A method of multiplexing lasers to increase the available power is set to develop Extreme Ultraviolet Lithography (EUVL) source into a workable solution for high volume manufacture of semiconductor chips.
Researchers have for the first time used multiple lasers to generate a Laser Produced Plasma (LPP) EUV light source, a promising technology for producing semiconductor chips with 32nm or better accuracy.
Powerlase, manufacturers of powerful nanosecond Q-switched, diode-pumped solid state (DPSS) lasers, and the
Dr Samir Ellwi, Powerlase’s Vice President of Strategic Innovations, said, ‘We decided to build on this success with a second phase with the purpose of doubling the EUV power. We have more than doubled the output to 23W, which has the major benefit of increasing the power scalability of the LPP EUV approach.’
The conversion efficiency of the laser light into 13.5nm wavelength EUV light from both lasers is similar to when they are used independently of each other. This shows more lasers can be added to the process without lowering the conversion efficiency, an important element in making the LPP EUV source a viable solution for high volume manufacture.
Ellwi explained, ‘UCF has an EUV target that is run at a very high rep rate of 30kHz or higher and contains a small amount of tin. Our laser hits a few droplets from the available target, but not every single one. Potentially, if you aim another laser module at the target, you could hit another droplet.
‘What this means is you double the amount of power generated from the EUV source in a process called temporal multiplexing. Each laser hits a droplet independently, but by averaging what you get every second you generate the power required. Two laser modules hitting the droplets independently gives more laser power on the target, hence you get more EUV power.’
Currently, 193nm laser beams are used to generate chips at 65nm accuracy. EUVL increases this accuracy to 32nm and below, providing the semiconductor industry with the ability to print chips more accurately.
In addition to the key benefit of the scalability of this approach, having multiple lasers circumvents the maintainability issue. ‘With a large laser, if one single optic fails the whole machine would be down,’ said Ellwi. ‘Having a modular system, if one laser fails, another five or six lasers could still be running, decreasing down time for the industry.’
Ellwi said the study ran smoothly due to a lot of foregoing research. ‘We didn’t face any significant challenges using the two lasers as we have been studying this area for the last five or six years, preparing a lot of experimental and theoretical understanding. Because our lasers are so small, we didn’t have any problems putting them together, although with more units we might possibly face more of a challenge.’
In theory, there is no limit to the number of lasers that can be added to scale up to the required power. ‘By multiplexing and adding more lasers to satisfy the requirements set by the stepper (photolithography machine) manufacturers, if they say we need only 100W, we put in 10 lasers. If they want 200W, we put in more,’ said Ellwi. ‘But what we have to bear in mind is the cost; it would be prohibitively expensive to add a huge number of lasers.’
If phase 2 is successful, the researchers intend to initiate another phase to satisfy the stepper manufacturers’ beta tool requirements of providing 50W at the intermediate focus.
‘We believe very strongly that we are capable of producing this power,’ said Ellwi.