Achieving a semiconductor accuracy of 32 nanometre

Paul Boughton
Dr Samir Ellwi discusses advances in extreme ultra violet lithography.

Although many may never see one, semiconductors are incredibly common and important components of modern electronic devices. The average person will most likely carry several with them everyday; in a laptop, mobile phone or MP3 music player.

As such technologies continue to become more sophisticated, faster and more efficient semiconductors need to be developed.

Semiconductor development requires the doubling of accuracy and efficiency of components once every two years.

Moore's Law, initially coined by Intel founder Gordon E Moore and affirmed by the International Semiconductor Roadmap, has set a target for semiconductors to reach an accuracy of 32nanometre(nm) by 2012.

Immersed in high wavelengths

Currently, semiconductors are manufactured using microlithography printing. A lithography 'stepper' machine is used to focus a bright light source onto a wafer (a very thin sheet of material usually made of silicon) through a patterned mask.

The light source passes through the mask and a series of optical lenses within the stepper before reaching the wafer. This prints the mask image on the wafer. Successful application of the process requires an extremely bright light source.

Manufacturers employing this technique currently use excimer lasers with a wavelength of 193nm a light source.

While 193nm has been sufficient to keep pace with the increasing demand for more efficient components, continuing industry advancement means it will soon reach the end of its technological life.

Work has been completed on adapting the technology to Immersion Lithography, which replaces the air between the optical lenses with a liquid to focus light more effectively, but this has only seen success in creating chips as low as 45nm. Significant improvements to printing accuracy require a much lower wavelength source than 193nm.

Ultra violet lithography

Extreme ultra violet lithography (EUVL) is considered to be the most likely next-generation lithography (NGL) technology to supersede 193nm microlithography.

Currently in development, the technology uses a much lower wavelength light at 13.5nm. Successful application of this will result in a great advance in printing accuracy.

Structure sizes should reach 32nm by 2012 (meeting Moore's requirement), with the capacity to refine the process further to as low as 22nm or even 16nm.

The most significant challenge faced in EUVL development is successful creation of the 13.5nm light.

Unlike 193nm Immersion, a laser cannot directly generate light at the required wavelength.

Two methods of creating an EUV source are currently being researched; laser produced plasma (LPP) and discharge produced plasma (DPP).

LPP is created by heating a very small droplet with a powerful laser pulse. Heating the droplet creates an environment know as the plasma medium, which emits light at various wavelengths including 13.5nm. Lasers initially used to create LPP were low powered units.

The application of diode-pumped, solid-state (DPSS) lasers with high power and a high repartition rate provide a high average power at the EUV, greatly improving throughput potential.

13.5nm source

By contrast, DPP has a similar capacity in creating a 13.5nm source, but achieves this thorough an entirely different method. With DPP the EUV target, which can be a droplet or ablated material, is 'pinched' in an electronic field created between two electrodes (a cathode and an anode).

When targeted with a laser beam, a discharge is triggered, creating a high temperature plasma medium that emits light at the critical 13.5nm.

With either technique, the amount of 13.5nm generated by the plasma medium accounts for just 2.5 per cent of light generated.

The limited amount emitted means capturing as much light as possible within the stepper machine is vital.

Here, LPP takes the lead over DPP. The LPP approach requires the droplet target to be suspended in a vacuum when creating the plasma medium, the light generated is emitted into the open space within the vacuum chamber.

As there is no physical obstruction to absorb or block the light, the maximum amount of light can be captured for printing.

With DPP, the electrodes used to generate the electronic field act as obstacles in light collection; meaning a smaller amount 13.5nm light can be exploited. This essentially lowers the brightness of the source, reducing throughput potential.

Multiplexing for more power

The University of Central Florida (UCF) and Powerlase have been working on the development of an efficient method of producing an EUV source using the LPP method.

Since the work began, the partnership has seen significant breakthroughs in the successful application of LPP. UCF began using a one kilowatt laser from Powerlase to irradiate tin-droped micro-droplet EUV targets. Initially results demonstrated a high conversion of laser pulse into useful 13.5nm light (the EUV conversion efficiency).

It became clear successful application of the LPP approach relied on increasing the average power delivered to the EUV target to 100 watts. This would allow a greater amount of light to be generated, without damaging the stepper machine.

To achieve this, researchers worked on multiplexing a second laser with the original.

When using a single laser, only a single beam of laser pulse energy can irradiate the EUV targets. By multiplexing two lasers together, a process known as Temporal Multiplexing, one laser pulse will target a single tin-droplet, while the second pulse targets a second droplet.

The amount of energy aimed at the droplets can be better regulated, maximising the amount of laser light converted into 13.5nm.

Multiplexing two lasers doubled the average power at the target to 23watts, compared with 10 achieved with a single laser.

When using the DPP technique there is no way of reducing the power to the electrodes without lowering the amount of 13.5nm source generated, and thus adversely effecting throughput potential. With LPP there is no limit, in theory, to the amount of lasers that can be added to generate the source. The only real limit is the stepper manufacturer's budget.

Multiplexing lasers

The capacity to multiplex lasers together through the LPP approach increases the flexibility of the lithography process.

A hypothetical system targeting six droplets with six separate lasers would still be running at 85 per cent efficiently if one of the lasers were to fail and need adjustment or replacing.

The ability to minimise complete stepper downtime in this manner is a great benefit to semiconductor manufacturers, as even a short period of downtime results in a profit loss.

Summary

Although both laser produced plasma and discharge produced plasma to extreme ultra violet lithography represent great advances in semiconductor manufacturing technology, the LPP approach has distinct advantages for high-volume manufacturers.

Near unlimited sociability, greater flexibility and a high throughput potential means the technology will meet the 32nm requirement while also allowing chip manufacturers to keep their businesses in profit.

The result will be the capacity to create semiconductors at a much greater accuracy than the 45nm currently in production.

- Dr Samir Ellwi is Vice President of Strategic Technology, Powerlase, Crawley, West Sussex, UK. www.powerlase.com