Extraction of interfacial thermal resistance across an organic/semiconductor interface using optical-interference contactless thermometry

We have developed an experimental method to extract interfacial thermal resistance (ITR) at an organic/semiconductor interface based on optical-interference contactless thermometry. The proposed technique was applied to a SU-8/SiC bilayer sample, and clear oscillations in reflectivity induced by optical interference during pulse heating and cooling were observed. After fitting the observed reflectivity waveform with simulation results by a two-dimensional (2D) double-layer heat conduction model and multi-reflection calculations, ITR was extracted as 190 mm2 K W−1, which resulted in a temperature drop of 11 K at the interface. Moreover, the 2D transient temperature distribution of the sample throughout pulse heating and cooling was obtained.


R
ecently, the demand for higher-performance power devices has become urgent due to the rapid growth of electric vehicles.Heat management and design play an important role in maintaining high performance and ensuring safety for power devices. 1)In conventional power device packages, [2][3][4] devices are attached to a heat sink or heat spreader with the interface filled with thermal interfacial material (TIM) [5][6][7] to avoid the formation of a thermal insulation layer as a result of air voids.Among TIMs, organic types have gained widespread use due to their cost-effectiveness, electrical insulation, flexibility and environmental friendliness.7) However, the use of TIM introduces additional thermal resistance in the bonding region, known as interfacial thermal resistance (ITR), [8][9][10][11][12][13][14][15][16][17][18][19][20][21][22][23] which is an important factor in determining the thermal performance of a device package.Since TIM is the first component to contact power devices, impacts of ITR between TIMs and semiconductor wafers on heat dissipation are evident.Therefore, a technique for measuring ITR for TIM/semiconductor interfaces is necessary, and this work focuses on ITR measurement for organicbased TIMs.
Hitherto, various ITR measurement techniques [11][12][13][14][15][16][17][18][19][20][21][22][23] have been proposed, including the steady-state heat-flow method [11][12][13] and transient measurement techniques such as the transient thermoreflectance technique, [14][15][16][17][18] the 3ω method, 19,20) the laser flash method [21][22][23] and the atomic force microscope method.24) In this work, we propose an alternative and useful ITR measurement method using optical-interference contactless thermometry (OICT).[25][26][27][28] This method can extract ITR by directly measuring the transient temperature distribution around the interface based on the principle of multiple reflection and interference and known thermo-optic coefficient (TOC) values.29,30) OICT can clarify the temperature distribution inside the sample including the interface, but so far it can only be used for single substrates. In thi study, the simulation model of OICT was improved to a double-layer structure and ITR was set as a fitting parameter that determines the heat conduction coefficient at the interface. The prposed method made it possible to measure the transient temperature distribution inside a bilayer sample and extract the ITR at the interface.To demonstrate the capabilities of the proposed technique, the ITR of a SU-8/SiC bilayer sample was measured. SU-8 i a typical polymer that is a matrix material for organic-based TIMs and has a mature preparation process, while SiC is a promising material for next-generation power devices.
In the fabrication of the SU-8/SiC bilayer sample, the SU-8 layer was formed on a pre-cleaned and hexamethyldisilazanetreated 4H-SiC wafer (thickness 350 μm, n-type, 4°off, 0.02 Ω cm) by spin-coating.This was followed by baking and UV exposure to solidify the SU-8 film, and the sample was further hard-baked at 200 °C for 30 min to stabilize the properties of SU-8.The thickness of the fabricated SU-8 film was 53 μm as measured by scanning electron microscopy.A 100 nm thick H-patterned Ni electrode was then deposited on SU-8 by sputtering.Figure 1 schematically illustrates the OICT experimental system.CW He-Ne laser light (632.8nm) was irradiated to the center of the sample through a beam splitter from the backside, and the reflected light was measured by a photodiode.Cu strips were mounted at the ends of the H-shaped nickel electrodes, with copper mesh inserted at the connection to reduce contact electrical resistance.The circuit was connected to the copper strip via alligator clips, and a single pulse voltage (100 ms, 6 V) was applied to generate Joule heat in the H-patterned channel during OICT measurements.The voltage (U′) of the 1 Ω resistor was measured to calculate the current (I) in the circuit, while the voltage (U) between two copper strips was also measured to calculate the input power to the sample.A two-dimensional (2D) simulation using the finite difference method 31,32) was used to calculate the transient temperature distribution inside the sample and corresponding reflectivity change at the laser-irradiated area based on heat conduction equation and multiple reflection and interference principle. 33) thermal conductivity of 0.3 W m −1 K −1 was used for SU-8 and a temperature-dependent anisotropic thermal conductivity measured by Zheng 34) et al. was used for 4H-SiC.Energy transfer efficiency (η) and ITR are two fitting parameters of the simulation.Heat (Q) transferred into SU-8 in unit time (Δt) is calculated by Q For simplicity, only the heat transferred from the channel of the H-patterned electrode is accounted for in the simulation, and it is assumed that this heat is uniformly transferred to each of the cells in the first top layer of SU-8.
The influence of ITR is reflected in the continuity condition 32) for temperature and heat flux at the interface, determining the interfacial heat transfer coefficient.η and ITR were optimized until the simulated reflectivity waveform matched the experimental observation.As a result, the transient temperature distribution and ITR of the sample were obtained as the most probable ones.
Figure 2(a) shows the voltage and current waveforms of the Ni electrode when a voltage of 6 V was applied.The gradually decreasing current indicates increasing resistance of Ni due to an increase in temperature.The transient change in reflectivity measured during the pulse heating is shown in Fig. 2(b).Oscillations in reflectivity were observed due to the variation in optical thickness induced by the temperature change in both SU-8 and SiC.The measured reflectivity waveform was then fitted by the simulation, and the influence of ITR and η on reflectivity waveforms is illustrated in Figs.2(c) and 2(d), respectively.Since the TOC of SU-8 is nearly one order magnitude greater than that of SiC, 30,35) temperature change in SU-8 can more easily cause oscillations in reflectivity.As shown in Fig. 2(c), a higher ITR prevents heat flow from SU-8 to SiC, thus leading to an increasing number of oscillations in the reflectivity an ITR of 300 mm 2 K W −1 between polymer and copper wire.The magnitude of ITR is greatly influenced by the difference in thermal conductivity of the materials on either side of the interface and is sensitive to the interface morphology.Given that our sample exhibited an ITR value of the same order of magnitude as those in the literature, we consider our measurement results to be reliable.Additionally, the 2D transient temperature distribution of the SU-8/SiC sample throughout pulse heating and cooling (see Fig. 3) was obtained from the best-fit simulation result shown in Figs.2(e) and 2(f).Heating starts at 0 ms, then cooling starts at 100 ms.Intuitively, the existence of ITR makes it difficult to conduct heat from SU-8 to SiC, which leads to the accumulation of heat in the SU-8 layer and thus a temperature drop at the interface.(Note that this is because the heat flow into SiC is so small that there is no observable temperature gradient in the through-thickness direction of SiC.) In this work, the spatial and temporal resolutions of the proposed technique are applied as ∼50 μm and ∼10 μs, respectively.Spatial resolution is determined by the size of the laser beam incident on the sample, which can be further improved by using a condenser lens with higher magnification power (a convex lens equivalent to 13× was used in this work).Time resolution is the reciprocal of the photodiode sampling rate, and can be up to tens of picoseconds for a Si photodiode. 37)The quality of the fitting result has a significant impact on the accuracy of ITR measurements.The sensitivity of the fitting parameters and the influence of physical properties, including film thickness, thermal conductivity and specific heat, on the fitting procedure will be further clarified in future work.In addition, the double-layer heat conduction model developed in this study assumes that ITR remains constant during both heating and cooling processes.However, many studies [38][39][40][41][42][43] have reported that ITR varies with temperature and surface-parallel stresses at various interfaces. I practical power device packages, the device operating temperature can be up to ∼250 °C, and thus significant temperature gradients can occur in the throughthickness direction, which can lead to considerable thermal stress near the TIM/semiconductor interface.Therefore, a model with variable ITR is needed for more realistic fitting and more accurate measurements.In principle, it can also measure transient ITR changes as well as temperature distributions and thermal stress distributions simultaneously.This will be quite useful for investigating interfacial phonon scattering and transmission and thermal design in power device packaging.In addition, we are dedicated to applying the proposed technique to practical power devices to evaluate the influence of ITR on device self-heating and thermal runaway.
In summary, we have developed a technique to extract ITR at an organic/semiconductor interface based on OICT.By using the developed technique, the 2D transient temperature distribution inside the SU-8/SiC bilayer sample was measured and ITR across the interface was extracted to be 190 mm 2 K W −1 , which is in agreement with the literature.Compared with conventional ITR measurement techniques, the proposed method involves simple equipment and provides direct measurements of transient temperature distribution around the interface.This will be a useful experimental tool for the design of heat dissipation for power device packages and will provide valuable insights into the thermal transport mechanism at the interface.