Creating metal and nonmetal nanosystems using conductive jettable inks
04/01/2006
There is a great deal of interest in the development of ink-jettable systems for RFID tags and other types of interconnection. This paper reviews the chemistries, processing, and performance of jettable inks based on metals and conductive oxides, and outlines potential application areas.
Alan Rae, Dana Hammer-Fritzinger, NanoDynamics Inc., Buffalo, New York
Recently, an enormous amount of attention has been devoted to printing circuits by both startup and established companies that are supported by government and private funds worldwide. Yet the concept of printing circuit traces is not new. In fact, printing techniques have been utilized for ceramic hybrid circuits and for flexible circuits used in membrane switches and keypads for many years. Nevertheless, the printed electronics market has the potential to reach at least $10 billion by 2010 [1]. A combination of new materials, new circuit structures, and new market opportunities is driving the growth. Although many of the markets are nascent, with the structures not optimized and the materials needing further development, all areas are actively being explored.
Printing technologies
Printing techniques are of interest for a number of reasons, including the following:
Environmental. Many circuit-forming processes are additive-subtractive and typically can require up to 8kg of material to produce a 1kg circuit board. Conversely, printing processes are strictly additive and, therefore, more environmentally efficient.
Flexibility. A printing process can be either a digitally driven serial deposition, or a massively parallel deposition using flexographic or lithographic printing. Materials can be deposited on 3D surfaces, such as casings using inkjet or transfer printing. A digital process gives flexibility, whereas a parallel one is low cost.
Cost. A printing process can be adapted to be low-cost, for example, reel-to-reel on flexible substrates. For the last several years, the fastest growing sector of the substrates business has been in flexible substrates-traditionally, polyimide for solderability and polyester for use as a low-cost substrate for keyboards and membrane switches.
Low-temperature processing. Nonfired composite curing at <200°C or low-temperature fired systems <250°C can be used to create functional circuit elements.
Materials that can be printed include:
- conductors for use as low-cost substrates, such as polyester and even paper, instead of epoxy, polyimide, or ceramic (but process temperatures must be reduced to <200°C);
- semiconductors, i.e., polymers or polymer composites that can be components of structures such as solar cells (Graetzl cells), light emitting diodes for displays, or transistors;
- dielectrics, for example, high-k materials, such as for embedded capacitors, or low-k materials for insulation; and,
- phosphors and other functional materials.
Competing processes include plating techniques, which are well established, but use aggressive chemical baths; etching laminated planar copper on glass-epoxy to develop traces and pads, which is also a well-established, low-cost, mature technology; and semiconductor processes, including spin-on, lithographic, CVD, ALD, etc.
Traditional printing normally does not require as high a level of precision in registration as for electronics. Newspapers, for instance are frequently printed with misregistrations approaching nearly 1mm, whereas sophisticated screen printers used in the printing of solder paste on circuit board on a single-panel basis contain extremely complex optical registration and measurement systems [2, 3].
According to the iNEMI 2004 roadmap [4], commercial boards typically use a 100µm (4mil) linewidth and space with state-of-the-art micro-via substrates running at 50µm, line and space. The tendency toward finer resolution for smaller components is balanced by the move to higher clock speeds and wireless systems in which larger conductors are needed to optimize impedance at high frequencies.
The limit for stencil printing is ~50µm, as is the minimum stencil thickness to allow reasonable life. Conventional color lithographic or flexographic prints normally run with a registration of 400µm, but can be “tightened” somewhat for electronics use, and dot sizes for inkjet can allow a resolution of 50µm line and space with very thin conductors. Xerographic and electrophoretic printing processes have already been demonstrated with compatible accuracy and precision.
The resolution required will depend on the application. Printed RFID antennas on cardboard packaging or keyboard networks require much less precision than trimmed electrodes.
Conductive ink fillers are particulate, and for inkjet applications are screened to eliminate agglomerates, or particles >1µm. In general, smaller is better, and an average particle size of 30-50nm seems to be the accepted range. Other printing techniques with lower resolution can tolerate larger particles.
The inkjet process
Inkjet processes are only part of a wide range of printing tools available to the industry including flexographic, lithographic, stencil, etc. Different techniques will be appropriate depending on the speed, precision, and thickness of deposition desired.
The sweet spot for data-driven printing can be described in several ways: products that cannot be readily miniaturized to the nanoscale, e.g., keyboards; antennas; short-run products, e.g., prototypes; thin and 3D structures (repeated print applications can increase thickness at the expense of speed); and trimmable resistors, capacitors, and inductors, potentially with a metrology feedback loop to allow tunabilty to the desired value (a major issue with passives because engineers are used to the tight tolerances obtained by testing and sorting every discrete component).
The primary data-driven technology is inkjet, which is in widespread industrial use for product marking and in the office and home environments for low-cost color printing. Electrophoretic or xerographic processes also have the potential to be data-driven, but are less commercially advanced than inkjet.
 Figure 1. The two main types of inkjetting: a) continuous inkjet; and b) drop-on-demand. |
There are two main types of inkjet: continuous and drop-on-demand (Fig. 1). In the continuous process, a stream of ink droplets-usually produced by a piezo device creating pressure pulses through an orifice-is ejected and subsequently caught in a collection cup. The stream can be deflected electrostatically, at which point droplets reach the substrate. These jets are frequently used for high-speed marking of rapidly moving food packaging, and generally have lower dot size precision (~50µm) than drop-on-demand.
Drop-on-demand inkjet systems can use a piezo or a thermal engine. Electrostatic and acoustic, as well as electrostrictive engines, are also possible. Generally, piezo systems are lower resolution (~125µm) than thermal engines (~20µm) but are often favored in industrial inkjet systems. Variable drop size metering can increase precision and multiple nozzles enhance throughput.
Conductive Inkjet Technology (CIT) [5] has developed a laser-assisted drop-on-demand technology to lay down fine lines down to 5µm, creating essentially invisible conductive lines. The highest precision so far appears to have been achieved at the U. of Cambridge [6], where 100nm features were developed in organic polymer circuits. These techniques rely on the deposition of organic rather than purely metallic structures; in the case of CIT, the polymer is subsequently metallized on a catalyst, while at Cambridge, a conductive polymer is used.
Inkjet heads are attached to reservoirs, which may be sealed
efillable cartridges, similar to consumer inkjet cartridges, or custom-designed with features such as recirculating pumps to keep solids suspended. Water-based inks are typically formulated to a viscosity of ~10 centipoise (0.01 Pascal-sec). The diameter of the ink ejected from the orifice is usually (but not always) about the same in diameter as the orifice diameter. Typical orifice sizes are in the range of tens to hundreds of microns and are manufactured by photolithographic or laser ablation techniques. A detailed description of inkjet structures has been given by H.P. Le [7].
Marking inks can be water-based, hot-melt, or solvent-based. Water-based or hot-melt inks are generally preferred for environmental and safety reasons, but solvent systems tend to be more flexible. Pigment contents are typically 5-10%, but for functional systems, the solids content needs to be pushed as high as possible; 50% is a target that is not readily achievable.
With high concentrations of very small particles, the particle-to-particle distance becomes very small, and it is difficult to prevent agglomeration. There are a million times as many particles in a 10nm suspension than in a 1µm suspension of the same mass. To put this in perspective, solder pastes typically are ~90% metal by weight (and 50% by volume). Increasing the surfactant content can increase the loading at the expense of burnout difficulties at a layer stage, as many surfactants contain significant inorganic moieties.
Smaller particles are less susceptible to Stokes Law settling-even at 80nm under an optical microscope, significant movement through Brownian motion is visible-but they need to be prevented from 1) agglomeration through management of the “solvation sheath” and 2) oxidation or other undesirable reaction with the environment.
Silver concentration is a typical compromise because it is a key conduction path, forming material with a specific gravity of 10.5. This material tends to settle quickly at high concentrations at the low viscosities needed for fine inkjetting according to Stokes’ law, unless the particle size is <50nm. A typical water-based ink viscosity, as mentioned above, is ~10 centipoise, similar to blood, whereas the viscosity of solder paste at rest is similar to peanut butter, or 150,000 centipoise.
Ink materials for conductors
The choice of conductor pigment is determined by the resistivity required and the process parameters. Although conductive oxides, such as indium tin oxide, and other non-metal pigments, such as carbon blacks, can be used, their relatively low conductivity means that they are really resistors rather than conductors. Because there is no easy mechanism to coalesce the particles, percolation conduction is the only practical mechanism. Carbon nanotubes promise higher conductivity but are still relatively expensive and difficult to disperse and align into useful conductor structures.
With respect to metals, precious metals such as gold are effective, but not economical, and low-melting metals are scarce or toxic (mercury, gallium, indium), oxidize readily, or do not consolidate at practical processing temperatures to form high-conductivity films.
Nanosilver ink has received attention because of its potential to be made in stable suspensions that can be inkjetted to form conductive deposits by processing below 200°C.
Many materials exhibit a change in properties as they move toward the nanoscale due to the increase in the relative proportion of higher energy surface atoms. This change can be exhibited as a change in reactivity, e.g., sinterability, or electromagnetic properties driven by band gap changes resulting in dramatic changes of electronic properties or optical properties, such as color and transparency.
The “tipping point” at which these changes appear typically occurs below the 100nm scale, but it is a function of the individual element or compound and its environment (Fig. 2). In the case of silver, a dramatic increase in sinterability takes place at low temperatures (<200°C).
Nanosilver powders can be produced in essentially a monodisperse form; one example is the precipitation technology developed at Clarkson U. (Fig. 3). The particles are precipitated using a bottom-up approach where nuclei grow under a protective polymer coat that allows metal atoms to accumulate while acting as a charge director to keep the embryo crystals separated. The crystals are allowed to grow until the desired particle size is achieved. Then the reaction is stopped and the crystals are suspended in an appropriate vehicle or dried. The polymer coating, which can be made hydrophilic or hydrophobic, aids re-dispersion and prevents the spontaneous sintering of the dried silver particles.
 Figure 3. Near-monosize metal particles produced by bottom-up synthesis using precipitation from liquids: 10nm Ag dispersion (left); and 60nm dried Ag powder (right). |
Additional developments include the precipitation of nano-sized Ag platelets (Fig. 4), which can be made compatible with inkjetting through a narrow orifice. (In comparison, regular silver flake in conducting adhesives is typically 10µm dia.) A general rule is that for inkjet inks, the maximum pigment particle size should be <1/10 of the orifice diameter to avoid bridging and blockage, although the ratio may need to be greater depending on the nozzle profile and operational cycles.
 Figure 4. Nano-sized precipitated silver platelets are of interest because of their aspect ratio, processability, and reactivity in nonfired and fired conductors. |
Inks can be prepared by adding the particles to a suitable vehicle, applying them to the substrate, and then consolidating them in a polymer matrix, or by sintering particles together at a temperature compatible with substrate integrity. Significant silver-to-silver conductivity is achieved when the polymer coating is removed at 175-250°C, or if the polymer is a permanent matrix rather than a process binder, percolation paths are ensured by the very large number of nano particles in the composite (Fig. 5). This process temperature is well below the traditional melting temperature of 961°C for silver, and the remarkable silver mobility can be explained by the energy of particles with a high surface-to-volume ratio-10,000× higher at 10nm compared with 1µm.
 Figure 5. Conductive silver inkjet ink using a) 80nm nano-sized Ag as-deposited; and b) thermally processed at 250°C. (Courtesy: Heraeus Inc.’s Circuit Materials Division) |
The processing temperature can be matched to the substrate by modifying the silver size and surface preparation, as well as the ink formulation.
Nanosilver can be dispersed in alcohol for compatibility with styrene-acrylic resins or in glycols for compatibility with water-based systems. Curing is accomplished by the cooling of hot-melt resins, evaporation of solvents, heat crosslinking, or UV crosslinking. Scanning laser techniques as a way of selectively applying heat have also been suggested.
Conclusion
Printed electronics is a rapidly developing area with applications in displays, antennas, and other structures for RFID, etc., as well as in existing applications in keyboards, membranes, and flexible circuits. One of the most rapidly developing areas of commercialization of printed electronics will be in the printing of conductors using silver-based inks enabled by controlled precipitation of nanosized silver particles and platelets. The use of jettable inks promises soft tooling as a way to increase resolution, 3D capability, on-the-fly thickness control, and low-temperature circuit development on a range of substrates.
References
- Refer to data at http://www.nanomarkets.com.
- Examples of advanced printer types are at http://www.speedlinetechnologies.com.
- Examples of advanced printer types are at http://www.dek.com.
- iNEMI 2004 roadmap is at http://www.inemi.org.
- http://www.xennia.com/conductive_inkjet.
- http://www.technologyreview.com/InfoTech/wtr_14676,294,p2.html.
- http://www.imaging.org.
Alan Rae received his honors BSc in chemistry from Aberdeen U., and his PhD in metallurgy and engineering materials and his MBA from the U. of Newcastle upon Tyne, UK. He is the VP of market and business development at NanoDynamics Inc., 901 Fuhrmann Blvd., Buffalo, NY 14203; e-mail [email protected].
Dana Hammer-Fritzinger received her BSc in chemistry from SUNY at Geneseo, her MSc in analytical chemistry from the U. of Helsinki, Finland, and her MBA from Canisius College. She is a product manager at NanoDynamics Inc.