Hi Density QFN Packages and Modules Incorporating Windings for Inductors, Chokes and Antennae

Package, noun, (dictionary.com), a finished product contained in a unit that is suitable for immediate installation and operation. Another definition, a group, combination, or series of related parts or elements to be accepted or rejected as a single unit. As a verb, package is defined as to group or combine (a series of related parts) into a single unit.

Both definitions are used interchangeably in the semiconductor world. One might suggest an available package for a device while in another thought one might discuss “packaging” a device. Regardless of which form of sentence, in the semiconductor world, packaging is the term used to describe how electrical contact with the outside world is achieved while protecting die from environmental influences that could cause electrical, mechanical or electrochemical failure.

A number of options are available for packaging integrated circuits and related devices. Bumped die, leadless packages, leaded packages and modular solutions are all options available for housing die and components. This paper will focus on leadless QFN packages.

Leadless packages are not actually leadless. In most lead-like packages, die are attached to thin patterned metal foils, leadframes, or mini substrates. The die are typically mounted using epoxy (electrically conductive or non-conductive). The die are then electrically contacted to the leadframes/modules using small diameter gold, copper or aluminum wires. The assembly is then overmolded in plastic. The protruding leads are shaped to conform to an industry standard footprint. To reduce the footprint the protruding leads are eliminated. These are the leadless devices - the leads are not completely eliminated of course but they are truncated at the body of the package.

(Modular packages are usually larger because they integrate multi-die, inductors, passives and/or other components into the QFN footprint in order to save board space and improve functionality.)

Figure 2 is an example of a QFN package offered by one of the large semiconductor suppliers. QFN's are fabricated by first forming the leadframe. Leadframes can be fabricated in various processes including etching, laser cutting or stamping. Stamped leadframes are fabricated using a hydraulic press that stamps a lead patter from a sheet of blank metal. These leadframes are limited in minimal feature size and are typically reserved for larger pitch (+800μm) and prototype builds. To fabricate higher density packages requires reduction in leadframe geometry.

Laser cut and etched leadframes can produce tighter geometries but laser cut leadframes can have heat damage on the edges (burrs and slag) while etched leadframes can be over or under etched outside the target geometry. Both are cost effective solutions and provide acceptable leadframes for the 600um to 700μm pitch range used in QFN's. But a high density QFN leadframes require better etch control. Adding traces and winding that are dimensionally smaller than the QFN landing pads require even higher control.

In addition to controlled etch, holding the small features in a common plane is a challenge. That is to say metal traces and windings might short if not “fixtured” in place. These small geometries have to be steadfast during cutting, storage, processing and overmold.

The lead pitch in the 24 pin QFN above shows a 700μm pitch with pad size at 400μm. Higher pin count die require larger package sizes to account for the number of pads on die to pads on QFN. To accommodate higher pin counts the package has to grow in the x and y dimension. As discussed later in the paper, with improved controls and handling the number of leads per package can be doubled without growing the package size. With controlled and improved etch solutions the lead pad and pitch can be reduced by over 50% allowing additional features such as metal traces and windings (coils and antennae).

While coils and antennae can be added directly to the die, the goal of the IC is to pack as much memory and computing power into as small a space as possible. Removing coils and antennae from the die frees up space that is better used for additional transistors. Coils also require special considerations during die design as to not induce signals in adjacent wiring structures. This functionality can be added to a module as stand-alone components (die and passives) but each component added to the board or package increases the footprint.

By integrating the winding in the package the need for these additional features and components is eliminated. Multiple geometries can be added in the same package to capture varying EM fields. Multiple die, passives and routing complete the offering for a module or stand-alone sensor.

There are a few different methods to fabricate inductors. The windings in the TQFN-M utilize a type of inductor that use an air core winding called flat spiral coils. The beauty if inductors is that the coiled geometry of an inductor can be tuned to be an antennae. Flat spiral coils are common in RFID tags and proximity detectors. Other geometries are possible including planar square spiral coils, planar rectangular spiral coils, planar hexagonal spiral coils and octagonal spiral coil. An air core inductor is an inductor that does not depend upon a ferromagnetic material to achieve its specified inductance. Equation (1) below is used to calculate, model and verify the inductor values.

Flat spiral coil inductor calculator [1]...where:

Di

coil inner diameter

s

distance between windings

w

wire diameter

r

radius

N

number of turns

Do

outer diameter

• Inductance (L):

• Outer diameter (Dout):

• Wire length (Wl)):

Air cores have some advantages when compared to wire wound ferrite inductors. One major advantage is the inductance is unaffected by the current it carries. This contrasts with the situation with coils using ferromagnetic cores whose inductance tends to reach a peak at moderate field strengths before dropping towards zero as saturation approaches. Because the iron core is eliminated, the package can perform better at frequencies as high as 1 GHz since most ferromagnetic cores tend to be rather lossy above 100 MHz. Better Q-factors, greater power handling, greater efficiency and less distortion is possible with the iron loss of ferromagnetic cores are eliminated.

Though typically a drawback, flat spiral coils will have higher chances of stray field radiation pickup. But because the air coils are used in targeted applications with a specific inductance (proximity sensor, loop antenna, induction heater, Tesla coil, electromagnet, magnetometer head or deflection yoke etc.) the external radiated field is desirable. (In a cored inductor as the diameter increases towards a wavelength (lambda = c / f), loss due to electromagnetic radiation becomes significant.

The following show conceptualizations and real images of QFN-HD (QFN-high density). Fig 5a shows the detailed traces. Because such fine lines can be maintained, integration of high density pads and inductors, with much higher values than passive 0201 components are possible. Whereas a standard QFN might only have 28 pads, the module below demonstrated 56 pads are achieved in a similar package size while also including a inductor/choke in the package. Figure 6 shows the number of windings that can be integrated into one package to take advantage of varying frequencies in the EM field. By capturing EM energy it possible to build a battery-free solution.

Table 1 list the inspection results of the line traces. (Not listed is the pad size at 90μm; similar to a bump pad.) The pitch is 175μm; similar to that of a bumped die. In one iteration of this device the pads were eliminated to produce a wireless solution. Power was supplied using a Q-charge pads for cell phones. A small board with caps were added to power an ulta-low power MEMS sensor.

Figures 6a and 6b demonstrate fabrication of a multi winding module. Probe pads were added to test the coils using a daisy chained die. Theoretical values were within a 6 percentage points and repeatable. The standard equation used to model the devices will have an error factor added as more components are fabricated to build a baseline.

Table 2 shows the theoretical values obtained and dimensional results. It should be noted that the coils were also bonded out in a manner to capture wireless energy as well and while not tuned for a given frequency, a measurable voltage was obtained when the caps were discharged.

An improved QFN footprint with integrated metal features is manufacturable beyond the current 600–700μm pitch QFN design.

With improved etching pad width can be reduced from 400μm to 175μm allowing for higher density in the same xy geometry. In addition to the reduction in pad size thin traces can be added to form circuitry as well as features such as coils and antennae. This design frees up real estate for additional functionality on the die, IO's in the package, components, inductors or RF antennae. While it is true these features have been added to packages they are typically post processing features that require additional manufacturing steps.

Two objectives were sought. To reduce pad pitch and to integrate features into the package at lead fabrication. The results:

With inductors in the package, inductive coupling can become a close proximity power solution. Integrated coils or antennae can be tuned as charging inductors picking up energy from Wi-Fi, RFID, Bluetooth, NFC, microwave and various other sources of EM fields. Capacitors provide energy storage and can be charged indefinitely for the life of the device. Adding capacitors, microprocessors, sensors, etc. enables the engineer to create solutions never before possible with traditional battery powered and wired devices.

The author would like to thank Mr. Keith Bradshaw for his assistance in testing and validation of the integrated coil package. His +40 years of experience in the electronics industry was critical to our validation.