In situ Thermoreflectance Characterization of Thermal Resistance in Multilayer Electronics Packaging

Abstract

High-performance, high-reliability microelectronic devices are essential for many applications. Thermal management is required to ensure that the temperature of semiconductor devices remains in a safe operating range. Advanced materials, such as silver-sintered die attach (the bond layer between the semiconductor die and the heat sink) and metal-diamond composite heat sinks, are being developed for this purpose. These are typically multilayered structures, with individual layer thicknesses ranging from tens of micrometers to millimeters. The effective thermal conductivity of individual layers likely differs from their bulk values due to interface effects and potential material imperfections. A method is needed to characterize the thermal resistance of these structures at the design optimization stage to understand what effect non-idealities may have on the final packaged device temperature. We have adapted the frequency-domain thermoreflectance technique to measure at low frequencies, from 10 Hz to 10 kHz, enabling multiple layers to be probed at depths from tens of micrometers to millimeters, which is tailored to assess novel device packaging and heat sinks. This is demonstrated by measuring the thermal resistance of a sintered silver die attach.

Thermal Characterization of Metal-Diamond Composite Heat Spreaders Using Low-Frequency-Domain Thermoreflectance

“…. 23,29 Here, the sample surface is coated with a 10 nm chromium (Cr) adhesion layer and a 150 nm gold transducer layer prior to the FDTR measurement. The high thermoreflectance coefficient of gold (CTR = 2.3 × 10 −4 K −1 ) at the chosen 520 nm probe laser ensures a high measurement sensitivity.…”

“…Measured phase versus frequency results are analyzed by using the n-layer 2D axisymmetric heat diffusion model. 25,29 Clad Metal-Diamond Composite Measurements. Three different clad metal-diamond composite samples were measured by using FDTR.…”

Transducer-Less Thermoreflectance Technique for Measuring Thermal Properties of the Buried Buffer Layer and Interface in GaN-based HEMTs

“…Thermal conductivity and thermal boundary conductance (TBC) of multilayers have been widely characterized with the optical pump–probe thermoreflectance techniques. , The traditional thermoreflectance methods, such as time-domain thermoreflectance (TDTR) and frequency-domain thermoreflectance (FDTR), have shallow thermal penetration depths (∼0.2 to 3 μm) under standard operating conditions. , Therefore, only the part of the sample near the sample surface can be probed, leading to the poor sensitivity of measuring deeply buried layers and interfaces. The recently reported steady-state thermoreflectance (SSTR) , and low-frequency FDTR methods have been demonstrated for the measurement of buried layers and interfaces. For the SSTR, intentionally controlling the pump spot size can extend the thermal penetration depth to tens of micrometers, maximizing the measurement sensitivity for the target properties of the buried layers and interfaces .…”

“…For the SSTR, intentionally controlling the pump spot size can extend the thermal penetration depth to tens of micrometers, maximizing the measurement sensitivity for the target properties of the buried layers and interfaces . For the low-frequency FDTR, measuring from 10 Hz to 10 kHz enables multiple layers to be probed at depths from tens of micrometers to millimeters . Although thermal characterization of deeply buried layers was achieved by the recently developed methods, the pump–probe thermoreflectance methods are often limited for the challenge of simultaneously extracting the multiple unknown parameters when characterizing the multilayer structure .…”

“…Although thermal characterization of deeply buried layers was achieved by the recently developed methods, the pump–probe thermoreflectance methods are often limited for the challenge of simultaneously extracting the multiple unknown parameters when characterizing the multilayer structure . To measure the GaN HEMT structure, investigated in the study, with the traditional thermoreflectance methods where a metal transducer is routinely deposited on the sample surface, ,,− at least five parameters (TBC at metal transducer/GaN interface, thermal conductivity of undoped GaN, thermal conductivity of doped GaN buffer, TBC at GaN/substrate interface, and substrate thermal conductivity) need to be extracted simultaneously in the fitting. The large number of fitting parameters limits the measurement accuracy.…”