PACKAGING
Enabling High-Performance RF Devices with Photosensitive Glass Ceramics
A lower cost, highly accurate way to integrate passive devices.
by JEB FLEMMING
Photosensitive glass was invented in November 1937 by Dr. Donald Stookey of the Corning Glass Works. It was made public 10 years later, on June 1, 1947, and patented in 1950. Most will know glass ceramics from their glass stove top or the iPhone 12 Corning Ceramic Shield screen.

Glass is amorphous, meaning it has no crystalline structure. It’s just a random assortment of molecules in a solid matrix. Ceramics, on the other hand, are crystalline structures of various types and compositions. Glass ceramics can exist in both the amorphous glassy phase and the crystalline ceramic phase. Glass ceramics are used in either one of those two states: 100% glass or 100% ceramic. For example, a Brown stove top is 100% ceramic, and the Samsung Gorilla Glass screen is 100% glass.

Photosensitive glass ceramics are a small subset, where microscopic regions of glass are converted into the ceramic phase. We call this a “dual-phase” or “two-phase” ceramic/glass structure, mostly glass with little bits of ceramic in them.

3-D Passives
Passive components have a rather boring name, which undervalues their role in electronics. In fact, passives are an essential part of any RF circuitry. One way to think of them is as the circulation system of RF chips. Historically there have been two major paths for manufacturing passive components.

The first path is discrete surface mount technology, in which a manufacturer will make a chip inductor and chip capacitor, and then assemble those on a PCB.

close view of SIP with integrated passives and wire-bonded IC
Figure 1. SIP with integrated passives and wire-bonded IC for fully functional system-in-package.
The second path, called integrated passive devices (IPDs), is where an inductor and capacitor are combined in a manufacturing process into a single device, not attached individually on a substrate such as FR-4. The value in this approach is due to the shorter connections between components, and the elimination of solder and solder pads, which cause significant system losses. And with IPDs, efficiencies go up and power consumption goes down. Both historical processes are planar and two-dimensional, which limits the quality factor, called Q (higher quality factor means lower loss), especially in an inductor.

The 3-D IPD (three-dimensional integrated passive device) offers the best performance of all. The integration of inductors, capacitors and resistors embedded into a single chip results in low system losses (FIGURE 1). 3-D inductors provide larger inductance value per given surface area and higher Q due to enhanced magnetic storage. Reduced parasitic capacitance increases the self-resonant frequency (SRF), permitting higher operating frequencies. An added benefit is precision 3-D control of distance between inductor coil windings. This improves the tolerance of inductance and SRF, both essential to higher-frequency applications.

PSG (photosensitive glass-ceramic), the backbone of 3-D IPDs from 3D Glass Solutions, has exceedingly small through-holes (vias) down to 10µm that are used for electrical redistribution through the glass. This is done to spread out the inductor windings to both the top and bottom of the PSG to enhance the magnetic storage without increasing the surface area. Alternately, for a given inductance and surface area, the inductor turns could be spaced farther apart, reducing the parasitic capacitance between turns to boost the SRF. Benefits of PSG for electrical devices include getting the RF properties of glass with the strength of ceramics, and a low-cost, high-precision batch manufacturing process to produce the vias in glass. Additionally, PSGs have an intermediate thermal expansion, so vias can be filled with copper to make low-loss connections between the top and bottom of the device.

Being able to perform additional processing adds flexibility. For example, an inductor may be integrated by making a cavity and filling it with a magnetic material. These are second, third and fourth degrees of freedom that aren’t available with other types of materials or processes. PSG enables these types of additional features (cavities and filling with magnetics) that can’t be done at scale with ceramic boards, PCBs, or low-temperature co-fired ceramic (LTCC). PSG materials enable additional mask, etch and metal filling manufacturing steps, permitting these second, third and fourth degrees of freedom.

Another benefit of PSG is the cost. The cost is the same to make one hole or a million holes. Competitive technologies that make through-holes in glass rely on lasers, literally lasering every single hole individually.

In the novel technology, holes are not drilled. Instead, they are formed using ultraviolet (UV) light and a mask to start a chain reaction in the glass (FIGURES 2 to 6). In this chain reaction certain molecules accept a photon and donate an electron to a different molecule. This creates a chemical change at the nanometer scale. Baking, the next phase of the process, converts the exposed regions into a ceramic micro-pattern. In the final phase, acid is used to preferentially eat away the ceramic phase, while not touching the glass phase.

The ability to create exceedingly small and precise vias enables much higher density interconnects compared with the traditional PCB process. Density can be increased by stacking several glass layers. Also, the ability to create cavities with ease presents a unique opportunity to tune the properties of the substrate, which is not practical when working with traditional substrates.

Repeatable Manufacturing
It all goes back to chemistry 101. The two fundamental principles are accuracy and precision, and they’re very different. Both are necessary when manufacturing high-performance devices.

On one end of the spectrum is a very precise, repeatable manufacturing process that can make a single through-hole or a million, all going at the same 50µm size. The precision of the novel technology is tight, less than 1µm, compared to other technologies that might achieve +/-7 to 10µm. Building devices with precision ensures that if a million are built, they all behave the same. Such repeatability leads to high part yields and cost savings.

close view of PSG wafers
Figure 2. PSG wafers used in assembly of the novel processes.
The other manufacturing principle, accuracy, means hitting the bullseye on the dartboard every time. The accuracy is possible by utilizing standard lithographic techniques used in the semiconductor industry. This level of manufacturing accuracy, when coupled with accurate 3-D electromagnetic (EM) simulation, leads to fewer design iterations and shorter time to production.

With precision and accuracy, systems can be designed to meet real world requirements. That means RF designs can be manufactured to accurately meet the requirements of a wide array of frequencies, from 400MHz to 300GHz.

Applications Everywhere
Potential applications for 3-D IPD technology span a variety of fields, from 5G/6G wireless, optical transceivers in data centers, wi-fi, internet of things (IoT), ADAS radar sensors for autonomous driving, to gesture recognition, healthcare diagnostics, and much more.

Almost too many potential applications exist for this technology. Recently, we have focused on applications within the biggest markets, as well as some of the most challenging applications. Recently, most requests are around 3GHz to 7GHz because that is where a lot of market demand is. The high millimeter wave (mmWave) frequency, around 70GHz or higher, is another area of focus. They are sweet spots because it is difficult for other technology to create devices that perform well at these frequencies.

The Design Process
With the onset of 5G and other advanced technology requirements, designers are faced with the daunting task of creating circuit designs that operate anywhere from <1GHz to the extremely high frequency mmWave spectrum above 100GHz, sometimes simultaneously, for signal coexistence.

Designs include lumped component IPDs in the sub-6GHz band, ranging in complexity from basic inductors and capacitors to low/high/band-pass filters, diplexers, baluns, couplers, and to complex integrated designs in mmWave frequencies going all the way up above 100GHz.

PSG plate with empty cavities
Figure 3. Mask, bake and etch PSG for vias, slots or other shapes to form 3-D structures.
PSG plate with copper filled cavities
Figure 4. Plate and fill cavities with copper.
PSG plate with interconnections
Figure 5. Plating of front and backside interconnections.
PSG plate with IPD created by metal sputtering
Figure 6. Create IPD by means of metal sputtering.
The novel substrate technology is amenable across this entire range of frequencies. Like any substrate technology, it comes with its own set of process rules that need to be followed to yield reliably manufactured parts.

3D Glass Solutions (3DGS) decided early on to provide designers with resources on how to use the technology. The company built a robust set of design rules and published these to the broader market, so design engineers would have all the information needed to successfully utilize the technology.

Then, 3DGS embarked on a one-year strategic project with Cadence AWR Software to move these design rules to full software-based Product Development Kits (PDKs). PDKs exist for a variety of applications, allowing customers to accelerate their designs without having to spend a lot of time learning the design rules.

Invested in Technology
3D Glass Solutions chose to tackle the issue of scaling manufacturing within its own facility and with factory part- nerships. That way the technology can be scaled up in terms of volume and in terms of location or geography.

3D Glass Solutions is located in Albuquerque, NM. Not every product can be made there, however. In some cases, overseas production is required. A global supply chain is being created that has flexible, elastic capacity, adjusting to demand where needed.

Two of the main investors in 3D Glass Solutions, Murata and Lockheed Martin, are strategic partners. Coincidentally, they are on fundamentally different ends of the spectrum as far as what is needed from the relationship.

Murata is the world’s leading producer of RF devices, mak- ing billions of inductors and capacitors each day. 3D Glass Solutions works with Murata on the next generation of these simple devices.

At the other end of the spectrum is Lockheed Martin in the aerospace industry. Lockheed is principally concerned with size, weight and performance (SWaP) and needs to take evolutionary leaps for its customers, around 5% per year for the next 30 years. 3D Glass Solutions works with Lock- heed Martin on systems-level integration to build complete systems with 3-D IPDs, cavities, and embedded transistors, with five to eight layers of glass bonded together. This system-on-a-chip (SoC) approach significantly reduces the footprint and weight of the device, as well as cost.

Multiple 3DGS PDKs in AWR enable designers to rapidly build up full 3-D EM circuits in the technology, and provide a quick design turn-time. Designers spend less time learning the individual process rules and building parameterized models from scratch. 3-D models of various types of inductors, capacitors and transmission lines are available for drop-in placement in the circuit. Also, the in-built circuit models permit easy tuning of layout parameters. For instance, an inductor could be tuned by modifying its trace width, core width, number of turns, spacing and via diameter, to generate a custom component that fits holistically with the end-requirements; e.g., low-loss, small-size, high SRF or power handling. The PDK also contains quasi-static component models that provide EM-like performance estimates at a fraction of the time and CPU resources needed for a full 3D E-M simulation.

Measured vs. 3-D EM simulated results of manufactured designs have shown exceptionally good correlation between the two.

Conclusion
Simple chemistry and physics expanding on standard semiconductor processes permit next-generation RF devices to perform well. Applying physics to the known chemistry of photosensitive glass has changed the microstructure. This fundamental change permits precision RF designs in glass substrates, improving the performance of high-frequency devices, reducing parasitic losses, and driving miniaturization advancements into electronics.
close view of Antenna-in-package (AiP)
Figure 7. Antenna-in-package (AiP) with transmit and receive chips mounted directly to the novel integrated substrate.
Jeb H. Flemming is chief technology officer and founder of 3D Glass Solutions (3DGS) (3dgsinc.com); jeb.flemming@3dgsinc.com. While at Sandia National Laboratories, he was principal developer of micropost technology, which won a 2007 R&D 100 award. He is inventor or coinventor on more than 30 patent applications.