Aerogel, a high performance material for thermal insulation

Aerogel, a high performance material for thermal insulation - A brief overview of the building applications

Larisa Meliță* and Cristiana Croitoru

Check now

E3S Web of Conferences, ()

Technical University of Civil Engineering of Bucharest, code . Lacul Tei Bvd. 122-124, Bucharest, Romania

* Corresponding author:

Abstract

In this paper data regarding the utilization of aerogel as a promising material for thermal insulation of the residential and commercial buildings are presented. Also, research work and developments in synthesis, properties and characterization of silica aerogels will be addressed. Aerogel is a synthetic porous ultralight material derived from a gel in which the liquid component of the gel has been replaced with a gas. The result is a solid with extremely low density and low thermal conductivity. Sol-gel is the most used method of preparation. Aerogel melts at ºC and the thermal conductivity is almost 0. Is a solid material with the smallest density because contains about 99.8% air. This material has almost unlimited potential, believing that they might find application in most human activities and areas. Aerogel insulation is a good choice because nearly neutralizes all three methods of heat transfer: convection, conduction and radiation. The resistance to convective transfer is given by the fact that air does not circulate in the material structure. The resistance to thermal transfer by conduction is given by the majority of gaseous components. If using a carbon based gel, a high resistance to radiation transfer is obtained. Therefore, the most used aerogel for thermal insulation is the silica aerogel with carbon as nanostructured material. The high price makes it currently inaccessible and less used material. But, inevitably, the aerogel will quickly become one of the most attractive materials in the future.

Sample synthesis

Ceramic nanofibrous aerogels were prepared using a turbulent-flow-assisted electrospinning (TFAE) method. On the basis of a typical fabrication of polyacetylacetonatozirconium (PAZ)50, we developed a stepwise dissolution process to synthesize the electrospinning precursor. First, the zirconium&#;silicon precursor sol, was produced from a mixture with a molar composition of PAZ:yttrium nitrate hexahydrate:triethoxysilane:solvent (methanol, ethanol or water)&#;=&#;(about 1.50&#;1.75):0.095:(about 0.77&#;0.92):(about 0.65&#;0.85). The mixture was stirred for 1&#;h to obtain a clear and golden-transparent solution. Then, polyethylene oxide (PEO, weight-averaged molecular mass Mw&#;=&#;1,) was added to the zirconium&#;silicon precursor sol with a sol:PEO weight ratio of 1,250:1 at a temperature of 60&#;°C, and stirred for 80&#;min to obtain the eventual electrospinning solution. After defoaming, the resulting solution was used directly for TFAE to obtain the as-spun ceramic nanofibres. We developed a coaxial TFAE system. Forty millilitres of acquired solution was fed into plastic syringes fixed in a microsyringe pump, and the spinning solution was injected through the 28G-nozzle spinneret with a speed of 2&#;ml&#;h&#;1. The electrified liquid droplet was stretched and elongated to nanofibres from the spinneret via the direct-current, constant-high-voltage (25&#;kV) power and the coaxially high-speed air flow (airflow velocity of 15&#;m&#;s&#;1), generating the ceramic nanofibres (Supplementary Figs. 5, 6). The as-spun nanofibres were collected on the aluminium collector at a distance (about 15&#;20&#;cm) from the nozzle. Then, we prepared the as-spun ceramic nanofibres with a zig-zag architecture by a controllable mechanical folding process (Supplementary Fig. 7). The resulting 3D zig-zag ZAGs were sintered to pre-crystallization in a Muffle furnace under flowing air to form the hypocrystalline ZAGs. The sintering process can be divided into two stages: slow heating at 2&#;°C&#;min&#;1 from room temperature to 1,100&#;°C; and maintaining at 1,100&#;°C for 60&#;min. Next, with a secondary sintering treatment (same as above conditions) to crosslink the ZAG units, we fabricated the well assembled ZAGs with a controllable density (about 15&#;55&#;mg&#;cm&#;3) and diverse shapes (Supplementary Fig. 7). Except as noted, all of the material and physical property investigations were performed using a sample with a density of 20&#;mg&#;cm&#;3. In addition, we fabricated the typical SiO2, zirconium dioxide and mullite nanofibrous aerogels using the same TFAE method, and the hBN aerogels were prepared through the 3D graphene template-assisted chemical vapor deposition method19.

Material characterization

The morphology was investigated by SEM and TEM on a ZEISS, Merlin Compact at 20&#;kV and a FEI, Tecnai G2 F30 with EDS at 300&#;kV. The elemental and structural analyses were carried out by XPS, XRD and Raman spectroscopy on a ThermoFisher, ESCALAB 250Xi, a Panalytical, X&#;Pert and a Renishaw, inVia-Reflex with a 532-nm laser. The mechanical properties of the ceramic aerogels were studied using an Instron universal testing machine with 100-N load cells at a loading rate of 2&#;mm&#;min&#;1. The weight of the samples was measured by a Sartorius analytical balance (BSA124S-CW) with an accuracy of 0.1&#;mg. The thermal-shock tests of the ceramic aerogels were carried on a homemade pneumatic system with a cold source (25&#;°C) and a hot source (1,000&#;°C). The thermal expansion coefficient of the ceramic aerogels was investigated using a Thermal dilatometer (L75VDLT, Linses). The thermal conductivity of the ceramic aerogels was measured by a steady-state thermal conductivity tester (DRPL-V, Xiangtan Xiangyi Instruments, heater at 75&#;1,350&#;°C and chiller at 20&#;°C in air). The infrared absorption spectra in the 0.25&#;2.5&#;μm range were recorded using an ultraviolet&#;visible near-infrared spectrometer (Lambda 950) with an integrating sphere unit and automated reflectance measurement unit (sample size of 30&#;×&#;30&#;×&#;1&#;mm3 and density of 20&#;mg&#;cm&#;3). The infrared images of ZAGs were recorded using a Flir A615 (&#;40&#;°C to 2,000&#;°C) infrared thermal camera on an optical platform. The capacitance change was measured by a Precision LCR digital bridge (VCA).

MD simulation

Design of hypocrystalline ceramics

We investigated the physical properties of crystal, amorphous and hypocrystalline ceramics at different scales by MD simulations. First, the ν of three types of bulk sample with 2,304&#;atoms were calculated by using the Voigt&#;Ruess&#;Hill51 formalism at 25&#;°C. To be specific, a crystal sample with a size of 2.64&#;×&#;2.64&#;×&#;3.59&#;nm3 was duplicated from a 24-atom unit cell. An amorphous sample with a size of 3.17&#;×&#;3.18&#;×&#;3.15&#;nm3 was generated by melting the crystal sample at 5,000&#;°C for 100&#;ps and then quenching to 25&#;°C at a cooling rate of 1&#;°C&#;ps&#;1. A hypocrystalline sample with a size of 2.81&#;×&#;3.36&#;×&#;3.21&#;nm3 was generated by duplicating a 576-atom unit cell of amorphous and crystal ceramics. Second, we performed thermal expansion simulations for short fibres of three different phases with a slenderness ratio of about 5, as shown in Fig. 1a. To be specific, a crystal sample with a size of 11.55&#;×&#;1.99&#;×&#;1.98&#;nm3 containing 3,888 atoms, an amorphous sample with a size of 10.91&#;×&#;2.11&#;×&#;1.94&#;nm3 containing 3,456 atoms, and a hypocrystalline sample with a size of 11.23&#;×&#;2.06&#;×&#;2.1&#;nm3 containing 3,316 atoms were used for the thermal expansion simulations. Starting from 25&#;°C, the temperature was increased to 1,300&#;°C at a heating rate of 5&#;°C&#;ps&#;1 and then equilibrated at 1,300&#;°C for 200&#;ps. A time step of 1&#;fs was used for the integration of the motion equations.

Goto bestlink to know more.

We further investigated the values of ν and α in the crystal, amorphous and hypocrystalline cells consisting of four crosslinked long fibres, as shown in Fig. 1a. Fibres of three different phases with a slenderness ratio of about eight were first generated, and each fibrous sample was then replicated and rotated into four crosslinked fibres, which together made up the square cells. Specifically, four crosslinked fibres made up a crystal cell with a size of 16.4&#;×&#;1.99&#;×&#;16.4&#;nm3 containing 18,792&#;atoms, an amorphous cell with a size of 16.58&#;×&#;2.11&#;×&#;16.58&#;nm3 containing 17,984&#;atoms, and a hypocrystalline cell with a size of 16.67&#;×&#;2.06&#;×&#;16.67&#;nm3 containing 17,398&#;atoms. The thermal expansion simulations were performed the same as those with the short fibres mentioned above. To perform the compression simulations, we fixed the atoms at the two ends of the cells in the longitudinal direction and applied the displacement over the fixed atoms with the compression rate of 1&#;×&#;109&#;s&#;1.

We increased the slenderness ratio of the fibres to about 20 and generated larger square cells with crystal, amorphous and hypocrystalline ceramics with various crystal:amorphous volume ratios. To be specific, four crosslinked fibres made up a crystal cell with a size of 45.0&#;×&#;1.99&#;×&#;45.0&#;nm3 containing 53,312&#;atoms, an amorphous cell with a size of 45.0&#;×&#;2.11&#;×&#;45.0&#;nm3 containing 54,038&#;atoms, a hypocrystalline cell (crystal:amorphous volume ratio of 1:1) with a size of 45.0&#;×&#;2.06&#;×&#;45.0&#;nm3 containing 50,514&#;atoms, a hypocrystalline cell (crystal:amorphous volume ratio of 1:2) with a size of 45.0&#;×&#;2.06&#;×&#;45.0&#;nm3 containing 50,289&#;atoms, and a hypocrystalline cell (crystal:amorphous volume ratio of 2:1) with a size of 45.0&#;×&#;2.06&#;×&#;45.0&#;nm3 containing 49,833&#;atoms (Supplementary Fig. 1a). We performed the compression and thermal expansion simulations in triangle, pentagon and hexagon cells with a crystal:amorphous volume ratio of 1:1. A triangle cell with a size of 45.0&#;×&#;2.06&#;×&#;39.79&#;nm3 containing 37,603&#;atoms, a pentagon cell with a size of 67.04&#;×&#;2.06&#;×&#;62.9&#;nm3 containing 62,848&#;atoms, and a hexagon cell with a size of 78.16&#;×&#;2.06&#;×&#;71.14&#;nm3 containing 75,165&#;atoms were used for the calculation (Supplementary Fig. 1b).

Crystallization simulations

We performed the WTMetaD simulations for 100&#;ns under temperatures ranging from 1,300&#;°C to 1,700&#;°C. To distinguish the states of tetragonal zirconium dioxide and amorphous zircon, we chose two different collective variables with the number of bridging oxygens as \({s}_{1}\) and the dot product of the local Steinhardt&#;s order parameter52,53,54 \({q}_{4}^{{\rm{dot}}}\) of zirconium as \({s}_{2}\). We started the WTMetaD simulations with 576 atoms of amorphous zircon with a molar ratio of Zr:Si:O&#;=&#;2:1:6. The isothermal&#;isobaric MD simulations was performed using a Nose&#;Hoover thermostat and barostat55,56. The short-range interaction for WTMetaD simulations was described by the Buckingham form potential57 with a cut-off distance of 8&#;Å. For the long-range Coulombic term, the cut-off was set to 10&#;Å with the Ewald sum method. The bias factor \(\gamma \) of the WTMetaD was set to 50. The Gaussians with a height of 40&#;kJ&#;mol&#;1 and width of 2 and 0.04&#;collective variable units for \({s}_{1}\) and \({s}_{2}\) were introduced to construct the bias potential of the WTMetaD. All the simulations were carried out with the Large-scale Atomic/Molecular Massively Parallel Simulator (LAMMPS)58 code, and the WTMetaD simulations were carried out with an additional plugin code59 PLUMED 2. We constructed the model of complete hypocrystalline zircon by adding carbon atoms to the system obtained from WTMetaD at 1,700&#;°C, which showed the highest crystallinity from the local-entropy-based fingerprint60,61 \(\bar{{\rm{s}}}\). We performed the rest of the MD simulations of hypocrystalline zircon using the ReaxFF reactive force field62,63. The ReaxFF force field was constructed by combining the parameters of the C/Zr/O force field64 and the parameters of the C/Si/O force field65.

Mechanical properties

We duplicated the system from a 576-atom cell into a 2.67&#;×&#;3.25&#;×&#;3.47&#;nm3 sample with 2,304&#;atoms and replaced atoms inside a spherical region (radius of 0.2&#;nm) in amorphous zircon part with 20 randomly distributed carbon atoms. The 2,315-atom sample was further duplicated to a 2.67&#;×&#;3.25&#;×&#;111.12&#;nm3 fibrous sample with 74,080&#;atoms. We first performed bending simulation of the 74,080-atom fibrous sample with slenderness ratio of about 30. The fibre was compressed to the maximum strain of 60% and recovered to its original length. We then performed fracture simulation of the fibre sample along the c coordinate axis. The tensile strength was obtained by increasing the box size \({L}_{{\rm{c}}}\) of the system stepwise by 0.1% of the initial value \({L}_{{\rm{c}}0}\) until the fracture of the fibre. The tensile stress along the c axis was then computed using the virial theorem66.

Thermal properties

We first performed α simulation for the 74,080-atom fibrous sample and calculated the α by MD simulation. The temperature was increased stepwise by 50&#;°C from 0&#;°C to 400&#;°C at a heating rate of 5&#;°C&#;ps&#;1, and relaxed at each step for 20&#;ps to obtain a statistical average of α. We then performed two sets of non-equilibrium MD simulations and calculated κ using a direct method to show the reduction of κ in hypocrystalline zircon caused by the atomic arrangement and the crosslinking pattern. We calculated κ of bulk single-crystal zircon and the hypocrystalline zircon samples. To be specific, for crystal zircon, we duplicated the unit cell of zircon into two 6,144-atom samples with a size of 10.57&#;×&#;2.64&#;×&#;2.39&#;nm3 and 2.64&#;×&#;2.64&#;×&#;9.56&#;nm3 for the calculation of κ along the a and c directions. For hypocrystalline zircon, we replaced atoms inside a spherical region (radius of 0.2&#;nm) in the amorphous zircon part (576 atoms) with 20 randomly distributed carbon atoms, and then duplicated the hypocrystalline zircon to three samples with sizes of 10.42&#;×&#;2.67&#;×&#;3.25&#;nm3, 3.47&#;×&#;10.68&#;×&#;1.62&#;nm3 and 3.47&#;×&#;2.67&#;×&#;9.72&#;nm3 to obtain κ along the a, b and c directions. Hot and cold regions were set vertical with the longitudinal direction and located at two sides of the system with a thickness of 2&#;Å. Furthermore, we calculated κ of two nanofibrous samples overlapping vertically, compared with samples placed side by side. We used two 14,088-atom fibrous systems with the same size of 2.67&#;×&#;3.25&#;×&#;20.84&#;nm3 duplicated from a 587-atom hypocrystalline system. Hot and cold regions were set at the middle of each fibre along the longitudinal axis with a thickness of 2&#;Å.

Carbon retainment

The reactive MD simulations were performed in a canonical ensemble, and a Berendsen thermostat67 was used to control the temperature of the system. Periodic conditions were used along each boundary of simulation box. A time step of 0.25&#;fs was used for the integration of the motion equations. First, we generated a sample containing 288 randomly distributed atoms with a molar ratio of Zr:Si:O&#;=&#;1:2:6 and melted the sample at 5,000&#;°C for 100&#;ps to ensure that the atoms were fully relaxed. The sample was then quenched to 25&#;°C at a cooling rate of 1&#;°C&#;ps&#;1. The sample was duplicated to cubes with different lengths of 2&#;nm, 3&#;nm and 4&#;nm. The zircon atoms inside a cylinder region in the centre of the cubes with radius of 0.5&#;nm were replaced with 50, 75 and 100 randomly distributed carbon atoms, respectively. Furthermore, 100, 225 and 400 oxygen atoms were placed randomly in the oxygen regions (space of 2&#;nm above and below the amorphous zircon region) in the three systems, respectively. In contrast, for the hypocrystalline zircon sample, we generated a quarter of a cylinder of tetragonal zirconium dioxide with a radius of 1.5&#;nm at each corner of the cubic amorphous zircon. We performed 10&#;ps of simulation at a temperature of 25&#;°C for each system and then sintered at a temperature of 1,100&#;°C for 500&#;ps.

Additional details of the MD simulation can be found in the Supplementary Information.

FE simulation

In the FE simulations, we designed a zig-zag architecture in the aerogel to trigger an additional high-order deformation at the macroscale to mitigate the structure-derived increase of ν and α to achieve the near-zero responses (Fig. 1b, Supplementary Fig. 2). In geometric morphologies, a 2D porous zig-zag architecture was constructed by laminated fibrous layers and corner joints with high-order deformation fibrous cell as building blocks. Three shapes of cells (triangle, pentagon and hexagon) were used to assemble the fibrous aerogel with zig-zag architectures. To simplify the simulation process, a 2D FE model was established and the characteristics of fibrous units were simulated using a beam element with the geometric parameter set as 13:1 for the slenderness ratio. For ν, compression loadings were added by a displacement of the top layer at 5% strain with the model fixed at the bottom. For α, the temperature was gradually increased from 0&#;°C to 400&#;°C to investigate the thermally excited response. The material attributes were set similar to those of the hypocrystalline ceramics from MD simulations.

Computational fluid dynamics simulation

In the numerical simulation, a large eddy simulation with kinetic-energy transport subgrid-scale model was adopted to solve the Navier&#;Stokes equation, which obtains 3D unsteady flow motions. The computational domain, boundary conditions and mesh are shown in Supplementary Fig. 3. The coaxially circular nozzle was 30&#;mm in length with an inner diameter of 5&#;mm. The dimension of the collector was 560&#;×&#;480&#;×&#;450&#;mm3. The pressure inlet with a gauge total pressure of 10&#;kPa was assigned to the coaxially circular nozzle, and no-slip boundary conditions were used for the other surfaces. The simulations were carried out using commercial software Fluent Solver  R1. The finite-volume method was used in the solver, and the pressure&#;velocity coupling was achieved with the pressure implicit with the splitting of operators method. Least-squares cell-based, bounded central differencing and second-order upwind schemes were adopted to discretize the gradient, momentum and subgrid kinetic energy terms, respectively. The second-order implicit scheme was adopted for the temporal discretization. The time step was adaptive with a courant number of 1. Supplementary Fig. 4 presents the instantaneous turbulent structures and fibrous streamlines of the 3D turbulent flow. The turbulent structures were obtained by the Lambda2 (λ2) criterion.

If you are looking for more details, kindly visit Aerogel Thermal Insulation.