February_EDFA_Digital

edfas.org 21 ELECTRONIC DEVICE FAILURE ANALYSIS | VOLUME 21 NO. 1 also changes. Thermoreflectance thermal imaging (TTI) depends on an accurate measurement of the relative change in surface reflectivity as a functionof thematerial’s temperature. [3,4] The change in reflectivity is dependent on the thermoreflectance coefficient, a basic material prop- erty that is a function of the illumination wavelength, the material and its surface characteristics, and the ambient temperature. Fortunately, the thermoreflectance coef- ficient, C th , can be considered constant for the range of pertinent ambient temperatures. The basic concept is shown in Fig. 2. The device under test (DUT) is illumi- nated with an LED at a wavelength in the visible range and the reflected signal is captured by the CCD camera. Longer wavelengths in the near infrared range are used for through-the-substrate thermal imaging. A first-order relationship between the normalized change in illumination reflectivity and the change inmate- rial temperature can be approximated as: ∆R(x,y, λ ) = C th ( x,y, λ )∆ T (Eq 2) where C th is the thermoreflectance coefficient and ∆ T is the temperature change of the material. As shown in Fig. 3, C th is quite small—in the order of 10 -6 to 10 -3 —for materials typically encountered with semiconductor devices. Therefore, a lock-in technique is employed to enhance the signal-to-noise ratio (SNR). With time averaging and pixel-by-pixel calibration over the region of interest, <0.1°C temperature resolution can be achieved. With illumination wavelengths in the 400 to 900 nm range, TTI can achieve submicron spatial resolution—as predicted by the Abbe diffraction limit—to meet the imaging requirements of today’s advanced devices. In contrast, traditional infrared imaging based on black-body radiationmeasures emissions in the infrared range with a resolution of 3 to 5 µm. TRANSIENT THERMAL ANALYSIS In addition to gathering information about time-dependent thermal events, transient thermal analysis can provide the information required to fully evaluate the total heat conductionpath in a typical device assembly. [5] The plot in Fig. 5 is derived from the time-dependent thermal analysis after application of bias to the device. A finite time is required for heat to travel from the R(x,y, λ ) Fig. 2 Key components comprising the microscope head with thermoreflectance thermal imaging: An Si CCD camera is used for illumination wavelengths in the visible range and an InGaAs CCD is used for wavelengths in the near-IR range. Fig. 4 Timing diagram shows how time-dependent thermal data is collected. By pulsing the device excitation with a low duty cycle and the LED illumination at a controlled rate and relative time delay, the device temperature is allowed to reach a maximum level and subsequently return to ambient level between device excitation pulses. Fig. 3 Thermoreflectance coefficient vs. wavelength for various materials typically encountered with semiconductor devices.