Computers will act and perform a lot differently by 2032. Those differences will revolutionize the efficiency, size, and speed of these devices. Getting there is a long road.
Deep Jariwala, Ph.D., is an Assistant Professor of Electrical and Systems Engineering at the University of Pennsylvania’s Device Research and Engineering Laboratory. The lab focuses on the study, design, and development of nanometer and atomic-scale devices, materials, and interfaces for applications in computing, sensing, information technology, and renewable energy.
His role has evolved over the years.
“I'm a materials engineer by training and then I have slowly ventured into applied physics and then electrical engineering and device research,” he said. “I'm materials agnostic. So, I like to investigate any and all interesting nanoscale materials that appear on the research horizon.”
Much of these nanoscale materials have features embedded within them or are themselves of dimensions that are just one to a few hundred nanometers. So, investigating the local, atomic structure, electronic structure, optical properties, and vibrational structure, are quite interesting to him. And these materials are important for our devices as well.
It is fairly well known at this point that atomic force microscopy with tip-enhanced Raman scattering (AFM Raman/TERS) is the easiest and most versatile technique that allows us to visualize nanoscale structures in a few minutes.
It's even more powerful than electron microscopy in some respect, especially if you're trying to image something on a surface. AFM Raman/TERS has also evolved to a point that it can help visualize local optical properties. We tend to use a combination of all these techniques to integrate all kinds of nanomaterials, starting from quantum dots, which are polarly synthesized to carbon nanotubes to two-dimensional materials (2D Materials).
“One area of research where we use AFM Raman/TERS heavily is the investigation of 2D Materials and its heterojunctions within them,” he said. “One of the reasons we want to investigate that is 2D Materials are extremely promising. They are poised to be the semiconductors of the future that drive our microelectronics and various other things.”
But one of the challenges with these materials is that, because it is so thin, it has many inhomogeneities in it from time to time. There could be defects in certain regions, which are oxidized or come from other defects. Looking at those regions, which are nanoscale in their dimensions is quite important. Defects in particular have different optical properties, and also have different vibrational properties. And so, if you were to use TERS or tip enhance photoluminescence (PL) to probe these materials, it becomes visible.
“You would be able to see how spatially localized or delocalized the defects are. What is the density of defects? You can also do other interesting things like you can make interfaces of these semiconductors with metal, and probe how good these interfaces are. As many of us know in modern semiconductor microelectronics, the interface between the metal and the semiconductor is one of the hardest to optimize and is the source of most losses. This interface has yet not been optimized for two-dimensional semiconductors. And so, by using tip-enhanced Raman scattering, we can look at how good these interfaces are that you're forming and what are the ways you can understand the defects or other deformations to make these interfaces better.”
How does knowing these defects advance the materials characterization?
“Because we are dealing with semiconductor materials, it is important that semiconductors are produced at the highest quality and purity. That’s because they need to perform at a very high level when they are inserted into devices. So, quantifying the defect structure both visually as well as in terms of numbers is quite important. When you are making devices out of these semiconductors or materials, or when you are growing them and then evaluating them after growth, having this kind of characterization tool lets you rapidly evaluate at the nanoscale what the defect types, properties, and densities are. And tip-enhanced Raman scattering and AFM Raman are some of the most powerful techniques in that regard.”
How does this bridge technology into applied engineering?
There are a few ways. Semiconductors made from 2D Materials are being looked at by various industries, including the biggest semiconductor companies as a potential replacement for Silicon. If we want to replace Silicon with some of these emerging materials, we want to understand everything about it and need standard characterization routes for it too.
“This is where these semiconductors are extremely powerful, and I don't think they will just impact the semiconductor industry or the mainstream semiconductor electronics industry. The two-dimensional material’s optical properties are so fascinating that will also impact electronics industries, such as, for example, displays or LEDs, lasers and optical modulators, telecom, and things like that. So, in all of these places understanding the material, property, understanding its interfaces with other materials is very, very important. And this is where these characterization techniques are paramount in understanding the fundamental nature of these materials and its structure.”
Prof. Deep Jariwala with a HORIBA LabRAM HR Evolution Raman microscope with TERS
What beneficial optical and electrical properties do 2D Materials possess?
“The optical properties are mainly their ability to luminesce. When you shine light on these semiconductors, they absorb some of the light and then fluoresce or send out light again. This can also be done in a different way, where if you send in electrons and emit light, this is the principle of LEDs as we know it. Or light-emitting diodes. And so many of these 2D Materials that we’re investigating, emit light very efficiently, which means that you could use them for a display or lasers. The other interesting thing about them is that because they are so thin, it is also easy to modulate their optical properties, which means you can change the amount of light going through them as a function of some electric field that you can apply to them.”
This has potentially transformative applications for optical modulation, which is the basis of light-based communication, which is also the basis of our internet and all our communication technology. Those are the interesting optical properties on the electronic property side. The ability to modulate current flow in these semiconductors as a function of electric field allows you to do electrical switching. And that is the basis of all of computing.
“These are the two central properties, on the optical side, on the electronic side that makes these semiconductors so exciting for the future,” Jariwala said.
With all this innovation, where are we going to be in 10 years, 2032, when it comes to 2D materials?
“I would say that I was not very optimistic about (2D Materials) just a couple of years ago, because of the pandemic effects, but I was also slowly believing that people will keep making Silicon just better and better. I didn’t think these materials would ever have the potential to replace Silicon.”
“But what has happened in the past two to three years is people have really figured out ways to grow these materials over large areas or wafers with very high quality. And at the same time, people like myself and others have also figured out a way to make devices out of them that are really good, even beating some of the Silicon devices.”
Now many companies, including some of the biggest semiconductor chipmakers are seriously thinking about 2D Materials as a next-generation semiconductor, either for doing the computational logic or for storing data, such as memory applications. He predicts it might happen in about eight to 10 years from now.
“There are research pipelines or development programs, and we are starting several pilot projects. If successful, these companies will launch, research projects worth multi-hundreds of millions of dollars to drive these materials and devices towards commercial technology.”
So, things are looking more and more promising now, and Jariwala has been interacting closely with some of the biggest semiconductor groups in this regard.
What does that mean? He predicts that by 2032, we can expect to see the beginnings of nanotechnology computing. Although, one could say that computing already reached the nanoscale years before 2010, with all of the transistors nanoscopic in dimensions.
“I think the key question to ask is when would bottom-up grown nanomaterials actually enter computing hardware? And I think the answer to that is yes, I think in 10 years, in one or the other layer in the microprocessor, some of these bottom-up grown nanoscale materials would have been added as functional units in computing hardware.”
That would add essential properties to the microprocessor or storage medium – efficiency and speed.
“You know, it's a trillion-dollar industry, and when things are at those scales, profits and certainly the turnover in the business matters the most,” he said. “What is driving our computing industry to make better and better chips now is the energy efficiency in computing and the speed. These are the two most important things.”
“Energy efficiency has become a huge problem because we are making so many tiny devices, all crammed onto a single chip. So, energy efficiency is the key thing, and once energy efficiency kicks in, to some extent, speed also is sort of related. Efficiency helps out with some of the speed too since it reduces heating up of the components.’
The push behind this research, besides energy efficiency and speed, is size, and the drive to make smaller devices.
2D Materials can be so well tuned with an electrostatic field that we can keep shrinking the dimensions while not driving up energy consumption.
With Silicon, as manufacturers are designing devices shorter and shorter, it is also leading to more steady-state energy consumption, which means the processors heat up a lot more. In other words, a metric that is important is how many numbers of computations you can do per Joule of energy or per watt of power.
“This number has flattened out with Silicon, and we somehow need to get this number up again. And that is only possible if at an individual device level, you drive down the power consumed per computation, and that is possible when you shrink these devices further and make them even thinner, which is what you naturally get from two-dimensional semiconductors.”
Research occupies almost 70 percent of Jariwala’s time at the University of Pennsylvania. He uses a HORIBA LabRAM HR Evolution with an attached TERS microscope, along with a variety of lasers to hit the TERS system to investigate his samples.
“We have a 633-nanometer laser and a 785-nanometer laser, both of which can do TERS. We are hoping to add one more laser to the system. It's been working very nicely for us, and we have been publishing some really good papers using the system. So, we are very happy about it overall.”
These tools, along with the ingenuity to apply the results into tangible technologies will undoubtedly contribute to the next generation of computing.
Do you have any questions or requests? Use this form to contact our specialists.
