Negative Expansion Materials Resist Warpage

Molding compounds and underfill can improve thermal stability. Key Takeaways: Thermally induced warpage, caused by differences in the rate at which materials expand when heated, is becoming increasingly problematic in large chips and packages. The solution may be to integrate a new class of materials that actually shrink when heated. Rather than relying solely on the few substances that can shrink, companies are focusing on mixtures of materials with a lattice structure that can resist the urge to expand. Using such materials for underfill and as the packaging molding compound can add rigidity for larger packages while keeping warpage in check. Warpage as the enemy Warpage, especially in large packages and panels, occurs because of mismatches between how, for example, a substrate, a die, and surrounding molding compound behave when heated. It’s one of the biggest issues holding back the constant enlargement of package substrates, interposers, and panels. The larger the bonded area, the greater the likelihood that edge connections may be poor or fail altogether. Even before bonding, too much warpage can defeat standard manufacturing equipment features such as vacuum chucks. “The vacuum chucks will pull the warpage down, but if the warpage is too high, then you won’t get a good vacuum,” said Knowlton Olmstead, senior manager of wafer services at Amkor. Anywhere two materials meet with different coefficients of thermal expansion (CTE), warping is likely to some degree. For example, given two layers of material, if the bottom layer has a higher CTE, then it will try to expand to its natural extent. But the material above it will restrict it, assuming they’re well adhered. That will cause the edges of the bottom material to want to curl up. It’s theoretically impossible to stop all warpage if there is a CTE difference. The question becomes how much is too much. If it’s too much, what can you do about it? “We do a certain level of modeling to model the warpage beforehand, and then there’s certain levers you can pull to control the warpage,” said Olmstead. If silicon is one of the layers (as it often is), thinning the silicon can reduce the amount of warping. “You reduce the impact of the low CTE silicon if there’s thinner silicon,” said Mike Kelly, vice president, chiplets/FCBGA integration at Amkor. “In the extreme, you can make it super thin, and then it can follow the shape of whatever I’m attached to. But that ends up being a problem, because nobody wants really thin silicon for managing high power.” Warpage also impacts how new materials may be received, such as glass. “The larger the panel you’re trying to make, the more warpage you will see,” said Lang Lin, principal product manager at Synopsys. “Today we talk about micrometers of warpage, but with glass, it could be even larger — millimeters, maybe.” One approach being explored is to identify or design materials that contract when heated, giving them a negative CTE. “Negative CTE materials could be utilized in a variety of ways to offset expansion,” said Luke Prenger, research scientist IV team lead at Brewer Science. “It might depend on why the expansion needs to be offset to begin with. For things like stress management and warpage reduction, negative CTE materials can be utilized to help better match the CTEs of different materials. An example case of this would be to add filler to a polymer material, which typically has a very high CTE compared to inorganics, to reduce the overall CTE of the polymer matrix so it is closer to that of the metal layers and other components in the device. Reducing or modulating the CTE of organic layers to be closer to the inorganic layers helps in reducing warpage and overall stress build-up in a device.” Going against the grain One of the material interfaces that results in warping is the epoxy molding compound (EMC) used for fillers and underfill. It fills all the major gaps inside the package, so when the temperature changes, it can affect everything it touches. Altering the composition of that compound can mitigate the warping. The key is changing the CTE, with a negative CTE ideal for counteracting an adjacent positive one. Alternatively, mixing a positive CTE material with a negative one can alter the overall CTE of the combination. A variety of materials can exhibit true negative thermal expansion (NTE) behavior, although it often occurs within specific temperature ranges. It may also be associated with a phase change. For example, water behaves normally above 3.98 °C, below which it moves toward freezing. Ice is famously less dense than water due to its molecular structure, and so as more of the water takes on that configuration between 3.98°C and 0°C, it expands up until the final phase transition to ice. Unlike materials exhibiting true NTE behavior, other substances may consist of normal materials encased in a lattice that restricts expansion. The structure of that lattice is important in determining how anything encased in it behaves. “The key is the structure of the filler itself,” said Sanjiv Bhatt, senior director of global marketing and business development for the semiconductor business at Mitsubishi Chemical Group. Solids may have a wide range of molecular arrangements. Whereas polymers can be amorphous, inorganic materials typically have some sort of lattice structure. Materials scientists categorize the different structures and study their rheological properties (which refers to the study of how they flow). They’ll describe lattice structures, such as house-of-cards, rigid-rod, or tetrahedral. Each of these structures behave differently and can establish rigidity in one or more dimensions. What’s within the lattice (also called a matrix or a network) can also affect the flow, and the combined NTE materials reflect such a structure — a functional material within a restraining matrix. “It’s more of a network thing than a material thing,” explained Bhatt. “Not everything will form the same network. A carbon powder forms a flocculated network. If you look at a carbon fiber network, that is a reinforcing rigid-rod network. So if you are going to put a carbon fiber into a matrix, it’s like logs in a river. If I use the logs in water, it is going to have little impact on the ability of water to flow. It will restrict it some, but the water is still going to find room between them. If I put a whole bunch of clay in the water, it will muddy it up, and it will literally restrict the water movement because now you have changed the viscosity.” Caught in a cage Such a structure, a matrix encasing another material, can restrict the tendency for the enclosed substance to expand. It can even contract a little when heated by reorienting the molecules in the structure. Rheologically, expansion and contraction are studied as flows, so in the restricted case, the flow is redirected in a way that avoids the warpage that might otherwise occur. “A lot of people will talk about shrinkage and expansion, but to me, those are low shear-rate flows rheologically,” said Bhatt. “If you’re looking at something in a matrix, when you heat it up, the reason it shrinks or expands is because it has to flow. The molecules want to relax. That flow finds the easiest path. Now you integrate something to cage it. Then you change the dynamics of that flow.” “In these [filler] materials you have a house-of-cards, so it’s a lattice that does not allow movement, and the network arranges itself so the resultant matrix turns out to have a negative CTE,” Bhatt said. “Self-assembly is where chemistry comes into the picture — you have to make sure that these things are functionalized appropriately so they arrange themselves in a certain way.” Having one material trying to expand but being restrained by another structure raises the question of whether the matrix could ever burst under pressure from the expanding material. The answer is yes. This is called melting, at which point the entire configuration falls apart. “If you have a matrix and you melt it and it starts flowing, that’s the bursting analogy,” he said. “But by that time, the material has the matrix thermal capability.” Even though this matrix changes the flow characteristics of the material it encases, it doesn’t change the characteristic temperatures of that material. “The reinforcement does very little for the thermal transitional behavior of the matrix material,” noted Bhatt. “It will never change the glass transition or the melt transition temperature.” This also isn’t about establishing a zero-CTE material. The idea is to tune the compound’s CTE to minimize differences between one material and another. “Our negative-CTE filler contracts as temperature rises, offsetting the natural expansion of the resin,” said Bhatt. “By lowering and tuning the effective CTE of the molding compound, it reduces thermal-expansion mismatch among the die, substrate, and encapsulant, helping minimize stress and package warpage.” NTE materials debut NTE materials are unusual, which is a downside. “The number of negative CTE materials is very limited,” confirmed Prenger. “β-eucryptite is actively used as ceramics for chunks and parts in certain tooling, and zirconium tungstate is being utilized as an additive for various types of mold compounds and underfills.” While eucryptite (meaning “well concealed”) is found in nature (if you look hard enough, apparently), zirconium tungstate is not. With NTE in all three dimensions, it’s one of the human-made NTE substances. Such materials are very challenging to design. “On top of negative CTE materials already having a limited selection, designing new materials is very complicated and exists only in the realm of inorganic materials,” explained Prenger. “For organic materials, for almost every polymer material that is noted to have negative CTE, it is negative only in one dimension, not in three dimensions. This leads to organic materials being exclusively limited to adding inorganic fillers.” Although NTE materials have been a research topic for a while, they’ve recently become available from Mitsubishi Chemical Group. It’s an inorganic material that can be mixed with organic epoxy resins to customize the epoxy’s CTE. It’s also available pre-mixed with resin. Fig. 1: Mitsubishi Chemical’s NTE material can be mixed into epoxy resins for electronics packaging. Source: Mitsubishi Chemical Group If this proves out and other materials companies follow suit, this could become a standard formulation for semiconductor packaging. But for commercial success, these materials must perform over a wide temperature range to cover not only processing but also higher-temperature manufacturing operations. They must be easily and uniformly mixed into resins. They must also not introduce unexpected effects from alpha-particle emissions caused by impurities. If all works as planned, this may be yet another step toward enabling larger packages and panels. — Gregory Haley contributed to this article. Related Articles Backside Power Delivery Creates Fab Tool, Thermal Dissipation Barriers Moving the power delivery network to the backside of a chip reduces congestion, but it introduces new challenges for fabs. Advanced Packaging Limits Come Into Focus Mechanical and process control limits are now shaping what can be manufactured at scale. Panel-Level Packaging’s Second Wave Meets Engineering Reality The cost case is strengthening, but glass, warpage, and bonding yield stand in the way. Leave a Reply