10.1117/2.1200801.0975

Bubble and antibubble defectsin 193i lithographyYayi Wei

Air bubbles and topcoat particles act as microlenses, distorting patterns

projected on the resist. Fourth in a series.

The difference between 193nm immersion lithography and con-

ventional photolithography is that, in the former, water fills

the space between the front lens and the wafer. Replacing air

with water enhances the depth of focus and enables the nu-

merical aperture (NA) to be increased to values greater than

1.? However, physical and chemical interactions can occur be-

tween the water and the resist/topcoat stack. This leads to

immersion-related defects, of which the major types are bub-

ble and ‘anti-bubble’ types, as well as water marks, particles,

and microbridges.?, ?, ? These are observed on almost all 193nm

immersion-processed wafers and account for more than 90% of

the total defects. This article describes these bubble and anti-

bubble defects—other kinds are discussed in a later article.

Bubble defects

Air in the immersion water causes bubbles: an air bubble reflects

or refracts the exposure light and thereby distorts the local image

projected on the resist. Usually, bubble defects are circular and

have diameters ranging from one to several hundred microme-

ters. The impact of bubbles on the image depends on their size

and distance from the resist surface. It is generally accepted from

various theoretical and experimental results?, ? that free-floating

air bubbles in the water have less impact because they are out

of the imaging focus. Furthermore, floating bubbles have very

short lifetimes.? The real threat comes from those attached to the

surface.

An air bubble on the resist surface works like a small lens

during exposure: it diverges the exposure light and distorts the

pattern underneath and around itself (see Figure 1). After de-

velopment, the distorted pattern forms a circular defect. The ef-

fects of bubbles on imaging were simulated by several differ-

ent groups.?, ?, ? In particular, work by Peter De Bisschop, An-

dreas Erdmann, and Andreas Rathsfeld helps us to recognize

bubble defects and differentiate them from other circular defects

Figure 1. Air bubble on the resist surface works like a lens during ex-

posure.

in experiments.? According to their simulations, the most notice-

able effect caused by a bubble is the strong reduction in image

intensity at the bubble location on the resist surface, whereas

at certain locations around the bubble the image intensity in-

creases. The bubble also seems to magnify the incident line and

space pattern.

Figure 2(a) shows a typical bubble defect on a line and space

pattern. It is circular, with a diameter of about 9μm. As predicted

by the simulation, the bubble causes under-exposure in the lo-

cal area. In the central region of the defect, image modulation is

visible, but the pattern is magnified. The pitch of the pattern is

220nm in the well-patterned area as designed; however, in center

of the defect, it is measured as ∼308nm. The magnification ratio

is ∼1.4. Moving from the center to the edge, the imaging modu-

lation gets weaker and weaker, finally completely disappearing

at the edge of the bubble.

In general, the area that surrounds the bubble defect is over-

exposed. Looking closely, we can observe intensity fringes in

this area. Figure 2(b) shows an amplified scanning electron mi-

croscope (SEM) image taken at the edge of another bubble de-

Continued on next page

10.1117/2.1200801.0975 Page 2/3

Figure 2. (a) Scanning electron microscope (SEM) image of a big bub-

ble defect about 9μm across. The defect is symmetric in both X and Y

directions. In the circled areas on opposite edges of the defect, residuals

of the resist lines have the same shape. (b) SEM image of the edge of

a bubble defect, where the resist is completely developed. (c) SEM im-

age taken at the edge of the bubble defect at low magnification shows a

fringe effect. The pitch of the line and space pattern is 220nm.

fect. In the surrounding area, the resist is completely developed,

making the edge of the bubble defect very sharp and clear. A

little away from the edge, resist lines appear, which indicates a

lower local dose. Further away, the resist lines gets thinner—as

though pinched—indicating a higher local dose. The variation in

resist linewidth shows spatial periodic behavior corresponding

to the dose modulation. Further away from the defect, the dose

modulation weakens and finally becomes imperceptible. Figure

2(c) shows a low-magnification SEM image. The fringe pattern

caused by the dose modulation can be better observed outside of

the bubble defect. Obviously, the dose modulation derives from

exposure light reflection and refraction by the bubble.

From our experience with defect inspection, we find that big

bubble defects with diameters greater than 2μm are relatively

easy to identify. One or more of the features described above are

usually visible. The imaging modulation is clear, leaving ample

information on the resist pattern. This is because reflection and

refraction of the exposure light by a big bubble is stronger than

that of a small bubble. The challenge is to identify small bubble

defects for which the features described in previous paragraphs

are not clearly visible. For example, a defect caused by a roughly

100nm-diameter bubble may look just like a small bridge con-

necting two resist lines.

The source of bubbles in immersion water is still under in-

vestigation. It seems that they are most likely trapped by a non-

optimized exposure head design.? Also, surface topography has

proven to have minor effects on bubble formation.? Therefore,

better exposure head design appears to be the key to reducing

these defects. Recognizing this, scanner vendors have produced

new designs and now claim ‘no bubble’ processes.?, ? Complete

degassing of the immersion water and use of low outgassing re-

sists are helpful for further reducing bubbles.

Figure 3. (a) A transparent particle on the resist surface works like a

lens during exposure. In the circled areas, light is reflected, rather than

focused. (b) A sketched diagram shows the dose distribution caused by

the particle.

Anti-bubble defects

The ‘microlens effect’ can also be caused by a transparent particle

or bump, for example, formed by topcoat particles. Figure 3(a)

shows a diagram of a spherical topcoat particle attached to the

surface. The refractive index of the topcoat is ∼1.55, which is

larger than that of water. The transparent particle focuses the

exposure light (the opposite effect to that of an air bubble). The

center area underneath the particle tends to get overexposed, as

indicated by the gray scale in Figure 3(b). Near the edge of the

particle, the exposure light is mostly reflected due to the high

incident angle at the interface of water and particle. A heavily

underexposed ring is expected at the edge of anti-bubble defects.

This was first observed in a deep-UV-processed wafer. It was

attributed to the resist particles.?

Figure ??(a) shows a SEM image of an anti-bubble defect. It is

circular, with a diameter of about 1.7μm. As predicted, we ob-

serve the underexposed ring surrounding the defect. In the cen-

ter of the defect the resist lines are slightly narrower than near

the edge, which indicates convergence of the exposure light. The

resist lines in this area shift towards the center, and their pitch is

smaller than in the unaffected area. This is due to the demagni-

fying effect of the particle ‘lens’. The shape of the particle de-

fines the extent of the demagnification effect. A particle with

high curvature generates a more pronounced demagnification

effect and underexposure ring. Figure ??(b) shows another anti-

bubble defect. Although it is smaller than that in Figure ??(a),

the pattern demagnification and under-exposure ring are very

pronounced. The two resist lines in the defect area are hardly re-

solved. As suggested by Figure ??(b), the shapes of anti-bubble

defects are not necessary perfect circles. They can also be oval.

The source of the transparent particles is attributed mainly to

the topcoat material and a non-optimized coating process. After

Continued on next page

10.1117/2.1200801.0975 Page 3/3

Figure 4. SEM images of anti-bubble defects: (a) a typical one with a

diameter of about 1.7μm; (b) a small oval-shaped defect. The pitch of

the line and space pattern is 220nm in both images.

the topcoat solution is stored for a long time, chemical segrega-

tions can generate particles and clusters. When these are deliv-

ered to the wafer surface, they form bumps embedded in the

topcoat film. During the coating process, excessive topcoat or re-

sist spins off the wafer and lands on the wall of the coating bowl.

The topcoat sticking to the wall dries and forms particles. In the

next wafer process, the high spin speed will create a vacuum on

the wafer surface and suck at the particles.

Water uptake has been demonstrated as another way to form

bumps in the resist and topcoat.? Due to surface imperfections—

for example, the existence of pinholes—water may penetrate

through the topcoat and reach the interface between it and

the resist. The water swells the topcoat film and forms circu-

lar bumps. Filtering the topcoat solution before applying it to

the wafer surface and cleaning the coating bowl frequently have

proven to effectively reduce anti-bubble defects. Improving the

coating uniformity also helps to reduce defect counts.

Bubble defects are unique to the immersion process, while

anti-bubble defects are also observed on dry-processed wafers,

but the immersion process provides additional sources of par-

ticles. The mechanism by which immersion-related defects are

formed may involve the scanner, the materials (including bot-

tom anti-reflection coating, resist, and topcoat), and the process

setup. The defects we addressed are introduced by the scanner.

The author thanks Dr. Stefan Brandl of Qimonda AG for his collabora-

tion.

Author Information

Yayi Wei

AZ Electronic Materials

Somerville, NJ

References

1. B. W. Smith, Immersion Optical Lithography at 193nm, Future Fab Int’l 15, July2003.2. U. Okoroanyanwu, J. Kye, N. Yamamoto, and K. Cummings, Defectivity in waterimmersion lithography, Microlithography World, Nov. 2005.3. S. Brandl, R. Housley, P. Benson, F. Goodwin, Yayi Wei, C. Robinson, D. Corliss,and K. Cummings, Immersion Defect Studies, SEMATECH 2nd 193nm ImmersionSymp., September. 14 2005. Brugge, Belgium4. M. Kocsis, D. Van Den Heuvel, R. Gronheid, M. Maenhoudt, D. Vangoidsen-hoven, G. Wells, N. Stepanenko, M. Benndorf, H. W. Kim, S. Kishimura, M. Ercken,F. Van Roey, S. O’Brien, W. Fyen, P. Foubert, R. Moerman, and B. Streefkerk, Im-mersion specific defect mechanism: Findings and recommendations for their control, Proc.SPIE 6153, 2006. doi:10.1117/12.6604325. Y. Fan, N. Lafferty, A. Bourov, L. Zavyalova, and B. W. Smith, Study of Air BubbleInduced Light Scattering Effect On Image Quality in 193 nm Immersion Lithography,Proc. SPIE 5377, pp. 477–486, 2004. doi:10.1117/12.5372556. C.H. Lin and L. A. Wang, Simulation of air bubble scattering effects in 193nm immer-sion interferometric lithography, J. Vac. Sci. Technol. B 23 (6), pp. 2684–2693, 2005.7. A. Kawai, T. Niiyama, H. Endo, and M. Yamanaka, Adhesion and removalof micro bubbles for immersion lithography, Proc. SPIE 6153, pp. 61531S–1, 2006.doi:10.1117/12.6554448. P. De Bisschop, A. Erdmann, and A. Rathsfeld, Simulation of the effect of a resist-surface bound air bubble on imaging in immersion lithography, Proc. SPIE 5754, pp. 243–253, 2005. doi:10.1117/12.5997929. Yayi Wei, K. Petrillo, et al., Selection and evaluation of developer-soluble topcoat for193 nm immersion lithography, Proc. SPIE 6153, doi:10.1117/12.655725 2006.10. M. Kocsis, P. De Bisschop, et al., A methodology for the characterization of topogra-phy induced immersion bubble defects, Proc. SPIE 5754, 2005. doi:10.1117/12.59932911. B. Streefkerk, C. Wagner, et al., Advancements in system technology for the im-mersion lithography era, SEMATECH 2nd 193nm Immersion Symp., Sept.14 2005.Brugge, Belgium12. K. Nakano, S. Nagaoka, and S. Owa, Defectivity data taken with a full-field immer-sion exposure tool, SEMATECH 2nd 193nm Immersion Symp., Sept.14 2005. Brugge,Belgium13. S.-F. Tsai, C.-Y. Chen, C.-C. Chang, T.-W. Huang, H.-Y. Gao, and C.-Y. Ku,Microlens-induced pattern defect in DUV resist , Proc. SPIE 5376, pp. 1149–1156, 2004.doi:10.1117/12.53448214. S. Kanna, H. Inabe, K. Yamamoto, S. Tarutani, H. Kanda, W. Kenji, K. Kodama,and K. Shitabatake, Non topcoat resist design for immersion defectivity reduction, Int’lSymp. Immersion Lithography, Oct. 2006. Kyoto, Japan

c© 2008 SPIE