InsightFab
Knowledge Base/Critical Quality Indicators for Lithography Process
Semiconductor Process6 min read

Critical Quality Indicators for Lithography Process

This article highlights critical quality indicators in the lithography process, beginning with a real-world incident that underscores their importance. It explains the core challenges of CD control and overlay accuracy, demonstrating why nanometer-level precision is crucial for process yield in semiconductor manufacturing.

That Day, When the CD Report Came Out, the PM's Face Turned Green

I remember one time, during the night shift at the wafer fab, I was about to hand over equipment when suddenly the production line alarm blared, and red lights flashed incessantly. I looked at the machine screen, and damn! The Litho CD report for the previous batch was out, and several points had directly shot out of specification! The PM's face turned green when he got the call, because this batch was an urgent order from a major client, originally scheduled for shipment the next morning. The entire team immediately tensed up; everyone knew that if lithography wasn't managed well, the subsequent several process steps would be futile, and millions worth of wafers in that entire batch could be directly scrapped. At that moment, you truly understand what "critical quality indicators" are.

Where Was the Problem? If the Dimensions Are Wrong, They Are Wrong!

To put it simply, the critical quality indicator for the lithography process is to ensure that the pattern you etch is "exactly the same as your design." Sounds simple, right? But in reality, maintaining nanometer-level precision is incredibly difficult.

There are two main "bosses" (major challenges):

  1. CD (Critical Dimension) Control: This refers to the linewidth and aperture size you etch. Think about it: if the design specifies 50 nanometers, but you etch 55 nanometers, the transistor will short-circuit and simply won't function properly.
  2. Overlay Accuracy: The chip manufacturing process isn't completed in one etching step; it requires stacking layer by layer. Imagine building with LEGOs: if each layer isn't aligned properly, the final structure will be crooked. The lithography process is the same; you must ensure that the pattern of the current layer is perfectly aligned with the pattern of the previous layer.

So the key is: CD must be accurate, and Overlay must also be accurate. If either of these two items has a problem, no matter how much you try to remedy it later, it's useless—just like a burnt-out battery, no matter how much you charge it, it won't hold power.

How Is It Actually Done? Just Look at the Numbers!

What we engineers constantly monitor are these numbers.

  1. CD Monitoring: We regularly sample and measure CD values on wafers. Suppose your spec is 50 ± 5 nanometers, but you measure an average of 53 nanometers for a batch, with a standard deviation of 1 nanometer. At this point, we calculate an index called Cpk. If the Cpk is only 1.08, it means your process capability is not very good, and it's easy for products to fall out of specification. We aim for a Cpk of at least 1.33 or higher; the greater, the better.
  2. Overlay Measurement: This is even more direct. We design special "overlay marks" on the wafer and use the machine to measure the alignment deviation between the upper and lower layers. Suppose your specification states that the deviation cannot exceed 10 nanometers, but you measure a deviation of 12 nanometers at a certain point; then that point is immediately "No Good" (NG). We look at the Mean deviation and Sigma (standard deviation), and of course, smaller values are better.

So the key point is: these numbers must be monitored daily, every shift, and for every machine. As soon as a trend is incorrect, immediate intervention and adjustment are required.

The Most Common Pitfall: Adjusting Parameters and Making Things Worse

I still remember when I first joined the company, there was a time when CD values started to drift. My senior colleague told me to adjust the exposure energy. I, being clever, thought I'd adjust it a bit more at once to see if the effect would be better. What happened? Although the CD values came back, the Overlay started to fluctuate wildly because too much energy adjustment changed the photoresist sidewall shape, affecting overlay accuracy. That time, my senior colleague gave me a stern lecture, saying I was "fixing one thing while neglecting another."

Frankly, many times, parameter adjustments are like pulling one hair and moving the whole body. You think you've solved one problem, but you might have created another, bigger one. Therefore, every parameter change must be data-supported, executed in small, rapid steps, and not attempted to be perfect in one go.

One Thing You Can Do Today

Go back and check your quality reports: are the Cpk values for CD and Overlay meeting their targets?

Want to try it yourself?

Every tool mentioned in this article is available on InsightFab — just upload a CSV to analyze.

Go to Tools