INR: A negative tone I-line chemically amplified photoresist

INR, an I-line negative photoresist, is described. Acid catalyzed cross-linking of phenolic resins using a nonmetallic photoacid generator, 2,6-bishydroxymethyl-p-cresol as a cross-linker, and 9-anthracene methanol as an I-line sensitizer results in a very high photospeed aqueous TMAII developable photoresist. Poly(p-hydroxystyrcne) was found to have advantages over novolac resins for formualtion of a high performance negative I-line photoresist. Advantages obtained by using PHS rather than novolac include higher thermal stability, elimination ofundercut on nuleophilic surfaces and compatibility with 2.38% TMAH puddle develop processes A high resultion version, INR-X, is described. Resolution to O.3Oum and linearity to O.35um was obtained using a O.54NA ASML I-line stepper. O.35um line-spaces array's had 1 .2um depth of focus and O.4Oum line-space array's had a depth of focus greater than I .6um. An unusual characteristic found in INR-X is a veiy low sensitivity to variation in PEB temperature. A 3nm/°C line-width dependency was found. 1.0 INTRODUCTION 1.1 BACKGROUND Negative tone photoresists were once common in IBM's semiconductor manufacturing fabs. These resists were typically based on cyclyied rubber resins and used organic solvents as developers. Major concerns with negative resists of this type include photospeed sensitivity to the presence of oxygen and pattern distortion caused by swelling of resist images during develop' . Additionally, manufacturing concerns over the use of organic solvents as developers made most production facilities replace these negative resists with non-swelling aqueous base developable positive tone resists. One approach for developing a non-swelling negative photoresist used cross-linking of phenolic resins with photosensitive bis-azidc cross-linkers2. Undercut profiles were typically obtained due to poor develop latitude and high absorbance. A negative tone deep tJV chemically amplified resist was developed by IBM and implemented on IBM Burlington's 1 Mb l)RAM line3. The resist was based on the now well known t-butoxy carbonyl ( t-BOC ) protected poly(phydroxystyrene) which was developed by Ito, Wilson and Frechet4 . This approach depended on polarity reversal of a hydrophobic resin. On exposure and baking the t-BOC group is cleaved converting the resin from hydrophobic to hydrophilic. A non-polar solvent was then used, typically anisole, to develop a negative tone pattern. While this system did not suffer from many of the problems associated with earlier negative resists, it still had the handicap of an organic solvent O-8194-1490-5/94/$6.oo SPIE Vol. 2195 /307 Downloaded From: http://ebooks.spiedigitallibrary.org/ on 09/07/2016 Terms of Use: http://spiedigitallibrary.org/ss/termsofuse.aspx development process. Feely's development of cross-linking aqueous developable negative tone resists solved many of the problems associated with negative tone resists5. By using acid catalyzed cross-linking ofphenolic resins a non-swelling negative resist was developed which incorporated chemical amplifiction. Thackeiy et al. have based Shipley's ANR series of resists on this chemistry8. 1NR is an I-line Negative Resist developed for use by IBM's semiconductor production facilities. INR is a chemically amplifed resist which uses a cross-linking chemistiy developed jointly by Frechet et al. and IBM. 1NR-31 10 is a high volume product which has been qualified on IBM Burlington's 4M DRAM line and is being qualified on IBM Sindlefigen's 4M DRAM line. Manufacturing experiences have allowed INR's performance and chemistry to be refmed which has led to improved versions called INR-3210 and INR-X. These new versions are now undergoing evaluation in IBM's manufacturing facilities. 1.2 INR APPLICATIONS 1NR's current main application is as an ion-implant mask on non-critical levels. The rational for using 1NR in place of conventional positive tone resists as an ion implant mask involves apply processes, resist thermal properties, and ionimplant performance. Using 1NR negates the need for top edge bead removal ( TBR ) during apply. TBR has historcally been a source for defects. Improved thermal flow property of 1NR eliminated UV hardening prior to ion-implant reducing cycle time and tooling costs. Lower out gasing during ion-implant has shortened pump down times reducing cycle time. Additionally, photospeed ofINR is very high relative to positive resists due to chemical amplification. This has significantly improved throughput for block levels increasing stepper capacity. These results and observations are summarized by Puttlitz etal.t1. 1NR is also being developed for critical level lithography. The primary application would be for gate level lithography where post-etch nested v/s isolated pattern biasing is a concemt2t3. Positive resists typically print isolated features smaller than nested features of the same dimension. During etch, RIE bias causes the nested v/s isolated bias to increase resulting in significant line-width variation. l3Onm post etch bias' have been found for 500nm stmctures using a positive resist process. Negative resists typically print isolated features slightly larger than nested features and as a result the photobias subtracts from the etch bias . This effect should significantly reduce isolated v/s nested linewidth variation after etch. 1.3 ff4R CHEMISTRY Figure 1 shows INR-31 10's and INR-3210's composition. 2,6-bis(hydroxymethyl)-p-cresol is used as an acid activated cross-linking agent. A sulfonate ester of an n-hydroxy compound is used as a photoacid generator. 9-anthracene methanol is used an I-line sensitizer14. INR-3110 uses a m/p-cresol novolac resin and INR-3210 uses poly(phydroxystyrene). CH2OH iioij R—NO—R Sensitizer: 9-anthracene methanol Photoacd generator