Radio Frequency Filters For 5G: What They Are And Why They’re Worth The Trouble

By David Haynes, Daniel Shin, and Lidia Vereen­

In the recent blog article “Our wireless world – how Wi-Fi 6 will seamlessly integrate with 5G to keep us connected,” David Haynes from our Customer Support Business Group (CSBG) explained how this new generation of wireless technologies will improve our connectivity by using higher frequencies and greater bandwidth than current 4G and Wi-Fi solutions.

But he also concluded by saying that the coexistence of 5G and Wi-Fi 6 and the need to operate at these higher frequencies, above 5 GHz, posed technical challenges for our customers; specifically, the technical demands on the filters that allow signals in the band to be separated.

In this blog, we will explore how these radio frequency filters work, why they’re so important, the challenges chipmakers face when building cellular devices, and how Lam is helping to solve them.

Frequencies and filters 101

Frequencies can be found everywhere in nature, necessitating that you identify their unique ranges to filter out those you don’t want to listen to and isolate the ones you do.

Filters work by turning down – or, ideally, eliminating – frequencies that we’re not interested in. For example, a stereo can enable you to filter out treble frequencies and focus on the bass when listening to music; with a camera, you can filter out ultraviolet (UV) light to improve image quality.

In the cellular spectrum, the massive range of frequencies available is divided into channels, so a conversation can take place on one channel without interference from conversations happening concurrently on others. However, this only works if you can isolate a channel from all of the other frequencies in that spectrum.

Radio frequency (RF) filtering is what enables isolation and use of specific frequencies in a channel without having to address all the other channels that exist at the same time.

There are four ways to filter these frequencies:

1. Filter out the high frequencies and pass through only low ones.
2. Filter out the low frequencies and pass through only high ones.
3. Isolate some range of frequencies, eliminating all frequencies above and below that range – this range is referred to as a “band,” and such a filter is a “band-pass” filter.
4. Eliminate only one range of frequencies, keeping all others intact – this is referred to as a “band-stop” filter.

RF filters are therefore critical to our modern cellular data systems. Each channel is a band and some modern cellular phones may have as many as 60 band-pass filters in them, each isolating a single channel.

Cellphone filters

There are two main types of cellphone filters. The first has vibrations that run along the surface of the filter – these are called surface acoustic-wave, or SAW, filters. They tend to be less expensive to build, and they work best on the lower frequencies of the cellular range.

The second type has vibrations that run through the entire material, not just the surface – they’re called bulk acoustic-wave, or BAW, filters. While somewhat more expensive to make, they handle the upper reaches of the cellular spectrum.

RF filter manufacturing challenges

Manufacturing of RF filters presents a number of challenges, as there is a continuous pressure to reduce device size, especially for mobile and IoT applications, and better performing, more complex filters require a higher degree of precision. Adding to this, filter architectures and the materials used are evolving to take advantage of both the higher frequencies and greater bandwidths available in 5G.

Let’s look in a little more detail at one of the critical steps in RF filter manufacturing.

Depositing and etching Sc-doped layer with high throughput

Leading developers are looking to increase filter bandwidth by modifying a critical aluminum nitride (AlN) layer with the addition of scandium (Sc). This improves the piezoelectric properties of the AlN layer, and the final filter performance. Solmates, the company that Lam recently acquired in The Netherlands, is focused on deposition of these AlScN films with best-in-class Sc doping levels and film properties.

The addition of Sc creates a far more difficult material to etch, which can negatively impact throughput. Furthermore, the etch process must stop on the bottom electrode layer with high selectivity – any etching into the bottom electrode will negatively affect device yield. Finally, the bottom electrode will be thinner when compared to previous generations of the device, presenting the challenge of achieving a uniform etch without impacting the bottom electrode, which can negatively affect device yield.

Lam’s Kiyo family of etch tools offers both the high etch rate and selectivity needed to overcome these challenges. Available for both 200 mm and 300 mm diameter wafers and used in high volume manufacturing today, Kiyo has the high bias power required to achieve the required profile while maintaining a competitive etch rate.

Filters are only one part of the solution

RF filters are one key component of these new RF systems, but they are not the only one. Filters are combined with other devices such as RF switches, low noise amplifiers, power amplifiers, and antenna tuners to form complex RF module solutions. Many of these other RF devices are fabricated using RF-CMOS (complementary metal oxide semiconductor) or RF-SOI (silicon on insulator) technologies, but have specific fabrication schemes that allow capacitor and inductor elements to be integrated into the process back end of line (BEOL). These elements are essential for the efficient operation of the devices at high frequencies. Just like the challenges in RF filter manufacturing, these additional BEOL integration steps also pose new challenges for Lam process tools.

Depositing high quality MIMCAP

Metal-insulator-metal capacitors (MIMCAPs) are now commonly integrated into RF devices. As the name implies, MIMCAPs are made of metal layers that conduct electrical signals and power, and dielectric layers that provide insulation between the metal layers. The dielectric layer, typically silicon nitride, must be high quality and have excellent adhesion to the metal layers.

Lam’s VECTOR Express provides the high-quality film deposition needed. Its multi-station sequential deposition (MSSD) architecture tightens wafer-to-wafer non-uniformity and maintains superior within-wafer uniformity.

Thick passivation layer deposition with low CoO

An often overlooked challenge involves the final passivation layer: it has to be thick to completely seal the device, protecting it from the environment. Any breaks or pinholes can affect device performance, so depositing films thick enough to provide the desired hermeticity can require multiple passes, greatly reducing throughput and increasing cost of ownership.

VECTOR Express delivers high quality deposition of thick USG (undoped silicon glass) films with high productivity. As in the MIMCAP application, the MSSD architecture ensures excellent wafer-to-wafer non-uniformity and pinhole-free deposition for thick passivation layers.

Daniel Shin is senior customer technology manager for Reliant Systems in Lam’s Customer Support Business Group.

Lidia Vereen­ is business development director of Reliant systems in Lam’s Customer Support Business Group.