Stability of Photo Resist Coating Performance of Small Dispense

Nozzle Size in Photolithographic Spin Coating Process

Xiao Li, Tom Lehmann and Warren Greene
LSI Logic Corporation, Gresham, Oregon, USA

ABSTRACT

Reduction of photoresist consumption to reduce costs while maintaining resist coating quality is becoming a major
challenge for process and equipment engineers in the semiconductor industry. This challenge can be met by reducing
dispense nozzle diameter to maintain a constant dispense rate at a reduced dispense volume.
In this study, two small dispense nozzle sizes (0.5 and 0.6 mm in diameter) and two resist dispense volumes (0.4 and
0.5cc per coating) were evaluated during the resist spin coating process. Stability tests of five resist thickness means and
ranges of three photo resists types with various resist viscosities were performed using small dispense nozzles and small
resist dispense volumes. Each stability test consisted of both 25 wafer continuous resist coats and one wafer per coating
for 15 days. Coat defects from the coat process using a small dispense nozzle and small resist dispense volume were
analyzed on the layers of Island, Poly, Metal and Contact in a manufacturing fab. The effect of the resist coat process
using a small dispense nozzle and a small resist dispense volume on critical dimension (CD) performance of Island,
Poly, Metal and Contact layers before and after etch was reported. Resist thickness uniformity data , coating defect data
and CD data from the small dispense nozzle size and reduced resist dispense volume coating process were also compared
with a normal resist coating process with dispense nozzle size of 1.5mm and resist dispense volume of 0.6 to 0.75cc per
coating.

Keywords: photo resist coating, photo resist consumption, resist dispense, resist viscosity, dispense nozzle

1. INTRODUCTION

As photolithography processes in semiconductor industry have grown more complex, the cost of photo resists used in
these procedures has increased greatly. Reduction of photo resist consumption to reduce costs while maintaining resist
coating quality is becoming a major challenge for process and equipment engineers.
The spin coating process of photo resist has been widely used in modern semiconductor manufacture because the process
can be easily controlled to obtain high quality coating. The biggest disadvantage of spin coating process of photo resist is
its lack of material efficiency because an excessive amount of photo resist is used to prevent coating discontinuities
caused by the photo resist fluid front drying prior to it reaching the wafer edge [1]. Typical spin coating process utilizes
only 2-5% of material dispensed onto the wafer substrate, while the remaining 95-98% is flung off into the coating bowl
and disposed [2]. Several technologies has been developed and implemented to reduce photo resist consumption per
wafer in semiconductor industry recently [3-6].
One method to control and reduce resist volume is to use a small resist dispense nozzle diameter. Investigation of this
method has been performed and reported previously [7]. This study is to continue to evaluate the stability of small resist
dispense nozzle coating process in manufacture fab. Defects from the coat process using a small dispense nozzle and
small resist dispense volume were analyzed on the layers of Island, Poly, Metal and Contact in a manufacturing fab. The
effect of the resist coat process using a small dispense nozzle and a small resist dispense volume on critical dimension
(CD) performance of Island, Poly, Metal and Contact layers before and after etch is reported. Resist thickness
uniformity data , coating defect data and CD data from the small dispense nozzle size and reduced resist dispense volume
coating process are also compared with a normal resist coating process with normal dispense nozzle size of 1.5mm and
resist dispense volume of 0.6 to 0.75cc per coating.

Advances in Resist Technology and Processing XXIII, edited by Qinghuang Lin

Proc. of SPIE Vol. 6153, 61533A, (2006) · 0277-786X/06/$15 · doi: 10.1117/12.655360

Proc. of SPIE Vol. 6153 61533A-1

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 07/03/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx

 2. EXPERIMENTAL

The resist coating, bake and develop experiments in this study were performed on a Tokyo Electron Clean DUV ACT8
track with 200 mm silicon wafers. Two small dispense nozzle sizes (0.5 and 0.6 mm in diameter) and two resist dispense
volumes (0.4 and 0.5cc per coating) were evaluated during the resist spin coating process. The bake temperatures and
times varied from 90 to 205C and from 60 to 90 seconds, respectively. The measurement of resist thickness and
reflectivity was performed on a KLA1070DB or a KLA1280SE. The distance between resist dispense nozzle tip and
wafer surface was 5 mm. The dispense pump used was an RDS pump with a 0.1um filter. KLA2139 was used to
quantify defects at 0.39 um pixel size. The scanner utilized in this experiment was an ASML 5500/700. Deep
Ultraviolet Bottom Antireflective Coating (BARC) and DUV resists utilized in this study were from Brewer Science,
Sumitomo Chemical and JSR. All wafers were developed with Tokyo Ohka NMDW 2.38% TMAH. The critical
dimension (CD) was measured on KLA8100 targeting bottom profile.

3. RESULTS AND DISCUSSION

3.1 Stability of Small Dispense Nozzle Size Process for BARC and DUV Resist Coating

The stability of a process is critical in semiconductor manufacturing. A process with wide process window not only
provides high quality product but also reduces tool maintenance and increases productivity. Stability is also one of the
most important factors to determine if a new process is suitable to production or not. The stability tests were performed
on bare silicon wafers and TEL ACT8 track with volume consumption of 0.4cc per coating for a Brewer Science BARC
using 0.5mm in diameter dispense nozzle and 0.5cc per coating for a resist of Sumitomo Chemical and a resist from JSR
using 0.6mm in diameter dispense nozzle. The measurement of BARC and resist thickness was performed on
KLA1280SE with 49 points of measurement for each test wafer. The thickness mean is an average value of 49 points of
measurement, and thickness range is the difference between maximum value and minimum value of 49 points of
measurement. The volume consumption was measured by graduated cylinder. Figure 1-2 shows the relationship
between Brewer Science BARC thickness measurement and volume consumption per coating using dispense nozzle size
of 0.5mm in diameter. It is noted from Figure 1 and 2 that both acceptable thickness mean and range may be obtained
when volume consumption per wafer reached to the range of 0.2-0.3cc with thickness mean of 700+10 Angstrom and
range less than 15 Angstrom. The relationship between Brewer Science BARC thickness range and pump variation was
also investigated. Figure 2 indicates that BARC thickness range was very stable (less than 2 Angstrom) within the pump
variation range (0.38-0.42 cc per coating). Figure 3-4 shows the relationship between a Sumitomo Chemical resist
thickness measurement and volume consumption per coating using dispense nozzle size of 0.6mm in diameter. It is
noted from Figure 3 and 4 that both acceptable thickness mean and range may be obtained when volume consumption
per wafer reached to the range of 0.4-0.5cc with thickness mean of 6300+40 Angstrom and range less than 40 Angstrom.
The relationship between the Sumitomo Chemical resist’s thickness range and pump variation was also investigated.
Figure 5 indicates that Sumitomo Chemical resist thickness range was stable (less than 5 Angstrom) in the pump
variation range (0.48-0.52 cc per coating). Figure 5-6 shows the relationship between a JSR resist thickness
measurement and volume consumption per coating using dispense nozzle size of 0.6mm in diameter. It is noted from
Figure 5 and 6 that both acceptable thickness mean and range may be obtained when volume consumption per wafer
reached to the range of 0.4-0.5cc with thickness mean of 5650+40 Angstrom and range less than 40 Angstrom. The
relationship between JSR resist’s thickness range and pump variation was also investigated. Figure 6 indicated that JSR
resist thickness range was stable (less than 5 Angstrom) in the pump variation range (0.48-0.52 cc per coating).
The process stability tests also were performed to run 25 wafers continuously one day and also 2 wafers each day for 15
days using volume consumption of 0.4cc for Brewer Science BARC and 0.5cc for Sumitomo Chemical resist. These
results are shown in Figure 7-12. It is found that the small dispense nozzle process was stable.

Proc. of SPIE Vol. 6153 61533A-2

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 07/03/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx

 750

740

730
Thickness Mean (Angstrom)

720

710

700

690

680

670

660

650
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
Dispense Volume Consumption per Coating (cc)

Figure 1. The relationship between a Brewer Science BARC thickness mean and volume per coating of small dispense
nozzle process (each point presented the average of 2 wafer samples)

35

30
Thickness Range (Angstrom)

25

20

15

10

+ 0.02cc pump variation

0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
Dispense Volume Consumption per Coating (cc)

Figure 2. The relationship between a Brewer Science BARC thickness range and volume per coating of dispense nozzle
process (each point presented the average of 2 wafer samples)

Proc. of SPIE Vol. 6153 61533A-3

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 07/03/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx

 6400

6380

6360
Thickness Mean (Angstrom)

6340

6320

6300

6280

6260

6240

6220

6200
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
Dispense Volume Consumption per wafer (cc)

Figure 3. The relationship between a Sumitomo Chemical resist thickness mean and volume per coating of small
dispense nozzle process (each point presented the average of 2 wafer samples)

100

90

80
Thickness Range (Angstrom)

70

60

50

40

30

20

10
+ 0.02cc pump variation
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
Dispense Volume Consumption per wafer (cc)

Figure 4. The relationship between a Sumitomo Chemical resist’s thickness range and volume per coating of dispense
nozzle process (each point presented the average of 2 wafer samples)

Proc. of SPIE Vol. 6153 61533A-4

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 07/03/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx

 5700

5690

5680
Thickness Mean (Angstrom)

5670

5660

5650

5640

5630

5620

5610

5600
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
Dispense Volume Consumption per Wafer (cc)

Figure 5. The relationship between a JSR resist thickness mean and volume per coating of small dispense nozzle process
(each point presented the average of 2 wafer samples)

60

50
Thickness Range (Angstrom)

40

30

20

10

+ 0.02cc pump variation

0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
Dispense volume Consumption per wafer (cc)

Figure 6. The relationship between a JSR resist thickness range and volume per coating of dispense nozzle process (each
point presented the average of 2 wafer samples)

Proc. of SPIE Vol. 6153 61533A-5

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 07/03/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx

 740
0.4cc/coating
730
Thickness Mean (Angstrom)

720

710

700

690

680

670

660
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Wafer Number

Figure 7. Brewer Science BARC thickness mean data of running 25 wafers using small dispense nozzle (0.5mm in
diameter) process on bare silicon wafer.

20

18 0.4cc/coating

16
Thickness Range (Angstrom)

14

12

10

0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Wafer Number

Figure 8. Brewer Science BARC thickness range data of running 25 wafers using small dispense nozzle (0.5mm in
diameter) process on bare silicon wafer

Proc. of SPIE Vol. 6153 61533A-6

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 07/03/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx

 740
0.4cc/coating
730
Thickness Mean (Angstrom)

720

710

700

690

680

670

660
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
Days

Figure 9. Brewer Science BARC thickness mean data of 2 wafers each day for 15 days using small dispense nozzle
(0.5mm in diameter) process on bare silicon wafer

20

18
0.4cc/coating

16
Thickness Range (Angstrom)

14

12

10

0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
Days

Figure 10. Brewer Science BARC thickness range data of 2 wafers each day for 15 days using small dispense nozzle
(0.5mm in diameter) process on bare silicon wafer

Proc. of SPIE Vol. 6153 61533A-7

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 07/03/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx

 6400

6300

6200
Thickness Mean (Angstrom)

6100

6000

5900

5800

5700

5600

5500 Sumitomo Chemical 0.5cc/wafer

JSR 0.5cc/wafer
5400

5300
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Days

Figure 11. A Sumitomo Chemical resist and a JSR resist thickness mean data of average value of 2 wafers each day for
15 days using small dispense nozzle (0.6mm in diameter) process on bare silicon wafer

50
Sumitomo Chemical 0.5cc/wafer
45
JSR 0.5cc/wafer
40
Thickness Range (Angstrom)

35

30

25

20

15

10

0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Days

Figure 12. A Sumitomo Chemical resist and a JSR resist thickness range data of average value of 2 wafers each day for
15 days using small dispense nozzle (0.6mm in diameter) process on bare silicon wafer

Proc. of SPIE Vol. 6153 61533A-8

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 07/03/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx

 3.2 The effect of small dispense nozzle process on critical dimension (CD) performance of Island, Poly, Metal
and Contact layers

The effect of small dispense nozzle resist coating process on critical dimension (CD) performance of Island, Poly, Metal
and Contact layers was also investigated in this study. Photo resist coating consisted of a Sumitomo Chemical resist
with a Brewer Science Barc for Island layer (6800A resist with 700A Barc), a Sumitomo Chemical resist with a Brewer
Science Barc for Poly layer (6300A resist with 700A Barc), a Sumitomo Chemical resist with a Brewer Science Barc for
Metal layer (8500A resist with 700A Barc) and a JSR resist with a Brewer Science Barc for Contact layer (5650A resist
with 700A Barc). Table 1 lists the data of develop inspect critical dimension (DICD) and final inspect critical dimension
(FICD) of Island, Poly, Metal and Contact layers by use of normal dispense nozzle size (1.5mm in diameter for all resists
and Barc) and small dispense nozzle size (0.5mm in diameter for Brewer Science Barc, 0.6mm in diameter for Sumitomo
Chemical resist and JSR resist). Resist consumption of Barc was 0.4cc per coating and resist consumption of both resists
was 0.5cc per coating. Each data was an average of 6 wafers for each layer and each wafer had 9 points measurement.

Table1. The results of DICD and FICD of Island, Poly, Metal and Contact layers using normal and small resist dispense
nozzle size.

Normal dispense Island Layer Poly Layer Metal Layer Contact Layer
nozzle size A Brewer Science A Brewer Science A Brewer Science A Brewer Science
Barc/A Sumitomo Barc/A Sumitomo Barc/A Sumitomo Barc/A JSR Resist
Chemical Resist Chemical Resist Chemical Resist
DICD Mean (nm) 247.0 216.8 246.3 253.5
DICD Range (nm) 5.6 7.0 8.29 6.27
FICD Mean (nm) 275.5 192.7 256.9 280.6
FICD Range (nm) 12.0 9.5 9.0 10.6
Small dispense Island Layer Poly Layer Metal Layer Contact Layer
nozzle size A Brewer Science A Brewer Science A Brewer Science A Brewer Science
Barc/A Sumitomo Barc/A Sumitomo Barc/A Sumitomo Barc/A JSR Resist
Chemical Resist Chemical Resist Chemical Resist
DICD Mean (nm) 247.35 218.8 246.8 253.0
DICD Range (nm) 4.0 8.1 4.1 4.9
FICD Mean (nm) 275.8 195.1 253.8 282.7
FICD Range (nm) 11.0 7.0 8.7 11.0

It is found from Table 1 that there was no significant different between normal dispense nozzle coating processes and
small dispense nozzle coating process. It is also shown in Table 1 that DICD range was less than 9 nm and FICD range
was less than 12 nm for all three layers by use of small dispense nozzle coating process. The DICD and FICD data
showed that there was not a significant difference between normal dispense nozzle and small dispense nozzle coating
process, and both coating process provided acceptable CD results.

3.3 The effect of small resist dispense nozzle coating process on resist coating defect of Island, Poly, Metal and
Contact layers

Resist coating defects are another important factor that should be considered when a new process is implemented in
production. KLA2139 was used to quantify defects at 0.39 um pixel size. The defect analysis was performed before and
after etching process for layers of Island, Poly, Metal and Contact. The comparison of resist coating defect number and
volume consumption per coating of normal dispense nozzle and small dispense nozzle coating processes are listed in
Table 2. Each data was an average of 3 wafers for each layer. It is found from Table 2 that no significant difference in
defect number exists between normal dispense nozzle and small dispense nozzle coating processes for all layers of
Island, Poly, Metal and Contact. The defect analysis also indicated that no new defect type was found in small dispense
nozzle coating process compared to normal dispense nozzle coating process.

Proc. of SPIE Vol. 6153 61533A-9

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 07/03/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx

 Table2. Comparison of resist coating defect number and volume consumption per coating of normal dispense nozzle
(1.5 mm) and small dispense nozzle coating processes

Layer Dispense Nozzle Size Volume Total Defect Resist Thickness

(mm) Consumption Number (A)
(cc/coating)
Island 1.5 mm for both 0.6 cc for Barc 41.5 before Etch 700A Barc/6800A
Barc and Resist 0.75 cc for Resist 3 after Etch Resist
Island 0.5 mm for Barc and 0.4 cc for Barc 43 before Etch 700A Barc/6800A
0.6 mm for Resist 0.5 cc for Resist 4 after Etch Resist
Poly 1.5 mm for both 0.6 mm for Barc 1263 before Etch 700A Barc/6300A
Barc and Resist 0.75 mm for 1153 after Etch Resist
Resist
Poly 0.5 mm for Barc and 0.4 cc for Barc 95.5 before Etch 700A Barc/6300A
0.6 mm for Resist 0.5 cc for Resist 100.2 after Etch Resist
Metal 1.5 mm for both 0.6 mm for Barc 29.5 before Etch 700A Barc/8500A
Barc and Resist 0.75 mm for 35 after Etch Resist
Resist
Metal 0.5 mm for Barc and 0.4 cc for Barc 30 before Etch 700A Barc/8500A
0.6 mm for Resist 0.5 cc for Resist 25.5 after Etch Resist
Contact 1.5 mm for both 0.6 mm for Barc 24 before Etch 700A Barc/5650A
Barc and Resist 0.75 mm for 25.5 after Etch Resist
Resist
Contact 0.5 mm for Barc and 0.4 cc for Barc 17 before Etch 700A Barc/5650A
0.6 mm for Resist 0.5 cc for Resist 19.5 after Etch Resist

4. CONCLUSIONS

Stability of small dispense nozzle size with reducing resist consumption in spin coating process of photo resist was
studied. The uniformity results of three photo resist coatings with various resist thickness and viscosity showed that
small dispense nozzle coating process was stable in manufacturing. Resist consumption of three investigated photo
resists could be reduced up to 33% by use of small dispense nozzle coating process compared to normal dispense nozzle
coating process. Coating defect analysis indicated that no significant difference in defect number and type existed
between normal dispense nozzle and small dispense nozzle coating processes for all evaluated layers of Island, Poly,
Metal and Contact. The effect of small dispense nozzle resist coating process on critical dimension (CD) performance of
Island, Poly, Metal and Contact layers were also investigated. DICD and FICD data showed that there was no significant
different between normal dispense nozzle coating processes and small dispense nozzle coating process.

Proc. of SPIE Vol. 6153 61533A-10

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 07/03/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx

 5. ACKNOWLEDGEMENTS

The authors would like to thank Doug Skinner and Ryan Pond for providing technical support, and manufacture
managements for providing finance and equipment support on LSI Gresham Campus.

6. REFERENCES

1. Gregory A. Luurtsema, “Spin coating for rectangular substrates”, Thesis of Master Degree, Department of
Electrical Engineering and Computer Sciences, University of California, Berkeley, July, 1997, Berkeley, California,
USA.
2. Haaland, P; McKibben,J.; and Paradi, M. “Fundamental Constraints on Thin Film Coatings for Flat-Panel Display
Manufacturing”, SID, pp.79.
3. Bob Orifice, Dalton Chen, Bette Mullen and Emir Guerra, “How to Minimize Resist Usage During Spin Coating”,
Semiconductor International, 6/1/1988, pp. 32-43.
4. X. Li, W.Greene, C. Bowker, “Double Prewet RRC (reducing resist consumption) process for deep- ultraviolet
bottom antireflective coatings (BARC)”, Proceedings of SPIE, vol.5376, February, 2004, pp. 729-738.
5. Michael A. Rodrigues, “Optimized Photoresist Dispense Method”, United States Patent, No.5405813, April 11,
1995.
6. Emir Gurer, Tom Zhong, John Lewellen, Murthy Krishna and Eddie Lee, “Photoresist Processing Tool-Based
Advanced Technologies for DUV Lithography and Low-k Spin on Dielectrics of 200/300 Wafers”, Semiconductor
Fabtech, 12th Edition, pp. 161-166.
7. X. Li, T. Lehmann and W.Greene, “Evaluation of Photo Resist Coating Performance of Small Dispense Nozzle Size
in Photolithographic Spin Coating Process”, Proceedings of SPIE, vol.5753, February, 2005, pp. 1094-1101.

Proc. of SPIE Vol. 6153 61533A-11

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 07/03/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx