High Thermal Conductivity of Submicrometer Aluminum Nitride Thin Films Sputter-Deposited at Low Temperature.

Aluminum nitride (AlN) is one of the few electrically insulating materials with excellent thermal conductivity, but high-quality films typically require exceedingly hot deposition temperatures (>1000 °C). For thermal management applications in dense or high-power integrated circuits, it is important to deposit heat spreaders at low temperatures (<500 °C), without affecting the underlying electronics. Here, we demonstrate 100 nm to 1.7 µm thick AlN films achieved by low-temperature (<100 °C) sputtering, correlating their thermal properties with their grain size and interfacial quality, which we analyze by X-ray diffraction, transmission X-ray microscopy, as well as Raman and Auger spectroscopy. Controlling the deposition conditions through the partial pressure of reactive N2, we achieve an ∼3× variation in thermal conductivity (∼36-104 W m-1 K-1) of ∼600 nm films, with the upper range representing one of the highest values for such film thicknesses at room temperature, especially at deposition temperatures below 100 °C. Defect densities are also estimated from the thermal conductivity measurements, providing insight into the thermal engineering of AlN that can be optimized for application-specific heat spreading or thermal confinement.

Techno-economic feasibility analysis of an extreme heat flux micro-cooler.

An estimated 70% of the electricity in the United States currently passes through power conversion electronics, and this percentage is projected to increase eventually to up to 100%. At a global scale, wide adoption of highly efficient power electronics technologies is thus anticipated to have a major impact on worldwide energy consumption. As described in this perspective, for power conversion, outstanding thermal management for semiconductor devices is one key to unlocking this potentially massive energy savings. Integrated microscale cooling has been positively identified for such thermal management of future high-heat-flux, i.e., 1 kW/cm2, wide-bandgap (WBG) semiconductor devices. In this work, we connect this advanced cooling approach to the energy impact of using WBG devices and further present a techno-economic analysis to clarify the projected status of performance, manufacturing approaches, fabrication costs, and remaining barriers to the adoption of such cooling technology.

Energy Efficient Neuro-Inspired Phase-Change Memory Based on Ge4 Sb6 Te7 as a Novel Epitaxial Nanocomposite.

Phase-change memory (PCM) is a promising candidate for neuro-inspired, data-intensive artificial intelligence applications, which relies on the physical attributes of PCM materials including gradual change of resistance states and multilevel operation with low resistance drift. However, achieving these attributes simultaneously remains a fundamental challenge for PCM materials such as Ge2 Sb2 Te5 , the most commonly used material. Here bi-directional gradual resistance changes with ≈10× resistance window using low energy pulses are demonstrated in nanoscale PCM devices based on Ge4 Sb6 Te7 , a new phase-change nanocomposite material . These devices show 13 resistance levels with low resistance drift for the first 8 levels, a resistance on/off ratio of ≈1000, and low variability. These attributes are enabled by the unique microstructural and electro-thermal properties of Ge4 Sb6 Te7 , a nanocomposite consisting of epitaxial SbTe nanoclusters within the Ge-Sb-Te matrix, and a higher crystallization but lower melting temperature than Ge2 Sb2 Te5 . These results advance the pathway toward energy-efficient analog computing using PCM.

Unveiling the Effect of Superlattice Interfaces and Intermixing on Phase Change Memory Performance.

Superlattice (SL) phase change materials have shown promise to reduce the switching current and resistance drift of phase change memory (PCM). However, the effects of internal SL interfaces and intermixing on PCM performance remain unexplored, although these are essential to understand and ensure reliable memory operation. Here, using nanometer-thin layers of Ge2Sb2Te5 and Sb2Te3 in SL-PCM, we uncover that both switching current density (Jreset) and resistance drift coefficient (v) decrease as the SL period thickness is reduced (i.e., higher interface density); however, interface intermixing within the SL increases both. The signatures of distinct versus intermixed interfaces also show up in transmission electron microscopy, X-ray diffraction, and thermal conductivity measurements of our SL films. Combining the lessons learned, we simultaneously achieve low Jreset ≈ 3-4 MA/cm2 and ultralow v ≈ 0.002 in mushroom-cell SL-PCM with ∼110 nm bottom contact diameter, thus advancing SL-PCM technology for high-density storage and neuromorphic applications.

A novel hardmask-to-substrate pattern transfer method for creating 3D, multi-level, hierarchical, high aspect-ratio structures for applications in microfluidics and cooling technologies.

This letter solves a major hurdle that mars photolithography-based fabrication of micro-mesoscale structures in silicon. Conventional photolithography is usually performed on smooth, flat wafer surfaces to lay a 2D design and subsequently etch it to create single-level features. It is, however, unable to process non-flat surfaces or already etched wafers and create more than one level in the structure. In this study, we have described a novel cleanroom-based process flow that allows for easy creation of such multi-level, hierarchical 3D structures in a substrate. This is achieved by introducing an ultra-thin sacrificial silicon dioxide hardmask layer on the substrate which is first 3D patterned via multiple rounds of lithography. This 3D pattern is then scaled vertically by a factor of 200-300 and transferred to the substrate underneath via a single shot deep etching step. The proposed method is also easily characterizable-using features of different topographies and dimensions, the etch rates and selectivities were quantified; this characterization information was later used while fabricating specific target structures. Furthermore, this study comprehensively compares the novel pattern transfer technique to already existing methods of creating multi-level structures, like grayscale lithography and chip stacking. The proposed process was found to be cheaper, faster, and easier to standardize compared to other methods-this made the overall process more reliable and repeatable. We hope it will encourage more research into hybrid structures that hold the key to dramatic performance improvements in several micro-mesoscale devices.

Multiobjective Optimization of Graded, Hybrid Micropillar Wicks for Capillary-Fed Evaporation.

As electronic device power densities continue to increase, vapor chambers and heat pipes have emerged as effective thermal management solutions for hotspot mitigation. A crucial aspect of vapor chamber functionality depends on the properties of the microporous wick that drives heat and mass transport within the device. While many prior studies have focused on the optimization of these porous structures to increase the maximum capillary-limited dryout heat flux, an equally important aspect of porous wick design is the minimization of the thermal resistance above heated areas. Segmented wicks with geometries that vary along the length of the wick are attractive candidates that can potentially be used to fulfill these simultaneous design goals. Previous studies on bisegmented wicks with only two distinct adiabatic and heated region geometries, however, have shown mixed results regarding the degree of performance benefit over homogeneous wicks. In this work, we present a systematic modeling approach to investigate the optimal composition of segmented micropillar wicks comprising multiple, discrete regions of graded geometry. Using a genetic algorithm, we generate Pareto fronts of optimal segmented wick distributions that maximize the dryout heat flux and minimize the thermal resistance for a given heating configuration. We find that optimal, graded segmented wicks are capable of dissipating dryout heat fluxes more than 200% higher than baseline homogeneous wicks with significantly lower thermal resistance. The sensitivity of the wick performance to the total number of geometry segments is found to vary depending on the desired heat flux and thermal resistance operating regimes.

Thermal Characterization of Metal-Oxide Interfaces Using Time-Domain Thermoreflectance with Nanograting Transducers.

Metal-oxide thermal boundary conductance (TBC) strongly influences the temperature rise in nanostructured systems, such as dense interconnects, when its value is comparable to the thermal conductance of the amorphous dielectric oxide. However, the thermal characterization of metal-amorphous oxide TBC is often hampered by the measurement insensitivity of techniques such as time-domain thermoreflectance (TDTR). Here, we use metal nanograting structures as opto-thermal transducers in TDTR to measure the TBC of metal-oxide interfaces. Combined with an ultrafast pump-probe laser measurement approach, the nanopatterned structures amplify the contribution of the thermal boundary resistance (TBR), the inverse of TBC, over the thermal resistance of the adjacent material, thereby enhancing measurement sensitivity. For demonstration purposes, we report the TBC between Al and SiO2 films. We then compare the impact of Al grating dimensions on the measured TBC values, sensitivities, and uncertainties. The grating periods L used in this study range from 150 to 300 nm, and the bridge widths w range from 72 to 205 nm. With the narrowest grating transducers (72 nm), the TBC of Al-SiO2 interfaces is measured to be 159-48+61 MW m-2 K-1, with the experimental sensitivity being 5× higher than that of a blanket Al film. This improvement is attributed to the reduced contribution of the SiO2 film thermal resistance to the temperature signal from TDTR response. The nanograting measurement approach described here is promising for the thermal characterization of a variety of nanostructured metal-amorphous passivation systems and interfaces common in semiconductor technology.

Uncovering Thermal and Electrical Properties of Sb2Te3/GeTe Superlattice Films.

Superlattice-like phase change memory (SL-PCM) promises lower switching current than conventional PCM based on Ge2Sb2Te5 (GST); however, a fundamental understanding of SL-PCM requires detailed characterization of the interfaces within such an SL. Here we explore the electrical and thermal transport of SLs with deposited Sb2Te3 and GeTe alternating layers of various thicknesses. We find up to an approximately four-fold reduction of the effective cross-plane thermal conductivity of the SL stack (as-deposited polycrystalline) compared with polycrystalline GST (as-deposited amorphous and later annealed) due to the thermal interface resistances within the SL. Thermal measurements with varying periods of our SLs show a signature of phonon coherence with a transition from wave-like to particle-like phonon transport, further described by our modeling. Electrical resistivity measurements of such SLs reveal strong anisotropy (∼2000×) between the in-plane and cross-plane directions due to the weakly interacting van der Waals-like gaps. This work uncovers electrothermal transport in SLs based on Sb2Te3 and GeTe for the improved design of low-power PCM.

Thermal Interface Enhancement via Inclusion of an Adhesive Layer Using Plasma-Enhanced Atomic Layer Deposition.

Interfaces govern thermal transport in a variety of nanostructured systems such as FinFETs, interconnects, and vias. Thermal boundary resistances, however, critically depend on the choice of materials, nanomanufacturing processes and conditions, and the planarity of interfaces. In this work, we study the interfacial thermal transport between a nonreactive metal (Pt) and a dielectric by engineering two differing bonding characters: (i) the mechanical adhesion/van der Waals bonding offered by the physical vapor deposition (PVD) and (ii) the chemical bonding generated by plasma-enhanced atomic layer deposition (PEALD). We introduce 40-cycle (∼2 nm thick), nearly continuous PEALD Pt films between 98 nm PVD Pt and dielectric materials (8.0 nm TiO2/Si and 11.0 nm Al2O3/Si) treated with either O2 or O2 + H2 plasma to modulate their bonding strengths. By correlating the treatments through thermal transport measurements using time-domain thermoreflectance (TDTR), we find that the thermal boundary resistances are consistently reduced with the same increased treatment complexity that has been demonstrated in the literature to enhance mechanical adhesion. For samples on TiO2 (Al2O3), reductions in thermal resistance are at least 4% (10%) compared to those with no PEALD Pt at all, but could be as large as 34% (42%) given measurement uncertainties that could be improved with thinner nucleation layers. We suspect the O2 plasma generates stronger covalent bonds to the substrate, while the H2 plasma strips the PEALD Pt of contaminants such as carbon that gives rise to a less thermally resistive heat conduction pathway.

Direct Quantification of Heat Generation Due to Inelastic Scattering of Electrons Using a Nanocalorimeter.

Transmission electron microscopy (TEM) is arguably the most important tool for atomic-scale material characterization. A significant portion of the energy of transmitted electrons is transferred to the material under study through inelastic scattering, causing inadvertent damage via ionization, radiolysis, and heating. In particular, heat generation complicates TEM observations as the local temperature can affect material properties. Here, the heat generation due to electron irradiation is quantified using both top-down and bottom-up approaches: direct temperature measurements using nanowatt calorimeters as well as the quantification of energy loss due to inelastic scattering events using electron energy loss spectroscopy. Combining both techniques, a microscopic model is developed for beam-induced heating and to identify the primary electron-to-heat conversion mechanism to be associated with valence electrons. Building on these results, the model provides guidelines to estimate temperature rise for general materials with reasonable accuracy. This study extends the ability to quantify thermal impact on materials down to the atomic scale.

Tunable Dielectric and Thermal Properties of Oxide Dielectrics via Substrate Biasing in Plasma-Enhanced Atomic Layer Deposition.

The ability to control the properties of dielectric thin films on demand is of fundamental interest in nanoscale devices. Here, we modulate plasma characteristics at the surface of a substrate to tune both dielectric constant and thermal conductivity of amorphous thin films grown using plasma-enhanced atomic layer deposition. Specifically, we apply a substrate bias ranging from 0 to ∼117 V and demonstrate the systematic tunability of various material parameters of Al2O3. As a function of the substrate bias, we find a nonmonotonical evolution of intrinsic properties, including density, dielectric constant, and thermal conductivity. A key observation is that the maximum values in dielectric constant and effective thermal conductivity emerge at different substrate biases. The impact of density on both thermal conductivity and dielectric constant is further examined using a differential effective medium theory and the Clausius-Mossotti model, respectively. We find that the peak value in the dielectric constant deviates from the Clausius-Mossotti model, indicating the change of oxygen fraction in our thin films as a function of substrate bias. This finding suggests that the increased local strength of plasma sheath not only enhances material density but also controls the dynamics of microstructural defect formation beyond what is possible with conventional approaches. Based on our experimental observations and modeling, we further build a phenomenological relation between dielectric constant and thermal conductivity. Our results pave invaluable avenues for optimizing dielectric thin films at the atomic scale for a wide range of applications in nanoelectronics and energy devices.

Quasi-Ballistic Thermal Transport Across MoS2 Thin Films.

Layered two-dimensional (2D) materials have highly anisotropic thermal properties between the in-plane and cross-plane directions. Conventionally, it is thought that cross-plane thermal conductivities (κ z) are low, and therefore c-axis phonon mean free paths (MFPs) are small. Here, we measure κ z across MoS2 films of varying thickness (20-240 nm) and uncover evidence of very long c-axis phonon MFPs at room temperature in these layered semiconductors. Experimental data obtained using time-domain thermoreflectance (TDTR) are in good agreement with first-principles density functional theory (DFT). These calculations suggest that ∼50% of the heat is carried by phonons with MFP > 200 nm, exceeding kinetic theory estimates by nearly 2 orders of magnitude. Because of quasi-ballistic effects, the κ z of nanometer-thin films of MoS2 scales with their thickness and the volumetric thermal resistance asymptotes to a nonzero value, ∼10 m2 K GW-1. This contributes as much as 30% to the total thermal resistance of a 20 nm thick film, the rest being limited by thermal interface resistance with the SiO2 substrate and top-side aluminum transducer. These findings are essential for understanding heat flow across nanometer-thin films of MoS2 for optoelectronic and thermoelectric applications.

Tailoring Permeability of Microporous Copper Structures through Template Sintering.

Microporous metals are used extensively for applications that combine convective and conductive transport and benefit from low resistance to both modes of transport. Conventional fabrication methods, such as direct sintering of metallic particles, however, often produce structures with limited fluid transport properties due to the lack of control over pore morphologies such as the pore size and porosity. Here, we demonstrate control and improvement of hydraulic permeability of microporous copper structures fabricated using template-assisted electrodeposition. Template sintering is shown to modify the fluid transport network in a manner that increases permeability by nearly an order of magnitude with a less significant decrease (∼38%) in thermal conductivity. The measured permeabilities range from 4.8 × 10-14 to 1.3 × 10-12 m2 with 5 µm pores, with the peak value being roughly 5 times larger than the published values for sintered copper particles of comparable feature sizes. Analysis indicates that the enhancement of permeability is limited by constrictions, i.e., bottlenecks between connecting pores, whose dimensions are highly sensitive to the sintering conditions. We further show contrasting trends in permeability versus conductivity of the electrodeposited microporous copper and conventional sintered copper particles and suggest these differing trends to be the result of their inverse structural relationship.

A method for quantifying in plane permeability of porous thin films.

The in-plane permeability of porous thin films is an important fluid mechanical property that determines wicking and pressure-driven flow behavior in such materials. This property has so far been challenging to measure directly due to the small sidewall cross-sectional area of thin films available for flow. In this work, we propose and experimentally demonstrate a novel technique for directly measuring in-plane permeability of porous thin films of arbitrary thicknesses, in situ, using a manifold pressed to the top surface of the film. We both measure and simulate the influence of the two dimensional flow field produced in a film by the manifold and extract the permeability from measurements of pressure drop at fixed flow rates. Permeability values measured using the technique for a periodic array of channels are comparable to theoretical predictions. We also determine in-plane permeability of arrays of pillars and electrodeposited porous copper films. This technique is a robust tool to characterize permeability of thin films of arbitrary thicknesses on a variety of substrates. In Supplementary material, we provide a solid model, which is useful in three-dimensional printer reproductions of our device.

Impact of thermally dead volume on phonon conduction along silicon nanoladders.

Thermal conduction in complex periodic nanostructures remains a key area of open questions and research, and a particularly provocative and challenging detail is the impact of nanoscale material volumes that do not lie along the optimal line of sight for conduction. Here, we experimentally study thermal transport in silicon nanoladders, which feature two orthogonal heat conduction paths: unobstructed line-of-sight channels in the axial direction and interconnecting bridges between them. The nanoladders feature an array of rectangular holes in a 10 µm long straight beam with a 970 nm wide and 75 nm thick cross-section. We vary the pitch of these holes from 200 nm to 1100 nm to modulate the contribution of bridges to the net transport of heat in the axial direction. The effective thermal conductivity, corresponding to reduced heat flux, decreases from ∼45 W m-1 K-1 to ∼31 W m-1 K-1 with decreasing pitch. By solving the Boltzmann transport equation using phonon mean free paths taken from ab initio calculations, we model thermal transport in the nanoladders, and experimental results show excellent agreement with the predictions to within ∼11%. A combination of experiments and calculations shows that with decreasing pitch, thermal transport in nanoladders approaches the counterpart in a straight beam equivalent to the line-of-sight channels, indicating that the bridges constitute a thermally dead volume. This study suggests that ballistic effects are dictated by the line-of-sight channels, providing key insights into thermal conduction in nanostructured metamaterials.

Direct Visualization of Thermal Conductivity Suppression Due to Enhanced Phonon Scattering Near Individual Grain Boundaries.

Understanding the impact of lattice imperfections on nanoscale thermal transport is crucial for diverse applications ranging from thermal management to energy conversion. Grain boundaries (GBs) are ubiquitous defects in polycrystalline materials, which scatter phonons and reduce thermal conductivity (κ). Historically, their impact on heat conduction has been studied indirectly through spatially averaged measurements, that provide little information about phonon transport near a single GB. Here, using spatially resolved time-domain thermoreflectance (TDTR) measurements in combination with electron backscatter diffraction (EBSD), we make localized measurements of κ within few µm of individual GBs in boron-doped polycrystalline diamond. We observe strongly suppressed thermal transport near GBs, a reduction in κ from ∼1000 W m-1 K-1 at the center of large grains to ∼400 W m-1 K-1 in the immediate vicinity of GBs. Furthermore, we show that this reduction in κ is measured up to ∼10 µm away from a GB. A theoretical model is proposed that captures the local reduction in phonon mean-free-paths due to strongly diffuse phonon scattering at the disordered grain boundaries. Our results provide a new framework for understanding phonon-defect interactions in nanomaterials, with implications for the use of high-κ polycrystalline materials as heat sinks in electronics thermal management.

Thermal Conduction across Metal-Dielectric Sidewall Interfaces.

The heat flow at the interfaces of complex nanostructures is three-dimensional in part due to the nonplanarity of interfaces. One example common in nanosystems is the situation when a significant fraction of the interfacial area is composed of sidewalls that are perpendicular to the principal plane, for example, in metallization structures for complementary metal-oxide semiconductor transistors. It is often observed that such sidewall interfaces contain significantly higher levels of microstructural disorder, which impedes energy carrier transport and leads to effective increases in interfacial resistance. The impact of these sidewall interfaces needs to be explored in greater depth for practical device engineering, and a related problem is that appropriate characterization techniques are not available. Here, we develop a novel electrothermal method and an intricate microfabricated structure to extract the thermal resistance of a sidewall interface between aluminum and silicon dioxide using suspended nanograting structures. The thermal resistance of the sidewall interface is measured to be ∼16 ± 5 m2 K GW-1, which is twice as large as the equivalent horizontal planar interface comprising the same materials in the experimental sample. The rough sidewall interfaces are observed using transmission electron micrographs, which may be more extensive than at interfaces in the substrate plan in the same nanostructure. A model based on a two-dimensional sinusoidal surface estimates the impact of the roughness on thermal resistance to be ∼2 m2 K GW-1. The large disparity between the model predictions and the experiments is attributed to the incomplete contact at the Al-SiO2 sidewall interfaces, inferred by observation of underetching of the silicon substrate below the sidewall opening. This study suggests that sidewall interfaces must be considered separately from planar interfaces in thermal analysis for nanostructured systems.

Phonon Conduction in Silicon Nanobeam Labyrinths.

Here we study single-crystalline silicon nanobeams having 470 nm width and 80 nm thickness cross section, where we produce tortuous thermal paths (i.e. labyrinths) by introducing slits to control the impact of the unobstructed "line-of-sight" (LOS) between the heat source and heat sink. The labyrinths range from straight nanobeams with a complete LOS along the entire length to nanobeams in which the LOS ranges from partially to entirely blocked by introducing slits, s = 95, 195, 245, 295 and 395 nm. The measured thermal conductivity of the samples decreases monotonically from ~47 W m-1 K-1 for straight beam to ~31 W m-1 K-1 for slit width of 395 nm. A model prediction through a combination of the Boltzmann transport equation and ab initio calculations shows an excellent agreement with the experimental data to within ~8%. The model prediction for the most tortuous path (s = 395 nm) is reduced by ~14% compared to a straight beam of equivalent cross section. This study suggests that LOS is an important metric for characterizing and interpreting phonon propagation in nanostructures.

Quasi-ballistic Electronic Thermal Conduction in Metal Inverse Opals.

Porous metals are used in interfacial transport applications that leverage the combination of electrical and/or thermal conductivity and the large available surface area. As nanomaterials push toward smaller pore sizes to increase the total surface area and reduce diffusion length scales, electron conduction within the metal scaffold becomes suppressed due to increased surface scattering. Here we observe the transition from diffusive to quasi-ballistic thermal conduction using metal inverse opals (IOs), which are metal films that contain a periodic arrangement of interconnected spherical pores. As the material dimensions are reduced from ∼230 nm to ∼23 nm, the thermal conductivity of copper IOs is reduced by more than 57% due to the increase in surface scattering. In contrast, nickel IOs exhibit diffusive-like conduction and have a constant thermal conductivity over this size regime. The quasi-ballistic nature of electron transport at these length scales is modeled considering the inverse opal geometry, surface scattering, and grain boundaries. Understanding the characteristics of electron conduction at the nanoscale is essential to minimizing the total resistance of porous metals for interfacial transport applications, such as the total electrical resistance of battery electrodes and the total thermal resistance of microscale heat exchangers.

Thermal Conduction in Vertically Aligned Copper Nanowire Arrays and Composites.

The ability to efficiently and reliably transfer heat between sources and sinks is often a bottleneck in the thermal management of modern energy conversion technologies ranging from microelectronics to thermoelectric power generation. These interfaces contribute parasitic thermal resistances that reduce device performance and are subjected to thermomechanical stresses that degrade device lifetime. Dense arrays of vertically aligned metal nanowires (NWs) offer the unique combination of thermal conductance from the constituent metal and mechanical compliance from the high aspect ratio geometry to increase interfacial heat transfer and device reliability. In the present work, we synthesize copper NW arrays directly onto substrates via templated electrodeposition and extend this technique through the use of a sacrificial overplating layer to achieve improved uniformity. Furthermore, we infiltrate the array with an organic phase change material and demonstrate the preservation of thermal properties. We use the 3ω method to measure the axial thermal conductivity of freestanding copper NW arrays to be as high as 70 W m(-1) K(-1), which is more than an order of magnitude larger than most commercial interface materials and enhanced-conductivity nanocomposites reported in the literature. These arrays are highly anisotropic, and the lateral thermal conductivity is found to be only 1-2 W m(-1) K(-1). We use these measured properties to elucidate the governing array-scale transport mechanisms, which include the effects of morphology and energy carrier scattering from size effects and grain boundaries.