For the PDF version of this article, click here.
The electrical and mechanical stability of components in power train applications has always been a major issue. The drive to lower costs and produce smaller-sized components has generally removed some of the safety margin in the designs, or caused increased parasitic losses under normal application conditions. To compensate, various capacitor-type (tantalum, aluminum electrolytic and ceramic) manufacturers have begun adding polymer elements to their products to enhance electrical and mechanical performance. But, rather than using polymer hybrids, those requiring the highest performance, most stable capacitors commercially available should consider pure polymer film capacitors for their critical applications.
Critical electronic systems used in markets such as military, flight or telecommunications require the use of components with inherent reliability. No matter how much circuit redundancy or accelerated screening testing is done, there is always a golden nexus that can produce a single point of failure (SPoF). That condition requires circuit designers to spend a great deal of time trying to minimize the probability of such failures.
Due to their low mass, outstanding performance capabilities and unmatched inherent reliability, polymer film capacitors have long been established as the capacitors of choice in high-performance, mission-critical applications. These applications require units with an established track record of durability and reliability, rather than components that simply meet minimum performance standards.
Industries such as telecom learned decades ago that while other capacitor technologies have their viable uses, in pivotal applications, only polymer film capacitors have the inherent performance, stability and reliability needed. The telecom industry's commitment to using polymer film capacitors is such that all the major telecom companies once produced their own capacitors, until the commercial market was finally able to produce the quality and volume levels that telecom required. Other industries, such as aerospace, the military and automotive, also produced polymer film capacitors to meet their special needs.
Despite their longevity, polymer film capacitors are not a stagnant technology; instead, they have evolved to become smaller, more reliable and ever-higher performing. Polymer film capacitors have gone from a simple wound film-foil construction to using metallized plates to stacked construction and, finally, to their latest iteration of multilayer polymer (MLP) construction. Polymer film capacitors are available in axial and radial lead configurations, as well as special surface-mountable constructions to meet various assembly needs.
The polymer film market is not a cottage industry producing job shop quantities of highly specialized capacitors. But rather, it's a high-volume, highly automated industry supplying billions of capacitors to every conceivable market. Polymer film capacitors can be found in markets ranging from the automotive industry to zero-current-switching power converters.
Ceramic Capacitor Cracking
For many years, there has been a long-standing debate about the use of multilayer ceramic capacitors (MLCCs) versus polymer film capacitors (wound, stacked or MLP). As circuit designs have shrunk in physical size, multilayer ceramic capacitors have been winning the design-in race with available case sizes down to 0201. Unfortunately, while ceramic capacitors do an outstanding job in small sizes and at low voltages, inherent problems quickly arise in applications where the case size and applied voltages are increased.
Multilayer ceramic capacitors made from large slips — large dielectric plates — tend to crack (Fig. 1) due to circuit-board flexure and mismatched coefficients of thermal expansion (CTEs, also known as TECs). In contrast, MLP polymer capacitors have CTEs almost identical to FR-4 pc boards. While the MLCC crack alone is not catastrophic to the capacitor, it becomes a nexus, allowing moisture and other ionic contaminants to enter the capacitor. At some point, the crack will become a conductive parallel path that ultimately results in a low-resistance path or short-circuit failure (Fig. 2).
To prevent these problems, using empirical observations, many industries have simply decreed that they will not use multilayer ceramic capacitors with greater than an 1812 case size or voltage ratings higher than 50 Vdc to 100 Vdc. In other words, in low-voltage signal processing, multilayer ceramic capacitors are usually the capacitor of choice, while in power-handling applications of 42 V and higher, polymer film capacitors provide the best performance and reliability (see the “MLPs Versus MLCCs”).
To emulate the inherent mechanical flexibility of pure polymer film capacitors, various ceramic capacitors are adding an additional termination layer of conductive polymer to provide a cushioning effect during soldering and board flexing (Fig. 3). According to manufacturers' specifications, adding this layer allows for a board flexure of up to 5 mm before ceramic capacitor cracking begins.
Flexure or pc-board bending is a significant source of stress that can lead to component failure. Ceramic capacitors are inherently brittle and can exhibit catastrophic failure when cracked during pc-board bending if the crack propagates across opposing electrodes and there is sufficient energy present in the power supply. MLP capacitors are made with polymer films, which are not brittle under normal conditions and are more forgiving when physically stressed.
The most common test procedures for this type of robustness follow EIAJ specification RC3402, where a capacitor is reflow soldered to pads on a test PWB. The assembly is mounted component face down, supported on the PWB ends and bending stress is applied to the backside of the assembly with a ram directly behind the component under test. The basic setup is shown in Fig. 4a.
Capacitance shift is used to detect failure under test conditions, but this might not detect cracking of ceramic capacitors. The standard also uses a 1-mm deflection as an acceptance level for no failures. A test PWB with 1 mm of deflection is shown in Fig. 4b. A maximum deflection of only 1 mm is difficult to achieve at every step of PWB assembly and final product manufacturing to eliminate flexure cracking of ceramic capacitors.
Flexure testing has found that, while all nonflexible-termination 1812-ceramic chip capacitors tested in this particular test set failed between 3 mm and 4 mm of deflection, MLP chip capacitors flexed at 7 mm and subjected to 500 hr of accelerated life testing showed no failures or degradation. Throughout the testing, it was evident that MLP capacitors did not exhibit failure or degradation when tested at or beyond deflection values that cracked ceramic capacitors of similar size and values.
While the new flexible-termination MLCC capacitors are guaranteed to withstand up to 5 mm of deflection, it comes at the cost of adding an additional nonmetallic interface in the power stream of the capacitor.
Anyone familiar with the use of conductive epoxies has firsthand knowledge of the potential pitfalls of adding series impediments to current flow, especially in high-power applications.
While the use of flexible terminations on MLCC capacitors might help with preventing cracking from board flexure, these capacitors are still subject to cracking from internal differential stresses caused by both thermal and piezoelectric influences.
Think of the flexible terminations like the rubber tires on your car. They might help cushion you from the bumps and irregularities of the road, but they do not help when you hit a tree at high speed (voltage surge, or extreme thermal excursion).
Base Metal Electrodes
Although there are a variety of metals used as the electrode material in polymer capacitors, the most common one has always been basic aluminum. Pure aluminum eliminates voltage potential differences caused by using dissimilar metals that can result in potentially damaging galvanic reactions.
While the polymer capacitor's use of a base like aluminum was done to enhance the performance and improve the reliability of the product, MLCCs are aggressively seeking base metal electrodes for a far different reason. Because of the escalating costs of the old tried-and-true precious metal electrode materials, such as palladium, the ceramic capacitor industry has been driven to convert to using much cheaper base metals, like nickel (Fig. 5).
Materials will only handle so much stress and, no matter how much clever engineering is done, there eventually reaches a point where performance and reliability start to suffer. The base metal conversion on multilayer ceramic capacitors is a good case in point. The use of nickel is not new to ceramic capacitors; this metal has been used for many years as a pacification layer on the capacitor termination internal endface to prevent leaching.
While many of the problems of getting proper adhesion between the dielectric plates and nickel electrodes have been overcome, physics is still physics. A layer of nickel oxide is physically larger than a layer of nickel. As oxidation takes place within the capacitor, stresses will be applied to the dielectric plates that will reduce the capacitor's long-term reliability. There is a tremendous difference in performance behavior between using nickel on the outside of a component and having it trapped between ceramic layers internal to the capacitor.
In its primary market of small case sizes and low voltages, use of base metal electrodes will not present much of a problem, but when used in power-handling applications, the effects could be catastrophic.
Aging
Like batteries, electrolytic capacitors, such as aluminum electrolytic and tantalum, operate through a chemical reaction. Because of entropy, time will eventually slow, stop or reverse that reaction, and the capacitors will become nonfunctioning. Electrostatic capacitors — ceramic and polymer film — do not function through chemical reactions. Nevertheless, ceramic capacitors contain certain base elements and dopants that can radically affect their longevity.
The most commonly used ceramic capacitors are based on a barium-titanate dielectric that exhibits aging, which reduces capacitance values over decade-hours of time. Basically, the dielectric relaxes, or transforms, to a lower dielectric constant material, thereby decreasing the capacitance of the MLCC. Decade-hours are industry time periods of 1 hr to 10 hr, 10 hr to 100 hr, 100 hr to 1000 hr, etc. For example, an X7R ceramic capacitor loses between 6% and 7.5% of its capacitance value in its initial 1000 hours, while a Z5U capacitor can lose almost 20% (Fig. 6).
Ceramic-capacitor manufacturers carefully guard band their capacitance value testing to compensate for the capacitance loss, so that by the time the capacitors are first used by the capacitor user, they are initially within tolerance. Unfortunately, capacitor aging occurs both on the shelf and while in use, meaning that in applications that require the capacitance to remain within a specific range, in-plant testing by the capacitor user can produce acceptable results, while end-user testing at a later time could produce unacceptable results.
Ceramic capacitors also can be “de-aged” by subjecting the capacitors to temperatures above the dielectric's Curie point (greater than 120°C) after which the aging process starts anew. This de-aging characteristic could prove to have interesting repercussions in applications using the new high temperature, no-lead solder assembly operations. For example, initial capacitance readings on the MLCCs could be well within the acceptable range, but after being subjected to high soldering temperatures, the capacitance values could be outside of a workable tolerance range, causing equipment malfunction. After a thousand hours or so, the capacitors may age back to acceptable values.
MLCCs also exhibit increasing capacitance instability with increasing applied voltage (Fig. 7). An applied voltage of 100 Vdc can result in capacitance losses of up to 40%, while an applied voltage of 400 Vdc produces losses of up to 80%. Fig. 8 shows how the dissipation factor of MLCCs increases with increasing applied ac voltage.
One of the most important attributes of a capacitor used in power circuits is its equivalent series resistance (ESR). ESR determines the I2R heating losses for the capacitor, which in turn establishes the efficiency, pulse handling and, indirectly, the reliability of the circuit. Fig. 9 shows comparisons of the ESR of various dielectric systems and how they vary with frequency. A summary of the performance benefits offered by MLP capacitors is provided in the table.
MLPs Versus MLCCs
Because of the differences in their dielectric constants, it is possible to pack more capacitance per unit volume with ceramic material than with film. This benefit helps account for the great popularity of MLCCs in small case sizes. Cost is another factor because pricing of the small MLCCs has been driven down through competition and the use of base metal electrodes.
Nevertheless, in the larger case sizes (greater than 1812) and higher voltages (greater than 100 V) film capacitors can match the board footprint of the ceramic capacitor by migrating to higher profiles. Meanwhile, film capacitors weigh less than MLCCs because of the heavier dielectric and electrode materials used by ceramics. Barium titanium oxide (BaTiO3), the commonly used dielectric in MLCCs, has a density of about 5.85 g/cm3 versus 1.39 g/cm3 for a polymer dielectric. The greater weight of the ceramic capacitor's electrodes adds considerably more weight to this difference. And since the large MLCCs typically employ electrodes made from noble metals rather than the less-expensive base metals, film capacitors in the larger cases sizes are less costly than ceramics of the same size.