High numerical aperture EUV lithography: a revolutionary technology leading the next generation of chip manufacturing
The next generation technology that enables smaller transistors is high numerical aperture EUV lithography.
Moore's Law states that the number of transistors on a given area of silicon wafer doubles approximately every two years, and this gain drives the development of computing technology. For the past half century, we have viewed this law as an inevitable natural process like evolution or aging. However, the reality is very different. Keeping up with Moore's Law requires an almost unimaginable amount of time, energy and human wisdom, involving countless people and the most complex machines on the planet.
Perhaps the most important of these is the extreme ultraviolet (EUV) lithography machine. EUV lithography is the product of decades of research and development, the driving technology behind the past two generations of cutting-edge chips, and has been found in every high-end smartphone, tablet, laptop and server for the past three years. However, Moore's Law must continue to advance, and chipmakers continue to advance their roadmaps, which means further shrinking of device geometries is required.
The next generation of lithography technology, high numerical aperture EUV lithography, involves a major transformation of the internal optical system of the system. Before the end of this decade, the chip industry is expected to rely on it for continued progress.
Resolution and Numerical Aperture
The maintenance of Moore's Law depends on the improvement of the resolution of photolithography technology, which allows chip manufacturers to create increasingly fine circuits. Over the past 35 years, engineers have improved resolution by two factors by studying a combination of the wavelength of light, a factor k1 related to the packaging process, and the numerical aperture (NA), a measure of the angular range over which a system emits light. Magnitude.
The critical dimension is the smallest product size that can be printed using a certain lithography exposure tool and is proportional to the wavelength of light divided by the numerical aperture of the optical element. Therefore, we can achieve smaller critical dimensions by using shorter light wavelengths or larger numerical apertures, or a combination of both. For example, through improved manufacturing process controls, the k1 value can be as close as possible to its physical lower limit of 0.25.
Generally speaking, the most economical way to increase resolution is to increase the numerical aperture and improve tooling and process controls to achieve a smaller k1. Chip manufacturers will find ways to shorten the wavelength of the light source only when they cannot further increase the numerical aperture and k1.
However, the semiconductor industry has modified the wavelength many times. The history of wavelengths has evolved from 365 nanometers produced using mercury lamps, to 248 nanometers produced by krypton-fluorine lasers in the late 1990s, to 193 nanometers produced by argon-fluorine lasers in the early 2000s. With each wavelength generation, the numerical aperture of lithography systems gradually increases until the industry adopts shorter wavelengths.
For example, as the use of 193 nanometers comes to an end, a new method has been introduced to expand the numerical aperture: immersion lithography. By placing water between the bottom of the lens and the wafer, the numerical aperture can be significantly increased from 0.93 to 1.35. Since its introduction around 2006, 193nm immersion lithography has been the industry-leading workhorse lithography technology.
Improving numerical aperture is urgent
However, as demand for products printed below 30nm increases and the numerical aperture limits of 193nm lithography are reached, keeping up with Moore's Law becomes increasingly complex. To make products smaller than 30 nanometers, you either need to use multiple patterns to make single-layer chip products (a technology with high technical and economic costs), or you need to change the wavelength. After more than 20 years and unprecedented development research, the new wavelength of 13.5nm EUV has come online.
EUV requires a completely new way of emitting light. It's a very complex process that involves hitting droplets of molten tin in mid-air with a powerful carbon dioxide laser. The laser vaporizes the tin into a plasma, which emits a spectrum of photon energy. EUV optics take the desired 13.5-nanometer wavelength from this spectrum and guide it through a series of mirrors that reflect it onto a patterned mask, projecting the pattern onto the wafer. The entire process must be completed in an ultra-clean vacuum because the 13.5 nanometer wavelength is absorbed by air. (Previous generations of lithography directed light through a mask, projecting a pattern onto the wafer. But EUV is very easily absorbed, so the mask and other optical components must be reflective.)
The shift from 193nm light to EUV shrinks the critical size to some extent. An approach called "design for manufacturability" significantly reduces k1. This method sets the design rules for circuit modules and takes advantage of the limitations of photolithography technology. Now it's time to increase the numerical aperture again, from the current 0.33 to 0.55.
In a vacuum chamber, EUV light (violet) is reflected by multiple mirrors before reflecting off the photomask. From there the light continues to reflect until it is projected onto the wafer, carrying the pattern of the photomask. This illustration shows a current commercial system with a numerical aperture of 0.33. In future systems with a numerical aperture of 0.55, the optics will be different.
To increase the numerical aperture from 0.33 to the target value of 0.55 inevitably requires a series of other adjustments. Projection systems such as EUV lithography have numerical apertures on both the wafer and the mask. When you increase the numerical aperture on the wafer, the numerical aperture on the mask also increases. Therefore, the light cones entering and exiting the mask become larger and must be angled away from each other to avoid overlap. Overlapping light cones can produce asymmetric diffraction patterns, resulting in undesirable imaging effects.
But there are limits to this angle. Because the reflective mask required for EUV lithography is actually made of multiple layers of materials, we cannot ensure proper reflection above a certain reflection angle. The EUV mask has a maximum reflection angle of 11 degrees. Although there are other difficulties, reflection angle is the biggest challenge.
The only way to overcome this challenge is to add a feature called "zooming." As the name suggests, downscaling involves taking the reflective pattern out of the mask and shrinking it. In order to compensate for the reflection angle problem, the reduction factor had to be increased to 8. As a result, the portion of the mask that is imaged on the wafer is much smaller. A smaller imaging field means it takes longer to produce a complete chip pattern. In fact, this requirement would reduce the throughput rate of our high numerical aperture scanners to below 100 wafers per hour, a productivity level that would make chip manufacturing uneconomical.
Happily, the researchers found that they only needed to increase the reduction in one direction, which is the direction with the largest reflection angle, while the reduction in the other direction could remain unchanged. This results in an acceptable-sized imaging field on the wafer—about half the size used in today’s EUV systems, which is 26 mm by 16.5 mm instead of 26 mm by 33 mm. This direction-dependent or distorted reduction forms the basis of our high numerical aperture system.
To ensure the same productivity levels for half-size fields, the system's mask stage and wafer stage (the two stages that hold the mask and wafer respectively) also had to be redeveloped and moved synchronously during the scanning process. High numerical aperture EUV is a key component in maintaining Moore's Law, but achieving a numerical aperture of 0.55 is not the ultimate goal. The entire semiconductor ecosystem will move towards better, faster, and more novel technologies on this basis.
Review Editor: Huang Fei
#High #numerical #aperture #EUV #lithography #revolutionary #technology #leading #generation #chip #manufacturing
- Advantages and applications of Aigtek power amplifiers
- Microchip further expands its mSiC solutions with the introduction of 3.3 kV XIFM plug-and-play mSiC gate drivers,
- What is NAND type Flash memory?
- Data leaks can sink machine learning models
- What does the AC voltage regulator indicate when it is overvoltage?
- Adjustable voltage stabilized power supply circuit diagram based on LM317
- Demystifying Qualcomm Domain Controller Level 1 Power Design: Power Supply Design and Computation
- EU consumers challenge Meta paid service as privacy ‘smokescreen’
- Qualcomm Snapdragon
- Perform complex motor drive and interactive control testing with the MDA8108HD Motor Drive Analyzer