Semiconductor companies, today, are in a difficult position as after years of regular innovation with silicon, which allowed them to generate consistent profits and performance improvements over many years, they are now struggling to generate more value out of silicon.
The question for many has been what material could replace silicon and graphene has been touted as having the potential to equal or surpass silicon’s performance.
In a nutshell, the concept of deploying materials as near perfect sheets just one atom thick, which have totally predictable electrical characteristics, offers a completely new way of constructing the semiconductor devices we’re all dependent on.
In recent years potential performance improvements in silicon have slowed and as described by one analyst, silicon is now becoming mortal.
In response, the semiconductor industry has been experimenting with a number of exotic new materials, such as silicene, germanene, and black phosphorous, but graphene is seen as having the greatest potential. However, until this point graphene has only ever been grown in small quantities in laboratories by the scientific community and graphene’s commercialisation could be anything up to 25 years away and will require substantial investment in both R&D and capital costs to bring it into production.
Without developing industrial processes, tools and equipment, there’s a risk graphene will never deliver its potential and will forever be caught in a catch 22 of hyperbole.
What do Intel, TSMC and other major chip manufacturers want from graphene? Transistors based on 2D materials are firmly on the International Technology Roadmap for Semiconductors (ITRS). As a result, these main manufacturers are extremely interested in finding solutions around how to manufacture 2D materials, either in-house or from the ecosystem for which the industry is famous.
Adopting new materials in the semiconductor industry is a hugely disruptive process and some of the physics involved is yet to be fully understood. However, because the potential is so huge the industry will make it its business to find the right solutions.
Right now, 2D materials are not ready to be introduced to mainstream production but the teams who are responsible for planning investment and infrastructure – particularly at the world’s semiconductor leaders like Intel and TSMC – are very keen to understand the manufacturing implications. And these same companies are developing relationships with the material and equipment suppliers they’ll need to partner with in order to compete in the future.
Rather than rushing to publish yet another research paper on the amazing potential of graphene, the industry is now shifting focus on developing the tools and processes necessary to manufacture high quality graphene at an industrial scale. So that it can be adopted widely by the semiconductor industry.
What does that look like? There are some nuts to crack to industrialise graphene and integrate it into mainstream manufacturing at industrial scale: material growth, material transfer, and integration.
If the industry can meet these requirements, 2D material manufacturing will evolve to play a critical part of the global wafer and FAB equipment market, which is currently expanding at a massive rate, +30% to $86bn this year according to SEMI.
Current suppliers grow graphene using chemical vapor deposition (CVD), a widely used process in the semiconductor industry. It involves exposing a heated substrate to carbon-containing gases in a vacuum. As the gas settles on the hot substrate surface, carbon grows into the famous honeycombed network pattern of graphene.
Very similar procedures are used to grow other 2D materials, such as hexagonal boron nitride. The critical aspect of the process is rigorously controlling multiple aspects of the CVD process to grow high-quality wafers of these materials to attach to the 200mm to 300mm diameter wafers required by semiconductor manufacturing processes.
Once the 2D material has been grown and inspected, it needs to be separated from its substrate and moved onwards in the manufacturing cycle. Remember this is a sheet of material one atom thick which needs to be structurally perfect and free of impurities to perform. The task of transferring a 200 or 300mm wafer of graphene is analogous to, grasping something the size of a football pitch but as thin as cling-film, picking that up whole and depositing it on another football pitch without breaking it or contaminating it.
ANL has integrated the end-to-end production of 2D material manufacture. Each step entails specially engineered equipment and processes. The degree of integration needed to produce and integrate graphene and other 2D materials requires extensively customised (for 2D) production tools that handle industry standard 200mm and 300mm wafers. Also, the many subtle interactions of high temperature growth and room temperature dry transfer are exploited in ANL automated systems and process control to ensure highly consistent material results.
Lastly, having developed the processes to transfer the 2D materials as entire sheets to another wafer or substrate, ANL is already preparing an Application Development kit to enable customers to shorten their path to new product development.
What now for graphene?
With the long-term IRDS as a guide and several discussions with leading semiconductor players we continue to focus on ensuring that all the stepping-stones are in place to assure the industry has the tools and processes needed to meet the 2D transition to HVM.
The more immediate opportunities for graphene to prove itself as a commercial technology are in Back End of Line (BEOL) – typically the last processes are added to the semiconductor wafer relating to special functions like sensing, photonic switching, or advanced interconnect where multiple devices are stacked or linked together.
It is BEOL where customers are already being engaged with examples including photonics and sensors on top of active silicon dies.
Sensors represent another key area for development because the technology is experiencing amazing growth driven by autonomous systems and especially biosensing for point of care diagnosis.
By their nature, sensors tend to be produced in foundries which have already transitioned to added BEOL technologies such as MEMS and have processes that easily accommodate planarized 2D layers.
In a related more current application in leading edge semiconductor manufacture, graphene is being investigated as material to be used for an EUV mask protection pellicle within EUV machines that image advanced semiconductors during manufacture. A pellicle is a transparent protective screen that protects the expensive photo mask from particulate contamination in the extremely hostile environment inside the EUV machine, extending the mask’s operating life.
Today ANL is collaborating with key players in the ecosystem and supplying engineering volumes of graphene to companies developing prototype devices.
It continues working with them very closely so that when they are ready to go to production, the industrial growth, transfer and integration propositions will all scale with them.
Over the next 10 to 25 years, graphene could end up replacing silicon as the primary material in semiconductors and could be used in applications where its technical merits (such as high speed, low-loss requirements, small scale, and flexibility) are better suited for electronic applications than alternative materials.
Market analysis has suggested that the total addressable market for graphene could be worth in excess of $190 billion across data processing, wireless communications, and consumer electronics, so whatever the challenges associated with its commercialisation it will prove well worth the challenge.
Author details: Paul Hedges, CEO of Applied Nanolayers