We all understand that higher circuit density and increasingly compact microelectronic devices have created a need for more efficient cooling. Vapor chambers have emerged as a widely employed solution in some of the most demanding microelectronics applications. They are already found in high-performance computers, such as servers, workstations, and personal gaming desktops and laptops, as well as in mobile devices, including smartphones and tablets.
Vapor chambers are a form of planar heat pipe and operate in a manner like that of traditional heat pipes. Specifically, the operating circuitry heats a liquid within the heat pipe converting it into its gas phase. The heated gas physically moves away from the region where heat is produced through channels, carrying the heat with it. It then reaches a condenser located away from the active circuitry where the latent heat of the vapor is transferred out of the device, and the gas recondenses. The fluid circulates back to its original location to begin the process again.
Vapor chambers owe their success primarily to two key advantages as compared to other cooling methods, including traditional heat pipes. First, their thin, flat shape occupies less volume than a heat pipe. This makes them well-suited for use in highly space-constrained systems. Second, vapor chambers generally offer greater thermal transfer efficiency for a given area as compared to heat pipes. Furthermore, they typically can handle higher power densities (>50 W/cm²) are better and particularly useful for cooling circuitry with “hot spots.” Vapor chambers offer numerous other advantages, as well. For example, the thin shape of the vapor chamber makes them relatively insensitive to orientation, which is particularly useful for handheld and other mobile devices.
But, as the microelectronics miniaturization trend continues, there is a demand to make vapor chambers even thinner than current designs. This creates some challenges in terms of fabrication. One of these involves the welding process used in assembly of the most common vapor chamber configuration. This article explores how welding methods based on newly developed blue laser sources may provide a solution to this issue.
Vapor Chamber Production
Most vapor chambers in use today are formed from two separate, thin pieces of high thermal-conductivity metals. These are stamped to create the desired pattern of channels for fluid and gas flow. During assembly, these pieces are matched together, and the periphery of the assembly is sealed to complete the device.
For the device to function properly, it is essential that the two separate plates are joined with a hermetic seal. Additionally, the seam must be of sufficient mechanical strength to ensure that the integrity of the device is maintained over the lifetime of the product. In many cases, this means the seal must remain unbroken even in the presence of mechanical shock and vibration in addition to withstanding substantial temperature changes.
Currently, the most widely used production technique for joining the two plates is solid-state diffusion bonding. This process involves pressing the two plates together in a vacuum under high mechanical pressure and at an elevated temperature (but below the melting point of any of the materials involved). The seam area on both plates must be prepared so that they are smooth and debris free.
Under these conditions, solid-state diffusion will cause atoms from the two surfaces to intersperse with each other creating a permanent bond. While diffusion bonding doesn’t form a full fusion weld, the mechanical characteristics of the joint are still adequate for typical vapor chamber applications.
While diffusion bonding has long been used extensively and successfully, it does pose some drawbacks. One is the relatively long cycle time for evacuating the oven and heating the parts. It is possible to process several parts simultaneously, but as oven capacity increases, so does its capital cost and the amount of production floorspace it occupies.
Looking forward, the biggest issue with diffusion bonding is that thermal cycling tends to impart a permanent warpage in very thin copper vapor chambers. And, copper, with its high thermal conductivity, is by far the predominant material used for vapor chamber construction. This substantially limits the ability to continue using diffusion bonding as vapor chamber thicknesses shrink.
Laser Welding
Laser welding offers a possible alternative. This technique is already being successfully employed in several high-precision joining applications including numerous tasks, particularly in e-mobility and battery manufacturing, that involve joining thin metals.
For vapor chambers in particular, laser welding can deliver a substantially reduced cycle time. This is true even though it welds a single part at a time, rather than processing multiple parts at once, because the processing time per part happens within seconds.
While laser welding also requires fixturing for holding and clamping parts, these are generally much simpler than the tooling associated with diffusion bonding. This is because laser welding doesn’t have the same requirement for exerting high pressure evenly over the entire weld path. The computer programmable weld path of a laser also makes it easier to create new tooling for new designs. Plus, it allows the flexibility of switching geometries on the fly in production to accommodate changing the clamp.
Laser welding produces a true fusion weld while diffusion bonding does not. Thus, the seal integrity and strength are generally better, although diffusion bonding generally delivers adequate results for this application.
Finally, laser welding is generally a “greener” method than diffusion bonding. Specifically, much less energy is required to run the laser system than needed for the large oven and pumps utilized in diffusion bonding.
Blue Laser Welding
Given the apparent advantages of laser welding, why hasn’t it been widely adopted for vapor chamber production in the past? The problem again is associated with copper.
Copper is very highly reflective at the infrared wavelengths produced by the fiber or solid-state industrial lasers commonly used for metal welding. This is shown in the absorption graph in Figure 1.
This makes welding copper with infrared industrial lasers an inefficient process. A significant amount of power must first be delivered to the work surface melt it. This melted material absorbs infrared more strongly than the solid form. Continued absorption of laser light then vaporizes the copper. This initiates a process called “keyhole” welding, that’s name is derived from the narrow aspect ratio that is seen in the shape of the interaction zone (the melt pool and associated vapor).
The large amount of energy that must be supplied to maintain the keyhole welding process makes it inherently chaotic and unstable. Bubbles within the melt pool can eject molten material, causing spatter on the material surface leaving voids or porosity within the weld seam itself. The process window is relatively narrow, and the total heat load introduced into the part is relatively high. This can cause permanent warping, making the method completely unusable with thin metals.
The graph also shows that both copper and steel (the second most popular material for vapor chamber construction) exhibit higher absorption at blue wavelengths. The absorption of copper, in particular, is 13X higher at 450 nm than at 1 μm. This means much less laser energy is required to initially melt the material and then keep it molten.
The result is that blue lasers can weld copper and steel in both the keyhole and more versatile “conduction” modes. The latter involves melting the material, but not vaporizing it. This reduces the overall heat input so that part warping is avoided. Plus, the lower energy input makes the process more stable, consistent, and easy to control. The welds produced are free of spatter and voids.
joining two 100 μm thick phosphor bronze plate. No spatter is seen.
bronze plates. The weld is free of porosity and voids.
Blue Laser Implementation and Future Trends
Blue lasers in the 500 W output power range have been successfully used for vapor chamber welding. These are focused on to the workpiece, and either the laser or part is moved so that the beam traces along the desired weld path.
The weld quality achieved with this approach is good and consistent, and the overall production throughput possible is competitive with diffusion bonding.
However, as vapor chambers get thinner, even blue lasers have faced limitations. Now, a new generation of higher brightness blue lasers has become available that can overcome this problem.
But what is high brightness, and why is it needed to weld thinner parts? High brightness is essentially a combination of high laser output power, small apparent source size, and low beam divergence. A high brightness laser can be focused to very small spot sizes to achieve high power densities, which is essential for many materials processing tasks. Furthermore, this tight focus can generally be achieved more easily (with simpler and less costly optical systems) and over longer focusing distances.
A key benefit of high brightness is that it provides more flexibility in terms of exactly how laser power (and therefore heat input) is delivered and spatially distributed at the work surface. This provides the finesse required to weld very thin materials without warping. In particular, a high brightness blue laser is uniquely able to weld very thin copper because its high absorption minimizes heat input, and its high brightness enables that small amount of heat input to be carefully managed.
Another important advantage of higher brightness is laser scanner compatibility. A scanner moves the weightless laser beam using galvanometer mirrors rather than by bulk motion of the workpiece or optics. This can deliver a dramatic increase in processing speed. It also facilitates the processing of larger sized vapor chambers, enabling complex weld seam shapes and the ability to switch product geometries on the fly.
Getting Higher Brightness
Creating this new class of higher brightness blue lasers has required a different approach than the ones used in the past.
Virtually all commercial blue lasers operate by combining the output from numerous Gallium nitride (GaN) semiconductor laser diodes that output at about 450 nm (blue). First generation blue lasers used either multi-emitter laser bars or commercially available modules containing multiple single emitters. In either case, the individual emitters are in fixed positions spaced relatively far apart. The inability to reduce the spacing between these emitters, and most importantly, to control the pointing and divergence of each individual source, limits the achievable brightness of these designs.
To overcome this limitation, manufacturers are now building blue lasers using large numbers of chip-on-submount laser diodes. Each chip-on-submount laser diode can be individually placed with high positional precision into an array with relatively small spacings between the sources. Furthermore, separate collimating optics can be placed over every single diode source. Each of these are independently adjusted to precisely control the divergence and pointing of every laser diode. This arrangement delivers a much higher brightness combined source.
Blue lasers based on individual chip-on-submount devices have now been successfully scaled up to the 1 kW level. They have sufficient brightness to weld very thin copper parts using scanning systems, which delivers the speed and flexibility benefits mentioned previously. This makes blue lasers a practical alternative to diffusion bonding for high volume welding of vapor chambers now and in the future as product dimensions decrease.
