Integrate multiple discrete capacities into a solid-state device

Power system designers, whether data servers or data centers for the Internet of Things, are constantly required to achieve greater power densities and improved conversion efficiencies. While semiconductor switching devices have received a lot of attention to make these advances, capacitors are also an important design component that helps engineers meet their energy storage, filtering, leveling, and tuning needs.

However, the development of capacitors has not kept pace with the changes seen in the semiconductor world and even cutting-edge technologies such as multi-layer ceramic capacitors (MLCC). MLCCs are monolithic electronic devices that consist of alternating layers of metal electrodes and ceramic dielectrics. High temperatures are used during the manufacture of MLCC laminates to create sintered volumetrically efficient capacitive devices. Next, a conductive termination barrier is built into the exposed edge to complete the connection.

Empower semiconductors (Empower) has developed its own 220 nF capacitor technology (E-CAP), a series of Integrated voltage regulator (IVR) After recognizing the shortcomings of conventional capacitors. In an interview with Power Electronics News, Steve Shultis, Senior Vice President of Sales and Marketing at Empower, talks about how the benefits of E-CAP can help businesses benefit from advances in IVR systems. did. “E-CAP combines different discrete capacitances into one solid-state component,” he said. “To take advantage of superior performance, size, configurableness, durability, and stability, Empower currently offers E-CAP silicon capacitor solutions in many difficult application areas.

“When we were developing the first IVR platform four years ago, we spent a lot of time not even maintaining the 100MHz or 200MHz switching frequencies of the first generation IVR, even with the highest performance ceramic capacitors.” He added. “Thus we realized that we needed something new, and together with our partner TSMC, we found a technology that used this technology and became a specialist in its design. Like IVR, power management. It includes and runs it on routes that were not previously considered. “

Empower has revealed that its partner is TSMC and is helping to implement E-CAP. “The intellectual property of the design is ours, but because we can use the process through TSMC, we can acquire the intellectual property and apply it to another process,” says Shultis.

Last week, Empower Semiconductor announced that it has expanded the E-CAP family of silicon capacitors with new technologies that further improve density and performance.The latest E-CAP solution provides a density of 1.1 µF / mm2.. In addition to density, thickness levels with an overall height of less than 50 µm can be achieved. You can combine multiple matched capacitance values ​​from 75pF to 5µF (2V) into a single die to create a custom integrated capacitor array. Package options based on bumps, pads, and pillars allow designers to choose the best solution based on specific system constraints (Figures 1 and 2).

Figure 1: ECAP-based solutions provide more than five times the density compared to standard MLCCs. (Source: Empower Semiconductor)

E-CAP technology

E-CAP integrates multiple capacitors into a single solid-state device, providing the flexibility and efficiency of silicon. According to Shultis, this technology combines the characteristics of enhanced Equivalent Series Inductance (ESL) and Equivalent Series Resistance (ESR) to significantly reduce parasitic capacitance and nearly five times the capacitor density of major MLCCs. Achieve density.

Shultis emphasized the ability of E-CAP technology to achieve thickness levels of less than 50 µm. It is ideal for supporting next-generation data-intensive systems that require high-frequency operation from the smallest form factor and maximum efficiency. IoT, wearable, mobile, and processor sectors where size, performance, and flexibility are important. E-CAP solutions help designers reduce the cost of BoMs and the risk of circuit failures.

“The more capacitors that are integrated into a particular region, the more capacitors can be placed in a smaller region, giving us the flexibility to accommodate a variety of common applications below 4V,” says Shultis. “High voltage work requires special processing with trade-offs. Empower is scouting the landscape for potential new perspectives.”

Figure 2: E-CAP and MLCC (Source: Empower Semiconductor)
Figure 3: In the design example, 10 MLCCs are replaced with 1 E-CAP die. A single die solution for 9 capacitor requirements for ultra-dense applications. (Source: Empower Semiconductor)

Moreover, unlike MLCCs, which require multiple devices to account for derating from voltage, temperature, and elapsed time, E-CAP has minimal or no AC or DC bias derating requirements. .. As a result, capacity does not need to be “over-engineered” and derating does not need to be considered.

“While using MLCCs, calculations are needed to determine the proper temperature, voltage, DC rating, etc.,” says Shultis. “E-CAP does not need to be over-engineered due to its low voltage, temperature, and aging derating. To demonstrate this, we excel in terms of qualification, reliability, and client production. Performance may be quoted. ESL is another non-negligible component, so ESL is used in high performance decoupling systems because it requires very little inductance and is provided by capacitors throughout the line. There are also low-voltage ceramic capacitors, but because they are made of highly customized special ceramics, they usually cannot achieve single-digit pico-henry values ​​and are costly. “

Figure 4: Empower E-CAP design example (Source: Empower Semiconductor)
Figure 5: MLCC structure and silicon trench capacitor (Source: Empower Semiconductor)


With an ultra-low ESL of only 15 pH and a package height of 50 µm, flexible capacitors are available via E-CAP for use in data center servers and IoT devices from mobile and wearable devices. E-CAP has a good frequency response and a strong ESL, so its impedance at high frequencies is low. Figure 3 contrasts the two implementations. One uses E-CAP instead of high frequency capacitors and the other uses a large number of MLCC capacitors to improve the frequency response of the decoupling solution. The results show that E-CAP can reduce the number of components by 40% while reducing the impedance at high frequencies to half the impedance achieved with traditional MLCC-based solutions. Is shown.

“According to Figure 4, the roadmap aims to offer a variety of options, but as we talk to more and more customers and applications, a significant percentage of the market is 200nF or 400nF in general. I understand that there are requirements, “Shultis said. “We are already practicing some 2nd generation multi-cap type designs. Those with 18 capacitors are most likely to be regular items. Arrays are about 2x2mm , 400 µ pitch, total capacitance is about 4.8 µF. Therefore, it can be used in most PCB mounting situations. Second generation can be less than 50 µ thick under the board. The very small decoupling capacitors in are so thin that they fit within the height limits even after soldering or when mounted inside a board or PCB layer, so these packaging techniques and methods It is working. “

Shultis argued that E-CAP capacitors are safe to use in magnetic field conditions and resistant to low frequency noise (Figure 5). “This is a challenge for some low frequency applications that use MLCC ceramic capacitors,” he said. “We have encountered this problem several times. In addition, nickel plating the pads of ceramic capacitors makes them cumbersome to use in medical applications because they are sensitive to magnetic fields.”

post Integrate multiple discrete capacities into a solid-state device First appeared Power Electronics News..

https://www.powerelectronicsnews.com/integrating-multiple-discrete-capacitances-into-a-solid-state-device/ Integrate multiple discrete capacities into a solid-state device

Back to top button