Servizio Comunitario di Informazione in materia di Ricerca e Sviluppo - CORDIS

CARL resist process

The majority of chip manufacturing companies and resist vendors started research in the field of 157nm photoresists in 1999 and declared this to be the main lithography technology for the 65nm node. Due to the lack of transparency of the current 248nm and 193nm polymer generations at 157nm new polymer platforms had to be developed in order to follow the requirements of the Semiconductor Industry Association’s Technology Roadmap (ITRS). Common single layer resists (SLR) require maximum absorbances below 1/micron in order to generate relatively thick (>150nm) and etch resistant films to be used in 157nm lithography. For this, much effort was spent to synthesize new fluoropolymers with enhanced transmission.

The CARL-principle (Chemical Amplification of Resist Lines) enables reduced film thicknesses by means of a bilayer approach and offers the advantages of biasing trenches and holes in an additional wet chemical biasing step. Thus, smaller critical dimensions (CDs), i.e. widened lines as well as shrinked contact holes with reduced line edge roughness (LER) can be generated if cross linking silylating agents are covalently incorporated into the reactive polymer films after the wet development. Simultaneously, an increase of film thickness can be generated by this method enhancing the etch resistance of the generated profiles. Unfortunately the previous polymers designed for 248nm or 193nm lithography for the CARL process show absorbances at 157nm that are too high (8.5/micron), even when used for TFI processes.

During the phase 1 of the UV2Litho project, Infineon started model polymer screening and developed several new synthetic approaches towards more transparent CARL model polymers by the use of fluorinated comonomers. A novel polymer platform has been synthesized by means of radical copolymerisation, comprising a 157nm absorbance of 3.3/mircon.

CARL Wet Silylation Experiments
Feasibility tests have been carried out by measuring vertical film thickness increase of non-patterned resist layers. Almost 50nm (ca. 50%) increase could be found only after 15 seconds. Further experiments are needed on structured wafers to optimise the lateral pattern shrink.

Exposure Results:
Only a small number of exposure experiments have been performed with the new polymer platform due to limited access to early 157nm exposure tools. Resolution of a resist formulated from CG1 is limited by adhesion problems of the polymer on bare Silicon. As the 157nm Microstepper at International SEMATECH is a non-clustered tool, a large amount of quencher had to be used in order to prevent from T-topping caused by airborne contamination.

Adhesion of the polymer can be improved by the use of antireflective coatings (ARC). Several 193nm exposures have been performed on the AT/1100 at IMEC. A maximum resolution of 90nm dense Lines/Spaces could be obtained. All exposures were done with a binary reticle. The current 157nm resist had to be adjusted to 193nm with regard to sensitivity and film thickness. Still, contamination and line edge roughness (LER) remain an issue.

Process Development:
Since no usable 157nm exposure capability for process development had been available, CARL process optimisation was carried out on standard DUV (non-fluorinated) polymer platforms with a high technological maturity. Prior to investigations of gate patterns, the work was focused on shrinking contact holes, the greatest challenge of future litho nodes.

The key steps of the CARL shrink concept are described here: The thin top resist layer of the bilayer resists system is exposed and developed, followed by a wet silylation process at room temperature resulting in a typical pattern shrink (so called “chemical biasing”). The generated structures are transferred via oxygen reactive ion etch (RIE) into the bottom resist, serving as etch mask. A lateral bias of 50nm could be achieved from shrinking 200nm CH patterns via the CARL process at the 193nm node.

Special attention was spent on shrink investigations of unsymmetrical patterns such as T-shape or L-bar structures as well as rectangular contacts. When using half tone phase shifting masks (HTPSM), side lobe printing, especially at defocus, becomes an issue. Side lobe defects after development may be cured (“healing”) during the silylation process. Depending very much on silylation conditions and the photoresist chemistry those defects can also result in severe bridging defects, so called “peanutting” originating from the swelling of the polymer matrix during the silylation reaction. This would require very sensitive balancing of all process and materials parameters and possibly would become another road blocker at future litho nodes, which would have to be thoroughly investigated.

Due to the high chemical complexity and the outlined process limits of the CARL shrink technology Infineon decided not to use the process for DRAM manufacturing.

More information on the UV2LITHO project can be found at:

Reported by

Infineon Technologies AG
Paul-Gossen-Str. 100
91052 Erlangen
See on map