|
|
·
Non-invasive temperature measurement of
microscopic features of electronic devices
·
Probing spot diameter smaller than 0.8 mm
·
Less than ± 2°C uncertainty
over a 200°C temperature
range
·
Automated probing of
a sample’s surface along a prescribed trajectory
·
Five (5) MHz tracking capability
of temperature variations
|
The Stationary Thermo-Reflectance (STR) system
at the SMU Submicron Electro-Thermal Sciences Laboratory is designed to
non-invasively measure the surface temperature of active features in
high-density integrated circuits. The STR experimental system shares
many of
the components of the TTR system. The STR setup does not require the
use of the
Nd:YAG heating laser, but adds an intensity stabilizer for the Ar-Ion
probing
laser and uses a differential photo-detection approach in order to
minimize the
low frequency fluctuations in the laser beam power. These enhancements
are
critical for a measurement approach based on thermo-reflectance because
the
thermo-reflectance coefficient, CTR,
is small, typically of order 10-4 K-1. The
stabilization
stage combines the use of a liquid crystal modulator to condition the
irradiation of the linearly polarized laser with the use of feedback
circuitry
on the photodiode (PD) sensor to reduce the long-term intensity
fluctuations.
Measurements over an 8-hour experiment have shown that the stabilizer
attenuates the long-term fluctuations of the Ar-Ion probing laser from
6% down
to 0.3%. The STR performance is further improved by the use of the
differential
photo-detection approach, which consists of two identical silicon PIN
photodiodes monitored by two identical multimeters. A pulse generator
is used
to trigger both sides of the differential setup and to synchronize the
start of
a measurement sequence. While each of the two PD signals would
naturally track
the intensity fluctuations of the laser beam, the ratio of the two
signals, Rt = EPD1 / EPD2,
remains constant as long as the surface reflectivity of the sample does
not
change. Test measurements have shown that the combination of
stabilizing the
laser power and utilizing the differential approach reduces the
experimental
uncertainty down to 0.02%, or 2 K (given a CTR
value of 10-4 K-1)
In order to check the actual performance of the
STR system, measurements have been made on gold-covered samples
provided by
Raytheon and Marlow Industries. The samples consist of a silicon
substrate, a
silicon dioxide thin layer, and a thin gold cover layer. The
temperature of the
sample was varied by the use of the thermochuck from 293 K (20 °C)
through 373
K (100 °C) with 10 K increments. The results of these STR
experiments, shown in
the figure on the right, can also be interpreted as a calibration curve
for
future temperature measurements. The data for the two samples, which
were provided
by two different sources, show good correspondence, with both following
a
linear behavior with a CTR =
2.9× 10-4 K-1. The standard deviation of
the
experimental data from the linear curve is less than 6.45× 10-4
K-1,
corresponding to a temperature change of 2.2 K.
|
