INSTITUTE OF PHYSICS PUBLISHING
NANOTECHNOLOGY
Nanotechnology 16 (2005) 2436–2441
doi:10.1088/0957-4484/16/10/074
Inkjet printing of nanosized silver colloids Hsien-Hsueh Lee, Kan-Sen Chou1 and Kuo-Cheng Huang Department of Chemical Engineering, National Tsing Hua University, Hsinchu 30013, Taiwan E-mail:
[email protected]
Received 27 April 2005, in final form 15 August 2005 Published 2 September 2005 Online at stacks.iop.org/Nano/16/2436 Abstract A water-based conducting ink, composed of well dispersed nano-silver particles, has been successfully inkjet printed using an ordinary commercial printer. The silver colloids with diameter around 50 nm were dispersed in a water and diethylene glycol cosolvent system. The 25 wt% silver ink had a viscosity of about 7.4 cP and surface tension of 33.5 dyn cm−1 at 20 ◦ C, which is appropriate for printing jobs. Continuous and smooth lines of 130 µm width could be printed on ordinary glass substrates using an Epson R210 printer. After baking at 260 ◦ C for 3 min, these lines exhibited a resistivity of 1.6 × 10−5 cm, which could serve as conducting lines for electronic applications. (Some figures in this article are in colour only in the electronic version)
1. Introduction In the electronic industry, fabrication of conductive lines (circuit) is vital and inevitable. Traditionally, for example, electroplating and etching processes accompanied by lithography technology are widely adopted in the printing circuit board (PCB) industry for manufacturing its circuits. However, this method is not only time consuming but also very complicated, because many steps are required to construct one layer of circuit. Moreover, the conventional electroplating and etching processes also produce large quantities of waste water. For these reasons, the development of convenient and fast processing techniques to fabricate conductive lines has attracted more and more attention in recent years. One of the most promising alternatives is the ink-jet printing technology, by which the conductive lines can be drawn (i.e. printed) onto the substrate in one step. There are two important components of the ink-jet printing technology: one is the mechanical system, i.e. the printer, and the other is the conductive material, i.e. the ink. Studies focused on the former item discussed the formation of droplets, accurate control of dropping location, jetting parameters, etc [1–5]. On the research of conductive ink, several materials were studied including molten metal, conductive polymer, and metallic nanoparticle suspension [6]. Among these, metallic nanoparticle suspension gained significant interest in recent years because it can be operated at room 1 Author to whom any correspondence should be addressed.
0957-4484/05/102436+06$30.00 © 2005 IOP Publishing Ltd
temperature [7] as compared to the molten metal, and also has a better performance in terms of conductivity than conductive polymers on the other hand. Viscosity and surface tension are the two most important properties of general inks [8]. Size, dispersion and stability of metallic nanoparticles are also crucial for the conductive ink system, because the nozzle of the printer would be clogged with larger particles or due to the aggregation of nanoparticles. In addition, the physical properties of metallic nanoparticles undoubtedly affect the performance of the product to a great extent. Due to their high conductivity and thermal stability, gold and silver nanoparticle suspensions were widely used in the studies of conductive ink. Some operating parameters in these studies are listed in table 1 [9–14] for comparison. In the work by Fuller et al in 2002 [9], Ag nanoparticles suspended in α-terpineol were printed by the piezo-driven ink-jet printer onto a heated substrate at 100–300 ◦ C. After sintering at 300 ◦ C for 10 min, the fabricated lines showed a resistivity of approximately 3 µ cm, about twice that of bulk silver (1.51 µ cm). However, in this case α-terpineol was used as the solvent and it may cause the problem of volatile organic compounds (VOCs) in realistic industrial applications. In the work of Bieri et al in 2004 [12], Au nanoparticles suspended in toluene were used. Conductive ink droplets were generated by the piezo-driven ink-jet printer and deposited onto a moving substrate to form continuous lines. Then focused laser irradiation was utilized to evaporate the solvent and sinter these Au nanoparticles. The result showed that micro-lines
Printed in the UK
2436
Inkjet printing of nanosized silver colloids
Table 1. Summary of prior works on metallic nanoparticle conductive ink.
Particle
Particle size (nm)
Solvent
Conc. (wt%)
Ag
5–7
α-terpineol
10
Au
2–4
Toluene
30
Au
2–4
Toluene
30–35
Au
2–5
Toluene
30
Au Ag Ag
5–20 1–10 10–50
Toluene Toluene Water-DEG
30 30–35 25
Curing condition ◦
100–300 C (hot surface) 300–400 ◦ C (laser curing) 50–500 mW (laser curing) 200–1000 mW (laser curing) 300 ◦ C 300 ◦ C 150–260 ◦ C
with resistivity of 14 µ cm, which is six times larger than that of bulk gold, were fabricated. Similar to Fuller’s work [9], it also had the problem of VOCs; moreover, toluene is often classified as a toxic substance. In this work, a conductive ink was based on Ag nanoparticles using solvents mainly composed of water and diethylene glycol, which are currently used in commercial colour ink. This composition is expected to be more environmentally friendly. Besides, our conductive ink can be directly used in a general commercial ink-jet printer without any modification of the mechanical system because the solvent system is the same. In other words, the equipment would be cheaper and readily available. It is hoped that the silver nanoparticle conductive ink developed in this work can be useful for future applications in the electronic industry.
2. Experimental details 2.1. Ink preparation The nanosized silver colloids used in this work were synthesized by the chemical reduction method using formaldehyde as the reducing agent. 0.1 M silver nitrate was reduced at room temperature by formaldehyde ([HCHO]/[Ag+ ] = 2.96) in the presence of protective agent, polyvinyl pyrrolidone (PVP) (molecular weight = 40 000). About 7.95 g PVP/g Ag was used for this work. To ensure complete reaction, a stoichiometric amount of NaOH was also added. After completion of chemical reduction, the suspension was washed with acetone, which precipitated out the silver colloids. This washing step was repeated twice more to finally recover relatively pure silver colloids, having diameter about 50 nm. The residual PVP content at this stage was about 5% and was small enough not to cause any problem in providing electrical conductivity after inkjet printing and mild temperature treatment. Other details of the synthesis experiment can be found elsewhere [15]. The recovered and concentrated silver colloid was dispersed by diethylene glycol (DEG) as a cosolvent to water (50 wt%) to various concentrations in an ultrasonic bath to form the ink for testing on the inkjet printer. Here we used atomic absorption spectroscopy (AAS) (Model 5100 PC, Perkin Elmer, USA) to determine the exact Ag content in the
Line width (µm)
Line thickness (nm)
ρ ( cm) −6
Author
80
100
3 × 10
20
50
1.4 × 10−5
Bieri et al [10]
123
250
4.5 × 10−6
Chung et al [11]
20–200
1.4 × 10−5
Bieri et al [12]
600 1000 532
1 × 10−5 3.5 × 10−5 1.6 × 10−5
Szczech et al [13] Szczech et al [14] This work
17 1000 120 130
Fuller et al [9]
suspension. In order to know more about the properties of such an ink, the viscosity and surface tension were also measured as a function of silver content. Here, a viscometer (model DV-III+/spindle CPE-40, Brookfield, USA) and tensiometer (K 10ST, Kruss, Germany) were used. The measuring temperature was maintained at 20 ± 0.1 ◦ C by a temperature controller (B 402, Firstek, Taiwan). 2.2. Printer set-up The commercial printer used here is an Epson R210 with a reported resolution of 5760 dpi. According to its manual, this printer has a piezoelectric head with 90 openings of size about 28 µm, and each droplet volume is of the order of 3 pl. Though the printer provides six containers for different coloured inks, we used only the black or yellow container for our silver ink in this work, while the rest of the hardware as well as the software remains unchanged. The line patterns were drawn by Auto CAD having length of 5 cm and designated widths of 50, 100, 200 and 300 µm. At both ends of the lines, a touch pad of 2 mm was made for electrical measurements. The parameters of printing quality, paper choice and printing speed were adjusted for the optimal printing effects. As for the ink, its silver content was varied from 5 to 35 wt% before an optimal value was determined. However, when 35% ink was used, a few of the 90 openings became blocked after printing for a while. A DEGwater solution (50:50) was used to clean it up. In general, our concentrated silver ink of 30 wt% or less was stable for up to at least two months, capable of producing successful printing results after a long period of storage. Nevertheless, after an overall evaluation of printing quality, the 25% ink was chosen for most of the printing experiments reported here. 2.3. Baking and electrical measurement A glass slide of 2.5 cm × 7.5 cm was used as substrate throughout this work. It was placed on top of a CD for the printing work. After printing the line patterns, the substrate was first dried in an oven at 70 ◦ C for 5 min to remove solvents. It was then further heated at various temperatures (150–260 ◦ C) for various times to follow the changes in electrical conductivity (milliohmmeter, model 502 BC, Zentech, Taiwan) as functions of temperature and 2437
H-H Lee et al
time. The samples were also observed by both optical and scanning electron microscopy (Model S-4700, Hitachi, Japan) techniques to examine their microstructures as well as measurements of line width and thickness.
22 20 shear rate = 150 (sec–1), 20o C
18
η (cp)
16 14
3. Results and discussions
12
3.1. Ink properties
10 8 6 4 2 5
10
15
20
25
30
35
40
Ag content (wt.%)
Figure 1. The increase of ink viscosity with concentrations of Ag colloids.
34.0
20oC
σ (dyne/cm)
33.8 33.6 33.4 33.2 33.0 0
10
20
Shown in figures 1 and 2 are the results on viscosity and surface tension of our silver ink at 20 ◦ C as a function of silver content. Clearly both properties increase with silver content. The use of DEG as a cosolvent helps to reduce the evaporation rate of solvent and thus avoid blocking at the head opening. It also helps to provide the ink with proper values of viscosity and surface tension for printing purposes. At low concentrations of colloids, the suspension’s viscosity basically follows the Einstein equation, µ/µs = 1 + 2.5φ (where φ is the solid volume fraction); in other words, the viscosity is linearly proportional to solid content. However, as the colloidal concentration continues to increase, the interaction between these colloids is no longer negligible and the viscosity increases rapidly. There are many empirical equations proposed in the literature to describe the relationship between measured viscosity and solid content [16]. As for our silver ink, it exhibited a similar trend to other ceramic suspensions. The relatively high viscosity of 35% silver ink (meaning higher resistance to flow) is probably one of the reasons for occasional blockage at the opening. As a reference, the viscosity and surface tension of the original inks for the Epson printer are in the range of 2–6 cP and 30–34 dyn cm−2 respectively.
30
Ag content (wt.%)
Figure 2. The increase of ink surface tension with concentrations of Ag colloids.
Premium Glossy Photo Paper High Speed
3.2. Printer parameter selection There are at least three parameters in the Auto CAD program to be chosen for different printing jobs. After a series of trial
Premium Glossy Photo Paper Low Speed
Figure 3. Comparison of printed features under different printing speeds.
2438
Inkjet printing of nanosized silver colloids
Figure 4. Actual line width in comparison with defined line width by Auto CAD as viewed by an optical microscope.
0.9
Line thickness (µm)
0.8
0.7
0.6
0.5
0.4 100
200
300
400
500
Line width (µm)
Figure 5. Line thickness as measured by SEM versus actual line width.
and error, it was found that the choice of ‘best printing quality’ provides a better resolution, probably due to more ink dots. As for the choice of paper, the selection of ‘premium glossy photo paper (PGPP)’ helps to avoid the break-up of printed lines, also due to more ink dots. Finally, the printer is usually set at high speed for the printing work. Under this circumstance, there appeared more satellite points. When this function was ‘off’ (i.e. giving low speed), the printed figure exhibited continuous and smooth lines. A comparison is shown in figure 3. It took about 35 s to print 20 lines of various widths in the area of 2.5 cm × 5 cm under this condition under the above settings. 3.3. Line width and thickness Shown in figure 4 are the actual widths of printed lines in comparison with the widths set by the Auto CAD software. Clearly, when the width was set at 50 µm, one could only obtain lines of 100–130 µm with many breaking points along the line. When the line width was set at 100 µm or higher, one could always obtain continuous lines without any breaking points as exhibited in figure 4. However, the actual widths as observed under the optical microscope were greater to some extent than the setting values. It is believed that this discrepancy
between setting value and actual width is dependent upon the substrate. In other words, if an ordinary paper is used here, this difference might be smaller because some of the solvent would be adsorbed by the substrate here. Next shown in figure 5 are the thickness data of these lines as observed under SEM. It showed a monotonic increase with line width from about 530 to 790 nm, due to more inks inkjetted during printing. The results presented here are thickness after baking treatment (260 ◦ C, 3 min). The original thickness should be somewhat greater because of shrinkage after thermal treatment. A simple calculation shows that the cross section area of each line, i.e. equivalent to the quantity of ink, is proportional to the line width. 3.4. Thermal treatment and electrical resistivity Here an ordinary oven was used to treat our printed lines to remove the solvent and to cause some sintering effect between nanosized silver colloids in the ink. The electrical resistivities of these lines after heat treatment as functions of temperature and time are exhibited in figure 6. Generally, the resistivity decreased with either heating temperature or time, showing a sintering effect for nanosized silver colloids 2439
H-H Lee et al
Although one could prolong the treatment time to reach low resistivity using low temperatures, it was still felt that heating the sample at 260 ◦ C for only 3 min would be a reasonable choice. The electrical resistivity was 1.6 × 10−5 cm after this thermal treatment. Finally, in figure 8, the test patterns having different widths on a glass slide were shown for an overall view of the results achieved here. When the line width setting is above 100 µm, the results are repeatable within the normal experimental error range (±5%) in terms of standard deviation of electrical resistivity. This repeatability improves with either heating temperature of heating time.
7e-5 150oC
6e-5
Resistivity (Ω-cm)
200oC 5e-5
260oC
4e-5 3e-5 2e-5 1e-5 0 0
10
20
30
40
50
60
Time (min)
Figure 6. Electrical resistivities after heating at different temperatures for various times (actual line width = 130 µm for these data).
in the ink. Exhibited in figure 7 are the SEM pictures of surface morphology after different temperature treatments for 45 min each. Here one could easily notice the sintering effect with temperature. At 70 ◦ C, each silver particle seemed to be unchanged and could be clearly observed. Yet after heating at 150 ◦ C for 45 min, the sintering effect became obvious and the size of particles increased significantly. Above 200 ◦ C, most boundaries between particles disappeared due to sintering and consequently the electrical resistivity decreased to low values.
4. Conclusion Nanosized silver colloids, synthesized by chemical reduction with formaldehyde, were dispersed by a cosolvent system made of diethylene glycol and water to get a 25 wt% silver ink. This ink was stable up to at least two months and could be successfully inkjet printed by a commercial printer (Epson R210) onto glass substrates. Continuous and smooth lines were formed when the line width was set at 100 µm or greater. The actual width would always be somewhat greater than the setting value. For a line with actual width of 130 µm, thickness of 530 nm and heated at 260 ◦ C for 3 min, it exhibited a resistivity of 1.6 × 10−5 cm, a value close to the bulk resistivity of silver (1.51 µ cm), and could serve as conducting lines for electronic applications.
70°C
150°C
200°C
260°C
Figure 7. SEM pictures of surface morphology after different temperature treatments.
2440
Inkjet printing of nanosized silver colloids
Figure 8. Test patterns having various line thicknesses on a glass slide.
Acknowledgments The authors wish to thank Mr Yun-Chien Chang of Beader Technology Co., Ltd, Taiwan, for valuable information and suggestions. Financial support from the National Science Council (NSC 90-2214-E007-010) and Chun-Shan Institute (grant BD 93020P) are also appreciated.
References [1] Percin G and Khuri-Yakub B T 2003 Rev. Sci. Instrum. 74 1120 [2] Huang D and Kim E S 2001 J. Microelectromech. Syst. 10 442 [3] Tseng F G, Kim C J and Ho C M 2002 J. Microelectromech. Syst. 11 427 [4] Tseng F G, Kim C J and Ho C M 2002 J. Microelectromech. Syst. 11 437
[5] Meinhart C D and Zhang H 2000 J. Microelectromech. Syst. 9 67 [6] Calvert P 2001 Chem. Mater. 13 3299 [7] Chou K S, Huang K C and Lee H H 2005 Nanotechnology 16 779 [8] Kang H R 1991 J. Imaging Sci. 35 179 [9] Fuller S, Wilhelm E J and Jacobson J M 2002 J. Microelectromech. Syst. 11 54 [10] Bieri N R, Chung J, Haferl S E, Poulikakos D and Grigoropoulos C P 2003 Appl. Phys. Lett. 82 3529 [11] Chung J, Ko S, Bieri N R, Grigoropoulos C P and Poulikakos D 2004 Appl. Phys. Lett. 84 801 [12] Bieri N R, Chung J, Poulikakos D and Grigoropoulos C P 2004 Superlattices Microstruct. 35 437 [13] Szczech J B, Megaridis C M, Zhang J and Gamota D R 2004 Microscale Thermophys. Eng. 8 327 [14] Szczech J B, Megaridis C M, Gamota D R and Zhang J 2002 IEEE Trans. Electron. Packaging Manufact. 25 26 [15] Chou K S and Lai Y S 2004 Mater. Chem. Phys. 83 82 [16] Reed J S 1995 Principles of Ceramic Processing 2nd edn (New York: Wiley) chapter 17
2441
