Modeling anomalous depth dependent dissolution effects in. chemically amplified resists

Transcription

1 Modeling anomalous depth dependent dissolution effects in chemically amplified resists Mosong Cheng, Jacek Tyminski*, Ebo Croffie, Andrew Neureuther Electronic Research Laboratory Department of Electrical Engineering and Computer Sciences 231 Cory Hall, University of California, Berkeley, CA 9472 Phone: (51) , Fax: (51) *Nikon Precision Inc Shoreway Rd, Belmont, CA Phone: (65) , Fax: (65) Abstract A post exposure bake and dissolution model for JSR KRF-K2G resist has been established for bake temperature and time effects from large area exposures on production equipment. Data from TARC coated K2G resist were well behaved and provided key understanding of the PEB processes. Data from K2G without TARC showed inhibition of surface dissolution that is possibly due to photoacid evaporation. Both processes showed an intrinsic thickness reduction of 5 nm in the first 3 seconds of PEB and then a continued decrease of 4nm per minute of bake time. An algebraic dissolution model for resist showing an S-shape in the logarithm dissolution rate versus exposure dose is developed using a barrier model. This model was fit with the dissolution rates deduced from the coated K2G as a function of activated site concentration. The rates agree well with DRM data and the DRM data showed further details of depth dependent effects. The methodology of large area exposure allowed the simulation

2 parameters for PEB temperature and time effects to be quantitatively determined very effectively. Keywords: chemically amplified resist, post exposure bake, dissolution rate monitoring, dissolution model, fitting process, simulation I. Introduction Chemically amplified resists are based on the acid catalytic deprotection of a polymer matrix. During the standard post exposure bake (PEB) step, several mechanisms are involved, including a deblocking reaction, photoacid diffusion, acid loss, free volume generation, polymer matrix densification, evaporation of small species such as acid and volatile groups. Some of the systems also suffer from substrate or air contamination. As a result, the dissolution rate distribution within resists is not uniform and can distort the formation of final patterns after development [1]. Dissolution Rate Monitor (DRM) measurements have been widely used as a tool to quantify development pattern formation process. While DRM data is a great input for resist modeling, the data in Fig. 1 presents a formidable challenge. It shows that the dissolution rate in JSR KRF-K2G resist depends on depth. However, this dependence is anomalous in that the rate increases from the top of the resist to the bottom of the resist when the exposure dose is low, while the exposure energy decreases from top to bottom. Note that if a top anti-reflective coating (TARC) is applied before exposure, then this anomalous behavior disappears, as shown in Fig 1 (b).

3 The goal of this paper is to establish a PEB and development model for JSR KRF- K2G resist that accounts for PEB time and temperature effects. The approach is to utilize the large area exposure methodology developed by M. Zuniga et al. [2] and to compare result with the DRM data shown above. This paper begins with observation of intrinsic compaction during PEB. Data on thickness loss versus exposure dose after development are then presented. The data is then normalized using activated site concentration as a common basis. The dissolution parameters of the resist are then fit and compared with DRM data. Finally an isolated line profile simulation of the resist is presented. II. Post exposure bake model During the post exposure bake (PEB) process, several chemical and physical reactions occur. Photoacid catalyses the deblocking/deprotection process [3], in which the blocked polymer is converted to a soluble hydroxyl group and a volatile group. The volatile group then generates free volume. Meanwhile, the photoacid can diffuse and its diffusivity is enhanced by the amount of free volume. Furthermore, if the PEB temperature approaches the glass transition temperature T g, the deprotected polymer matrix starts densifying, which reduces the acid diffusion [3]. As a result of PEB, the evaporation of solvent / free volume can lead to compaction of resist. A model for these effects is forth coming from collaboration work of S. Postinkov and E. Croffie [4]. Based on M. Zuniga s work [6], and given the acid evaporation phenomena which exists in some resists, we assume the following model: C as = K1(1 t C a =.( D t C D = D e as C C a as ) ) C m a K 2 C a - K e (x)c a (3) (1) (2)

4 Where C as is the activated site concentration, C a is the photo acid concentration, K 1 is the reaction rate constant, K 2 is the acid loss rate within resist, K e is the acid evaporation rate at the surface, D is the acid diffusion coefficient in protected polymer and is a constant. Equation (1) describes the deblocking reaction and (2) describes the acid diffusion and acid loss process. There are several mechanisms, which contribute to acid loss. One is free volume loss mechanism given by K 2. Another is surface evaporation described by K e. Here we assumed the acid concentration in air is, and therefore the evaporation rate is only proportional to the resist surface acid concentration. Equation (3) gives the acid diffusion coefficient which is assumed to increase exponentially with the concentration of activated site. III. Dissolution model S-shape dissolution rate versus dose curve has been observed and reported for polyvinyl phenol [7] which is the resin in K2G.. When dose is low, R is flat. When dose reaches some threshold, R increases dramatically, which causes a high contrast. And with dose increasing, R approaches its saturation value. Since there are no models in the literature which give such an S-shape dissolution rate curve, we begin by establishing an algebraic model. KRF-K2G consists of poly(p-vinyl)phenol partially blocked with a blocking group and a photoacid generator [5]. Fig. 2 from [8] depicts how protected groups prevent penetration of developer as a basis for a model for positive DUV chemically amplified

5 resist. We assume three mechanisms are involved in the development of a positive chemically amplified resist: the penetration of developer into the surface layer of resist, the reaction of the developer with the resist and the transportation of the product back into the bulk developer. A block group can prevent the penetration of developer into the resist, hence blocks the dissolution. To establish a model, we assume that the rate limiting step is the penetration of the developer which is governed by a barrier effect with activation energy E u for unprotected sites and activation energy E p for fully protected sites. We assume for partially deprotected sites, the activation energy is a linear combination (1-C as )E u +C as.e p, where C as is the normalized amount of deprotected sites. Therefore the developer penetration rate in partially deprotected polymer is R p ( E u as u p = ( R + R e )( D D ) = ( R + R p p1 C ( E E )) / kt s p p2 e wcas )(D - D s ) (4) Here D and Ds are the developer concentration in the bulk developer and at the surface of polymer, respectively, D>Ds. T is the temperature. Also, R p is the intrinsic penetration rate without depreotection, R p2 and w are the deprotection enhanced penetration components. Both of them are functions of T. The dissolution rate is also given by the surface reaction rate of polymer with the developer, also the dissolution rate, and is R s = kd s (5)

6 Here k is the removal rate of polymer. At steady state, R s =R p. After substitution of (4) to (5) and denoting new parameters, the dissolution rate can be simplified as r R = (1 + ae 1+ be wcas wcas ) (6) Here C as is the activation level, r is the development rate for the unexposed resist, a, b and w are constant. The parameters r, a, b and w have temperature dependence. This model predicts a S-shape curve for a logarithm of dissolution rate versus C as or dose. IV. Experiment Large area exposures of JSR KRF-K2G were made on a Nikon exposure tool. The nominal processing conditions were: softbake 9 o C, 9sec; resist thickness 75nm; post exposure bake 1 o C, 6sec; developed with.261n TMAH, 21 o C, 6sec. The large-area exposures were followed by post exposure bake at different temperatures. Wafers with and without TARC were exposed with doses varying from 6 to 35 mj/cm 2, and baked for the matrix of bake times and temperatures shown in Table 1. Note that a star pattern of various bake temperature for fixed bake time and then varying bake time for fixed temperature was used. Table I. PEB conditions for the experiment Temp. o C Time(sec) After PEB and development, the remaining resist thickness in the exposed and unexposed areas was measured.

7 V. Results and Analysis We now proceed to establish a quantitative model for the effects of PEB and dissolution on the thickness of resist developed. However, we noticed a considerable thickness change of even unexposed resist which could not be accounted for by development. Thus we begin in section 1 with an intrinsic compaction. We then temporarily assume a threshold model in section 2 to estimate reaction parameters. In section 3, with these reaction parameters, a non-threshold model for calculating resist thickness loss is then developed and used to extract dissolution rate parameters. Finally in section 4, DRM curves are fit to obtain the acid diffusion parameters. 1. Intrinsic compaction Significant resist thickness changes were observed during PEB and before development. We call this change intrinsic compaction. The resist thicknesses are shown in Fig 3 and Fig 4 for unexposed regions after development. It can be seen that the resist thickness is decreased. The thickness reduction is about the same for both topcoated and non-top-coated resists. The resist thickness loss is fairly linear with PEB time. It shrinks about 5nm within the first 3sec of PEB. Then it continues to shrink with the rate of 3 4nm per minute as can be seen in Fig. 4. Note that the resist thickness decreases sharply when the PEB temperature is above 1 o C. Since this thickness loss is very large for unexposed films while the dissolution rate for unexposed resist is only.15nm/s, we believe this effect is due to a shrinkage of the materials. Possible mechanism for this intrinsic compaction is resist densification in which the residual solvent and/or free volume are driven out by PEB, which could be considered as a continuation of the soft bake. The effect of high temperature can be

8 explained by the thermal decomposition effect. When the PEB temperature is high, some deprotection of the resin can take place that results in free volume and subsequent compaction. 2. Extraction of reaction parameters K 1,K 2 and m An example of the photoacid concentration as a function of depth after exposure is shown in Fig. 5. This plot was obtained by using the BLEACH program in SAMPLE3D with the Dill s parameters A=-.16 m -1, B=.91 m -1, C=.13mJ -1, the resist layer thickness 75nm, refractive index n=1.56, k=-.2. To obtain the reaction/acid loss rates of the resist, the method described by M. Zuniga et al. [2] was used which is based on a threshold of activated site concentration for development. Ignoring diffusion and assuming uniform acid concentrations in resist, the relationship between acid concentration and bake time to achieve a given activation level is given by m mk 2 Ca [1 exp( mk 2t)] = ln(1 K1 Casth ) if K 2 Or (7) ln(1 Casth ). t = if K (8) K1 Fig. 6 shows the resist thickness after development versus dose for different PEB m Ca 2 = temperatures and times for both TARC and non-tarc processes. It can be seen that JSR KRF-K2G has a very high contrast. When the PEB temperature is below 8 o C, almost no deprotection process occurs even at high exposure doses. Also, varying the temperature by 1 o C from the nominal 1 o C temperature will vary the dose-to-clear by about 1 mj/cm 2. Given the nominal dose-to-clear of 16 mj/cm 2, this means the sensitivity is

9 varied by about 6%. When the PEB time is doubled, the dose-to-clear is reduced by 1 mj/cm 2 which corresponds to a sensitivity increases of about 5%. Assuming identical resist thickness loss corresponding to identical C as, and using average acid concentration obtained from BLEACH as the initial C a, we calculated the different amount of C a to achieve a certain resist thickness loss for different bake times. Then fitting these C a and times with equation (7), we were able to obtain the reaction rate. We fitted the reaction parameters to the thickness developed in the ranges from 5nm to 45nm and obtained the reaction parameters at a PEB temperature of 1 o C. The values in Table II show a difference primarily in the K 2 value. Table II. reaction parameters when PEB temperature is 1 o C. K 1 (sec -1 ) K 2 (sec -1 ) m No TARC TARC This indicates acid evaporation exists. As a first order approximation, the evaporation rate K e is the difference and is.2 sec -1. Given the resist thickness loss at other PEB temperatures [6], we have obtained the reaction parameters, which are shown in Table below. Table III. Reaction parameters LnK 1 (sec -1 ) E K1 (ev) lnk 2 (sec -1 ) E K2 (ev) m K e (sec -1 )

10 3. Extraction of dissolution parameters Having obtained the reaction parameters in section 2, we then extracted the dissolution parameters by fitting the resist thickness developed versus C as data with equation (6) for 1 o C, 6sec PEB. Note that when C as reaches.9 which is enough to clear the resist, the experimental dissolution rate saturates at 7nm/6sec 12nm/sec. We fitted the data in the region C as <.9 with an optimization program that is on the basis of Method of Feasible Direction and can reach the global optimum with certain constraints. Finally we obtained the following dissolution parameters: Table IV. Dissolution parameters r (nm/s) a b w * For TARC-coated, PEB 1 o C and 6sec, The resist thickness curve versus C as generated by the above parameters and the experimental data are both shown in Fig 7. In order to justify the model, Fig 8 shows the thickness developed versus exposure dose curves from experimental data and simulation with the above parameters for 1 o C, 3 sec PEB, TARC-coated process. It can be seen that the two curves fit at low, moderate and high doses. However, the transition of the experimental curve is higher exponentially. 4. Fitting DRM data and extracting acid diffusion parameters With the reaction rate and dissolution rate parameters obtained above, we can now fit the dissolution rate versus depth with the DRM data and also extract the possible acid diffusion parameters with reaction and dissolution parameters fixed. For this purpose, a

11 fast program capable of simulating 1-dimensional transportation-reaction systems was developed. And the MFD optimization program described in section 3 was also applied. Table V summaries the diffusion parameters obtained from the fitting process. Fig. 9 shows the simulation results for no-tarc resist, PEB 1 o C, 6sec, it fits considerably well with the DRM data. Table V Diffusion parameters D (nm 2 /s) 25 6 VI. Resist profile simulation This anomalous behavior of JSR KRF-K2G resist without TARC may cause severe distortion of pattern during development of the resist. An example is T-top. To overcome this effect, a TARC is applied to the resist and Fig 1 shows the resist profile after development, for TARC-coated, dose 7mJ/cm 2, PEB 1 o C, 6sec, develop 1sec. VII. Conclusion In this paper a methodology for extracting PEB and dissolution parameters is used to deduce a model for the JSR KRF-K2G resist. PEB reaction rate and dissolution rate versus chemical state parameters were extracted from large-area exposure and resist thickness measurements. The simulated dissolution rate curves were then compared to the DRM data and this allowed further modeling of acid evaporation.

12 Intrinsic compaction of the resist during PEB process is described. A linear scale transform of space coordinate system is suggested for modeling this compaction phenomenon. We also quantified the PEB temperature and time effects on resist sensitivity. A 1 o C increase of bake temperature causes a 6% increase on sensitivity. Doubling the bake time only increases the sensitivity by 5%. The S-shape dissolution rate versus dose curve is developed using energy barrier model, which is able to explain the dissolution of poly-vinyl phenol. Since DRM data provide very detailed information about both PEB and dissolution effects, fitting DRM data could be a solid basis for extracting and calibrating bake and dissolution models. The parameters extracted from large area exposures agree with DRM data but the DRM data was independently matched in attempting to model depth dependent effects. Reference: [1] H. Miyamoto, et al., Study for the design of high resolution Novolak-DNQ photoresist: the effects of low molecular weight phenolic compounds on resist systems, Proc. SPIE, Vol. 2438, 1995 [2] M. Zuniga, A. Neureuther, Post Exposure Bake Characterization and Parameter Extraction for Positive Deep-UV Resists through Broad Area Exposure Experiments, Proc. of the SPIE, Vol. 2724, [3] B. Mortini, S. Tedesco, B. Dal Zotto, P. Paniez, Specific behavior of chemically amplified systems with low activation energy under electron-beam exposure: implementation of 248 and 193 nm resists, J. Vac. Sci. Technol. B, 15(6), Nov/Dec 1997, pp

13 [4] See papers by S. Postinkov and E. Croffie in this proceedings. S. Postinkov et al, A study of resolution limit due to intrinsic bias in chemically amplified resists, E. Croffie et al, New moving boundary transport model for acid diffusion in chemically amplified resists. [5] A. Tritchkov, R. Jonckheere, L. Van den hove, Use of positive and negative chemically amplified resists in electron-beam direct-write lithography, J. Vac. Sci. Technol. B, 13(6), Nov/Dec 1995, pp [6] M. Zuniga, A. Neureuther, Reaction Diffusion Modeling and Simulation in Positive Deep Ultraviolet Resists, J. Vac. Sci. Tech. B, Dec 95 (2957). [7] T. Itani, H. Iwasaki, M. Fujimoto, K. Kasama, Proc. SPIE, 2195, 126(1994) [8] T. Itani, H. Iwasaki, H. Yoshin, M. Fujimoto, K. Kasama, Proc SPIE, 2438, 91(1995) [9] M. Cheng, E. Croffie, A. Neureuther, Methodology of Modeling and simulating line-end shortening effects in deep-uv resist, Proc. SPIE, Vol 3678 (1999) [1] J. Sheats, B. Smith, Microlithography: science and technology, Marcel Dekker Inc., 1998

14 K2G on Bare Silicon 9 Development Rate [ nm/sec ] z-position [ nm ] (a) K2G; ARC 7 Development Rate [ nm/sec ] z-positon [ nm ] (b) Fig. 1 Dissolution rate vs depth into JSR KRF-K2G resist. In (a), the resist is coated on bare silicon without ARC, in (b), resist is coated with top ARC.

15 TMAH TMAH Block Group Block Group Block Group Block Group Block Group Block Group Block Group (a) (b) Fig. 2 A proposed dissolution model for positive DUV chemically amplified resist. Two situations are considered: (a) before deblocking where TMAH penetration is prevented by block group. (b) after deblocking where TMAH penetration is enhanced by absence of block group.

16 Unexposed resist thickness reduction varying PEB temperature Thickness (nm) Bake temperature(c) w/o TARC w/ TARC Ref Fig. 3 Unexposed resist thickness vs. PEB temperature

17 Thickness (nm) Unexposed resist thickness reduction varying PEB time Bake time(sec) w/o TARC w/ TARC Ref Fig. 4 Unexposed resist thickness vs PEB time

18 .25 Normalized acid concentration w/o TARC w/ TARC Depth(nm) Fig 5 Acid concentration in resist, the exposure dose is 8mJ/cm2, note that there is very little difference between TARC-coated and non-tarc-coated layers.

19 Resist thickness after develop, varying bake temperature, no TAR C Resist thickness after develop, varying bake time, no TARC Thickness(nm) Dose(mJ/cm^2) 34 2C 4C 6C 8C 9C 1C 11C 12C Resist thickness(nm) Dose(mJ/cm^2) 3s 6s 12s 24s 48s (a) (b) Resist thickness after develop, varying bake temperature, with TARC Resist thickness after develop, varying bake time, with TARC Thickness(nm) Dose(mJ/cm^2) 34 2C 4C 6C 8C 9C 1C 11C 12C Resist thickness(nm) Dose(mJ/cm^2) 3s 6s 12s 24s 48s (b) (d) Fig. 6. Resist thickness after development versus dose. The curves in (a) and (c) are for different bake temperatures. The curves in (b) and (d) are for different bake times at 1 o C.

20 Dissolution rate vs. normalized activated site concentration dissolution rate(nm/s) normalized activated site concentration Calculate Experimental Fig. 7 Resist thickness vs Cas, PEB 1 o C, 6sec. Experimental data are marked with * and simulated results are marked with o.

21 Resist thickness loss vs. dose Thickness developed(nm) Calculation Experimental Dose(mJ/cm^2) Fig. 8 Resist thickness loss vs. dose curves, PEB 1 o C, 3sec.

22 K2G without TARC Development Rate [ nm/sec ] Simulation for dose 6.7mJ/cm^2 DRM curve for dose 6.7mJ/cm^ z-position [ nm ] Fig 9 DRM date, experimental and simulation. PEB 1 o C, 6sec, dose 6.7mJ/cm 2.

23 Fig. 1 Resist profile simulation