Cubechips
Cubechips is the common term for a common integrated circuit packaging architecture enabling high volumetric density integration of computation circuitry. The name is derived from the shape of the completed chip (not including external cooling systems) often being roughly cubical.
Structure
Cubechips consist of several stacked layers of semiconductor dies, typically utilizing dual-side lithography techniques, interspersed by layers of micro-scale support columns creating channels between dies. These so-called open layers are used to circulate an ultra-low viscosity cooling fluid to achieve evenly distributed cooling throughout the circuit. Some products include a fully integrated micromachinery cooling loop and radiator, allowing them to be used without any external cooling system, while others are designed to interface with external coolant circulation systems, allowing multiple Cubechips to be run off a shared cooling loop. Typically, cubechips have external connections on one side only, but there are standardized heights for dual-sided connectivity, allowing the chips to be mounted between two carrier boards for more compact implementation. In these cases, typically one side is heavier in power supply connections while the other is primarily used for signal connections.Applications
Cubechip architectures are typically used to produce highly integrated systems in a single package. Most Cubechips include several processors, memory, dedicated hardware accelerators, high-speed interfacing, and often reconfigurable hardware as well. Cubechips therefore often form the centerpieces of compact high performance electronics such as Processing Modules or Virtual terminals. There are also more specialized Cubechips, typically including only processors and memory, which are used in ultra-high performance computation systems such as those required for starship navigation.Manufacturing
There are multiple methods commonly used to create the inter-layer structures found in Cubechips, the most common being the so-called microdie approach where multiple smaller dies fabricated using similar methods as the primary logic dies are interposed to provide interconnection and structural support. This is generally a fast and cheap method, but it is limited in that the minimum size of the microdies is relatively large compared to other methods.
The second common methode uses precision material deposition similar to Universal Solid Fabricator technology to form interconnections and structural support. This method is significantly more expensive, but allows for much more control over support structures and greater coolant contact area.
Regardless of the method used for creating the support and interconnect structures, the coolant layers are typically filled with a soluble resin after the initial formation process, which is then dissolved at the end of fabrication. This adds structural stability during the fabrication process and improves yield.
Access & Availability
Very Common
Complexity
High
Comments