What does integration level mean? The concept of integration level
What does integration level mean? The concept of integration level
Integration level refers to the minimum line width in a pattern. The integration level of an integrated circuit (IC) refers to the number of components housed on a single chip. Higher integration level means more components can be accommodated. Therefore, many improvements have been made to traditional photolithography methods to meet resolution requirements and increase the integration level of ICs.
What does integration level mean? The linewidth of an IC is generally understood as the minimum line width in the fabricated circuit pattern. However, in MOS circuits, gate length is also commonly used to define linewidth. There is a corresponding relationship between integration level and linewidth; that is, the higher the integration level, the smaller the linewidth. Therefore, linewidth is often used to indicate the level of IC manufacturing technology.
With the development of large-scale integrated circuits (LSI) and very large-scale integrated circuits (VLSI), the requirements for photolithography resolution are becoming increasingly higher. In 1995, it was 0.35μm; in 1998, it was 0.18μm; the current goal is 0.18μm, followed by 0.13μm.
The Concept of Integration Level Since the birth of integrated circuits, their integration level has been on an ever-increasing, irreversible path. Experts have summarized this trend as "Moore's Law." However, Moore's Law only describes the phenomenon of increasingly higher integration levels; what truly drives this phenomenon is the fierce competition driven by the pursuit of profit.
Simply put, the benefits of increased integration level include: 1. The area of a chip with the same complexity can be reduced by about half (regardless of core-limit or pad-limit, focusing only on logic). This is equivalent to reducing the manufacturing cost of a chip with the same price by about half using a new process (not considering NRE);
2. Or, a chip with the same area can provide twice the computing power. This is equivalent to selling a chip with the same cost for twice the price;
3. Better manufacturing processes lead to lower power consumption because the driving voltage is reduced, and dynamic power consumption is proportional to the square of the voltage. For the same performance, lower power consumption means longer standby time, allowing for a higher selling price;
4. But I think the most important benefit is the increase in clock frequency. The more advanced the manufacturing process, the shorter the latency, and the higher the maximum clock frequency a chip can achieve. Do you remember the era when each generation of CPUs had a higher clock speed than the last? Moreover, low power consumption is even more significant for handheld devices. For desktop chips like CPUs and GPUs, the pursuit of performance far outweighed the pursuit of low power consumption. Back then, we only heard about researching how to cool CPUs and GPUs; nobody cared about low power consumption. You probably remember that the latest CPUs were always very expensive, with Intel's gross margin exceeding 60%.
The semiconductor industry is a highly competitive industry. To survive and pursue higher profits, participants are forced to constantly upgrade, keep up with trends, and avoid being eliminated. This leads to the phenomenon described by Moore's Law, while also benefiting consumers. Therefore, we can say that the higher the integration, the better, and the smaller the transistor size, the better.
It's important to note that technological upgrades require significant investment in research and development. In previous generations, people hoped that profits would exceed investment. However, as sizes shrink further, investment costs increase, and perhaps one day it will become unprofitable. That day should be the end of Moore's Law. As long as scientific and technological advancements continue to provide us with ways to generate profits, Moore's Law may well continue.
