Build to Print: An Introduction to Thin-Film Component Design  

by Peter Matthews, Director of Technical Marketing, Knowles Precision Devices

Components keep shrinking while delivering higher capacity and enhanced connectivity in what might be called “thin is in” regarding high-frequency RF devices. The challenge is getting the most out of these tiny products, and build-to-print is a good choice for achieving it.

The build-to-print approach to component manufacturing is a step beyond build-to-spec. It is a collaborative effort with a components manufacturer and customer working together through every step of the production process. At a high level, a customer brings a drawing or schematic, after which engineers will review the spec drawings and look for areas of improvement before moving into design creation. Those schematics are used to create a portion of the circuit and, ultimately, a prototype for production.

What is Thin Film?

Thin film is a layer of material used to fabricate electronic components ranging from fractions of a nanometer to several nanometers in thickness. Many companies that offer a build-to-print model facilitate thin-film product design, manufacturing, and testing from prototype to high-volume production. Constructing circuits using thin film offers several advantages over other techniques, including precision patterning and the ability to make components much more compact versus other material options.

Customization is the main theme of build-to-print, which is why many companies use thin film to create their products. Thin-film substrates offer a sufficiently high dielectric constant (K) well suited for miniaturization and temperature-stable performance. Thin-film structures can also be built to exact specifications, particularly when size, weight, and power (SWaP) are major concerns, as it reduces complexity and size while optimizing performance.

Who Opts for Build-to-Print?

A customer will partner with a component manufacturer specializing in build-to-print when it needs a product so specific that it can only be created using this approach. Build-to-print is an excellent option for many applications, including heat sinks and standoffs, integrated passive components, resistor-capacitor networks, Lange couplers and power combiners, EMI and other filters, and MMICs. It is also desirable for bias decoupling and filtering, lumped-element impedance matching networks, power amplifier stabilization, and power-combining networks (Figure 1).

How the Build-to-Print Process Works

Once a company decides to proceed with build-to-print for its component, the next steps include working with a manufacturer to create a customized solution. The first step is determining the most appropriate substrate material for an application. Companies often stock common materials, including alumina, aluminum nitride, high-dielectric titanates, quartz, and sapphire, as well as a wide variety of custom ceramics. They can work with standard thickness materials and offer surface finishes of “as fired,” lapped, or polished.

To determine the right fit for a build-to-print application, consider the following characteristics of the substrate material and the desired application:

• Operating frequency
• Size
• Dielectric constant
• Material type
• Thickness
• Transmission line impedance
• Mechanical requirements (TC, CTE)

Once a customer chooses the appropriate substrate, the next parameter to consider is the selection of metallization (Table 1). This includes making decisions about solderability, solder types, solder hierarchy, circuit attach, skin depth, conductivity, and thermal extremes. Typically, metallization is the largest factor in determining lead times and process compatibility.

Next, it’s time to establish the best processes to make everything fit together and function appropriately. The following laser-cutting techniques may be required depending on what type of project is requested:

• Laser scribing – The laser perforates the ceramic substrate so the customer can break apart the ceramic at the appropriate point when ready to use.
• Laser machining – Used to cut the ceramic materials to make the “puzzle pieces” that form each plate.
• Laser trimming – A precision technique in which lasers carefully remove small bits of resistive metallization on the ceramic to make final adjustments to bring the part into spec. This includes techniques such as plunge, scan, and serpentine cuts.
• Laser obliteration – This is like laser trimming but can remove gold, nickel, and copper to achieve precision tolerances.

With lasers, the smaller the wavelength, the more capable it is of performing high-precision cuts. CO2, fiber, and UV lasers can perform different jobs. Invented in 1964, the CO2 laser is one of the oldest gas lasers. As it delivers high power levels and is relatively low cost, it is a common machining workhorse. Similarly, fiber lasers use doped fiber as the gain medium. UV lasers operate at much smaller wavelengths than CO2 and fiber lasers, so they can provide much tighter control over feature size, making them suited for tiny cuts like those needed for laser trimming and obliteration. The latter two types of lasers offer a range of wavelengths and different levels of precision compared to CO2 lasers.

In general, the conductor layer for all circuit designs contains critical dimensions necessary for compliance and performance. It is important to consider these dimensions when designing thin-film circuits to achieve appropriate functionality to reduce costs.

The Via  

A via is a small opening in a circuit that provides a conductive connection between different layers in the circuit or a connection to the groundplane. In any circuit medium, vias can offer many advantages (Figure 2). That said, many designs tend to overpopulate ground areas, and poorly placed vias can degrade yield due to reduced substrate durability in manufacturing. Consequently, it is critical in hard substrate manufacturing to consider via quantity and location carefully.

Tantalum nitride (TaN) is a widely used resistor material because it offers a higher maximum exposure temperature, a wider annealing window, superior resistance to the harsh environment encountered in soldering and other processing, and well-established self-passivating characteristics. For applications where minimum resistor change over temperature extremes is critical, nickel-chromium (NiCr) resistor films are a better choice.

When using solder in a circuit, a solder mask or dam must usually be incorporated. This is an area of non-solderable material applied to the patterned traces of a thin-film circuit to prevent solder from flowing away from surface-mounted components during solder reflow. Unwanted solder flow can cause surface-mounted components to move during soldering or cause solder joints to be thinner.

Many microwave circuits require components that combine or split power or a spiral inductor for filtering and tuning purposes. However, when using traditional connection methods such as wire bonding, these components can create challenges for the thin-film assembly process and are a common source of damage during manufacturing and testing. This is mainly because the parasitic effects of some wire bonds make tuning and stable performance difficult to achieve, especially over broad frequency ranges.

When specific components are critical to a design, it is best to use a connection option known as a supported bridge, as it provides a consistent, reliable method for incorporating these structures into a design. With supported bridges, functionality testing can be completed on these components to ensure performance before integrating high-cost active devices.

In general, the build-to-print process and thin-film technology are uniquely suited for designing and building components and microwave-integrated passive devices (IPDs) based on planar transmission lines, which are conductors or, in some cases, dielectric strips that are flat ribbon-shaped lines (Figure 3). There are several types of planar transmission line designs, including microstrip, stripline, and coplanar waveguides. More complex integration between transmission line elements and discrete capacitors, resistors, and inductors to create IPDs is a challenging task. If this is not done correctly, condensing multiple passive components into a small device can lead to problems.

Biasing is the process of getting DC voltage from point A to point B in the most appropriate way. A bias network assists with this by combining capacitors and resistors to meet the needs of the device in which the circuit will be used. To make a bias network, the components are fabricated on the same substrate to save space and assembly costs. Three of the most common types of bias networks are the bias filter network, the self-bias network, and the bias tee. For build-to-print, if a customer brings a specific design, the manufacturer should focus on developing bias filter networks and self-bias networks.

It’s not standard practice to test beyond validation in most build-to-print shops. However, Knowles Precision Devices and others offer comprehensive testing because they also offer build-to-spec and customized component designs in-house. From basic quality tests to qualification testing to classified tests, some government contractors must comply with DD254, which is a legal document that provides the security requirements and classification guidance required to perform a classified contract.

(52)