In this blog post we discuss the key solver capabilities that OnScale provides to RF engineers designing acoustic filters. We describe the process of simulating different resonators based on surface acoustic wave (SAW) and bulk acoustic wave (BAW) technology.
Mobile device manufacturers and carriers alike are currently racing to be the first to deliver a fully 5G enabled device. However, there are some significant challenges in the way. The RF front ends and resonators that they contain are extremely complex devices. Full 3D studies of these devices are critical in understanding and optimizing RF filter designs, and shortening time to market. Legacy CAE tools are simply incapable of performing complete 3D design studies, due to limitations both in terms of memory and solution time. To support design engineers in overcoming these challenges, OnScale has developed a fast and efficient finite element method (FEM)-based software package for full 3D simulation of these devices. OnScale is fully cloud-enabled, empowering engineers with high-performance computing (HPC) resources for rapid exploration of design spaces that would be impossible with competing CAE platforms. One of OnScale’s key capabilities that specifically helps RF engineers design acoustic filters, is our coupled piezo-electric solvers without the requirement of a separate license. Engineers are able to directly simulate piezo-electric materials with full anisotropy in a single simulation seamlessly..
Why would one wish to use finite element analysis (FEA) for the design of RF filters? Well as engineers, we're constantly looking for ways to gain more insights about problems and FEA allows us to do this with an approach that is much faster than imperial testing. It also has the added advantage of interrogating our results much more thoroughly and can give us a deeper understanding of the physics and the challenges that we're facing in this problem. As a result, FEA is an indispensable part of modern R&D for an engineer.
We are going to go through a few examples of what you can do in OnScale for the design of RF filters. We look at a full 3D simulation and the computational metrics behind, a Solid Mounted Resonator (SMR), a Thin Film Bulk Acoustic Resonator (FBAR) and some of the interesting things that you can do with a full 3D SAW resonator model.
Above is an example of a typical Surface Acoustic Wave (SAW) resonator configuration. In the centre we have 100 interdigital transducers (IDT pairs), the IDT’s will act as the electrodes. On the edge of the active area there is a bragg grating reflector which will keep the energy within the resonator. There are several materials within, first being the piezoelectric substrate which for this example is made of lithium niobate (LiNbO3). The IDT’s and the bragg reflectors deposited on top of the LiNbO3 substrate are made of gold. Meshing the surface acoustic waves that we need to resolve in this simulation will require quite a lot of elements. This specific model requires around 34 million degrees of freedom. Good News! This can be solved very efficiently. We can solve this in just over four hours on a standard 16 core system and we can do it in less than five gigabytes of RAM.
In this 3D simulation, there are some very common metrics that would definitely interest a SAW designer! For example, we are able to see metrics like electrical impedance profile where you can see the evidence of spurious modes before and after the pass-band. If you wanted to look at a broader frequency spectrum, you are able to see some additional modes. You can also pull out any one of these modes and look at the resonance of it. This can give insights into the origin of these spurious modes , what physical mechanisms are creating them and methods to optimize the design to suppress these modes
Solidly Mounted Resonators (SMR’s) are another type of resonator that are commonly found in an RF acoustic filter design. It features an active piezoelectric layer that sits on top of a bragg reflector, which consists of multiple layers of alternatively high impedance and lower acoustic impedance materials. Similar to SAWs we would look at the filter performance of the SMR’s to see the pass band and look at the presence of spurs or spurious modes and how they affect the Q and the overall performance. Our solvers are particularly useful as OnScale can choose regions of the domain to apply a particular type of physics. So in this case we can choose the piezoelectric layer as the electrical mechanical solver, which is computationally intensive rather than computationally expensive. We can locate it just at that region where it's needed, making it less expensive and a more efficient mechanical solver for any frame substrate or bragg you wish to include in the simulation. This type of optimization gives us the ability to greatly reduce the amount of RAM and the amount of computing time that we need to solve these problems in full 3D.
You might wonder why one would want to explore 3D simulations for SMR’s. When exploring design concepts for SMR’s, often there is a frame that's applied on the outside of the active layer. This frame design can be tweaked and optimized to ensure that the performance of the filter is optimal.
Shown in the figure above, is a very simplistic frame design which will help to demonstrate how we can use an arbitrary shape of frame design with an arbitrary slope of curve between the active layer on the top of the frame design. This can affect the filter performance. On the right hand side you can see in the blue line - this is the phase profile for an SMR that has no frame and just a really simple resonator. The black line shows the impact of the frame on that. It has caused a dip in the performance of the filter, however, it has smoothed out the pass-band a little bit and we no longer have those ripples. This is just one of the concepts that you can explore with a 3D simulation tool where you can employ arbitrary materials and geometries.
Lastly, we look at 3D simulations of FBARs. All of these resonators are a function of the material properties and their 3D geometry, which can drastically affect the performance of the filter. In the above example you will see we have a pentagonal electrode shape that we use on top of the resonator for energy confinement. To further optimize energy confinement by using the pentagonal shape while keeping the same electrode area - we set up a design sweep for different orientations of pentagonal shape (shown above) to identify the optimal orientation of the pentagonal electrode for the best possible FBAR performance.
All of these simulations can run simultaneously in OnScale. We can now take a look at the impedance results. The grey line shows the results that we can obtain from a simple square electrode shape on a FBAR and the blue line shows what we can get from an optimized pentagonal design. This illustrates the concept of how 3D simulation can help to improve filter design quite rapidly.
OnScale has the ability to rapidly optimize RF filter design and has been described as a game changer by our clients. Our ability to deploy multiple parallel simulations in the cloud, without the restrictions of hardware and licensing has allowed engineers to dramatically compress the simulation aspect of the design cycle. Combined with the ability to run full 3D simulations, drastically removes risks from any design before going out to a hardware prototype.