Dutch language English language

Samples - The power of one hundred electron beams (Terrific Technology 2008)


Electron beams are widely used in lithography – which is a preparatory step in the production of microchips. However, electron beams have a very low throughput. Scientists at the Department of Imaging Science and Technology (IST) are designing a very clever system that solves this problem.

Nienke Beintema

Most of us are familiar with the highly magnified, black-and-white images of objects like an insect’s eye or a human hair. These amazingly sharp and detailed images do not stem from traditional light microscopy, but rather from scanning electron microscopy – a technology that has been commercially applied since 1965. Inan electron microscope, a high-energy electron beam is focused by electrostatic or magnetic lenses, and then projected and scanned over a sample to generate a highly magnified image.

The advantage of electron microscopy in comparison to light microscopy is that its resolution is unlimited – at least for practical purposes. The resolution of a start-of-the-art electron microscope is around 1 nanometer. Focused electron beams are also used for high-resolution patterning, or electron beam lithography. This type of lithography is usually referred to as ‘maskless’ lithography because it doesn’t require a mask to generate the final pattern. Instead, the final pattern is created from a digital representation on a computer, and the beam is switched on and off as it scans across the resist-coated substrate to write the pattern.

Large burden

“The problem with this technology,” says Yanxia Zhang, PhD student at IST, “is that the ‘serial’ nature of electron beam lithography makes the pattern generation very slow compared to a ‘parallel’ technique such as photolithography.” In photolithography the final pattern is produced at once by projecting the photomask - a media that stores the pattern information - onto the resist-coated substrate. This technology has countless potential applications. Electron beam lithography, however, due to its low throughput, is primarily used for manufacturing the photomasks needed in photolithography.

Nowadays, a set of masks costs several million dollars. It constitutes a large burden in terms of cost and turnaround time between design and market. Zhang and her colleagues are therefore working on developing a lithography system that has a high throughput, and eliminates the need for a photomask. Rather than just one beam, they use multiple beams at the same time. Their prototype is a 100-beam source: an array of 10 by 10 beams. Eventually they aim to employ as many as 13,000 beams. “If you have multiple beams writing simultaneously, the throughput of electron beam lithography will be greatly increased,” says Zhang.

The idea behind the multi-beam technology is surprisingly simple. After leaving the electron emitter, the electron beam encounters multiple aperture lenses, rather than just one. This array of apertures splits the electron beam into multiple beamlets. The beamlets then pass through an electrostatic lens array which focuses the multiple beamlets onto a third array, which is called a blanker array. “This blanker array is used to switch individual beamlets on and off,” explains Zhang. “This is done by changing the voltage applied onto two separate plates. The electrostatic field between the two plates will influence the course of the beamlet, which determines whether or not the beamlet will be able to pass through its respective hole in the array. Our aim is to be able to create beamlets with identical properties, in terms of beam size and beam current, and to control each beamlet individually.”

Why hasn’t this idea been applied many years ago? “The theory may sound simple, but making such a set-up is harder than you think,” says Zhang. The first challenge is to accept as much of the emitted current into the sub-beamlets as possible. It is the strength of the current that determines the system throughput. “Usually the diameter of the beam can only be around 10% of the lens diameter,” she says, “otherwise the image will be too blurred. In other words, you can only admit a limited amount of electrons into the beam.” When aberrations – image defects – are small, however, the amount of electrons can be strongly increased. Aperture lenses – the lenses used by Zhang – have superior properties in terms of aberrations. However, their application in traditional set-ups is impaired due to the fact that other optical components are influenced by the uniform electric field. The secret of Zhang’s system design is that the uniform field for the aperture lens is confined between two macro-electrodes, which prevents it from reaching the otheroptical components. By utilizing this type of aperture lenses, the filling of the lenses can increase from 10% up to 90%. What’s more, aperture lenses can be more closely packed than the conventional 3-electrode lenses. The total transmitted current, therefore, is around 15 times larger. Another advantage of the aperture lens is that it comprises only one electrode and doesn’t require alignment.

Entirely new technologies

Obtaining the same beam size for all the beamlets is essential for a faithful pattern transfer from the digital representation to the resist-coated substrate. For outer beamlets, however, the image-forming rays are not parallel to the optical axis. This inclination gives rise to additional image defects – off-axial aberrations. In addition, the amplitude and shape of these defects are different for each beamlet, depending on its position relative to the optical axis.

Zhang reduces the off-axial aberrations by changing the aperture design (see illustration). “Ideally, I want to design the apertures in a way that each ray, regardless of the angle in which it falls upon the aperture lense, only passes through its centre. This will reduce the aberrations to a minimum.”

By fine-tuning the voltage and shape of the macro-electrodes, Zhang has ultimate control over the electric field in front of the aperture lenses. This even allows her to ensure that all beamlets are focused onto one plane, rather than onto a curved surface – an off-axial aberration named field curvature. As she underlines, this was a breakthrough in her work. “No-one has ever been able to apply this field curvature correction to a multi-beam source,” she says proudly. “Compared to what others are doing, our set-up is simpler yet it has much better results.”

So far Zhang’s work has been merely theoretical. She still has to verify her findings experimentally. The necessary elements of her experimental set-up have been fabricated and she is now aligning the optical column. The next challenge will be to assemble the system as a whole and test its performance.        

Zhang is very optimistic about the impact that her technology will have on electron beam lithography. “Tens of thousands of beams writing at the same time,” she states, “will not only significantly improve the speed of current production technologies, but it will also open up possibilities of entirely new technologies, for instance in microchip fabrication, that we simply cannot envision today.”

Another application that would benefit from this technology is electron beam-induced deposition (EBID), a method to deposit materials directly onto a substrate without an intermediate material. Multi-beam technology, as Zhang suggests, could reduce the time to deposit a single dot. “It would be a great opportunity to merge these technologies,” she says.

After completing her PhD project, Zhang will start working at MAPPER Lithography – a spin-off company of TU Delft – where she will further develop her multi-beam design. “In theory it is working perfectly well,” she concludes. “There are some challenges ahead, including on the microfabrication side, but I am confident about the design itself. There is really no reason why it wouldn’t work.”

< Back to the samples
Copyright © 2010 - All rights reserved - Webdesign Laurens Mast Freelance Webdesign