Samsung’s I-Cube is a heterogeneous integration technology that horizontally places one or more logic dies (CPU, GPU, etc.), and several High Bandwidth Memory (HBM) dies on top of a silicon interposer making multiple dies operate as a single chip in one package.
I-CUBE4
The successor of the I-Cube2 incorporates four HBMs and, one logic die.
With these arrangements, the I-Cube4 is expected to bring fast communication and power efficiency between logic and memory and provide better power efficiency.
The I-Cube4 will be ideal for high-performance computing (HPC) applications such as AI, 5G, cloud, and large data centers.
In general, to accommodate more logic dies and HBMs, the area of the silicon interposer should be increased. However, the silicon interposer in the I-Cube is thinner (around 100㎛ thick) than paper, as per the company.
“With strong expertise and knowledge on semiconductors, Samsung has studied how to control interposer warpage and thermal expansion through changes to material and thickness, succeeding in commercializing the I-Cube4 solution,” says the company.
As a result, Samsung has developed its mold-free structure for I-Cube4 to efficiently remove heat and enhanced its yield by conducting a pre-screening test that can filter out defective products during the fabrication process.
This approach provides additional benefits such as a reduction in the number of process steps, which result in cost savings and shorter turnaround time.
Samsung is currently developing more advanced packaging technologies to I-Cube6 and higher by using a combination of advanced process nodes, high-speed interface IPs, and advanced 2.5/3D packaging technologies.