, i -..,. Polvmer. Effects on Acid Generation Efficiency Using EUV and DUV Exposures - The mbmitkd amnuaaripihas b Paul Dentinger,* Robert L. Brainard,2 Joseph F. Mackevich,2 Jeffrey M. Guevremont? and Charles R. Szmanda2 adhcmed by acontractor of the VE: States Government under cmntr Acmxdiigly the Uniked Statea G ernment retains a non- exchw royalty-free Ibrmse to publish or Sandia National Laboratories, Livermore, CA 94550 prodme the pablished form +ef mntrbztion, or allow othexs to de 2Shipley Company, 455 Forest Street, Marlborough, MA 01752 for U?likld%tea ~mt ~ Thin resist films (< 1500 ~) based on DUV chemical approaches ha=been demonstrated for use in EUV lithography. Resists with good sensitivity (5-6 mj/cm2) were observed but imaging mechanisms, in particular as they affect sensitivity, are poorly understood. To clarify mechanisms leading to photosensitivity, acid-generation efficiency at both EUV and DUV wavelengths was measured for our most promising EUV resist compositions as well as initial radiation damage experiments. In previous work, [R. Brainard et al. J. Vat. Sci. Technol B 17(6), p. 3384, 1999] polymer composition was found to be more important in determining the relative dose to print of resists to EUV and DUV radiation than was PAG composition. Here, acid generating efficiency for several polymers upon exposure to EUV and DW are compared to gain insight into the role of the polymer and PAG in converting the incident EUV photon energy into resist images. It is shown that acid generation efficiencies at EUV do not track efficiencies measured on identical films with DUV exposures, and is attributable to polymer and polymer/pag interactions. No particular structural feature of the polymer could be. correlated to the acid generation results. Radiation damage studies showed that polymers that create acid in different yields at EUV do not show differences in radiation damage, - - as detected by dissolution rate changes. In addition, it is shown that no significant dissolution altering mechanisms occur with EW radiation at relevant exposure doses. - We conclude that photospeed differences between EUV and DUV are quantitatively attributable to acid generation ei%ciencies for the compositions studied. RECEIVED Au$u.42@ I. INTRODUCTION QsT/ Recently, chemically amplified EUV photoresists have been developed which utilize phenolic ESCAP polymers and thin films (<150 nm). 1~2 These new resists can be imaged using DUV or EW, and in some cases, DW imaging has been used as a screening tool for developing these EW resists. 1 IVhen exposed to DUV light, these &p*-- resists printed lines at an average of 12 mj/cmz with 9Y~ When exposed with EW light, the resists printed lines at an average of 6.7 mj/cm2 with 30% absorption. Dose to print, therefore, was approximately 200 J/cm3 for EW and 100 J/cm3 at DUV. A significant 1
DISCLAIMER This report was prepared as an account of work sponsored by an agency of the United States Government. Neither the United States Government nor any agency thereof, nor any of their employees, make any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof.
DISCLAIMER Portions of this document may be illegibie in electronic image products. Images are produced from the best available original document.
. portion of the absorbed energy was not utilized in EUV. Despite this inefficiency, resists with good sensitivity (5-6 rnj/cm2) have been demonstrated. Nonetheless, the mechanism for obtaining photospeed when DUV-based materials are exposed to EUV radiation, and the ultimate sensitivity limits of this approach are not understood. The latter has very important implications for throughput and cost-of-ownership. In previous work,l it was found that polymer composition was more important in determining the relative sensitivities of resists exposed to EUV and DUV radiation than was PAG composition. Figure 1 shows a plot of EUV vs. DUV sizing energies for 19 resists. These resists were prepared from three phenolic ESCAP polymers that vary in monomer composition but not monomer type. Despite a wide range of formulation variables including the use of 5 photoacid generators (PAGs), the nineteen resists broke into three bands depending upon polymer composition. Resists prepared with Polymer C were comparatively more sensitive at EUV radiation than the other two. This polymer appeared to utilize the EUV energy more efficiently than the other two polymers. In this paper, we investigate the role of the polymer in determining sensitivity i-n EUV, to gain insight into why EUV exposure is in general less el%cient than DUV exposure and to better understand why certain polymers are more efficient at EUV than would be expected based on DUV sensitivity. We utilize a method for determining the acid generation efficiency or Dill C-Parameter described elsewhere.3)4 Nine resists were prepared from five phenolic ESCAP polymers. The five polymers were chosen so that we could first test the hypothesis put forth by otherss that hydroxystyrene (HS) content in the polymer resin is responsible for acid generation under radiolysis. Next we test whether the polymer alone determines the acid generation, or whether polyrner/pag 2
., interactions predominate. Finally, we test radiation-induced damage to the polymers to assess more completely possible photospeed differences between DUV and EUV. II. EXPERIMENTAL A. Processing Thirty-six chemically amplified resists were prepared in a 9x4 experimental array utilizing nine polymer-pag combinations and four levels of base (O, 3.33,6.66, and 10 mole Avs. PAG) as shown in Table I. All polymers are hydroxystyrene-based ESCAP systems with the only difference being the ratio of monomers. All wafers were spun to 1000 ~ thickness, had a post-apply bake (PAB) of 130 C/60 sec. and a post-exposure bake (PEB) of 130 C/90 sec. and were developed in a single puddle for 45 sec. with Microposit@ LDD-26W, unless otherwise noted. B. Acid generation efficiency The method used to measure the first-order rate constants for acid generation or C-parameters was discussed previously.s~ 1 In brief, several resists are formulated and all are identical except that they contain different concentrations of the base ([B]), which is usually varied between O-107O (mol/mol) of the PAG concentration. Some indication of the threshold exposure dose (EOor EjO) is then measured using standard techniques for each resist in the set. In previous work, it was shown that over a concentration range of approximately O-10?ZO,[B] and E. are related linearly. In a simple plot of [B] vs. E., the slope of the line gives the uncorrected C-parameter. The C-parameter corrected for absorbance is obtained by multiplying the uncorrected value by a/[1 -e- ) where a is the absorbance of the film in base e units. For EUV exposures, the correction is approximately 1.23 and for DUV exposures the correction is 1.047 for all of the polymers and photoresists chosen. Uncorrected C-parameters are reported unless otherwise stated.
!.. EW C-Parameters were determined by plotting base loading against the energy required to allow for only 50 XOfilm thickness remaining (E50in mj/cm2). DW C- parameters were measured by determining the slope of base loading against E. in mj/cm2. The difference in C-Parameters determined using E. vs. E50at DW is estimated to be less than lyo. C. Exposures EW exposures were performed at Sandia National Laboratories by filtering the output of Xe laser-produced plasma source with a 1 pm Be membrane. Beryllium removes the IR, visible, DUV, and VW portions of the spectrum as well as energies greater than 111 ev. The resulting spectrum at the wafer plane is a broad band of EUV photons centered near an energy of 92.5 ev (13.4 rim), similar to that used for imaging, Film thickness measurements were performed with a Nanospec film thickness monitor. Films exposed at 248 nm were first coated with 60 nm of AR3 DUV antireflective coating and exposed with a GCA XLS 78000.53 NA DUV stepper at Shipley. Error b ars shown on the graphs are given as 1.63 times the standard deviation, from the fit unless otherwise stated. Because DUV and EUV exposures both utilized 4 points to fit the line, the pooled standard deviation is an average of the two standard deviations. A difference in the slopes can be attributed at 99$Z0confidence as 3.26 times the pooled standard deviation using a combined degrees of freedom of 4. Utilizing 1.63x standard deviations allows for visual comparison: lack of overlap indicates the slopes are different with 99% confidence. III. RESULTS AND DISCUSSION
A. Acid generation efficiencies of polymers Acid generation efficiencies for the five polymers, A-E, were measured and the results are shown in Figure 2a. It should first be noted that the numerically similar uncorrected C-parameter values at DUV and EUV for polymer A was fortuitous. The value based per unit of energy absorbed would be different. At DUV, there was little affect of the polymers chosen on the efficiency. However, it was clear that polymer C and D have higher acid generation efficiencies at EUV than their DUV counterparts. Polymer C was the polymer used in Fig. 1 that printed faster photoresists. In addition, it has been proposedq that the acid generation efficiency under radiolysis is driven by the HS concentration in t-boc protected polymers. Fig. 2b plots the same data vs. relative HS mole concentration in the ESCAP polymers. At DUV, it appears as if there was a slight decrease in acid generation as the HS content of the polymer increased. However, in EUV, it was clear that no monotonic increase in acid generation efficiency occurred with increased HS content. Figure 3 shows the threshold acid concentrations (intercept of C-parameter plots) for DUV and EUV exposures (Fig. 3a) as well as the dark loss corresponding to each exposure (Fig. 3b). Although polymer radiolysis might be expected to influence the threshold acid concentration in the EUV samples, statistically significant differences were not observed. Furthermore, the unexposed film thickness loss (UFTL) plot indicates substantial agreement between our two laboratories. In fact, because the C- parameter for B and C are similar at DUV wavelengths, as well as the threshold acid concentration and the UFTL, we conclude that the polymers B and C should have
,,... comparable photospeeds at DUV. The reason for the difference between polymer B and I C in Fig. 1 appears to be that polymer C created acid more efficiently at EUV. B. Polymer/PAG interactions It appears from Fig. 2 that the polymer can have a significant effect on the acid generation efficiency, and that this can produce comparatively fast photoresists at EW wavelengths. It is not clear whether the polymer alone dominates the acid generation at EUV wavelengths or whether the PAG and polymer together alter the acid generation efficiency at EUV. To test this, C-parameters for polymers B and C were measured for a range of photoacid generators at both DUV and EUV wavelengths, and the results are shown in Figure 4. PAG 1 is the same result as shown in Fig. 2 and is included for comparison. Fig. 4 compares the difference between the polymers using different PAGs and DW vs. EUV radiation. At DUV again the acid generation efficiency is relatively independent of polymer across the three PAGs. In addition, the photoacid generators behave similarly with the exception of ND-Tf,6 which is essentially transparent at 248 ~ nm. On the other hand, it is clear that there can be a dramatic effect of the polymer at EUV wavelengths. For the ND-Tf PAG at EUV, there was not a statistical difference betweenpolymers B and C since only 3 points were obtained for the C-parameter plot. However, for the DTBI-PFOS PAG,T there is a large difference between polymers B and C, polymer C again being much more efficient. In fact, there is a significant difference at EUV when comparing across PAGs, DTBI-PFOS being the most efficient. From the small sample set of three photoacid generators, we cannot conclude that polymer C generates acid more efficiently than polymer B over all PAGs. However, it is 6
., - clear from the data, that simply considering polymer alone or PAG alone in the process of optimizing acid generation at EUV will be misleading, and that polymer/l?ag interactions are predominant. C. Radiation damage It has been shown that polymer C can produce more acid per unit dose than B at EUV wavelengths, and that polymer C produces faster photoresists as a result. However two questions remain. First, is efficiency of acid generation the only factor that determines the relative photospeed between DUV and EUV exposures of these polymers, or do photoinduced side reactions such as chain scission or crosslinking make significant contributions? Second, why does one polymer which has the same monomer groups, albeit at different concentrations, create acid with different efficiency? As an attempt to answer these questions, polymers and photoresists were exposed to radiation without performing deblocking chemistry, and dissolution rate changes between unexposed and exposed regions were measured. Deblocking chemistry was not performed so as not to overwhelm more subtle, radiation-induced effects. For instance, if significant crosslinking reactions were occurring, we would expect that the resins in the exposed regions would dissolve more slowly, and more rapidly if scissioning reactions were predominant. Formulations 2 and 3 have the same PAG but differ by polymers B and C, respectively. Figure 5a shows the results of the experiment where formulations 2 and 3 were exposed at EUV and thickness measurements were made. In this experiment, the development time was extended to 145 sec. in order to differentiate between unexposed
and exposed portions of the film. Fig. 5a shows first that there was no detectable difference between formulations 2 and 3 that have the same PAG but differ only by polymer. Second, the behavior of the photoresist to EUV radiation at doses up to = 5 times E,im was entirely positive tone. The positive tone could be attributed to either dissolution inhibition effects of the photoacid generator or radiation-induced reactions of the polymer. To differentiate between these two mechanisms, the pure polymers without PAG were exposed, and the results are shown in Fig. 5b. The EUV exposures no longer show the positive tone behavior found in Fig. 5a, and again no significant difference between B and C were found. We consider the small amount of apparent negative tone behavior found in Fig. 5a to be irrelevant for imaging at EUVwavelengths for two reasons. First, negative tone behavior was found in the DUV results at approximately the same energy of exposure relative to sizing energies. Second, at 5 times E,iz,, the amount that the dissolution rate slows down is =2-3 ~sec., which is much less than the =26 ~sec. for unexposed film. Since sizing doses produce maximum dissolution rate changes of about 10,000 ~sec. when deblocking chemistry is performed (Table I), this slight negative tone behavior is negligible at typical imaging exposure doses. In addition to the results shown in Fig. 5b, polymers B and C were exposed to z. similar amounts of radiation at both EUV and DUV wavelengths followed by a PEB of 130 C and 90 sec. There were no significant differences between the results with or without PEB, except that the PEB wafers dissolved slightly slower in all areas. The absence of any significant change in the exposure curves by PEB indicates that the polymer alone generates no acid.
It is clear, therefore, that the sensitivity improvements from polymer C as shown in Fig. 1 are due entirely to acid generation efficiency. It is not clear, however, why the resist prepared from polymer C generate acid more efficiently. Polymers B and C show little difference at DUV because direct photon absorption by the PAG is largely responsible for acid generation, and the subtle changes in polymer are unlikely to affect this process. However, at EUV atomic absorption dominates, and the polymer constitutes approximately 95% of the atoms in the photoresist, so acid generation must go through the polymer. If hydrogen abstraction from the polymer by PAG fragments were the difference between polymers B and C, then some dissolution rate differences may have been observed, as hydrogen abstraction is critical in crosslinking and chain scission reactions. However, dissolution rate changes by direct exposure did not indicate different mechanisms of radiation damage and consequently did not indicate a mechanistic difference in acid generation between the polymers. Energy transfer from the polymer to the PAG may be critical to determining acid generation at EUV. We might expect then, that affects such as the added PAGs inhibition of the pure polymer dissolution rate could be correlated with the polymer acid generation efficiency of a resist. Dissolution inhibition by PAGs ostensibly works by enhancing or reducing hydrogen bonding interactions within the photoresist, and may provide some indication of the capability of the PAG and polymer to mix. Potentially, this could be a metric for predicting how effectively the polymer transfers energy to the PAG. Dissolution inhibition effects of the PAG were different between the polymers B and C studied, but there were few data, so this is an area of future work. 9
IV. CONCLUSIONS Photospeed differences between EUV and DW can be quantitatively assigned to different acid generation efficiencies of polymer/pag combinations at EUV which are not seen with DUV exposures. No structural feature of the polymer explored in this work (e.g. monomer concentration) could explain the acid generation results. Simple consideration of PAG or polymer properties alone cannot explain all of the results; instead polymer/pag interactions dominate the process. Photolysis damage studies at DUV and EUV indicate that a) no significant dissolution altering mechanisms occur at relevant sizing energies at EUV for the polymers studied and b) that no significant dissolution altering mechanisms occur between polymers which otherwise produce acid in different yields. Polymers with similar chemical properties produce acid at different rates upon EUV exposure. V. ACKNOWLEDGEMENTS The authors would like to thank Scott Gunn for assistance, C. Henderson, G. Cardinale, and D. O Connell of Sandia Corp., J. Cobb of Motorola, and G. Taylor of the Shipley-Co. for useful discussions. The work performed at Sandia National Laboratories, a Lockheed Martin Company, was supported by the Extreme Ultraviolet Limited Liability Company (EUV LLC) and by the U.S. Department of Energy under contract DE-AC04-94AL85000 VI. REFERENCES 1. R.L. Brainard, C. Henderson, J. Cobb, V. Rae, J.F. Mackevich, U. Okoroanyanwu, S. Gunn, J. Chambers, and S. Connolly, Journal of Vacuum Science and Technology, 1999. B 17(6): p. 3384-3389. 2. P.M. Dentinger, C. Henderson, G. Cardinale, A. Fisher, and A. Ray-Chaudhuri. Proc. SPIE. in press, 2000,.
3. T. Kozawa, S. Nagahara, Y. Yoshida, S. Tagawa, T. Watanabe, and Y. Yamashita, Journal of Vacuum Science and Technology, 1997. B 15(6): p. 2582-2586. 4. C.R. Szmanda, R.J. Kavanaugh, R.J. 130hland, J.F. Cameron, P. Trefonas, and R.F. Blacksmith. Proc. SPIE. 3678, 1999, p. 857. 5. C.R. Szmanda, et al., Journal of Vacuum Science and Technology, 1999. B 17(6): p. 3356. 6. W.R. Brunsvold, C.J. Knors, R.W.L. Kwong, S.S. Miura, M.W. Montgomery, W.M. Moreau, and H.S. Sachdev,.1990: European Patent 90-48001619900202. 7. Y. Suzuki and D.W. Johnson. Proc. SPIE. 3333,1998, p. 735-746. -..- 11
El A O Polymer A Polymer B Polymer C o-l I I I I I I I I I I 1 i I 02468 10 12 14 16 18 29 22 24 26 DUV Dose to Size (mj/cm ),.=. Figure 1: Correlation between DUV dose to size and EUV dose to size for 19 formulations.
Table L Resist Compositions at Dissolution Rates for Formulations Used Polymer IPAG Formulation Polymer Relative PAG Base ~ at L at Amount of 107o base 10% base HS (A/see.)* (Msec.) * 1 2 3 4 5 6 7 8 9 * Determined for 1 pm thick films. A 1.00 PAG 1 I Base 1 1.9 7700 B 1.08 PAG 1 Base 1 2.7 3400 c 1.09 PAG 1 Base 1 2.6 17600 D 1.22 PAG 1 Base 1 4.9 15600 E 1.13 PAG 1 Base 1 3.9 28000,, i a! B I 1.08 I ND-Tf i DBU I - I I I I 1 I c 1.09 I ND-Tf I DBU I - I - I B 1.08 I+ PFOS DBU - - c 1.09 I+ PFOS DBU - -
.. 75 El 70 65 60 / 55 0 50 84x10-3 80 76 A B c D E Polymer i -1 72 68! 1 e 1 52- I I I I I I I 1.00 1.02 1.04 1.06 1.08 1.10 1.12 1.14 1.16 1.18 1.20 1.22 Hydroxystyrene concentration ( mole % rel. to polymer A) n Figure 2: Top a): C-parameter measurements at DUV and EUV for formulations 1 through 5. Polymer C is significantly more efficient than A and B, and also prints faster photoresists, as shown in Fig. 1. See text for explanation of error bars. Bottom b): C- pammeter vs. normalized mole % HS in polymer at EUV and DUV.
, 90X10-3 85: DEal 80 75 3 70 D J 65 g 4 60 =i 55 50 45 ; ~, 40! 35 f A B c D E Polymer :~ EEEa$- $160- $ 140 3 $ 120- * k s ~oo_ 2 80, ii 60 40 -i!! A B c D E Polymer Figure 3: Top: Intercept on C-parameter plots for EUV and EUV exposures. The intercept is interpreted as the threshold acid concentration, and is independent of photons used. Bottom: Unexposed film thickness loss (UFTL) for the 5 resists shown in Fig. 2. The error bars are estimated as 7 ~.
I 1, E21 DUV Polymer B m EUV Polymer B Cl DUV Polymer C U EUV Polymer C PAG 1 0 1 Ez 3 Figure 4: C-parameters of polymers B and C at EUV and DUV wavelengths across three photoacid generators. Note the small difference between polymers at DUV, but a very significant difference at EUV. The PEB of all ND-Tf formulations was 110 C and 90 sec.
.. ~. 1.0 0.9 * ~ 0.8 a) S 0.7 : 0.6.1 % ().5 N ~ 0.4 E 0.3 202 Polymer B with PAG 1 0 Polymer C with PAG 1 @w@ l!!? 6 o 0.1 I 0.04 1 1 1 I t 1 1 1 I I 1 1 1 t 1 i 1 2 3 4567891 2 3 4 567891 2 0.1 10 Jose (mj/cm2j 0.80 0.75- VJ 0.70- j o,65 ~ + 0.60- ~ 0.55 N ~ 0.50 o E ~0.45 0 O Polymer B Film, EUV Polymer C Film, EUV Polymer B Film, DUV Polymer C Film, DUV E 0.40. 0.35 : n2n-1.j../ L I 1 I 1 1 1 1 1 I 0.1 1.000u@m- I 1 1 1 1 I I t I Y Dose (mj/cm J I 1 & 1 1 1 t 1 I 100 1 Figure 5. Top a): EUV exposures of formulations 2 and 3 without performing a PEBstep. The typical dose to size for these resists is roughly 6 rnj/cm2. Note how only positive-tone behavior is seen in the exposed areas. The development time was 155 sec. for this experiment. Bottom b): EUV and DUV exposures of polymers B and C only. Some negative tone behavior is seen at EUV and DUV exposures.