Experts describe the trends for this emerging technology.
Chipmakers are ramping up a new class of chip architectures, such as 3D NAND and finFETs. Measuring and characterizing the tiny structures in these technologies is a major challenge. It will not only take the traditional metrology tools, but also various X-ray techniques.
To get a handle on X-ray metrology, Semiconductor Engineering recently discussed the trends with the following experts: David Rossi, president of the Semiconductor Division at Bruker; Isaac Mazor, vice president and general manager of the X-ray semiconductor business at Bruker; and Adam Feinstein, director of product management at Bruker. Bruker is a supplier of AFMs, X-ray metrology tools and other products. What follows are excerpts of that conversation.
SE: Bruker has made several moves, including the recent acquisition of X-ray metrology tool specialist Jordan Valley Semiconductors. What was behind that deal?
Rossi: It broadens our customer base. Also, when Bruker formed the Bruker Semiconductor division in January of this year, it was to align our X-ray and AFM technologies and products to be able to solve our customers’ metrology requirements. We also saw that we could serve our customers by joining forces with Jordan Valley in X-ray metrology. By combining Jordan Valley’s product and technologies with Bruker’s X-ray technologies, we really saw the ability to not only meet our customers’ current needs and requirements, but also their future requirements at 7nm and 5nm.
SE: What are the big challenges in metrology today?
Mazor: The challenges are increasing. The number of parameters of critical value is growing. And the complexity of these new structures is driving metrology.
SE: Chipmakers are using traditional metrology tools. What about X-ray metrology?
Mazor: As we move to 28nm and below, and down to 10nm and 7nm, we will see a substantial increase in X-ray metrology. It’s a must and a significant advantage. The two technologies that will require X-ray for 3D geometries are 3D memory and finFET structures. Both create a need for 3D metrology. X-ray can play a significant role.
SE: Didn’t X-ray metrology begin to make inroads in the semiconductor business at 45nm?
Mazor: There were several nodes that made X-ray a more viable option for metrology. The first one was perhaps copper metallization, which started at 180nm. Then, you are right when mentioning the 45nm node. When several companies adopted germanium as a strain, you started using strain-silicon. That brought in another X-ray metrology for photometric and concentration analysis. That significantly extended the X-ray market.
SE: Why do we need X-ray metrology?
Mazor: We are dealing with angstrom-level measurements.
SE: Where is X-ray metrology being used?
Feinstein: Now, we are seeing (it used) even more in strained-silicon with silicon-germanium. We are seeing it with raised source/drain at 14nm and 10nm. We will likely see it with III-V or SiGe fins as we go to 7nm and 5nm.
SE: Which X-ray techniques are required for these applications?
Feinstein: Generally speaking, we can think about XRD, which is X-ray diffraction, at the need to really qualify SiGe and the epitaxial quality of those films, as well as a way to directly quantify the targeted strain engineering in the channel.
SE: Does X-ray metrology replace the traditional techniques like CD-SEM and OCD?
Feinstein: It’s not necessarily that you are replacing anything. But as we integrate more materials, and we have more complex processes, whether that’s multiple patterning or three-dimensional structures, there’s the need for new and more advanced metrology. This overcomes where some of the traditional techniques may be hitting some of their limits.
SE: What types of X-ray metrology tools are being used today?
Feinstein: Originally, it was XRR, which is X-ray reflectivity. That was for simple thin-film measurements. Now, we are really seeing a transition to XRD and XRF.
Mazor: XRR is important, but actually XRF was adopted long ago, especially in Japan. They have been using it for a long time. XRR and XRF, in a way, have become viable metrology technologies starting 10 years ago. XRF is growing. It is a photometric technique. It’s more capable for 3D geometries.
SE: XRF, or X-ray fluorescence, looks at surface contamination. What about XRD?
Mazor: XRD is back in play. It was sometimes useful at one time, but then it was put aside. XRD is now becoming a critical metrology technique for the selective growth of epi. The market will use it for II-VI and III-V. Those atoms will require selective growth. XRD is a technique to look at the quality of the epi and to qualify the repeating patterns of finFETs. XRD is emerging for 10nm and below. We believe it will be adopted as an in-line metrology, not just a metrology for ramp up.
SE: What are some of the differences between these X-ray techniques?
Mazor: X-ray techniques have three modes of operation in general: X-ray in and X-ray out; X-ray in and electron out; and electron in and X-ray out. XPS is the one with X-ray in and electron out. It’s for composition and thickness at the angstrom level. At Bruker, we primarily deal with X-ray in and X-ray out. XRF, XRD and CD-SAXS are X-ray in and X-ray out. This is where we dominate. E-beam in and X-ray out have different applications. This is different than what we do.
SE: For years, the industry has been working on CD-SAXS. It’s still not ready. What about CD-SAXS?
Feinstein: If we go beyond epitaxial, this is really a XCD concept. This could be GI-SAXS or CD-SAXS, which is small-angle X-ray scattering. That is applicable to any type of material. So we can go from crystalline epitaxial, polycrystalline and resists. That’s really where X-ray enters into some of the traditional optical scatterometry spaces.
Mazor: Basically, with CD-SAXS, you can do metrology with a 7nm or 5nm structure with no calibration. That saves you a huge amount of cost. The problem was, and still is, the productivity of this technique. But there is a lot of progress on this front. We will continue to catch up on the productivity. Once the productivity is achieved, there is no match for this powerful, non-destructive technology. We think this is the next technology for 3D memory, advanced epi and finFETs.