Page 304 - System on Package_ Miniaturization of the Entire System

278 Cha pte r F i v e

the frequency characteristics of the magneto-dielectric materials. The ferrite composite
is a composite of NiZn ferrite powder and dielectric resin. The magneto-dielectric
substrate has ~5.3 of relative permittivity from 100 MHz to 10 GHz. The dielectric loss
is less than 0.1. The relative permeability of this material is ~5 to 8 in the VHF band with
a reasonably low loss.
Though the concept of using a magneto-dielectric material to miniaturize antennas
is very compelling, the properties of these materials degrade at frequencies beyond 1 GHz,
as shown in Figure 5.17, where the material almost becomes nonmagnetic at high
frequencies. Hence, the synthesis of new materials that have matched properties is
required beyond the 1-GHz range for WiFi applications.
Additionally, novel soft-hard surfaces (SHS) have been researched for the miniaturi–
zation and elimination of the substrate-crosstalk modes improving the efficiency of
patch antennas and suppressing the backside radiation by 10 to 15 dB [32]. Stacked patch
configurations have been explored for the realization of broadband antennas [33]. As in
the examples provided here, a combination of novel design and implementation
techniques needs to be explored to develop fully integrated, high-performance RF SOP
systems.

5.4.4 Inductors
The integration of RF front-end modules has been a key driver for the development of
next-generation wireless communication systems. An important component of these
high-performance RF modules is the inductor, which forms a critical part in the design
of filters, voltage-controlled oscillators, power amplifiers, and low-noise amplifiers.
The figures of merit when dealing with inductor performance are

• Inductance (L)
• Self-resonant frequency (SRF)
• Quality factor (Q)
Efficient RF inductors need a very high Q associated with a considerable high SRF.
Since Q is affected by the package parasitics, a careful choice of the substrate material
such as silicon, LTCC, and organics, used to model and fabricate these inductors, is of
utmost concern.
Silicon and ceramics have been the primary materials for inductors until recently. Si
is the most mature technology because of its use in wafer fabs, and it offers excellent
surface and planarity for a variety of needs including flip-chip bumping and bonding
technologies, has good thermal conductivity, and can be multilayered. However, silicon
has not been a very successful platform for the design and fabrication of high-Q
inductors, primarily due to the losses associated with the semiconducting substrate.
Therefore, certain modifications need to be done. Some of these approaches use high-
resistive silicon oxides such as silicon-on-insulator (SOI), thick dielectric layers, or thick
and multilayer conductor lines. Single- and double-level inductors have been fabricated
on silicon by many investigators [34]. A typical silicon substrate used for the single- and
double-level inductors is shown in Figure 5.18.
In this approach, a 3D process has been implemented in single layers and multilayers
for a spiral inductor. As expected, the inductance observed in the case of a single-layer
inductor was higher when compared to that of a double-layer inductor. For a line width
of 20 μm, spacing of 5 μm, and substrate thickness of 625 μm, a quality factor of 52.8 was