Langmuir 2005, 21, 8099-8102
8099
Site-Selective Direct Photochemical Deposition of Copper on Glass Substrates Using TiO2 Nanocrystals Kensuke Akamatsu,*,†,‡ Akihiro Kimura,‡ Hiroaki Matsubara,‡ Shingo Ikeda,‡ and Hidemi Nawafune*,†,‡ Faculty of Science and Engineering, Graduate School of Science, Konan University, 8-9-1 Okamoto, Higashinada, Kobe 658-8501, Japan Received May 16, 2005. In Final Form: July 14, 2005 Deposition of copper thin films was achieved by a photocatalytic reaction of site-selectively adsorbed TiO2 nanocrystals for direct fabrication of copper circuit patterns on glass substrates. The nanocrystal monolayers absorbed on hydrophobic surface templates serve as an effective photocatalyst, producing metallic copper and formic acid via oxidation of methanol in solution. The formic acid generated has also been suggested to serve as an electron donor that accelerates copper deposition through a UV-mediated autocatalytic reaction, even after nanocrystals are embedded into the grown copper films. The thickness of the deposited copper films was easily controlled by varying the UV irradiation time, irradiation power, and initial concentration of methanol as a hole scavenger. The process presented herein provides an effective methodology for resist-free, direct metallization of insulating substrates.
Introduction Fabrication of metallic circuit patterns on insulating substrates is extremely important for the manufacturing processes of electronic, optical, and mechanical devices ranging from displays to biosensors.1,2 These metallic patterns are generally fabricated using photolithography, electron beam lithography, and focused ion beam lithography.3-5 Although these methods are well established and capable of precise fine patterning, the development of a facile process for fabrication of metallic patterns still remains a great challenge. Typical conventional approaches for the deposition of metals include electroless deposition, physical vapor deposition, and sputtering. Among these, the simplicity of wet chemical deposition process and its compatibility with various different kinds of substrates with complex shapes make it particularly attractive for the low-cost fabrication of metallic interconnects in the microelectronics industry.6 Therefore, there are a variety of reports for fabricating metallic patterns and devices on both rigid and flexible substrates, based on site-selective metal deposition.7 Copper is well-known to be the most useful candidate for metallization, due to its low electrical resistivity and * To whom correspondence should be addressed. E-mail:
[email protected]. † Faculty of Science and Engineering, Konan University. ‡ Graduate School of Science, Konan University. (1) Xia, Y.; Rogers, J. A.; Paul, V. R.; Whitesides, G. M. Chem. Rev. 1999, 99, 1823. (2) Holdcroft, S. Adv. Mater. 2001, 13, 1753. (3) VLSI Fabrication Principle; Gandhi, S. K., Ed.; Wiley: New York, 1994. (4) The Science and Engineering of Microelectronic Fabrication; Campbell, S. A., Ed.; Oxford University Press: New York, 1996. (5) Fundamentals of Microfabrication, 1st ed.; Madou, M. J., Ed.; CRC Press: Boca Raton, FL, 1997. (6) Modern Electroplating, 4th ed.; Schlesinger, M., Paunovic, M., Eds.; Wiley: New York, 2000. (7) (a) Hidber, P. C.; Helbig, W.; Kim, E.; Whitesides, G. M. Langmuir 1996, 12, 1375. (b) Rogers, J. A.; Bao, Z.; Baldwin, K.; Dodabalapur, A.; Crone, B.; Raju, V. R.; Kuck, V.; Katz, H.; Amundson, K.; Ewing, J.; Drzaic, P. Proc. Natl Acad. Sci. U.S.A. 2001, 98, 4835. (c) Delamarche, E.; Vichiconti, J.; Hall, S. A.; Geissler, M.; Graham, W.; Michel B.; Nunes, R. Langmuir 2003, 19, 6567. (d) Huang, D.; Liao, F.; Molesa, S.; Redinger, D.; Subramanian, V. J. Electrochem. Soc. 2003, 150, G412. (e) McAlpine, M. C.; Friedman, R. S.; Lieber, C. M. Nano Lett. 2003, 3, 443.
high electromigration resistance, as compared with aluminum. Deposition of copper circuit patterns on dielectric substrates has been of great importance to electronic applications, and therefore has been extensively investigated.6,8 However, since the conventional plating bath for electroless copper deposition involves toxic chemical reagents, the discharge from the deposition bath can be a serious environmental issue. Therefore, the development of a facile process that enables resist-free, direct fabrication of metallic patterns as well as low environmental toxicity is an important challenge for the fabrication of circuit patterns. Low-temperature fabrication of metallic circuits is also desirable for the realization of processes with lowenergy consumption. Herein, we report a novel process for the fabrication of copper circuit patterns on glass substrates. This approach relies on a photocatalytic reaction using TiO2 nanocrystals adsorbed on surface templates of organic monolayers. TiO2 is a photocatalyst that is well-known to exhibit a strong oxidizing ability,9 whereby organic species are oxidized to carbon dioxide, and the redox chemistry of TiO2 nanocrystals has been studied in detail.10 In addition, there are several studies concerning the use of TiO2 as catalyst for reducing metallic ions, e.g., reduction of copper ions in TiO2 colloidal suspensions,11 reduction of cadmium ions on nanoparticles modified with amino acids,12 and formation of copper thin films on ion-doped poly(amic acid).13 (8) (a) Hanna, F.; Hamid, Z. A.; Aal, A. A. Mater. Lett. 2003, 58, 104. (b) Kim, J. J.; Cha, S. H.; Lee, Y. S. Jpn. J. Appl. Phys. 2003, 42, L953. (c) Zangmeister, C. D.; van Zee, R. D. Langmuir 2003, 19, 8065. (d) Delamarche, E.; Vichiconti J.; Hall, S. A.; Geissler, M.; Graham, W.; Michel B.; Nunes, R. Langmuir 2003, 17, 6567. (e) Vaskelis, A.; Norkus, E.; Stalnioniene, I.; Stalnionis, G. Electrochim. Acta 2004, 49, 1613. (f) Wang, Z.; Yaegashi, O.; Sakaue, H.; Takahagi, T.; Shingubara, S. J. Electrochem. Soc. 2004, 151, C781. (g) Li, J.; Hayden, H.; Kohl, P. A. Electrochim. Acta 2004, 49, 1789. (h) Kim, Y.; Bae, D.; Yang, H.; Shin, H.; Wang, G. W.; Senkevich, J. J.; Lu, T. M. J. Electrochem. Soc. 2005, 152, C89. (9) Fujishima, A.; Honda, K. Nature 1972, 238, 37. (10) (a) Dimitrijevic, N. M.; Rajh Tijana.; Saponjic, Z. V.; Garza, L.; Tiede, D. M. J. Phys Chem. B 2004, 108, 9105. (b) He, Chun.; Xiong, Y.; An, T.; Zhu, X. J. Environ. Sci. Health 2003, A38, 949. (11) Yamazaki, S.; Iwai, S.; Yano, J.; Taniguchi, H. J. Phys. Chem. A 2001, 105, 11285. (12) Ruvarac-Bugarcic, I. A.; Saponjic, Z. V.; Zec, S.; Rajh, T.; Nedeljkovic, J. M. Chem. Phys. Lett. 2005, 407, 110. (13) Ikeda, S.; Akamatsu, K.; Nawafune, H. J. Mater. Chem. 2001, 11, 2919.
10.1021/la051298f CCC: $30.25 © 2005 American Chemical Society Published on Web 08/04/2005
8100
Langmuir, Vol. 21, No. 18, 2005
Scheme 1. Schematic Representation of the Present Process for Fabrication of Copper Patterns on Glass Substrates
Letters the glass surface and the quartz plate was set at 50 µm using a polymer spacer film. The lamp was at a distance of approximately 5 cm above the substrate surface (260 mW cm-2). The surface morphology of the films was observed using a scanning electron microscope (SEM, JSM-6340F, JEOL) operating at 15 kV. The amount of deposited copper was quantified by inductively coupled plasma (ICP) spectroscopy after dissolving the deposited copper film in dilute nitric acid solution (5%). The chemical species in the deposition solution were characterized by a Fourier transform infrared spectrometer (FT/IR 670, Japan Spectroscopic Co.) equipped with an attenuated total reflection (ATR) attachment. The spectra were recorded at a resolution of 4 cm-1 with 50 accumulations. X-ray photoelectron spectroscopy (XPS) was used to characterize the TiO2 nanocrystals using ESCA 750 spectrometer (Shimadzu) with a Mg KR X-ray source. For all of the XPS measurements, we used Au 4f photoelectron peak (84.0 eV for Au0) as reference to correct electron charging up effects by coating approximately a half of sample surface with thin gold films deposited by vacuum vapor deposition.
Results and Discussion
These reports have motivated the attempt to use the TiO2 as photocatalyst for fabrication of metallic circuits on the insulating substrates. The present approach works by the formation of surface templates of hydrophobic monolayers on a glass substrate, selective adsorption of TiO2 nanocrystals on the hydrophobic monolayers, followed by photocatalytic deposition of copper (Scheme 1). The process described herein is thought to be an entirely additivebased photochemical method, enabling the direct fabrication of copper circuit patterns on glass substrates, without requiring posttreatment modifications. Experimental Section All reagents used were of analytical grade and used as received. In a typical procedure, bare glass substrates (nonalkali glass AN100, Asahi Glass Co., Japan) were treated with n-octadecyl trimethoxysilane (OTS) by chemical vapor deposition (CVD) according to the procedure previously described.14 The substrates were sealed with OTS solution in a Teflon box at 10% humidity, followed by heat treatment at 170 °C for 90 min. The substrates were then rinsed with toluene and methanol three times and dried using a stream of nitrogen. The deposition of OTS layers was characterized by water contact angle measurements (102° after CVD process), the value of which was in good agreement with that reported for OTS monolayers deposited on Si substrate.14b The OTS-coated substrates were then subjected to irradiation with vacuum ultraviolet (VUV) light through a metalon-quartz photomask for 20 min at 10 Pa using a VUV illumination system (main wavelength: 176 nm) in order to selectively decompose the OTS monolayers. The patterned substrates were immersed into 3.0% sol solution (Ishihara Techno. Ltd., STS-01 anatase TiO2 with ca. 20 nm diameter) at pH 8.0 for 1 min in order to adsorb the TiO2 nanocrystals. A quantity of 100 µL of water/methanol mixed solution containing 0.1 M (CH3COO)2Cu and 50 mM CH3COOH/CH3COONa buffer (pH 7.0) was dropped on the TiO2-adsorbed substrate. After being placed on a quartz plate (without metallic patterns), the samples were irradiated with ultraviolet light at various irradiation times in air at room temperature using a fluorescent lamp (Spot Cure ST-7, Ushio, Inc., main wavelength: 365 nm) with a heat ray cut filter. The distance between (14) (a) Sugimura, H.; Hozumi, A.; Kameyama, T.; Takai, O. Surf. Interface Anal. 2002, 34, 550. (b) Hong, L.; Sugimura, H.; Furukawa, T.; Takai, O. Langmuir 2003, 19, 1966.
The surface templates used consisted of OTS monolayer patterns on glass substrates in which the OTS in the area exposed to VUV light decomposed,14 resulting in the formation of the glass substrate bearing hydroxyl groups. In the current approach, the TiO2 nanocrystals were adsorbed only to the surface of the OTS monolayers and not to the bare glass surface. This is thought to be due to electrostatic repulsive interactions between hydroxyl groups on the glass substrate (negative surface potential at pH 8.0) and on the surface of the TiO2 nanocrystals (negative potential at pH 8.0).15 This site-selective adsorption was confirmed by X-ray photoelectron spectroscopy (XPS) after both substrates were immersed in solution containing TiO2 nanocrystals. Results indicated that Ti 2p photoelectron signals were successfully detected for OTS-modified substrates, whereas signals were not evident for the VUV-irradiated substrate (results not shown). The amount of TiO2 nanocrystals adsorbed on the OTS layer was estimated to be 9.1 × 10-14 mol cm-2 using ICP measurements16 and was not found to be dependent on the pH during adsorption. Under these conditions, the photocatalytic reaction to reduce copper ions could occur only on the surfaces in which the nanocrystals were physisorbed (OTS layers), thus enabling site-selective deposition of copper thin films. The motivation lies in the assumption that the reduction of copper ions occurs preferentially to the generation of hydrogen molecules due to the more positive standard electrode potential of copper ions. The successful site-selective reaction was verified by SEM (Figure 1A). The image confirms that copper (5 µm in width) was deposited on the surface of OTS monolayers with adsorbed TiO2 nanocrystals. Pattern dimensions were in good agreement with the photomask specifications, being essentially dependent on the resolution of the metal-on-quartz photomask used for VUV lithography. The same deposition processes were conducted in the absence of both UV irradiation and TiO2 nanocrystals as a control. In these control experiments, no deposition was observed, thus indicating a photocatalytic reaction initiated by TiO2 nanocrystals. The present approach can be described as a resist-free, completely (15) Bourikas, K.; Vakros, J.; Kordulis, C.; Lycourghiotis, A. J. Phys. Chem. B 2003, 107, 9441. (16) The amount of adsorbed Ti (atom) determined by ICP measurements was 1.3 × 10-7 mol cm-2 after dissolution of the adsorbed nanocrystals in hydrochloric acid. We estimated the amount of the adsorbed nanocrystals using the amount of adsorbed Ti, unit cell volume of anatase TiO2, and mean nanocrystal size (20 nm).
Letters
Langmuir, Vol. 21, No. 18, 2005 8101
Figure 2. Changes in the thickness of deposited copper as a function of UV irradiation time for solution containing different concentrations of methanol.
Figure 1. (A) SEM image of the copper patterns on glass substrate. The copper thin film was deposited via UV irradiation for 120 min. The brighter regions correspond to the metallized regions. Scale bar: 20 µm. (B-D) SEM images of the copper thin films on glass substrates. The films were deposited via UV irradiation for 0.5 (B), 2 (C), and 4 h (D). Initial methanol concentration (0.1 vol %) Scale bar: 1 µm.
additive-based photochemical strategy for the direct fabrication of metallic circuit patterns on glass substrates. Figure 1B-D shows SEM images of the deposited films grown under various UV irradiation times. Initially, small copper nanoparticles (mean diameter of ca. 50 nm) can be seen forming on the surface of the glass substrate (Figure 1B). As the reaction proceeds, the nanoparticles gradually grow, until a continuous film is formed (Figure 1C,D). No cracks or peeling was observed for these films.17 If desired, thicker films could be achieved by subsequent electro- or electroless plating reactions. The deposited copper thin films obtained after UV irradiation for 4 h exhibited an electrical conductivity of 1.5 × 105 S cm-1. The value is relatively lower than that of bulk copper (6.0 × 105 S cm-1). This can be due to relatively large grains of deposited copper, which may be loosely connected each other.18 To explore the possibility for decomposition of OTS layers in nonirradiated regions due to the diffusion of ozone generated during VUV-lithography, we compared the amount of the adsorbed nanocrystals per unit area of “OTS region” for substrates modified with OTS layers (without VUV lithography) and its patterns (after VUV lithography, the area of OTS region was determined by known photomask features). Consequently, the amount was almost the same for both samples (9.1 × 10-14 mol cm-2 and 8.9 × 10-14 mol cm-2 for samples before and after VUV patterning, respectively), indicating that this effect is not so significant in the present experimental conditions regarding with the amount of adsorbed TiO2 nanocrystals on OTS monolayers. We also tried to form patterns of adsorbed TiO2 nanocrystals by VUV irradiation for glass substrate (17) The Scotch-tape adhesion test revealed that the films were not so strongly adhesive, possibly because of physical interaction between TiO2 nanocrystlas and OTS monolayers. We believe that the films having reliable adhesive strength could be obtained by appropriate selection of the functional groups of the template monolayers, e.g., -OH or -NH2 groups. The study of this effect is now in progress and the results will be published elsewhere. (18) Preliminary experiments revealed that the annealing of the deposited films at 150 °C in an inert atmosphere resulted in slight increase in conductivity up to 3.5× 105 S cm-1.
without OTS monolayers. However, although patterns of the hydrophilic surface were created, selective adsorption of the nanocrystals was not achieved. Only a small amount of the nanocrystals was adsorbed on the nonirradiated region (bare glass surface), possibly due to nonspecific adsorption of the nanocrystals. This also resulted in nonuniform deposition of small amount of copper after UV-mediated electroless deposition. Therefore, OTS monolayer templates play a role in uniformly adsorbing TiO2 nanocrystals. Additionally, copper patterns could also be formed by irradiating UV light through the metal-onquartz photomask onto a substrate with TiO2 nanocrystals adsorbed over the entire surface (without VUV patterning of OTS layers). However, pattern resolution was not clear, due to spreading of the irradiated light through the deposition solution. This observation demonstrates that initial patterning of OTS monolayers is essential to confine the region at which the photocatalytic reaction could occur in order to achieve the formation of high-resolution circuit patterns. In addition, the deposition rate is dependent on several parameters including the irradiation power (distance from substrate), and the concentration of the solution; in general, higher irradiation power and higher concentrations of (CH3COO)2Cu and methanol gave a greater deposition rate. Figure 2 shows typical results of the changes in the thickness of deposited copper thin films calculated from the amount of deposited copper (from ICP measurement) as a function of irradiation time. The film thickness gradually increases as the irradiation time increases, and then becomes saturated after 4 h of irradiation. This finding demonstrates that the thickness of the deposited copper films can be controlled by simply varying the UV irradiation time. Deposition of copper films of ca. 1.1 µm could be achieved after 4 h of irradiation at initial methanol concentration of 0.3 vol %. In an effort to explore the reaction mechanism of the growth of copper films, FTIR-ATR measurements were performed in order to characterize the chemical species in the solution during reaction. The results are shown in Figure 3. As the UV irradiation time increases, the CdO stretching band of acetic acid (1640 cm-1) remained unchanged, whereas intensity of the methanol C-O stretching band (1020 cm-1) decreased and a new band, which could be assigned to aldehyde CdO stretching, appeared at 1700 cm-1. These results suggest the following local cathodic reaction on the nanocrystals
Cu2+ + 2e- f Cu0
(1)
8102
Langmuir, Vol. 21, No. 18, 2005
Letters
Figure 3. FTIR-ATR spectra of the solution used for UVmediated electroless copper plating before (a) and after UV irradiation for 1 h (b), and 4 h (c). Initial methanol concentration (0.1 vol %).
and local anodic reactions
CH3OH + HO• f HCOOH + 3H+ + 3e-
(2)
or
CH3OH + HO• f CO2 + 5H+ + 5e-
(3)
The copper ions can be reduced by electrons generated in the TiO2 nanocrystals via photoexcitation (eq 1), and methanol rather than carboxylic acid can be oxidized by holes or hydroxyl radicals. Subsequent agglomeration of copper atoms produces copper thin films. The methanol radicals generated react with water to produce formic acid (eq 2), which is the origin of the aldehyde band in the FTIR spectra (Figure 3). The reaction also generates carbon dioxide (eq 3), which was confirmed by gas chromatography (no hydrogen was detected). In addition to photoexcited electrons, the electrons generated via reactions 2 and 3 can also work effectively for the reduction of copper ions during UV irradiation, supporting the fact that the deposition rate increased as the methanol concentration increased (Figure 2). To investigate the reaction process in detail, XPS measurements were conducted for deposited films after several different UV irradiation times. Representative results are shown in Figure 4. Notable among these results, Ti 2p photoelectron signals could be observed after UV irradiation for 30 min, whereas the signals were no longer evident after 1 h. This indicates that the TiO2 nanocrystals were embedded in the grown copper films. Further, these findings demonstrate that electroless deposition of copper could continue even after the nanocrystals were embedded (the films were observed to grow even after 1 h, Figure 2). It should be noted here that in preliminary experiments the deposition of copper took place on copper films under UV irradiation without TiO2 nanocrystals when a solution
Figure 4. XPS Ti 2p spectra of the TiO2-adsorbed substrate after UV irradiation for 30 min (a), 1 h (b), and 4 h (c). Initial methanol concentration (0.1 vol %).
containing formic acid was used (no deposition occurred in the absence of UV irradiation). Therefore, the deposition of copper after embedding of the nanocrystals can be explained by the following reaction:
Cu2+ + HCOOH f Cu0 + CO2 + 2H+
(4)
This reaction is thought to be catalyzed by deposited copper with the aid of UV irradiation. Upon initial UV irradiation, reactions 1-3 dominate, and the additional reaction 4 could effectively proceed to accelerate the deposition of copper thin films. Further experimental work is currently underway in order to fully understand the reaction mechanism. Conclusion Herein, the reduction of copper ions using TiO2 nanocrystals as a photocatalyst for the formation of copper thin film on glass substrates has been demonstrated. Siteselective copper deposition was also achieved for the fabrication of copper patterns by the initial selective adsorption of nanocrystals on organic monolayer templates. The addition of methanol as a hole scavenger into the deposition solution was observed to effectively accelerate the deposition rate. This process is thought to offer an attractive alternative to conventional patterning strategies, providing a new direct metallization scheme based on photocatalytic chemistry. The process can also be applied to other metals for direct metallization and the fabrication of metal wiring on other insulating substrates. The effort to explore such extensions is currently in progress. Acknowledgment. This work was supported by the Foundation for Technology Promotion of Electronic Circuit Board, Japan. We acknowledge Prof. T. Nishino for allowing us to use the X-ray photoelectron spectrometer. S. I. is grateful for research fellowship support from the Japan Society for the Promotion of Science (JSPS). LA051298F
