DoD-Printing of Conductive Silver Tracks Dominik Cibis, Klaus Krüger Institute of Automation Technology, Helmut-Schmidt-University / University of the German Armed Forces, Holstenhofweg 85, 22043 Hamburg, Germany Dominik Cibis: Phone +49/(0)40/6541-2740, Fax +49/(0)40/6541-2004, E-Mail: dominik.cibis@hsu-hh.de Klaus Krüger: Phone +49/(0)40/6541-2722, Fax +49/(0)40/6541-2004, E-Mail: klaus.krueger@hsu-hh.de Abstract The Inkjet-Printing principle shows high potential for new applications besides conventional graphic printing. Based on a drop-on-demand (DoD) micro feeding system, a colloidal silver ink printer was developed, which makes it possible to print conductive silver tracks. The printer has a stationary piezo-driven print head with a nozzle diameter of 100 microns. Its drop-rate can be adjusted between 50 and 5000 drops per second. Printing of defined structures is realized with an x-y-planar motor which positions the substrate. As a printing medium, a colloidal silver ink is used. The colloidal suspension consists of a dispergent (e.g. terpineol or ethylene glycole), a dispersant additive and silver particles. The main particle diameter of the silver powder is below 1 micron. The percentage of such powder in the colloid reaches up to more than 20% by weight. The system reacts sensitively to changes of the colloid properties. Besides the particle load, the viscosity and surface tension of the dispergent show significant influences. Size and duration of the piezoelectric pulse as well as nozzle temperature are suitable control parameters. Corresponding to the surface tension of the colloids, the spot size of drops printed on alumina substrate varies between 125 and 250 microns. By heating the substrate up to 150 °C, different surface profiles of the conductive tracks can be realized. Conductive paths are printed and then characterized by conductivity measurement, light microscopy and atomic force microscopy. The thickness of the fired conductive tracks is about 2 microns when printing single layer. Key words: piezo print head, drop-on-demand, ink-jet printing, conductive ink, piezo pulse, thick-film ______________________________________________________________________________________________ 1. Introduction Over the last years there has been a great deal of success in extending the use of ink-jet printers beyond graphic applications. They have been used for drug discovery, for multiple automated micro-titrations, for picoliter fluid dispensers and for combinatorial devices in medicinal devices. They have also been used to pattern functional nano- and microstructures from layers of surfactant and to deposit cell adhesion proteins in tissue engineering. It has also been possible to build three-dimensional structures by using ink-jet printable powders. Ink-jet printer technology has another application in the production of flat screen displays. Piezoelectric and thermal drop-on-demand printers have been used successfully for forming shapes with ceramic suspensions. High temperature drop-on-demand nozzles can be used for depositing molten wax-based ceramic suspensions with up to 40 vol-% powder onto alumina [1]. The assessment of these developments leads to the view that ink-jet printers have a great range of possible applications and many different inks can be delivered by the micro print head. Theoretical models suggest that the printability of a special ink is influenced by ink viscosity, surface tension and density. In fact, ink viscosity is the limiting factor for printability and has to be suitable to the used print head. By heating up the nozzle the ink viscosity can be adjusted in a special range. Furthermore, the suspension needs to be well-dispersed to avoid aggregation and agglomeration. In all drop-on-demand printers the shape and form of the electrical piezo pulse that activates the drop formation is control variable. Recent work has demonstrated the high potential of drop-on-demand printers and shows that printing of metal containing powder suspensions can be achieved by preparing an ink suitable to the used print head. An important part for the success of the printing process is the formulation of the ink, based on an organic dispergent. The present work shows the possibility to deposit silver containing ink on alumina substrate to produce conducting paths. 2. Experimental details 2.1. The printing system The printer developed at the Institute of Automation Technology (IfA) consists of a stationary piezodriven print head [2], an x-y-planar motor to position the substrate and an integrated heating plate such that it is possible to heat the substrate up to over 150 °C. A piezo-based ink-jet print head was taken because it offers a greater range of ink compatibility than for example thermal ink-jet heads, which are limited in the use of inks or require a new design for each different type of ink suspension [3]. For the assembly of defined structures, two possibilities arise: Moving the print head and leaving the substrate unmoved or moving the substrate along with the heating plate. In the case presented here the moving-substrate choice was preferred to prevent vibrations on the print head and to gain accuracy. There are two operation modes for the print head: Coupled with the planar motor to print defined structures or observation mode which means, droplet formation can be recorded. In observation mode the maximum drop rate of the printer is 5 kHz and can be adjusted continuously down to 50 Hz. When operating the printer in structure mode the droplet frequency goes down to 25 to 50 Hz, dependend on the center distance between single dots. Reason for that is the interconnection of the trigger pulse between the planar motor and the micro print head. Each trigger pulse signal generates one piezo pulse signal to form one droplet. Figure 1 shows the printer in structure mode configuration. The specifications of the planar motor allow for a positioning accuracy of 15 µm, a repeat accuracy of 1 µm and a resolution of 1 µm per step. The distance between nozzle and substrate is generally 1 mm when printing, but can be varied from 1.5 cm down to less than 1 mm. With the specifications described above, different printing patterns can be realized. The distances between single droplets can be adjusted as shown schematically in figure 2. Starting with pattern (a) center distances are reduced in x and/or y direction to finally arrive at pattern (e) for conductive paths. Because of the heating plate underneath the substrate which enables drying of the printed dots within seconds, multilayer printing is also possible. Figure 2: Printing patterns 2.2. Further instruments To complete the laboratory assembly at the Institue of Automation Technology further instruments used in the printing process are an ultrasonic pulse instrument to stir up the samples being printed, an uv/vis-spectral photometer for sedimentation tests, a sintering oven and a light and atomic force microscope for optical characterization of the printed spots and paths. The printed conductive paths are finally electrically characterized with a multimeter. 3. Results and discussion Figure 1: Printer in structure mode configuration 3.1. Characterization of the dispergent In usual solutions encountered on chemical level, the solute and solvent molecules are of comparable size and it is normally assumed that the solute molecules are, on average, dispersed uniformly throughout the solvent. There is an important class of materials, however, in which the kinetic units that are dispersed through the solvent are very much larger than the molecules of the solvent. Such systems are called colloidal dispersions. If a substance A is insoluble in substance B, it will usually be possible to break A down into small particles that can be distributed more or less uniformly through the substance B. Substance A is then called the disperse phase and substance B the dispersion medium or dispergent [4]. To print conductive paths first of all a conductive ink, a colloid suspension, has to be prepared. Therefore at least two components are necessary: Particles which allow for current conduction after the sintering process and a suitable dispergent as dispersion medium. Later on, to get a stable suspension with a homogeneous particle distribution, a dispersant additive is required. This ingredient acts as a helpful means to avoid or at least minimize aggregation, agglomeration and sedimentation of the particles in the dispergent. On the way to develop a printable conductive ink for the used piezo print head initially a dispergent has to be characterized and selected for the printing process. Six different dispergents are taken into account being potentially possible and compatible with the print head: Ethylene glycol, terpineol, tetradecane, limonene, ethanole and texanole. Characterization proceedings are pointing out ethylene glycole, terpineol and limonene being the dispergents showing the best printing features. Figure 3 shows the droplet and filament formation of ethylene glycole, terpineol and limonene on the print head nozzle. Figure 3: Droplet formation with pure dispergent The nozzle diameter is 100 µm. On the left side of figure 3 the droplet formation of ethylene glycole can be seen. The printability is rated by optical observation. This is necessary to ensure that drops are generated always in the same shape and quality which means no sputtering and no wobbling of the droplet during the formation process. A stable printing process, good filament and drop formation and a fast forming of the drop when detached from the nozzle are rating points. The dispergent most suitable for the print head is the one that ensures establishing a range of print parameters for the demands described above. Such print parameters are the voltage UP of the piezo pulse, the duration tw of the piezo signal and the nozzle temperature ϑ. With the used 100 µm nozzle the droplet diameter of ethylene glycole in air is about 95 µm when the drop has formed out completely. All three pictures are taken at a frequency of 1500 Hz. The optical observation pointed out limonene being printable but in comparison with ethylene glycole and terpineol showing worse results regarding filament formation and stability. In table 1 the viscosities, surface tensions and densities together with print parameter ranges of these three dispergents are given. Within the parameter range drop formation in case of the IfA print head is possible. Table 1: Viscosities, surface tensions, densities and parameter ranges of ethylene glycole, terpineol and limonene at 25 °C Dispergent Ethylene Glycole Terpineol Limonene Commercial ink η in mPas σ in mN/m ρ in g/cm3 17.3 46.0 0.8 4.8 46.8 31.4 27.3 38.0 1.11 0.93 0.84 1.06 Dispergent Ethylene Glycole Terpineol Limonene UP in V 135 - 155 155 - 165 138 - 148 tw in µs 14 - 32 18 - 26 40 - 45 ϑ in °C 25 - 35 37 - 47 27 - 45 The slightly different surface tensions of the dispergents compaired with the commercial ink [5] do not seem to impair droplet formation. Next step is to have a look at the filament length and droplet speed after detaching from the nozzle. The measurements mentioned here are for the three dispergents ethylene glycole, terpineol and limonene. As already described, ethanole, tetradecane and texanole do not show an acceptable printing stability, so they are not subject of any measurements here. Figure 4 shows the influence of the nozzle temperature against filament length. By heating up the print head nozzle the ink viscosity can be reduced if necessary. Ethylene glycole is the dispergent offering the best print properties when being printed at room temperature. Figure 4 shows ethylene glycole, terpineol as well as limonene having a nearly linear behaviour, and taking the filament length ranges of ethylene glycole and terpineol into account, they are much higher compared to limonene which means the filament length can be better controlled by the temperature. Furthermore the printing process can be affected by the piezo signal, that is the pulse voltage and the pulse duration. Figure 5 is depicting the droplet velocity against pulse voltage. Ethylene glycole offers a maximum area of droplet velocity ranging from 1.1 m/s to 3.3 m/s linearly from 135 V to 155 V pulse voltage. The measurements were taken at a frequency of 1500 Hz. Considering the results of figures 3, 4 and 5 ethylene glycole is turning out to offer the best possibilities for acting as a dispergent in the drop-ondemand printing process described here. By having the chance to control the droplet velocity, the question arises, if droplets may splash and be divided into many little droplets when having a velocity 1000 900 Filament Length in µm 800 700 600 500 400 300 200 100 Ethylene Glycole Terpineol Limonene 0 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 Nozzle Temperature ϑ in °C Figure 4: Influence of the nozzle temperature on the filament length 3,5 Droplet Velocity v in m/s 3 2,5 2 1,5 1 Ethylene Glycole 0,5 0 134 136 138 140 142 144 146 148 150 152 154 Pulse Voltage U p in V Figure 5: Influence of the pulse voltage on the droplet velocity Terpineol 156 158 160 Limonene 162 164 166 between 1.1 and 3.3 m/s (ethylene glycole). The surface energy Esur of a droplet is given by Esur = 4 ⋅ σ ⋅ π ⋅ r 2 . (1) σ is the surface tension of the droplet and r its radius. After the drop has detached from the nozzle and has formed out it has a kinetic energy Ekin according to Ekin = 12 ⋅ m ⋅ v 2 = 23 ⋅ ρ ⋅ π ⋅ r 3 ⋅ v 2 . (2) m is the droplet’s mass, v its velocity, ρ its density and r its radius. To make a drop break up into two little droplets on top of the substrate the gain of surface energy in the transition one drop-two droplets must be gathered up by the kinetic energy in the moment of impact, that means the drop can only burst into two droplets when Ekin has a bigger value than the gain in surface energy according to ∆Esur = 2 ⋅ Esur, new − Esur, old the dissipation energy should be even bigger. Beginning with six different dispergents the number was reduced to three and from now on focused on ethylene glycole being the basis for a colloidal silver suspension for conductive paths. 3.2. Dispergent Ethylene Glycole with particles After finding out ethylene glycole being suitable as a dispergent for DoD-printing colloidal suspensions are prepared. Starting with 7.5 M% solids content in ethylene glycole percentage is increased to 20 M% and 30 M%. The suspension is mixed and before printing homogenized with an ultrasonic pulse instrument. Figure 6 shows the droplet formation on the nozzle of the piezo print head. Print medium is a 7.5 M% colloidal suspension with Ag95/Pd5 silverpalladium particles (95 % silver, 5 % palladium). (3) and Ekin is completely turned into ∆Esur. The dispensed volume V of a droplet is V = 4 ⋅π ⋅ r3 3 (4) and keeps constant during the breakup process, means 2 Vnew = Vold. Table 2 shows the ratio Ekin/∆Esur for minimum, middle and maximum droplet velocity v for ethylene glycole. The drop volume and therefore the drop diameter and radius depend on the pulsevoltage of the piezo signal [3]. In the observed experiments with ethylene glycole the diameter of a droplet varied from 90 µm to 110 µm when increaseing the piezo pulse from 135 V to 155 V (cf. figure 7). Table 2: Ratio Ekin/∆Esur of ethylene glycole for minimum, middle and maximum droplet velocity v (energies given in nJ) 1.22 ∆Esur 0.32 Ekin/∆Esur 0.8 1.54 1.51 0.37 4.2 4.22 1.82 0.48 8.8 velocity Ekin Esur, old 1.1 m/s 0.26 2.3 m/s 3.3 m/s Theoretically the drop should splash for values of Ekin/∆Esur being greater than 1, but this effect could not be observed in experiments at the IfA. Reason for that is that the kinetic energy of a drop is dissipated at a great deal into thermal energy when attaching and wobbling on the substrate’s surface, so the remaining part of Ekin is not enough to let the drop break up into two droplets. When using particles in the dispergent, Figure 6: Droplet formation A PSD (particle size measurement) figured out a d90 of 1.5 microns and a d50 of 0.9 microns for the used particles Ag95/Pd5. The time τ specifies the point of observation and shows the time delay after the start of the piezo signal. By turning up the time τ the droplet formation can be observed step by step when the printer is arranged in observation mode. Watching closely at the droplet formation, a phenomena can be detected: The droplet filament is not perpendicular as compared with pure dispergent (cf. figure 3). Reason for that is an inhomogeneity of the particles’ allocation in the colloidal suspension. The better the suspension is homogenized the better the filament generation is stable and perpendicular. The skill in that situation is to find the special dispersant additive that prevents agglomeration and sedimentation of the particles in the suspension as well as possible. Besides it can be observed that a small droplet peeled off the main drop but this small droplet catches up with the main drop again. This phenomena is due to a high pulse voltage UP and its taking place can be affected by UP. If printing with solid material in the dispergent in contrast to pure dispergent it can be realized that the pulse voltage has to be adjusted to a higher value, but generally it can be said: If the pure dispergent allows printing then there is a printer setting that enables printing of a colloidal suspension as well. The next experiments presented here were with a 20 M% colloidal suspension of Ag95/Pd5. It has turned out that 20 M% is printable more stably than 30 M% which means the printing process does not stop after a certain time as in 30 M% case. Reason for stopping of the process is an inhomogeneity of the print medium and therefore nozzle clogging. The effort in keeping the 30 M% colloid homogeneous is much higher, but good results in terms of conductive paths could be achieved with a 20 M% colloid already. Stable operation of the print head with a particle loaded dispergent is a basic requirement. When the colloidal ink is kept homogeneous, there is almost no difference in printing stability of 7.5 M% or 20 M%, but the advantage of having a higher material flow rate when printing 20 M% Ag95/Pd5 gave rise to the idea to print the conductive tracks with such a percentage of solid material in the colloid. In figure 7 two spots of colloidal silver on alumina substrate are mapped, having a diameter of 150 microns. The printed dots were sintered for 15 minutes at 850 °C an then photographed under a light microscope. When having a look at the spots under such a light microscope a three-dimensional structure of the 20 M% spot can be identified, so the mass fraction 20 % is the basis for the silver tracks printed at the IfA. Figure 7: Sintered spots of an Ag95/Pd5 colloidal suspension with dispergent ethylene glycole on alumina substrate, 7.5 M% (left) and 20 M% (right) Another feature in the DoD-printing process is the drying behaviour of the dots on top of the substrate. Because of the high fraction of the dispergent ethylene glycole to keep the colloid liquid and thereby printable, dispensed dots are wet when the substrate is unheated. The option to use dispergents with a high evaporation rate such as e.g. ethanole is not a good solution because through the high evaporation rate, printing is possible only for a few minutes before the nozzle dries and clogs [3]. That is another point arguing for ethylene glycole as a dispergent because of its lower evaporation rate. When using this substance, heating of the substrate while printing is necessary. The use of a heating plate during the deposition process improves the deposit quality. By heating the substrate, the liquid in the droplets is flash-evaporated on contact which minimizes the feature size in comparison to room temperature. Figure 8 shows printed spots on heated alumina substrate dried through flash evaporation. The two spots are dried at different heating plate temperatures and show different surface profiles. In both cases the same colloidal suspension was used but there is a significant influence of the substrate temperature. Figure 8 : Dried spots of an Ag95/Pd5 colloid (20 M%) with dispergent ethylene glycole on heated alumina substrate, 100 °C (left) and 140 °C (right) In case the substrate is heated up to 140 °C, the dispergent evaporates too fast and so the drop dries too fast. The drop is slowed down, solid material is pushed to the edge and does not have the possibility to build a surface profile according to the interfacial tension between the colloid and the substrate. The drop boundaries contain almost the whole solid material in a ring while in the middle of the drop there is nearly no silver-palladium any more. When reducing the substrate temperature to 100 °C this phenomena could be eliminated. The drop is not drying and growing stiff that fast and the solid material can arrange through the whole dot to form a spot as shown in figure 8. 3.3. Printing of conductive silver tracks Structural material was fabricated out of a silverpalladium colloidal ink consisting of particles in the range of 500 nm to 2.5 µm diameter. Figure 9 shows a 4 cm square alumina substrate on the heating plate of the planar motor with different printed conductive paths. Tracks with one, two and three layers were printed. For the conductive paths the substrate is heated up to 110 °C and the step size of the planar motor is set to 50 µm. The droplets can flow together before the dispergent evaporates and can form a conductive line with a coherent cross section. By varying the pulse voltage and thus the velocity of the droplets in the air, conductive paths of different width can be achieved. Figures 10 and 11 show sintered conductive tracks of different widths, obtained out of a 20 M% Ag95/Pd5 colloid with ethylene glycole as dispergent. Figure 9: Alumina substrate with printed conductive silver tracks on planar motor’s heating plate The silver lines seen in figure 10 are single layer (top), double layer (middle) and triple layer printed. When printing single layer and sintering the line for 15 minutes at 850 °C, part of the silver-palladium exhales from the substrate and the surface is fragile and subsequently discontiguous. In case of a single layer line no conductivity could be established. The lines in figure 10 were set with a piezo voltage of 190 V and have a width of 270 microns (single layer) and about 300 to 310 microns (double and triple layer paths). The triple layer line boundary is more uneven and not as smooth as the double layer line which originates in the drying behaviour of the dispergent on a heated substrate. The surface area of the second Figure 10 : Sintered conductive silver paths on heated alumina substrate, 1 layer (top), 2 layers (middle) and 3 layers (bottom) under a light microscope layer has dried already before the third layer is dispensed and has left back an uneven surface upon which the third layer is printed. The influence of the surface roughness seems to have a bigger effect on the transition second to third layer than first to second. One remedy in this case to obtain a better path-boundary is an adaption and lowering of the pulse voltage to set the droplets with less power. In figure 11 a conductive silver line printed with the same colloidal suspension as the lines in figure 10 is mapped. The difference to the paths before is the piezo pulse voltage which was chosen to 166 V and thus 24 V lower than before. Figure 11 : Sintered conductive silver path (3 layers) on heated alumina substrate, width 150 µm, with millimeter paper under a light microscope Lowering the piezo voltage and reducing the droplet velocity from 2.9 m/s down to 1.3 m/s reduces the line width of the silver path substantially. The width could be halved, measuring now 150 µm, and the track boundary of a triple layer path looks smoother than if printing at a high pulse voltage. For making a rough measurement of the cross sectional area of a DoD printed line, an atomic force microscope (AFM) is used. The double layer line as shown in figure 10 is AFM investigated (see figure 12). The cross section could not be charted completely because of the small measurement sector of only 70 µm, whereas the conductive track has a width of rough 300 µm. Nevertheless, it can be noticed that from the boundary towards the center of the track no gaps seem to appear. The section of the AFM measured Z-range in the upper part of figure 12 is from outside to outside, middle to middle and between the two inner marks. Outside to outside measures a thickness of 2.62 µm, middle to middle marks a vertical distance of 2.51 µm and the vertical Z- range between the both inner marks is 2.44 µm. The conductive path is printed double layer and a thickness of 2.6 microns is reached before the path center i.e. not even at the path’s thickest section, so it can be guessed that the thickness of a double layer path is at least 3 microns at the path’s center section. AFM measurement methods enabling the mapping of a complete cross section of the paths are already under construction at the IfA. The bottom part of figure 12 shows the double layer path from above, on the left as an AFM shot, on the right under a light microscope with sideways illumination. The tracks presented here have a length of 25 mm and they feature the electrical properties illustrated in table 3. H the track thickness. Best results with variation in RS appear when printing 3 layers, whereas a conductive path cannot be achieved with one single layer (only). 4. Conclusions Figure 12 : AFM picture of the profile of a sintered conductive silver path (2 layers), and under a light microscope (right) With the relationship RS = R ⋅ W L (5) the sheet resistance RS of the paths can be determined after measuring R. Based on a rough estimation of the lines’ thickness, the resistivity ρ calculated according to ρ = RS ⋅ H (6) is approximately 200 mΩ m. Table 3: Electrical characterization of DoD-printed conductive silver tracks (25 mm length) Track Layers R in Ω W in µm RS in mΩ/□ 1 2 3 4 5 6 7 8 1 1 2 2 2 3 3 3 8.2 4.6 16.9 4.9 1.9 1.9 130 310 270 290 290 150 300 310 88 53 195 30 23 23 In equations (5) and (6) R is the measured track resistance over the length L, W is the track width and With a newly developed piezo printing system consisting of a DoD print head and a planar motor, conductive tracks of different widths and thickness could be printed. A 20 M% silver-palladium colloidal suspension with ethylene glycole as dispergent was used to print lines onto alumina substrate. For a stable printing process, the colloid has to be well-dispersed to avoid agglomeration and sedimentation. Through the possibility of heating the substrate during the printing process, different surface profiles could be achieved with the temperature being an evaporation control parameter for the used dispergent. The piezo pulse voltage acts as a parameter to control the droplet velocity, but velocity should not be too high to obtain a smooth path boundary. Conductive tracks printed at the IfA, were characterized by conductivity measurement, light and atomic force microscopy. Square resistances in a range complying with already existing values could be measured [6]. Next steps are to mingle a suitable dispersant additive to the colloid to obtain a capable printing process. Thanks to Prof. Rothe and his crew for providing AFM-measurements and especially to Robert Bosch GmbH for their support. References [1] H. M. Nur, “Ink-jet printing of gold conductive tracks”, Journal of Materials Science – Materials in Electronics Vol. 13, pp. 213-219, 2002 [2] Microdrop GmbH, Micro Dispensing Systems of liquids in the nano- to picoliter range, Norderstedt, Germany, http://www.microdrop.de/index.html [3] S. B. Fuller, “Ink-jet printed nanoparticle microelectromechanical systems”, Journal of Microelectromec. Systems, Vol.11, Issue 1, pp. 54-60, 2002 [4] R. J. Hunter, “Foundations of Colloid Science”, Oxford University Press, second edition, New York, Chapter 1, pp. 1-2, 2002 [5] M. Mott, “Microengineering of ceramics by direct ink-jet printing”, Journal of the American Ceramic Society Vol. 82, Issue 7, pp. 1653-1658, 1999 [6] J. B. Szczech, “Ink Jet Processing of Metallic Nanoparticle Suspensions for Electronic Circuitry Fabrication”, The American Society of Mechanical Engineers – First International Conference on Microchannels and Minichannels, pp. 795-800, April 24-25, 2003