Highly efficient continuous-wave (CW) solid-state lasers, operating in the sub-micron near-infrared (NIR) region, with radiances ${\gt}{10}\;{{\rm TWm}^{- 2}}\;{{\rm sr}^{- 1}}$ are not currently available via standard laser architectures. Applications calling for high radiance CW sources are reasonably wavelength agnostic. However, for remote sensing applications, sub-micron lasers potentially offer better systems performance through higher detector efficiency. Powerful sources in this spectral region are dominated by semiconductor diode lasers, which offer excellent conversion efficiency and average power scaling potential, though only via multiplexing many individual channel waveguide lasers [1], and therefore with limited radiance. Instead, rare-earth (RE) ion impurities in insulator materials, for which neodymium (Nd) and ytterbium (Yb) are the two prime candidates for radiance-scaling around 1 µm, are favored with advantageous spectroscopic properties. Moreover, their metastable excited-state lifetimes, nominally four orders of magnitude longer than those of semiconductors, offer significantly higher energy-storage potential.

${{\rm Yb}^{3+}}$-doped gain-media particularly when employed in laser architectures optimized for extreme thermal management, either in optical-fiber [2], thin-disk [3], or slab [4] geometries, have been scaled to multi-kilowatt average powers, with record radiance levels exceeding ${1000}\;{{\rm TWm}^{- 2}}\;{{\rm sr}^{- 1}}$ [5,6]. Notably, these achievements are all at wavelengths higher than 1 µm.

The improved spectroscopic and thermo-optical properties of active crystals at cryogenic temperatures have been exploited for many different laser materials. From uranium-doped ${{\rm CaF}_2}$, the second laser ever demonstrated [7], to the highest reported CW power of 455 W from a cryogenically cooled Yb:YAG laser oscillator [8]. The reported record CW power for a cryogenic master oscillator power amplifier is 963 W [9]. For ${{\rm Nd}^{3 +}}$-doped lasers though, due to the higher quantum defect for the strongest transition at 1.06 µm, the scope for high-power lasers is less promising than for ${{\rm Yb}^{3 +}}$ [10]. One result stands out for 1 µm cryogenic operation of Nd:YAG, which focused on the reduction in the thermal lensing effect, demonstrating 25 W of output power [11]. However, a Nd:YAG laser operating at 946 nm between the $^4{{\rm F}_{3/2}}\to {^4{\rm I}_{9/2}}$ energy levels, when pumped directly to the upper laser level, has an equivalent energy-level character to ${{\rm Yb}^{3 +}}$. The highest average power and slope efficiency at 946 nm yet reported are 60 W [12] and 78% [13], respectively.

In this Letter, we report a new benchmark power, radiance, and efficiency for the 946 nm Nd:YAG laser, achieved by closed-cycle cryogenically cooling the active crystal. Performance advances, with respect to previous results, are realized by using a bespoke crystal mount providing symmetric heat extraction and minimal mechanical stress-induced optical phase distortion upon cooling. At the same time, excellent thermal contact was maintained between the gain medium and heat sink at cryogenic temperatures. For a cold head temperature of 80 K, a maximum output power of 113 W with a near-ideal slope efficiency of 80% (with respect to the absorbed pump power) was obtained.

Our results represent a major improvement in performance for cryogenically cooled 946 nm Nd:YAG lasers compared with recent reports [12–14]. With respect to [12], we have improved thermal management through the realization of a more symmetric cooling of the crystal and pump distribution, and mitigation of mechanically induced stress upon cooling. To the best of our knowledge, this demonstration represents the highest slope efficiency and average output power for a 946 nm Nd:YAG laser achieved to date.

The experimental arrangement is shown schematically in Fig. 1, comprising a simple linear cavity containing the crystal fixed to the cold head of the cryostat, contained within an evacuated housing providing thermal isolation. A vacuum pressure of ${\sim}{{10}^{- 7}}\;{\rm mbar}$ was used during operation. Optical access into the housing was through 5-mm-thick Suprasil windows, anti-reflection (AR) coated for the laser and pump wavelengths with high-transmittance (HT) for 1060 nm. A plane pump in-coupling mirror [high-reflectance (HR) ${\gt}{99.8}\%$ for wavelengths between 920 and 1040 nm and ${\lt}{25}\% {\rm R}$ at 1060 nm] was positioned near the first vacuum-chamber window. Two concave output coupler (OC) mirrors were trialed, with radii of curvature (ROC) of 200 mm (1000 mm) and reflectivity of 90%R (80%R) at 946 nm, respectively. These respective OCs were coated for HT at 868 nm, 1060 nm, and 1320 nm and in use were positioned near the second window, opposite the first, resulting in a cavity length of 145 mm. At the maximum pump power, the calculated average fundamental-mode radii in the crystal for the 200 mm and 1000 mm ROC OCs were 204 µm and 230 µm, respectively.

 

Fig. 1. Schematic diagram of the experimental setup. PM1,2 are power meters, and OC is the output coupler mirror, with a dichroic mirror used to separate the residual pump and laser beams.

Download Full Size | PDF

The cuboid $\langle{111}\rangle$-cut Nd:YAG crystal was 15-mm-long, with an aperture of ${3} \times {3}\;{{\rm mm}^2}$ and doping concentration of 0.3 at. %. The end facets of the crystal were AR coated with a typical residual reflection of ${\lt}{0.05}\%$ at the pump and laser wavelengths and ${\lt}{2}\%$ around 1060 nm. As the crystal must be mounted at room temperature (RT) but used at cryogenic temperatures (e.g., 80–100 K), the mounting design needed to mitigate the mechanical stress induced by differential thermal contraction of the respective materials in the assembly. An indium foil was wrapped around the crystal before being clamped in its mount, which was then attached to the cold head. Exacting tolerances were used in the laser head design to enable daily cycling between 80 K and RT.

An acoustic Stirling cryostat (${Q}$-Drive 2s132K) was used, capable of providing ${\sim}{20}\;{\rm W}$ (${\sim}{35}\;{\rm W}$) heat extraction at a temperature of 80 K (100 K) for a nominal 600 W electrical input. The cold head temperature could be set anywhere between 60 and 240 K, with typical cooldown times to 80 K, from RT, on the order of 30 min.

The pump source was a 200 W CW diode laser bar (DILAS-Coherent EY-Series), with a nominal center wavelength of 866.5 nm and bandwidth of 3.2 nm at a 195A drive current and 25°C cooling water. A beam transformation system (BTS) micro-optic (LIMO GmbH) was used for efficient radiance conversion of the diode laser bar output. Positioned immediately after the BTS was a volume-Bragg-grating (VBG) (Optigrate) with a resonance wavelength of $({868.2}\;{\pm}\;{0.2})\;{\rm nm}$ and $({25}\;{\pm}\;{5})\%$ diffraction efficiency (reflectance). Feedback from the VBG stabilized the pump wavelength for optimal absorption by Nd:YAG when cooled to ${\sim}{80}\;{\rm K}$ [12]. A ${{\rm f}_y} = {40}\;{\rm mm}$ cylindrical lens collimated the vertical axis of the wavelength-stabilized diode laser output (i.e., the slow axis of the respective emitters). Subsequently the beam was expanded in the orthogonal axis with an afocal cylindrical telescope comprising two lenses ${{f}_{x1}} = {75}\;{\rm mm}$ and ${{f}_{x2}} = {150}\;{\rm mm}$, then bisected before polarization combining the respective parts, effectively halving the beam parameter product in this axis. This configuration resulted in a beam quality of ${M}_x^2 = {43}$ by ${M}_y^2 = {25}$ at 135 W pump power, in the respective axes, after two separate cylindrical focusing lenses for the $x$ and $y$ axes (${{f}_x} = {{ f}_y} = {200}\;{\rm mm}$). A maximum incident power of 152 W was available at a 200 A drive current. We measured the free-space pump beam caustic for incident powers of 45 W and 135 W, around the position of the beam waist(s) after the cylindrical focusing lenses. Figure 2 shows the location of the pump beam caustic, referenced to the input-coupler-mirror position. Note, we have accounted for refraction at the air/crystal interfaces, as well as beam propagation through the crystal.

 

Fig. 2. Pump beam caustics (measured in free space), and fundamental cavity modes for ${{\rm ROC}_{{\rm OC}}} = {200}\;{\rm mm}$, at 45 W and 135 W. $z$ coordinate referenced to the input mirror position.

Download Full Size | PDF

The shaded region in Fig. 2 shows the crystal location within the cavity and with respect to the pump beam caustic for the two measured conditions. As the pump laser was entirely free-space coupled to the crystal, the waist radii and position were pump power dependent. A consequence of the diode-laser array being driven further above threshold, thus increasing the higher-order mode content in each waveguide’s slow axis. The waist position in the $x$ axis was almost stationary, while for the $y$ axis it moved away from the focusing lenses by ${\sim}{10}\;{\rm mm}$ near full operating power (Fig. 2). Calculations of the fundamental cavity mode radius throughout the cavity are superimposed on the plots in Fig. 2, for predicted thermal lenses of 1000 mm and 250 mm [15], at 45 W and 135 W incident pump power, respectively.

During operation, the non-absorbed pump light transmitted through the Nd:YAG crystal, the OC, and a dichroic mirror, was measured by a water-cooled powermeter (Gentec, UP55N) (PM1), while the 946 nm laser beam was monitored by a second water-cooled powermeter (PM2) (see Fig. 1). A 200 µm/0.22 NA fiber, coupled to an optical spectrum analyzer (OSA, AQ6317B), collected light scattered off PM2. This allowed us to monitor the emission spectrum during alignment and operation. Figure 3 shows the evolution of the peak wavelength of the VBG locked pump laser. Measured in a similar way although from leakage through the beam conditioning optics before the cavity, a redshift at a rate of 3 pm/W was observed, with respect to the pump power incident upon the crystal, consistent with previous findings [12]. Pump absorption was calculated in function of the incident pump power, when the cryostat temperature was set to 80 K, while the laser was operating using the two individual OCs (Fig. 3). There was a significant change in absorption efficiency with increasing pump power due to the temperature rise in the VBG and associated wavelength chirp aligning the resonance peak with the absorption peak of the Nd:YAG crystal. The VBG resonance wavelength was chosen such that at full current the pump laser wavelength would overlap the absorption peak of the crystal at ${\sim}{80}\;{\rm K}$. Hence, the absorption efficiency initially is low due to the poor spectral overlap with the narrow absorption band of the crystal, and then it rapidly increases with pump power and saturates near 98% for powers ${\gt}{100}\;{\rm W}$. However, during the experiment with the 80% OC, the VBG temperature exceeded normal operating conditions, and the resonance wavelength moved beyond the peak absorption, at pump powers above 135 W.

 

Fig. 3. Pump absorption efficiency during lasing with two different OCs, and typical VBG-locked pump peak emission wavelength, versus incident pump power.

Download Full Size | PDF

As shown in Fig. 4 when using the 90%R OC, the laser output power reached 113 W, limited by the available pump power, with a slope efficiency of 80%. Instead for the 80%R OC, the slope efficiency increased to 82% with a maximum power of 104 W. As indicated in Fig. 3, for the latter cavity, the achievable maximum absorbed and, therefore, output power were reduced due to warming of the VBG. This can be mitigated by controlling temperature of the VBG. For the cavity employing the 200 mm ROC, 90%R OC, we measured the ${{\rm M}^2}$ beam quality parameter at different output powers, which increased from ${{\rm M}^2}\sim{1.3}$ at 35 W to 4.9 by 105 W. It is to be expected for the linear cavity used that higher-order modes would be supported at sufficiently high pump powers, as the fundamental cavity mode size in the crystal is nominally half that of the pump (see Fig. 2). More advanced cavity configurations are still to be explored to improve the beam quality. The laser emission wavelength, ${\lambda _{{\rm em}}}$, was observed to blueshift at ${\sim}{25}\;{\rm pm/W}$ (of absorbed pump power; see Fig. 4), which could be attributed to the peak wavelength shift in the emission cross section with temperature [16]. The laser emission wavelength was measured in real-time with the OSA, observed to operate solely at 946 nm once lasing was established. Noticeably in the experiment we could observe parasitic amplified spontaneous emission (ASE) at 1061 nm in the spectrum, when the laser was not well aligned. Nevertheless, it influences the available gain for 946 nm, as observed around threshold with dramatic reduction in ASE irradiance at 1061 nm above laser threshold.

 

Fig. 4. Measured laser emission wavelength (top-left axis) and respective calculated average local temperature (top-right axis), optical-to-optical efficiency (red, left axis), and output power (blue, right axis) versus absorbed pump power, for the two tested OCs.

Download Full Size | PDF

Despite the improvement of the thermal and thermo-optical properties of crystals at low temperatures, energy transfer upconversion (ETU), thermal effects, and aberration are still the main factors limiting the radiance and scalability of cryogenically cooled Nd:YAG lasers. In general, the major contributors to the spatial phase distortion of cryogenically cooled lasers are mechanical stress-induced birefringence, temperature dependent strain-induced birefringence, the thermo-optic coefficient, and deformation of the crystal facets. Nevertheless, the phase aberration due to the mechanical pressure due to the different contraction of the materials in laser head can be minimized by careful design of the crystal mount, whereas the other effects could be mitigated with appropriate resonator designs to control total phase degradation [17]. Upon mounting the Nd:YAG crystal in the cryostat and before pumping it, we investigated the phase distortion associated with mechanically induced pressure while cooling from RT to 80 K. This was achieved by monitoring the depolarization loss for a linearly polarized 912 nm probe laser beam passing through the middle of the mounted crystal that was positioned between two crossed polarizers. A maximum depolarization loss of 0.3% was observed during cooling.

ETU is a notorious source of extra heat that exacerbates degradation in laser performance for low-gain transitions like that of 946 nm in Nd:YAG [18]. Severely limiting at RT, derived from negative feedback involving increasing thermal loading and cavity losses, coupled with reducing gain, eventually leads to a rollover in performance that can only be mitigated through better thermal management. It has been demonstrated that although the macroscopic ETU parameter for Nd:YAG is doubled at 80 K compared to RT [19], the improvement in the spectroscopic characteristics due to cryogenic cooling has a much larger positive influence on laser performance. Moreover, it is known that the enhanced thermo-optical properties of Nd:YAG at cryogenic temperatures provide further advantage in terms of thermal management. Nevertheless, a qualitative study of the effects of ETU on laser performance dependent upon the pump-to-cavity mode overlap and heat loading distribution was undertaken following the methods detailed in Ref. [18]. In the experiment, the measured pump beam spot size is twice the calculated fundamental cavity mode; therefore, the oscillator will support higher-order modes, for which in modeling, it is nontrivial to allocate a proportional contribution in terms of the laser output. As the rate of ETU depends on ${N_U}^2$, where ${N_U}$ is the upper laser level population density, its contribution to heating of the crystal is dynamic, a function of the number of times the laser is above threshold. As such, the induced thermal lens is highly aberrated, mainly in the outer regions of the pump beam, leading to beam quality degradation [17,18], as evidenced by the larger-than-expected ${{M}^2}$ value obtained at higher pump powers.

One way to determine the average temperature rise in the crystal is to measure the emission wavelength, which has been correlated with gain medium temperature [16]. Using this method, the average local temperature of the portion of ions contributing to the stimulated emission was estimated in terms of pump power (Fig. 4). It was observed that for the 80%R OC, the absorption efficiency was slightly lower than for the 90%R OC. As such, the effective absorption length was increased, and the deposited heat would have also been pushed further into crystal, leading to a more uniform temperature distribution and heat extraction. Moreover, the 80%R OC had a larger ROC and, therefore, cavity mode, leading to greater saturation of ${N_U}$ near the edges of the pump beam. Consequently, heat loading associated with ETU in these regions would be lower, and the core temperature would rise [18].

Another critical challenge in achieving high radiance 946 nm operation of the Nd:YAG laser, even at cryogenic temperatures, is suppressing parasitic ASE or lasing at 1.06 µm. Nd:YAG’s strongest emission line and, therefore, gain switches from 1064 nm to 1061 nm below ${\sim}{200}\;{\rm K}$ [16]; moreover at 80 K, this later wavelength has an effective emission cross section of ${78}\;{{\rm pm}^2}$, compared with ${8.4}\;{{\rm pm}^2}$ for the 946 nm transition. The almost order-of-magnitude higher gain at 1061 nm can support strong growth of ASE if the pumping arrangements provide both a large solid angle and small-signal gain. Fortunately, for the investigated experimental setup, the single-pass small-signal gain at 946 nm was only 0.28 dB (0.54 dB) for the 90%R (80%R) OCs, based on a round trip loss of ${{L}_{946\,\,\rm nm}} = {0.02}$. Consequently, the single-pass small-signal gain at 1061 nm was 2.6 dB (5 dB). A spontaneously emitted photon at the center of the pump beam at one facet would see a solid angle defined by the pumped volume of $\Omega \sim{2} \times {{10}^{- 3}}\;{\rm sr}$. The resulting intensity of the ASE at 1061 nm would be 2–3 orders of magnitude smaller than the saturation intensity at this wavelength and, therefore, would have a negligible depletion effect on the upper laser level. Nevertheless, it influences the available gain for 946 nm, as observed around threshold with dramatic reduction in ASE intensity above laser threshold.

Suppressing parasitic lasing at 1061 nm required special coatings and attention to feedback into the pumped region. Feedback from the in-coupling mirror was the strongest of any of the cavity, or neighboring, optics; however, being sufficiently separated from the crystal, it had little impact on the resulting ASE. The round trip cavity loss at 1061 nm was designed to be in excess of 20 dB, substantially higher than the anticipated gain at this wavelength for the optimum output coupling for 946 nm operation. With optimal alignment in the experiment, that is when 946 nm stimulated emission saturated the available gain, ASE and parasitic lasing 1061 nm were effectively suppressed.

We have developed a highly efficient ${\gt}{110}\;{\rm W}$ closed-cycle CW cryogenically cooled Nd:YAG laser operating at 946 nm, enabled by in-band pumping with a bespoke high-brightness VBG stabilized diode laser and aided by a bespoke laser crystal mounting configuration. Up to the maximum absorbed pump power of 152 W, the corresponding slope efficiency was of 80% with an OC reflectivity of 90%R, increasing to 82% for an 80%R OC. A new benchmark has been set for output power equal to 113 W, corresponding to a radiance of ${5}\;{{\rm TWm}^{- 2}}\;{{\rm sr}^{- 1}}$, and optical-to-optical efficiency of 74%. No saturation phenomena or cavity instabilities were observed, implying that the system was limited by the available pump power. Future investigations will aim to improve radiance through reducing the beam parameter product and exploring the potential for pulsed operation for energy scaling in this NIR wavelength regime.