arXiv subjects

Zhe Cheng

Publications and source records attributed to Zhe Cheng.

At least 19 recordsLinked to original sources

arXiv2026

High Thermal Conductivity of Back-End-of-Line Compatible Diamond Films

Back-end-of-line (BEOL) thermal management requires electrically insulating heat-spreading dielectric that can be integrated within thermal budgets below 400 C. Here, we report polycrystalline diamond films grown directly on Si at a substrate temperature below 400C. Two films with average thickness of 760 and 1000 nm were characterized by Raman spectroscopy, scanning electron microscopy (SEM), and time-domain thermoreflectance (TDTR). Raman spectra show a sharp diamond peak with minor signatures of non-diamond carbon, while SEM reveals lateral growth and large grain size. Temperature dependent TDTR measurements were performed from room temperature to 100C. Sensitivity analysis indicates that the sensitivity of cross-plane thermal conductivity is comparative to the in-plane thermal conductivity. Accordingly, the films were analyzed using an isotropic thermal model by considering the nearly-isotropic grain structure, yielding room temperature effective thermal conductivity of 73 and 86 W m-1 K-1, respectively. These values are about two orders of magnitude higher than those of conventional dielectric materials and demonstrate the potential of diamond films grown at low temperatures as dielectric heat-spreading layers.

cond-mat.mtrl-sci↗

arXiv2026

Active Passivation Tunes Hotspot Locations in GaN Transistors with In Situ Thermal Mechanical Visualization

Efficient thermal dissipation has become critical in emerging electronic devices. However, most existing studies have primarily focused on engineering heat dissipation pathways, largely overlooking the intrinsic behavior of the heat source itself. We demonstrate an active passivation technology that proactively tunes hotspot locations in GaN transistors. By adjusting the active passivation layer length, the hotspot is shifted from the gate edge to the drain-side AP edge, establishing a clear one-to-one spatial correlation. In-situ thermal-mechanical visualization via micro-Raman thermography, combined with multi-physics electro-thermal-mechanical simulations, directly captures the spatial redistribution of both temperature and thermal stress profiles. Electrical analysis confirms that this hotspot migration is driven by the spatial shift of the peak electric field and localized Joule heating. This proactive heat-source tuning strategy provides critical design guidelines for power electronics.

physics.app-ph↗

arXiv2026

Mitigating Hallucinations in Large Vision-Language Models via Causal Route Gating

Large vision-language models (LVLMs) often hallucinate content that is fluent yet unsupported by the image, limiting their reliability in real-world deployment. We show that a key failure mode arises from route competition: even when visual tokens receive attention, the final token decision can be dominated by the textual pathway, causing the decoder to follow linguistic priors over visual evidence. To mitigate this, we propose a training-free, decision-aligned intervention that decomposes each attention head into a visual route and a text route, and estimates their token-level effects using an efficient one-forward/one-gradient approximation. These estimates reveal route conflict within heads and identify prior-dominant ones, enabling selective suppression of only the text route while keeping the visual route intact. Across five benchmarks spanning discriminative and generative settings, our method consistently reduces hallucination-related errors across models with limited impact on overall multimodal performance, while incurring a modest inference-time overhead.

cs.CV↗

arXiv2026

A Full-Stack Performance Evaluation Infrastructure for 3D-DRAM-based LLM Accelerators

Large language models (LLMs) exhibit memory-intensive behavior during decoding, making it a key bottleneck in LLM inference. To accelerate decoding execution, hybrid-bonding-based 3D-DRAM has been adopted in LLM accelerators. While this emerging technology provides strong performance gains over existing hardware, current 3D-DRAM accelerators (3D-Accelerators) rely on closed-source evaluation tools, limiting access to publicly available performance analysis methods. Moreover, existing designs are highly customized for specific scenarios, lacking a general and reusable full-stack modeling for 3D-Accelerators across diverse usecases. To bridge this fundamental gap, we present ATLAS, the first silicon-proven Architectural Three-dimesional-DRAM-based LLM Accelerator Simulation framework. Built on commercially deployed multi-layer 3D-DRAM technology, ATLAS introduces unified abstractions for both 3D-Accelerator system architecture and programming primitives to support arbitrary LLM inference scenarios. Validation against real silicon shows that ATLAS achieves $\le$8.57% simulation error and 97.26-99.96\% correlation with measured performance. Through design space exploration with ATLAS, we demonstrate its ability to guide architecture design and distill key takeaways for both 3D-DRAM memory system and 3D-Accelerator microarchitecture across scenarios. ATLAS will be open-sourced upon publication, enabling further research on 3D-Accelerators.

cs.AR↗

arXiv2026

High Thermal Conductivity in Back-End-of-Line Compatible AlN Thin Films

With thermal issues becoming a major challenge to the development of integrated circuits (ICs), high-thermal-conductivity (high-TC) materials are gaining interest from both the industry and academia, especially for high-density back-end-of-line (BEOL) structures. Aluminum nitride (AlN) is an insulating material with high TC, suitable for thermal management in electronic devices. Furthermore, polycrystalline AlN thin films can be deposited at BEOL-compatible low temperatures (lower than 400 C) while retaining relatively high TC, rendering AlN a promising candidate for BEOL heat dissipation. Still, AlN deposition should aim at achieving high TC on a variety of substrates used in IC processes. In this work, we obtained 600- and 1200-nm BEOL-compatible AlN thin film samples on Si, SiO, SiN, and Al2O3 substrates for structural and thermal characterizations. Specifically, time-domain thermoreflectance (TDTR) measurements revealed consistently high TC (higher than 45 W m-1 K-1) on different substrates. Moreover, finite element analysis (FEA) simulations of AlN capped on top of an indium-tin-oxide (ITO) transistor showed a reduction of up to 44% in peak device temperature. Our work provided experimental and calculational evidence for the practicality of leveraging AlN as a high-TC, BEOL-compatible heat spreader material.

cond-mat.mtrl-sci↗

arXiv2026

Ultralow and Tunable Thermal Conductivity of Parylene C for Thermal Insulation in Advanced Packaging

Parylene C thin films have significant applications in advanced packaging of microelectronics. Their thermal properties are critical for thermal management of electronic devices. However, a unified understanding of the tunable structure and the corresponding thermal conductivity is still missing. This study investigated parylene C thin films of varying thickness and post-annealing temperatures grown via thermal chemical vapor deposition. The ultralow thermal conductivity of as-deposited parylene C measured by time domain thermoreflectance (TDTR) is 0.10 W/m-K. The thermal conductivity can be tuned by post-annealing. Significant increase in thermal conductivity is observed in the annealed samples (0.18 W/m-K) which induces melting and recrystallization. The results of XRD and polarized Raman spectroscopy show that the enhanced thermal conductivity is due to improved crystalline quality and the change in chain orientations. The measured thermal conductivities of the as-deposited and annealed films are much lower than the values predicted by the Cahill minimum thermal conductivity model, which can be explained by the diffuson-mediated minimum thermal conductivity model. Parylene C is found to possess the lowest thermal conductivity among dense low-k materials. Our work provides guidance for the structural design of ultra-low thermal conductivity polymers and corresponding thermal design of electronics.

cond-mat.mtrl-sci↗

arXiv2026

Ion Implantation Enhanced Nucleation Facilitates Heat Transport across Atomically-Sharp Semiconductor Interfaces

Overheating is a critical bottleneck limiting the performance and reliability of next-generation high-power and high-frequency electronics. Interfacial thermal resistance constitutes a significant portion of the total thermal resistance. In this study, we report an ultrahigh thermal boundary conductance (TBC) of approximately 800 MW/m2-K at the atomically-sharp AlN-SiC interface, achieved through an ion implantation-enhanced nucleation epitaxy technique. This value is among the highest TBC values reported for semiconductor interfaces, confirmed by structural characterizations which show an ultrahigh-quality interface. Atomistic Green Function calculations reveal that elastic phonon transmission dominates the interface, with nearly half of the acoustic modes (0-15 THz) exhibiting near-unity transmission due to the atomically sharp structure. Furthermore, using high-energy-resolution electron energy loss spectroscopy, we probe vibrational properties with nanometer spatial resolution and identify unique interfacial phonon modes connecting the mismatched phonon spectra, confirmed by molecular dynamics simulations. The ultrahigh TBC is attributed to both the high elastic phonon transmission due to the high quality interfaces and the inelastic phonon scattering channel due to interfacial phonon modes. These findings not only advance the fundamental understanding of interfacial thermal transport but also provide a pathway for effective thermal management in emerging electronic devices.

cond-mat.mes-hall↗

arXiv2026

Thermal conductance across bonded SiOx-SiOx interfaces in hybrid bonding process

Hybrid bonding is a pivotal technology for enabling three dimensional integrated circuits. Among the foremost challenges facing 3D IC implementation is thermal management, where a deep understanding of heat conduction across bonded interfaces is essential for addressing heat dissipation and reliability issues. Nevertheless, the thermal conductance of bonded dielectric-dielectric interfaces remains poorly understood. In this study, we employ the low-temperature bonding technique integral to hybrid bonding to fabricate SiO-SiO interfaces and investigate their thermal boundary conductance using time domain thermoreflectance. Structural characterizations show high quality bonded interfaces. By fitting the data with an equivalent multilayer thermal model, we establish a lower limit TBC of 150 MW/m2-K for the SiO-SiO interfaces, which corresponds to a thermal resistance lower than that of a 9.2-nm-thick dielectric layer. These findings offer valuable insights into thermal transport in hybrid-bonded structures and provide critical guidance for the thermal design of advanced packaging solutions.

cond-mat.mes-hall↗

arXiv2025

M$^2$OE$^2$-GL: A Family of Probabilistic Load Forecasters That Scales to Massive Customers

Probabilistic load forecasting is widely studied and underpins power system planning, operation, and risk-aware decision making. Deep learning forecasters have shown strong ability to capture complex temporal and contextual patterns, achieving substantial accuracy gains. However, at the scale of thousands or even hundreds of thousands of loads in large distribution feeders, a deployment dilemma emerges: training and maintaining one model per customer is computationally and storage intensive, while using a single global model ignores distributional shifts across customer types, locations, and phases. Prior work typically focuses on single-load forecasters, global models across multiple loads, or adaptive/personalized models for relatively small settings, and rarely addresses the combined challenges of heterogeneity and scalability in large feeders. We propose M2OE2-GL, a global-to-local extension of the M2OE2 probabilistic forecaster. We first pretrain a single global M2OE2 base model across all feeder loads, then apply lightweight fine-tuning to derive a compact family of group-specific forecasters. Evaluated on realistic utility data, M2OE2-GL yields substantial error reductions while remaining scalable to very large numbers of loads.

cs.LG↗

arXiv2025

Experimental Observation of Extremely Strong Defect-Phonon Scatterings in Semiconductor Single Crystals

The role of doping in tailoring thermal transport in semiconductors is critical for efficient thermal management in electronic devices. While the effects of doping have been extensively studied to tune electrical properties, its impact on thermal transport has not yet been thoroughly explored, particularly with respect to experimental investigations into exceptionally strong non-Rayleigh defect-phonon scattering phenomena. Herein, by combining the high-quality growth and advanced characterizations of cubic silicon carbide single crystals with well controlled boron doping, we experimentally observe anomalous strong defect-phonon scatterings, among the strongest reported in common semiconductors, that exceeds the predictions of the classic mass difference model by tens of times in magnitude. The measured thermal conductivity of doped 3C SiC match excellently with those predicted by first principle calculations in which resonant scattering of low frequency phonon is considered. Our findings not only shed light on the fundamental understanding of defect-phonon interactions and will also impact applications such as thermal management of electronics.

cond-mat.mtrl-sci↗

arXiv2024

Thermal Conductivity of Cubic Silicon Carbide Single Crystals Heavily Doped by Nitrogen

High-purity cubic silicon carbide possesses the second-highest thermal conductivity among large-scale crystals, surpassed only by diamond, making it crucial for practical applications of thermal management. Recent theoretical studies predict that heavy doping reduces the thermal conductivity of 3C-SiC via phonon-defect and phonon-electron scattering. However, experimental evidence has been limited. In this work, we report the thermal conductivity of heavily nitrogen doped 3C SiC single crystals, grown using the top-seeded solution growth method, measured via time domain thermoreflectance. Our results show that a significant reduction (up to 30%) in thermal conductivity is observed with nitrogen doping concentrations around 1020 cm-3. A comparison with theoretical calculations indicates less intensive scatterings are observed in the measured thermal conductivity. We speculate that the electron-phonon scattering may have a smaller impact than previously anticipated or the distribution of defects are nonuniform which leads to less intensive scatterings. These findings shed light on understanding the doping effects on thermal transport in semiconductors and support further exploration of 3C SiC for thermal management in electronics.

cond-mat.mtrl-sci↗

arXiv2024

Interfacial reaction boosts thermal conductance of room-temperature integrated semiconductor interfaces stable up to 1100 C

Overheating has emerged as a primary challenge constraining the reliability and performance of next-generation high-performance electronics, such as chiplets and (ultra)wide bandgap electronics. Advanced heterogeneous integration not only constitutes a pivotal technique for fabricating these electronics but also offers potential solutions for thermal management. This study presents the integration of high thermal conductivity semiconductors, specifically, 3C-SiC thin films and diamond substrates, through a room-temperature surface-activated bonding technique. Notably, the thermal conductivity of the 3C-SiC films is among the highest for all semiconductor films which can be integrated near room temperature with similar thicknesses. Furthermore, following annealing, the interfaces between 3C-SiC and diamond demonstrate a remarkable enhancement in thermal boundary conductance (TBC), reaching up to approximately 300%, surpassing all other grown and bonded heterointerfaces. This enhancement is attributed to interfacial reactions, specifically the transformation of amorphous silicon into SiC upon interaction with diamond, which is further corroborated by picosecond ultrasonics measurements. Subsequent to annealing at 1100 C, the achieved TBC (150 MW/m2-K) is record-high among all bonded diamond interfaces. Additionally, the visualization of large-area TBC, facilitated by femtosecond laser-based time-domain thermoreflectance measurements, shows the uniformity of the interfaces which are capable of withstanding temperatures as high as 1100 C. Our research marks a significant advancement in the realm of thermally conductive heterogeneous integration, which is promising for enhanced cooling of next-generation electronics.

cond-mat.mes-hall↗

arXiv2022

Ionic Peltier Effect in Li-Ion Electrolytes

The coupled transport of charge and heat provide fundamental insights into the microscopic thermodynamics and kinetics of materials. We describe a sensitive ac differential resistance bridge that enables measurements of the temperature difference on two sides of a coin cell with a resolution of better than 10 uK. We use this temperature difference metrology to determine the ionic Peltier coefficients of symmetric Li-ion electrochemical cells as a function of Li salt concentration, solvent composition, electrode material, and temperature. The Peltier coefficients {\Pi} are negative, i.e., heat flows in the direction opposite to the drift of Li ions in the applied electric field, large, 30 kJ mol-1, and increase with increasing temperature at T > 300 K. The Peltier coefficient is approximately constant on time scales that span the characteristic time for mass diffusion across the thickness of the electrolyte, suggesting that heat of transport plays a minor role in comparison to the changes in partial molar entropy of Li at the interface between the electrode and electrolyte. Our work demonstrates a new platform for studying the non-equilibrium thermodynamics of electrochemical cells and provides a window into the transport properties of electrochemical materials through measurements of temperature differences and heat currents that complement traditional measurements of voltages and charge currents.

cond-mat.mtrl-sci↗

arXiv2022

Strengthened proximity effect at grain boundaries to enhance inter-grain supercurrent in Ba1-xKxFe2As2 superconductors

Iron-based superconductors have great potential for high-power applications due to their prominent high-field properties. One of the central issues in enhancing the critical current density of iron-based superconducting wires is to reveal the roles and limitations of grain boundaries in supercurrent transport. Here, we finely tuned the electronic properties of grain boundaries by doping Ba1-xKxFe2As2 superconductors in a wide range (0.25<x<0.598). It is found that the intra-grain Jcintra peaks near x~0.287, while the inter-grain Jcinter has a maximum at about x~0.458. Remarkably, the grain boundary transparency parameter defined as Jcinter/Jcintra rises monotonically with doping. Through detailed microscopic analysis, we suggest that the FeAs segregation phase commonly existing at grain boundaries and the adjacent grains constitute superconductor-normal metal-superconductor (SNS) Josephson junctions which play a key role in transporting supercurrent. A sandwich model based on the proximity effect and the SNS junction is proposed to well interpret our data. It is found that overdoping in superconducting grains largely strengthens the proximity effect and consequently enhances the intergrain supercurrent. Our results will shed new insights and inspirations for improving the application parameters of iron-based superconductors by grain boundary engineering.

cond-mat.supr-con↗

arXiv2022

Selective Direct Bonding of High Thermal Conductivity 3C-SiC Film to \b{eta}-Ga2O3 for Top-Side Heat Extraction

beta-Ga2O3 is a wide bandgap semiconductor with electrical properties better than SiC and GaN which makes it promising for applications of next-generation power devices. However, the thermal conductivity of \b{eta}-Ga2O3 is more than one order of magnitude lower than that of SiC and GaN, resulting in serious thermal management problems that limit device performance and reliability. This work reports selectively transferring of high thermal conductivity 3C-SiC thin film grown on Si to beta-Ga2O3 (001) substrate using surface activated bonding (SAB) technique at room temperature, to attempt extracting the heat from the surface of the devices. A 4.5-nm-thick interfacial crystal defect layer is formed at the as-bonded 3C-SiC/beta-Ga2O3 interface. The thickness of the interfacial crystal defect layer decreases with increasing annealing temperature, which decreases to 1.5 nm after annealing at 1000 C. No voids and unbonded area are observed at the interfaces, even after annealing at temperature as high as 1000 C. The thermal boundary conductance (TBC) of the 1000 C-annealed 3C-SiC/beta-Ga2O3 interface and thermal conductivity of the beta-Ga2O3 substrate was measured by time-domain thermoreflectance (TDTR). The 3C-SiC/beta-Ga2O3 TBC value was determined to be 244 MW/m2-K, which is the highest value ever reported for SiC/Ga2O3 interfaces, due to the high-quality heterointerface. Our works demonstrate that selective transferring of 3C-SiC film to the beta-Ga2O3 substrate is an efficient path to improve heat dissipation of the \b{eta}-Ga2O3 power devices.

cond-mat.mtrl-sci↗

arXiv2022

High Thermal Conductivity in Wafer Scale Cubic Silicon Carbide Crystals

High thermal conductivity electronic materials are critical components for high-performance electronic and photonic devices as either active functional materials or thermal management materials. We report an isotropic high thermal conductivity over 500 W m-1K-1 at room temperature in high-quality wafer-scale cubic silicon carbide (3C-SiC) crystals, which is the second highest among large crystals (only surpassed by diamond). Furthermore, the corresponding 3C-SiC thin films are found to have record-high in-plane and cross-plane thermal conductivity, even higher than diamond thin films with equivalent thicknesses. Our results resolve a long-lasting puzzle that the literature values of thermal conductivity for 3C-SiC are perplexingly lower than the structurally more complex 6H-SiC. Further analysis reveals that the observed high thermal conductivity in this work arises from the high purity and high crystal quality of 3C-SiC crystals which excludes the exceptionally strong defect-phonon scatterings in 3C-SiC. Moreover, by integrating 3C-SiC with other semiconductors by epitaxial growth, we show that the measured 3C-SiC-Si TBC is among the highest for semiconductor interfaces. These findings not only provide insights for fundamental phonon transport mechanisms, also suggest that 3C-SiC may constitute an excellent wide-bandgap semiconductor for applications of power electronics as either active components or substrates.

cond-mat.mtrl-sci↗

arXiv2021

Direct Observation of Reversible Heat Absorption in Li-ion Battery Enabled by Ultra-Sensitive Thermometry

The reversible heat in lithium-ion batteries (LIBs) due to entropy change is fundamentally important for understanding the chemical reactions in LIBs and developing proper thermal management strategies. However, the direct measurements of reversible heat are challenging due to the limited temperature resolution of applied thermometry. In this work, by developing an ultra-sensitive thermometry with a differential AC bridge using two thermistors, the noise-equivalent temperature resolution we achieve (10 uK) is several orders of magnitude higher than previous thermometry applied on LIBs. We directly observe reversible heat absorption of a LIR2032 coin cell during charging with negligible irreversible heat generation and a linear relation between heat generations and discharging currents. The cell entropy changes determined from the reversible heat agree excellently with those measured from temperature dependent open circuit voltage. Moreover, it is found that the large reversible entropy change can cancel out the irreversible entropy generation at a charging rate as large as C/3.7 and produce a zero-heat-dissipation LIB during charging. Our work significantly contributes to fundamental understanding of the entropy changes and heat generations of the chemical reactions in LIBs, and reveals that reversible heat absorption can be an effective way to cool LIBs during charging.

physics.app-ph↗

arXiv2021

Experimental Observation of Localized Interfacial Phonon Modes

Interfaces impede heat flow in micro/nanostructured systems. Conventional theories for interfacial thermal transport were derived based on bulk phonon properties of the materials making up the interface without explicitly considering the atomistic interfacial details, which are found critical to correctly describing thermal boundary conductance (TBC). Recent theoretical studies predicted the existence of localized phonon modes at the interface which can play an important role in understanding interfacial thermal transport. However, experimental validation is still lacking. Through a combination of Raman spectroscopy and high-energy resolution electron energy-loss spectroscopy (EELS) in a scanning transmission electron microscope, we report the first experimental observation of localized interfacial phonon modes at ~12 THz at a high-quality epitaxial Si-Ge interface. These modes are further confirmed using molecular dynamics simulations with a high-fidelity neural network interatomic potential, which also yield TBC agreeing well with that measured from time-domain thermoreflectance (TDTR) experiments. Simulations find that the interfacial phonon modes have obvious contribution to the total TBC. Our findings may significantly contribute to the understanding of interfacial thermal transport physics and have impact on engineering TBC at interfaces in applications such as electronics thermal management and thermoelectric energy conversion.

cond-mat.mes-hall↗