Methods
Ink composition and preparation
Polyethoxydisiloxane (PEDS) precursor^34. 173 ml Dynasylan 40 (ethyl
silicate with an SiO 2 content of 40–42 wt%, Evonik) was mixed with
189 ml 1-pentanol (abcr Schweiz) and 13.5 ml ultrapure water (double
distilled, >18 MΩ cm) at 35 °C. After stirring at 250 rpm for 10 min, the so-
lution was cooled to 25 °C, with continuous stirring at 250 rpm. A 0.06 M
HNO 3 aqueous solution was then injected dropwise (0.45 ml min−1) with
a syringe pump (LaboTechSystems). The silica precursor was stored at
4 °C for 24 h before use.
1-pentanol with pH indicator. 1-pentanol was selected as solvent to
limit solvent evaporation during printing; 0.1 wt% methyl orange indi-
cator (Sigma Aldrich) was added to determine the sol–gel transition.
Ramsdellite MnO 2 (R-MnO 2 ) microspheres^28. 10.04 g Mn(NO 3 ) 2 ·4H 2 O
(Sigma Aldrich) was dissolved in distilled water (50 wt%). Then, 100 ml
of a 3.06 wt% KMnO 4 aqueous solution was added dropwise and the
mixture was stirred at 60 °C for 12 h. The precipitate was filtered and
dried at 60 °C overnight.
Silica aerogel ink. The ink was prepared by first mixing 1-pentanol
(with pH indicator) with poly(propylene glycol) bis(2-aminopropyl
ether) (PPGNH; average Mn ≈ 4,000; Sigma-Aldrich) at room tempera-
ture (25 °C) for 5 min. PPGNH increases viscosity and prevents settling
of silica aerogel particles (Extended Data Fig. 2). Also, amino groups
from PPGNH improve the homogeneity of the gel structure during the
sol–gel transition (Extended Data Fig. 3). 12 M HCl (37%; Sigma-Aldrich)
was added dropwise until the colour changed from yellow to red. Then,
PEDS was added and mixed at 500 rpm for 5 min. Silica aerogel parti-
cles (amorphous, 5–20 μm; ENOVA, Cabot Aerogel), in mass ratios of
63–397 wt% with respect to the silica in PEDS (Extended Data Table 1),
were added to achieve the rheological properties required for direct
ink writing. The blend was mixed first by spatula, and then in a plan-
etary speed-mixer (DAC 150.1 FVZ; FlackTek) at 3,000 rpm for 5 min,
followed by de-foaming at 3,500 rpm for 2 min. For the R-MnO 2 -silica
inks, a fraction of the silica aerogel powder particles were replaced
with R-MnO 2 microspheres.
Printing procedure
The silica inks were loaded into syringe barrels and de-foamed at
2,500 rpm for 3 min to remove air bubbles. The inks were then mounted
to a direct ink writer from EnvisionTEC (Bioplotter) with conical noz-
zles (diameters of 100 μm, 250 μm, 410 μm and 1,190 μm; smoothflow
tapered tip, H. Sigrist & Partner). The inks were driven pneumatically
through the micronozzles at 0.3–4.0 bar, with a filament extrusion rate
of 12–18.4 mm s−1. Printing paths and STL files were generated by Mag-
ics Envisiontec 18.2, sliced, and converted into BPL files (Bioplotter RP
software package) to command the x–y–z motion of the printer head.
Solidification and drying
The solidification of the printed objects was achieved by a base-catalysed
sol–gel transition of the matrix sol (Fig. 1a). Printed objects were placed
in a closed polystyrene box together with a 5.5 M ammonia solution
(0.1 ml per 1 ml ink) that was not in direct contact with the objects. The
NH 3 gas atmosphere induces solidification. The gelation time depends
on diffusion, but typically ranges from 2 min (single-wall objects with
200–1,000 μm wall thickness) to 120 min (50 mm × 50 mm × 10 mm gel
blocks) at about 25 °C. After solidification (colour change from red to
yellow), ethanol (5% isopropanol, Alcosuisse) was used to cover the gels.
After solvent exchange into ethanol (≥99.5%), the matrix silica gel was
hydrophobized by soaking the printed objects in a 6-wt% hexamethyld-
isilazane (HMDZ) solution in ethanol at 25 °C for 24 h (20 ml of printed
gel into 100 ml of hydrophobization solution). Finally, the objects were
placed inside a SCF extractor (Separex), exchanged into liquid CO 2 over
48 h (61 bar, 23 °C), and supercritically dried from 180 bar and 48 °C.Rheology
The rheological properties of the inks at 25 °C were characterized using
a rotational rheometer (MCR502, Anton Paar) with a 50-mm-diameter
steel-plate geometry and a gap height of 0.5 mm. Apparent viscosities
were measured via steady-state flow experiments with a sweep of shear
rate (0.001–1,000 s−1). Shear storage and loss moduli were determined
as a function of shear stress via oscillation experiments at a fixed fre-
quency of 1 Hz with a sweep of stress (10–10,000 Pa). The aerogel inks
were equilibrated for 1 min before testing. The shelf-life was checked
after storage at 4 °C for 20 days, and the apparent viscosity and storage
moduli were tested and compared with fresh ink.Thermal conductivity
Two identical square planar boards (width, 50 mm; height, 10 mm)
were printed from ink SP1.6 with a conical nozzle diameter of 1,190 μm,
shown in Supplementary Video 5. After gelation and drying, the plates
were placed in a custom-built guarded hot-plate device for thermal
conductivity measurement (guarded zone, 50 mm × 50 mm; measuring
zone, 25 mm × 25 mm; 50% relative humidity, 25 °C)^35.Microstructural analysis
The dimension and shape of the filaments were imaged using opti-
cal (Leica DVM VZ 700C) and SEM, and the interface of the printed
objects was cut using a diamond saw. SEM images were recorded on a
FEI Nova NanoSEM 230 (FEI) at an accelerating voltage of 10 kV and a
working distance of roughly 5 mm. Nominally 15 nm of Pt (measured
with a flat piezo dector) was coated to avoid charging, but the effective
thickness on the aerogels, with their extreme topography and surface
area, will be much lower. For transmission electron microscopy (TEM),
printed aerogel objects were crushed and suspended in methanol. The
suspension was dispersed by ultrasound and dropped onto a lacey
Cu TEM grid (S166-2, Plano). High-resolution TEM (HRTEM) images
were recorded on a JEM-2200FS field-emission electron microscope
( JEOL) at 200 kV. Nitrogen sorption analysis was carried out on a TriFlex
nitrogen sorption analyser (Micromeritics) after degassing for 20 h
at 100 °C and 0.03 mbar. The specific surface areas (SBET; uncertainty
of around 20 m^2 g−1) were obtained using the BET method^36. The pore
volume (Vpore) and average pore diameter (d 0 ) were calculated from the
density of the printed aerogels and their SBET: Vpore = 1/ρ − 1/ρskeleton and
d 0 = 4Vpore/SBET, where ρ is bulk density and ρskeleton is the skeletal density.Synchrotron X-ray tomographic microscopy
A filament from ink SP1.3M0.9 printed through a 410-μm nozzle was
measured by tomography. This ink was selected because the MnO 2
microparticles distributed in the matrix sol–gel provide contrast
between the silica aerogel powder particles and the MnO 2 -loaded silica
aerogel matrix. Imaging was performed at the TOMCAT beamline of the
Swiss Light Source, situated on a 2.9-T bending magnet, and equipped
with a multilayer monochromator. X-ray images were acquired at 12 keV
and a propagation distance of 50 mm. The X-ray indirect detector com-
prised a LSO:Tb 5.8-μm scintillator, a 40× optical objective and a sCMOS
pco.EDGE camera (6.5 μm pixel size, 2,160 × 2,560 pixels), resulting in
an effective pixel size of 0.16 μm. During the continuous tomography
scan, 1,801 projections were collected over 180°, with an exposure
time of 150 ms per projection, as well as two series of 100 flats and a
series of 30 dark projections. The data were reconstructed using the
Gridrec algorithm^37.3D image analysis
Distinct phases are observed in the slices (Extended Data Fig. 4b), where
aerogel powder grains appear darkest. The MnO 2 -loaded silica aero-
gel matrix (binder phase) has a higher absorption and thus appears