A team at the Rensselaer Polytechnic Institute in New York came up with the adhesive by heat-treating glue material that was already commercially available. In doing so, it created a super-strong and sticky substance.
The basic technology is known as nanolayers, which are molecular chains with a carbon backbone that end with a certain element, such as silicon, oxygen or sulfur. These nanoylayers can improve adhesion while also preventing materials from mixing together.
The problem with nanolayers, however, has been they lose their effectiveness above 400 degrees Celsius. They degrade or completely detach from a surface.
The researchers, led by materials science and engineering professor Ganapathiraman Ramanath, had for about eight years investing ways to improve the structural integrity of semiconductor devices in computer chips.
They sandwiched a nanolayer between a thin film of copper and silica, thinking the extra support would help strengthen the nanolayer’s bonds and make it stickier. But they were pleasantly surprised to learn that the middle layer of the nano-sandwich did not deteriorate in any way – simply because it had nowhere to go. What’s more, the nanolayer’s bonds grew stronger and more adhesive between 400 degrees Celsius and 700 degrees Celsius. Sandwiched in between the copper and silica, the nanolayer’s molecules hooked onto an adjoining surface with unexpectedly strong chemical bonds.
Within a computer integrated circuit, copper wires are used to transport electrons within an insulating material, usually silica. While copper is an excellent diffuser of electrons, it does not stick well to dielectric material. The problem chipmakers now face, as devices and their respective chip geometries shrink, is that electrons tend to skip from one copper wire to the next, because they are so close together, causing current leakage. In turn, the computer suffers from reduced reliability and performance.
To get around this, the chip industry currently uses tantalum nitride or other compounds to stick the copper to the silicon. But these materials tend to mix together, so an additional layer is used to bind them. Problem is, these additional layers take up valuable real estate on chips that are becoming increasingly smaller.
Chipmakers are pushing the boundaries of how small these layers can be. Already they are almost inconceivably small – about 10 nanometers or 15 nanometers. (The dot in this letter i measures about 1 million nanometers wide.) Going any thinner means they don’t coat the copper deep or uniformly enough.
They don’t work with shrinking dimensions, Ramanath said. There will come a point where they will say, ‘Look we need to try something completely different.’
Ramanath is hoping they will try nanoglue, which he said could be used between the copper and the silicon. Just a single layer of nanoglue would form a strong bond and also prevent the two materials from mixing. You are killing two birds with one stone, he said.
And the nanoglue is cheap: 100 grams cost about $35, Ramanath said.
But what may be most attractive to the chip industry is nanoglue’s ability to self assemble as either a solution or a gas. In other words, if the molecules were put in a wet solution or vapor, they would organize themselves, like rank-and-file soldiers, into place. This means they would provide a continuous film on the silicon substrate, which would improve the reliability and performance of the chip.
This gives chipmakers greater flexibility, particularly with next-generation manufacturing methods, Ramanath said. Most materials don’t offer you this kind of choice, he said. The nanoglue also could be used in existing CMOS chip manufacturing, he said.
Having self-assembling properties also means the nanoglue might be used in next-generation chip materials, such as nanotubes or nanowires, which are both contenders to replace copper during the next 10 to 15 years. There, the problems are going to be more severe, Ramanath said. The smaller structures, more closely packed together… means a higher need for structural integrity.
Indeed, Ramanath said the concept of sandwiching molecules and enabling bonds to form is generic and could be used on a variety of materials, although his team has not yet tested additional material.
Just how the material would be commercialized is not yet known, he said. Ramanath himself has considered setting up a business. Several parties have also contacted him enquiring about the technology. Rensselaer Polytechnic also has an IP-commercialization center on campus.
Intel and IBM are both very interested and excited in the material, Ramanath said. But being excited is one thing. Whether they implement it or not, there’s a big gap in between that, he said.
