Novel Ag and AgPd Nanoparticles for MLCC s with Ultrathin Electrodes

Transcription

1 Novel Ag and AgPd Nanoparticles for MLCC s with Ultrathin Electrodes Brendan Farrell, Daniel Andreescu, and Dan V. Goia* Center for Advanced Materials Processing, Potsdam, NY , USA Tel: ; Fax: ; goiadanv@clarkson.edu Abstract The aggressive reduction in materials costs and the relentless drive to increase the specific volumetric capacitance are two of the most important trends that characterize the MLCC technology. In regard to the first trend, the shift from precious metals to base metals was unquestionably a disruptive technology that has led to dramatic cost savings. In regard to the second trend, however, the limitations of the existing metal powders and the inability of the screen-printing technology to yield ultrathin internal electrodes represent major obstacles to further increase the volumetric capacitance. This paper presents a new line of highly dispersed AgPd nanoparticles that can be used to generate ultrathin ( nm) uniform conductive layers and could pave the way for capacitors with a very high number of electrodes and, therefore, high volume capacitances. A novel deposition technique that can be integrated along with these materials in the existing MLCC manufacturing lines with minimal disruption and in a cost effective manner, is also proposed. Introduction Metallic particles are extensively used in the electronic industry to build conductive layers that can function as either intrinsic elements (electrodes) in various components (capacitors, varistors, actuators, etc.) or as connecting paths between electronic components in complex circuits. The large majority of these metallic structures are obtained via thick film technology, an approach in which metallic particles are dispersed in high viscosity liquid vehicles (pastes) and then deposited in the desired patterns by screen-printing [1-3]. The resulting deposits of closely packed metallic particles are subsequently converted into solid, continuous conductive layers following the removal of the organic matter first and then sintering at appropriate temperatures. In a more sophisticated version, this technology can be used to build multilayer ceramic capacitors (MLCC) consisting of hundreds of alternating layers of dielectric as thin as 2 µm and metallic layers as thin as 0.8 µm. As a result, the volumetric density of a given electrical property (capacitance in the case of MLCC's) can be dramatically increased allowing a more efficient use of the board's real estate and, therefore, further miniaturization of electronic components and devices. While the level attained by the existing MLC technology is impressive, there are still many compelling economic and technical arguments to further reduce the thickness of the metallic layers. For example, in the case of noble metal based MLCC s, a four-fold reduction in the thickness of the metallic electrodes (from ~800 nm in the present state-of-the-art to 200 nm) would reduce dramatically the cost of the metal incorporated. Such significant reduction in metal lay-downs in conjunction with a drastic reduction in the palladium content in the electrode can make the PM-based technology competitive in cost with the BM technology. More importantly, thinner metallic layers also offer the possibility to increase the number of electrodes and decrease the thickness of the dielectric layers, as the mechanical stresses developed in the multi-layered structures are greatly diminished. As a result, the specific volumetric capacitance will be dramatically increased and further miniaturization of the components and electronic devices in general and multilayer ceramic capacitors in particular will be possible. Unfortunately, the inability of the screenprinting technique to further decrease metal laydowns represents a great barrier in obtaining thinner metallic layers with the present state-of-the-art

2 thick film technology. For this reason, the efforts to decrease the size of the metallic particles and improve their dispersion and uniformity have been held back due to the realization that they would not necessarily translate into thinner metallic layers [4]. Nevertheless, the electronic industry continues to look for suitable means to further reduce the dimensions of the metallic layers. One alternative often considered is the thin film technology, an approach in which very thin, dense, and conductive metallic films are assembled atom by atom by condensation from gas phase. However, the inability of the chemical or physical vapor deposition methods to directionally control the flux of atoms and to obtain sophisticated patterns on a desired substrate without significant metal losses makes them unsuitable for low cost, high throughput mass production of multi-layer structures. Furthermore, the sintered metallic layers deposited by vapor deposition cannot be used in conjunction with a green ceramic substrate since the subsequent sintering of the latter would develop stresses at the metal/ceramic interface that will make it impossible to construct complex multi-layer structures. This paper proposes a method to build conductive metallic layers with thickness of less than 200 nm by using highly dispersed uniform nanosize metallic particles and a novel deposition technique capable of assembling them into wellpacked green deposits that can be subsequently converted into ultrathin continuous metallic films. A novel approach to stack these ultrathin electrodes and the dielectric layers in a cost effective manner is also presented as n alternative to the existing MLCC manufacturing process. I. Preparation of nanosize Ag and AgPd particles a. Characterization techniques The size and morphology of the Ag and AgPd particles were studied by field emission scanning electron microscopy using a Jeol JSM 7400 instrument. The oxidation behavior was evaluated using a Perkin Elmer Pyris 1 TGA analyzer, the temperature of the samples being increased at 10 0 C/min to C. A Bruker D8 Focus instrument was used to generate the XRD patterns. The particle size distributions of the metallic particles were measured by the laser diffraction technique using a Malvern Mastersizer 2000 instrument and confirmed by dynamic light scattering determinations performed with a Brookhaven Zeta-Plus analyzer. The content of carbon and oxygen in the powders was determined using a LECO combustion instrument. b. Design criteria for the metallic particles Several important aspects must be considered in the design and preparation of the metallic particles to ensure that they would indeed yield ultrathin continuous metal electrodes. First, the optimum particle diameter needed for a desired final electrode thickness can be estimated based on theoretical considerations pertaining to the theories of sintering and packing of monodispersed spheres. Assuming a random loose packing in the green body with a packing efficiency of 0.52% [5] and a final sintered density of 95%, the relation between their diameter d and the thickness of the sintered layer δ can be expressed by: d δ = 0.7n( ) 1 3 ( ) where n represents the number of particle layers in the arrangement and 0.7 is a correction factor reflecting the geometrical interpenetration of adjacent layers. Since a minimum of five superposed particle layers are usually recommended to ensure a good tridimesional sintering, the calculated particle diameter for target electrode thicknesses of 150 and 200 nm is 55 and respectively 72 nm. Secondly, it is essential that the nanosize AgPd particles are truly nonagglomerated to ensure their ability to move freely and occupy the optimal positions in a highly ordered close packed three-dimensional arrangement. For such cases the ratio between the mean value of the particle size distribution (PSD50%) and the particle diameter determined from electron microscopy measurements must be very close to unity, a larger value indicating the presence of undesired agglomerates. Finally, it is desirable that the metallic particles are highly uniform and spherical or isometric so they can assemble in three-dimensional structures with high 1 3 (1)

3 packing densities [5,6]. The high particle coordination number and uniform pore size of such arrangements ensure a uniform sintering and the formation of continuous conductive metallic layers. c. Particles preparation Highly dispersed metallic particles can be obtained by either reducing the size of the bulk metal (atomization, milling), by converting thermally or chemically finely divided metal compounds [7-9], or by assembling the metal atoms in either gas (CVD, PVD) or liquid phase (chemical precipitation). While the vapor techniques offer some distinct advantages (formation of highly crystalline metallic particles, synthesis of truly alloyed composite metallic particles), they rarely yield highly uniform, agglomerate-free powders and their scale-up involves expensive equipment. In contrast, the methods based on the chemical reduction in liquid phase, are known to be capable of producing monodispersed, highly non-agglomerated metallic particles over a very wide range of sizes and at very affordable costs. The highly dispersed, uniform AgPd nanoparticles presented in this work were generated using a novel precipitation method described in a recently submitted patent. This process uses only environmentally friendly chemicals (solvents, reductants, additives) and can yield pure silver and silver/palladium powders with a content of Pd as high as 30% at manufacturing costs comparable to those of commercially available AgPd powders. d. Particles properties Although the size of the Ag and AgPd particles can be varied between 50 and 120 nm by adjusting the parameters of the precipitation process, this paper deals with metallic particles having a diameter of ~65 nm, a value which represents the middle of the calculated size range needed to obtain fired electrodes with thickness between nm. Once the experimental conditions for a given diameter are selected, the size and the size distribution of the resulting metallic particles do not change when the AgPd ratio is modified, as demonstrated by the FESEM images in Fig. 1 and the corresponding PSD distributions in Fig. 2. For all compositions there is a very good agreement between the diameter measured by electron microscopy and the average size distribution, an indication that the particles are completely nonagglomerated. b a c Particle Number (%) Particle Number (%) nm Fig. 1. SEM images of a) Ag 100%, b) AgPd 95%/5%, c) AgPd 90%/10%, and d) AgPd 80%/20% nanoparticles d = 65 nm 100% Ag 0 1E a c Log Diameter (µm) AgPd 90/10 d = 68 nm 0 1E Log Diameter (µm) Particle Number (%) Particles Number (%) E d AgPd 95/5 d = 68 nm Log Diameter (µm) d = 73 nm AgPd 80/20 0 1E Log Diameter (µm) Fig. 2. PSD data for a) Ag 100%, b) AgPd 95%/5%, c) AgPd 90%/10%, and d) AgPd 80%/20% nanoparticles The highly non-agglomerated character of the powders is confirmed by the high values of the tapped density (TD) of the dry powders (over 3.0 g/cm 3 ), which is comparable with the values recorded for existing commercial products. d b

4 Regardless of their composition, all AgPd particles consist of a pure silver core surrounded by a Pd shell, the separation of the two phases being clearly displayed by the XRD patterns in Fig. 3. Intensity 40 Ag / Pd 80 / Ag / Pd 90 / 10 Ag / Pd 95 / 5 40 Ag 100% Theta (θ) Fig. 3. XRD spectra of different AgPd compositions This particle structure was intentionally chosen as it permits to obtain higher crystallinity phases for the two metals and offers the advantage of a limited migration of silver due to the external Pd shell. Since these materials are geared toward the low-fire applications, in which the Ag content is more than 80%, the oxidation of Pd and the associated volume expansion are less pronounced and, therefore, of less concern, as confirmed by the TGA graphs in Fig. 4. Weight (%) Temperature (C ο ) 90 / / / 5 Fig. 4. TGA curves for powders with various Ag/Pd ratios e) Preparation of stable metallic dispersions The deposition of the particles as ultrathin closepacked layers onto the surface of a substrate can be best realized using their dispersions in liquids (vehicles). While in the case of the screen-printing technology the vehicles are very viscous pastes, in this work we start from the premise that dispersions of lower viscosity are best suited to yield ultrathin films. The rheology and the evaporative properties of the metallic dispersions can be adjusted by manipulating the concentration of the metal phase, the properties of the solvent/vehicle, and through the addition of surfactants or polymers. While water is the most desirable solvent, its evaporative properties pose serious problems in the drying of thin films. This negative attribute can be eventually addressed through the addition of miscible solvents, preferably mono- and polyalcohols. The dispersion needs also to contain a polymer capable to act as an effective dispersant and to hold the assembly of particles together once the solvent has evaporated. However, the amount of polymer(s) used should be selected in a way that its volume in the dried deposit would not exceed the volume of the pores formed in the closely packed arrangement of spherical particles. A low molecular weight polymer is preferable in order to prevent a significant increase in the viscosity of the film and possibly the agglomeration of the particles at the late stages of the drying process. Based on these general considerations, stable aqueous dispersions of Ag and AgPd nanoparticles with concentrations as high as 30% by weight were successfully prepared. The size distribution measurements indicate that the metallic particles are fully dispersed and the dispersions are stable for long times. The stability of the dispersions is facilitated by the negative charge of the metallic particles (between 40 and -45 mv) and the presence of small amounts of dispersants and higher molecular weight polyols. f) Film formation and sintering properties Following a very basic dip coating method, these dispersions prepared as described above were used to lay down well-packed uniform deposits of metallic nanoparticles on smooth glass plates previously functionalized to provide a

5 positively charged surface. As illustrated in Figure 5, the deposition experiments resulted in uniform well-packed deposits of nanoparticles with a thickness of ~300 nm, which corresponds to a sintered layer ~270 nm. ~300 nm Glass slide Fig. 5. Cross-section of a layer of metallic particles deposited on a glass slide By optimizing the properties of the dispersion and the deposition technique, the thickness of the metal deposit can be further reduced to achieve the targeted fired electrode of less than 200 nm. The nanoparticles described in this paper tend to sinter at significantly lower temperature than the submicrometer or micrometer size particles used presently as precursors for MLC electrodes. For example, the sintering experiments carried out with the Ag films shown in Fig. 5 indicate an onset of sintering at ~175 0 C and complete film formation at ~260 0 C. The addition of Pd raises the sintering temperature although, for a given composition, the nanoparticles described in this paper would sinter at a significantly lower temperature than the existing commercial materials. II. Novel process for manufacturing MLCC with ultrathin electrodes The properties of the metal nanoparticles presented above coupled with the demonstrated ability to be deposited as well-packed thin deposits offer the premise for a novel, cost effective way to assemble the metallic and dielectric layers into multilayer ceramic capacitors. This novel approach involves a) the deposition of ultrathin layers of metallic nanoparticles particles on a carrier film followed by b) the transfer of the deposited layer in the desired pattern onto the dielectric tape by hot-stamping. a) Deposition of the metal layer on the carrier film Once suitable metallic particles are available as stable dispersions, several mass production techniques can be adapted to deposit them as ultrathin well-packed deposits onto carrier films. The selected method must be capable of depositing a thin volume of dispersion, which yields, after the evaporation of the solvent, a closely packed assembly of metallic particles consisting of ~4-6 layers. This is not a trivial matter considering that 5 closely packed layers of ~60 nm particles of AgPd 80%/20% (ρ = g/cm 3 ), which are needed to generate a 200 nm thick metallic film, correspond to a metal coverage of slightly less than 0.2 mg metal/cm 2 of substrate (For comparison, the lowest surface coverage obtainable in mass produced capacitors today by conventional screen-printing is ~ 0.8 mg/cm 2 ). Nevertheless, several technologies (spin coating [10-12], letterpress, offset lithography, gravure, flexography, recess printing, and inkjet printing) are known to have the capability to achieve this level of performance. However, these techniques have not been integrated yet in the existing MLCC manufacturing lines mainly because of their inability to match the capabilities of the screenprinting method to deposit patterned electrodes at the high speeds required by the production process. In the novel process proposed in this paper the metallic particles are deposited on the surface of a carrier polymer film by spin coating. Prior the deposition, the surface of the carrier needs to be coated with a thin layer of a release agent. The role of this compound is to ensure a good adherence of the metallic layer to the carrier film and allow its separation only from selected regions where the right combination of temperature and pressure is applied. After the metallic layer with the desired thickness is deposited, a sizing agent is applied on the external surface of the metallic film. The role of this coating is to ensure a good bonding between the metallic layer and the dielectric tape during the subsequent transfer process. Optionally, polymers that can promote strong interactions with the binder used in the

6 dielectric tape can be added in the dispersion of metallic particles. To achieve the optimum packing density of the metallic particles and the desired final thickness of the deposit, all characteristics of the metal dispersion (composition, drying properties, rheological behavior, concentration of solids, etc.) need to be specifically tailored to deposition technology selected. The deposition of the metallic layer onto the surface of the carrier film by spin coating is illustrated schematically in Fig. 6. remaining on the spent carrier film can be resuspended in the same solvent to form a re-usable dispersion. Supported metallic film Hot-stamping Release Heated die Dielectric tape Carrier film Release agent Spent metal film Printed dielectric Film deposition Metallic layer Sizing agent Fig. 6. Schematic illustration of the deposition of the AgPd particles on the carrier film b) Transfer of the metallic layer on the dielectric After they are deposited on the carrier films, the metal layers can be transferred onto the substrate of choice (i.e. ceramic tape) by hot stamping. This is achieved with the help of a die on the surface of which flat rectangular features identical in shape and size with the electrode are provided. Put it simply, the surface of the die is the negative image of the screen used in the existing manufacturing process: the wires of the mesh corresponding to the grooves in the die and the openings to the raised rectangular pads. The temperature of the die and the pressure applied is adjusted in a way that ensures the rapid release of the film from the areas where the die comes in contact with the carrier film (Fig. 7). The unused fraction of the metallic film, which remains on the carrier after hot-stamping, can be minimized through an optimized patterning of the die surface. It is also possible to design the binder in the metallic layer in such a way that the metallic film Fig. 7. Schematic representation of the hotstamping transfer of the metallic layer onto the dielectric tape The ideal way to solve the problem, however, is to deposit the metallic layers directly in a patterned manner on the carrier as in the existing screenprinting technology. In this regard, the inkjet printing, which is making lately significant progresses in reducing the volume of the deposited layers, and other deposition methods used in the printing industry (offset lithography, gravure) can be excellent alternatives. A major advantage of the direct patterned deposition is the fact that the die used to transfer the metallic layer does not need to have the rectangles corresponding to the electrodes engraved on its surface. c) Proposed MLCC assembly process The two elements of the proposed technology presented above, the metallic film supported on the carrier film and the transfer of the metallic layer by hot-stamping, can be combined in a new process that can be used to stack at high speed metallic and dielectric layers to build cost effectively the multilayer structure of the MLCC s (Fig. 8).

7 Pressing Hot-stamping Stacking Film transfer Green MLCC Fig. 8. Hot-stamping film transfer process for stacking metallic and dielectric layers Since this novel process is very similar to the existing state-of-the-art one illustrated in Fig. 9, its integration in existing manufacturing lines of MLCC s could be eventually done with minimal disruption of the existing infrastructure. Pressing Screen-printing Stacking Drying Green MLCC Fig. 9. Conventional MLCC making process Conclusions A novel series of uniform nanosize Ag and AgPd particles with an average diameter of ~65 nm and a high degree of deagglomeration were prepared using a new chemical precipitation process. These materials can be obtained either as dry powders or as stable dispersions in liquids. By various deposition techniques, the latter can be used to assemble the metallic particles onto a desired substrate as well-packed thin deposits, which can then be converted through subsequent sintering into ultrathin continuous metallic layers with a thickness below 200 nm. In conjunction with a suitable deposition technique, these novel nonomaterials can be used to build cost effectively multilayer ceramic capacitors with significantly thinner metallic layers than those obtained in the existing process based on the use of larger particles dispersed in high viscosity pastes and the screen-printing deposition method. The significant reduction in the cost of the precious metals incorporated, achieved by combining ultrathin electrodes and a lower Pd content, can potentially re-establish the PM MLCC technology as a viable alternative to the BM technology. References 1. Hamer, D.W., Biggers, J.V., Thick Film Hybrid Microcircuit Technology, Wiley-Interscience, New York (1972). 2. Masussek, L.I., Glang, R., Handbook of Thin Film Technology, McGraw-Hill, New York (1970). 3. Vosson, J.L., and Kern, W., Thin Film Processes, Academic Press, New York (1970). 4. Goia, D.V., Berube, G., Challingsworth, M., Mann, L.A., Chambergo, M., Monodispersed Low-fire AgPd Powders for Ultra-thin MLC Electrode, CARTS, October 16-20, 2000, Brussels, Belgium. 5. German, R.M., Particles Packing Characteristics, Metal Powder Industries Federation, Princeton, (1989). 6. Her, Y.-S., Matijević, E., Chon, M.C., Preparation of well Defined Colloidal Barium Titanate Crystals by the Controlled Double Jet Precipitation, J. Mater. Res., 10, (1995). 7. Goia, D.V., Matijević, E., Preparation of Monodispersed Metal Particles, New J. Chem., , (1998). 8. Ishikawa, T., Matijević, E., Formation of Monodispersed Spindle-type Pure and Coated Iron Particles, Langmuir, 4, (1988). 9. Porta, F., Hsu, W.P., Matijević, E., 'Preparation of Uniform Colloidal Metallic Ruthenium and Its Compounds', Colloids Surf., 46, 63-74, (1990). 10. Gu, J., Bullwinkel, M., Campbell, G. A., Measurement and Modeling of Solvent Removal for Spin Coating, Polymer Engineering and Science, 36(7), 1019 (1996).

8 11. Gu, J., Bullwinkel, M., Sukanek, P., Campbell, G. A., The Effect of Polymer Molecular Weight and Solvent Type on the Planarization of Spin- Coated Films J. of the Electrochem. Soc., 142(7) 2389 (1995). 12. Gu, J., Bullwinkel, M., Campbell, G. A., Spin Coating on Substrate with Topography" J. of the Electrochem. Soc., 142(3) 907 (1995)