Article
brighter. There are also distinct bright areas, probably related to MnO 2
aggregations. The size of these MnO 2 -enriched areas ranges from the
resolution limit of around 1 μm up to around 60 μm, but these areas
are not of primary interest because they do not exist in the MnO 2 -free
printed filaments. A volume (203 μm × 118 μm × 234 μm) free of large
MnO 2 -enriched areas was selected for image analysis.
The grey values were deconvoluted in three peaks, corresponding, in
order of increasing grey value, to silica aerogel particle grains, the silica
aerogel matrix with finely dispersed MnO 2 , and MnO 2 -enriched areas
(Extended Data Fig. 4a). A normal distribution of grey values within a
phase was assumed, and the fit was done using the nl2sol library^38 in
R^39. The position of the grey-value distribution of the MnO 2 -enriched
areas was restricted to be higher than for the binder, and the weight was
restricted to no more than twice the starting value, on the basis of the
remainder after peak fitting of only two peaks up to a grey value of 2,000.
For the majority of the binder and the MnO 2 -enriched phases, no
constant contrast is reached within the phase. Instead, the grey value
has a peak across the binder phase when plotted as a profile (Extended
Data Fig. 4b). This indicates that most of the binder structure is at the
edge of the resolution limit and the contrast gets smeared out over a
larger area. To counteract this effect, the threshold grey value for the
matrix aerogel (17,839) and MnO 2 -enriched phase (22,860) was chosen
such that, according to the phase deconvolution, the probability of a
voxel (pixel) being the binder (MnO 2 ) phase was more than 90%. Before
applying the threshold, a Gaussian blur with a width of 2 voxels or pixels
was applied. After applying the threshold, the matrix aerogel and the
MnO 2 -enriched phase were eroded twice to remove isolated voxels or
pixels and further counteract the smeared out contrast. An example
of the resulting segmentation is shown in Extended Data Fig. 4b. The
final segmented image consists of 57.9% SiO 2 particles, 40.5% matrix
aerogel and 1.6% MnO 2 -enriched areas. The final segmentation and
visualization were performed using GeoDict (Math2Market).
FTIR and NMR
FTIR spectra (400–4,000 cm−1) were measured on a Bruker Tensor 27
spectrometer in attenuated total reflectance mode, using a diamond
crystal, and corrected for the background signal. Solid-state NMR spec-
tra were acquired with a Bruker Avance III spectrometer equipped with
a 9.4-T magnet, corresponding to Larmor frequencies of 400.2 MHz
for^1 H, 100.6 MHz for^13 C and 79.5 MHz for^29 Si. Spectra were collected
in 7-mm zirconia rotors with magic-angle spinning, with a spin rate of
4 kHz.^1 H–^13 C and^1 H–^13 Si cross-polarization spectra were collected,
with contact times of 2 ms and 5 ms, respectively.
X-ray diffraction analysis
The X-ray diffraction analyses of the MnO 2 and MnO 2 –SiO 2 composite
were recorded on a PANalytical X’Pert PRO diffractometer equipped
with a Johansson monochromator (Cu Kα1 radiation, λ = 1.5406 Å).
Thermal stability
Differential thermogravimetry analysis was conducted on a TGA7 ana-
lyser (Perkin Elmer).
Contact angle
The surface wettability by water and 1-pentanol were evaluated using
a contact-angle measurement system, OCA (Dataphysics TBU 90E).
Liquid droplets (5 μl) were deposited either on a packed bed of silica
aerogel powder or on the surface of the printed membrane. Three
measurements were performed and averaged.
Mechanical testing
Three identical cylinders (diameter, 15 mm; height, 26 mm) were pre-
pared by casting ink SP1.6 in the cylindrical polystyrene boxes. The
processed aerogel cylinders were polished to even the surfaces for
uniaxial compression and Brazilian tensile testing. The specimens
were tested on a mechanical testing machine (Z010, Zwick/Roell) with a
2-kN force transducer (KAP-S, AST Gruppe) at a rate of 1 mm min−1. The
tensile strength σT was calculated from the sample geometry (diameter
D and height H) and Brazilian test compressive force F: σT = F/[π(D/2)H].Infrared thermal imaging and thermal management
The thermal insulation performance was evaluated using an infrared
camera (TH 3102 MR, NEC-San-ei, Japan) equipped with a Stirling-cooled
HgCdTe detector, with a temperature sensitivity of 0.08 at 30 °C and an
accuracy of ±0.5 °C. The emission was set to 1. Thermal images were ana-
lysed on a PicWin-IRIS system (version 7.3). A thermal management array
(Fig. 3b) was printed with a 410-μm nozzle. Practical demonstrations
were carried out on a Raspberry Pi 1 Model A+ circuit board (Fig. 3f). The
first demonstration was designed to illustrate anti-scalding of a heating
component. As a control, the voltage-controller high-power circuit
was imaged after 1 h of operation. Then, an aluminium thermal sink
(70 μm thick, including conductive adhesive, sprayed with an infrared
anti-reflective coating) was applied on the voltage controller to remove
heat, and imaged after 1 h of operation. Finally, a 3D-printed aerogel
cap (ink SP2.5, 410-μm nozzle) was placed on top and the tempera-
ture was imaged after 1 h of operation. The second demonstration was
designed to protect a thermally sensitive component. For reference,
a resistive heater was placed 5 mm above a capacitor (tantalum type A
476), with a T-type microthermocouple attached on top. After 30 min
of stabilization, the gap distance was reduced to 1 mm for 10 s. Finally,
the heater was placed directly on the capacitor and thermocouple. To
benchmark the aerogel performance, a 3D-printed aerogel cap (ink
SP2.5, 410-μm nozzle) and a polystyrene foam cap cut to size, both with
a nominal thickness of 1.2 mm, were placed on top of the capacitor, and
temperature was recorded using the same protocol.Gas pumping and VOC degradation setup
The micropump demonstrated here relies on thermal transpiration
through a porous silica aerogel membrane (Fig. 4a, b). Thermal tran-
spiration refers to the flow of gas molecules from the cold to the hot
side of a channel subjected to a temperature gradient, when the gas
flow is dominated by molecular flow^40 ,^41. The heat is generated by the
optical MnO 2 absorber layer under a 300-W halogen lamp. For thermal
transpiration to be substantial, the Knudsen number Kn = lm/d 0 needs
to be roughly 1–10, where lm is the mean free path of gas molecules and
here d 0 is the pore chord length (which is sometimes approximated by
average pore diameter d 0 above). Here lm is calculated from the
temperature (Tavg), pressure (P) and gas particle diameter (dg), as
lkmB=/Tdavgg(2π)^2 P, where kB is the Boltzmann constant and dg = 3.7 1 1 ×
10 −10 m for nitrogen gas^42. Kn = 2.04 for the aerogel membrane (Fig. 4 ).
The thermal transpiration and VOC degradation were tested in a
stainless-steel reactor (Extended Data Fig. 9). The bilayer MnO 2 –SiO 2
aerogel composite was printed onto a glass fibre sheet (LydAir; 0.36 mm
thickness) and the dried sample was sealed between two compartments
of the reactor. The upper compartment contains a window for light
irradiation on the MnO 2 absorber. The gas flow from the bottom part of
the reactor to the outlets in the upper part is driven by the temperature
gradient generated accross the aerogel insulator. The gas flow was
monitored using a mass-flow meter (MFC, AALBORG TIO Totalizer).
The VOC degradation of the functional micropump was monitored
in a closed system containing synthetic air with 25 ppm toluene. First,
toluene was introduced into the reactor and circulated using an exter-
nal pump to reach a steady-state condition. After that, the sample was
irradiated by a halogen lamp (300 W). The light intensity on the sample
surface was 344 mW cm−2 (S170C microscope power sensor and Thor-
labs PM100USB power/energy meter) and the resulting temperature
increase generated a gas flow through the MnO 2 -loaded catalytic layer
where toluene was degraded. The concentration of toluene was moni-
tored by gas chromatography (GC/FID). The surface temperature was
monitored by a thermocouple (type K, NiCr-Ni).