- Email: [email protected]
Status and future of DUV photoresists for the semiconductor industry
MICROELECTRONIC ENGINEERING
ELSEVIER
Microelectronic Engineering 41/42 (1998) 37-40
Statm and Future of DUV Photoresim for the Semiconductor Indmtry
J. W, Thackeray Shipley Co., 455 Forest St, Marlboro, MA 01752 e-mail: [email protected] This paper reviews the development of positive DUV resists at IBM and Shipley. The review begins with APEX-E the prototypical chemically amplified positive DUV resist used throughout the semiconduc~tor industry. The paper then follows the emergence of new resist technology to overcome the weaknesses of first generation DUV resists like APEX-E, These improvements include feature-specific resists for low k~ processing, improved delay stabih'ty, in~oved PEB sensitivity, better etch resistance, and reduced substrate dependence. A new series of annealing type resists, UVS, UV6, and Titan have demonstrated all of tbese improvements.
I. Introduction DUV photoresi.sts, are now rapidly being implemented into the manufacturing lines of advanced logic and memory device manafacturers. The implementation of these materials required tremendous technical achievements and rapid cycles of learning by resist vendors and the resist users. The current stablehorse of the DUV resist is APEX-E, which was developed at IBM in the early 90's [1], This material is based on t-betoxycarbonato-blocked polyhydroxystyrene. APEX-E has lmJven to be the material of choice for the advanced microprocessor industly. High resolution and smooth sidewalls down to 200rim isolated lines has been demoustrate~ The one manufacturing problem that APEX had was its intrinsic sensitivity to airborne base contamination. However, a series of innovative manufacturing approaches has overcome this problem [2]. More advanced resists have since been developed to overcome postexposure delay stability. The UVIIHS, UVIII photot~sists have incorporated Ito's annealing concept into the resist design strategy [3,4]. Alternatively, low activation energy resists have been developed which use acetal/ketal deblocking schemes to reduce airborne contamination effects [5]. For these systems, another delay problem is introduced which is linewidth slimming [6]. Still more advances in resist development are require~ Substrate footing or undercut remain as technical hurdles. Developing resist materials 0167-9317/98/$19.00 © Elsevier Science B.V. All rights reserved. PII: $0167-9317(98)00009-4
with good metal etch selectivity is also required. Feature~c resists for low k~ processing are also being rapidly designed and optimized. This paper will review some of the latest Shipley/iBM resists designed for these requirements. 2. Advanced DUV Resist S y ~ m s
A~ t-butyl ester-blocked annealing systems. Ito's design of a polymer system, poly (phydroxystyrene co-t-butyl acrylate), cleverly incorporates the requisite thermal stability of the t-butyl ester [210°C] with a lower glass transitiontemperature [153°C]. In this way, the firm can be annealed without destroying the blocking group. Figure 1 illustrates the improved delay stability of UVIIHS and other annealing resists over that of APEX-E, the first generation DUV resist. 7
'i
i
4 2
1 0
.~ .IdPl~-II ~
Call IJ~
dl UV6
Figure 1. Post-exposure delay stability for DUV resists tested in the Shipley cleanroom (est. 10 ppb NI-I3). The figure represents the order of magnitude improvement in delay stability due to the anneafing concept.
38
J. W. Thackeray/Microelectronic Engineering 41/42 (1998) 37-40
Another improvement realized in the UVIIHS, UV5 and UV6 resists was a dramatic reduction in the CD sensitivity to PEB temperature. Figure 2 illustrates the dramatic improvements over APEX-E. Even though the activation energy for deprotection is slightly higher for a t-butyl ester V t-boc blocking group, the PEB sensitivity is substantially reduced [7]. The reduction in PEB sensitivity is due to the higher temperature processing above the annealing temperature of the resist. More advanced resists like UV5, Titan, and UV6 operate somewhere in between the low activation energy resist systems, where shelflife and linewidth slimming are problems, and high activation energy resist systems, where PEB sensitivity is a problem.
!
i
|
a
Figure 2. PEB sensitivity of IBM/Shipley developed resists. The figure illustrates the steady improvement in PEB sensitivity since APEXoK Another improvement is realized in the design of Titan, a resist for metal levels, with no footing on TiN. Figure 3 illustrates the removal of footing with Titan vs UVIIHS, and APEX-E resists. A proprietary concept was used to eliminate footing. The figure shows that APEXE has difficulty resolving 250nm dense lines vs. the other resists. UVIIHS resolves 250nm lines easily with severe footin~ Titan shows good resolution, with no footing, and the post-etched images show more resist retention than UVIIHS. Note also that Titan resolves the isolated lines better than UVIIHS.
, UVIIHS Ire-etch
(F.) UVIIHS Post-etch Figure 3. Metal Pre- and Post-etch results for Titan, APEX- E, and UVIIHS resists. Titan resist shows minimal footing at 250nm with good etch transfer.
J W. Thackeray/Microelectronic Engineering 41/42 (1998) 37-40
APEX-E is an exceptional resist for gate level processing. UVIIHS and UVIII are exceptional resists for contact hole processing. Our goal was to design advanced resists so that they could improve dense line performance and also do isolated lines and contact holes. We collected dissolution rate vs exposure curves for APEX-E and UVIIHS and compared them to our next generation resists, UV5 and UV6. This data is compiled in Table 1. In this table, we are trying to establish correlations between bulk dissolution parameters and lithographic performance. All the lithographic data and dissolution rate data were run on 900 angstroms of Shipley BARL. UV3, UV5, and UV6 all outperform APEX-E for dense lines. We attribute this improved performance due to the larger developer selectivity, n, for these resists over APEX-E. This information tells us that
R,,~A/s) (A/s) n
2S0nm 1:1 l/s DOF 250rim Iso line DOF 250rim 1:1 I/s EL 25Onto Iso line EL 1:1 I/s ML (urn) iso line iso-dense bias 250rim C/H EL 25Onto Cfll DOF
652 1.11 4.2 0.6 0.6 7% 20% 250 170 4nm NR NR
39
resists for 180rim must have high n. UV3 has excellent performance for contacts due to its high developer selectivity and high R , ~ However, the price you pay for high I ~ is reduced lithographic capability for the isolated lines. With UV5, we can outperform APEX-E for isolated lines and still have reasonable dense line and contact hole performance. This is accomplished by dialing in a lower 1 ~ value to a more reasonable 2620 A/s. Lrv6 improves the dense line and contact hole performance over UV5 by increasing Rm~ to 3840 A./S. The Titan resisthas similar dissolutionproperties to UV6. As DUV resists are applied to lower k~ processing~ more extensive fine tuning of the dissolution properties of the resists will be required to optimize their performance.
28300 0.79 7,9 0.8 0.6 16% 14% 220 220 -25nm 16% 1.0
2620 1.92 7.7 1.1 1.1 18.2% 20% 220 150 -20nm 7°/o 0.6
3840 0.49 6.8 1.4 0.8 21.5% 18% 220 170 0rim 14.5% 0.8
J. W. Thackeray/ Microelectronic Engineering 41/42 (1998) 37-40
40
Next-generation DIYV resists must also have improved etch and shrinleage properties over and UVIIHS. Through reduction in the Olmishi lmameter for DUV polymers, we have reduced the etch rate of UV5 and UV6 resists [8,9]. Figure 4 illustrates the reduced etch rate of these polymers.
Figure 4. Etch rate in a CI2/HCI metal etch for recently developed DUV resists. The decreased etch rate of UV5 and UV6 photoresists is due to the reduced Olmishi parameter for these resists. 3. ~ r e
R~
l~ign
DUV resists will continne to evolve over the next few years. Now that DUV is more af~pted by manufacturing lines throughout the world, more emphasis will be placed on simple, high throughlxa processing that can be easily integrated into the modern lab. Issues such as shelflife, ~ and Pl/B sensitivity, substrate sensitivity, and environmental/health effects of DUV resists will become more imnoRant. Ultimately, larger process window margins will be necessary to apply DUV resists at lower design rules. Specialized resists for resolution enhancement teelmiques must be designed (PSM resists, OAI resists, resists designed to reduce proximi~ corrections, etc.). Negative resists may be necessary for specialized features. In order to achieve larger process windows, thinner resist thicknesses will be nm on antireflective layers. Future DUV resists will have higher etch selectivity, especially for metal, and they will have high transparency on antireflective substrates, customized dissolution properties for
the resist feature of intere~ and reduced diffusion. It is an exciting time to be designing DUV resists for use in the semiconductor indnsUy. 4.
Refereaces
I. H. Ito, C. G. Willson, Polym. Eng. Sci., 23, 1012 (1983). 2. J. Vigil, M. Barrick, Proc. SPIE, 2438, pp 626-643 (1994). 3. H. Ito, G. Breyta, D. Hofer, 1~ Sooriyakumamn, K. Petrillo, D. Seeger, "Environmentally Stable Chemical Amplification Positive Resist: Principle, Chemistry, Contamination Resistance and Lithographic Feasibility," J. Photopolym. Sci. Technol. 7, 433448 (1994). 4. W. Conley, G. Breyta, W. Brunsvold, R. Dipietro, D. Hofer, S. Holmes, H. Ito, R. Nunes, G. Fichtl, P. Hagerty, J. Thackeray, "The Lithographic Performance of an Environmentally Stable Chemically Amplified Photoresist," Pro¢, SPlE, 2724, pp. 34-60 (1996). 5. H.-T. Schacht, N. Mnenzel, P. Falcigno, H. Holzw~h, J. Schneider, "Acid-Labile Crosslinked Units: A New Concept for Deep-UV Photoresists," J. Photepolym. Sci. and Technol., 9(4), 573 (1996). 6. N. Munzel. H. Holzwarth, P. Falcigno, H.-T. Sehaeht, R. Sehulz, O. Nalanmsu, A.G. Timko, E. Rcichmanis, J. Kometani, D. R. Stone, T. X. Neenan, E. A. Chandross, S. G. Slater, M. D. Frey, A. Blakeney, " Advanced Positive DUV Photoresists for Practical Deep-UV Lithography," Prec. SPIE, 2195, 47 (1995). 7. G. Wallm~ J. Opitz, W. Hinsberg F. Houle, J. rhac~ray, T. Fedynrsh~ D. r, ane, M. Rajaratnam, "Reactivity and Kinetic Parameters of UVIIHS," Proc. SPIE, 3049, Iyp. 492-500 (1997). 8. R. R. Kunz, S. C. Palmeteer, A. 1~ Forte, R. D. Allen, G.M. Wallmff, R. A. Depietro, D. C. Hofer, " Limits to Etch Resistance for 193nm single layer Resists, " Prec. SPIE, 2724, pp. 365-376 (1996). 9. H. Gokan, S. Esho, Y. Ohnishi, J. Electrnehem. So¢., 130, 143 (1983).
ELSEVIER
Microelectronic Engineering 41/42 (1998) 37-40
Statm and Future of DUV Photoresim for the Semiconductor Indmtry
J. W, Thackeray Shipley Co., 455 Forest St, Marlboro, MA 01752 e-mail: [email protected] This paper reviews the development of positive DUV resists at IBM and Shipley. The review begins with APEX-E the prototypical chemically amplified positive DUV resist used throughout the semiconduc~tor industry. The paper then follows the emergence of new resist technology to overcome the weaknesses of first generation DUV resists like APEX-E, These improvements include feature-specific resists for low k~ processing, improved delay stabih'ty, in~oved PEB sensitivity, better etch resistance, and reduced substrate dependence. A new series of annealing type resists, UVS, UV6, and Titan have demonstrated all of tbese improvements.
I. Introduction DUV photoresi.sts, are now rapidly being implemented into the manufacturing lines of advanced logic and memory device manafacturers. The implementation of these materials required tremendous technical achievements and rapid cycles of learning by resist vendors and the resist users. The current stablehorse of the DUV resist is APEX-E, which was developed at IBM in the early 90's [1], This material is based on t-betoxycarbonato-blocked polyhydroxystyrene. APEX-E has lmJven to be the material of choice for the advanced microprocessor industly. High resolution and smooth sidewalls down to 200rim isolated lines has been demoustrate~ The one manufacturing problem that APEX had was its intrinsic sensitivity to airborne base contamination. However, a series of innovative manufacturing approaches has overcome this problem [2]. More advanced resists have since been developed to overcome postexposure delay stability. The UVIIHS, UVIII photot~sists have incorporated Ito's annealing concept into the resist design strategy [3,4]. Alternatively, low activation energy resists have been developed which use acetal/ketal deblocking schemes to reduce airborne contamination effects [5]. For these systems, another delay problem is introduced which is linewidth slimming [6]. Still more advances in resist development are require~ Substrate footing or undercut remain as technical hurdles. Developing resist materials 0167-9317/98/$19.00 © Elsevier Science B.V. All rights reserved. PII: $0167-9317(98)00009-4
with good metal etch selectivity is also required. Feature~c resists for low k~ processing are also being rapidly designed and optimized. This paper will review some of the latest Shipley/iBM resists designed for these requirements. 2. Advanced DUV Resist S y ~ m s
A~ t-butyl ester-blocked annealing systems. Ito's design of a polymer system, poly (phydroxystyrene co-t-butyl acrylate), cleverly incorporates the requisite thermal stability of the t-butyl ester [210°C] with a lower glass transitiontemperature [153°C]. In this way, the firm can be annealed without destroying the blocking group. Figure 1 illustrates the improved delay stability of UVIIHS and other annealing resists over that of APEX-E, the first generation DUV resist. 7
'i
i
4 2
1 0
.~ .IdPl~-II ~
Call IJ~
dl UV6
Figure 1. Post-exposure delay stability for DUV resists tested in the Shipley cleanroom (est. 10 ppb NI-I3). The figure represents the order of magnitude improvement in delay stability due to the anneafing concept.
38
J. W. Thackeray/Microelectronic Engineering 41/42 (1998) 37-40
Another improvement realized in the UVIIHS, UV5 and UV6 resists was a dramatic reduction in the CD sensitivity to PEB temperature. Figure 2 illustrates the dramatic improvements over APEX-E. Even though the activation energy for deprotection is slightly higher for a t-butyl ester V t-boc blocking group, the PEB sensitivity is substantially reduced [7]. The reduction in PEB sensitivity is due to the higher temperature processing above the annealing temperature of the resist. More advanced resists like UV5, Titan, and UV6 operate somewhere in between the low activation energy resist systems, where shelflife and linewidth slimming are problems, and high activation energy resist systems, where PEB sensitivity is a problem.
!
i
|
a
Figure 2. PEB sensitivity of IBM/Shipley developed resists. The figure illustrates the steady improvement in PEB sensitivity since APEXoK Another improvement is realized in the design of Titan, a resist for metal levels, with no footing on TiN. Figure 3 illustrates the removal of footing with Titan vs UVIIHS, and APEX-E resists. A proprietary concept was used to eliminate footing. The figure shows that APEXE has difficulty resolving 250nm dense lines vs. the other resists. UVIIHS resolves 250nm lines easily with severe footin~ Titan shows good resolution, with no footing, and the post-etched images show more resist retention than UVIIHS. Note also that Titan resolves the isolated lines better than UVIIHS.
, UVIIHS Ire-etch
(F.) UVIIHS Post-etch Figure 3. Metal Pre- and Post-etch results for Titan, APEX- E, and UVIIHS resists. Titan resist shows minimal footing at 250nm with good etch transfer.
J W. Thackeray/Microelectronic Engineering 41/42 (1998) 37-40
APEX-E is an exceptional resist for gate level processing. UVIIHS and UVIII are exceptional resists for contact hole processing. Our goal was to design advanced resists so that they could improve dense line performance and also do isolated lines and contact holes. We collected dissolution rate vs exposure curves for APEX-E and UVIIHS and compared them to our next generation resists, UV5 and UV6. This data is compiled in Table 1. In this table, we are trying to establish correlations between bulk dissolution parameters and lithographic performance. All the lithographic data and dissolution rate data were run on 900 angstroms of Shipley BARL. UV3, UV5, and UV6 all outperform APEX-E for dense lines. We attribute this improved performance due to the larger developer selectivity, n, for these resists over APEX-E. This information tells us that
R,,~A/s) (A/s) n
2S0nm 1:1 l/s DOF 250rim Iso line DOF 250rim 1:1 I/s EL 25Onto Iso line EL 1:1 I/s ML (urn) iso line iso-dense bias 250rim C/H EL 25Onto Cfll DOF
652 1.11 4.2 0.6 0.6 7% 20% 250 170 4nm NR NR
39
resists for 180rim must have high n. UV3 has excellent performance for contacts due to its high developer selectivity and high R , ~ However, the price you pay for high I ~ is reduced lithographic capability for the isolated lines. With UV5, we can outperform APEX-E for isolated lines and still have reasonable dense line and contact hole performance. This is accomplished by dialing in a lower 1 ~ value to a more reasonable 2620 A/s. Lrv6 improves the dense line and contact hole performance over UV5 by increasing Rm~ to 3840 A./S. The Titan resisthas similar dissolutionproperties to UV6. As DUV resists are applied to lower k~ processing~ more extensive fine tuning of the dissolution properties of the resists will be required to optimize their performance.
28300 0.79 7,9 0.8 0.6 16% 14% 220 220 -25nm 16% 1.0
2620 1.92 7.7 1.1 1.1 18.2% 20% 220 150 -20nm 7°/o 0.6
3840 0.49 6.8 1.4 0.8 21.5% 18% 220 170 0rim 14.5% 0.8
J. W. Thackeray/ Microelectronic Engineering 41/42 (1998) 37-40
40
Next-generation DIYV resists must also have improved etch and shrinleage properties over and UVIIHS. Through reduction in the Olmishi lmameter for DUV polymers, we have reduced the etch rate of UV5 and UV6 resists [8,9]. Figure 4 illustrates the reduced etch rate of these polymers.
Figure 4. Etch rate in a CI2/HCI metal etch for recently developed DUV resists. The decreased etch rate of UV5 and UV6 photoresists is due to the reduced Olmishi parameter for these resists. 3. ~ r e
R~
l~ign
DUV resists will continne to evolve over the next few years. Now that DUV is more af~pted by manufacturing lines throughout the world, more emphasis will be placed on simple, high throughlxa processing that can be easily integrated into the modern lab. Issues such as shelflife, ~ and Pl/B sensitivity, substrate sensitivity, and environmental/health effects of DUV resists will become more imnoRant. Ultimately, larger process window margins will be necessary to apply DUV resists at lower design rules. Specialized resists for resolution enhancement teelmiques must be designed (PSM resists, OAI resists, resists designed to reduce proximi~ corrections, etc.). Negative resists may be necessary for specialized features. In order to achieve larger process windows, thinner resist thicknesses will be nm on antireflective layers. Future DUV resists will have higher etch selectivity, especially for metal, and they will have high transparency on antireflective substrates, customized dissolution properties for
the resist feature of intere~ and reduced diffusion. It is an exciting time to be designing DUV resists for use in the semiconductor indnsUy. 4.
Refereaces
I. H. Ito, C. G. Willson, Polym. Eng. Sci., 23, 1012 (1983). 2. J. Vigil, M. Barrick, Proc. SPIE, 2438, pp 626-643 (1994). 3. H. Ito, G. Breyta, D. Hofer, 1~ Sooriyakumamn, K. Petrillo, D. Seeger, "Environmentally Stable Chemical Amplification Positive Resist: Principle, Chemistry, Contamination Resistance and Lithographic Feasibility," J. Photopolym. Sci. Technol. 7, 433448 (1994). 4. W. Conley, G. Breyta, W. Brunsvold, R. Dipietro, D. Hofer, S. Holmes, H. Ito, R. Nunes, G. Fichtl, P. Hagerty, J. Thackeray, "The Lithographic Performance of an Environmentally Stable Chemically Amplified Photoresist," Pro¢, SPlE, 2724, pp. 34-60 (1996). 5. H.-T. Schacht, N. Mnenzel, P. Falcigno, H. Holzw~h, J. Schneider, "Acid-Labile Crosslinked Units: A New Concept for Deep-UV Photoresists," J. Photepolym. Sci. and Technol., 9(4), 573 (1996). 6. N. Munzel. H. Holzwarth, P. Falcigno, H.-T. Sehaeht, R. Sehulz, O. Nalanmsu, A.G. Timko, E. Rcichmanis, J. Kometani, D. R. Stone, T. X. Neenan, E. A. Chandross, S. G. Slater, M. D. Frey, A. Blakeney, " Advanced Positive DUV Photoresists for Practical Deep-UV Lithography," Prec. SPIE, 2195, 47 (1995). 7. G. Wallm~ J. Opitz, W. Hinsberg F. Houle, J. rhac~ray, T. Fedynrsh~ D. r, ane, M. Rajaratnam, "Reactivity and Kinetic Parameters of UVIIHS," Proc. SPIE, 3049, Iyp. 492-500 (1997). 8. R. R. Kunz, S. C. Palmeteer, A. 1~ Forte, R. D. Allen, G.M. Wallmff, R. A. Depietro, D. C. Hofer, " Limits to Etch Resistance for 193nm single layer Resists, " Prec. SPIE, 2724, pp. 365-376 (1996). 9. H. Gokan, S. Esho, Y. Ohnishi, J. Electrnehem. So¢., 130, 143 (1983).