Industry is pushing for ever-denser packaging and improved performance from flex and rigid-flex circuits, as these are gaining importance in 3-D packaging and helping to solve space constraints, while meeting other design requirements. This has increased the demand for interconnections of a more complex nature. At the same time, apart from being a mechanical solution alone, flex and rigid-flex circuits (IPC-6013 Types 1 through 4) are also critical in meeting other electrical performance requirements. Similar to any other design, these circuits must also be manufacturable, reliable, and meet cost considerations.
Apart from meeting the above requirements, a successful rigid-flex (IPC-6013 Type 4) circuit involves meeting several additional requirements as well. Chief among them being mechanical and thermal requirements along with signal integrity, and the designers must consider them during development. In reality, the final design requires an optimum balance considering various trade-offs between the collaborative efforts of several teams involved with mechanical development, signal integrity modeling and input, qualification engineering, support from the sourcing team, and feedback from the fabricator concerning the manufacturability and cost.
As some parts of the rigid-flex circuit needs to bend, it has to meet the tight mechanical bend radius requirements. Rather than using a simple stack of circuit materials, it helps if the bend area uses an unbonded or loose-leaf, bookbinder cross section. This technique is ideally suitable for applications requiring high-density interconnections within a tight form-factor, where space is at a premium. For instance, a specific application may require bending more than 13 conductive layers at a right angle within a span of 15-16 mm.
Additional requirements of impedance control and good signal integrity properties on some or all the layers may complicate the difficulty of bending at such severe conditions. As good signal integrity properties and impedance control depend on the thickness of prepreg or the interlayer dielectric material, the thickness of the final flex may be significantly more compared to that of any standard non-impedance controlled circuit.
Considerations for Coefficient of Thermal Expansion
Building high-layer-count flexible printed circuits require special considerations. While standard multilayer construction is adequate for most applications, increasing the layer count, or more importantly, an increase in circuit thickness, calls for special attention to be paid to materials used on the stack. This specifically requires all material within the stack be matched for the coefficient of thermal expansion (CTE), as a large portion of the stack-up comprises acrylic adhesive.
Materials making up the typical flex multilayer are a successive layer of dielectric, adhesive, and conductive layers. With temperature rise, expansion of material in z-axis for plated through holes (PTH) is a major concern. With the material swelling in the z-axis, the strain may be enough to crack the PTH barrel, disrupting the electrical continuity. This is the reason all materials in the flex design require a close CTE match. However, although common flex materials show a good CTE match in the planar or X-direction, they do not match in the Z-direction.
For instance, going above the glass transition temperature of 40°C, the Z-expansion for acrylic adhesive rises to 4-times the rate of expansion of other materials in the stack. One advantage of rigid-flex type 4 construction is the use of cutback coverlay and acrylic bonding films.
Designers do not place vias in the flex areas of the circuit, as repeated bending during operation of the circuit can stress them and make the vias tear away from the substrate. Therefore, by eliminating use of acrylic adhesives in the via area, the CTE mismatch is also avoided. For this, the adhesive layer must be stopped at the flex/rigid transition zone, leaving only the prepreg in the via area.
Considerations for Bend Radius
According to the recommendations of IPC-6013, a flex circuit must use a minimum radius of 10-times its thickness at the bending area. Although the standard provides a safe guideline, this does not take into account the initial thickness or the material type. While it is possible to fold over a thin flex on itself, a thicker circuit would require a more relaxed bend. For instance, a 13-layer flex circuit would have a typical thickness of about 0.965 mm, and taking the recommendations of IPC-6013, require a minimum radius of 9.65 mm.
However, accounting for the real world scenario, where radii are never perfect, comfortable installation will require a longer flex length. For the example above, for instance, to form the 90-degree bend, the 13-layer flex would require a length of 15.25 mm. With a loose-leaf, bookbinder construction, it is possible to bend the flex at a much tighter finished radius of 2.54 mm, which is about 3.8 times tighter than the IPC allows.
The Loose-Leaf Approach
The loose-leaf approach does allow going below the minimum recommended bend radius. In this technique, individual layers or sub-composite of the board in the flex stack-up would remain unbonded and separated. Separating the layers into multiple sub-composites within the stackup allows the minimum bend radius to be calculated based on the thickness of the individual layers.
For instance, assuming the thickest sub-composite is 0.3 mm, calculating the minimum bend radius, as per the IPC guidelines, gives a figure of 3.0 mm. This is considerably less than the original design minimum.
However, even with this loose-leaf approach, although the flex region still required the same length as calculated earlier (15.25 mm in the example), the unbonded layers did not possess enough slack to allow for a stress-free bend.
The Unbonded Bookbinder
If each of the unbonded sub-composites is made somewhat longer than the one below it, it allows room for all sub-composites to bend without undue stress to itself or to the layer immediately adjacent to it. The design requires calculating the length of each sub-composite to allow it to rest under its adjacent part without interference.
Although the basic construction is easy to build and reliable, the flex part is more difficult to build because each sub-composite is independent of its neighbor and coverlay is required on both sides of each layer, with air-gaps in between.
Effects of Stress on the Bend Area
It is easy to calculate the stress forces on the bookbinder part of the rigid-flex. For simplicity, assuming a 2-D cross section allows plane strain, while modeling the contact between sub-composites allows observing the effect of sub-composites coming into contact with each other. As flex material does not have any direct modeling parameters available, the same modulus may be assigned to the different flex layers, while the rigid section is treated as steel.
By avoiding all vias in the bend area, and restricting them to the rigid areas alone, designers avoid placing undue stress on the sub-composites.
A loose-leaf, bookbinder construction in a rigid-flex circuit is a difficult construction, as it requires building a board that is not flat, since each sub-composite has a different length to suit the shape the flex part will have after it is bent. Manufacturing may be limited to a few boards per panel due to extra tooling required. This may significantly lower the manufacturing yield, leading to higher unit cost. The initial setup cost is also high, requiring individual forms for each flex board.
A rigid-flex circuit design includes several key elements, and all of them need special consideration during development. While in combination, they may present a significant challenge, successful implementation of a workable elegant solution requires a close collaboration between the designers, material supplier, flex supplier, and the user.