Behind the buzz and beyond the hype:
Our Nanowerk-exclusive feature articles
Posted: Aug 5th, 2008
Fullerene resist materials for the 32nm node and beyond
(Nanowerk Spotlight) The success of the semiconductor industry has been due in large part to its ability to continuously increase the complexity, and therefore the processing power, of integrated circuits at a given manufacturing cost. Moore’s Law observes that the number of transistors in a computer chip doubles every two years, whilst the cost of making the chip remains the same, due to miniaturization of the components.
In order to produce the next generation of computer chips it is necessary to continue to shrink
the size of the components on the chip. The miniaturization upon which Moore’s Law rests has been achieved through advances in
the photolithographic process used to pattern the components onto to the silicon wafer. A
beam of light is projected through a shadow-casting reticule and the light pattern is then
directed onto a silicon wafer coated with a photochemically sensitive material, known as a
resist. The solubility of the resist is modified by exposure to the light, allowing specific areas
of the resist film to be removed, whilst other areas remain as a mask, so that the silicon wafer
can be selectively etched, metallized or doped.
For many years it has been predicted that the end of photolithography is approaching, and
that further miniaturization will require next generation lithography techniques, such as EUV lithography. However, photolithography has proved remarkably resilient, and continues to
improve. Unfortunately, whilst the ability of photolithography to pattern small features
continues to improve, the industry is beginning to challenge the capabilities of the
The properties of the resist start to limit further miniaturization
There are several properties of resists that are important for commercial application. The
three primary characteristics are the resolution, the line width roughness (LWR) and the sensitivity.
The resolution dictates how small a feature you can pattern; the line width roughness is a
measure of the deviation of the pattern from the ideal, and thus limits the yield of the devices
you are trying to create by affecting their performance characteristics; and the sensitivity
determines how quickly you can make the pattern (which in turn relates to the cost of making
For commercially successful patterning it is necessary that you can pattern a small enough
feature accurately and quickly. In the past the photolithography machine has limited this, but it
is now becoming increasingly difficult to extend the resist. For a given resist material the
resolution, the line width roughness and the sensitivity are interrelated, so that beyond a
certain point it is only possible to increase the resolution at a cost to the sensitivity or the line
width roughness or vice versa. This is called the “RLS trade-off” (or the Lithographer’s
Uncertainty Principle). For a given resist:
Resolution × Line Width Roughness × Sensitivity ≥ Constant
It has become increasingly apparent that current commercial resist systems based on
chemically amplified polymers are limited to a resolution of around 25 to 30 nm for useful
values of line width roughness and sensitivity. This means that current resist materials cannot simultaneously meet the sensitivity, resolution and line width roughness requirements set out by the International Technology Roadmap for Semiconductors (ITRS) for the 32nm node and beyond.
To go to better resolutions with current resists it is necessary to sacrifice the sensitivity, or to suffer greater line width roughness, or both.
Typically photoresists are based on a polymeric material, due to the ease with which polymer
films can be produced on semiconductor wafers by spin coating. However, polymers are
generally very large molecules. The typical polymer in a resist may take up a volume of 10
nm diameter or more. This limits the size of the smallest feature that the resist can record.
Therefore in order to produce resists that can support the requirements of future lithography it
has become necessary to move away from the typical polymeric resists.
The move away from polymeric resists
Whilst the smallest feature size is also impacted by the type of resist you're a using, in simplified terms you can say
that the size of a polymer molecule sets the size of the 'pixel' you can use and thus limits the
smallest feature you could write. Molecular resists typically use molecules with a diameter up
to ten times smaller than a polymer – 1-2 nm versus 10 nm – which allows a smaller pixel size to write with. The
problem with non-polymeric materials is that after spin coating they tend to crystallize as they
dry, which roughens the film. Fullerene derivatives are one of a small number of nonpolymeric
materials that can form smooth spin-coated films.
Lines with a width of 15 nm defined in the resist. (Image: Dr. Robinson, Nanoscale Physics Research Laboratory)
"We have previously demonstrated that fullerene based molecular resists are capable of very
high-resolution lithographic patterning with electron beam lithography," (Ultrathin Fullerene Films as High-Resolution Molecular Resists for Low-Voltage Electron-Beam Lithography) Dr Alex P G Robinson, a Senior Scientist at the Nanoscale Physics Research Laboratory (NPRL) at the University of Birmingham in the UK, tells Nanowerk. "Furthermore, by virtue of the high carbon content of the fullerene materials these resists are extremely resistant to the plasma etching procedure used to transfer resist patterns into the semiconductor, which allows us to use very thin films of the resist. When patterning feature sizes less than 100 nm in width, the surface tension of the solvent used to develop the resist pattern can cause closely spaced features to collapse if the film thickness is more than three times the feature size, and hence the ability to use thinner films is beneficial."
Fullerene-based molecular resists
Robinson explains that, while it has become apparent that molecular resists look like a promising way to shrink the constant in the RLS trade-off and thus push lithographic capabilities further, the main disadvantage of molecular resists such as the fullerene resists the NPRL team has previously demonstrated has been their sensitivity. Whilst these materials can record very dense patterns it has typically required a large dose of light or electrons to expose them. That makes
these materials too slow for commercial use, where economic viability requires the patterning of a certain number of wafers per hour.
"In our new paper about (Fullerene Resist Materials for the 32nm Node and Beyond) we address that point" he says. "While we previously have demonstrated very high resolution with poor sensitivity in pure fullerene based resists, or alternatively good sensitivity but mediocre resolution in chemically amplified fullerene resists, this is the first time that we have managed to combine both properties into a single material."
"What we were able to demonstrate" he continues, "is that it is possible to apply the well-known photoresist sensitivity enhancement technique of chemical amplification to high resolution fullerene based molecular resists — without a degradation of the existing resolution and line width roughness, and whilst retaining a useful etch resistance."
In chemical amplification a sensitizing ingredient called a photoacid generator is added to the resist. When the resist is exposed to radiation
the photoacid generator releases an acid. This acid then reacts catalytically with the resist molecules to change their solubility. In this way one exposure event can expose several resist molecules – increasing the sensitivity.
The particular advance described in the new paper, a collaboration of Robinson, Francis P. Gibbons and Prof. Richard E Palmer from NPRL, and Dr Sara Diegoli, Dr Mayandithevar Manickam and Prof. Jon A Preece from the University of Birmingham's School of Chemistry, is that they have shown for the first time a resist that combines the high resolution of pure fullerene materials and the sensitivity of the chemically amplified resists.
"We believe that this material has the best combination of resolution, line width roughness and sensitivity currently available" says Robinson. "It also retains good etch durability – comparable with high durability commercial resists – and shows LWR values approaching those specified by the ITRS."
Robinson points out that, to date, their materials have mainly been characterized using electron beam lithography. "We are currently investigating the response of this materials using extreme ultraviolet lithography, and are also keen to investigate the use of our materials in current photolithography exposure tools. Additionally we are looking into ways in which we can replace the halogenated solvents currently used for our molecular resists, with more environmental friendly solvents."
For the field in general the main focus is on finding ways to minimize the RLS trade-off and therefore provide resists to meet the needs of next generation lithography. There has been a significant increase in interest in molecular resists to meet this challenge.