad-area diode lasers operating at 780 nm are essential for rubidium laser development. With the help of Littrow-type transmission gratings, we demonstrated a simple scheme with a narrow linewidth and the diode laser’s center wavelength locked without thermal drift, in contrast to volume Bragg gratings. By carefully collimating the diode laser beam, we realized a linewidth narrower than 0.17 nm and a side-mode suppression ratio over 20 dB. Furthermore, broad-spectrum tuning at 9 nm was demonstrated by grating angle tuning. This method could easily be adapted to other wavelength diode lasers.

1. Introduction

High-power 780 nm narrow-linewidth broad-area diode lasers are of great concern in the development of rubidium-diode-pumped alkali lasers (DPALs) [1,2]. Volume Bragg gratings (VBGs) play an essential role in DPAL engineering due to their high efficiency and simplicity for spectra narrowing [3,4,5,6]. However, due to the intrinsic absorption of the photo-thermal-refractive (PTR) glass, there will be a natural wavelength drift of approximately 8 pm/K, originating from the temperature rise in VBGs. To meet the needs of DPALs, careful temperature tuning and VBGs with complex temperature controlling systems are typically needed, which will make the whole system too complex in an engineering point of view [7,8].

To solve the problem, the Faraday anomalous dispersion optical filter (FADOF) was first demonstrated by Knize’s group with a 780 nm high-power diode stack in 2018, with the spectra strictly locked at the Rb D2 line and an extreme linewidth of 10 GHz [9,10]. Later, we extended this method to the standard diode module and further proposed a mutual injection optical loop to decrease its polarization dependence [11,12]. However, the current FADOF-based diode laser is still bulky, and more volume reduction work is still needed.

In fact, Littrow or Littman reflective grating external cavities were widely used in high-power laser diode array (LDA) spectrum narrowing at the early stage of DPAL development. Knize et al. achieved a 11 GHz linewidth and 10 W output diode laser [13,14], while we obtained a single broad-spectrum diode laser with a linewidth of 15 GHz at 780 nm [15], and Xemed LLC proposed a method using stepped mirrors and reflective plano diffraction gratings to narrow the linewidth of high-power stacks [16]. In conclusion, the traditional Littrow structure has a problem when tuning the central wavelength, which is the change in the direction of the output light [17]. For that reason, during the development of grating-based external cavity diode lasers, Littman structures are preferred. However, all of these reflective-grating-based methods face the common problem of folded optics, which greatly increase the system’s complexity for power scaling.

As early as 2002, Laurlia et al. utilized the transmission grating to narrow and tune a single-transverse-mode diode laser at 650 nm [18]. In recent years, transmission grating technology has greatly improved due to the various needs for spectra combination applications [19]. This provides a possibility for reconsidering its application in high-power broad-area diode lasers.

In this paper, we described a 100 μm strip broad-area laser with a 780 nm narrow linewidth and a wide tuning range based on Littrow-type transmission gratings. We achieved a wavelength tuning range of ~9 nm, a linewidth less than 0.17 nm and an output power of ~3 W. This scheme is as simple as the current VBG-based method with an even wider spectrum tuning range using convenient angular tuning.

2. Experimental Setup

The external transmission grating cavity laser includes a transmission grating, collimation systems, and a diode laser.

In this study, we utilized a self-made 2052 groove/mm fused-silica transmission grating measuring a size of 25 × 25 × 1.5 mm to build the tunable external cavity, with the grooves on one side parallel to the x direction, as shown in Figure 1, and an anti-reflection (AR) coating on the other side. The grating is TM-polarized, which means the field direction is perpendicular to the grating groove. According to the general grating equation of the Littrow configuration [20],

$$\lambda =2dsin\theta$$

(1)

where λ and d represent the wavelength of the incident light beam and the grating period, respectively, and θ stands for the angle between the incident light beam and the normal direction. Thus, the Littrow angle of the grating is calculated to be 53.16° at 780 nm.

When the diode laser beam passes through the transmission grating at a self-collimation Littrow angle, the back-diffracted Littrow-order (−1R) light feeds back to the diode chip with a grating efficiency, η, of 13%. The external cavity is constructed between the grating and the end surface of the diode chip. Since this self-made grating is un-optimized, the intrinsic geometric 0R reflection and 1T-order diffraction exist with a total fraction of ξ (~14% and ~21% separately by measurement), which amounts to the loss of this external cavity system. The narrowed beam is coupled through the 0T portion with an efficiency of 1 – η − ξ.

The grating resolution follows Equation [20]:

$${\displaystyle \frac{\lambda }{\mathsf{\Delta }\mathsf{\lambda }}}={\displaystyle \frac{mNW}{\mathrm{c}\mathrm{o}\mathrm{s}\theta }}$$

(2)

where ∆λ, m, and N, represent the spectrum bandwidth, the diffraction order, and the number of grating grooves, respectively, and W represents the beam size, which represents the fast-axis beam width of the diode laser in this experiment. When the linewidth of the diode laser is less than 0.2 nm, a sufficient efficiency can be obtained when pumping the DPAL [21,22,23]. According to this demand, if we set W to 1.5 mm, the theoretical spectrum resolution would be around 0.15 nm based on Equation (2).

The experimental setup is shown in Figure 1. The packaged 100 μm × 3 mm single-emitter broad-area diode laser was glued on an aluminum water-cooled block kept at 22.5 °C. A standard 350 µm effective-focal-length fast-axis collimator (FAC) was used to compress the divergence angle to 10 mrad. The slow axis (SA) beam quality factor of the diode laser was significantly larger than that of the fast axis (FA), with its M² factor measured to be 34. To ensure the feedback light was efficiently coupled to the diode chip, we used CL_SA (f = 50 mm) and CL_FA (f = 60 mm) to further collimate and expand the beam size to 12 mm × 1.5 mm (SA&FA).

After collimation, the diode laser beam passed through the transmission grating at the self-collimation Littrow angle, and the back-diffracted Littrow-order (−1R) light was fed back to the diode chip. Specific modes of light related to the characteristics of the grating were fed back to the diode laser, which changed the mode competition inside the laser cavity. Eventually, the desired wavelength of the laser mode was enhanced, with other modes of light suppressed. Thus, the compression and narrowing of the laser spectrum were achieved

The wavelength of the output light could be further tuned by grating angle rotation. The diode spectrum was monitored by an optical spectra analyzer (OSA) with a resolution and accuracy uncertainty of 0.02 nm, and the diode power was recorded by a power meter with a range of 10 W.

3. Results and Discussion

Theoretically, the longer the external cavity length, the narrower the linewidth. However, a long external cavity brings more losses to the feedback efficiency of the diode laser at the same time [24,25]. Therefore, we need to find a balance between the external cavity length and the efficiency. At a driven current of 7 A and central wavelength of 780.20 nm, when the cavity length changed from 80 mm to 560 mm, the maximum difference in the linewidth and power of the locked diode laser was only ~0.03 nm and ~0.6 W, respectively. Furthermore, the difference in tuning range between the shortest and longest cavity length was ~1 nm. When the broad-area diode laser was well collimated, we found the external cavity length showed little influence on the power, linewidth, and tuning range. Hence, we set the cavity length to 320 mm in the experiment after careful optimization, and we have to point out that a shorter cavity length could have been acceptable as well.

The free-running and locked diode laser spectra are shown in Figure 2, with the strongest peak centered at 778.56 nm. The locked spectral linewidth is 0.17 nm, which is significantly narrower than the free-running spectra, and a high side-mode suppression ratio (SMSR) better than 20 dB was easily obtained. The narrowed linewidth is very close to the theoretical estimation of Equation (2). In this experiment, to couple more output lasers with the OSA, a 50 μm core fiber rather than a 9 μm core fiber was used to sample the diode beam. However, only the latter fiber could match the highest resolution of the OSA (0.02 nm), which meant that the real spectral linewidth was is less than 0.17 nm.

Figure 3 shows the output spectrum tuning characteristics of the tunable diode laser at a driven current of 7A, with a cavity length of 320 mm and cooling temperature of 22.5 °C. In this case, wide-central-wavelength tuning from 773.5 nm to 782.5 nm was achieved with the output power changing by a little, as shown in Figure 4 The minimum output power is 2.83 W at the wavelength of 782.5 nm. The maximum output power is 3.05 W at 780.43 nm. The output power is ~3 W at the wavelength of 775.5 nm–780.5 nm, which is consistent with the free-running spectrum range.

Figure 5 shows that the tuning wavelength of the laser system varies with the rotation angle of the transmission grating. As the transmission grating rotates, the central wavelength of the external cavity diode laser (ECDL) shows a regular and approximately linear shift. This shift mainly results from the grating’s angular dispersion effect. The rotation changes the mode component, feeding back to the diode laser’s cavity and shifting the central wavelength. Compared with Ref. [3], the VBGs could achieve a tuning wavelength of ~0.4 nm. It is obvious that this method obtains a larger tuning range, and is easier to operate.

The PI curves of both the free-running and locked (at 780.20 nm) diode laser are shown in Figure 6. The diode laser has an output power of 5.9 W in the free-running case when the driven current is 7 A. The external cavity feedback is not fixed for the transmission grating external cavity. This structure’s wavelength is -tuned by rotating the angle of the transmission grating. As the grating rotates, the angle of light returning to the diode laser is changed. Due to the angle change, a portion of the feedback light cannot return to the cavity. As a result, there are changes in the feedback optical power, which will affect the output power of the laser at different wavelengths. The diode laser realized a maximal output of 3.05 W, which showed a 52% external cavity efficiency. At the same time, the slope efficiency dropped from 0.99 W/A in the free-running case to 0.52 W/A in the locked case.

The loss in the cavity was inevitable because the grating was not optimized, and the 0R geometric reflection and 1T-order diffraction light were both wasted. The measured power of the 1T-order diffraction and the 0R diffraction light were 1.26 W and 0.82 W, respectively. If these factors are subtracted, the external cavity efficiency could exceed 88%, which will be close to the VBG scheme if further optimization of the external cavity length and front-chip anti-reflection surface coating are carried out.

Furthermore, the FA size of the current-collimated diode laser beam is 1.5 mm, similar to the FA size of the standard diode laser bar’s beam (~1.8 mm). It is indicated that this method could realize efficient power scaling by increasing the power density of the diode laser, for example by replacing single-diode laser emitters with diode laser bars or stacks. Meanwhile, FACs (fast-axis collimators) with longer effective focal lengths are needed to ensure sufficient FA beam size, and SAC (slow-axis collimator) array are needed to ensure relatively good SA collimation. In addition, the diffraction efficiency of the transmission grating should be further enhanced. Though this is not an easy target to reach, thanks to the advances in the grating industry, the enhancement of grating transmission can be made possible in the future.

4. Conclusions

In conclusion, we demonstrated a simple narrow-spectrum 780 nm broad-area diode laser with a transmission grating, which could be used to realize more convenient and wider central wavelength tuning compared to that achieved with the traditional VBG method. Meanwhile, with the development of grating manufacturing, zero-order diffraction and higher losses will be further minimized. Therefore, a higher efficiency of up to 88% and a narrower linewidth of this structure could be achieved in the future.

The emergence of this kind of laser will offer great support and convenience to scientific research and engineering applications. First, compared to the VBG-based ECDL, this diode laser has better stability, which means that the rubidium laser can operate more stably when pumped by this diode laser. Furthermore, the wide tuning range of this diode laser gives it the potential for two-photon excitation in Rb vapor [26,27] and pumping the metastable rare-gas laser, which has more complex energy levels [28].

Furthermore, the method is scalable and usable, which makes it possible to realize a high-power output with a wide central wavelength tuning range. This kind of diode laser has potential for various applications, such as high-power alkali vapor lasers, high-power rare-gas lasers, spin-exchange optical pumping (SEOP), nonlinear optics, etc.