Post Exposure Baking Temperature Effect on Resist Profile

with
Bottom Anti-reflective Coating

Chang-Ming Dai, Chin-Lung Lin, Shi-Chang Tai, James E. Lamb JJJ* and M. Iida#

Submicron Technology Group, S000, ERSOIITRI. 195-4, Sec. 4 Clmng-Hsing Rd., Chutung, Hsinchu 3 10, Taiwan
* Brewer
Science, Inc. , Rolla, Missouri USA

# Nissan Chemical md., ltd., Japan

AB S TRACT
Thin film interference effects in photoresists are the most serious issues for device production in sub-half micron
patterning. These effects change the fraction of the energy available for photoresist absorption and subsequently cause
serious line width fluctuation. One of the most realistic candidate from the point of view for the device mass production is
the development of a organic bottom anti-reflective coating (ARC). Because organic ARC has high absorption
characteristics of incident light, the standing wave in photoresist could be diminished. However, up to now, organic ARC
still has some issue in resist profile (i.e. footing). Usually, the processes of organic ARC is optimized by tuning its
thickness. Very little effort has been done on the optimization of post exposure bake. In this paper, the effects of post
exposure bake temperature on resist profile and process windows, including energy latitude and focus latitude, of single
line and dense line features will be discussed. The swing ratio isimproved from 15.5% for the case without BARC to 1.2
% for the case with XHRi BARC of thickness 10 1OA (at top of reflectance curve). In term of PEB effect, PEB
temperature of 90°C is better than other conditions except it still has slightly standing wave. With BARC, it can not only
improve DOF process window from nothing to O.8.tm of O.35.tm dense and single line features but also reduce its
proximity effect. Comparing the resist profile between BARC at top and bottom of reflective swing curve, the footing is
much severe for the case ofBARC with thickness at bottom ofswing curve in HRi or XIi material.

1. INTRODUCTION
As device dimensions shrink to the sub-half-micron range, critical dimension (CD) control becomes increasingly
important, especially for polysilicon gate (polygate). In conventional or modified local oxidation of silicon (LOCOS), the
notching and necking effect on processing layers, caused by the light reflected off rough topography, usually makes the
CD control become a tough issues. The solutions to the topography problem can be divided into two methods. One is to
smooth out the topography by using chemicomechanical polishing (CMP). The other one is the use of an antireflection
coating (ARC) layer which reduces or minimizes the magnitude of reflected light and thus allows CD in photoresist to be
well controlled. In general the ARC material can be grouped into two types: the first one is organic high-absorption
material (e.g., die resist,1) top-ARC2) and bottom ARC3)) and the other one is inorganic material (e.g., TiN, TiW and
SiOxNy:H46)). In case of organic ARC, a dyed resist is the simplest solution, but dyeing is often accompanied by a loss
in resolution and process latitude. Top ARC can reduce the swing effect, but does not reduce notching problems. Bottom
ARC (BARC) can eliminate both swing and notching problems and the process is compatible with resist process.
Though inorganic BARC could provide acceptable swing reduction and notching protection, they either have to be
stripped from the substrate being patterned by a separated plasma etch step or to remain as an integrated part of the device.
However, in case of organic BARC material, once the patterning process is complete the organic materials are removed
from the wafer by the same process used to remove photoresist.

598 1 SP!E Vol. 2726 O-8194-2102-2/96/$6.OO

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 06/22/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx

 For resolution down to O.35tm, the linewidth swing due to thin film interference uses up the complete linewidth
budget, leaving no latitude for other factors such as Iinewidth variation from stepper lens, dose consistency, depth of
focus, resist process and the develop process. Anti-reflective coating (ARC) provide a number of well documented
benefits to the lithographic processes [7-8]. These include standing wave reduction, reflective notching reduction,
elimination of back scattered light from grainy substrates and most recently, the elimination of substrate chemical reaction
with chemical amplified resist system. However, organic BARC has drawbacks of degradation of resist profile and foot
formation at photoresist interfce. From previous report, the footing (or tailing) at the interface of resist and BARC is
usually a function of the BARC thickness and its refractive index, if BARC has good chemical compatibility with resist.
Therefore, the photoresist profile on BARC is very important before it can be used in IC manufacturing. In this paper, the
photoresist profile on BARC (Brewer science newest i-line BARC XHRi-1 1 and FIRi-1 I were used in this experiment)
with thickness at top and bottom of reflectance curve were investigated.

As linewidths continue to decrease, the size of the standing waves relative to the feature size will increase, and the
difficulty of linewidth control in fine-line lithography will become severe. As we knew from Wakers paper [9], the
standing wave effect could be diminished by post exposure bake and he also proposed that the mechanism of standing
wave reduction was thermally-induced diffusion of unexposed photoactive compound (PAC) from lightly exposed strata
into the highly exposed strata. The homogenization of the unexposed PAC by diffusion would help to equilibrate the
dissolution rates of the alternating strata. In general, diffusion of a mobile species occurs during time t at which the
exposed photoresist is at elevated temperature T and is characterized by the diffusion length, s=(2Dt)"2. Typically
diffusion lengths is in the range of 0.03 im to 0. lpiii for current i-line resist with bake temperature of 1 10°C and baking
time 90 sec [10]. Because of the importance of the standing wave on sub-half-micron lithographic process control, the
resist profile and process window under different PEB temperature were also studied.

2. EXPERIMENT
The exposure tool is PAS5000/50 produced by ASM Lithography with NA=O.48 and degree of coherence of 0.62.
The photoresist is SUMITOMO PFI-38A9 and its thickness is 0.85 tm (normal case). Before resist coating,, wafers were
baked for dehydration at 250°C for 45sec and BARC was applied on wafers. The BARC used in this studies are Brewer
science model XHRi-1 I and HRi-1 1. In the following, BARC baking temperature within 100-200 °C were tried to
optimize the BARC baking temperature. After that, resist coating and soft-bake at 90°C for 60 sec were done. To study
the PEB effect on photoresist profile, baking temperature at 0, 60, 90 and 110°C for 90sec were used. Finally, wafers
were developed by SD-i developer (TMAH 2.38°/) using spray development mode. Process window and latitudes were
studied by using scanning electron microscope (SEM). In summary, we will study the photoresist profile with BARC at
different baking temperature, the PEB temperature effect on standing wave reduction and process window of dense and
single line features and the comparison of photoresist profile between FIRi and XHRi with BARC thickness both at top
and bottom of reflectance curve.

3. RESULTS AND DISCUSSION

Organic BARC materials are supplied as low viscosiiy liquid solutions by spin coat processing. Once coated these
materials produce a highly absorbent film at the exposure wavelengths used for resist patterning. The BARC material
typically has a refractive index that matches the photoresist at exposure wavelengths. The optical properties of BARC
HRi include absorbent of 3.615 per micron at 36mm. The refractive index is n1.78 and k0.23. While for i, its
absorbent is 0.723 (1300A) at 365nm and its refractive index is n=I.72 and k=0.21. To realize BARC capability, the
reflectance (measured by Nanospec 216) versus the XI-[Ri BARC thickness was studied and is shown in Fig. 1. This
figure indicates that the BARC thickness around 1 500A has the lowest reflectance (denoted as bottom of reflectance curve)
and thickness around 1OIOA has the highest reflectance (denoted as top of reflectance curve). Hereafter, two thickness of
BARC corresponding top and bottom of reflectance curve were adopted in this experiment.

SPIE Vol. 2726 / 599

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 06/22/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx

 After the determination of BARC thickness, the swing curve of PF138A9 on BARC were studied. The swing curve of
resist with XHRi BARC at 10 IOA (top) and 1500A (bottom), Fi at 1500A (bottom) and without BARC is shown in
Fig.2. The swing ratio is about 15.5% for the case without BARC and reduced to 5.8% with I-ni BARC at bottom. While
for XHRI BARC, the swing ratio is reduced to 5.0 % and 1.2 % for BARC with thickness at bottom and top, respectively.
For the process control point of view, BAR C thickness at top of reflectance curve is much better than that at bottom.
Comparing these two BARC materials, new material XHRi definitely provides better performance. In term of photoresist
thickness, we chose O.85un (middle of swing curve) rather than O.87tm (top of swing curve without BARC) because
O.85.tm is located at the middle of swing curve.

3. 1 BARC baking effect

After the BARC ofXHRi-1 1 was dispensed on wafers with thickness of IOIOA (top of reflectance curve), the wafers
were baked at temperature from 1 10 to 220°C . Then photoresist with thickness of O.85im were coater on BARC followed
by a 1 10°C for 90 sec bake. The process window and resist profile were characterized. Fig.3 shows the resist profiles of
O.4im dense and single lines with BARC baked at temperature of 1 10 to 220°C . The exposure energy is written on the
right ofevery photographs. It was found that the resist profile is better for the baking temperature in the range of 155-185
OC . The resist profile is getting poor as the baking temperature of BARC above or below this range. In the case of
temperature higher than 185°C, the prominent footing phenomenon occurred. While for case ofbaking temperature lower
than 155 °C , the resist profile becomes worse than 220 °C case and optimal exposure energy increase tremendously. In
terms of energy latitude at various BARC baking temperature, the results are summarized in Table 1. It is very clear that
the optimal baking temperature of Ri-1 1 BARC is within 1 55-185 °C . The proximity effect is smaller comparing with
the results without BARC. The energy latitude and optimal exposure latitude for single and dense feature is nearly the
same. It is very curious that as the baking temperature of BARC below 155°C ,there is no process window at 13000 but
with process window at 1 10 °C . Besides the process performance in profile, we also studied the stability of BARC at
different baking temperature. Chemically, the coated film muse be inert to photoresist solvent system to prevent
interfacial mixing during the photoresist coating process. Lower the BARC baking temperature will degrade its
lithographic performance due to insufficient to remove solvent in BARC. This may also make the adhesion between
BARC and resist poor. Higher baking temperature of BARC will decompose the BARC and many pin holes occurred on
the surface of BARC.

Table I The process window of O.4.tm dense and single lines with BARC baked at temperature of 1 10 to 220 °C . The
optimal baking temperature of XHR1-1 1 BARC is within 1 55-1 85CC . The thickness of BARC is 10 IOA.

Baking temperature (°C) 1 10 130 155 170 185 200 220

320 3 10 200 200 200 190 190

Eop dense line (mj/cm2)

290 290 200 1 90 200 190 200

Eop single line (mj/crn2)

ELdenseline(mj/cin2) NA

EL single line (mj/crn2) 10 NA 25 25 25

note: Eop: optimal exposure energy NA: not available

EL: energy latitude

600 I SP!E Vol. 2726

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 06/22/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx

 3.2 PEB effect ofXHRi BARC baking at 165°C

In this study, the PEB temperature effect on photoresist profile was studied. The XHRi BARC with thickness of
1O1OA was baked at temperature of 165°C . After wafers had been exposed, four different PEB baking temperatures were
conducted to realized the photoresist profile and process window. Figure 4 shows the photoresist profile of dense and
single line ofO.35im features with PEB temperature at 110. 90, 60 and 0°C for 90 sec. All the resist profile is with best
focus condition. As you can see from these figures. the standing wave of dense and single line features is eliminated for
I 10 °C case, however, it still exists for other cases. Comparing the profile among 90, 60 and 0 °C cases, standing wave
exists for single line, while for dense line (especially the center one) the standing wave becomes stronger as the PEB
temperature decreases. To realize the overall process of various PEB conditions, their results are summarized in Table 2.
As you can see in this table, all the evaluated process items for 90°C case is better or the same as other conditions except it
still has slightly standing wave. We also like to point out that DOF process window of O.35.tm dense and single line
features with BARC is improved to O.8tm from nothing of the case without BARC. Comparing these performances in
every PEB temperature, 90 °C is much better than others. Besides the resist profile at best focus, Fig.5 summarize the
resist profile ofO.35p.m dense and single line features at pin defocus. In these figures, we also can conclude that 90
°C for 90 sec PEB is the best condition for Xl-IRi BARC with thickness of 10 IOA because it had larger DOF process
window.

Table 2 The influence of PEB temperature on process performance of 0.35jnn dense and single line features with BARC
baked at temperature of 165°C . The thickness of BARC is 10 1 OA.

items\PEB conditions 1 10°C/9Osec 90°C/90sec 60°C/90sec 0 w/o BARC/1 10°C

EL dense line (mj/cm2) 10 12 10

EL single line(mj/cm2) 15 15 15 10 10

DOF dense line (tm) 0.6 0.8 0.6 0.6 0

DOFsinglelineQ.trn) 1.0 1.0 0.8 0.6 0.6

Standing wave (single line) no slightly slightly slightly no

Standing wave (dense line) no slightly clearly clearly no

Note PEB: post exposure bake DOF: depth of focus

EL: energy latitude

3.3 PEB dependence ofphotoresist profile on XHRi and HRi BARC baking at 200°C

In this section, the influence of PEB temperature on photoresist profile of 0.40tm dense and single line features
with XHRi and HRi BARC baked at 200 °C was discussed. Both top (1 IOOA) and bottom (1600A) of reflective swing
curve of XHRi and HRi BARC material were studied. The process condition is the same as section 3.2, except the baking
temperature of BARC was changed to 200 °C . All the photoresist profile at different conditions at best focus were
summarized in Fig. 6. Comparing the resist profile between BARC at top and bottom of reflective swing curve, the
footing is more severe for the case of BARC with thickness at bottom of swing curve in HRi or IRi material. This is
because of a node point located at the interface of BARC and photoresist which decrease the exposure of the resists.
Regarding the PEB temperature effect, the standing wave of resist profile is nearly eliminated for BARC with thickness at

SPIE Vol. 2726 I 601

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 06/22/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx

 bottom whichever the PEB condition was used. But for case of BARC thickness at top, eveiy cases still have slightly
standing wave except temperature at 1 10°C . This result is consistent with that of section 3.2.

3 .4 The necking effect improvement of XHRi BARC

Summarizing all the results mentioned in previous section, XHRi BARC at thickness 10 1OA (at top of reflectance
curve) which was baked at 165 °C and photoresist thickness of O.85im with PEB at 90 °C for 9Osec were chosen as a
lithographic processes for O.25p.m polysilicon gate patterning. O.25.tin CMOS device with LOCOS isolation process
usually leaves 1000-2000 step between field oxide and active area if processes without planarization. This step will
generate severe necking or notching in photoresist patterns. With the optimal processes for BARC, the necking and
notching in step of LOCOS is eliminated as shown Figs.7(a) and 7(b). Fig.7(a) shows the test key of O.25.tm polysilicon
gate just across LOCOS step and Fig. 7(b) shows its photoresist profile. It is very clear that this process is suitable for
O.25m device studies. Figs.8 (a) and (b) shows the test key pattern after dry etch processes. Its CD variation across
LOCOS step still gets the demand of O.25pin CMOS device patterning requirements.

4. CONCLUSION
In conclusion, the effects of post exposure bake temperature on resist profile and process windows, including energy
latitude and focus latitude, of single line and dense line features were studied. From these discussion, organic BARC
processes were optimized to meet our 0.25 tm CMOS device requirement. The swing ratio is improved from 15.5% for
the case without BARC to 1.2 % for the case with XHRi BARC of thickness 1O1OA (at top of reflectance curve).
Regarding the baking temperature of BARC, the resist profile and process window are better for the baking temperature
in the range of 155-185 °C . In term of PEB effect, PEB temperature of 90°C is better than other conditions except it still
has slightly standing wave. With BARC, it can not only improve DOF process window from nothing to O.8tm of O.35im
dense and single line features but also reduce its proximity effect. Comparing the resist profile between BARC at top and
bottom of reflective swing curve, the footing is iuuch severe for the case of BARC with thickness at bottom of reflectance
curve in }{Ri or XHRi material. But for case of BARC thickness at top, every cases still have slightly standing wave
except temperature at 1 10°C . Based on experimental results. XHRi BARC at thickness 10 1OA (at top of reflectance curve)
baked at 165°C and photoresist thickness of O.85.Lm with PEB at 90°C for 9Osec were chosen as a lithographic processes
for O.25p.m polysilicon gate patterning. With this optimal processes, the necking and notching in step of LOCOS is
eliminated and suitable for sub-half-micron lithographic processes.
5. REFERENCES
[1J 5.5. Miura, CF. Lyons and TA. Brunner: Proc. SPIE 1674, pp. 147, 1992

[21 T.Tanaka, N.Hasegawa, H.Shiraishi and S. Okazaki: J. Electrochem. Soc. 137, pp.3900, 1990

[3] M.W.Horn: Solid State Technol.. Nov. 57, 1991

[4] T.Ogawa, M. Kimura, T. Gotyo, Y. Toino and T. Tsumori: Proc. SPIE 1927, pp.267, 1993

[5J C.M.Dai and S.HLiu, Jpn, J. Appl. Phys., Vol.34. pp.6611, 1995
[61 T.Ogawa, H.Nakano, T.Gocho and T. Tsumori: Proc. SPIE 2197, pp.772, 1994

[7] M. lida and J. E. Lamb III, SEMI technical Proceedings. SEMICON Japan, pp.3 59, 1994

[8] J. Sturtevant and B. Roman, Microlithography world. Autumn, pp. 13. 1995

[7] E.J. Walker, IEEE Trans Elect. Devices. ED-22, pp.464-466. 1975

[8] P.Trefoiias HI, B.K.Daniels, M.J.Eller and A.Zampini. SPIE, vol.920, pp.203, 1988

602 ISPIE Vol. 2726

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 06/22/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx

 Reflectance of XHRi-11 BARC versus its thickness 100

.5
40
90
—.—XHRi(1510
30 —.—365nrn —o—XHRi(IOIOA)
20 85 —--PRonly
—.—366nrn
80 —s— HRi (1 500A)
10 —a—367nrn
0 75
1031 1103 1193 1309 1472 1775 2382
70
BARC thickness (A) 8527 8680 8960 9194 9413 9628
Photoresist thickness (A)

Fig. 1 The reflectance of XHRi BARC within

thickness of 1000 to 2300A Fig. 2 The swing curve of photoresist on BARC
material with thickness at top or bottom of
reflectance curve.

""

(a) (b)

Fig. 7 The SEM photographs of lithographic patterns of O.25nii CMOS device polysilicon gate across LOCOS 2000A
step. Fig.7(a) shows the top view SEM photograph oftest key of O.25jtm polysilicon gatejust across LOCOS step and Fig.
7(b) shows its photoresist profile. Lithographic processes are with XHRi BARC 10 1OA baked at 165 °C and photoresist
0.85 rim.

•i Fig. 8 The SEM photographs of 0.25gm

CMOS device polysilicon gate across
LOCOS 2000A step after dry etching
processes. Fig.8(a) shows the top view
SEM photograph of test key of 0.25tm
polysilicon gate just across LOCOS step
and Fig. 8(b) shows polysilicon gate
__________ _____ J profile.
(a) (b)

SPIE Vol. 2726 / 603

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 06/22/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx

 Fig.3 The resist profile of O.4.tm dense and single line features with BARC baking at temperature of
1 1O—22O°C

3aking 0.4tm dense line 0.4jtrn single line

emperature ______
100°C

130°C

155°C

170°C

185°C

200°C

220°C

604 ISPIE Vol. 2726

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 06/22/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx

 Fig. 4 The photoresist profiles of 0.35 j.m dense and single line features with PEB temperature of 0, 60,
90 and 110°C for 90 sec. All the resist profiles are with best focus condition and with BARC baking
at 165°C.

PEB O.35trnL/S O.35tmSL

tempera4 _________ ________
110°C
I9Osec

90°C/90sec

60°C/90sec

0°C

SPIE Vol. 2726 / 605

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 06/22/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx

 0

Fig.5 The photoresistprofilesof 0.35tm dense and single features with PEB temperature at 0, 60, 90 and 110°C for 90 sec. The BARC baking
temperature is at 165°C and exposedat defocus.

I
I Ii

90°C/90s

Th

60°C/90s

0°C

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 06/22/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx

 Fig. 6 ThePEB temerature effect on the photoresist profiles of X1-lRi and HRi BARC materials with thickness at top and bottom of reflectancecurve.
BARC baking temperatureis 200°C and PEB baking temperature is 0, 60, 110°C for each cases.

PEB XHIRi-1600A (bottom) XHIRi-1 OOA (top) HRi-1600A(bottom) HRi-1 100A (top)
temper
ature
110°C

6000 1

0°C

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 06/22/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx

F\J
N
C'
C'
0
N