alternating phase shift mask (psm) phase defect … phase shift mask (psm) phase defect printability...

Alternating Phase Shift Mask (PSM) Phase Defect Printabilityfor 130 nm and 100 nm KrF Lithography

International SEMATECHTechnology Transfer # 00033918A-TR

© 2000 International SEMATECH, Inc.

SEMATECH and the SEMATECH logo are registered service marks of SEMATECH, Inc.International SEMATECH and the International SEMATECH logo are registered service marks

of International SEMATECH, Inc., a wholly-owned subsidiary of SEMATECH, Inc.

Product names and company names used in this publication are for identification purposes only and may be trademarks or service marks of their respective companies.

Alternating Phase Shift Mask (PSM) Phase Defect Printability for130 nm and 100 nm KrF Lithography

Technology Transfer # 00033918A-TRInternational SEMATECH

May 19, 2000

Abstract: This document covers a study to determine the maximum non-printable phase defects for 130 nmand 100 nm linewidths by using a KrF deep ultraviolet (DUV) scanner. It is part of an effort toextend optical lithography to the sub-wavelength region for very low K1 patterning processes.This study used resist simulations verified by wafer printing results to predict printable defects fordenser patterns. It also tested mechanical repair tools for phase bump defects and comparedelectromagnetic field (EMF) three-dimensional simulation with commercialized two-dimensionaltools for phase defect printability.

Keywords: Atomic Force Microscopy, Critical Dimension, Deep Ultraviolet Lithography, Phase ShiftingMasks, Defect Detection, Pattern Defects, Scanning Electron Microscopy

Authors: Juhwan Kim, Wang-Pen Mo, Ron Gordon, Alvina Williams

Approvals: Young-Sik Kim, Project ManagerDan McGowan, Technical Information Transfer Team Leader

iii

International SEMATECH Technology Transfer # 00033918A-TR

Table of Contents

1 EXECUTIVE SUMMARY....................................................................................................... 1

2 INTRODUCTION..................................................................................................................... 1

3 EXPERIMENTAL PROCEDURES ......................................................................................... 23.1 Overall Experimental Process Flow ................................................................................. 23.2 Programmed Defect Mask Design.................................................................................... 23.3 Mask Pattern Metrology ................................................................................................... 23.4 Wafer Printing Process Condition .................................................................................... 33.5 Definition of Printable Defects ......................................................................................... 3

4 RESULTS AND DISCUSSION ............................................................................................... 34.1 Programmed Defect Mask Fabrication ............................................................................. 34.2 Wafer Printing Results on Defect Printability .................................................................. 4

4.2.1 L/S=140 nm/360 nm Pattern .................................................................................. 44.2.2 L/S=100 nm/400 nm Pattern .................................................................................. 7

4.3 Aerial Image Simulation vs. Resist Process Simulation for Defect Printability............... 94.4 3D EMF vs. 2D Aerial Image Simulation for Phase Defect Printability........................ 114.5 Defect Printability Prediction for Denser Pattern of L/S=130 nm/130 nm .................... 134.6 Phase Bump Defect Repair ............................................................................................. 14

5 CONCLUSION AND FUTURE WORKS ............................................................................. 16

6 REFERENCES........................................................................................................................ 17

iv

Technology Transfer # 00033918A-TR International SEMATECH

List of Figures

Figure 1 Chrome Defects and 180° Phase Bumps...................................................................... 2

Figure 2 Defect Size Measurement ............................................................................................ 3

Figure 3 CD SEM Defect Size Measurement Results (Programmed Size vs. MeasuredSize) ............................................................................................................................. 4

Figure 4 Edge Phase Bump Defect Pattern; Chrome Etched Away from Line Edge ................ 4

Figure 5 Defect Size vs. Linewidth of L/S = 140 nm/360 nm Pattern with FocusVariation....................................................................................................................... 5

Figure 6 Expose/Defocus Window of L/S = 140 nm/360 nm Pattern........................................ 6

Figure 7 MNPD Comparison Between Defocus Only and Expose/Defocus Window ofL/S = 140 nm/360 nm Pattern ...................................................................................... 7

Figure 8 Defect Size vs. Linewidth of L/S – 100 nm/400 nm Pattern with FocusVariation....................................................................................................................... 7

Figure 9 Expose/Defocus Window of L/S = 100 nm/400 nm Pattern........................................ 8

Figure 10 MNPD Comparison Between Focus Only and Expose/Defocus Window ofL/S = 100 nm/400 nm Pattern ...................................................................................... 8

Figure 11 Defect Size vs. Linewidth Comparison Between Aerial Image and ResistSimulation of L/S = 140 nm/360 nm Pattern ............................................................... 9

Figure 12 Defect Size vs. Linewidth Comparison Between Aerial Image and ResistSimulation of L/S = 100 nm/400 nm Pattern ............................................................. 10

Figure 13 Simulated Result Comparison Between Defocus Only and Expose/DefocusWindow...................................................................................................................... 10

Figure 14 Aerial Image Comparison Between 2D Simulation and 3D EMF Simulationof L/S = 140 nm/360 nm Pattern with 390 nm Phase Defect .................................... 11

Figure 15 Defect Size vs. Linewidth Simulation Plot Comparison Between 2D AerialImage Simulation and 3D EMF Simulation of L/S = 130 nm/370 nm Pattern.......... 12

Figure 16 Defect Size vs. Linewidth Simulation Plot Comparison Between 2D AerialImage Simulation and 3D EMF Simulation of 1:4 Duty Ratio Pattern and 1:2Duty Ratio Pattern...................................................................................................... 12

Figure 17 Defect Size vs. Linewidth Simulation Plot comparison Between 2D AerialImage Simulation (Shadow) and 3D EMF Simulation of L/S = 30 nm/370 nmPattern ........................................................................................................................ 13

Figure 18 Defect Size vs. Linewidth Simulation Plot Comparison Between AerialImage and Resist Simulation of L/S = 130 nm/130 nm............................................. 14

Figure 19 Phase Bump Defect Pattern Images Before and After Repair by MechanicalRemoval Tool............................................................................................................. 15

Figure 20 Expose/Defocus Windows of Phase Bump Center Defect Pattern of L/S =140 nm/360 nm Before and After Repair by Mechanical Removal Tool .................. 16

Figure 21 Expose/Defocus Windows of Phase Bump Edge Defect Pattern of L/S =140 nm/360 nm Before and After Repair by Mechanical Removal Tool ................. 16

v


List of Tables

Table 1 MNPD Size Comparison Between Traditionally Defined Case and WindowOverlap Cases .............................................................................................................. 9

Table 2 MNPD Size Comparison Between Aerial Image Simulation and ResistSimulation .................................................................................................................. 10

Table 3 MNPD Size Comparison Between Traditionally Defined Case and WindowOverlap Cases by Resist Simulation .......................................................................... 11

Table 4 MNPD Size Comparison Between 2D Aerial Image and 3D EMF Simulation ........ 13

Table 5 Simulated MNPD Size for 1:1 Dense Pattern of L/S=130 nm/130 nm ..................... 14

vi


Acknowledgements

The authors wish to acknowledge Wally Carpenter and Wayne Smith at InternationalSEMATECH for their support this work. We also thank Barry Hopkins at RAVE for repairwork, and Asmita Shah for AFM images collection.

1


1 EXECUTIVE SUMMARY

Optical lithography is pushed more to extend to sub-wavelength region for very low K1patterning processes. In this, alternating phase shift mask (PSM) is the solution for isolatedpatterns, without changing the wavelength of exposure tools’ light source. With this prospect,critical issues such as design layout complexity, light intensity imbalance between shifted andunshifted space area, and phase defect controllability recently have been studied in order to applyalternating PSM for mass production of devices.

This report recounts a study to find out the maximum non-printable phase defects for 130 nm and100 nm lines by printing the wafer using a KrF deep ultraviolet (DUV) scanner. With thelimitations of the mask making process for very small programmed defects, masks with dutyratio around 1:3 have been made. After resist simulation for the test pattern was verified bywafer printing results, printable defects for denser pattern were predicted.

In addition to the defect printability study, mechanical repair tools for phase bump defects weretested using 248 nm AIMS, atomic force microscopy (AFM), and critical dimension scanningelectron microscopy (CD SEM) metrology, as well as wafer printing. Electromagnetic field(EMF) three-dimensional (3D) simulation also was compared with a commercialized two-dimensional (2D) simulation tool for phase defect printability.

2 INTRODUCTION

Optical lithography is pushed more to extend to sub-wavelength region for very low K1patterning processes, in which alternating PSM is the solution for isolated patterns withoutchanging the wavelength of exposure tools’ light source. With this prospect, the critical issuessuch as design layout complexity, light intensity imbalance between shifted and unshifted spacearea [1,2], and phase defect controllability [3,4] recently have been studied in order to applyalternating PSM for device mass production.

Especially, discriminating the non-printable defects from the printable defects is becoming moreimportant to mask manufacturers, not only for mask making process itself, but also fordetermining the specifications of inspection tools and repair tools within reasonable tooldevelopment cost.

However, while all the literatures have considered the exposure tool’s defocus for defectprintability study, few of them have included exposure dose as a factor for it. In addition, theyhave used aerial image simulation to predict the printable defects, but resist process factorscannot be neglected, especially for low K1 process.

T his project studied phase defect printability for 130 nm and 100 nm lines of 500 nm pitch byprinting the wafer using programmed defect masks and a KrF DUV scanner. Using experimentalresults, printability was compared by considering only the scanner’s focus factor, and by lookingat both exposure dose and focus factors.

Aerial image simulation and resist process simulation were compared to actual wafer printingresults, verifying the defect printability prediction for 130 nm and 100 nm lines of 500 nmpitched pattern using a two-dimensional (2D) simulation tool.

2


Two-dimensional simulation was compared to 3D electromagnetic field simulation for theprintability of phase bump defects and divot defects. Based on verified defect simulation for ~1:3duty ratio test patterns, printable phase defects for denser pattern were predicted.

Mechanical repair tool was tested for printable quartz bumps using 248 nm AIMS, AFM, andCD SEM metrology, as well as wafer printing.

3 EXPERIMENTAL PROCEDURES

3.1 Overall Experimental Process Flow

After the programmed defect mask was fabricated, experimenters measured the mask patternCDs and actual defect sizes using CD SEM. Based on this mask pattern information, waferprinting was done and analyzed to define printable defects. Duty ratio patterns of 1:3 wereverified from experimental data, aerial image simulation, and resist process simulation.Afterward, printable phase defects for denser pattern were predicted using a verified simulationapproach. Regarding phase defects repair tool evaluation, several printable quartz defects wereremoved and inspected using CD SEM, AFM, 248 nm AIMS, and wafer printing test, before andafter repair.

3.2 Programmed Defect Mask Design

With the limitations of the mask making process for very small programmed phase defects, themasks were made with duty ratio around 1:3. Wet etch processing after dry etch was omitted toget very small phase defects. Lines and space patterns at 130 nm/370 nm and 100 nm/400 nmwere applied for wafer printing test. Three different phases, 60/120/180º, were made for phasebump defects and divot defects, as well as chrome defects. All programmed defects were locatedat the center and edge of the patterns, as shown in Figure 1. The experiment was limited tochrome defects and 180º phase bump defects.

Figure 1 Chrome Defects and 180° Phase Bumps

3.3 Mask Pattern Metrology

A KLA8100R CD SEM was used to measure pattern CDs and defects sizes on the mask. Thedefect sizes on the mask were measured by diameter of the inside circles, as shown in Figure 2.To evaluate the mechanical repair tool’s capability of quartz bump defect removal, AFM, CDSEM and 248 nm AIMS images were captured both before and after repair. Those images werecompared to wafer printing results.

3


Figure 2 Defect Size Measurement

3.4 Wafer Printing Process Condition

An SVGL MS3 KrF DUV scanner and UV6 resist were used for printing the wafer. Illumination�� organic bottom antireflective coating (ARC). Exposure dose and focus matrices were shot on thewafer so that the process window of defective pattern linewidths could be compared to non-defective pattern linewidth.

3.5 Definition of Printable Defects

Traditionally, the standard wafer CD is defined as that printed line width included in the regionsurrounded by the lines corresponding to specified focus range and wafer CD target +/-10% onthrough focus plots. A 0.4 µm focus range was used for this study, based on the process windowof no defect patterns.

However, experimenters calculated the process window overlap between defect patterns and nodefect patterns for various defect sizes in order to include the expose dose process factor toprintability study. This process window overlap approach was compared with traditionalmethods to investigate the relationship between them. The term “maximum non-printable defect”(MNPD) was used because it is more useful to mask users rather than is minimum printabledefect. Also, this study measured wafer CD of the linewidth next to programmed defects.

4 RESULTS AND DISCUSSION

4.1 Programmed Defect Mask Fabrication

In case of phase bump defects, programmed defects were generated as small as 60 nm at centerand 40 nm at the edge of the pattern on the mask. Also, 180º phase bump defects weregenerated at almost the same sizes as chrome defects, as shown in Figure 3. Experimenterscould not get phase divot defects smaller than 700 nm on the mask because of the secondwriter’s resolution limit.

4


Cr dot defect (site_A)

0.0

50.0

100.0

150.0

200.0

250.0

0 50 100 150 200 250

target size (nm)

me

asur

ed s

ize

(nm

)

CE_DE

ED_DE

CE_ISO

ED_ISO

Quarts bump defect (site_A)

0.0

50.0

100.0

150.0

200.0

250.0

0 50 100 150 200 250

target size (nm)

me

asu

red

siz

e (n

m)

CE_DE

ED_DE

CE_ISO

ED_ISO

Figure 3 CD SEM Defect Size Measurement Results (Programmed Size vs. MeasuredSize)

It also was discovered that chrome layer was unexpectedly removed at the pattern where thephase bump edge defects are located, due to the limited registration ability of the second writer.This chrome removed pattern caused pattern position shift compared to non-attacked pattern, asshown in Figure 4.

(a) mask pattern (b) resist pattern (c) aerial image

Figure 4 Edge Phase Bump Defect Pattern; Chrome Etched Away from Line Edge

4.2 Wafer Printing Results on Defect Printability

4.2.1 L/S=140 nm/360 nm Pattern

The target pattern size was changed from L/S=130 nm/370 nm to L/S=140 nm/360 nm, based onmask CD measurement. With the traditional definition of printability, which considers only focusrange at optimum expose dose, MNPD sizes were 320 nm for chrome center defect, 305 nm for180º phase bump center defect, 270 nm for chrome edge defect, and 280 nm for phase bumpedge defect, as shown in Figure 5.

5


Cr defect (center, dense, siteA)

103

129

155

181

207

0 100 200 300 400 500

defect size (nm)

Lin

ewid

th (

nm 36/-0.6

36/-0.8

36/-1.0

36/-0.4

36-/-0.2

Phase defect (center_dense_siteA)

107

133

159

185

211

0 100 200 300 400 500

defect size (nm)

Lin

ew

idth

(n

m 36/-1.0

36/-0.8

36/-0.6

36/-0.4

36/-0.2Line

wid

th (

nm)

Line

wid

th (

nm)

(a) Chrome center defects 180 º. phase bump center defects

Cr defect (edge, dense, siteA)

107

133

159

185

211

0 100 200 300 400 500

defect size (nm)

Lin

ewid

th (

nm

)

36/-0.6

36/-0.8

36/-1.0

36/-0.4

36-/-0.2

phase defect (edge_dense_siteA)

112

138

164

190

216

0 100 200 300 400 500

defect size (nm)

Lin

ewid

th (

nm

)

36/-1.0

36/-0.8

36/-0.6

36/-0.4

36/-0.2Line

wid

th (

nm)

Line

wid

th (

nm)

(c) Chrome edge defects (d) 180º phase bump edge defects

Figure 5 Defect Size vs. Linewidth of L/S = 140 nm/360 nm Pattern with FocusVariation

When the process window overlap method was applied between defect pattern and no defectpattern, traditionally defined MNPD sizes showed 84% overlap for chrome center defect, 76%overlap for phase bump center defect, 81% chrome edge defect, and 63% overlap for phasebump edge defect, as shown in Figure 6.

6


Defect size/144 nmoverlap/95%

270 nm89%

NoDefects

320 nm84%

410 nm47%


265 nm82%

NoDefects

305 nm76%

415 nm12%

32 34 36 38 40-1 144 132 130 133 106

-0.8 156 148 138 139 135.7

-0.6 157 139 142 144 132-0.4 156 148 142 134 137

-0.2 147 124 133 135 115

32 34 36 38 40-1 150 139 133 133 114

-0.8 160 156 143 140 138

-0.6 161 143 148 149 133-0.4 161 151 147 143 139

-0.2 153 127 139 133 119

32 34 36 38 40-1 151 141 137 139 120

-0.8 158 157 145 143 143

-0.6 162 147 149 148 133-0.4 163 154 148 147 142

-0.2 160 128 140 138 126

32 34 36 38 40-1 154 147 139 142 121

-0.8 166 158 149 148 145

-0.6 169 152 154 155 140-0.4 169 156 151 150 144

-0.2 165 134 140 144 130

32 34 36 38 40-1 161 150 145 148 129

-0.8 170 166 157 154 150

-0.6 175 161 160 161 150-0.4 176 167 157 156 152

-0.2 178 146 149 146 134

32 34 36 38 40-1 144 125 133 123 108

-0.8 157 149 140 139 139-0.6 154 140 146 141 130-0.4 152 144 142 133 136-0.2 144 127 132 128 114

32 34 36 38 40-1 144 136 135 136 113

-0.8 157 153 147 143 137-0.6 157 141 146 145 132-0.4 158 149 144 137 137-0.2 144 124 135 132 116

32 34 36 38 40-1 158 156 142 140 130

-0.8 169 162 155 151 148

-0.6 169 155 157 155 143-0.4 171 160 151 151 144

-0.2 158 138 145 140 128

32 34 36 38 40-1 178 166 162 160 141

-0.8 183 179 175 168 159

-0.6 187 170 171 168 157-0.4 186 172 168 162 156

-0.2 173 150 157 153 145

32 34 36 38 40-1 157 145 141 142 122

-0.8 167 159 154 148 146

-0.6 168 150 157 150 139-0.4 163 155 151 145 143

-0.2 154 131 138 138 124

(a) Chrome center defects (b) 180º phase bump center defects


230 nm81%

NoDefects

270 nm81%

380 nm25%


230 nm88%

NoDefects

280 nm63%

380 nm25%

32 34 36 38 40-1 144 131 130 129 108

-0.8 152 145 140 136 134

-0.6 156 140 146 143 128-0.4 159 143 142 138 132

-0.2 144 119 135 132 113

32 34 36 38 40-1 145 133 131 136 111

-0.8 155 148 142 140 140

-0.6 155 144 147 146 128-0.4 159 148 145 143 137

-0.2 150 123 136 133 118

32 34 36 38 40-1 149 139 134 132 112

-0.8 160 157 152 144 142

-0.6 163 150 152 151 136-0.4 165 155 152 144 142

-0.2 151 127 135 131 119

32 34 36 38 40-1 152 145 140 137 117

-0.8 166 157 152 146 149

-0.6 166 152 156 154 145-0.4 166 155 151 147 145

-0.2 154 128 134 135 123

32 34 36 38 40-1 160 150 142 142 123

-0.8 173 166 161 155 156

-0.6 177 166 165 164 149-0.4 180 167 161 152 151

-0.2 161 136 145 145 129

32 34 36 38 40-1 152 143 139 140 120

-0.8 159 155 148 143 144

-0.6 160 141 151 148 133-0.4 158 139 138 135 131

-0.2 136 110 121 120 108

32 34 36 38 40-1 154 143 143 138 123

-0.8 162 159 155 148 146

-0.6 165 144 153 150 137-0.4 159 144 145 139 131

-0.2 139 110 121 122 102

32 34 36 38 40-1 163 153 150 150 122

-0.8 170 164 159 153 152

-0.6 173 151 158 154 143-0.4 160 143 143 140 130

-0.2 136 104 114 110 98

32 34 36 38 40-1 168 160 154 154 130

-0.8 176 169 164 159 153

-0.6 175 157 161 158 145-0.4 168 144 148 142 131

-0.2 130 104 108 109 100

32 34 36 38 40-1 184 176 165 167 143

-0.8 194 186 176 175 169

-0.6 192 171 174 169 160-0.4 175 155 158 152 139

-0.2 143 101 117 113 102

(c) Chrome edge defects (d) 180º phase bump edge defects

Figure 6 Expose/Defocus Window of L/S = 140 nm/360 nm Pattern

7


For 140 nm/360 nm lines and spaces pattern, chrome defects and phase bump defects smallerthan 230 nm keep the process window overlap at least 80% as shown in Figure 7.

MNPD (nm)

300

200

100

bestfocus

0.4 µmfocus range

70%overlap

80%overlap

90%overlap

Chrome center defect

180 deg. phase edge defect

180 deg. phase center defect

Chrome edge defect

Figure 7 MNPD Comparison Between Defocus Only and Expose/Defocus Window ofL/S = 140 nm/360 nm Pattern

4.2.2 L/S=100 nm/400 nm Pattern

For L/S=100 nm/400 nm pattern, MNPD sizes of traditional method were 150 nm for phasebump center and edge defect, as shown in Figure 8.

180 deg. edge bump

75

95

115

135

0 100 200 300 400 500defect size (nm)

linew

idth

(n

m)

0.60.40.20-0.2

180 deg. center bump

75

95

115

135

0 100 200 300 400 500defect size (nm)

Lin

ewid

th (

nm

)

0.6

0.4

0.2

0

-0.2

(a) 180º phase bump center defects (b) 180º phase bump edge defects

Figure 8 Defect Size vs. Linewidth of L/S – 100 nm/400 nm Pattern with FocusVariation

8


When the process window overlap method was applied, 150 nm phase bump defects showed72% overlap for center defect and 83% overlap for edge defect, as shown in Figure 9.


200 nm92%

NoDefects

300 nm25%

400 nm0%


200 nm50%

NoDefects

300 nm42%

400 nm25%

46 48 50 52 540.6 97 95 104 84 790.4 120 106 105 88 920.2 113 110 105 109 1150 117 117 112 106

-0.2

46 48 50 52 540.6 89 92 94 94 750.4 120 106 107 97 880.2 119 113 111 108 1080 114 112 110 103

-0.2

46 48 50 52 540.6 99 99 97 95 910.4 126 112 111 97 1030.2 114 115 115 115 1100 122 120 116 112

-0.2

46 48 50 52 540.6 104 101 107 112 1020.4 130 119 117 114 1130.2 119 120 117 116 1190 124 122 119 113

-0.2

46 48 50 52 540.6 126 127 119 113 1030.4 138 129 125 116 1130.2 131 130 127 128 1250 135 126 125 117

-0.2

46 48 50 52 540.6 97 95 104 84 790.4 120 106 105 88 920.2 113 110 105 109 1150 117 117 112 106

-0.2

46 48 50 52 540.6 87 91 97 93 760.4 122 107 110 102 990.2 113 111 111 110 110

0 119 111 113 108-0.2

46 48 50 52 540.6 88 91 98 95 920.4 129 109 111 95 980.2 118 118 119 115 1140 125 119 112 115

-0.2

46 48 50 52 540.6 98 98 108 94 850.4 132 115 119 101 980.2 126 126 122 122 1200 131 121 122 115

-0.2

46 48 50 52 540.6 101 104 107 100 870.4 140 124 126 109 1130.2 138 136 139 133 1320 141 133 133 127

-0.2


Figure 9 Expose/Defocus Window of L/S = 100 nm/400 nm Pattern

It was found that traditionally defined MNPD sizes for 100 nm/400 nm pattern also showedaround 80% overlap for phase bump defects, as shown in Figure 10.

MNPD (nm)

300

200

100

bestfocus

0.4 µmfocus range

70%overlap

80%overlap

90%overlap


180 deg. phase edge defect

Figure 10 MNPD Comparison Between Focus Only and Expose/Defocus Window ofL/S = 100 nm/400 nm Pattern

9


Based on the Table 1, 70% to 80% process window overlap can be expected between defectpattern and no defect pattern if an MNPD-sized defect exists at pattern of L/S=140 nm/360 nmand 100 nm/400 nm.

Table 1 MNPD Size Comparison Between Traditionally Defined Case and WindowOverlap Cases

L/S=140 nm/360 nm L/S=100 nm/400 nm

Chrome Phase bump Phase bump

Center Edge Center Edge Center Edge

Traditional way 320 270 305 280 150 150

70% overlap 320 270 305 230 150 150

80% overlap 320 270 265 230 — 150

90% overlap 144 120 155 190 — —

4.3 Aerial Image Simulation vs. Resist Process Simulation for Defect Printability

Aerial image simulation and resist simulation were done for the same patterns ofL/S=140 nm/360 nm and 100 nm/400 nm, and compared to experimental results. Experimentersused the UV6 resist parameters of baseline lithography process at International SEMATECH.The square-shaped defects were applied for the defect pattern simulation.

As a result, aerial image simulation data were very to the experimental data, while resistsimulation tended to overestimate defect printability compared to experimental data, as shown inFigure 11, Figure 12, and Table 2.

Cr center defect

101

127

153

179

0 100 200 300 400 500 600

defect size (nm )

Lin

ew

idth

(n

m

Aeria lR esist

Exp.

180 quartz defect

101

127

153

179

0 100 200 300 400 500 600

defect size (nm )

Lin

ew

idth

(n

m

Aeria lResist

Exp.

Lin

ewid

th (

nm

)

Lin

ewid

th (

nm

)


Figure 11 Defect Size vs. Linewidth Comparison Between Aerial Image and ResistSimulation of L/S = 140 nm/360 nm Pattern

10


C r center defect

70

90

110

130

0 100 200 300 400 500 600

d efect s ize (nm )

Lin

ew

idth

(n

m

A eria lR es is tE xp .

180 quartz defect

70

90

110

130

0 100 200 30 0 400 500 600

defect s ize (nm )

Lin

ew

idth

(n

m

Aeria lR esis tExp .

Lin

ewid

th (

nm

)

Lin

ewid

th (

nm

)


Figure 12 Defect Size vs. Linewidth Comparison Between Aerial Image and ResistSimulation of L/S = 100 nm/400 nm Pattern

Table 2 MNPD Size Comparison Between Aerial Image Simulation and ResistSimulation

L/S=140 nm/360 nm L/S=100 nm/400 nm

Chrome center Phase center Chrome center Phase center

2D aerial image 350 250 350 250

Resist simulation 220 205 200 150

Experiment 320 265 — 250

From these resist simulations, it was found that traditionally defined MNPD showed around 70%process overlap between defect pattern and no defect pattern, as shown in Figure 13 and Table 3.

MNPD(nm)

300

200

100

aerialimage

resistimage

70%overlap

80%overlap

90%overlap

300

200

100

aerialimage

resistimage

70%overlap

80%overlap

90%overlap


180 deg. phase edge defectchrome center defect

chrome edge defect

(a) L/S = 140 nm/360 nm pattern (b) L/S = 100 nm/400 nm pattern

Figure 13 Simulated Result Comparison Between Defocus Only and Expose/DefocusWindow

11


Table 3 MNPD Size Comparison Between Traditionally Defined Case and WindowOverlap Cases by Resist Simulation

L/S=140 nm/360 nm L/S=100 nm/400 nm

Chrome Phase bump Chrome Phase bump

Center edge Center Edge Center Edge Center Edge

Traditional way 220 120 205 150 200 200 150 150

70% overlap 220 120 205 150 200 200 130 150

4.4 3D EMF vs. 2D Aerial Image Simulation for Phase Defect Printability

Three-dimensional EMF simulation was applied for phase bump defects and divot defects tocompare with 2D simulation, as shown in Figure 14.

(a) 2D aerial image (b) 3D EMF image for bump (c) 3D EMF image for divot

Figure 14 Aerial Image Comparison Between 2D Simulation and 3D EMF Simulationof L/S = 140 nm/360 nm Pattern with 390 nm Phase Defect

12


It was found that 2D simulation underestimates defect printability for phase bump center defects,while it estimates similar to 3D EMF simulation for phase divot defects of L/S=130 nm/370 nm,as shown in Figure 15.

1 2 0

1 3 0

1 4 0

1 5 0

1 6 0

1 7 0

0 1 0 0 2 0 0 3 0 0 4 0 0

S h a d o w

E M F - 1 8 0b u m p

E M F - 1 8 0 d ive t

Figure 15 Defect Size vs. Linewidth Simulation Plot Comparison Between 2D AerialImage Simulation and 3D EMF Simulation of L/S = 130 nm/370 nm Pattern

When a denser pattern of L/S=100 nm/200 nm is considered, the printability difference between2D simulation and 3D EMF simulation increases as shown in Figure 16.

1 0 0

1 0 5

1 1 0

1 1 5

1 2 0

1 2 5

1 3 0

0 1 0 0 2 0 0 3 0 0 4 0 0

S h a d o w C DE M F C DC D S p e c

100

105

110

115

120

125

130

0 100 200 300 400

Shadow CDEMF CDCD Spec

(a) L/S = 100 nm/400 nm pattern (b) L/S = 100 nm/200 nm pattern

Figure 16 Defect Size vs. Linewidth Simulation Plot Comparison Between 2D AerialImage Simulation and 3D EMF Simulation of 1:4 Duty Ratio Pattern and1:2 Duty Ratio Pattern

13


A 2D simulation showed that phase bump edge defects are more printable than phase centerdefects, but 3D EMF simulation did not show a difference between center and edge defects atpattern of L/S=140 nm/360 nm, as shown in Figure 17 and Table 4.

120

125

130

135

140

145

150

155

160

165

170

0 100 20 0 300 400

ShadowEM F

120

125

130

135

140

145

150

155

160

165

170

0 100 200 300 400

S hadowE M F


Figure 17 Defect Size vs. Linewidth Simulation Plot comparison Between 2D AerialImage Simulation (Shadow) and 3D EMF Simulation of L/S = 30 nm/370 nmPattern

Table 4 MNPD Size Comparison Between 2D Aerial Image and 3D EMF Simulation

L/S=140 nm/360 nm L/S=100 nm/400 nm

Phase bump Phase divot Phase bump

Center Edge Center Edge Center Edge

2D aerial image 280 230 280 — 300 —

3D EMF simulation 230 230 280 — 200 —

4.5 Defect Printability Prediction for Denser Pattern of L/S=130 nm/130 nm

Even though experimenters found out that aerial image simulation could predict the defectprintability for the pattern of L/S=140 nm/360 nm, 100 nm/400 nm, they are not sure if it worksfor dense pattern, such as duty ratio 1:1. With the limitations of the mask making process, it isdifficult to make programmed defect mask having small defects on 1:1 dense pattern. Therefore,2D aerial image and resist simulation was applied to see MNPD at denser pattern of L/S =130 nm/130 nm in this section. Aerial image simulation still predicted big MNPD, 350 nm forchrome defect and 250 nm for phase defect, at 1:1 dense pattern; in this, resist simulationpredicted MNPD smaller than 100nm, as shown in Figure 18 and Table 5. Aerial imagesimulation for defect printability of 1:1 dense pattern does not make sense, based on thesimulation and the experimental results from the pattern of L/S=140 nm/360 nm and100 nm/400 nm. One can predict conservatively that defects smaller than 60 nm will not beprinted on this pattern, but exact numbers of MNPD size should be revisited with more analysison 3D EMF simulation, using actual defect shapes and sizes.

14


MNPD (nm)

300

200

100

aerialimage

resistimage

70%overlap

80%overlap

180 deg. phase center defect180 deg. phase edge defectchrome center defectchrome edge defect

Figure 18 Defect Size vs. Linewidth Simulation Plot Comparison Between Aerial Imageand Resist Simulation of L/S = 130 nm/130 nm

Table 5 Simulated MNPD Size for 1:1 Dense Pattern of L/S=130 nm/130 nm

Chrome Phase bump

Center Edge Phase Edge

2D aerial image 350 — 250 —

Resist simulation 80 130 80 80

70% overlap 60 80 60 60

4.6 Phase Bump Defect Repair

In this section, experimenters tested the mechanical repair tool for quartz bump defect removal,which is being developed by RAVE. Defect pattern was L/S=140 nm/360 nm and repaireddefect sizes were 250 nm, 400 nm, and 500 nm for center defects and edge defects. The defectpattern was inspected using AFM, 248 nm AIMS, CD SEM and wafer pattern before and afterrepair. As a result, phase bump center defects were repaired well enough not to print on thewafer, but edge defects did not work well, as shown in Figure 19. When process windows ofdefect patterns was examined, it was found that they were fully recovered after repair for centerbump defects, as shown in Figure 20 but not for edge bump defects as shown in Figure 21.

15


Pre repair

Post repair

500 nm180 deg. phase

center bump


center bump


edge bump


edge bump

(a) AFM

Pre repair

Post repair

248 nm AIMS

Pre repair

Post repair

(c) wafer printing

Figure 19 Phase Bump Defect Pattern Images Before and After Repair by MechanicalRemoval Tool

16



415 nm12%

NoDefects

520 nm0%


415 nm95%

NoDefects

520 nm80%

32 34 36 38 40-1 144 125 133 123 108

-0.8 157 149 140 139 139-0.6 154 140 146 141 130-0.4 152 144 142 133 136-0.2 144 127 132 128 114

32 34 36 38 40-1 219 205 190 186 171

-0.8 231 224 207 193 188

-0.6 218 202 192 190-0.4 217 198 191 185 179

-0.2 196 177 177 170 163

32 34 36 38 40-1 157 145 141 142 122

-0.8 167 159 154 148 146

-0.6 168 150 157 150 139-0.4 163 155 151 145 143

-0.2 154 131 138 138 124

32 34 36 38 40-1 138 135 131 128 112

-0.8 154 149 138 138 132-0.6 153 150 138 138 138-0.4 157 151 132 131 138-0.2 150 141 135 130 120

32 34 36 38 40-1 140 135 133 133 117

-0.8 159 152 146 142 136-0.6 154 154 144 141 138-0.4 163 157 138 133 139-0.2 153 147 137 131 121

32 34 36 38 40-1 141 145 142 135 117

-0.8 164 160 147 146 141-0.6 161 158 152 150 142-0.4 167 158 140 139 147-0.2 155 152 138 128 127

32 34 36 38 40-1 147 137 128 128 128

-0.8 152 147 143 141 127-0.6 163 156 129 140 138-0.4 154 147 141 140 138-0.2 152 150 145 134 119

32 34 36 38 40-1 178 166 162 160 141

-0.8 183 179 175 168 159

-0.6 187 170 171 168 157-0.4 186 172 168 162 156

-0.2 173 150 157 153 145

(a) Pre-repair (b) Post repair

Figure 20 Expose/Defocus Windows of Phase Bump Center Defect Pattern of L/S = 140nm/360 nm Before and After Repair by Mechanical Removal Tool


480 nm12%

NoDefects


480 nm35%

NoDefects

32 34 36 38 40-1 163 153 150 150 122

-0 .8 170 164 159 153 152

-0 .6 173 151 158 154 143-0 .4 160 143 143 140 130

-0 .2 136 104 114 110 98

32 34 36 38 40-1 2 0 1 1 9 3 1 8 2 18 2 1 7 0

-0 .8 2 1 1 2 0 6 2 0 1 19 2 1 8 6

-0 .6 215 194 188 185 176-0 .4 1 9 5 1 6 6 1 6 8 15 7 1 4 6

-0 .2 1 4 4 9 8 1 0 4 9 5 9 6

32 34 36 38 40-1 152 143 139 140 120

-0 .8 159 155 148 143 144

-0 .6 160 141 151 148 133-0 .4 158 139 138 135 131

-0 .2 136 110 121 120 108

32 34 36 38 40-1 126 124 126 112 97

-0 .8 155 150 142 138 136-0 .6 181 169 166 164 162-0 .4 204 197 193 182 177-0 .2 218 210 203 191 191

32 34 36 38 40-1 127 126 135 120 110

-0.8 160 155 148 141 138-0.6 170 171 158 154 152-0.4 184 184 171 168 164-0.2 196 183 173 162 159

32 34 36 38 40-1 138 135 131 128 112

-0 .8 154 149 138 138 132-0 .6 153 150 138 138 138-0 .4 157 151 132 131 138-0 .2 150 141 135 130 120

(a) Pre-repair (b) Post repair

Figure 21 Expose/Defocus Windows of Phase Bump Edge Defect Pattern of L/S =140 nm/360 nm Before and After Repair by Mechanical Removal Tool

5 CONCLUSION AND FUTURE WORKS

Conclusions from this study are as follows:

• Traditionally defined MNPD sizes provided a 70~80% process window overlap betweendefect pattern and no defect pattern of L/S=140 nm/360 nm and 100 nm/400 nm.

• Two-dimensional aerial imaging showed good agreement with experimental results, whileresist process simulation overestimated it.

• Three-dimensional EMF simulation showed that 2D aerial image simulation underestimatesdefect printability for phase bump center defects, but estimates similarly for phase divotdefects.

• Conservative prediction of MNPD sizes for chrome defects and phase defects are ~60 nmfor L/S=130 nm/130 nm.

17


• Phase bump defects could be removed well enough not to print on the wafer by mechanicalrepair tool for center defects.

Future study will include 60º and 120º phase defects for defect printability. Wafer printingtesting will be done for phase divot defect patterns to verify 3D EMF simulation.

6 REFERENCES

[1] T. Yasuzato, S. Ishida, and H. Tanabe “Pattern Dependence of Mask Topography Effectin Alternating Phase-Shifting Masks,” SPIE Vol. 3748, pp. 363–370, 1999.

[2] S. Peng “Through-Focus Image Balancing of Alternating Phase Shifting Masks,” SPIEVol.3873, pp. 328–336, 1999.

[3] L. Liebmann, S. Mansfield, A. Wong, J Smolinski, S. Peng, K. Kimmel, M. Rudzinski, J.Wiley, and L. Zurbrick “High Resolution Ultraviolet Defect Inspection of DAP ReticlesDarkfield Alternate Phase,” SPIE Vol. 3873, pp. 148–161, 1999.

[4] S. Nagashige, K. Hayashi, S. Akima, H. Takahashi, K. Chiba, Y. Yamada, and Y.Matsuzawa “Detection and Repair of Multiphase defects on Alternating Phase-ShiftMasks for DUV Lithography,” SPIE Vol. 3873, pp. 127–137, 1999.

International SEMATECH Technology Transfer2706 Montopolis Drive

Austin, TX 78741

http://www.sematech.orge-mail: [email protected]

alternating phase shift mask (psm) phase defect … phase shift mask (psm) phase defect printability...

Documents