Issue



Thermal interface materials for high-power BGAs


04/01/2000







Careful planning can prevent increased manufacturing costs and field failures.

ROBERT KRANZ

High-power ball grid arrays (BGAs) present one of the greatest challenges for thermal interface materials, primarily because they generate much heat in a small area. Phase change materials have largely replaced grease in this application area because they are cleaner to use and provide much greater resistance to the liquid state being pumped out of the interface. But a complex series of tradeoffs must be considered to optimize the application of these devices. For example, low phase change temperature materials activate at low temperatures but increase the chance that the interface material will be damaged during shipment. Reducing material viscosity can lower assembly costs but increase the variability of thermal performance and, in extreme cases, the liquid phase change material may drip off of the assembly. This article explores these and other issues involved in selecting and applying thermal interface materials for high-power BGAs.

First, it is necessary to define high-power BGAs. For this article, high power refers to devices having greater than 40 watts per square inch of heat to dissipate. Some devices with BGA packaging are moving toward more than 100 watts per square inch, and the future promises even higher power dissipation levels. Another way of defining this is a thermal path that requires a maximum of 0.5°Cin2/watt thermal impedance.

It is also important to examine the material properties that are critical in defining the properties of thermal interface materials and BGA applications. Thermal conductivity is simply the material property that describes the relative ability of a material to conduct heat. This value is independent of thickness and is reported in watts/meter Kelvin and several other units. Thermal resistance is the material property that describes the temperature drop through a material at a given thickness. It is measured in °Cin2/watt or °Cmm2/watt. Thermal impedance is an application measurement that defines the increase in temperature over the thermal stack-up from the surface of the heat sink to the surface of the device. This figure is the sum of material thermal resistance, interfacial thermal resistance of the device to the interface material and interfacial thermal resistance of the heat sink to the interface material. Thermal impedance is reported in two forms: °C/watt of a specific device and °Cin2/watt normalized to an area.

A Full Performance Picture

The reason for defining these different properties is to make the point that thermal conductivity and thermal resistance, which are frequently used to compare alternate thermal interface materials, only provide a limited picture of their performance. These two parameters can be considerably important in applications where the thickness of the interface material after assembly is relatively high because of the need to fill a gap. In the more common situation where the heat sink is clip-mounted, the interface thickness is minimized and these values are less useful for decision-making (Figure 1).

In the case of a clip-attached heat sink, thermal conductivity and thermal resistance values do not provide an accurate prediction of how a specific material will perform in a given application. Availability of thermal impedance values for the same device or one of similar size is a better guide to thermal interface material selection. After this data has been reviewed, a true application-specific thermal impedance measurement of the design assembly is required.

It is necessary to consider the full range of options when selecting thermal interface materials for BGAs. Liquid bonding adhesives do a good job of conforming to the surface before cure and usually do not have problems with separation or pumping after curing. However, these materials can present problems during the assembly process because they require a cure cycle and cannot be reworked. Elastomers, on the other hand, are relatively easy to handle and resist separation. However, elastomers cannot provide the low values of thermal impedance required for most high-power BGA applications. In most cases, engineers have found that these concerns have disqualified these materials.

Thermal Grease vs. Phase Change Materials

For these reasons, the vast majority of high-power BGA applications use either grease or phase change materials. Grease usually meets thermal impedance requirements of high-power BGA applications, but problems often arise in the assembly area. Because greases are liquids, they can be messy in production, where workers can get grease on their fingers and transfer the grease to other surfaces that collect dust. This is usually not a functional problem, but perceived workmanship suffers. In addition, the responsibility for applying the interface material has shifted to the heat sink supplier, and it is difficult to ship the heat sinks with a liquid grease applied without protective covering. The filler and liquid that are combined to deliver the right thermal performance can sometimes separate in the field, especially under relatively hot conditions. In this situation, the resin leaks out or evaporates and the interface dries, which increases the thermal impedance, causing the device to run hot and can possibly lead to failures.

Phase change materials are preferred to grease in most BGA applications because they provide similar thermal performance while offering improved resistance to separation and a clean assembly process. The defining characteristic of phase change materials is their ability to convert from a solid to their lowest viscosity state over a narrow temperature (Figure 2).

A thermal interface pad can be placed like a label onto a heat sink in its solid phase. It will attain a grease-like consistency at a specific higher temperature. In its liquid state, the phase change compound flows to wet and conforms to the profile of the mating surfaces. This process lowers the thermal impedance of the assembly from that point forward.

Fillers are then added to control the thermal conductivity and the viscosity above the melting point of the phase change material. Viscosity should be high enough to prevent the interface material from flowing under vertical orientation; this prevents the compound from dripping or draining out of the assembly. This characteristic can be tested by attaching the phase change pad onto a heat sink or spreader and placing it in an oven vertically at the maximum device operating temperature. If the pad does not flow, then pad viscosity is high enough to prevent dripping.

Advantage of Heat-free Adhesion

However, the versatility that makes phase change materials so effective can also create complications in the selection and application process. This is particularly complex in the common scenario where a phase change material is applied to a heat sink by the heat sink supplier, shipped to the computer assembler and assembled to the computer processor on an as-needed basis. This is because thermal performance must normally be traded off for assembly and handling considerations. For example, some phase change materials require the heat sink be heated to provide adhesion so the heat sink and thermal interface can be shipped as a single component. This process is more costly to complete than materials that have intrinsic tackiness or a layer of pressure-sensitive adhesive. Therefore, it is important to consider manufacturability issues during the design process, rather than simply specifying the highest performance material without any regard to assembly cost.

Tradeoffs are also required when specifying the phase change temperature of the interface material. The advantage of a material with a low phase change temperature is that the device does not have to get very hot to activate the material. The downside to this is that specifying a material with the lower phase change temperature can cause seasonally high temperatures to destroy the material pad form during shipping. Yet another potential pitfall is that the processor assembly may operate above the phase change temperature and increase the potential of pump-out, leading to premature device failure.

Assembly Pressure vs. Handling

It may be necessary to trade off assembly pressure for performance variability and handling concerns. Reducing the amount of pressure used in the assembly process can lower the cost of the clip and reduce the possibility of the PCB bowing after attachment. The variability of the thermal impedance, on the other hand, can increase as pressure decreases. The magnitude of the thermal impedance is also reduced as pressure increases. Therefore, the higher assembly pressure often wins out by lowering processor mean temperature and result variability, cutting warranty returns.

Design viscosity of a material is another consideration when selecting materials. If lower viscosity is combined with a phase change temperature below 60°C (a standard maximum shipping temperature specification), there is a risk that the material will be destroyed during shipment by dripping off of the assembly. A higher viscosity material will require a little more pressure to optimize the thermal performance, thus decreasing the risk of material loss during shipment.

Overall, there are many factors to consider when selecting a thermal interface material for a high-power BGA. In all, an ideal material will have a carefully balanced mix of the following properties:

1. low thermal impedance to maintain the device below its maximum operating temperature;

2. the ability to adhere to the heat sink for shipment to the assembler without being heated;

3. a phase change temperature below the maximum operating temperature of the device but as high as possible to avoid shipping problems; and

4. the ability to flow under pressure of assembly but not drain or drip.

Addressing these issues upfront will avoid the potential pitfall of selecting an interface material on the basis of only one or two performance specifications, then to discover an increase in manufacturing costs or field failures. Of course, the good news is that the thermal performance of phase change materials is continually improving to accommodate industry requirements for thermal performance, handling and productivity. AP

ROBERT KRANZ, senior product manager, can be contacted at The Bergquist Co., 5300 Edina Industrial Blvd., Minneapolis, MN 55443; 612-820-6562; Fax: 612-835-0430; E-mail: [email protected].

Click here to enlarge image

Click here to enlarge image

Figure 1. Illustration of the diminishing effect of thermal conductivity on the thermal resistance as the thickness decreases, assuming the interfacial thermal resistances are constant.

Click here to enlarge image

Figure 2. Complex viscosity curve of phase change compound.