Fabricación de la “película mágica” que atrapa los fotones para obtener características del tamaño del ADN en el silicio

‘Behind the Builders:’ Senior principal engineer Anna Lio gives an inside look at the alchemy that turns ‘fuzzy images’ from EUV lithography machines into intricate chip designs.

author-image

Por

Anna Lio contends that you don’t just fire up a 200-ton, school bus-sized extreme ultraviolet (EUV) lithography machine and sit back and watch it print out chips. That would be like showing up for a mountain climb without a plan.

Rather, it takes a team of scientists and engineers to operate this amazing machine, not to mention create the chipmaking recipes it uses. Lio develops the ingredients (specialized chemicals called “photoresist”) used to imprint ever-more sophisticated and intricate integrated circuits onto silicon wafers — a critical step in maintaining Moore’s Law.

A senior principal engineer in Intel’s Technology Development group, Lio refers to photoresist as “magic film.” Lithography is essentially printing with light, and just as you need a special solution to expose a photograph onto paper, photoresists are baked onto wafers in layers that react when exposed to light patterns projected inside lithography machines.

Lio, who’s also an SPIE Fellow, lives in a world of mind-boggling precision. Photoresist layers are as thin as 30 nanometers — a sheet of paper is 100,000 nanometers thick — and as she’ll soon explain, help create electronic devices far smaller yet.

“There are a lot of similarities between what I do at work and what I do outside of work,” says Lio, who is also an enthusiast climber. “Not in the details, of course, but in the way my thought process works, my risk-taking, the way I analyze and dissect a problem. In mountaineering, when you go on a climb, there is a lot of preparation that is involved.”

Translating Fuzzy Images into Precise Chips

Lio has been working in lithography since she joined Intel out of graduate school in electrical engineering 25 years ago, and has spent the better part of the past 10 years focused on EUV. Using a dramatically shorter wavelength of light (13.5 nanometers) than previous-generation machines (193 nanometers), the massive tool produces smaller and denser microscopic features on chips (performed affordably, this is the essence of Moore’s Law).

Small is actually too big a word.

“The DNA molecule is about two nanometers in diameter,” Lio explains. With forthcoming “High NA” or high numerical aperture EUV machines — of which Intel plans to be the first into volume production — “we’re looking at high resolution on the order of a few nanometers … printing features in the size of three or four DNA molecules. That is impressive.”

Impressive, too, is the effort that went into making EUV machines a reality. Intel contributed significantly to the 30-year cross-industry global collaboration now realizing EUV — truly a milestone achievement in human technology. (Watch this recent “Behind this Door” video to wrap your head around the wonder of EUV.) Just the mirrors inside the machine are remarkable engineering marvels.

Photoresists require a similarly broad set of contributors, Lio says, starting with many different teams within Intel, new and established photoresist suppliers, academia, national labs and industry consortia.

When photoresists work correctly, “we can take this fuzzy image, even in the case of EUV, and translate it into a very sharp and detailed image.” Said image is then “etched” permanently into the surface, new materials are added on top, and the whole process repeats until a multidimensional chip is complete — later to be packaged for placement inside myriad possible devices from touchscreens to the cloud.

EUV Brings New Capabilities­ — and New Challenges

Developing photoresists is painstaking work. “We analyze hundreds and hundreds of resists in a single calendar year,” Lio says. Each one is coated onto a wafer, baked, sent into the lithography machine for “printing,” baked and developed again, and then analyzed and recorded.

“The big challenge is to balance all the requirements,” Lio notes. A good photoresist achieves the desired resolution — printing features and patterns in precise dimensions — perfectly and consistently across millions and millions of wafers.

EUV complicates things. Although the smaller wavelength means finer printing, it also sends a fainter signal — it’s like going from a thick black paintbrush to an extra-fine pen with grey ink. At a typical 20-millijoule-per-square-centimeter “dose” or shot of light, Lio explains, “in EUV we have 14 photons per nanometer square, while in 193 (immersion lithography) we get 14 times that number.”

“So, you’re now literally counting photons.” A bigger dose would send more photons per exposure, but that would slow the machine, decrease output and thus increase the cost of every wafer processed. Lio and her team focus on making the resists absorb and use as many photons as possible.

To accelerate development — Intel is working to deliver five process nodes in the next four years, including seamless integration of EUV technologies — the resist testing is augmented with simulation and statistical analysis.

“In EUV lithography, photons need to generate lower energy electrons before they can actually interact with the photoresist,” Lio explains. “The chemistry mechanisms are different versus 193 immersion, and also more complicated. So, we do a lot of work both internally and with national labs and academic partners to understand how the chemistry, interactions and these processes work at an extreme level of detail.”

Working on the Leading Edge, Playing on a Mountain Ledge

“From one technology to the next, some resists can be reused, and some cannot,” she says. “So, we’re in constant development mode to make more and more advanced materials. We need new advances, new inventions and new types of materials, because we are going to want to print smaller and smaller features.”

“It can be a very intense job,” Lio says. “What we do literally changes every day. I’ve been very fortunate to always work on the latest technology and the latest advances.”

“That engagement is part of what I like, and that constant risk and risk-taking and exploration, constant need to innovate,” she notes. “This is definitely one of the things that I love the most about the job.”

Lio asserts that her experience in climbing helps her handle the relentless pace ­— not only to disconnect and recharge but also to smartly manage risk.

“How do you assess risk?” Lio asks. “How do you decide that a certain level of risk is okay, and beyond that is too much? What can you do to lower the level of risk so that your confidence increases? In the mountains, it can literally be a matter of life and death. If you make the wrong decision, it could end badly.”

“I learned to really assess everything to make sure that I not only have fun and succeed in my climb, but of course that I return home safely.”

On to the next peak — mountains and that Moore’s Law curve.

EUV aporta nuevas capacidades y nuevos desafíos

Desarrollar fotorresistentes es un trabajo minucioso. "Analizamos cientos y cientos de resistencias en un solo año calendario", dice Lio. Cada uno se recubre en una oblea, se hornea, se envía a la máquina de litografía para su "impresión", se hornea y se desarrolla nuevamente, y luego se analiza y registra.

"El gran desafío es equilibrar todos los requisitos", señala Lio. Una buena fotorresistente logra la resolución deseada (imprimiendo características y patrones en dimensiones precisas) de manera perfecta y consistente en millones y millones de obleas.

EUV complica las cosas. Aunque la longitud de onda más pequeña significa una impresión más fina, también envía una señal más débil: es como pasar de un pincel negro grueso a un bolígrafo extrafino con tinta gris. A una típica "dosis" o toma de luz de 20 milijulios por centímetro cuadrado, explica Lio, "en EUV tenemos 14 fotones por nanómetro cuadrado, mientras que en 193 (litografía de inmersión) obtenemos 14 veces ese número".

"Entonces, ahora estás literalmente contando fotones". Una dosis mayor enviaría más fotones por exposición, pero eso ralentizaría la máquina, disminuiría la producción y, por lo tanto, aumentaría el costo de cada oblea procesada. Lio y su equipo se centran en hacer que las resistencias absorban y usen tantos fotones como sea posible.

Para acelerar el desarrollo, Intel está trabajando para entregar cinco nodos de proceso en los próximos cuatro años, incluida la integración perfecta de las tecnologías EUV, las pruebas de resistencia se aumentan con simulación y análisis estadístico.

"En la litografía EUV, los fotones necesitan generar electrones de menor energía antes de que puedan interactuar realmente con la fotorresistencia", explica Lio. "Los mecanismos químicos son diferentes frente a la inmersión 193, y también más complicados. Por lo tanto, hacemos mucho trabajo tanto internamente como con laboratorios nacionales y socios académicos para comprender cómo funcionan la química, las interacciones y estos procesos a un nivel extremo de detalle".

Trabajando en la vanguardia, jugando en una cornisa de montaña

"De una tecnología a otra, algunas resistencias se pueden reutilizar y otras no", dice. "Por lo tanto, estamos en modo de desarrollo constante para hacer materiales cada vez más avanzados. Necesitamos nuevos avances, nuevos inventos y nuevos tipos de materiales, porque vamos a querer imprimir características cada vez más pequeñas".

"Puede ser un trabajo muy intenso", dice Lio. "Lo que hacemos literalmente cambia todos los días. He sido muy afortunado de trabajar siempre con la última tecnología y los últimos avances".

"Ese compromiso es parte de lo que me gusta, y ese riesgo constante y la toma de riesgos y la exploración, la necesidad constante de innovar", señala. "Esta es definitivamente una de las cosas que más me gusta del trabajo".

Lio afirma que su experiencia en la escalada la ayuda a manejar el ritmo implacable, no solo para desconectarse y recargarse, sino también para administrar el riesgo de manera inteligente.

"¿Cómo se evalúa el riesgo?" Lio pregunta. "¿Cómo decides que un cierto nivel de riesgo está bien, y más allá de eso es demasiado? ¿Qué puede hacer para reducir el nivel de riesgo para que su confianza aumente? En las montañas, puede ser literalmente una cuestión de vida o muerte. Si tomas la decisión equivocada, podría terminar mal".

"Aprendí a evaluar realmente todo para asegurarme de que no solo me divierto y tengo éxito en mi escalada, sino que, por supuesto, regreso a casa a salvo".

Al siguiente pico: las montañas y esa curva de la Ley de Moore.