PROGRESS REPORT Polymer Foams www.advmat.de Hollow-Structured Materials for Thermal Insulation Feng Hu, Siyu Wu, and Yugang Sun* conversion, conservation, and storage. The performance of a thermal insulation is characterized by the thermal conductivity of a material that is determined by its physical structure and chemical composition. Materials with intrinsic low thermal conductivities such as fiberglass, asbestos, and rock wool are widely used to achieve high thermal insulation performance.[2] However, their performance are still below the requirements in residential applications such as energy-efficient buildings and in aeronautic and astronautic applications under extreme conditions. Tackling this challenge requires an urgent development of new insulation (composite) materials, in particular, with unique engineered structures. For instance, silica aerogels with high porosities and low mass densities have been recognized as a class of excellent materials with superior thermal insulation even though solid silica has a high thermal conductivity (1.2–1.4 W m−1 K−1). The low thermal conductivities of aerogels mainly originate from the restricted heat transfer across the gas phase confined in the voids, which are formed from aggregated assembly of silica nanoparticles. The dependence of high thermal insulation performance of aerogels on the existence of voids (or hollow structures) has stimulated great efforts in developing new thermal-insulation materials by generating hollow structures (e.g., voids and bubbles) in bulk materials and composites. Although some review articles have summarized the progress in fabricating various thermal-insulation materials including aerogels, polymer foams, and composite films,[3] a review with a perspective on the general structure– property relationship between the hollow structures and reduction of thermal conductivity in all insulation materials is still absent. Herein, we present a comprehensive and timely overview on the present thermal-insulation materials containing hollow structures with an emphasis on the structure–property relationship. A hollow structure is typically defined as a solid structure with a void space inside a distinct shell.[4] The engineering bulk materials, such as the thermal-insulation materials, are considered as assemblies of many fused or aggregated hollow structures. They appear as bulk materials containing large amount (or high density) of individual voids, denoted as “hollow-structured materials” or HSMs in this article. The voids in HSMs could be produced from inheriting the hollow interiors of shells, bubbling of dissolved gases in these insulation materials, and assembling solid materials into 3D networks Heating and cooling represent a significant portion of overall energy consumption of our society. Due to the diffusive nature of thermal energy, thermal insulation is critical for energy management to reduce energy waste and improve energy efficiency. Thermal insulation relies on the reduction of thermal conductivity of appropriate materials that are engineerable in compositions and structures. Hollow-structured materials (HSMs) show a great promise in thermal insulation since the existence of high-density gaseous voids breaks the continuity of heat-transport pathways in the HSMs to lower their thermal conductivities efficiently. Herein, a timely overview of the recent progress in developing HSMs for thermal insulation is presented, with the focus on summarizing the strategies for creating gaseous voids in solid materials and thus synthesizing various HSMs. Systematic analysis of the documented results reveals the relationship of thermal conductivities of the HSMs and the size and density of voids, i.e., reducing the void size below ≈350 nm is more favorable to decrease the thermal conductivity of the HSMs because of the possible confinement effect originated from the nanometersized voids. The challenges and promises of the HSMs faced in future research are also discussed. 1. Introduction Energy consumption and environmental pollution have become global concerns due to the rapid depletion of fossil fuel and the surge of greenhouse gas emission. An agreement has been reached to alleviate these challenges by exploring renewable energy resources as well as improving the energy efficiency of traditional supplies/technologies. Currently, renewable energy resources, e.g., solar, wind, biomass energy, etc., support about 14% of the total world energy demand with a remarkable potential of continuous growth.[1] The related technologies such as photovoltaics still suffer from the drawbacks of low-efficiency energy conversion and high-cost storage. On the other hand, research on energy management and minimization of energy consumption becomes more vital than ever in recent years. Thermal insulation, which reduces heat flow with thermal resistant materials, plays a critical role on the improvement of energy efficiency involved in essentially every process of energy Dr. F. Hu, S. Wu, Prof. Y. Sun Department of Chemistry Temple University 1901 North 13th Street, Philadelphia, PA 19122, USA E-mail: ygsun@temple.edu The ORCID identification number(s) for the author(s) of this article can be found under https://doi.org/10.1002/adma.201801001. DOI: 10.1002/adma.201801001 Adv. Mater. 2019, 31, 1801001 1801001 (1 of 17) © 2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.advancedsciencenews.com www.advmat.de with interconnected openings. These three strategies can be described in details: i) hollow shells are synthesized with the assistance of sacrificial templates followed by assembly of them into HSMs, ii) turning the gases dissolved in the bulk materials into voids through a gas foaming process forms better insulating materials called cellular foams, and iii) assembling solid nanoparticles into or directly converting appropriate precursors to solid networks in the presence of solvent results in HSMs after the solvent is removed to leave voids. A number of HSMs include aerogels, polymer foams, and hollow spherebased composite films have been widely explored to serve as the thermal-insulation materials with enhanced performance compared to their solid counterparts without voids. The existence of hollow voids in HSMs of various compositions significantly influences their thermal insulation property. An HSM is summarized with the scheme of Figure 1 regardless of its type. An HSM comprises phase I of a continuous matrix, phase II of void-confined gas, and the interfacial phase III that bridges I and II. The compositions of individual phases can be varied independently or simultaneously to form different HSMs. For example, in a CO2-foamed cellular poly(methyl methacrylate) (PMMA) film phases I and II are PMMA matrix and CO2 gas, respectively, while interfacial phase III is considered as only a solid/gas (PMMA/CO2) interface without a physical thickness. In a silica aerogel the interfacial phase III is the solid silica network while both phases I and II are air. The total volume of the voids (phase II), which is determined by the average size and number of the voids, in an HSM with a unit of volume is used to describe the porosity of the HSM. These chemical and physical parameters are closely related to the spatial distributions (or assembly) of individual phases (i.e., I, II, and III), which influence the thermal conductivity of the corresponding HSMs. Therefore, we summarize in this report the HSMs with high thermal insulation performance, discuss the correlation of thermal conductivity and the hollow structures, and wrap up the report with the personal perspectives regarding the challenges and opportunities of developing HSMs for high-performance thermal-insulation materials. 2. Heat Transfer in HSMs Thermal transport in a thermal-insulation material follows three major modes of heat transfer, i.e., convection, radiation, and conduction. Convection is the macroscopic movement of the fluid (e.g., liquid and gas), which mixes the cold part and hot part of the fluid to enable heat transfer in the fluid. In HSMs convection becomes possible in the gases confined in voids only when the Grashof number is greater than 1000. The Grashof number (Gr) describes the ratio of the buoyant force driving convection to the viscous force opposing it with an expression[5] Gr = g ⋅ α ⋅ ΔT ⋅ d 3 ⋅ ρ g (1) η3 where g is the gravitational acceleration constant (9.81 m s−2), α is the volume expansion coefficient of the gas, ΔT is the temperature difference across individual voids, d is the diameter of Adv. Mater. 2019, 31, 1801001 Feng Hu received his Ph.D. degree in chemistry from the University of Chinese Academy of Sciences (UCAS) in 2017. He is currently a postdoctoral fellow in Dr. Sun’s group in Temple University. His main research interests focus on the developments of novel composite nanomaterials for efficient energy management, with an emphasis on exploring high optically transparent and insulating composite films. Siyu Wu received his B.S. degree from the University of Science and Technology of China (USTC) in 2017. He is currently a Ph.D. student in the Department of Chemistry at Temple University. His current research activities include the synthesis and in situ high-energy synchrotron X-ray characterization of efficient catalysts. Yugang Sun received his B.S. and Ph.D. degrees in chemistry from the University of Science and Technology of China (USTC) in 1996 and 2001, respectively. He is currently a professor at the Department of Chemistry, Temple University. His current research interests focus on design of quantumsized nanoparticles for energy applications and understanding of nanoparticle growth kinetics using in situ synchrotron X-ray techniques. the voids (assuming spherical shape), and ρg and η represent the density and dynamic viscosity of the gas, respectively. Convection of air with a pressure of 1 atmosphere (at Gr ≥ 1000) requires the void diameter to be larger than 10 mm. Therefore, gas convection in typical HSMs with void sizes of several micrometers or less is suppressed completely, making essentially no contribution to thermal conductivity. Thermal radiation in an HSM film is the heat transfer refers to the transmission of electromagnetic waves emitted by objects and surfaces with temperatures higher than 0 K. Heat transfer in this way highly depends on the optical responses to electromagnetic waves mainly in the infrared region that corresponds 1801001 (2 of 17) © 2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.advancedsciencenews.com www.advmat.de in both solid and gas, which strongly depends on the hollow structures of an HSM, makes major contribution to the overall thermal conduction of the HSM, representing the focus of this review. 2.1. Thermal Conduction of Gases in HSMs The contribution of the gas in an HSM to thermal conductivity originates from the collisions of gas molecules in the voids (phase II in Figure 1) and the collisions between the gas molecules and the solid walls of the interfacial phase III. The collisions can be described with the kinetic theory model, giving the thermal conductivity of a bulk gas by[8] 1 kg,0 = ⋅ C g,0 ⋅ ν g,0 ⋅ Λ g,0 ⋅ ε (3) 3 Figure 1. Illustrative diagram showing the morphology of a typical HSM that is constructed with three immiscible phases. Phase I represents the continuous component responsible for the heat transport pathways. Phase II is the gas confined in voids. The interfacial phase III bridges phase I and phase II, and it can be composed of a solid component but this is not necessary. to temperatures of hundreds of K. Typically, an HSM for thermal insulation is optically thick compared to the mean free path of infrared photons,[6] making the radiative heat transport a local phenomenon mainly influenced by the scattering and absorption in the HSM. In this case, the radiative conductivity can be described by[7] kr = 16 ni2 ⋅ σ ⋅ T 3 (2) 3 e (T ) ⋅ ρ where σ is the Stefan–Boltzmann constant, ni is the mean index of refraction of the specimen, T is the temperature, e(T) is the specific Rossland mean extinction coefficient with a unit of m2 kg−1, and ρ is the apparent density of the HSM. The thermal radiation should not be overlooked since the e(T) of HSM varies significantly from components to components. For example, silica aerogels show small absorption in mid-infrared region (3–5 µm) but organic aerogels made of polymers such as resorcinol–formaldehydes (RF) are nearly opaque to this spectral band. Therefore, one can effectively decrease the thermal radiation by doping the solid component of an HSM with strong infrared opacifiers such as carbon black.[6] This strategy depends more likely on the composition rather than the geometrical structure of the HSM, which is out of the scope of this review focusing on overviewing the construction of hollow structured materials and the structure–property relationship. Thermal conduction relies on the motion of microscopic energy carriers including molecules, atoms, as well as free electrons and phonons. Thermal conduction in a solid mainly depends on the lattice vibration of solid molecules around their equilibrium positions, while thermal conduction in a gas originates from the collision of gas molecules. Thermal conduction Adv. Mater. 2019, 31, 1801001 where Cg,0, vg,0, and Λg,0 represent the volumetric specific heat of the gas, the root-mean-squared velocity of gas molecules, and the mean free path of gas molecules, respectively. ε is the correction factor describing the influence of viscosity on the thermal conductivity determined with the kinetic theory model. According to the kinetic theory, the mean free path of bulk gas molecules is Λ g,0 = kB ⋅ T (4) 2 ⋅ π ⋅P ⋅ l2 where l is the collision diameter of the gas molecules and P, T, and kB represent the gas pressure, the temperature, and the Boltzmann constant, respectively. In an ideal bulk gas, thermal conductivity is independent of gas pressure since the volumetric specific heat, Cv,0, is proportional to the pressure (corresponding to the number density of gas molecules) and the mean free path, Λg,0, is inversely proportional to the pressure. In contrast, the movement of gas molecules is limited by the size of voids in an HSM, reducing their mean free path due to the collisions of gas molecules with the solid matrix when the average void size is smaller than Λg,0.[8a,9] The reduced mean free path represents a typical spatial confinement effect, which requires the use of an apparent effective mean free path, Λg,eff, to describe the corresponding lowered thermal conductivity, i.e., effective thermal conductivity of gas in a confined space 1 kg,eff = ⋅ C g,0 ⋅ v g,0 ⋅ Λ g,eff ⋅ ε (5) 3 The value of Λg,e is determined by following the Mattiesson rule[8a] 1 1 1 = + (6) Λ g,eff Λ g,0 d where d represents the characteristic size of the void to highlight the spatial confinement effect. Combining Equations (1)–(6) derives the effective thermal conductivity confined in voids with a characteristic dimension of d as 1801001 (3 of 17) © 2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.advancedsciencenews.com kg,eff = C g,0 ⋅ v g,0 ⋅ ε 2 ⋅π ⋅ P ⋅ l 2 1 + k ⋅ T d 3 B www.advmat.de −1 or kg,eff = kg,0 (7) 1 + Λ g,0 /d where Λg,0/d represents a useful parameter to evaluate the confinement effect on thermal conductivity of the gas in the voids with sizes smaller than Λg,0. The ratio of Λg,0/d is reported as the Knudsen number (Kn). According to Equation (7), the effective thermal conductivity decreases as the value of Kn increases. Therefore, increasing the mean free path of the corresponding bulk gas molecules and/ or reducing the characteristic dimension of the voids in an HSM, which can effectively increase the Knudsen number, represents a feasible strategy to lower the thermal conductivity of gas confined in small voids. Many other models based on the Knudsen number have also been proposed to empirically describe the influence of spatial confinement on the gas thermal conductivity in small voids of HSMs.[3d,10] The “corrected” effective thermal conductivity of gases confined in small voids allows us to more accurately describe the contribution of gases in an HSM to the overall thermal conductivity of the HSM. For an HSM with a defined shape and volume, the voids occupy only a fraction of volume of the HSM, which is usually described as porosity, Π. In general, the small voids uniformly distribute throughout the HSM, resulting in highfrequency collision between the gas molecules and the solid walls because the characteristic dimension of the voids are smaller than Λg,0. The collisions lead to an energy transfer from the gas molecules of the void phase II to the solid phase I (or interfacial phase III), which is described with a parameter, β. Considering these two factors, i.e., Π and β, the contribution of gas in the void phase II to the total thermal conductivity of the HSM (or gas thermal conductivity) can be predicted by the Kaganer’s model[7a,10c,11] kg,HSM = kg,0 ⋅ Π (8) 1 + 2 ⋅ β ⋅ Kn Increasing the mean free path of gas molecules can be realized by reducing the gas pressure in the voids of an HSM. For example, Kistler first reported the reduced thermal conductivity of silica aerogels under lower pressure.[12] The low thermal conduction of 0.006 W m−1 K−1 was achieved in the resorcinol– formaldehyde aerogel monoliths with a density of 80 kg m−3 when the monoliths were evacuated to 10 mbar.[11a] As a consequence, the thermal conductivity of the gas in an aerogel can be obtained by subtracting the thermal conductivity of the aerogel measured at vacuum from that measured at the ambient pressure of a specific gas. According to Equation (8), the Knudsen number can be determined to estimate the average size and size distribution of the voids in the aerogel with the consideration of the mean free path of the specific gas at the ambient pressure.[7a,10d,13] Moreover, filling the voids of an HSM with a gas that exhibits larger mean free path can also reduce the gas thermal conductivity. Kistler exchanged the air in silica aerogels with CO2 and CCl2F2 to lower the thermal conductivity of the aerogels.[12] Reducing the size of voids in an HSM is more promising since we can avoid the use of expensive long-mean-freepath gases and the vacuum process that may cause materials Adv. Mater. 2019, 31, 1801001 Figure 2. Effect of void size on thermal conductivity of gases in open cell polymeric foams and resorcinol–formaldehyde aerogels (symbols). The curve represents the calculation result according to the Knudsen equation, showing the good agreement with the experimental measurements. PMMA: poly(methyl methacrylate); MAM: block copolymer poly(methyl methacrylate)-co-poly(butyl acrylate)co-poly(methyl methacrylate); PE: polyethylene. Reproduced with permission.[15] Copyright 2014, Elsevier Ltd. failure.[14] For instance, the thermal conductivity of polyetherimide (PEI) foams is reduced to as low as 0.015 W m−1 K−1, a value even lower than that of air, when the size of the voids is reduced to ≈100 nm.[14b] The small voids largely restrict the motion and collision of the gas molecules, accounting for the reduced thermal conductivity. The experimental observations are consistent with theoretical studies using the finite and molecular dynamics models for a wide range of materials including silica aerogels, polyolefin foams, PMMA foams, and PEI foams. The reduction of thermal conduction of gases confined in small spaces is recognized as the Knudsen effect described by Equation (8). The Knudsen effect was, for the first time, demonstrated by Notario et al. in PMMA foams.[15] Figure 2 compares the thermal conductivities of gases in various aerogels, PE foams, and PMMA-based foams, highlighting their dependence on the size of voids and independence of the composition of solid phases. 2.2. Thermal Conduction of Solids in HSMs Heat transfer in solids mainly relies on the molecular vibrations in the materials. Different from the gas molecules that can move freely, molecules in solid materials can only vibrate around their equilibrium positions upon thermal excitation. In quantum mechanics, the energies of molecular vibrations are quantized and discrete. The vibrations with the minimum energy correspond to phonons. A phonon is a quantum mechanical description of an elementary vibrational motion in which a lattice of atoms or molecules uniformly oscillates at a single frequency. The property of phonon transport in a solid determines the thermal conductivity of the material. The thermal conductivity of a bulk solid can also be described with the kinetic model (Equation (9))[16] 1801001 (4 of 17) © 2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.advancedsciencenews.com www.advmat.de 2Λ ks,HSM = (1 − Π) ⋅ ks,eff = (1 − Π) ⋅ ks,0 ⋅ 1 + s,0 d +δ Figure 3. Schematic illustration showing the influence of void size on heat transport in a polymer foam. Phase I and phase II correspond to the polymer matrix and void-confined gas, respectively. The characteristic size (δ) of the polymer and the size (d) of the voids are defined in the left frame. The heat conduction pathway in the solid phase I is illustrated with the multiple arrows in the middle frame, highlighting the dependence on the value of δ. 1 ks,0 = ⋅ Cs,0 ⋅ v ph,0 ⋅ Λ s,0 (9) 3 where Cs,0, vph,0, and Λs,0 represent the volumetric specific heat of the solid, the root-mean-squared velocity of phonons in the solid, and the mean free path of phonons in the solid, respectively. Phonon scattering takes account in the thermal conductivity when the mean free path of the phonons is comparable to the characteristic dimension of the solid.[17] For example, the characteristic dimension of the solid in an HSM is determined by both the void size, d, and the dimensional parameter, δ, which is the smallest thickness of the walls of phase I between adjacent voids shown in Figure 3. Similar to the influence of spatial confinement of gas molecules on thermal conductivity, the size effect on thermal conductivity of the solid in the HSM can also be described by using an effective mean free path of phonons, Λs,eff [17a] 1 1 2 = + (10) Λ s,eff Λ s,0 d + δ Accordingly, the thermal conductivity of the porous solid is[17,18] 1 1 2 ks,eff = ⋅ Cs,0 ⋅ v ph,0 ⋅ + 3 Λ s,0 d + δ −1 −1 2Λ s,0 or ks,eff = ks,0 ⋅ 1 + d + δ (11) Equation (11) clearly shows that the thermal transport mechanism of a porous solid is consistent with the corresponding bulk solid when the dimensions of the voids are much larger than the intrinsic mean free path of phonons in the solid. Since the solid represents only a fraction of volume of the HSM, the contribution of the solid material to the total thermal conductivity of the HSM (or solid thermal conduction) is expressed by Adv. Mater. 2019, 31, 1801001 −1 (12) Therefore, simultaneously decreasing the size of the voids and reducing the value of δ in an HSM can lower the contribution of the solid to the total thermal conductivity of the HSM when the porosity of the HSM is constant.[14b,19] For instance, when a foamed polymer contains spherical voids packed in face-centered geometry with a porosity (Π) of 80%, the characteristic size (δ) exhibits a relationship of δ = 0.13d with the 3 π d , diameter (d) of the spherical voids according to Π = 6 d +δ highlighting the promise in reducing δ through the generation of small voids in an HSM. Moreover, the collision frequency of gas molecules in the voids of an HSM decreases with the decrease of the size of the voids because fewer gas molecules are confined in individual voids. For example, collisions of gas molecules are largely excluded in the voids with sizes on the order of the mean free path of the gas molecules, e.g., ≈70 nm for air under ambient conditions,[20] significantly reducing the contribution of heat conduction by the gas in the voids according to Equation (8). The mean free path of (Λs,0) in amorphous solids (e.g., polymers, sol–gel silica, etc.) is usually at a sub-nanometer scale, much less than both the void size (d) and dimensional parameter (δ) of HSMs that are often larger than 10 nm or even at the micrometer scale.[21] Such significant difference in length scale causes the term 2Λs,0/(d +δ) close to zero, indicating that the possible interfacial phonon scattering is negligible in typical HSMs. Therefore, the contribution of the solid to the total thermal conductivity of an HSM is mainly determined by the porosity (Π) and the intrinsic thermal conductivity of the corresponding bulk solid (ks,0) ks,HSM = (1 − Π)ks,0 (13) The surfaces and edges of a porous HSM behave differently from the solid material in contact with the voids inside the HSM, in particular when the HSM is thin enough to expose a significant fraction (fs) of solid to the surfaces and edges. Such a difference leads to different contributions to thermal conductivity for the solid material at different locations in the HSM. Glicksman developed a model to take the difference into account and describe the solid thermal conductivity in an HSM containing cubic voids as[22] 2 f ks,HSM = − s (1 − Π)ks,0 (14) 3 3 According to the minimum thermal conductivity model (MTCM),[23] the value of ks,0 of an amorphous solid is proportional to its atomic density (ρ0) and the average sound velocity (vsound,0). If a porous HSM exhibits an apparent density of ρ and sound velocity of vsound, the contribution of the solid to the total thermal conductivity can be described in an alternative way[13a,24,25] ks,HSM = ks,0 ⋅ 1801001 (5 of 17) ρ ⋅ ν sound (15) ρ0 ⋅ ν sound,0 © 2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.advancedsciencenews.com www.advmat.de Figure 4. A) Illustrative diagram showing a polymer foam consisting of the solid polymer (phase I) and the void-confined gas (phase II). B–D) Scanning electron microscopy (SEM) images of nanocellular open cell foams constituted 50 wt% of PMMA and 50 wt% of MAM (B), microcellular open cell foams constituted by 25 wt% of PMMA and 75 wt% of MAM (C), and nanocellular closed cell foams constituted by 95 wt% of PMMA and 5 wt% of MAM (D). E) Relationship between the thermal conductivity divided by the relative density (ρrel) of the foamed samples and the pore size. The relative density is the value of foam density divided by the density of the corresponding solids. F) Calculated thermal conductivity as a function of the relative density for PMMA foams with cells of varying sizes. The calculations were processed by considering three contribution of gas conduction (Kaganer’s model), solid conduction (Ashby’s model), and radiation (Williams and Aldao’s model). B–F) Adapted with permission.[15] Copyright 2014, Elsevier Ltd. Although various models have been reported in literatures, all of them consistently indicate that maintaining high porosity and small size (e.g., <100 nm) of voids in an HSM can simultaneously reduce the thermal conductivities in both the gas and solid phases. This property enables HSMs to be a class of unique thermal-insulation materials with significantly enhanced thermal insulation performance compared to the corresponding solid materials. Representative HSMs of various compositions are discussed in the following sections to highlight their superior thermal insulation performance. 3. Polymer Foams Polymer foams represent the widely used thermal-insulation materials because of their good thermal insulation performance, Adv. Mater. 2019, 31, 1801001 mechanical flexibility, durability, and low manufacturing cost. A polymer foam is a stable solid polymer monolith or film with trapped pockets of a desirable gas. The polymer and gas correspond to phase I and phase II, respectively. The interfacial phase III is the interface between phase I and phase II without a physical thickness in a polymer foam (Figure 4A). The solid phase I typically consists of one polymer or potentially of a polymer mixture containing multiple miscible components. Low-concentration additives are sometimes added to polymers to facilitate the foaming process or tune mechanical properties of the polymers. The trapped gas (phase II) can be confined as individual bubbles (closed-cell foams, Figure 4B,C) or interconnected channels (open-cell foams, Figure 4D). Forming polymer foams with inert gases such as CO2 and nitrogen has been intensively studied due to the environmentfriendly manufacturing process. An appropriate gas is first dissolved in a polymer at high pressures, usually at the supercritical 1801001 (6 of 17) © 2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.advancedsciencenews.com www.advmat.de conditions of the gas. Removing the supercritical conditions (e.g., decreasing pressure) triggers a solid–gas phase separation between the polymer and the gas, leading to the nucleation and growth of gas bubbles in the polymer matrix. The key to form polymer foams with low thermal conductivities is to boost the nucleation of gases, which potentially increases the density of voids and thus reduces the size of individual voids. According to the classical nucleation theory, increasing the supersaturation of a gas in a polymer that can be achieved by dissolving the gas at higher pressure for a longer time benefits the fast homogeneous nucleation. The nucleation of the gas can also be promoted through heterogeneous nucleation process by mixing high concentration of small nanoparticles, which serves as nucleation sites, in the polymer. The density, size, and morphology of the gas bubbles (or voids) in the polymer foam strongly depends on the process condition including pressure, temperature, saturation time, and depressurization rate, as well as the type of polymer and the density of nanoparticle fillers.[3a–c,26] In general, heterogeneous nucleation with a high concentration of nanoparticle fillers in polymers can efficiently reduce the void size and increase the void density compared with that from homogeneous nucleation. The porosity of polymer foams is usually higher than 50%, while the size of voids in polymer foams varies in a wide range from thousands of micrometers to tens of nanometers. The values of both parameters influence the thermal conductivity of polymer foams, based on the above theoretical discussion. The formation of small gas voids in a polymer reduces the contribution of the polymer to the thermal conductivity of the corresponding HSM due to the increased porosity, meanwhile the small void size enables the Knudsen effect to lower the contribution of the gas to the thermal conductivity of the HSM. A number of polymer foams have been synthesized and studied for thermal insulation, such as PS foams,[27] PU foams,[27e,28] PE foams,[14a,29] polyetherimide foams,[14b] PMMA foams,[15,30] polyether block amide (PEBA) foams,[31] and poly(vinylidene fluoride) (PVDF) foams.[32] Table 1 summarizes the sizes and densities of voids in the polymer foams and their corresponding thermal conductivities. The void fraction has a key effect on the effective thermal conductivity of a polymer foam. For example, PMMA foam with a porosity of 51.7% has a thermal conductivity of 0.107 W m−1 K−1, which can be largely reduced to ≈0.035 W m−1 K−1 by increasing the porosity of PMMA foam to 95.2%.[15,30] Park and co-workers have recently achieved extremely low thermal conductivities of 0.030 W m−1 K−1 for PEBA foam and 0.027 W m−1 K−1 for PVDF foam which have the porosities of 83.3% and 96.4%, respectively.[31,32] These values are close to the thermal conductivity of air (i.e., ≈0.026 W m−1 K−1), highlighting the great promise of polymer foams in thermal insulation. Thermal conductivity of the polymer foams with a constant porosity indeed decreases with the decrease in the size of the voids, in particular when the diameter of the voids is on the sub-micrometer scale. Thermal conductivity of the polymer foams with voids larger than 100 µm is dependent on the porosity and almost independent of the size of the voids. For example, polyurethane (PU) foams with void sizes from 147 to 341 µm and similar porosities (95.7–97.1%) exhibit similar thermal conductivities varying in a very narrow range (0.0334–0.0342 W m−1 K−1).[28b] Significant reduction of thermal Adv. Mater. 2019, 31, 1801001 Table 1. Thermal conductivities and void parameters of polymer foams formed from homogeneous nucleation processes.a) Polymer matrix (phase I) Thermal conductivity Year[Ref.] [W m−1 K−1] Void size [µm] Porosity [%] PS foam ≈200 97.2 0.0331 PU foam ≈200 96.9 0.0267 PU foam 300 96.3 0.0342 214 96.2 0.0339 147 95.8 0.0334 194 98.2 0.037 880 97.6 0.043 PE foam EVA/PE foamb) PEI foam PMMA/MAM foam 528 96.6 0.044 3294 ± 185 97.5 0.0460 1053 ± 87 96.9 0.0387 630 ± 52 97.1 0.0376 0.26 74 0.019 0.24 77 0.016 0.086 80 0.015 0.290 60.1 0.0884 0.235 56.8 0.0925 0.150 48.3 0.0948 0.0782 8.07 72.0 PMMA foam 0.82 51.7 0.1072 PMMA foam 445.9 95.24 0.0373 202.0 94.76 0.0363 208.5 95.55 0.0347 108.2 94.64 0.0353 2015[27e] 1998[28a] 2001[14a] 2008[29b] 2013[14b] 2015[15] 2017[30] 127.8 95.22 0.0343 PEBA ≈32 83.33 ≈0.030 2018[31] PVDF ≈150 96.4 0.027 2018[32] a) Phase II was air in voids; b)EVA: ethylene–vinyl acetate copolymer. conductivity of polymer foams requires the size of voids to be comparable to the mean free path of gas molecules in the voids, which is usually on the order of tens of nanometer. Therefore, recent research focuses on the synthesis of polymer foams with voids smaller than 200 nm, which are generally called nanocellular foams, to achieve high insulation performance.[3b,c,e] In nanocellular foams collisions of gas molecules are largely limited in the small voids, leading to the significant suppression of the gas thermal conductivity due to the Knudsen effect.[15] Notario et al. demonstrated the strong relationship between the void size and thermal conductivity in PMMA foams, as shown in Figure 4E. The ratio of thermal conductivity (k) and relative density (ρrel = ρfoam/ρsolid) generally shows a decrease trend with the void size from 1 µm to 100 nm, which can be attributed to the Knudsen effect. Simulation results show the PMMA foam with a void size of 10 nm has the lowest thermal conductivity compared with PMMA foams with a larger void size in the range of the relative density from 0.05 to 0.6 (Figure 4F). A theoretical study recently points out the good insulating 1801001 (7 of 17) © 2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.advancedsciencenews.com www.advmat.de property with a nanocellular polymer foam requests the void size less than 200 nm and a porosity in the range of 90–95%.[19] The synergistic effects of nanometer-sized voids on heat transport in both gases and polymers enable the synthesis of polymer insulation materials with very low thermal conductivities. For instance, Sundarram and Li successfully reduced thermal conductivity of the PEI foam to 0.015 W m−1 K−1 by forming CO2 voids of 86 nm in diameter with a porosity of 80%.[14b] Despite the successful tuning ability of foaming process on the size of voids in the homogeneous nucleation synthesis, it is still extremely challenging to reduce the void size down to the ideal values while maintaining a high porosity. The difficulty originates from the low nucleation efficiency, i.e., low density of nuclei formed from the spontaneous homogeneous nucleation. This challenge has been tackled by blending polymers with nanoparticles of different compositions that exhibit stronger affinity toward gases than that of polymer. The nanoparticle fillers can serve as nucleation sites of gas to promote the nucleation efficiency. When the concentration of the nanoparticle fillers is high enough, the density of gas voids can be increased to the appropriate values. At a specific porosity (i.e., the total volume of voids remaining constant), the size of individual voids decreases accordingly, benefiting the reduction of thermal conductivity of the resulting polymer foams. Various nanoparticles have been evaluated as fillers to improve thermal insulation performance of polymer foams. Typical nanoparticles fillers include clay nanoparticles,[33] TiN nanoparticles,[34] silica nanoparticles,[35] carbon fibers or nanotubes,[36] organic domains,[37] and so forth. Table 2 summarizes the detailed information of the composition and structure of the composite foams as well as their corresponding thermal conductivities. For example, the presence of clay nanoparticles with a mass concentration of around 2% (≈2 wt%) in PU could reduce the size of voids in the PU foams from 514 to 360 µm under the same foaming conditions.[33b] High dispersion of nanoparticle fillers in polymers is essential to fully benefit the heterogeneous nucleation in which the density of gas voids is improved compared to the homogeneous nucleation in polymers without nanoparticle fillers.[35a,36a] Surface modification of inorganic nanoparticle fillers with appropriate organic molecules that are compatible with the polymers represents a practical strategy to increase the dispersity of nanoparticles. Typical examples include the modification of clay nanoparticles with polymeric 4,4ʹ-diphenylmethane diisocyanate (PMDI) and the modification of silica nanoparticles with 3-(trimethoxysilyl)-propyl methacrylate (MSMA). The modified silica nanoparticles expose vinyl groups to the surroundings and are labeled as vinyl-modified silica nanoparticles or VMS nanoparticles. Using the modified nanoparticles decreases the size of voids and improves the uniformity of the voids in polymer foams. For instance, Yeh et al. dispersed the VMS nanoparticles in PMMA through a polymerization reaction of methyl methacrylate (MMA) in the presence of VMS nanoparticles.[35a] The well-dispersed VMS nanoparticles significantly increased the heterogeneous nucleation efficiency, resulting in an increase of density of voids and a decrease of void size compared with the foams with unmodified silica nanoparticles. In the presence of 2 wt% VMS nano­particles the resulting PMMA foam exhibited a void density 2.69 × 1012 cm−3 and a void size of 10.90 µm, which were 116.9% higher and 34.2% smaller, respectively, compared with the foam formed in the presence of 2 wt% unmodified silica nanoparticles that exhibited a void density of 1.24 × 1012 cm−3 and a void size of 16.57 µm. In contrast, the pure PMMA foam formed under the same condition exhibited a much smaller void density of only 5.48 × 1011 cm−3 and a larger void size of 22.43 µm. The comparison again highlights that adding monodisperse nanoparticle fillers to polymers can reduce the size of voids in the polymer foams, favoring the Table 2. Thermal conductivities and void parameters of polymer foams formed from homogeneous nucleation processes in the presence of nanoparticles. Composite matrix (phase I) PU/clay foam Gas (phase II) d) HFC-365mfc PU foam Filler mass concentration [wt%] Void size [µm] 3 360 Porosity [%] Thermal conductivity [W m−1 K−1] Year[Ref.] 0.0185 2007[33b] 0 514 PU/clay foam Air 5 163 94.0 0.0166 2008[33c] PU/clay foam Air 5 390 95.1 0.0332 2016[33d] PU/CNF foama) Air 0.5 86 0.0157 2010[36a] PMMA foam Air 0 22.43 61.6 0.0944 2009[35a] PMMA/VMS foam 1 12.71 50.3 0.0860 2 10.90 47.0 0.0801 PMMA/RS 2 16.57 57.4 0.0904 PS/CNT foamb) Air 1 5.8 Phenolic/TiN foam Air 1 102.1 ± 4 PMMA/PU foam Air 1 PP/PTFE foam BPP/PTFE a) foamc) 0.0194 0.0328 2015[36b] 0.0312 2016[34] 0.205 0.0248 2017[37a] 0.930 0.0369 97.6 Air 5 22.5 0.0365 2017[37b] Air 1 45 0.0324 2018[37c] CNF: carbon nanofiber; b)CNT: carbon nanotube; c)BPP: branched polypropylene; d)1,1,1,3,3-pentafluorobutane. Adv. Mater. 2019, 31, 1801001 1801001 (8 of 17) © 2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.advancedsciencenews.com www.advmat.de Figure 5. A) Illustrative diagram of an HSM made of blended polymer and hollow silica particles. Phases I, II, and III correspond to the polymer, air, and silica, respectively. Cross-sectional B) SEM and C) TEM images of organosilicate hybrid films containing 10 wt% of hollow silica nanoparticles with an average diameter of 50 nm and a shell thickness of 8–10 nm. B,C) Adapted with permission.[40] Copyright 2013, Springer Nature. D) Optical photograph of the transparent polyethersulfone composite film containing 2.5 wt% of hollow silica spheres with 200 nm in diameter and 6.2 nm in shell thickness. The film had a thickness of 35 µm, which is highlighted inside the rectangle outlined with dotted lines. D) Adapted with permission.[38d] Copyright 2016, American Chemical Society. reduction of their thermal conductivities. The PMMA foams containing 1 and 2 wt% VMS nanoparticle fillers exhibited low thermal conductivities of 0.0860 and 0.0801 W m−1 K−1, respectively, which were 8.9% and 15.1% lower than that of the PMMA foam. Besides inorganic nanoparticles, immiscible polymers are also possibly used as fillers to promote heterogeneous nucleation of gas. In a synthesis of PMMA foams with CO2 gas, PU that could form nanosized domains in PMMA matrix was added as filler to promote nucleation of CO2 voids,[37a] forming PMMA/PU composite foams with a void size of 205 nm and porosity larger than 85%. The nanosized voids and high porosity reduce the thermal conductivity of the PMMA/PU foams down to 0.0248 W m−1 K−1. 4. Blends of Polymers and Hollow Silica Particles Voids in a polymer matrix can be achieved by directly adding hollow particles (or shells), which are synthesized independently, to a solution or melt of the polymer followed by solidifying the polymer. Such a direct mixing process does not require the supercritical conditions (e.g., high pressure and high temperature) of Adv. Mater. 2019, 31, 1801001 gases used in the synthesis of polymer foams. In a blend of a polymer and hollow particles of an inorganic material (e.g., silica), the polymer and the empty interiors of the hollow particles represent the solid phase I and gas phase II, respectively, of an HSM (Figure 5A). The inorganic walls of the hollow particles represent the interfacial phase III of the HSM. According to the discussions in Section 2, the gas thermal conductivity in the void phase II can be reduced by using small hollow particles. The presence of the interfacial phase III makes a positive contribution to thermal conductivity because the inorganic solids are usually more thermally conductive than polymers. This undesired contribution can be alleviated by reducing the wall thickness of the hollow particles. Therefore, availability of small hollow particles with thin walls becomes important to synthesize composite polymer insulation materials with low thermal conductivities. Several types of blended polymers containing hollow inorganic shells are summarized in Table 3 with detailed information (e.g., component, filler size and concentration, porosity, etc.) and their corresponding thermal conductivities. Hollow silica (or glass) particles represent a class of widely explored materials to bring voids into composite polymer insulation materials because of the low cost of silica particles and their optical transparency.[38] For 1801001 (9 of 17) © 2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.advancedsciencenews.com www.advmat.de Table 3. Thermal conductivities and other parameters of blended polymers and hollow inorganic shells (glass, ceramic, and TiO2). Polymer matrix (phase I) Interfacial phase III Porosityc) [%] Filler Thermal conductivity [W m−1 K−1] Year[Ref.] Size [µm] Wall thickness [µm] Concentration PU Silica 0.375 0.040 0.25 wt% 0.14 0.05 2012[38b] PU Silica 0.1 0.010 10 wt% 6.11 0.029 2015[38c] PES Silica 0.181 0.006 12 vol% 9.77 0.030 2016[38d] Epoxy resin (E51) Glass 40 1.05 0 0 0.226 2015[38e] 10 vol% 8.51 0.211 20 vol% 17.01 0.196 30 vol% 25.52 0.174 40 vol% 34.02 0.163 50 vol% 42.53 0.156 Polyacrylate TiO2 0.394 0.043 0.1825 2017[41b] PVDFa) Glass 72.5 ± 42.5 1.0 ± 0.5 0.047 2017[38f ] Ceramic 91.6 ± 60 10 vol% 0.1815 Silica 27 ± 16 20 vol% 0.1493 Glass 52 ± 29 30 vol% 0.1362 b) EPON 862 Silicone rubber 0.110 2017[42] a) PVDF: poly(vinylidene fluoride); b)EPON 862: epoxy resin; c)The porosity was calculated from dividing the volume of voids in fillers by the total volume of blended composites using the parameters including the wall thickness, concentration, and densities of fillers as well as densities of polymers. example, Fuji and co-workers synthesized composite films by dispersing 10 wt% hollow silica nanoshells (with a diameter of around 100 nm and a wall thickness of ≈10 nm) in PU, achieving a thermal conductivity of 0.029 W m−1 K−1, which is less than one tenth of that of the pristine PU film (0.30 W m−1 K−1).[38c] Similar to the polymer foams, a higher number density of hollow silica nanoshells in a polymer reduces the characteristic size (δ) of the polymer network when the geometrical parameters of the silica nanoshells are constant, leading to a lower thermal conductivity of the polymer/silica composite material.[38a,39] Hollow silica nanoshells have also been dispersed in varying polymers to synthesize insulating materials with low thermal conductivities. Figure 5B,C shows the organosilicate (OS) composite films containing well-dispersed hollow silica nanoparticles due to the strong affinity between silica and OS matrix leading to a reduced thermal conductivity.[40] Ernawati et al. prepared composite polyethersulfone (PES) films containing 2.5 wt% of well-dispersed silica nanoshells (with a diameter of 181 nm and a shell thickness of 6.2 nm). The film with a thickness of ≈35 µm exhibited a thermal conductivity of 0.03 W m−1 K−1, much lower than that of the pure PES film (0.09 W m−1 K−1).[38d] Owing to the small size of hollow silica nanoshells and the intrinsic optical transparency of silica, the PES/silica composite film exhibited a high optical transmittance of ≈90% in the visible region of 400–800 nm (Figure 5D). The type of inorganic hollow nanoshells can also be independently tuned to synthesize composite polymer insulation materials. For example, hollow TiO2 particles were dispersed in polyacrylate to lower thermal conductivity of the polyacrylate film.[41] Zhao et al. dispersed three types of commercial hollow microparticles made of ceramic, silica, and glass in silicone rubber.[42] Thermal conductivities of all the resulting composites were lower than the pristine silicone rubber. Adv. Mater. 2019, 31, 1801001 It is worthy of pointing out that high dispersity of inorganic hollow particles in polymers is critical to efficiently lower thermal conductivity of the polymers especially when the loading of hollow particles is high. A high concentration of hollow particles in a polymer matrix can easily aggregate to form continuous chains that provide additional heat transport pathways with much higher thermal conductivity than that of polymer.[43] This effect could erase the benefit brought by the high-density voids towards reduction of thermal conductivity of the composite thermal-insulation materials. Ruckdeschel et al. have investigated the thermal transport in a binary colloidal mixture of hollow silica particles and P(MMA-co-nBA) polymer particles with varying ratios.[44] The assembled monoliths of hollow silica (441 nm in diameter) and polymer particles (423 nm in diameter) exhibit thermal conductivities of 0.013 and 0.057 W m−1 K−1 in vacuum, respectively. The thermal conductivities of the monoliths of binary colloidal mixtures increase monotonically from 0.013 to 0.057 W m−1 K−1 as the ratio of polymer particles increases from 0% to 100%. Melting the polymer particles at a temperature above the glass transition temperature fuses the polymer particles into a continuous network, creating additional percolation pathways for thermal transport and thus increasing thermal conductivity accordingly. Moreover, the structural integrity of hollow shells is also important to achieve high concentration of voids and thus low thermal conductivity of the corresponding polymer composites.[45] This issue becomes more challenging when using hollow shells with large sizes and thin walls that are easily broken. Once the hollow shells are broken, the voids will disappear and the resulting inorganic fragments will increase the possibility to form inorganic heat transport pathways, both of which prevent the reduction of thermal conductivity of the polymer–hollow-shell composites.[45] 1801001 (10 of 17) © 2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.advancedsciencenews.com www.advmat.de 5. Assembly of Hollow Silica Particles Inorganic hollow particles can be directly assembled into selfsupported macroscopic thermal-insulation materials. In an HSM made from assembled hollow particles both the quasi-continuous phase I and isolated void phase II are composed of gas, while the particle shells form the interfacial phase III (Figure 6A). The adjacent particles in a closely packed assembly are in tight contacts, which break the spatial continuity of the phase I to individual gas pockets with much smaller volumes. Therefore, the contribution of phase I to thermal conductivity of the HSM becomes minimum and similar to that of the void phase II. In contrast, the tight contacts of the hollow particles create a high density of inorganic heat transport pathways along the particle shells, which dominates the thermal conductivity of the corresponding HSM. The accuracy in controlling the size and shell thickness of hollow particles as well as their assembly behavior is beneficial for correlating thermal conductivities of the corresponding HSMs to the properties of the hollow particles.[46] The relationship is useful to guide the synthesis of appropriate hollow particles for making HSMs with superior thermal insulation performance. The HSMs formed from assembly of inorganic hollow particles can be operated at high temperatures due to the absence of easy-burning polymers. Gao et al. demonstrated the assembly of hollow silica particles with 150 nm in void diameter and 10–15 nm in shell thickness to form an HSM with a thermal conductivity of ≈0.02 W m−1 K−1 (Figure 6B).[46d] The significantly lowered thermal conductivity compared to that of bulk silica (1.2–1.4 W m−1 K−1)[47] was attributed to the small contact areas of adjacent particles that narrows the heat transport pathways. Puckdeschel and co-workers designed experiments to systematically study the dependence of thermal conductivity of the silica-particle HSMs on the interfacial contacts of hollow silica particles.[46e] Beside the contact interfaces, the defects (e.g., mesopores, organic residuals retained from synthesis, nonuniform mass distribution, surface roughness, etc.) in individual silica shells also influenced the thermal conductivity of the HSMs due to their ability of scattering phonons. To differentiate the contributions of the interfacial contacts between silica shells and the structural defects in individual silica shells, two groups of HSMs made of the carefully synthesized silica nanoshells were fabricated and evaluated in thermal conductivities. The monodisperse silica nanoshells with a diameter of 316 nm and a shell thickness of 44 nm were initially deposited on uniform polystyrene (PS) beads via a sol–gel reaction, forming PS@silica core–shell particles. Calcination of the PS@silica core–shell particles at elevated temperatures (≥500 °C) burned away the templating PS beads encapsulated in the silica nanoshells, leaving empty interiors in the silica nanoshells (i.e., hollow silica nanoshells, Figure 6B,C). Calcination of the silica nanoshells at elevated temperatures in air can reduce the density of structural defects in individual nanoshells and a higher Figure 6. A) Illustrative diagram of an HSM formed from the assembly of hollow silica nanoshells. Both phase I and phase II are gas while the interfacial phase III is silica. B) Transmission electron microscopy (TEM) image of assembled hollow silica nanoshells with a diameter of 150 nm and a wall thickness of 10–15 nm. B) Adapted with permission.[46d] Copyright 2013, American Chemical Society. C) SEM image of assembled hollow silica nanoshells. D) Thermal conductivities of the HSMs of assembled hollow silica nanoshells as a function of their Brunauer–Emmett–Teller (BET) surface areas and annealing temperature (inset). The thermal conductivities were measured at 25 °C. E) Schematic illustration highlighting the difference in the density of structural defects in the wall of individual nanoshells and the interfacial bonding (or interfacial contact) between adjacent nanoshells. C–E) Adapted with permission.[46e] Copyright 2015, The Royal Society of Chemistry. Adv. Mater. 2019, 31, 1801001 1801001 (11 of 17) © 2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.advancedsciencenews.com www.advmat.de temperature leads to a more significant result. When the silica nanoshells are in contact and well packed, thermal annealing can also increase the interfacial bonding between the adjacent nanoshells, resulting in possible enlargement of the interfacial contact area and enhancement of phonon exchange between the adjacent nanoshells. Both effects indicate that thermal conductivity of the assembled monoliths of silica nanoshells increases with the annealing temperature. To differentiate the influence of structural defects and interfacial bonding on thermal conductivity of the silica nanoshell monoliths, two groups of samples were compared comprehensively. The first group of HSMs were made from assembly of the PS@silica core–shell particles following by calcination at different temperatures resulting in both changes of interfacial bonding and structural defects between/in the silica nanoshells. High calcination temperature increases the interfacial bonding but reduces the density of structural defects, which leads to the increase of thermal conductivity (solid symbols, Figure 6D) corresponding to a structural transition from (1) to (3) (Figure 6E). Compared to the HSM formed from the assembly of the PS@silica nanoshells following by calcination at 500 °C, thermal conductivity of the HSM with the calcination of 950 °C increased by about 100%, i.e., from 0.071 to 0.140 W m−1 K−1 (solid symbols, inset of Figure 6D). The increase of thermal conductivity is caused by both the reduced density of structural defects in individual nanoshells and the enlarged interfacial bonding between nanoshells (structural 3 vs structure 1, Figure 6E). The second group of HSMs were fabricated from the assembly of hollow silica nanoshells synthesized at 950 °C, followed by thermal annealing at different temperatures. Since the hollow silica nanoshells were synthesized at 950 °C which significantly reduced the structural defects, the assembly of silica nanoshells followed by different annealing temperatures only brought the changes of interfacial bonding. Thermal conductivity of the HSM annealed at 950 °C increased by 64%, i.e., from 0.094 to 0.154 W m−1 K−1, compared to the HSM annealed at 450 °C (closed symbols, inset of Figure 6D). The increase in thermal conductivity of the second group samples was mainly contributed by the enlarged interfacial bonding between nanoshells at higher temperatures, corresponding to a structural transition from (2) to (3) (Figure 6E). Assuming thermal annealing of the assembled PS@silica core–shell particles and the assembled silica hollow nanoshells (formed from calcination of the PS@silica core–shell particles) at the same temperature produces similar interfacial bonding between the silica nanoshells, the variation of internal density structural defects in individual silica nanoshells is responsible for the difference of thermal conductivity between these two groups of samples shown in the inset of Figure 6D. Lowering the density of structural defects in individual silica nanoshells corresponds to the increase of thermal conductivity from 0.071 to 0.094 W m−1 K−1, which is reflected as the structural transition of (1) → (2) shown in Figure 6D,E. This difference is equivalent to 32%, which is on the similar scale to the difference caused by increasing interfacial bonding in the (2) → (3) transition even though the absolute value is smaller (32% vs 64%). The comparisons indicate that both reducing the interfacial bonding between adjacent silica nanoshells (i.e., narrowing the heat transport pathways) and increasing the density of structural defects in individual silica nanoshells can lower thermal conductivity of an HSM assembled from the silica nanoshells despite the former is more effective. The same author in another work proposed to reduce the thermal conductivity by reducing the interparticle contact points from 12 to 6 by changing the assembly order from ordered packing to random packing.[46f ] The geometry of the hollow silica particles (i.e., size and shell thickness) also plays significant roles on the thermal conductivity of the ensemble.[46f ] Two different series of hollow silica particles with different diameters (267, 387, and 469 nm) or shell thicknesses (14, 27, and 40 nm) are used to systematically identify the correlation between the geometry parameter with the thermal conductivity of the corresponding ensembles (Figure 7A,B). Thermal conductivity measurements in Figure 7C show that the thermal conductivity decreased as the walls of the silica nanoshells became thinner. Increasing the diameter of silica nanoshells with a similar shell thickness further lowered the thermal conductivity. A low value of 0.027 W m−1 K−1 was achieved in vacuum for the assembly of hollow silica nanoshells with 469 nm in diameter and 17 nm in shell thickness (HS-469/17), which increased up to 0.082 W m−1 K−1 for that of HS-266/40 (smaller nanoshells with thicker walls). Figure 7. A) TEM images of hollow silica nanoparticles with various diameters and wall thicknesses. B) SEM images of the monolithic HSMs assembled from the corresponding hollow silica nanoparticles shown in (A). C) Thermal conductivities of the HSMs of B) measured in helium and air at 1000 mbar as well as vacuum (i.e., in the presence of 0.05 mbar of air). Adapted with permission.[46f ] Copyright 2017, Wiley-VCH. Adv. Mater. 2019, 31, 1801001 1801001 (12 of 17) © 2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.advancedsciencenews.com www.advmat.de The self-support macroscopic films of hollow silica structures can also be synthesized from a one-step process, in which formation of hollow silica nanoshells and their self-assembly simultaneous occur at a two-phase interface (e.g., gas/water or oil/ water interface).[48] For example, Mille and Corkery synthesized films of assembled hollow silica shells by using a microemulsion method that involved the formation of silica shells at an oil/water interface.[48b] The walls of the silica shells were mesoporous, and the existence of a high density of nanometer-sized pores lowered the mass density of the silica shells to only 0.58 g cm−3. The thermal conductivity of the film was 0.041 W m−1 K−1, which represents only 3% of the thermal conductivity of bulk silica. 6. Aerogels The 3D structure of HSMs made from assembly of hollow silica shells discussed in Section 5 can be mimicked by forming aerogels of silica nanoparticles. In a silica aerogel, the solid network composed by silica nanoparticles is the interfacial phase III highlighted by the dotted line in Figure 8A. The phase I and phase II of aerogels are composed of gas. The characteristic size, δ, could be below zero due to the high void fraction and quasi-continuous solid network, indicating the interconnection of voids. The rendered 3D model of a typical silica aerogel is schematically shown in Figure 8B, highlighting the interlaced network of assembled chains of silica nanoparticles. In an silica aerogel HSM shown in Figure 8C, the voids and gas pockets are usually less than 100 nm in size, i.e., on the order of several nanometers to ten of nanometers, which significantly reduces the contribution of trapped gas to the thermal conductivity of the HSM.[49] The solid heat transport pathways are formed from the intricate network of silica nanoparticle chains. The small size (i.e., 2–5 nm) of the silica nanoparticles defines the characteristic size of the heat transport pathways that is very small and can lower the thermal conductivity. The high density of internanoparticle interfaces favors the strong phonon scattering and thus reduce the thermal conductivity. Therefore, highquality silica aerogels can have very high porosities (>97%) and low thermal conductivities around 0.03 W m−1 K−1 at ambient atmospheric condition. The thermal conductivity can be further reduced by vacuuming the aerogels to a value that is even lower than the thermal conductivity of still air (0.026 W m−1 K−1). Silica aerogels were invented by Kistler in 1931[50] and have attracted great attention due to their unique properties including extremely low mass density, high surface area, and low thermal conductivity.[51] Successful synthesis of high-quality silica aerogels is challenging. The critical step is the removal of entrapped solvents from the wet gels without breaking the integrity of the 3D networks of silica nanoparticles. Direct drying of a wet silica gel under ambient conditions leads to shrink its volume due to collapse of the voids. Such a mechanical failure originates from the poor mechanical properties of the fractal silica network and the strong capillary forces exerting on the solid network during solvent evaporation.[52] Kistler and Caldwell used supercritical organic solvents (e.g., dichlorodifluoromethane) to wet a silica gel followed by drying the wet gel at the critical point of the solvent, at which the conversion from liquid to gas was imperceptible without involvement of capillary compressive forces.[53] Complete removal of organic solvent left the intact skeleton of silica nanoparticles that maintained almost the same apparent volume as that of the wet gel. An alternative way to prevent the mechanical failure is to eliminate the surface tension force by switching the surface property of the silica nanoparticles from hydrophilic to hydrophobic. A silylation process can graft methylsilane reagents such as methyltriethoxysilane (MTES), trimethylethoxysilane (TMES), and Figure 8. A) Schematic illustration of the microscopic geometry of a silica aerogel slice. Both phase I and phase II are gas, and phase III is a network of the chains of assembled silica nanoparticles. B) Rendered 3D illustration of a silica aerogel. C) TEM and SEM (inset) images of a slice of typical silica aerogel. C) Adapted with permission.[54b] Copyright 2008, Springer Nature. D) SEM and TEM (inset) images of an aerogel made from singlewalled carbon nanotubes. E) Comparison of thermal conductivity of the SWCNT aerogels and conventional carbon aerogels with various densities. D,E) Adapted with permission.[58] Copyright 2013, Wiley-VCH. Adv. Mater. 2019, 31, 1801001 1801001 (13 of 17) © 2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.advancedsciencenews.com www.advmat.de trimethylchlorosilane (TMCS), to the silica nanoparticles, resulting in hydrophobic surfaces.[54] For example, Wei et al. modified the silica nanoparticle surfaces with TMCS and the corresponding silica wet gel was dried essentially without suffering of volume shrinkage (<1%) even under ambient conditions.[55] Avoiding the significant volume shrinkage retained the high porosity of the dry silica aerogel, enabling a low thermal conductivity of 0.036 W m−1 K−1, much lower than that formed from the unmodified silica nanoparticles (0.417 W m−1 K−1). These successes shed light on the commercialization of silica aerogels to the market. The silica aerogels with high porosities are mechanically fragile, limiting their applications. Their poor mechanical strength can be improved by forming composite aerogels with other materials (e.g., PMMA, fabrics and xonotlite) that exhibit good thermal insulation and excellent mechanical flexibility.[56] Moreover, nonfragile aerogels made of different materials such as 1D carbon nanotubes and mechanically flexible organics are receiving equivalent attention.[8a,57] Zhang et al. fabricated aerogels from single-walled carbon nanotubes (SWCNTs),[58] showing an ultralow mass density of ≈6.6 kg cm−3 that was much lower than the conventional carbon aerogels.[59] The atomic-level continuity and excellent mechanical strength/ flexibility of the SWCNTs makes the 3D entangled network to be more rigid (Figure 8D), facilitating the accommodation of more gas to exhibit a lower average density. As a result, the thermal conductivity of the SWCNT aerogel was measured to 0.056 W m−1 K−1[58] although the thermal conductivity of individual SWCNTs is high (Figure 8E).[60] An organic RF aerogel exhibited a thermal conductivity of only 0.012 W m−1 K−1 under ambient condition.[11a] Luo et al. synthesized the crosslinked glycidyl methacrylate (GMA)–divinylbenzene (DVB) copolymer monoliths (aerogels) containing voids with tunable sizes.[61] Decreasing the void size from 889 to 190 nm led to a decrease of thermal conductivity from 0.038 to 0.022 W m−1 K−1. 7. Conclusion and Outlook Hollow-structured materials are a class of macroscopic monoliths containing numerous microscopic voids filled with gases, in which the spatial distributions of the immiscible gaseous and solid phases are controlled to deliver new properties that are difficult or even impossible to achieve using the corresponding solid materials only. For example, the existence of high-density small gaseous voids in a polymer foam breaks the continuity of the polymer as well as the transport behavior in the polymer. Heat transfer in a polymer foam can be dramatically interrupted since the thermal conduction mechanisms in a gas and a solid polymer are different, i.e., molecular collision in the gas versus molecular vibration in the polymer. Simultaneously decreasing the size of the gaseous voids and increasing the density of the voids enhances the discontinuity of thermal transport pathways, favoring a significant reduction of thermal conductivity in the foamed polymer compared with the solid ones. In particular, thermal transport behavior may fall in a confinement region to further lower the thermal conductivity of a foamed polymer when the voids are small enough, e.g., on the scale of less than 100 nm. Such a discontinuity of thermal transport in HSMs makes the HSMs a class of promising materials for superior thermal insulation (e.g., comparable or even lower than the thermal conductivity of ambient air). The recent intensive efforts on various types of HSMs are well acknowledged to develop thermal-insulation materials with extremely low thermal conductivity for high-end applications. Figure 9 summarizes thermal conductivities of different HSMs Figure 9. Dependence of the normalized thermal conductivities of various HSMs on the void size. The normalization was calculated from dividing the recorded thermal conductivities (kHSM) by both the volume fraction and the intrinsic thermal conductivity of the solid components in the HSMs, i.e., kHSM/[(1 − Π)ks,0]. The colors differentiate the type of HSMs, i.e., black, red, green, and blue correspond to the polymer foams formed through (black full-filled symbols) homogeneous nucleation and (black half-filled symbols) heterogeneous nucleation (samples in Section 3), blend of polymer and hollow silica nanoparticles (samples in Section 4), assembled hollow silica nanoshells (samples in Section 5), and aerogels (samples in Section 6), respectively. The clustered distribution of the normalized thermal conductivities highlights the importance of the void size in influencing the thermal conductivity of HSMs. The voids smaller than ≈350 nm bring a positive effect to reduce the thermal conductivity of an HSM in addition to the effect of lowered mass density. In contrast, the HSMs with larger voids cannot completely benefit the lowered mass density to reduce their thermal conductivity. Adv. Mater. 2019, 31, 1801001 1801001 (14 of 17) © 2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.advancedsciencenews.com www.advmat.de with varying porosities and void sizes. The recorded thermal conductivities (kHSM = ks,HSM + kg,HSM) of the HSMs are normalized against the volume fractions of the solid components (1 − Π) and the intrinsic thermal conductivities (ks,0) of the solid components in the HSMs by following kHSM/[(1 − Π)ks,0]. When the thermal conductivities of the HSMs are dominated by the solid heat transport pathways, the normalized thermal conductivities would be close to 1 if the heat transport behavior is independent of the voids. The broad distribution of thermal conductivities of HSMs summarized in Figure 9 indicates that the void size represents an important parameter to influence the heat transport properties in the HSMs. As the voids are smaller than ≈350 nm, the HSMs exhibit normalized thermal conductivities lower than 1 regardless of the types of the HSMs (e.g., those discussed in Sections 3, 4, 5, 6). It indicates that reducing the void size may induce size-dependent phenomena (e.g., confinement effects) to further lower the thermal conductivity of an HSM in addition to the contribution made by the lowered mass density. In contrast, the normalized thermal conductivities of the HSMs with voids larger than ≈350 nm are usually larger than 1. The summary of Figure 9 highlights the importance of reducing the void size on fabricating high-quality insulation materials when the same porosity can be achieved in the HSMs. Besides low thermal conductivities, additional functions and fabrication cost should also be well considered to bring an HSM as a thermal-insulation material for practical applications. For instance, silica aerogels possess extremely low thermal conductivities and are stable at extreme temperatures, but their mechanical fragility and costly manufacturing process hinder their wide applications.[62] While we benefit from the low thermal conductivity originated from the breakage of continuity of thermal transport pathways in HSMs, the existence of gaseous voids also breaks the continuity of light transport pathways due to scattering, leading to optical opacity of HSMs. The light scattering caused by the gaseous voids can be eliminated by reducing the void size below 100 nm to form optically transparent HSMs, which are useful for thermal insulation requiring transparency, such as laminating films of single pane windows.[63] The function of blocking UV irradiation in the transparent insulation films can be added by introducing TiO2, which is transparent in the visible region but strongly absorbs UV light, to composite HSMs. For instance, an organosilicate film containing dispersed silica/TiO2 hybrid hollow nanoshells (with a size of 50 nm) exhibits a thermal conductivity 20% lower than the film without nanoshells.[40] The film is highly transparent in the visible region and is efficient in blocking the UV light. Doping HSMs with components that strongly absorb infrared irradiations (e.g., carbon fibers) is useful to reduce the radiative thermal conductivity of the HSMs.[64] Acknowledgements This work was completed with Government support under Award No. DE-AR0000739, awarded by the Advanced Research Projects Agency– Energy (ARPA-E), U.S. Department of Energy, through a subcontract from the University of Chicago. Conflict of Interest The authors declare no conflict of interest. Adv. Mater. 2019, 31, 1801001 Keywords aerogels, hollow-structured materials, polymer foams, silica nanoshells, thermal-insulation materials Received: February 11, 2018 Revised: July 31, 2018 Published online: October 31, 2018 [1] N. L. Panwar, S. C. Kaushik, S. Kothari, Renewable Sustainable Energy Rev. 2011, 15, 1513. [2] E. Cuce, S. B. Riffat, Renewable Sustainable Energy Rev. 2015, 41, 695. [3] a) A. Zhang, Q. Zhang, H. Bai, L. Li, J. Li, Chem. Soc. Rev. 2014, 43, 6938; b) C. Forest, P. Chaumont, P. Cassagnau, B. Swoboda, P. Sonntag, Prog. Polym. Sci. 2015, 41, 122; c) S. Liu, J. Duvigneau, G. J. Vancso, Eur. Polym. J. 2015, 65, 33; d) G. Tang, C. Bi, Y. Zhao, W. Tao, Energy 2015, 90, 701; e) B. Notario, J. Pinto, M. A. Rodríguez-Pérez, Prog. Mater. Sci. 2016, 78–79, 93. [4] X. Wang, J. Feng, Y. Bai, Q. Zhang, Y. Yin, Chem. Rev. 2016, 116, 10983. [5] L. J. Gibson, M. F. Ashby, Cellular Solids: Structure & Properties, Pergamon Press, Oxford, UK 1988. [6] L. Glicksman, M. Schuetz, M. Sinofsky, Int. J. Heat Mass Transfer 1987, 30, 187. [7] a) X. Lu, R. Caps, J. Fricke, C. T. Alviso, R. W. Pekala, J. Non-Cryst. Solids 1995, 188, 226; b) R. Siegel, J. R. Howell, Thermal Radiation Heat Transfer, Taylor & Francis, New York 2002. [8] a) S. N. Schiffres, K. H. Kim, L. Hu, A. J. H. McGaughey, M. F. Islam, J. A. Malen, Adv. Funct. Mater. 2012, 22, 5251; b) J. Jeans, An Introduction to the Kinetic Theory of Gases, Cambridge University Press, Cambridge, UK 1982. [9] T. K. Tokunaga, J. Chem. Phys. 1985, 82, 5298. [10] a) J. Zhao, Y. Duan, X. Wang, B. Wang, J. Nanopart. Res. 2012, 14, 1024; b) K. Denpoh, IEEE Trans. Semicond. Manuf. 1998, 11, 25; c) M. G. Kaganer, Thermal Insulation in Cryogenic Engineering, Israel Program for Scientific Translations, Jerusalem, Israel 1969; d) G. Reichenauer, U. Heinemann, H. P. Ebert, Colloids Surf., A 2007, 300, 204. [11] a) X. Lu, M. C. Arduinischuster, J. Kuhn, O. Nilsson, J. Fricke, R. W. Pekala, Science 1992, 255, 971; b) E. Solórzano, M. A. Rodríguez-Pérez, J. Lázaro, J. A. de Saja, Adv. Eng. Mater. 2009, 11, 818. [12] S. S. Kistler, J. Phys. Chem. 1934, 39, 79. [13] a) O. Lee, K. Lee, T. J. Yim, S. Y. Kim, K. Yoo, J. Non-Cryst. Solids 2002, 298, 287; b) C. Bi, G. Tang, W. Tao, J. Non-Cryst. Solids 2012, 358, 3124. [14] a) O. Almanza, L. O. Arcos y Rábago, M. A. Rodríguez-Pérez, A. González, J. A. de Saja, J. Macromol. Sci., Part B: Phys. 2001, 40, 603; b) S. S. Sundarram, W. Li, Polym. Eng. Sci. 2013, 53, 1901. [15] B. Notario, J. Pinto, E. Solorzano, J. A. de Saja, M. Dumon, M. A. Rodríguez-Pérez, Polymer 2015, 56, 57. [16] G. Chen, Nanoscale Energy Transport and Conversion: A Parallel Treatment of Electrons, Molecules, Phonons, and Photons, Oxford University Press, New York 2005, Ch. 1. [17] a) P. K. Schelling, S. R. Phillpot, P. Keblinski, Phys. Rev. B 2002, 65, 144306; b) J. Lee, J. Lim, P. Yang, Nano Lett. 2015, 15, 3273. [18] F. Yang, C. Dames, Phys. Rev. B 2013, 87, 035437. [19] G. Wang, C. Wang, J. Zhao, G. Wang, C. B. Park, G. Zhao, Nanoscale 2017, 9, 5996. [20] Y. S. Touloukian, P. E. Liley, S. C. Saxena, Thermophysical Properties of Matter – The TPRC Data Series. Vol. 3. Thermal 1801001 (15 of 17) © 2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.advancedsciencenews.com [21] [22] [23] [24] [25] [26] [27] [28] [29] [30] [31] [32] [33] [34] [35] [36] [37] [38] www.advmat.de Conductivity – Nonmetallic Liquids and Gases, Defense Technical Information Center, Fort Belvoir, VA, USA 1970. P. B. Allen, J. L. Feldman, J. Fabian, F. Wooten, Philos. Mag. B 1999, 79, 1715. a) L. R. Glicksman, Low Density Cellular Plastics: Physical Basis of Behaviour, Springer, Dordrecht, The Netherlands 1994; b) E. Placido, M. C. Arduini-Schuster, J. Kuhn, Infrared Phys. Technol. 2005, 46, 219. a) X. Xie, K. Yang, D. Li, T. Tsai, J. Shin, P. V. Braun, D. G. Cahill, Phys. Rev. B 2017, 95, 035406; b) S. Kommandur, S. K. Yee, J. Polym. Sci., Part B: Polym. Phys. 2017, 55, 1160. a) W. Hsieh, M. D. Losego, P. V. Braun, S. Shenogin, P. Keblinski, D. G. Cahill, Phys. Rev. B 2011, 83, 174205; b) X. Xie, D. Li, T. Tsai, J. Liu, P. V. Braun, D. G. Cahill, Macromolecules 2016, 49, 972. C. Bi, G. H. Tang, Int. J. Heat Mass Transfer 2013, 64, 452. a) A. I. Cooper, Adv. Mater. 2003, 15, 1049; b) L. Chen, D. Rende, L. S. Schadler, R. Ozisik, J. Mater. Chem. A 2013, 1, 3837; c) S. Costeux, J. Appl. Polym. Sci. 2014, 131, 41293. a) R. J. J. Williams, C. M. Aldao, Polym. Eng. Sci. 1983, 23, 293; b) C. Wong, M. Hung, J. Cell. Plast. 2008, 44, 239; c) J. Schellenberg, M. Wallis, J. Cell. Plast. 2010, 46, 209; d) C. V. Vo, F. Bunge, J. Duffy, L. Hood, Cell. Polym. 2011, 30, 137; e) M. Arduini-Schuster, J. Manara, C. Vo, Int. J. Therm. Sci. 2015, 98, 156. a) J. Wu, H. Chu, Heat Mass Transfer 1998, 34, 247; b) J. Wu, W. Sung, H. Chu, Int. J. Heat Mass Transfer 1999, 42, 2211. a) J. A. Martínez-Díez, M. A. Rodríguez-Pérez, J. A. de Saja, L. O. A. Rabago, O. A. Almanza, J. Cell. Plast. 2001, 37, 21; b) M. Alvarez-Lainez, M. A. Rodríguez-Pérez, J. A. de Saja, J. Polym. Sci., Part B: Polym. Phys. 2008, 46, 212. G. Wang, J. Zhao, G. Wang, L. H. Mark, C. B. Park, G. Zhao, Eur. Polym. J. 2017, 95, 382. G. Wang, G. Zhao, G. Dong, Y. Mu, C. B. Park, G. Wang, Eur. Polym. J. 2018, 103, 68. B. Zhao, C. Zhao, C. Wang, C. B. Park, J. Mater. Chem. C 2018, 6, 3065. a) M. Okamoto, P. H. Nam, P. Maiti, T. Kotaka, T. Nakayama, M. Takada, M. Ohshima, A. Usuki, N. Hasegawa, H. Okamoto, Nano Lett. 2001, 1, 503; b) Y. H. Kim, S. J. Choi, J. M. Kim, M. S. Han, W. N. Kim, K. T. Bang, Macromol. Res. 2007, 15, 676; c) T. U. Patro, G. Harikrishnan, A. Misra, D. V. Khakhar, Polym. Eng. Sci. 2008, 48, 1778; d) S. Estravís, J. Tirado-Mediavilla, M. Santiago-Calvo, J. L. Ruiz-Herrero, F. Villafañe, M. A. Rodríguez-Pérez, Eur. Polym. J. 2016, 80, 1. Q. Li, L. Chen, X. Li, J. Zhang, K. Zheng, X. Zhang, X. Tian, J. Appl. Polym. Sci. 2016, 133, 43765. a) J. Yeh, K. Chang, C. Peng, B. Chand, S. Chiou, H. Huang, C. Lin, J. Yang, H. Lin, C. Chen, Polym. Compos. 2009, 30, 715; b) S. Costeux, L. Zhu, Polymer 2013, 54, 2785. a) G. Harikrishnan, S. N. Singh, E. Kiesel, C. W. Macosko, Polymer 2010, 51, 3349; b) P. Gong, P. Buahom, M. Tran, M. Saniei, C. B. Park, P. Pötschke, Carbon 2015, 93, 819; c) P. Gong, G. Wang, M. Tran, P. Buahom, S. Zhai, G. Li, C. B. Park, Carbon 2017, 120, 1. a) G. Wang, J. Zhao, L. H. Mark, G. Wang, K. Yu, C. Wang, C. B. Park, G. Zhao, Chem. Eng. J. 2017, 325, 632; b) J. Zhao, Q. Zhao, C. Wang, B. Guo, C. B. Park, G. Wang, Mater. Des. 2017, 131, 1; c) J. Zhao, Q. Zhao, L. Wang, C. Wang, B. Guo, C. B. Park, G. Wang, Eur. Polym. J. 2018, 98, 1. a) S. N. Patankar, Y. A. Kranov, Mater. Sci. Eng., A 2010, 527, 1361; b) Y. Liao, X. Wu, Z. Wang, R. Yue, G. Liu, Y. Chen, Mater. Chem. Phys. 2012, 133, 642; c) M. Fuji, C. Takai, H. Watanabe, K. Fujimoto, Adv. Powder Technol. 2015, 26, 857; d) L. Ernawati, T. Ogi, R. Balgis, K. Okuyama, M. Stucki, S. C. Hess, W. J. Stark, Langmuir 2016, 32, 338; e) Y. Qiao, X. Wang, X. Zhang, Z. Xing, J. Reinf. Plast. Compos. 2015, 34, 1413; f) Z. Wang, T. Zhang, B. K. Park, W. I. Lee, D. J. Hwang, J. Mater. Sci. 2017, 52, 6726. Adv. Mater. 2019, 31, 1801001 [39] a) J. Liang, F. Li, Polym. Test. 2006, 25, 527; b) X. Lu, G. Xu, P. G. Hofstra, R. C. Bajcar, J. Polym. Sci., Part B: Polym. Phys. 1998, 36, 2259; c) B. Li, J. Yuan, Z. An, J. Zhang, Mater. Lett. 2011, 65, 1992; d) G. P. Srivastava, Mater. Res. Soc. Symp. Proc. 2011, 1172, T08. [40] H. Woo, K. Char, Macromol. Res. 2013, 21, 1004. [41] a) Y. Bao, Q. Kang, J. Inorg. Mater. 2017, 32, 581; b) Q. Kang, Y. Bao, M. Li, J. Ma, Prog. Org. Coat. 2017, 112, 153; c) Y. Bao, Q. Kang, J. Ma, Colloids Surf., A 2018, 537, 69. [42] X. Zhao, C. Zang, Y. Sun, Y. Zhang, Y. Wen, Q. Jiao, J. Appl. Polym. Sci. 2018, 135, 46025. [43] a) Y. Agari, T. Uno, J. Appl. Polym. Sci. 1985, 30, 2225; b) H. Zhou, S. Zhang, M. Yang, Compos. Sci. Technol. 2007, 67, 1035; c) J. Liang, F. Li, Polym. Test. 2007, 26, 419; [44] P. Ruckdeschel, A. Philipp, B. A. F. Kopera, F. Bitterlich, M. Dulle, N. W. Pech-May, M. Retsch, Phys. Rev. E 2018, 97, 022612. [45] Y. Hu, R. Mei, Z. An, J. Zhang, Compos. Sci. Technol. 2013, 79, 64. [46] a) Y. Liao, X. Wu, H. Liu, Y. Chen, Thermochim. Acta 2011, 526, 178; b) R. V. Rivera-Virtudazo, R. T. Wu, T. Mori, MRS Adv. 2016, 1, 3947; c) R. T. Wu, R. V. Rivera-Virtudazo, T. Mori, MRS Adv. 2016, 1, 3965; d) T. Gao, B. P. Jelle, L. I. C. Sandberg, A. Gustavsen, ACS Appl. Mater. Interfaces 2013, 5, 761; e) P. Ruckdeschel, T. W. Kemnitzer, F. A. Nutz, J. Senker, M. Retsch, Nanoscale 2015, 7, 10059; f) P. Ruckdeschel, A. Philipp, M. Retsch, Adv. Funct. Mater. 2017, 27, 1702256; g) P. Ruckdeschel, M. Retsch, Adv. Mater. Interfaces 2017, 4, 1700963; h) Z. Jia, Z. Wang, D. Hwang, L. Wang, ACS Appl. Energy Mater. 2018, 1, 1146. [47] D. G. Cahill, R. O. Pohl, Phys. Rev. B 1987, 35, 4067. [48] a) Q. Yue, Y. Li, M. Kong, J. Huang, X. Zhao, J. Liu, R. E. Williford, J. Mater. Chem. 2011, 21, 12041; b) C. Mille, R. W. Corkery, J. Mater. Chem. A 2013, 1, 1849. [49] a) R. J. Corruccini, Vacuum 1959, 7–8, 19; b) L. Su, L. Miao, S. Tanemura, G. Xu, Sci. Technol. Adv. Mater. 2012, 13, 035003. [50] S. S. Kistler, Nature 1931, 127, 741. [51] a) L. W. Hrubesh, R. W. Pekala, J. Mater. Res. 1994, 9, 731; b) N. Hüsing, U. Schubert, Angew. Chem., Int. Ed. 1998, 37, 22; c) A. C. Pierre, G. M. Pajonk, Chem. Rev. 2002, 102, 4243; d) M. A. Aegerter, N. Leventis, M. M. Koebel, Aerogels Handbook, Springer, New York 2011; e) J. P. Randall, M. A. B. Meador, S. C. Jana, ACS Appl. Mater. Interfaces 2011, 3, 613; f) R. Baetens, B. P. Jelle, A. Gustavsen, Energy Build. 2011, 43, 761; g) J. Schultz, K. Jensen, F. Kristiansen, Solar Energy Mater. Solar Cells 2005, 89, 275. [52] a) G. W. Scherer, J. Non-Cryst. Solids 1989, 109, 171; b) A. C. Pierre, Introduction to Sol–Gel Processing, Kluwer Academic Publishers, Boston, MA, USA 1998. [53] S. S. Kistler, A. G. Caldwell, Ind. Eng. Chem. 1934, 26, 658. [54] a) A. V. Rao, R. R. Kalesh, Sci. Technol. Adv. Mater. 2003, 4, 509; b) A. P. Rao, A. V. Rao, U. K. H. Bangi, J. Sol–Gel Sci. Technol. 2008, 47, 85; c) A. A. Pisal, A. V. Rao, J. Porous Mater. 2017, 24, 685. [55] T. Wei, T. Chang, S. Lu, Y. Chang, J. Am. Ceram. Soc. 2007, 90, 2003. [56] a) M. Venkataraman, R. Mishra, J. Militky, L. Hes, Fibers Polym.. 2014, 15, 1444; b) G. Wei, Y. Liu, X. Zhang, F. Yu, X. Du, Int. J. Heat Mass Transfer 2011, 54, 2355; c) G. Hayase, K. Kanamori, K. Abe, H. Yano, A. Maeno, H. Kaji, K. Nakanishi, ACS Appl. Mater. Interfaces 2014, 6, 9466; d) A. Demilecamps, M. Alves, A. Rigacci, G. Reichenauer, T. Budtova, J. Non-Cryst. Solids 2016, 452, 259; e) H. Zhang, W. Fang, X. Wang, Y. Li, W. Tao, Int. J. Heat Mass Transfer 2017, 115, 21; f) G. Luo, X. Gu, J. Zhang, R. Zhang, Q. Shen, M. Li, L. Zhang, J. Appl. Polym. Sci. 2017, 134, 44434. [57] a) B. Wicklein, A. Kocjan, G. Salazar-Alvarez, F. Carosio, G. Camino, M. Antonietti, L. Bergstrom, Nat. Nanotechnol. 2015, 10, 277; 1801001 (16 of 17) © 2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.advancedsciencenews.com [58] [59] [60] [61] www.advmat.de b) K. Guo, Z. Hu, H. Song, X. Du, L. Zhong, X. Chen, RSC Adv. 2015, 5, 5197; c) C. Yue, J. Feng, J. Feng, Y. Jiang, RSC Adv. 2016, 6, 9396; d) Y. Cheng, S. Zhou, P. Hu, G. Zhao, Y. Li, X. Zhang, W. Han, Sci. Rep. 2017, 7, 1439; e) K. H. Kim, S. Park, S. Hwang, M. D. Cho, Chem. Commun. 2012, 48, 386; f) S. Samitsu, Macromolecules 2018, 51, 151; g) D. Schmidt, V. I. Raman, C. Egger, C. du Fresne, V. Schädler, Mater. Sci. Eng., C 2007, 27, 1487. K. Zhang, A. Yadav, K. H. Kim, Y. Oh, M. F. Islam, C. Uher, K. P. Pipe, Adv. Mater. 2013, 25, 2926. a) X. Lu, O. Nilsson, J. Fricke, R. W. Pekala, J. Appl. Phys. 1993, 73, 581; b) V. Bock, O. Nilsson, J. Blumm, J. Fricke, J. Non-Cryst. Solids 1995, 185, 233; c) J. Feng, J. Feng, C. Zhang, J. Porous Mater. 2012, 19, 551. J. Hone, M. Whitney, A. Zettl, Synth. Met. 1999, 103, 2498. Y. Luo, C. Ye, Polymer 2012, 53, 5699. Adv. Mater. 2019, 31, 1801001 [62] a) D. M. Smith, A. Maskara, U. Boes, J. Non-Cryst. Solids 1998, 225, 254; b) H. Thorne-Banda, T. Miller, in Aerogels Handbook: Advances in Sol–Gel Derived Materials and Technologies (Eds: M. Aegerter, N. Leventis, M. Koebel), Springer, New York 2011. [63] a) R. Baetens, B. P. Jelle, J. V. Thue, M. J. Tenpierik, S. Grynning, S. Uvslokk, A. Gustavsen, Energy Build. 2010, 42, 147; b) U. Berardi, Appl. Energy 2015, 154, 603. [64] a) X.-D. Wang, D. Sun, Y.-Y. Duan, Z.-J. Hu, J. Non-Cryst. Solids 2013, 375, 31; b) G. Wei, Y. Liu, X. Zhang, X. Du, J. Non-Cryst. Solids 2013, 362, 231; c) J.-J. Zhao, Y.-Y. Duan, X.-D. Wang, B.-X. Wang, Int. J. Heat Mass Transfer 2012, 55, 5196; d) D. Lee, P. C. Stevens, S. Q. Zeng, A. J. Hunt, J. Non-Cryst. Solids 1995, 186, 285; e) R. Rettelbach, J. Sauberlich, S. Korder, J. Fricke, J. Phys. D: Appl. Phys. 1995, 28, 581; f) Y. G. Kwon, S. Y. Choi, E. S. Kang, S. S. Baek, J. Mater. Sci. 2000, 35, 6075. 1801001 (17 of 17) © 2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim