Page 41 - MEMS Mechanical Sensors

30 Materials and Fabrication Techniques

because it was developed at the RCA Company). Two wafers can be joined together
at room temperature, resulting in an immediate weak bond due to van der Waals
forces. The bond is then strengthened by heat treating in a furnace or by RF or micro-
wave heating at temperatures above 800°C [31]. At this temperature an hermetic seal
is formed between the silicon wafers making it possible to fabricate sealed cavities.
The exact nature of the bonding chemistry is not yet fully understood, but it is
believed that Si-O-Si bonds and water molecules form at the interface as the tempera-
ture is increased. The water molecules subsequently break and diffuse into the silicon.
Silicon fusion bonding is also used to bond silicon to oxidized silicon wafers as in the
manufacture of SOI wafers. The oxide thickness in this process can be as much as 4
µm, and typically temperatures of 1,100°C are used to obtain a permanent chemical
bond. The formation of silicon dioxide to silicon dioxide fusion bonds at 1,100°C
has also recently been reported [32]. Silicon-to-silicon bonding can also be done with
a thin intermediate layer of sputtered, evaporated, or spun on glass. The wafers are
clamped together and heated to the melting point of the glass, typically between
415°C and 1,150°C depending upon the glass. The assembly is then cooled and the
glass solidifies and the process can be used to form a hermetic bond. The relatively
thin layer of glass minimizes the residual stresses in the assembly.

2.3.5.2 Anodic Bonding
Otherwise known as electrostatic bonding, this process is used to bond silicon to
glass. The method uses electrostatic attraction between the glass and the silicon to
facilitate bonding, enabling this to be done at much lower temperatures than would
otherwise be possible [33]. Circular glass substrates with a thermal expansion coeffi-
cient closely matched to that of silicon are readily available. For example, Pyrex,
–6
which has a coefficient of thermal expansion of 3.25 ×10 /°C, is often used for this
purpose. The process is carried out by placing the silicon on a grounded hotplate
with the glass placed in contact with the silicon, as shown in Figure 2.16. The hot-
plate is used to heat the silicon and glass to a temperature of between 350°C and
500°C and a negative potential of about 1 kV is applied to the glass. Extremely
mobile positive ions, mainly sodium, in the glass drift towards the negative electrode
leaving a negative charge on the glass side of the silicon-glass interface. A high
electric field is generated between this fixed negative charge and positive charge in
the silicon, thus pulling the glass and silicon together and facilitating the chemical
bond. Hermetically sealed cavities can be obtained without difficulty. Residual
stresses do occur and bonding temperatures as low as possible are recommended to
keep these to a minimum. Narrow metal tracks, 100 to 200 nm thick, on the silicon
do not compromise the hermiticity of the bond making it possible to run electrical
feedthroughs into sealed cavities. Successful bonding even with a thin layer of
silicon dioxide on the silicon (up to 100 nm thick) can be achieved using this
technique.
Furthermore, the anodic bonding technique can be used to bond silicon to sili-
con by sputtering or evaporating a thin layer of glass onto one of the surfaces. Alter-
natively, glass layers can be spun on to wafers using a spin-on-glass. Wafers with
sputtered glass layers of 0.5- to 4.0-µm thickness have been bonded to silicon [34].
The applied voltages required for this are much less than for bonding to glass sub-
strates. In some cases bonding was achieved with as little as 30V applied. However,