understanding quencher mechanisms by considering … · c litho tech japan, 2-6-6-201 namiki,...

*[email protected]; This work was done at Department of Electrical Engineering and Computer Sciences, University of California, Berkeley as a visiting research topic; The current address: IMEC vzw, Kapeldreef 75 B-3001 Lueven- Belgium **[email protected]

Understanding quencher mechanisms by considering photoacid-dissociation equilibrium

in chemically-amplified resists

Seiji Nagaharaa,b*, Lei Yuana, Wojtek Jacob Poppe a, Andrew Neureuther a**, Yoshiyuki Konoc, Atsushi Sekiguchic,

Koichi Fujiwarad, Tsuyoshi “Gary” Watanabed, Kazuo Tairae, Shiro Kusumoto f, Takanori Nakano f, Tsutomu Shimokawa f

a Dept. of EECS, Univ. of California, Cory Hall, Berkeley, CA 94720-1774;

b NEC Electronics Corporation, 1120 Shimokuzawa, Sagamihara, Kanagawa 229-1198, Japan; c Litho Tech Japan, 2-6-6-201 Namiki, Kawaguchi, Saitama 332-0034, Japan;

d JSR Micro, Inc., 1280 N. Mathilda Ave., Sunnyvale, CA 94089; e JSR Corporation, 5-6-10 Tsukiji, Chuo-ku, Tokyo 104-8410, Japan;

f JSR Corporation, 100 Kawajiri-cho, Yokkaichi, Mie 510-8552, Japan;

ABSTRACT The quencher mechanisms in Chemically-Amplified (CA) resists have been investigated. To explain the acid distribution with a variety of acid strengths in the presence of quencher, a new full Acid-Equilibrium-Quencher model (AEQ model) is proposed and examined in solid-model-CA-resist systems. To observe the reactions in the CA resists, real-time Fourier-Transform-Infrared Spectroscopy (FTIR) is employed during post-exposure bake (PEB). The FTIR peaks of the protection groups are detected to measure the reaction kinetics during PEB. The solid-model-CA resists used in this work consist of both a KrF-acetal-type resist with a diazomethane Photo-Acid Generator (PAG) (weaker-photoacid system) and an ArF-ester-type resist with a sulfonium-salt PAG (stronger-photoacid system). The obtained FTIR results are analyzed using conventional Full-Dissociation-Quencher model (FDQ model) and the new AEQ model. The kinetic analysis of the model resists was performed for different quencher loadings. For the weaker-photoacid system, the AEQ model much more accurately predicts the deprotection-reaction kinetics than the FDQ model with the change of quencher content. This suggests the necessity of introduction of the acid-dissociation concept in the case of the weaker photoacid. For the stronger-photoacid system, both the AEQ and conventional FDQ models adequately predict the kinetic results. This shows that the conventional FDQ model is accurate enough to simulate the super-strong photoacid system. Finally, the new AEQ model is introduced in the UC Berkeley STORM resist simulator. Some simulation examples are shown in the paper. Keywords: quencher mechanism, chemically-amplified resist, FTIR, strength of acid, full-dissociation, acid-equilibrium-quencher (AEQ) model, acid dissociation, STORM, resist simulator

1. INTRODUCTION Bases such as air-borne amine contaminants are a critical issue for CA resists1. However, if an amine is formulated as a so called quencher in CA resists in advance, it helps CA resists in various respects 1-14. For example, the quencher helps to realize the higher environmental stability of CA resists1-5, 13, 14. Also the quencher often helps to obtain better resist contrast6, 11, resist resolution7, 12, exposure latitude (EL) 8, and line-edge roughness (LER) 8, 9. Also it is often used to precisely control the vertical resist shape, iso-focal bias, and resist sensitivity. The purpose of this work is to understand the quencher mechanisms in CA resists and to model the quencher effects for the use in a CA-resist simulator.

Advances in Resist Technology and Processing XXII, edited by John L. Sturtevant,Proceedings of SPIE Vol. 5753 (SPIE, Bellingham, WA, 2005)

0277-786X/05/$15 · doi: 10.1117/12.598949

338

By titration experiments using liquid models of CA resist, the quencher effects have been examined13, 14. The pH behavior in organic solvents showed that the acid dissociation should be considered to explain the observed buffer effects induced by the addition of quencher. This is because even the strong acids in water can work as much weaker acid in organic solvents. From the results, a new Acid-Equilibrium-Quencher model (AEQ model) was proposed to understand the acid concentration in CA resists. In this work the AEQ model is examined in solid model-CA-resist systems. The AEQ model is also expanded to describe the solid-CA- resist system. In section 3.1 of this paper, the concepts of conventional Full-Dissociation-Quencher (FDQ) model and new AEQ model are briefly summarized. Then, in section 3.2, the AEQ model is described in the form of differential equations for the implementation in Berkeley Simulation-Tool-Of-Resist Model (STORM). Then the AEQ model is verified in CA resists by the experiments using real-time Fourier-Transform-Infrared Spectroscopy (FTIR) during Post-Exposure Bake (PEB) in section 3.3. The AEQ model is shown to more accurately predict the deprotection reaction during the PEB than FDQ model especially when the weak photoacid is used to catalyze the deprotection reaction in the CA resists. Finally in section 3.4 some examples of quencher-effect simulation based on AEQ model are shown.

2. EXPERIMENTAL 2. 1. Resist Materials To verify the quencher mechanisms, the model resists are formulated. One is an acetal-type-KrF resist formulated with diazomethane Photo-Acid Generator (PAG) in Fig.1 (left). The other is the ArF resist16 with sulfonium-salt PAG in Fig.1 (right). In those resists, the amine quencher is added with the different loadings (0, 10, 25, 40, 60 mol% to PAG mol %).

Hydroxystyrene/Ethylvinylether-hydroxystyreneCo-polymer

Tri(n-octyl)amine

PGMEA

OH O

O

3367 S CN2

SO

O

O

O

Bis(t-butylsulfonyl)diazomethaneContents: 6 wt% vs. Resin

KrF Resin PAG

Mw=282.37

Mw=23000Mw/Mn=1.85

Quencher

Mw=353.68

H3C(H2C)6H2CN

H3C(H2C)6H2CCH2(CH2)6CH3

Contents: 0, 0.75, 1.89, 3.0, 4.5 wt% vs. Resin (0, 10, 25, 40, 60 mol% vs. PAG)

SolventTri(n-octyl)amine

PGMEA

TriphenylsulfoniumnonafluorobutanesulfonateContents: 6 wt% vs. Resin

ArF Resin PAG

Mw=562.47

Quencher

Mw=353.68

H3C(H2C)6H2CN

H3C(H2C)6H2CCH2(CH2)6CH3

Contents: 0, 0.75, 1.89, 3.0, 4.5 wt% vs. Resin (0, 10, 25, 40, 60 mol% vs. PAG)

Solvent

OO

O

O

OO

50 50

Mw=10000Mw/Mn=2.0

2-Methyl-adamanthyl-2-methacryrate/2,6-norbornenecarbolactone-5-methacrylate copolymer

S O S

O

OC4F9

KrF resist (non-fluorinated-weaker-photoacid system) ArF resist (fluorinated-stronger-photoacid system)

Figure 1: Model-resist formulation.

2.2. Measurement The FTIR measurements during PEB were done with the Litho Tech Japan (LTJ)’s PAGA-100 FTIR system which uses Bio-Rad’s FTIR spectrometer (Fig. 2) according to the literature flow17.

Exposed Wafer (KrF, ArF)

Bake plate

IR

50 µm

Liquid nitrogen cooled MCT

photoconductivedetector

IR

Bake plate

Figure2: Real time PEB FTIR measurement system.

Proc. of SPIE Vol. 5753 339

The infrared light goes through 10-mm-diameter hole which is opened in the center of the PEB hot plate. N2 pursing through the optical path was done to eliminate the effect of CO2. As a FTIR detector, a newly improved liquid-nitrogen-cooled MCT (Mercury Cadmium Telluride) photoconductive detector was installed and used. The resists were coated on the low-crystal-defect wafers with 500-nm thickness without using bottom antireflective coatings (BARCs) and were exposed with KrF or ArF light. After the wafer exposure, the FTIR spectra were measured during PEB time.

3. RESULTS AND DISCUSSION 3.1 Concepts of Full-Dissociation-Quencher (FDQ) model and Acid-Equilibrium-Quencher (AEQ) model In the conventional quencher model15 which we call “Full-Dissociation-Quencher (FDQ) model”, PAG generates the acid (HAgenerated) by UV light (reaction 1).

UV generatedPAG HA + -by product→ (1)

Acid is assumed to be strong enough to realize the full dissociation (reaction 2).

+generated activeHA H + A− → (2)

Here we use “H+

active” for the “catalytic” acid in CA resists in order to distinguish it from the photo-generated acid (HAgenerated). The A− in reaction 2 is a dissociated-counter anion of acid (a conjugate base of the acid). In this model, the H+

active is neutralized by quencher (Q) (reaction 3) independently of the acid-dissociation reaction (reaction 2) and loses the catalytic property to the deprotection reaction in a CA resist.

+

activeH Q H Q++ → (3)

Therefore simply the concentration of H+

active in the FDQ model is simply the difference between the photo-generated-acid concentration [H]generated and the quencher-concentration [Q] (equation 4).

active generated

[H ] [H] [Q]+ −= (4)

We examined this model by questioning whether the acid always fully dissociated in the resists (reaction 2) or partially dissociated in the resists in the case of a weaker acid system (reaction 5).

d

cactive

HA H + Ak

k

+ − →← (5)

In the partial-dissociation case, it is necessary to consider the acid-dissociation equilibrium (reaction 5). We believe that it is necessary to have a general description that includes the partial dissociation to understand and simulate the weaker-acid system. To support this consideration, we will show that the acid in resists can become weaker and that the quencher will help push the equilibrium reaction (reaction 5) to the left hand side. To describe the acid strength, Ka or pKa is often used (equation 6). The Ka is a ratio of the acid dissociation-reaction-rate constant (kd) and combination-reaction-rate constant (kc). Concentration relationship among HA, H+

active and A− in the equation 6 is assumed to be constant at a fixed temperature.

+

d active

a

c

[H ] [A ]

[HA]

kK

k

−

= ≅ (6)

340 Proc. of SPIE Vol. 5753

The lower acid pKa (=minus log of Ka) means stronger acid (equation 7).

a 10 ap = logK K− (7) Figure 3 shows the pKa range of fluorinated-super-strong acid and non-fluorinated acid. In Fig. 3, it can be seen that the pKa becomes higher in non-aqueous solvents than in the water system. This indicates that even very strong acids in water can act as a weak acid in organic matrices. Even for the fluorinated-super-strong acid, the acid becomes weaker in organic solvents and pKa becomes positive in Fig.3. In the case of the non-fluorinated sulfonic acid, the acid works as strong acid in water but in organic solvents the acid works as a very weak acid. The change in acid-dissociation constants due to the matrix is amazingly more than 10 orders of magnitude. The experimental consideration about the range of pKa in CA resists will be shown in Session 3.3. To include the partial dissociation, we have proposed an Acid Equilibrium Quencher (AEQ) model including the effect of photoacid-dissociation equilibrium14. In this model, at first a neutralization in reaction 8 occurs between HA and Q. By the neutralization, the conjugate base of acid (A−) (an anion of acid) is increased in the system. The reverse reaction in reaction 8 can be neglected because the pKa of the quencher is usually much higher than the acid pKa.

Conjugate base

+

Q + HA QH + A− → (8)

The increased A− pushes back the photoacid dissociation in reaction 9.

active

Buffered acid Conjugate base

HA H + A

+ −→← (9)

This coupling of the reactions 8 and 9 is called the “common-ion effect” 19. This induces more buffered acid. The more buffered acid results in a more stable system. The buffered acid can suppress the change of [H+]active due to the consumption of the acid such as by base contamination. This system is quite similar to the pH buffer solutions19 which consist of a weak acid and a conjugate base of the acid. If the buffer system is realized in a CA resist system then a similar effect of a more constant pH will occur in the resist. When the concentration relationships in quencher systems with the acid-dissociation equilibrium are discussed, it is necessary to consider two types of the reaction, that is, neutralization (reaction 10) and acid dissociation (reaction 11). Neutralization

Q + HA QH A+ −+ → (10)

generatedStarting [Q] [H] 0 0

Change [Q] [Q] − −

generated

[Q] [Q]

After Reaction 0 [H] [Q] [Q] [Q]

+ +−

Acid dissociation after neutralization

active HA H + A+ −→← (11)

generatedStarting [H] [Q] 0 [Q]

Change

[H

−

+− +

active active active

+ + + generated active active active

] [H ] [H ]

Equilibrium [H] [Q] [H ] [H ] [Q] [H ]

+

+

− −

+ +

Figure 3: pKa range of fluorinated-super acid and non-fluorinated weaker acid. Acid becomes weaker in non-aqueous matrices.18, 19

pKa

Weaker

in water(εrel = 78)

-10 -5 0 5 10

in propylenecarbonate (PC)(εrel = 64)

in acetonitrile(AN) (εrel = 36)

2.62.2-14

in dimethylsulfoxide(DMSO)(εrel = 46)

0.3

in waterin AN

10in PC

8.3-2.6

in DMSO

1.6

CF3SO3H (Fluorinated super acid)

CH3SO3H (Non-fluorinated acid)

Stronger

O O

O

H3CS

CH3

OHO

H

H3C C N


From the above consideration, the following relation is obtained (equations 12 and 13).

+ generated active

[HA] [H] [Q] [H ]− −= (12)

active+[ ] = [Q] + [H ]−Α (13)

If the concentration relationship in equations 12 and 13 is substituted in the equation 6, relationship in equation 14 is obtained.

+ generated active

active a a +active

[H] [Q] [H ] [HA][H ]

[A ] [Q] [H ]K K

− −+− +

= = (14)

With the assumption that the [H+]active is small enough that [HA] = [H]generated − [Q] − [H+] active ≈ [H+]generated − [Q] and [A−] = [Q] + [H+]active ≈ [Q], the [H+]active is described by equation 15.

( )active generated when [H ] [HA] generated

active a [H] [Q]

[H ] [Q]

K +−+ = � (15)

The [H]generated − [Q] is the same as the FDQ model, but here the value is divided by the quencher concentration [Q] and multiplied by the dissociation constant Ka. More precisely, when the equation 14 is solved, the [H+]active is described with the equation 16.

2active a a agenerated

1[H ] [Q] + 4 [H] ( [Q])

2K K K+ = − − + −

(16)

The AEQ model shown above has been verified by the non-aqueous pH-measurement experiment using the model-resist-liquid system13, 14. It should be noted here that the AEQ model in equation 15 or 16 is the steady-state modeling based on the concept that the equilibrium is fully realized. With the steady-state AEQ model in equation 16, the trend of the acid dissociation ratio with a quencher can be evaluated. In Fig.4, the acid-dissociation percentage is plotted. As can be seen in Fig. 4, the acid-dissociation ratio decreases with the increase of quencher loading. The effect is significant for the acid which has a pKa of higher than about zero. When pKa of photoacid is more above zero, it seems necessary to consider the effect of the quencher in the modeling. In solid-CA resists the AEQ model needs to be extended to describe the transient states which include the change of the system during the PEB time. The extended model will be described in Section 3.2. 3.2 Implementation of AEQ model in UC Berkeley resist simulation tool (STORM) We extend the steady-state AEQ model to describe the transient state in solid CA resists for the implementation of the AEQ model in the UC Berkeley resist-simulation tool, STORM.

0.01

0.10

1.00

10.00

100.00

-10 -5 0 5 10

pK a

Dis

soci

ated

pho

toac

id (

%)

0mM0.1mM1mM10mM100mM500 mM

Quencher

@[HAun-neutralized ] = 10 mM

Que

nche

r

Figure 4: Change in acid dissociation ratio by quencher. (AEQ Model) When 10 mM un-neutralized acid [H+]active is

assumed to be necessary, the acid of pKa<~0 can be regarded as a full dissociated acid in the presence of quencher.


The acid generation is assumed to obey the equation 17, where [PAG]0, C and E are the initial loading concentration of PAG, a acid-generation-reaction-rate constant (Dill’s C-parameter) and the exposure dose, respectively.

0generated [PAG] ( ( [HA] 1 exp ))CE≅ − − (17)

The kinetics of [H+]active in the AEQ model are described by equation 18. The [H+] active increases with the dissociation of HA and decreases with the recombination with A− in the equation 18. The other terms for diffusion, neutralization and acid loss are the same used in earlier models12.

2activeactive n2 active d c active loss1 activeH

Diffusion Neutralization

[H ]= [H ] [Q][H ] [HA] [H ] [A ] [H ]D k k k k

t+

++ + + − +∂

∇ − + − −∂

Dissociation (equilibrium) Combination (equilibrium) Acid loss (18)

To calculate the [H+]active, it is necessary to describe the kinetics of [HA], [Q], and [A−]. The reaction-rate equations for these concentrations under the dissociation equilibrium are shown in equations 19, 20 and 21

2

HA n1 d c active loss2

Diffusion Neutralization Dissociation (equilibrium)

[HA]= [HA] [Q][HA] [HA] [H ] [A ] [HA]D k k k k

t+ −∂

∇ − − + −∂

Combination (equilibrium) Acid loss

(19)

2 +

Q n1 n2 active qloss

Diffusion Neutralization Neutralization Quencher loss

[Q]= [Q] [Q][HA] [Q][H ] [Q]D k k k

t

∂∇ − − −

∂ (20)

2

n1 d c active A loss

Diffusion Neutralization Dissociation (equilibrium) Combinat

[A ] A [Q][HA] [HA] [H ] [A ] [A ]

AD k k k k

t− −

−− + − −∂

= ∇ + + − −∂

ion (equilibrium) A loss−

(21)

Using the resulting active-acid concentration, the kinetics of the deprotection reaction can be calculated using equation 22 where kr, m and [P] are the deprotection-reaction rate constant, reaction order, and relative protected-group concentration, respectively.

+

active

[P]= [H ] [P] m

rkt

∂−

∂ (22)

The important change here from former models is that the dissociated photoacid (active photoacid) is used in equation 22, instead of the total generated photoacid. This is because in this AEQ model, it is assumed that only the active acid can catalyze the deprotection reactions in CA resists. As to the relationship between the above Full AEQ model (equation 18) and conventional FDQ model (equation 23) is considered.

2activeactive n active loss activeH

Diffusion Neutralization Aci

[H ]= [H ] [Q][H ] [H ]D k k

t+

++ + +∂

∇ − −∂

d loss

(23)


When the dissociation is full in equation 18, kc becomes zero and [HA] becomes zero. This means that the equilibrium terms disappear and equation 18 turns into equation 23. Consequently, the conventional FDQ model can be considered as a simplified case of AEQ model.

3.3 AEQ model verification in CA resists − Real-time-FTIR measurement during PEB 3.3.1 Deprotection reaction in CA resist with weaker photoacid We tested the hypothesis in the full AEQ model using model resists. The model acetal-KrF resists consist of a diazomethane PAG which generates non-fluorinated-weaker sulfonic acid. The acid generation reaction and deprotection reaction are shown in Figure 5.The quencher is added to the resists with five different loadings of 0, 10, 25, 40, 60 mol% to the PAG.

OH O

O

3367

PEB

OH OH

+

by-products

H+

CH3CHO

CH3OH

KrF Resin Reaction

PAG Reaction

S CN2

SO

O

O

OSO3H +

by-products

SO3 + H

SO2CH , N2 , etc.

Exposure

Catalyticreaction

Deprotection

Acid Generation

Acid Dissociation

KrF resist PEB 100°C, Q=0

-0.03

-0.02

-0.01

0

0.01

0.02

0.03

0.04

0.05

7008009001000110012001300

Wavenumber (cm-1)

Ab

sorb

ance

PEB 1 secPEB 10 secPEB 100 sec

Area 960-925 cm-1

Area 855-805 cm-1

C-O Stretch of ether(deprotection)

944 cm-1

Ref peak (C-H of p-disubstituted benzen:

no reaction) 830 cm-1

OH O

O

3367

C-O Stretchof phenol1275-1200

cm-1C-O Stretch

of ether1275-1020 cm-1

IR peak decay (after nomalization)

0 50 100 150 200PEB time (sec)

Nor

ma

lize

d IR

pea

k a

rea

(a

rb. u

nit)

944 cm-1 peak1050 cm-1 peak1119 cm-1 peak

The samedecay behavior

PEB 100ºC, Quencher 0%Exp. 4 mJ/cm2

Figure 5: The reactions in model KrF resist. Figure 6: FTIR Spectra for model KrF resist.

The deprotection behavior is measured using the real-time-FTIR measurement during PEB as shown in Fig. 6. The peaks which are attributed to the protection group are at 944 cm-1, 1050 cm-1, 1119 cm-1 as well as other wavenumbers in Fig.6. The peaks show almost the same decay behavior (inset of Fig.6). Among the several peaks, the peak at 944 cm-1 (in the calculation, the sum of the area from 925 to 960 cm-1) is monitored for the deprotection reaction because this peak is less affected by surrounding peaks than others. Figure 7 shows the summary of the decay behaviors of the acetal-protective group in the KrF resists with the five different quencher loadings. Each plot has three different KrF exposure doses, which are 0.5×, 1× and 2× clear dose E0 (except Q = 40, 60% where the doses are fixed at high exposure doses because the E0 could not observed in the pre-testing due to too high E0 or negative-type reactions). For the decay curves in Fig. 7, the fitting was done with the AEQ model. In the fitting using equations 17-22, the simplification of the parameter is made by assuming kn1=kn2=kn, kloss1=kloss2 = kloss, kloss1[H

+]active= kA-loss[A−] and klossq = 0.

Also in the fitting, the diffusion terms are not used because the wide area exposures were done for FTIR experiments, where the diffusion terms can be neglected. In the acetal resist, some of the deprotection reaction happens before the PEB due to the lower activation energy of the deprotection reaction. Therefore, we assume the reaction starts at −2.5 sec (2.5 sec before the start of FTIR measurement) from one certain level (0.75 in the graph value) which corresponds to the assumed value when no deprotection reaction happens. The most important thing here is that the same one set of parameters (Ka, kr, kn, C-parameter) is used for five of the graphs because these parameter should stay the same in the physical modeling. The fitting finishes when the total differences between the experimental values and fitting curves become the minimum after iterating the parameters. We produced similar fitting for the FDQ model as well.


0% Quencher 10% Quencher 25% Quencher

40% Quencher 60% Quencher

PEB Time (sec) PEB Time (sec) PEB Time (sec)

PEB Time (sec) PEB Time (sec)

Pro

tect

ion

Pro

tect

ion

Pro

tect

ion

Pro

tect

ion

Pro

tect

ion

Figure 7: AEQ fitting of deprotection reaction (KrF resist / 90ºC PEB). Figure 8 is the summary of the experimental and fitting results for the both the FDQ and AEQ models for 0 to 60 sec of PEB. The protection group concentrations are plotted with the photo-generated acid minus initial-quencher loading. Triangles are the experimental results and the circles are the simulation results with different quencher loadings. With FDQ model fitting, it is found that the fitting results and experimental results have big differences. Even after the repeated iterations of the fitting, the FDQ model failed to fit the data. On the other hand, the AEQ model can track data that is not a single-valued function of [H]generated − [Q] and a good fit is obtained.

FDQ model AEQ model

Pro

tect

ion

Gro

up C

onc.

(ar

b. U

nit)

[H]generated – [Q]0 Pro

tect

ion

Gro

up C

onc.

(ar

b. U

nit)

[H]generated – [Q]0

Figure 8: Prediction of protection group concentration at 60 sec of 90°C PEB (FDQ model vs. AEQ model). Triangles are experimental data and circles are the fitting results. [H]generated: photo generated acid, [Q]0: initial quencher loading.

FDQ model parameter: kr = 1.0, kl = 0.24, kn = 3.0, m = 0.8, C=0.03; AEQ model parameter: Ka = 0.03, kr = 3.6, kl = 0.275, kn = 4.0, m = 0.72, C=0.03.

Figure 9 gives the results for a higher PEB temperature of 110°C. At this temperature, we also used the one set of parameters to fit all the different quencher loadings. The AEQ model again gives very good fitting results while FDQ failed to fit them. The results of the other PEB temperature of 100°C also showed that the AEQ model gives a much better prediction.


: simulation

FDQ model AEQ model

Pro

tect

ion

Gro

up C

onc.

(ar

b. U

nit)


Pro

tect

ion

Gro

up C

onc

. (ar

b. U

nit)


Figure 9: Prediction of protection group concentration at 60 sec of 110°C PEB (FDQ model vs. AEQ model). Triangles are experimental data and circles are the fitting results.

FDQ model: kr = 1.0, kl = 0.12, kn = 7.0, m = 0.8, C=0.03; AEQ model: Ka = 0.03, kr = 15.4, kl = 0.22, kn = 18.0, m = 1.02, C=0.03.

From the parameter extracted from fitting, the activation energies of deprotection reaction and neutralization reaction are calculated using the Arrhenius plot as shown in Fig. 10. The calculated value was Ea = 0.87eV = 20.06 kcal/mol for the deprotection reaction. The reaction rate of neutralization also roughly follows the Arrhenius plot with Ea = 0.91eV =20.98 kcal/mol. This Arrhenius type behavior further supports the validity of the AEQ model. 3.3.2. Deprotection reaction in CA resist with stronger photoacid To clarify the quencher mechanisms in the case of super-strong acid generators, we also conducted similar experiments using a second model-resist system. It consists of methacrylate ArF resists with onium-salt-type PAG which releases super strong acid (Fig.1). The reactions are shown in Fig. 11. Quencher concentrations are also changed as 0, 10, 25, 40, 60 mol % to PAG loading. The deprotection behavior is measured using the real-time-FTIR measurement during PEB as shown in Fig. 12. The peak at 1101 cm-1 is attributed to the ester-protection group and used to monitor the deprotection reaction (in the calculation, the sum of the area from 1198 to 1109 cm-1 is used).

Figure 13 shows the time profile of the ester-protection group by FTIR during PEB. Each plot has three different ArF exposure doses, which are roughly 0.5×, 1× and 2× clear dose E0. The similar fitting is done for ArF resists. In the fitting using equations 17-22, the simplification of the parameter as was done in the section 3.3.2 was also performed (kn1=kn2=kn, kloss1=kloss2 = kloss, kloss1[H

+]active= kA-loss[A−] and klossq = 0). Also in the fitting, the diffusion terms are not

considered due to the wide exposure area. The absorbance of the deprotection group started from roughly one same value. This means the reactions before PEB in this ArF resists can be ignored due to its higher Ea of deprotection. Therefore, the same value (0.7 in the figures) is used as the start value for the fitting. Again, one set of parameters is used for all the fittings as was done for the model-KrF resists (weaker-photoacid system). When the super strong photoacid in ArF resists is used, it is possible to fit them with FDQ as shown in Fig. 14. Naturally, this means it could be fit by AEQ model because the AEQ model includes the FDQ model. The PEB temperature was changed from 90ºC to 140ºC with 10ºC steps and all the results could be fit by FDQ model s(as well as AEQ model). This means that for the fluorinated-super-strong acid in ArF resists can be considered as the full-dissociated acid . Thus the FDQ model is still a good model for resists with super-strong acids. It is interesting to observe that, under many exposure conditions of ArF-resist samples, the photoacid concentration is less than quencher, e.g. [H]generated −[Q]0 < 0. Even with this condition, the deprotection reactions still occur to some extent prior to the competition of the neutralization reaction. After the acid is neutralized, the reactions stop as shown in

Figure 10: Arrhenius plot for deprotection reaction and neutralization reaction.

1

10

100

0.0025 0.0026 0.0027 0.0028

1/T

Reaction Rate

kr

kn

Rea

ctio

n R

ate

Con

stan

t


Fig.13. It is also interesting to see in Fig.13 that the reaction speed for a lower-dose sample without quencher is actually slow but gradually continuing without an apparent stop to the reaction as is observed in the presence of the quencher.

S O S

O

OC4F9 C4F9SO3H +

by-products

C4F9SO3 + H

S S , etc.,

OO

O

O

OO

50 50

PEB OO

O

O

HOO

50 50by-products

H+

H2C+

ArF Resin Reaction

PAG Reaction

Deprotection

Acid Generation

Exposure

Catalyticreaction

Acid Dissociation

ArF resist PEB 130°C, Q=0%

-0.03

-0.02

-0.01

0

0.01

0.02

0.03

0.04

0.05

9001000110012001300

Wavenumber (cm-1)

Ab

so

rba

nce

PEB 1 sec

PEB 200 sec

OO

O

O

OO

50 50

C-O Stretch (deprotection)

Area 1098-1109 cm-1

PEB 130ºC, Quencher 0%, Exp. 0.8 mJ/cm2

Figure 11: The reactions in model ArF resist. Figure 12: FT-IR spectra for model ArF resist.

0% Quencher 10% Quencher 25% Quencher

40% Quencher 60% Quencher

PEB Time (sec) PEB Time (sec) PEB Time (sec)

PEB Time (sec) PEB Time (sec)

Pro

tect

ion

Pro

tect

ion

Pro

tect

ion

Pro

tect

ion

Pro

tect

ion

Figure 13: FDQ fitting of deprotection reaction (ArF resist / 120ºC PEB).

PEB 120ºC: kr = 2.2, kl = 0.0, kn = 6.0, m = 1.0, C=0.05. 3.4 Examples of quencher effect simulation based on the AEQ model The coding of the AEQ model is done in UC Berkeley’s STROM resist simulator20. Some examples of quencher effect simulation based on the AEQ model are shown in this section. The first example is the process-latitude calculation with AEQ model (Fig. 15). As can be seen in Fig. 15, with the small addition of the quencher, the process window becomes wider. And then it gradually decreases.



Pro

tect

ion

Gro

up C

onc.

(ar

b. U

nit)

Pro

tect

ion

Gro

up C

onc.

(ar

b. U

nit)


Figure 14: Prediction of protection at 60sec PEB (FDQ model); PEB 120ºC (left) and 90ºC (right) for 60 sec. Triangles are experimental data and circles are the fitting results. [H]generated: photo-generated acid, [Q]0: initial quencher loading,

PEB 120ºC: kr = 2.2, kl = 0.0, kn = 6.0, m = 1.0, C=0.05; PEB 90ºC: kr = 0.4, kl = 0.0, kn = 1.25, m = 0.84, C=0.05. As a second example, the environmental stability of the CA resist is also investigated by evaluating CD changes in the presence of external base contaminants as shown in Fig. 16. The plots are the ones under best dose and best focus for each quencher loading. As can be seen in Fig. 16, the higher the quencher loading, the more stable the CD in general. This result is consistent with the frequently observed results in real CA-resist processing. Several phenomena might be responsible for the process latitude and environmental stability. The first one is the acid-buffering mechanism modeled by AEQ model. The second one is resist-contrast change due to the increased exposure doses. The third one is the effect of diffusion of both quencher and acid on the neutralization points. To simulate the effects of the quencher on the fine pattern precisely is beyond this purpose because the correct quencher diffusivity in the model resist is not known. However, using this new AEQ model, all of those effects can be simulated.

Dose variation (%)

Def

ocus

(µm

)

0

20

40

60

80

100

0 0.2 0.4 0.6

Quencher loading(Qencher/PAG)

CD

cha

nge

(nm

)

Base =0.01Base = 0.02Base = 0.03

External basic contaminant level（/PAG）

Figure 15: Process latitude calculation (AEQ model).* Figure 16: Environmental stability calculation (AEQ model).*

*Simulation details (Fig. 15 and 16): Illumination constants: λ=248 nm, NA = 0.60, σout=0.75, σin = 0.40. Line width 160 nm (wafer scale) for an iso-line (just one line). Best doses are determined so that 160-nm line is printed on wafer at best defocus for each quencher loading. Defocus range of 0.0 µm ~ 0.50 µm and +/−10% dose variation are then simulated to generate a set of line width. Process latitude is determined by requiring that the linewidth falls into the range of 150 nm ~ 170 nm. Resist parameters: kr = 8.8, kl =

0.24, kn = 15.0, m = 0.88, kd = 20, kc = 632, Ka = 0.03. Assumed acid-diffusion constant: D0 = 4.5 × 10-4 µm2/sec, ω = -5. Assumed

quencher-diffusion constant: DQ = 1 × 10 -4 µm2/sec


4. SUMMARY An Acid-Equilibrium-Quencher (AEQ) model is proposed to predict the photoacid concentration in CA resists with quencher. The AEQ model introduces the concept of the acid-dissociation equilibrium into resist modeling. The AEQ model successfully predicts the FTIR-experimental results for the KrF resist with a weaker photoacid under different quencher loadings and different PEB temperatures while conventional Full-Dissociation-Quencher (FDQ) model cannot explain these FTIR measurements that typically show multiple values of deprotection for a given [H]generated − [Q] value. With the AEQ model, a variety of photoacid strength is covered in the resist modeling. Still, the FDQ model is found to be accurate for predicting the photoacid concentration for the model-ArF-resist with a super-strong acid. The FDQ model is considered as a simplified case of AEQ and computationally advantageous for the super-strong acids. The full AEQ model is included in UC Berkeley STORM resist simulator. Simulations using the AEQ model show that the process window and environmental stability are improved in the presence of quencher loadings. These results are consistent with the observed results in real CA-resist processing.

ACKNOWLEDGEMENT We would like to thank the Advanced Lithography group at the University of California, Berkeley and Lithography Department at IMEC for valuable discussion on the results. We also thank Ms. Tomoko Kai at Litho Tech Japan for her help in FTIR experiments. We are also grateful to Mr. Shuichi Inoue, Dr. Kunihiko Kasama and Mr. Hidenobu Miyamoto, Mr. Takayuki Uchiyama, Mr. Makoto Tominaga, Mr. Masashi Fujimoto, Mr. Mitsuharu Yamana, and Mr. Hideo Kobinata of NEC Electronics Corporation for their advice and encouragement in this work. This research is partially supported by SRC/DARPA under grant SRC 01_MC_460 and DARPA MDA972-01-1-0021.

REFERENCES 1. D. J. H. Funhoff, H. Binder, and R. Schwalm, Proc. SPIE, 1672 (1992) 46. 2. Y. Kawai, A. Otaka, J. Nakamura, A. Tanaka, and T. Matsuda, J. Photopolym. Sci. Tech., 8 (1995) 535. 3. K. Asakawa, T. Ushirogouchi, and M. Nakase, Proc. SPIE, 2438 (1995) 563. 4. S. Funato, Y. Kinoshita. T. Kudo, S. Masuda, H. Okazaki, M. Padmanaban, K. J. Przybilla, N. Suehiro, and G. Pawlowski, J.

Photopolym. Sci. Tech., 8 (1995) 543. 5. S. Saito, N. Kihara, T. Naito, M. Nakase, T. Nakasugi, and Y. Kato, Proc. SPIE, 3049 (1997) 659. 6. J. Hatakeyama, S. Nagura and T. Ishihara, J. Photopolym. Sci. Technol., Vol.13, (2000) 519. 7. D. Van Steenwinckel, M. Shimizu and J. H. Lammers, Advances in Imaging Materials and Processes, edited by H. Ito, P. R.

Varanasi, M. M. Khojasteh and R. Chen (2004) 149. 8. T. Kai, S. Nishiyama, A. Saitou, and T. Shimokawa, J. Photopolym. Sci. Tech., 16 (2003) 447. 9. R. L. Brainard, J. Cobb, and Charlotte A. Cutler, J. Photopolym. Sci. Technol., Vol.16, (2003) 401. 10. K. Hattori, J. Abe, and H. Fukuda, Proc. SPIE, 4691 (2002) 1243. 11. T. B. Michaelson, A. T. Jamieson, J. Byers, and C. G. Willson, Proc. SPIE, 4691 (2002) 1243 12. M. D. Smith, C. A. Mack, J. D. Byers, Proc. SPIE, 5977 (2004) 1482 13. S. Nagahara, M. Fujimoto, M. Yamana, and M. Tominaga, J. Photopolym. Sci. Technol., Vol.16, (2003) 351. 14. S. Nagahara, M. Fujimoto, M. Yamana, and M. Tominaga, Advances in Imaging Materials and Processes, edited by H. Ito, P. R.

Varanasi, M. M. Khojasteh and R. Chen (2004) 233. 15. T. Ohfuji, A. G. Timko, O. Naramasu and D. R. Stone, Proc. SPIE, 1925 (1993) 213; For example as a reference book, C. A.

Mack, “Inside PROLITH: A Compreshensive Guide to Optical Lithography Simulation”, Finle Technologies (Austin, TX: 1997) 16. K. Maeda, K. Nakano, S. Iwasa and E. Hasegawa, Photopolym. (2000) 161. 17. A. Sekiguchi, Y. Kono and Y. Sensu, J. Photopolym. Sci. Technol., Vol.16, (2003) 209 18. J. A. Dean, “LANGE'S HANDBOOK OF CHEMISTRY Fifteenth Edition” McGRAW-HILL INC., 8.63.; M. B. Smith, J. March, “MARCH’S ADVANCED ORGANIC CHEMISTRY−REACTIONS, MECHANISMS, AND STRUCTURE, 5th edition” A Wiley-Interscience Publication John Wiley & Sons, Inc. (2001) Ch.8 19. K. Izutsu, “Hisuiyoueki no denkikagaku”, Baifukan, (1995), p48, p53 (in Japanese); K. Izutsu, “Electrochemistry in Nonaqueous

Solutions”, Wiley-VCH, Weinheim, (2002); Nihon Kagakukai, “Kagaku Binran Kisohen II,” Maruzen, 322 (in Japanese); K. Izutsu, “Acid-Base Dissociation Constants in Dipolar Aprotic Solvents,” Blackwell Sci.

20. L. Yuan, M. Cheng, E. H. Croffie, and A. R. Neureuther, Proc. of SPIE, 4345 (2001) 922.


understanding quencher mechanisms by considering … · c litho tech japan, 2-6-6-201 namiki,...

Documents