Introduction. The pioneering discovery of InGaN-based light sources by Nakamura [1,2] has opened innovative routes to build portable high-power diode lasers [3–6] with great potential for new applications [7–10]. Previously, blue lasers emitting several hundred-watt outputs have been demonstrated [11–15]; however, these systems essentially required refrigerant/coolant-based chillers, which are bulky in size as well as heavy. Furthermore, fluid-based chillers require periodic maintenance and (refrigerant) consumables which may not be environmentally friendly. Therefore, it is essential to develop an all-solid-state air-cooled high-power continuous wave (CW) blue-laser system. A portable air-cooled diode laser system may be desired for applications such as material processing and soldering battery packs. Thermal management of the 100 W scale blue lasers is challenging because with typical wall-plug efficiency around 25${\% }$, the generation of waste heat is several hundreds of watts which must be continuously removed for the thermally safe steady-state operation of the laser system.

The Peltier-based cooling system has been a preferred thermal management solution [16–18]; however, it has been mostly used for low-power diode lasers [19–22]. Table 1 summarizes selected solid-state cooled diode lasers which highlight the lack of Peltier-based air-cooled lasers above 10 W. One may wonder how to combine high-power laser arrays with powerful modern Peltier chips/thermoelectric coolers (TEC) on a compact platform to develop a 100 W class laser system. Additionally, the potential for unique applications of intense blue lasers is also under active research.

Table 1. Comparison of Selected All-Solid-State Air-Cooled Diode Lasers

View Table

Here, we design and demonstrate an all-solid-state 100 W scale CW blue laser by combining diode banks with Peltier-based air cooling on a compact platform. The thermal performance of the laser system is validated by 3D simulations of the laser head which agreed with the experiments. We achieved stable thermal operation and good electrical and optical performance. We demonstrate precision metalwork on copper such as cutting, bending, and fine soldering on battery banks. Additionally, we find that 4 kW/cm$^2$ blue-laser annealing boosts the mechanical resilience of copper strips to withstand extreme bending cycles by avoiding nanoscale cracks.

Laser design. Figures 1(a) and 1(b) show a front view and top view of our laser head, respectively. This setup combined three laser-diode banks (LDB), highlighted by dark brown rectangles with 8 blue circles, representing individual laser diodes. The LDB is sandwiched between copper plates, and a cluster of Peltier chips is placed between the copper plates and aluminum heat sinks. The total designed cooling capacity of the Peltier system, with forced air cooling, was 3–4 times the maximum laser output, as validated with our thermal simulations. A set of high-performance fans was used to force the air cooling of the heat sinks. The entire system was enclosed within a protective steel enclosure of dimension 35 $\times$ 35 $\times$ 25 cm$^3$ [Fig. 1(c)]. No additional chiller accessory was required, thereby making this laser portable. The 24 collimated beams from individual diodes were focused by two lenses. The spot size in the focal plane was measured about $1.2 \pm 0.1$ mm in the $x$ direction and $2.1 \pm 0.1$ mm in the $y$ direction using a knife edge technique at P=50 W [Figs. 1(e) and 1(f)]. The ellipticity of the focus is due to different divergences of each diode (Figs. S1 and S2 in Supplement 1). The laser generated up to 4 kW/cm$^2$ CW maximum intensity at the focus.

Thermal and Optical characterization. It is essential to validate the thermal design of the laser system. For this, we solved the following 3D time-dependent heat-transfer equation for the geometry of the laser head, $\rho C_{p}\left (\frac {\partial T(\vec {r},t)}{\partial t}+\mathbf {u}_{\text {trans}} \cdot \nabla T(\vec {r},t)\right )+\nabla \cdot \left (\mathbf {q_{cond}}+\mathbf {q_{conv}}\right ) = Q(\vec {r},t)$. Here $\rho$ is the density of materials and $C_p$ is the specific heat capacity at constant stress. $T(\vec {r},t)$ denotes the absolute temperature, $\mathbf {u}_{\text {trans}}$ is the velocity vector of translational motion, and $\mathbf {q_{cond}}$ and $\mathbf {q_{conv}}$ are the heat fluxes by conduction and convection, respectively [23,24]. The conduction term is $\mathbf {q_{cond}}=k\nabla T$ where k is the thermal conductivity of the material. The forced and natural air convection is modeled by $\mathbf {q_{conv}}=h(T_{ext}-T)$ with a heat-transfer coefficient $h=200$ Wm$^{-2}$K$^{-1}$ and $h=15$ Wm$^{-2}$K$^{-1}$, respectively [24–26]. The radiative cooling is expected to be negligible and therefore ignored. The heat source $Q(\vec {r},t)$ comprises of spatially distributed 24 identical laser diodes whose heat generation is parametrized by their efficiency and operating power.

 

Fig. 1. Schematic of all-solid-state air-cooled blue diode laser system: (a) front view and (b) top view. (c) Picture of the working laser. (d) Laser beams are focused using two lenses. Power versus position of the knife edge in the focal plane along the (e) $x$ direction and (f) $y$ direction. The red lines are the error function fit. The black dotted lines show 95% and 5% of the maximum power. The focal spot sizes are marked.

Download Full Size | PDF

We used Comsol (Multiphysics 5.3 Software) to solve the above time-dependent heat-transfer equation for the 3D laser head operating at different laser powers. The electric current module was combined to model the TEC chip in the system [27,28]. The Peltier modules were modeled using a set of P–N junctions using the electric current module in Comsol with Seebeck coefficients as specified by their manufacturer. We used adaptive meshing with a finer mesh size (1 µm) in the proximity to the laser diodes, whereas the maximum element size was 10 µm. The laser diodes were thermally modeled as spatially distributed heat sources, and their heat generation $Q(\vec {r},t)$ was parameterized according to the measured laser efficiency $\eta =24{\% }-30{\% }$ for a given output power. The thermal properties of various materials (Al, copper, air, and alumina) were taken from the built-in parameters. The initial temperature of the laser head was at room temperature 20° Celsius.

Figure 2(a) shows a comparison of the 3D thermal simulation of our design with the experimentally measured temperature of the laser head. Both simulations and experiments showed that the system reached a steady-state temperature in 10 min after being switched on. The maximum operating temperature was dependent on the laser power and always remained below the thermal damage temperature (75 $^\circ$C) ensuring reliable and repeatable laser operation. A typical snapshot of the steady-state isothermal lines of the laser head (at P=50 W) is shown in Fig. 2(b). The heat flux [front view and top view, Fig. 2(c) and 2(d)] of the laser head confirms that this design allows effective heat removal from the diodes and keeps all the diodes thermally safe (<75 $^\circ$C).

 

Fig. 2. Thermal assessment of the laser head of three LDBs. (a) Variation of temperature with time. (b) Isothermal contour over model geometry at 50 W output power. Simulated normalized vector of total heat flux within the geometry volume: (c) front view and (d) top view.

Download Full Size | PDF

Figure 3 shows the optical characterization of the laser system at CW operation. The P–V and I–V curves of the laser system are shown in Fig. 3(a), showing that the total power was adjusted from 40 W to 100 W. The laser demonstrates 98% output power stability over 180 min [Fig. 3(b)]. The spectra of all 24 diodes taken with Thorlabs CCS100 spectrometer at T = 45 $^\circ$C are depicted in Fig. 3(d). The histogram of the central wavelength of 24 diode shows a laser wavelength to be $\lambda =452.2\pm 2.5$ nm [Fig. 3(e)]. The electrical-to-optical efficiency of the laser was about $\eta \sim 25$% up to 70 W and was slightly lower for higher powers due to an increase in operating temperature [Fig. 3(c)]. These measurements establish this all-solid-state blue laser.

Applications. We exploit the high absorption coefficient of blue light by copper ( 60%) [10,29] and high energy of 2.7 eV per photon for precision metalwork on copper. Our laser source generates a 4 kW/cm$^2$ of intensity over an mm-size focus [see Fig. 4(a)] which cuts copper strips of 150 micron thickness [Fig. 4(b)]. Additionally, we demonstrate soldering copper strips onto a battery pack (sub-kW/cm$^2$ exposure) using a commercial solder wire [see Figs. 4(c) and 4(d)]. Importantly, the copper–solder welding joint exhibits nanoscale adhesion as shown by scanning electron microscopy (SEM) images [Figs. 4(e)–4(g)]. These SEM images confirm that the laser soldering is free from any microscale or nanoscale cracks and is mechanically robust.

 

Fig. 3. (a) P–V and I–V curves of the laser system. (b) Stability of laser system at P = 100 W with time. A 2% reduction in power is reversible. (c) Efficiency of laser system with voltage. (d) Spectra of all 24 diodes taken with Thorlabs visible spectrometer (resolution sub-1nm) at P = 50 W and T = 45 $^\circ$C. (e) Central wavelength of 24 diodes and their distribution with FWHM.

Download Full Size | PDF

 

Fig. 4. (a) Typical spot size on a test surface. (b) Laser cutting of a copper sheet (t = 150 µm). (c) Picture of the Cu–solder welding. (d) Copper sheet-battery welding. (e)–(g) SEM images of Cu–solder interface after welding.

Download Full Size | PDF

We demonstrate another interesting application of the laser–copper interaction by making copper micro-strips mechanically resilient against failure when subjected to extreme cyclic loading. For this purpose, we first show that a pristine copper strip easily fails only after N=1 to 2 cycles of 180$^\circ$ bending as expected for a metal deformed beyond its elastic limit (inset Fig. 5(a) shows schematics of experiments). A series of SEM images of pristine work-hardened copper (1<N<2) show that the bending strain induces several microscopic and nanoscale cracks along the folding line which causes structural failure. In contrast, a blue-laser-annealed copper strip at 0.5–4 kW/cm$^2$ intensity (1 mm/s scan speed) significantly enhances its ability to withstand cyclic loading without failure. The maximum number of bending cycles before failure progressively increases with laser intensity and could reach up to $N_c=29$ cycles in this case (Fig. 5). To understand the origin of this mechanical response, we take a closer look at SEM images along the folding line after N = 7 cycles. Remarkably, the laser-annealed sample shows the absence of nanoscale cracks along the folding line. Rather, contrasting nano-bumps with 20–50 nm edge-line are visible suggesting enhanced malleability of the copper.

 

Fig. 5. (a) Critical loading cycle for a copper strip exposed to different laser intensities over folding lines. Inset: schematic of a single-loading cycle. (b)–(e) SEM image of pristine copper (N=1–2) with magnification from 100 µm, 10 µm, 1 µm, and 100 nm, respectively. (f)–(i) SEM image of a laser-exposed copper after N=7. Some nanoscale cracks, highlighted by yellow arrows, are absent in laser-annealed sample after N=7.

Download Full Size | PDF

Previously, thermal annealing above the recrystallization temperature ($T_{rc}\sim 325$ $^\circ$ C) of a copper film on a substrate was shown to increase the grain size leading to improved malleability [30,31]. We provide further insight into the absence of nanoscale cracks in laser-annealed samples using grain growth in copper. It is known that thermal annealing increases the grain size $d$ following $d^2-d_0^2 \sim t_{on}e^{-U/kT}$ [32] where $d_0$ is the initial grain size, $t_{on}$ is the laser annealing time, U is the barrier height, and k is the Boltzmann constant. This equation correctly accounts for the recrystallization temperature of copper which is about one-third of its melting temperature. For our experimental parameters (U = 0.87 eV [33], $t_{on} = 30$ s) grains are roughly an order of magnitude larger for $I=3$ kW/cm$^2$. Larger recrystallized grains allow a more homogeneous distribution of stress which reduces the possibility of nanoscale cracks upon loading [30,31]. It is also worth mentioning that the observed response of copper after laser annealing is in contrast to the blue-laser-induced hardening of steel [13]. The demonstrated blue-laser annealing of copper has advantages since it is spatially localized and rapid at kW/cm$^2$ intensity.

In summary, we demonstrated an innovative all-solid-state portable CW blue laser up to 100 W power and validated its thermal and optical performance. Our laser system does not require any coolant fluid, refrigerant, or ozone-depleting chemicals. We demonstrated the cutting, bending, and soldering of copper strips with nanoscale adhesion. Additionally, we show that copper strips after 4 kW/cm$^2$ laser exposure show resilience to withstand cyclic loading, owing to their malleability to mitigate nanoscale cracks despite cyclic loading.

We envision that it should be possible to scale up our laser design by adding diode banks and Peltier chips to multi-100 W laser output, limited by the maximum heat-removal capacity of the forced air cooling. It should be possible to employ recently available piezo-based thin air-chip technology to design laser heads without moving parts for unique applications.