AC Imaging and Thermal Imaging

Thermal Imaging Camera:

James Christofferson and Ali Shakouri, "Thermoreflectance based thermal microscope", Review of Scientific Instruments, 76, 024903-1-6, Jan., 2005

Thermal Imaging Through Substrate:

James Christofferson and Ali Shakouri, "Thermal Measurements of Active Semiconductor Micro-Structures Acquired Through the Substrate Using Near IR Thermoreflectance" Microelectronics Journal - Circuits and Systems, vol.35, No. 10, pp.791-796, October 2004

Temperature measurements on integrated circuits (IC) and semiconductor devices with submicron spatial resolution, and temperature resolution of one tenth of a degree are non-trivial.  Many different techniques can be employed, but generally require a fairly sophisticated experimental setup and data analysis.  After a review of thermal characterization methods, a technique known as the thermoreflectance method is chosen to be the basis for a testing platform that can be used for thermal characterization of micro-scale semiconductor samples. The temperature profile is determined by measuring the temperature-induced change in the surface reflectivity. Lock-in technique is needed to detect the small change in the reflectivity (DR/DT~2x10^-5 /^oC). One can generate the thermal image of the device by illuminating the whole surface with a white light source and using a photodiode array for lock-in detection of the thermoreflectance signal at different pixels. Visible wavelength thermoreflectance imaging can give submicron resolution.

To try this method a sample was illuminated with a white light source and excited with current pulses at 200Hz, to allow for heterodyne filtering.  The detector was the Hammamatsu 16x16 photodiode array, and several National Instruments data acquisition boards were used for parallel processing of the 256 channels. Dr. Philip Melese at SRI International provided the detector and associated circuitry. Calibration was done using the thermocouple measurements on larger devices.  Presented are the images of a 30-micron microrefrigerator at 25C.  First figure (a) shows a CCD image of the cooler.  Surface reflection and interpolated thermal reflectance image are shown in (b) and (c).  Each data point represents a 1Hz bandwidth FFT at the cycling frequency averaged over 20 seconds. We see that there is cooling, but also heating at the junction between the contact layer and the cooler. This thermoreflectance microscope can create real time thermal images with a higher spatial resolution then any commercial available infrared microscope.  Another advantage of the thermoreflectance microscope is that it can be used at very low ambient temperatures where a blackbody infrared thermal camera cannot be used.  For example, the thermoreflectance measurement has been used on an active semiconductor samples at 10K.   Also, we can measure the transient thermal response of devices because the thermoreflectance probe has high bandwidth. 

In the second figure we see a higher resolution scanned thermal image of a 40x40 HIT micro-cooler, indicating a good cooling distribution.  Figure below shows the modulator section of an integrated electroabsorption modulator with a DBR laser.  When the modulator is active, the laser power is absorbed causing significant heating, as shown in the thermal image.

(a) CCD image of 35 micron diameter micro refrigerator. (b) Reflection image measured by 16x16 pin array detector. (c) Thermo reflectance image when the device is cycled at 200Hz with a current of 100mA
(James Christofferson, Daryoosh Vashaee, Ali Shakouri, Philip Melese, Xiaofeng Fan, Gehong Zeng, Chris Labounty, and John E. Bowers, Edward T. Croke III, ”Thermoreflectance Imaging of Superlattice Micro Refrigerator”, SEMITHERM XVII symposium proceedings, San Jose, Ca, March 2001)

Ali Shakouri, James Christofferson, Zhixi Bian, and Peter Kozodoy, “High Spatial Resolution Thermal Imaging of Multiple Section Semiconductor Lasers,” (invited talk) Proceeding of Photonic Devices and System Packaging Symposium (PhoPack 2002), pp22-25, IEEE CPMT and LEOS, July 2002, Stanford CA.

Diffraction-Limited Imaging

The Rayleigh criterion is generally regarded as a fundamental limit and due to its practical accuracy in predicting the performance of optical imaging systems, it has unfortunately become accepted as a de-facto physical law. In a joint research with Prof. Milanfar at UCSC, we have shown that this limit is simply a very good rule of thumb, which under proper conditions typically related to the signal-to-noise (SNR) of the sensor, can be overcome.

When two point sources are separated by a distance, d, their image by an incoherent optical system is made of two sinc² functions (assume one dimensional case). According to the Rayleigh criterion of resolution, the two sources are “barely resolved” when the center of one sinc² function falls exactly on the first zero of the second sinc² function (d=1 in the units below, see figure inset). The question of whether one or two peaks are present in the measured signal and what is their separation can be formulated in statistical terms. Calculations described in the reprint book show the smallest peak separation, d, which can be detected with very high probability (say 0.99), and very low false alarm rate (say 10^-6) as a function of SNR (see figure below):

In deciding the minimum resolvable separation between two sources from sampled image data, two main factors enter into play; first, and foremost, is the SNR per sample of the imaging array. A second parameter of importance is the sampling rate, the increase of which can also improve performance. (Milanfar, P., and A. Shakouri, "A Statistical Analysis of Diffraction-limited Imaging", Proceedings of the International Conference on Image Processing, pp. 864-867, September 2002)