Graduation Semester and Year
Fall 2025
Language
English
Document Type
Thesis
Degree Name
Doctor of Philosophy in Mechanical Engineering
Department
Mechanical and Aerospace Engineering
First Advisor
Dr. Dereje Agonafer
Second Advisor
Dr. Miguel Amaya
Third Advisor
Dr. Aaron Ameri
Fourth Advisor
Dr. Yogesh Fulpagare
Fifth Advisor
Dr. Tushar Chauhan
Abstract
The continued expansion of artificial intelligence, high-performance computing, and cloud-scale services has driven substantial increases in power density and heat generation in electronic systems. As rack-level power rises, conventional air cooling faces practical constraints associated with limited heat removal capability, increased fan power, and elevated facility energy overhead. Single-phase liquid immersion cooling (SPLIC) has therefore attracted significant interest as an alternative thermal management approach because it enables direct convective cooling by submerging electronic assemblies in dielectric liquids. While SPLIC can offer favorable thermal performance at the system level, reliable implementation at scale requires a thorough understanding of long-term performance and durability of power electronics, packaging materials, and thermal interfaces that remain in continuous contact with dielectric fluids. In particular, prolonged exposure to immersion fluids combined with electrical loading and thermal cycling can introduce performance drift and materials changes that must be quantified using reliability-relevant metrics and accelerated methodologies.
This dissertation presents a comprehensive reliability assessment of power modules and thermal interface materials (TIMs) in a single-phase immersion cooling environment through thermal, electrical, and materials-degradation analyses. The work is organized into four integrated investigations: (i) thermal and electrical performance characterization of power modules in immersion environments to establish baseline behavior and comparative trends, (ii) long-term reliability assessment of power modules through high-temperature operating life (HTOL) testing in immersion under dynamic loading, (iii) thermomechanical property characterization of molded and unmolded packaging substrate materials after exposure to immersion fluids and air, coupled with fluid reliability assessment, and (iv) evaluation of immersion-compatible TIM reliability using accelerated stress modes representative of practical operating conditions.
The first investigation establishes baseline thermal and electrical performance of power modules operated in single-phase immersion conditions. Benchmarking and in-situ immersion testing are used to quantify thermal response and electrical performance across representative loading conditions, enabling direct comparison across package constructions and dielectric fluid environments. In addition to in-situ measurements, extended thermal aging is incorporated to examine whether prolonged exposure results in measurable performance shifts, thereby providing a foundation for interpreting long-duration drift observed in subsequent accelerated reliability testing.
The second investigation evaluates long-term reliability of power modules in immersion using a high-temperature operating life methodology adapted to immersion operation. Six evaluation modules (EVMs) representing two package constructions (three molded and three unmolded) are operated for 1000 hours in immersion. The environmental temperature is maintained at 85 °C with a target device junction temperature of 115 °C. A dynamic electrical load is applied to represent repeated operational transients and to enforce thermal excursions: a 40 A load is cycled with a 5-minute ON / 5-minute OFF profile, producing 4800 total load cycles across the test duration. Key parameters are recorded throughout the test, including a junction temperature proxy (TAO), input current (), and output current (). Reliability is quantified through a performance-based failure criterion defined as output current degradation exceeding 20% from baseline. This criterion connects the accelerated test outcome to a functional limit relevant to practical power delivery capability and qualification decision-making.
The HTOL results demonstrate stable operation without catastrophic failures across the full test duration for all six EVMs. Input-side electrical behavior remains stable with no evidence of runaway current or major control instability. Measurable drift is observed in the junction-temperature proxy trends and in delivered output current, indicating gradual degradation under combined electro-thermal stress in immersion. Output current decreases by several amperes relative to the 40 A setpoint over the test duration, corresponding to an approximate 8–12% degradation depending on package construction. Molded devices exhibit slightly higher output-current degradation than unmolded devices under the tested conditions. However, all devices remain below the defined 20% degradation threshold, indicating that functional performance remains within the defined end-of-life criterion through 1000 hours of immersion HTOL operation. These findings provide quantitative evidence of gradual performance drift mechanisms that can be captured through immersion-adapted HTOL, while also indicating robust operation within the applied failure definition.
The third investigation examines materials compatibility and stability by characterizing molded and unmolded packaging substrate coupons after accelerated thermal aging in air and in immersion fluids. Packaging substrates provide structural support and electrical interconnection for power module architectures, and changes in thermomechanical properties can influence stress evolution, warpage, and overall reliability. To evaluate immersion-driven property shifts, substrate coupons are subjected to thermal aging at 80 °C for 336 hours using sealed exposure conditions, with parallel air-aged controls. Multiple dielectric fluids are included to assess fluid-dependent behavior. Thermogravimetric analysis (TGA) is used to quantify thermal stability and mass loss behavior from 25 °C to 260 °C at 5 °C/min. Dynamic mechanical analysis (DMA) is used to evaluate temperature-dependent storage modulus () and glass transition temperature () from –50 °C to 250 °C at 5 °C/min. The substrate materials exhibit negligible mass loss (less than 0.2%) across all conditions, indicating strong thermal stability over the analyzed range. DMA results show strong property retention with minimal shifts in storage modulus trends and limited change in following immersion exposure. Unmolded substrates exhibit across conditions, while molded substrates exhibit –, with variations remaining small and within measurement scatter. These results indicate that the examined substrate materials retain thermomechanical integrity under the accelerated exposure conditions considered.
In addition to substrate property characterization, fluid reliability is assessed by Fourier transform infrared spectroscopy (FTIR) to identify chemical stability indicators following thermal aging in contact with materials. Comparative spectral analysis is used to detect changes consistent with mild residue formation or surface leaching signatures. The evaluated fluids remain largely chemically stable under the aging conditions, with only minor spectral changes observed. This combined materials-and-fluid assessment provides evidence that, under the applied aging conditions, material property drift and fluid chemical change are limited, while also establishing a framework for detecting subtle interactions that may become more pronounced under extended exposure durations, higher temperatures, or additional stress factors.
The fourth investigation evaluates reliability of immersion-compatible thermal interface materials, which are critical in minimizing interfacial thermal resistance between heat sources and heat rejection hardware. In immersion-cooled assemblies, TIMs may experience additional stressors relative to air-cooled conditions, including repeated thermal excursions, potential fluid ingress pathways, interfacial pumping effects, and long-duration contact with dielectric liquids. Two candidate immersion-compatible TIMs (a graphene pad and an indium foil) are assessed using two accelerated stress methodologies: (i) immersion power cycling to impose repeated thermo-mechanical loading representative of operational transients, and (ii) thermal aging with periodic thermal resistance characterization across a matrix of immersion boundary conditions.
For power cycling, a thermal test vehicle is subjected to a 0–400–0 W square-wave power profile with a 20-minute cycle period for 1500 cycles (approximately 500 hours) in a PAO6 immersion environment. The fluid inlet temperature is maintained at 35 °C with a 2 LPM flow rate. Junction temperature and heat sink surface temperature measurements are used to compute TIM thermal resistance (). A performance-based failure criterion is defined as a greater than 20% increase in from baseline. Under these conditions, neither TIM approaches the failure threshold. Instead, both TIMs exhibit a reduction in over cycling, consistent with a conditioning or bedding-in effect that improves conformity and interfacial contact. The graphene pad maintains lower than the indium foil throughout the test window, and both show stable trends with small statistical dispersion.
To complement cycling-driven stress and to assess sensitivity to immersion boundary conditions, a thermal aging methodology is employed in which TIM assemblies are aged at 105 °C for 720 hours. Thermal resistance is periodically characterized every 144 hours, and each characterization is conducted across a matrix of inlet temperatures and flow rates. This approach enables separation of intrinsic TIM/interface changes from apparent changes driven by external convection conditions, providing a more robust framework for immersion TIM evaluation than a single-condition measurement. Together, the cycling and aging methodologies establish performance trends and expectations for immersion-compatible TIM solutions and provide a basis for future qualification protocols.
Overall, this dissertation develops and applies reliability-oriented methodologies to quantify device-level performance drift, package material stability, fluid chemical stability indicators, and interface-level thermal resistance evolution in single-phase immersion environments. The results demonstrate stable operation of power modules through 1000 hours of immersion HTOL testing without catastrophic failures and with functional degradation remaining below a defined 20% output-current criterion. Packaging substrate coupons retain thermomechanical integrity after accelerated exposure, with negligible mass loss and minimal changes in modulus and glass transition behavior, while fluid FTIR analysis indicates largely stable chemistry with only minor spectral changes. Immersion TIM evaluations demonstrate robust thermal resistance behavior under extensive power cycling, with thermal resistance improving rather than degrading over the observed window, and establish a multi-condition thermal aging framework for broader assessment. Collectively, these outcomes provide a holistic reliability perspective and practical metrics that support selection and qualification of dielectric fluids, package constructions, and immersion-compatible TIM solutions for high-density electronics in immersion-cooled environments.
Keywords
Single-phase immersion cooling; dielectric fluids; power modules; high-temperature operating life (HTOL); reliability; packaging materials; thermomechanical properties; DMA; TGA; FTIR; thermal interface materials (TIMs); thermal resistance; graphene pad; indium foil.
Disciplines
Electrical and Electronics | Electro-Mechanical Systems | Electronic Devices and Semiconductor Manufacturing | Energy Systems | Heat Transfer, Combustion | Mechanics of Materials | Other Mechanical Engineering | Polymer and Organic Materials | Power and Energy | Semiconductor and Optical Materials
License
This work is licensed under a Creative Commons Attribution 4.0 International License.
Recommended Citation
Suthar, Rohit K., "Reliability Assessment of Power Modules & Thermal Interface Materials in Single-Phase Immersion Cooling Environment: Thermal, Electrical and Materials Degradation Analysis" (2025). Mechanical and Aerospace Engineering Dissertations. 443.
https://mavmatrix.uta.edu/mechaerospace_dissertations/443
Included in
Electrical and Electronics Commons, Electro-Mechanical Systems Commons, Electronic Devices and Semiconductor Manufacturing Commons, Energy Systems Commons, Heat Transfer, Combustion Commons, Mechanics of Materials Commons, Other Mechanical Engineering Commons, Polymer and Organic Materials Commons, Power and Energy Commons, Semiconductor and Optical Materials Commons