By Isamu Miyamoto

The author started laser materials processing in 1965 with developing the CO₂ laser, and since then has been involved in a variety of materials processing using CW (continuous wave) to USLP (ultrashort laser pulse). Among them laser welding has attracted a lot of the author’s interest because of its profound and interesting physics from linear to nonlinear processes. The author’s work on laser welding is classified into two groups; depending on the laser absorption process. In the first group, CW laser welding process is studied based on linear absorption process, and the research efforts are directed to understanding keyhole welding process, and expanding the process limits. In the second group, USLP welding is analyzed where the laser energy is absorbed by nonlinear process, and is shown to provide excellent welding performances so that the most process limits found in the first group are in principle removed. In this paper, laser welding technology is presented starting from its origin using a CO₂ laser to a new wave of laser welding brought by USLP.

CW LASER WELDING: LINEAR ABSORPTION PROCESS

Advantages and Limitations of Keyhole Welding

In the first group, CW laser welding of opaque material is analyzed to understand the laser-matter interaction mechanism, and thereby expand the process limits caused by weld defects. The weld bead obtained using our in-house CW CO₂ laser is, to our best knowledge, the origin of laser welding in the world, although the welding was of thermal conduction type and most laser energy was lost by Fresnel reflection resulting in low weld efficiency. Keyhole welding, realized later on by the advance of laser technology, reduced the reflection loss dramatically due to the multi-reflections of the laser beam in the keyhole, and weldable thickness increased by the penetration of the laser energy into the work. Without any doubt, keyhole laser welding is one of the most successful technologies in laser materials processing.

However, the performance of keyhole welding is limited by keyhole instabilities, since keyhole is maintained by a delicate balance between evaporation recoil pressure and surface tension and hence easily becomes unstable or even collapses. For instance, in the early period of laser welding where CO₂ laser was widely used, absorbing plasma produced above the keyhole had to be controlled by assist gas. Then the keyhole easily collapsed by the excess pressure of the assist gas. Thus we had to develop an in-process monitoring system by detecting the light emission from the plasma in early 1980s, which is, to our best knowledge, one of the earliest in-process monitoring systems though.

High-Speed Welding by Single-Mode Fiber Laser

The plasma absorption problems were solved by the development of high-power solid-state lasers to which the laser-induced plasma is transparent. However, a laser power of at least 1 kW was needed for keyhole welding, since lump-pumped solid-state lasers were of a mainstream of industrial solid-state lasers for welding in the beginning of 21^st century. Meanwhile keyhole welding was realized using a newly developed high brightness disk laser at nearly one order lower laser power of 200 W by a German group. Shortly we realized keyhole welding at laser power further one order below (25 W) using a single-mode fiber laser. The welding speed attained by the single-mode fiber laser was also amazingly fast, over 2 m/s, which was the highest welding speed at that time. We also showed that the maximum humping-free welding speed increases as the keyhole diameter decreases.

Laser welding of very thin foil is also a challenging technique. Our model for simulating surface energy indicates that welding changes into cutting at a condition of d>h (d=keyhole diameter and h=foil thickness) due to the instability of the keyhole, and fine welding of 10 µm thick stainless steel foil by single mode fiber laser was demonstrated.

Direct Observation of Keyhole Welding in Glass

Direct observation of keyhole welding in glass is also an interesting topic to us, since much clearer pictures than X-ray transmission method are available. The experiments showed that keyhole became less stable as the viscosity of the glass decreased, suggesting that the stability of the keyhole is reduced as the melt flow speed increases in accordance with the results in the keyhole welding of metal. Interestingly no cracks are developed during welding in all the glasses examined independently of the coefficient of thermal expansion (CTE). However, after cooling down to room temperature, while no cracks are developed in small CTE glass like fused silica, cracks cannot be avoided in large CTE glass such as soda-lime glass and borosilicate glass due to the shrinkage stress in the molten region (Fig. 2a).

USLP WELDING: NONLINEAR ABSORPTION PROCESS

Embedded Molten Pool

In the second group, USLP of fs to ps regimes is used for welding transparent material. When USLP is tightly focused into bulk glass, plasma is ignited by multiphoton ionization (MPI), and expands toward the incoming laser source by avalanche ionization, which contributes as a heat source for internal melting of glass (Fig. 3). This is a big contrast to CO₂ laser welding of metal where the plasma produced outside the work, and shields the keyhole. Very short Rayleigh length (typically 2-3 µm) of high NA lens used in USLP welding is converted into much longer plasma, making focus adjustment easier. Then the plasma embedded in bulk glass produces embedded molten pool, providing numerous advantages. One of the most significant advantages is that no cracks are produced in internal melting during and after laser irradiation independently of CTE.

A New Wave of Laser Welding

Now a question arises. Why is crack-free melting of glass possible by USLP and not by CO₂ laser welding? The answer is found in the molten pool embedded in bulk glass. In the heating process before melting, compressive and tensile stresses are produced in welded region and surrounding region, respectively, whether or not the molten pool is embedded. In traditional welding having free surface in the molten pool, the compressive stress is released on melting since the molten pool is plastically deformable. Thus the tensile stress is produced eventually in the molten region by the thermal shrinkage when cooled down to room temperature, and thus cracks are developed due to the brittle property of glass.

The situation in USLP welding of glass where the molten pool is embedded is completely different after melting, while no difference is found from traditional welding before melting. No shrinkage stress is produced when the molten region is cooled down to room temperature, because the molten pool embedded in bulk glass is not plastically deformable, and hence behaves like an elastic body.

In addition to the advantage of the stress field to prevent cracking, the embedded molten pool provides other numerous advantages. In the keyhole welding process, weld defects including humping, sputtering and keyhole collapse are caused by the instability of the molten pool. This means that in principle there exist no process limits in USLP welding caused by the instabilities, since no free-surface and no melt flow exist, and hence no instability occurs in embedded molten pool. Additionally some other advantages are found.

• Only joint interface is selectively melted unlike the case of keyhole welding, allowing not only energy saving but integration of low heat-resistance parts near the interface.
• Welding can be done even in a clean room due to no emission of fumes and no melt sputter.
• In principle, no limitation of thinnest weldable thickness exists, since no change in the surface energy occurs in embedded molten pool.

Properties of weld joint are also excellent. The weld joint provides mechanical strength as high as base material (see ICALEO 2012, M401; Miyamoto et al.) and hermetic sealing with leakage rates below the resolution limits of standard Helium-leakage test. Even multi-path welding is possible in high CTE glass without pre- and post-heating using USLP as shown in Fig. 2(b). Now a new welding field has been created by a “new wave of laser welding” using USLP.

USLP Welding of Si/Glass

The new wave brought by USLP can be utilized not only for glass/glass welding, but for welding of dissimilar materials such as Si/glass joining, for instance, which is important for sensing and actuating microsystems. While anodic bonding is widely used for this purpose due to its excellent joint strength and throughput, there are some disadvantages that joining has no space selectivity and needs high temperature and electric field. Although the laser-based joining procedure has a possibility of high space selectivity, ns laser pulse cannot provide weld joint with high joint strength due to the splash of molten Si, which is caused by superheating of the Si surface during ns laser pulse, resulting in the increase in absorption coefficient due to temperature rise.

We have developed Si/glass joining technique using USLP with high joint strength as high as 70-90 MPa, which is competitive to anodic bonding. The melt splash can be avoided since the temperature rise is delayed after laser pulse in USLP. Excellent joining of Si/glass with local space selectivity is shown by drawing a grid pattern by weld lines and cutting by a standard dicer along the street of the laser-welded grid sample; no damage of the weld lines is produced, demonstrating the applicability of our joining procedure to wafer level packaging (Fig. 4). In addition, USLP provides a  joining rate at least competitive or even superior to anodic bonding. Using typical values of pulse energy 3 µJ and number of pulse overlap N=10, spot diameter D=14 µm and pulse repetition rate f=2 MHz, for instance, provide joining rate as fast as approximately 40 mm²/s (see ICALEO 2012, M402; Miyamoto et al.). Since the joining rate increases in proportion to D and f, Si wafer of 6 inches can be completely welded in as short as 90 seconds if f=10 MHz is used. No damage is found with keeping hermetic sealing by accelerated life test (500 cycles between -40˚C and 85˚C) in laser-welded silicon-Borofloat 33 sample.

Dr. Isamu Miyamoto is a professor of engineering at Osaka University.