Massachusetts University team achieves new breakthrough in photolithography chip

Recently, a research team from the University of Massachusetts Amherst has pioneered a new technology that uses laser irradiation on concentric superlenses on chips to generate holograms, thereby achieving precise alignment of 3D semiconductor chips.

This research result, published in the journal Nature Communications, is expected to not only reduce the production cost of 2D semiconductor chips, but also promote the feasibility of 3D photonics and electronic chips, and lay the foundation for the development of other low-cost, compact sensor technologies.

Semiconductor chips are the core of electronic devices for processing, storing, and receiving information, and their functions are determined by the specific component layout embedded in the chip. However, as the potential of 2D design approaches its limit, 3D integration technology is seen as a key path to breaking through bottlenecks.

To build a 3D chip, multiple 2D chips need to be stacked and each layer needs to be accurately aligned in the three-dimensional direction (i.e. front back, left and right, and vertical gaps corresponding to the x, y, and z axes), with an error controlled within tens of nanometers (1 millimeter equals 1 million nanometers).

Amir Arbabi, Associate Professor of Electronic and Computer Engineering at the University of Massachusetts Amherst and senior author of the paper, explained that traditional alignment methods involve observing the markings on each layer (such as corners or crosshairs) under a microscope and attempting to overlap, but this method is not suitable for 3D chip manufacturing. The first author of the paper and doctoral student Maryam Ghahremani pointed out that microscopes cannot clearly focus on two layers of markers at the same time because the interlayer gap can reach hundreds of micrometers, which may cause chip movement and misalignment during the refocusing process.

The research team simulated and measured different lateral misalignment scenarios ranging from 150 nanometers to 1 micrometer (1000 nanometers). Maryam Ghahremani also mentioned that the resolution of microscopes is limited by the diffraction limit, which is about 200 nanometers.

The new calibration technology of Amir Arbabi team does not require moving parts and can detect misalignment between two layers in a smaller range. They expected an accuracy of 100 nanometers, but in actual testing, the measurement error along the x and y axes was as low as 0.017 nanometers, and the error on the z axis was only 0.134 nanometers, far exceeding expectations. Amir Arbabi stated that this technology can detect movements equivalent to the size of an atom by observing the changes in light passing through an object. The naked eye can recognize errors of a few nanometers, while computers can read even smaller errors.

To achieve this goal, the research team embedded alignment marks composed of concentric metal lenses on semiconductor chips. When the laser passes through these marks, two interference holograms will be formed. Kahramani pointed out that these interferometric images can intuitively display whether the chip is aligned, as well as the direction and degree of misalignment.

For semiconductor tool manufacturers, chip alignment is a daunting and costly challenge, "Amir Arbabi said." Our approach solves one of these challenges. "In addition, this technology reduces costs and provides more opportunities for small startups seeking semiconductor innovation.

Amir Arbabi also mentioned that this technology can be applied to manufacture displacement sensors for measuring physical quantities such as displacement. He said, "Many physical quantities to be detected can be converted into displacement with just a laser and camera." For example, pressure sensors can be made by measuring the motion of a membrane, and any phenomenon involving motion, such as vibration, heat, acceleration, etc., can theoretically be tracked through this technology.

Summary
In current chip manufacturing, the alignment of multi-layer patterns relies on microscopic imaging technology, but due to the large distance, traditional methods are difficult to achieve the required nanometer level accuracy. The research team at the University of Massachusetts Amherst has successfully achieved sub nanometer precision alignment by introducing metasurface alignment markers, combined with lasers and cameras, greatly improving accuracy and ease of operation.

The experimental results show that this method can achieve lateral accuracy of one fifty thousandth of the laser wavelength and axial accuracy of one sixty three hundredth of the laser wavelength. This technology is expected to drive the production of a new generation of 3D integrated optical and electronic chips, paving the way for applications such as high-precision displacement sensors.

Source: OFweek