ETTC'2003 - SEE

Therefore, the sensor response in the linear regime
should be proportional to the acceleration, to the heater
power or heater temperature rise ∆T, to the square of gas
pressure, to the cube of a linear dimension which could
be the cavity volume and inversely proportional to the
square of the gas thermal diffusivity.
3. Microstructure design and fabrication
Heater and thermal detectors are made of platinum
thin film on a low stress (σ ~ 0) silicon rich silicon nitride
membrane SiNx [15]. The thicknesses of the SiN x and
platinum layers are 5000 Å and 3000 Å respectively. To
improve adhesion, adhesion-promoting layers such as Ti
or Cr are used but they tend to reduce the TCR [16].
Different methods of Pt deposition (AC sputtering,
magnetron and electron beam evaporation) and postannealing
conditions were tested to improve the TCR and
to prevent the layer from pealing off during KOH etching
[16, 17]. Best results have been obtained with electron
beam evaporation and vacuum annealing : good adhesion
is obtained even after 5 hours in KOH etching solution,
the electrical resistivity is about 15 µΩ.cm and the TCR
is 3.3×10 -3 /°C. After Pt deposition the resistors are
patterned by ECR (Electron Cyclotron Resonance)
etching. Then, the SiN x is etched by ECR to obtain
resistors on SiN x bridges by KOH etching at 85°C.
Manufacturing stages are summarized on figure 2,
figure 3 shows a SEM (Scanning Electron Microscope)
image of a sensor with 3 pairs of detectors and figure 4
presents its cross section with its different dimensions:
the silicon cavity depth is 400 µm, length is 2000 µm and
width is also 2000 µm. The heater and detectors widths
are 100 and 30 µm respectively and the distance between
the detectors and the heater is 100, 300 or 500 µm.
Finally, it is packaged with a TO16 to obtain a quasiisothermal
hermetic chamber : its height and diameter are
about 5 and 10 mm respectively.
Figure
2: Manufacturing stages of accelerometer : SiNx deposition by
LPCVD (1), Pt deposition (2), ECR etching of Pt (3), ECR etching of
SiNx (4), KOH etching (5).
F. Mailly et Al. ETTC 2003
Figure 3: SEM image of a sensor with 3 detectors pairs.
4.
Experimental results
Figure 4: Sensor cross section.
4.1
Sensitivity according to the distance heater/detector
Figure 5 presents the sensor sensitivity ∆Tdet. according
to the distance heater/detector x for an
acce leration of 1g and an heater temperature rise ∆T =
238°C. We assume that the sensitivity is close to zero if
the distance from the detector to the heater (or to the
substrate) is very low because the thermal resistance of
the gas layer between these elements can then be
neglected and sensor temperature would always equal the
heater’s (or substrate’s) one. The optimum distance x
between the sensors and the heater is about 400 µm, in
good agreement with the simulated optimum distance
which was 300 µm [13] : then, ∆Tdet. is about 3 °C/g for
∆T = 238°C.
∆T det. (°C/g)
4
3
2
1
0
0 200 400 600 800 1000
x (µm)
Figure 5: Sensor sensitivity ∆Tdet. for a heater temperature rise ∆T =
238°C vs. distance x.
2