Available online at www.sciencedirect.com

www.elsevier.com/locate/jnoncrysol

Journal of Non-Crystalline Solids 354 (2008) 2879–2884

Ink stamping lithography using polydimethylsiloxane stampby surface energy modification

Byoung-Kwon Choo, Na-Young Song, Ki-Hwan Kim, Jung-Su Choi,Kyu-Chang Park, Jin Jang *

Department of Information Display and Advanced Display Research Center, Kyung Hee University, Hoegi-dong 1, Dongdaemun-gu,

Seoul, 130-701, Republic of Korea

Available online 12 February 2008

Abstract

We studied a direct ink stamp (IS) process as a next generation patterning method using polydimethylsiloxane (PDMS) soft stampand low viscosity ink resist. A propylene glycol monomethyl ether acetate (PGMEA) diluted novolak resist was used as an ink resist. Thesoft stamp and metal coated glass were treated by UV ozone to achieve hydrophilic surface. The process conditions such as the resistthickness on PDMS stamp, resist viscosity, surface energy of stamp and substrate were optimized in this work. A stripe pattern with4.3 lm width Cr pattern was made.� 2008 Elsevier B.V. All rights reserved.

PACS: 68.35.Md; 42.82.Cr; 85.40.Hp

Keywords: Devices; Spin coating; Optical spectroscopy; Surfaces and interfaces

1. Introduction

Photolithography is widely used for the manufacturingof thin-film transistor (TFT) displays. However, photoli-thography is a high cost process so that it should bereplaced with a cheaper technology to reduce the manufac-turing cost of TFT backplane. To reduce the number ofphotomasks required for the TFT array process is an effec-tive method to decrease the manufacturing cost.

More fundamental method for low cost is to use smartlithography replacing photolithography for TFT-LCD[1]. Many kinds of smart lithography methods have beenproposed to make TFT display backplane. IS process isone of the direct lithography and a candidate for next gen-eration patterning method. Printing techniques were pro-posed as low cost lithography such as direct inkjet ofTFT material [2], inkjet of wax for resist pattern [3], gra-

0022-3093/$ - see front matter � 2008 Elsevier B.V. All rights reserved.

doi:10.1016/j.jnoncrysol.2007.10.094

* Corresponding author. Tel.: +82 2 961 0270; fax: +82 2 961 9154.E-mail address: jjang@khu.ac.kr (J. Jang).

vure method [4], direct printing of polymer [5] and flexibleletterpress stamping (FLEPS) [6]. The FLEPS method hadbeen used to pattern films of polystyrene (PS) and otherthermoplastic polymers with micrometer scale feature sizes.But FLEPS needs carrier substrate and substrate heatingup to glass transition temperature of resist. And its patternresolution and pattern shape are not accurate.

On the other hand, imprint is a powerful technology fornano-size patterning of first one layer, but it needs highcost process for applications to nanopatternings. Thereare many issues such as registration between the layers,need of high pressure, high thermal expansion of resistmaterial, bubble in recessed area, and contamination ofmaster stamp due to the contact process, etc.

Combined-nanoimprint-and-photolithography (CNP)[7] and imprint-photo hybrid method [8] were introducedto make the zero residual layer process. But it needs a postprocess to remove the residual layer remained.

In this paper we proposed the IS process which is similarto letter press printing of Millar et al. [6]. The PGMEA

2880 B.-K. Choo et al. / Journal of Non-Crystalline Solids 354 (2008) 2879–2884

diluted novolak resist was coated onto the soft stampdirectly without heating the substrate. The surface energyof the PDMS stamp was controlled to enhance the adhe-sion between resist and stamp.

Fig. 2. Surface energy changes of h-PDMS and s-PDMS surface asincreasing the UV ozone treatment time; solid line is linear regression ofthe circle shape plots and its slope is 0.133, dotted line is linear regressionof triangle shape plots and its slope is 0.167 (unit of slope; surfacemodification speed (mN/m s)).

2. Experimental

The master stamp pattern with stripe shape was madeby photolithography with pattern thickness of 5–7 lmon glass for IS process. Taper angle of the master patternis 70–80� and pattern pitch is 60 lm with its width of16.4 lm. Soft-PDMS (s-PDMS) stamp was made using aSylgard 184 silicon elastomer (Dow Corning), mixed with10 wt% of curing agent (cross linked agent based on aPt-catalyst, Dow Corning). After stirring, it was pouredinto a container which has the master stamp on thebottom. The mixture was evacuated in vacuum desiccatorto remove all bubble in the liquid s-PDMS mixture. Thes-PDMS was cured for 2 h at 60 �C. Then, the hardeneds-PDMS was detached from the master stamp, and thethickness was found to be 4–5 mm. S-PDMS stamp hasthe reverse structure of master stamp and its size anddimension are almost the same as the master stamp inmicro scale. Four materials [methylhydrosiloxane (HMS-301, Gelest), vinylmethylsiloxnedimethylsiloxane (VDT-731, Gelest), Platinumdivinyltetramethyldisiloxane(SIP6831.1, Gelest), and 2,4,6,8 tetramethyltetravinyl-cyclotetrasiloxane (Fluka)] were mixed for hard-PDMS(h-PDMS). Then it was coated onto the master stampand hardened at 60 �C for 30 m. Then, s-PDMS was castand cured onto hardened h-PDMS at 60 �C for 2 h. Pat-terned h-PDMS was detached from the master stampand was cut into a square of 3 � 3 cm2 for IS process.

The 172 nm UV was irradiated for 6 min in air onto thesoft and hard-PDMS stamps to modify the surface fromhydrophobic to hydrophilic. In addition, the Cr deposited

Fig. 1. Process flow of IS process; dropping ink resist on UV ozone treatedconformal contact onto the Cr substrate (c), ink resist pattern on the Cr substraafter ink resist removal.

by RF sputtering was also irradiated 172 nm UV for3 min to increase the surface energy. The contact anglewas measured by a surface tension analyzer produced bySurface-Electro Optics Cooperation using de-ionizedwater. Surface energy was achieved from the contact angledata.

A propylene glycol mono methyl ether acetate(PGMEA) diluted novolak resist was coated onto themodified PDMS stamp as shown in Fig. 1(a) and (b).The novolak resin wt% and surface energy of PDMSstamp were changed for optimizing the IS process. Then,a PDMS stamp was put on the Cr surface for 1 min with-out external pressure onto the PDMS stamp as shown inFig. 1(c). The ink resist was transferred to the Cr substrateby IS method. The transferred resist pattern was used as

PDMS and spin coating of resist (a), PDMS coated with ink resist (b),te (d), Cr etching through the ink resist pattern (e), Cr pattern by IS process

B.-K. Choo et al. / Journal of Non-Crystalline Solids 354 (2008) 2879–2884 2881

etch barrier for patterning of Cr by wet etching as shownin Fig. 1(d)–(f).

The IS patterns with stripe shape were seen by an opticalmicroscope and atomic force microscope (AFM). Resist

Fig. 3. Optical (up) and AFM (down) images of the resist pattern by IS procePGMEA diluted resist.

pattern deviation between the PDMS stamp width andresist width was measured to study the IS process. And wecompared h-PDMS with s-PDMS as a master stamp mate-rial in IS process by an optical microscope to find better one.

ss using 1 wt% (a), 6 wt% (b), 10 wt% (c), and 20 wt% (d) novolak resin in

2882 B.-K. Choo et al. / Journal of Non-Crystalline Solids 354 (2008) 2879–2884

3. Results and discussion

Surfaces of s-PDMS and h-PDMS were modified to behydrophilic by the UV ozone treatment. The PDMS mod-ification to hydrophilic by the UV irradiation or by theplasma treatment was reported [9,10]. Surface of s-PDMSchanged into SiO2 like by UV ozone irradiation [11]. In thiswork the h-PDMS was used for IS process to avoid the col-lapse of the mold [12]. H-PDMS surface can also be chan-ged into hydrophilic as the s-PDMS surface. Modificationof h-PDMS needs more time than that of s-PDMS asshown in Fig. 2, indicating that the slope of h-PDMS is0.133 and s-PDMS is 0.167 (mN/m s). This means thatthe surface of h-PDMS changes slowly by the UV ozonetreatment than that of s-PDMS. Finally, the surface energyof s-PDMS and h-PDMS increases up to 85–90 mN/m

Fig. 4. Optical images of the resist pattern by IS process using PDMS stamp61.60 mN/m (d), 79.41 mN/m (e), and 89.56 mN/m (f).

(contact angle � 0�). It was found that the ink resist couldnot be coated well if the s-PDMS or h-PDMS had lowersurface energy below the 34 mN/m (contact angle > 68�).Note that the most of the resist diffused out of the sampleduring the spin coating.

The amount of novolak resin was changed from 1 to20 wt% in the resist solution. Note that the PDMS stampwas modified with a surface energy higher than 85 mN/m. All other process conditions were fixed; resist coatingrpm of 4500 for 35 s and contact time of 1 min. Fig. 3shows the resist patterns with stripe shape taken by opticaland AFM images as a function of the novolak resin wt%.The viscosity of novolak resin was high (500–1800 cP)but that of PGMEA solvent was low (1.1 cP). The viscosityof the resist increases with increasing novolak resin. Thismeans that intermolecular force of ink resist increases with

with surface energy of 17.76 mN/m (a), 33.44 mN/m (b), 48.33 mN/m (c),

B.-K. Choo et al. / Journal of Non-Crystalline Solids 354 (2008) 2879–2884 2883

increasing the novolak resin wt%, making resist patternsharper. When the novolak resin is less than 6 wt%, thecontact area could not be completely filled so that the resistpattern shape is not good as shown in Fig. 3(a) and (b).This means that the intermolecular force of ink resist islower than the adhesive force between the resist and Crbecause the surface energy of Cr is modified to be high(85 mN/m; contact angle of Cr � 0�) by UV ozone treat-ment. Intermolecular force of the resist increases withincreasing the novolak resin wt% so that the resist patternfilled with the resist as shown in Fig. 3(c) and (d).

The surface energy of the PDMS stamp was changedusing the UV ozone treatment and Fig. 4 shows the resistpattern made by the IS process as a function of surfaceenergy of PDMS stamp. All the other process conditionsare fixed; resist coating rpm of 4500 for 35 s and contacttime of 1 min. The pattern shape is getting better withincreasing the surface energy of the PDMS. If surfaceenergy of PDMS stamp is less than 33.44 mN/m, the resistcould not be coated well onto the PDMS surface so thatthe resist pattern shape was not good as shown inFig. 4(a) and (b). The resist did not fill the contact area

Fig. 5. Pattern width deviation from PDMS stamp width with increasing the inslope is �0.0012 (a), IS process diagram shows the pattern width deviation (DL

stamp (b).

Fig. 6. Optical images of the resist patterns by IS pr

because the amount of resist on the PDMS stamp is notenough for IS process. The key point of our experimentis the control of surface energy for IS process. Patternshape by IS process is dominantly affected by the surfaceenergy of the PDMS stamp.

We measured the width of the resist pattern deviation(DL) as a function of the resist coating revolution per min-utes (rpm) as shown in Fig. 5(a). All the other process con-ditions are fixed; contact time of 1 min and novolak resinof 20%. Fig. 5(b) indicates the DL, the resist pattern widthfrom PDMS stamp edge of contact point. Fig. 5 shows thatDL linearly decreased as increasing the resist coating rpmand its slope is �0.0012 (DL(lm)/rpm). The amount ofthe resist on the protrusion area of the PDMS stampdecreases with increasing the rpm. Therefore, the resist ofcontact area results in pattern deviation. The pattern devi-ation in IS process is significantly dependent on the amountof the resist during the process.

Fig. 6 shows the optical images of resist patterns usingh-PDMS and s-PDMS stamps. The collapse of the PDMSstamp make undesired resist pattern as shown in Fig. 6(a).Pattern collapse can be avoided by increasing the pattern

k resist coating rpm; dotted line is the linear regression of the plots and its) which means the patterned resist length from the contact edge of PDMS

ocess using s-PDMS (a) and using h-PDMS (b).

Fig. 7. Optical images of s-PDMS stamp with 3.75 lm line width of protrusion area and 8 lm pitch with 4.8 lm thickness (a), resist pattern made by ISprocess with 4.39 lm line width and 8 lm pitch (b), and Cr stripe pattern with 4.3 lm line width and 8 lm pitch after sub-layer etching and removing theink resist (c).

2884 B.-K. Choo et al. / Journal of Non-Crystalline Solids 354 (2008) 2879–2884

height. However, increasing the pattern height of the stampleads to the pattern deviation from the master pattern.Therefore, we used h-PDMS due to higher Young’s modu-lus [10]. A clear gate pattern was obtained by the h-PDMSstamp with 4.8 lm thick pattern. The pattern collapse didnot occur in 1000 lm width between patterns as shown inFig. 6(b).

We finally made a Cr stripe pattern with 4.3 lm linewidth and 8 lm pitch by IS process after sub-layer etchingand removing the ink resist as shown in Fig. 7(c). The finalpattern deviation between the PDMS mold and Cr patternwidth was 0.275 lm. And the pattern deviation (DL) of inkresist between the PDMS stamp and resist pattern was0.32 lm. This result indicates that the IS process can beused as an alternative method for photolithography tomake TFT backplane for LCD.

4. Conclusion

We introduced an ink stamp technique using a soft moldfor a new lithography technique. To increase adhesionbetween thin film and resist, the surface was modified byUV ozone treatment. Novolak resin wt% was adjustedand thickness of the resist was controlled by the spin coat-ing for process optimization. In this experiment, a striperesist pattern with a pattern deviation (DL) of 0.32 lmand 4.3 lm stripe pattern of Cr were achieved. Stampand substrate should be hydrophilic and 15 wt% PGMEAdiluted novolak resist was coated with appropriate spinspeed.

Acknowledgment

This research was supported by MOCIE (Ministry ofCommerce, Industry and Energy) of Korea, through nextgeneration growth engines and development strategies.

References

[1] S.S. Yoo, H.L. Cho, O.N. Kwon, S.H. Nam, Y.G. Chang, K.Y. Kim,S.Y. Cha, B.C. Ahn, I.J. Chung, in: The 5th International Meeting onInformation Display, vol. 22, 2005 p. 948.

[2] P. Calvert, Chem. Mater. 13 (2001) 3299.[3] W.S. Wong, S. Ready, R. Matusiak, S.D. White, J.-P. Lu, J. Ho,

R.A. Street, J. Non-Cryst. Solids 299 (2002) 1335.[4] Y. Mikami, Y. Nagae, Y. Mori, K. Kuwabara, T. Saito, H. Hayama,

H. Asada, Y. Akimoto, M. Kobayashi, S. Okazaki, K. Asaka,H. Matsui, K. Nakamura, E. Kaneko, IEEE Trans. Electron Devices41 (1994) 306.

[5] A.A. Darhuber, S.M. Troian, S. Wagner, J. Appl. Phys. 90 (2001)3602.

[6] S.M. Miller, S.M. Troian, S. Wagner, J. Vac. Sci. Technol. B 20(2002) 2320.

[7] X. Cheng, L.J. Guo, in: Proceedings of the 48th InternationalConference Electron, Ion, and Photon Beam Technology andNanofabrication, vol. 441, 2003 (Abstract).

[8] H. Kawata, Y. Hirai, H. Kikuta, Jpn. J. Appl. Phys. 43 (2004) 4027.[9] V.M. Graubner, R. Jordan1, O. Nuyken, R. Kotz, T. Lippert,

B. Schnyder, A. Wokaun, Polym. Mater: Sci. & Eng. 88 (2003) 488.[10] G.S. Ferguson, M.K. Chaudhury, H.A. Biebuyck, G.M. Whitesides,

Macromolecules 26 (1993) 5870.[11] B. Schnyder, T. Lippert, R. Kotz, A. Wokaun, V.M. Graubner,

O. Nuyken, Surf. Sci. 532 (2003) 1067.[12] T.W. Odom, J.C. Love, D.B. Wolfe, K.E. Paul, G.M. Whitesides,

Langmuir 18 (2002) 5314.