Laser diodes are one of a number of different types of electronic devices that generate heat during normal operation. Some power applied to such devices is lost as heat energy. To ensure stable performance and high reliability, excess heat must be channeled away from devices such as laser diodes, microprocessors, and power transistors and safely dissipated within materials surrounding the heat-generating devices, such as circuit boards or equipment enclosures, or within the environment.
No active device is 100% efficient, meaning supply power or signal energy will inevitably be lost as heat generated by the device. If not properly managed, the resulting rise in device temperature can reduce the operating lifetime of an active device such as a laser diode. In fact, with a typical high-power laser diode operating typically at 50% efficiency, it produces as much heat as it does light.
As the output power levels of laser diodes increases, the amount of heat produced per unit device area also increases, requiring efficient and homogeneous cooling of the laser diodes. By understanding the thermal behavior of different materials and how various three-dimensional (3D) structures, such as microchannel coolers (MCCs), can contribute to heat flow, it is possible to channel the heat away from a heat-generating laser diode through its package and heatsink to the surrounding environment. MCCs have been fabricated from a number of different materials, including copper and copper-on-ceramic substrates, with some MCCs, such as manifold MCCs, small enough to be embedded into the package of an active device for which they are providing a heat-transfer function (Figure 1).
The thermal conductivity of a laser diode package, or its packaging material or heat sink, refers to its capability to conduct heat, and is measured in watts per meter per degree Kelvin (W/m-K). Heat is transferred more efficiently through materials with high thermal conductivity or, conversely, with low thermal resistance. Copper, for example, has very high thermal conductivity, about 400 W/ m-K. Thermal resistance, or Rth, is the reciprocal of thermal conductivity, with lower values more desirable when high flow of heat is needed.
A value of Rth for a given material can be found as a function of the change in temperature, ΔT, that occurs with a laser diode at full output power and the power loss (Ploss) or thermal power dissipated (Pthermal) by the laser diode (Figure 2):
Rth = ΔT/Ploss
where ΔT is the difference between the hottest temperature on the laser diode, Tjunction, and some reference or ambient temperature, Tambient, such as room temperature:
T = Tjunction — Tambient
The area across which heat flows from a laser diode to a thermally conductive material also determines the thermal resistance of the material and the size of a heat sink required to dissipate a given amount of heat. When the power loss and T are the same for a smaller versus a larger heatsink, it can be concluded that the larger heatsink has lower thermal conductivity since the same amount of heat is being dissipated through a larger area. The thermal flow through a 3D structure, such as a heat sink or MCC, is rarely a simple, linear function. As a result, thermal management solutions for laser diodes must take into account mechanical designs of cooling structures, as well as the thermal properties of materials used to construct those cooling structures.
As the optical power levels (and heat) generated by compact laser diodes increase, various cooling approaches are needed to maintain the operating temperatures of laser diodes as close to ambient as possible. Due to the cooling limits of materials with even low values of thermal resistance, cooling structures required to dissipate the heat generated by high-power laser diodes can be excessively large unless fabricated with materials having outstanding thermal properties; novel cooling structures are also often needed to effectively flow heat away from an active device such as a laser diode or high-power transistor. Mechanical configurations such as MCCs with flow-through cavities allow the use of cooling fluids to transfer heat through the MCC and away from the source of the heat. Through the use of such a structure, the effective thermal resistance can be reduced by increasing the flow rate of a cooling liquid, such as water.
Although copper possesses excellent thermal properties, it is typically combined with different types of ceramic substrate materials to achieve the blend of thermal and electrical performance needed for achieving high reliability with high-energy active devices such as laser diodes. Typical ceramic substrate materials used for cooling structures include alumina (Al2O3), aluminum nitride (AlN), and silicon nitride (Si3N4). For efficient dissipation of heat, cooling structures employ several layers of ceramic insulator and multiple layers of copper to facilitate the flow of heat away from the active device (Figure 3). As noted earlier, these cooling structures may also employ cavities within the cooling structure with some form of cooling liquid at a predetermined flow rate that has been optimized for efficient transfer of heat away from the source.
The material properties of these ceramic insulators, as well as the bonding process used to attach the copper layers to each other and to the ceramic substrates, are critical to the long-term effectiveness of a cooling structure. Because of the thermal stresses created by the heat of a laser diode, for example, surrounding materials such as ceramic insulators will undergo expansion as a function of temperature, described by a material's coefficient of thermal expansion (CTE). CTE essentially defines how the size of a material will increase with temperature, with larger values denoting greater increases in size with temperature. Potential thermal stress at the interfaces between different materials can be minimized by using materials with closely matched CTEs. For this reason, silicon semiconductor devices are often directly mounted on alumina substrates, which exhibit very similar CTE values as silicon.
Similarly, the bond between copper layers and ceramic insulator layers in a cooling structure will be subject to some amount of stress at high temperatures, depending upon the differences in their CTE characteristics. Any potential thermal stress can be minimized by using ceramic materials with CTEs closely matched to that of copper. For example, diamond substrates, with a typical CTE of 1 ppm/K, are often used as foundations for heat sinks because of their stability with temperature, although copper, with a typical CTE of 17 ppm/K, would undergo stress at the interface of a bond between the copper and the diamond because of the large difference in their linear expansion rates with temperature.
For practical cooling structures formed of copper and ceramic material layers, however, the matches in CTEs between copper and the insulator materials noted earlier are much closer, resulting in less thermal stress with temperature. For Al2O3 ceramic substrates, for example, the CTE of 6.8 ppm/K is fairly close to that of copper, compared to a CTE of 2.5 ppm/K for Si3N4 and 4.7 ppm/K for AlN. However, the thermal conductivities of Si3N4 and AlN, at 90 and 170 W/m-K, respectively, are higher than the 24 W/m-K of Al2O3, for improved dissipation of heat in thermal cooling structures.
Additional material characteristics can guide a circuit or system designer in the selection of a ceramic insulator material for use in the construction of a laser diode cooling structure. For example, Si3N4 materials feature high mechanical strength compared to other ceramic insulator materials, with increased durability compared to conventional substrates treated with either active metal brazing (AMB) or direct bond copper (DBC) processes. Such material properties indicate that a ceramic substrate such as AlN, when used as the basis for a well-designed cooling structure with sufficient copper layers, can provide excellent thermal dissipation even for active devices with high power densities, such as laser diodes.
Thermal cooling structures rely on effective bonding methods for attachment of copper-to-copper and copper-to-ceramic layers. Two well-established attachment methods for attaching copper to ceramic substrates are DBC and AMB methods. DBC is a high-temperature process by which pure copper is melted and diffused onto the ceramic substrate. AMB is also a high-temperature process, in which pure copper is brazed onto a ceramic substrate. Many complex cooling structures may also require a large number of thin copper layers to enhance the flow of heat away from a heat source, and multiple thin pre-oxidized copper layers are typically attached by means of eutectic melts at extremely high processing temperatures, such as +1083°C.
Selecting materials with characteristics well suited for efficient thermal management is only part of the task of maintaining manageable temperatures for high-power devices such as power transistors and laser diodes; mechanical layout and design is often essential for achieving an effective heat flow away from a heat source. A number of factors must be considered in the design of a cooling structure, including the minimum or maximum physical dimensions required for a particular application, the expected operating lifetime, the cost, the required T between the inlet and outlet of a cooling structure, and even the type of liquid and flow rate to be used in a liquid-filled cooler.
MCCs are heat-exchanging structures that also provide electrical connections to active devices such as laser diodes. Multiple MCCs based on thermally conductive materials such as copper or copper/ceramic combinations, are used in arrays of laser diodes, to minimize thermal effects. The multiple MCCs may be arranged in serial or parallel configurations, with inlet and output holes in each MCC to enable the flow of a cooling fluid (Figure 4). In a series configuration, cooling occurs from the inlet to the outlet of an MCC. Heat builds towards the outlet, as the thermal energy of each laser diode adds together. In a parallel cooling configuration, heat flows in a direction crosswise to the inlet and outlet of an MCC. To minimize unwanted electrical connections, cooling fluids with low electrical conductivity, such as deionized (DE) water, are typically used to cool the MCCs.
This combination of water and copper, while it provides a low-thermal-resistance path for the flow of heat away from the laser diode, also has long-term effects that are detrimental to copper. Water, which is often referred to as “the universal solvent,” has oxidation effects which can wear down copper over time and shorten the lifetime of a copper-based MCC, with or without electrical potential applied to the MCC. Due to the effects of corrosion and erosion, the copper walls of an MCC containing the cooling fluid can break down into microparticles which can move throughout the fluid and deposit on other surfaces within the MCC, even eventually clogging the flow of fluid through the MCC. By adjusting the pH of the cooling fluid and by strict particle filtering control, it is possible to minimize the effects of a cooling fluid on the deterioration of an MCC's copper.
In addition, the choice of fluid velocity or flow rate of the cooling fluid can determine the rate of deterioration of the copper walls within an MCC's liquid cooling chamber. The flow rate is a tradeoff between achieving improved thermal dissipation at higher flow rates and causing faster deterioration of the cooler's internal copper walls. Quite simply, the greater the flow rate, the greater the flow of heat away from the source, but also the greater the amount of copper corrosion. Although MCCs are designed for typical liquid velocity as high as 5 mL/s, flow rates of 2 mL/s or less will result in less corrosion of copper surfaces exposed to the flowing fluid.
Testing and Simulating
As with many electronic designs, development of an effective MCC for high-power laser diodes involves a weighing of various tradeoffs including size, cost, complexity, and operating lifetime. Measurements can help in the analysis of the cooling requirements of a particular application, while 3D electromagnetic (EM) simulation software can greatly speed the optimization of possible MCC configurations for high-power-density cooling applications.
In addition, by reinforcing the simulation power and accuracy of the software with models based on measurements using a specialized test setup with pressure sensors to precisely measure the flow rates of cooling liquids through different MCC structures (Figure 5), it is possible to efficiently develop optimum cooling solutions based on custom requirements, without the time and expense of trial-and-error prototyping.
This article was written by Manfred Goetz, Product Marketing Manager, Rogers Power Electronics Solutions GmbH (Eschenbach, Germany). For more information, contact Mr. Goetz at