New blend of materials in nanowires results in better transistors
2021-09-14
The semiconductor industry is driven forward by improvements in digital processors, of which the transistor has been the fundamental component. Recently, the field has been broadened to include components that are specially designed for efficiency in areas such as artificial intelligence, 5G networks and the internet of things. Adam Jönsson’s research is linked to this development. The III-V material variants, i.e. a combination of elements from the third and fifth columns of the periodic table, have better electron transport properties than silicon, the industry standard.
On Friday 17 September at 9:00, Adam Jönsson will present his doctoral thesis entitled “Vertical Heterostructure III-V MOSFETs for CMOS RF and Memory Applications” at LTH.
Adam Jönsson’s research is based on Lund University’s long experience with nanowires made of what is known as III-V material, i.e. material created by blending elements from column three of the periodic table with elements from column five.
“We try to use both new materials and new physical design of transistors and adapt them for example to digital computations, analogue high frequency amplifiers and memory matrices”, says Adam Jönsson.
Commercial applications of III-V material include radio-based communication with high frequencies such as mmwave (>60 GHz).
“We want to combine high frequency amplifiers and digital logic. We are trying to create a platform to produce complementary circuits, i.e. circuits that are optimised to transport both electrons and electron holes – and to do this completely using III-V material. This makes it easy to integrate digital logic with the best high frequency amplifiers for fast and smart signal processing.”
Another area in which Adam has been working is the development of a method to produce a nanowire-based selector, which can now be combined with what is known as a memristor – a component the research team had already been working on.
“This is a new component. By combining a classic transistor with a memory cell known as a memristor, we expect to be able to reduce energy consumption for machine learning, for example. And thanks to the new 3D design, we can also fit a very large number of memory cells on a small chip surface.”
High frequency amplifiers for 5G communication always have problems with resistance, above all unwanted capacitance. In order to overcome these problems, Adam developed various types of spacers that separate the close contact surfaces of the transistors.
Transistor production. (a) Nanowire after growth in MOCVD chamber. (b) Applied top metal, and etched channel. (b) Final transistor with three electrodes: Source, Gate and Drain.
Why are you doing a PhD – and why Lund?
¬“I looked for local opportunities in Lund and started to study engineering physics. Over the years, I got increasingly motivated by the subject and being offered a doctoral studentship felt like proof that I had actually performed well. When I met Lars-Erik [Wernersson] and heard about how successful his former students were, it became even more obvious that I should pursue a doctoral degree. The mixture of lab work and theoretical study was also an attraction.”
Competing against giants
Adam explains that it is becoming increasingly expensive to develop and produce electronic components such as transistors on the nano scale. This in turn has entailed the consolidation of a large part of commercial production to a few stakeholders operating on a global level.
“So there are only a few producers in the world, such as Samsung, Intel, GlobalFoundries, that produce pretty much all components. And in that context, I am proud that we, in Lund, can compete against these giants and actually beat them when it comes to certain performance indicators.”
What will happen to your research findings?
“I like to say that research is successful if its findings are used. I have tried to write my doctoral thesis in such a way as to inspire other doctoral students in similar subjects to continue the work in the same spirit. When it comes to commercialisation, the hurdle is a little higher. But it is possible that the technology could end up in processors or high frequency amplifiers within ten years or so. For memory circuits, a lot of commercialisation is already underway in the form of Static Random-Access-Memory (SRAM).”
---
