ARTICLE
All-day fresh water harvesting by microstructured
hydrogel membranes
Ye Shi
1
✉
, Ognjen Ilic
1,2
, Harry A. Atwater
1
& Julia R. Greer
1
✉
Solar steam water puri
fi
cation and fog collection are two independent processes that could
enable abundant fresh water generation. We developed a hydrogel membrane that contains
hierarchical three-dimensional microstructures with high surface area that combines both
functions and serves as an all-day fresh water harvester. At night, the hydrogel membrane
ef
fi
ciently captures fog droplets and directionally transports them to a storage vessel. During
the daytime, it acts as an interfacial solar steam generator and achieves a high evaporation
rate of 3.64 kg m
−
2
h
−
1
under 1 sun enabled by improved thermal/vapor
fl
ow management.
With a homemade rooftop water harvesting system, this hydrogel membrane can produce
fresh water with a daily yield of ~34 L m
−
2
in an outdoor test, which demonstrates its
potential for global water scarcity relief.
https://doi.org/10.1038/s41467-021-23174-0
OPEN
1
Division of Engineering and Applied Science, California Institute of Technology, Pasadena, CA, USA.
2
Department of Mechanical Engineering, University of
Minnesota, Minneapolis, MN, USA.
✉
email:
yeshi119@utexas.edu
;
jrgreer@caltech.edu
NATURE COMMUNICATIONS
| (2021) 12:2797 | https://doi.org/10.1038/s41467-021-23174-0 | www.nature.com/naturecommunications
1
1234567890():,;
W
ater scarcity is among the most serious global chal-
lenges of our time, and signi
fi
cant efforts have been
dedicated to harvesting fresh water from alternative
sources
1
–
3
. For example, interfacial solar steam generation utilizes
sunlight as energy source to purify saline or contaminated water
by directly heating water and driving its evaporation at the water-
air interface
4
–
10
. Its ef
fi
ciency depends on water transport and
thermal management, and various materials have been studied for
this application. For example, nanostructured carbon materials
have been designed to absorb light and facilitate water
transport
11
–
13
, and plasmonic materials
14
,
15
and ceramics
5
were
used to ef
fi
ciently convert sunlight to thermal energy. Yu et al.
demonstrated a polyvinyl alcohol (PVA)/polypyrrole (PPy)
hydrogel-based solar steam generator to have a vapor generation
rate of 3.2 kg m
−
2
h
−
1
enabled by expedited water transport
through porous matrix and a reduced water vaporization
enthalpy within the polymeric mesh
16
,
17
. Recently they
improved the rate to a high record of ~3.6 kg m
−
2
h
−
1
by
modifying the hydrogel systems with highly hydratable polymers
or light-absorbing
fi
llers
18
,
19
. All of these solar steam
generators have major drawbacks in that they can only work
under suf
fi
cient solar irradiation, and their output is limited by
the solar energy density at the earth
’
s surface and the size of
energy consumption required for water evaporation. The merit of
these materials will be greatly improved if they can harvest other
fresh water resources and continuously produce clean water
around the clock.
Fog frequently occurs in the coastal and post-sunset arid areas,
and is a source of water that is complementary to solar water
puri
fi
cation
20
–
22
. Fog collection presents a promising and low-
cost approach to water harvesting, which has been widely studied
and employed
23
–
25
. Polymer mesh materials are commonly used
to capture fog
26
–
28
, but their ef
fi
ciency is adversely affected by re-
entrainment of deposited droplets and clogging of the mesh with
pinned droplets
29
. Certain natural structures with distinctive
functions have been discovered that avoid these problems and
collect fog more ef
fi
ciently
30
–
32
. For example, the hierarchically
assembled conical structures of Cactus spine are able to con-
tinuously harvest fog by driving directional movement of
droplets
33
. Several bio-inspired fog collection motifs have been
explored in which devices are constructed with metals
34
, metal
oxides
35
and polymers
36
,
37
, all of which lack light-into-thermal
energy conversion ability, thus rendering them incompatible with
solar steam generation. Developing a structured material that
could support both technologies would provide an avenue
for exploiting both water collection mechanisms around the
clock and would have a signi
fi
cant impact on global water
scarcity relief.
We designed and fabricated a PVA/PPy hydrogel membrane
populated with three-dimensional (3D) tree-shaped surface
microstructures. Our choice of a hydrogel membrane stems from
its ability to serve as an effective interfacial solar steam generator
for water puri
fi
cation. Coupled with the excellent processability of
hydrogels and their compatibility with advanced manufacturing
techniques, these viscoelastic materials are easily shaped into
microstructures that can mimic biological systems at relevant
length scales to facilitate fog collection. When placed under
controlled fog generation conditions, this PVA/PPy gel mem-
brane ef
fi
ciently captures fog droplets at a rate of ~5.0 g cm
−
2
h
−
1
and drives droplet transport while providing directional control.
Using experiments and modeling, we also demonstrate that the
tree-shaped surface micro-topologies enable ampli
fi
cation of
thermal and
fl
uidic management for interfacial solar steam gen-
eration by maximizing light absorption ef
fi
ciency and guiding
vapor escape, thus enabling a high solar vapor generation rate of
3.64 kg m
−
2
h
−
1
under 1 sun irradiation. In outdoor tests, this
device is capable of all-day fresh water harvesting and delivers a
daily water collection rate of ~34 L m
−
2
.
Results
Design of PVA/PPy gel membrane with micro-tree array for
bi-functional water collection
. Figure
1
a is a schematic of the
fresh-water-collecting membrane. At night, the hydrogel mem-
brane is exposed to fog, and the surface microstructures con-
tinuously capture fog droplets and transport them to a storage
vessel. During the daytime, the hydrogel membrane acts as an
interfacial solar steam generator to purify saline or contaminated
water.
To develop this unique bifunctional water collection mem-
brane, PVA-based hydrogel was selected as the building material.
This material choice stems from its favorable solar steam
generation ability, water af
fi
nity, and processability. PVA
hydrogel provides hierarchically porous pathways within its
matrix for ef
fi
cient water transport and it reduces the evaporation
enthalpy of water owing to interactions between its hydroxyl
groups and water molecules, thus enabling high-performance
solar steam generation
16
. Its hydrophilic nature also favors water
capture on its surface. Hydrogel materials are compatible with
various processing techniques and can be easily shaped into
desired structures.
Though PVA hydrogel captures water ef
fi
ciently, a smooth
membrane surface inhibits its fog collection ability since captured
droplets will be pinned on its hydrophilic surface. To enable
optimally ef
fi
cient water collection from fog, the surface structure
needs to be modi
fi
ed to continuously remove deposited
droplets
38
,
39
. Cactus spine-inspired conical structures
were adopted for this purpose. Water droplets attached to the
sides of conical structures experience a Laplace pressure
difference,
Δ
P,
40
Δ
P
¼
dP
dz
Ω
¼
2
γ
ð
r
þ
R
0
Þ
2
sin
α
ð
1
Þ
where
Ω
is droplet volume,
γ
is surface tension,
r
is the local
radius,
R
0
is the droplet radius and
α
is the half apex angle. This
Laplace pressure difference drives droplets towards the wider
base, thus re-exposing the gel surface to more incoming vapor.
According to Eq.
1
, the apex angle in our design is the smallest
possible within the constraints of the fabrication process and
mechanical strength of PVA hydrogel to increase
Δ
P and causes
the droplets to move faster.
To increase the surface area and thus provide a bene
fi
t for both
fog capture and interfacial solar steam generation, we assembled
the gel cones in a hierarchical way by building branched small
cones on a cone trunk and then arrayed these tree-like structures
into a dense forest on a membrane surface. Light absorption is
improved in the gel forest and water droplets collected on
branches are able to merge together for quick drainage. The
density of these gel trees is also carefully tuned to facilitate the
escape of generated vapor during steam generation and water
drainage during fog collection.
To realize all-day water collection in natural environments, a
fl
oating prototype is built to support hydrogel membranes and
store collected water. As shown in Fig.
1
b, c, a foldable cover is
designed in our
fl
oating prototype. During night, it
’
s open and the
gel samples can be supported to face the fog
fl
ow. During
daytime, it
’
s closed and acts as a re-condensation structure.
The micro-tree array structure was designed in CAD software
(Supplementary Fig. 1) and fabricated on a PVA/PPy gel
membrane using stereolithography 3D Printing, followed by a
simple molding method (Fig.
2
a). The photomicrographs in
Fig.
2
b illustrate a typical gel membrane with a projected area
ARTICLE
NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-23174-0
2
NATURE COMMUNICATIONS
| (2021) 12:2797 | https://doi.org/10.1038/s41467-021-23174-0 | www.nature.com/naturecommunications
(A
p
) of ~5.5 cm
2
, which contains 100 hexagonally arranged
micro-trees on a supporting layer. Each tree is ~4 mm tall, has a
bottom diameter of ~0.8 mm, and contains nine 45-degree tilted
conical branches at 1/3, 1/2, and 2/3 of the tree height (Fig.
2
c).
All the branched cones have the same conicity as the trunk.
Scanning electron microscope images reveal that the smallest
dimension of the conical structure is ~20 μm at the tip (Fig.
2
d).
The cross-linked hydrogel is hierarchically porous and contains
inter-dispersed PPy particles, which may be bene
fi
cial in enabling
ef
fi
cient water transportation within the matrix (Fig.
2
e and
Supplementary Fig. 2)
16
. The chemical composition and mechan-
ical properties of PVA/PPy hydrogels were also investigated by
FTIR spectroscopy and rheological measurements (Supplemen-
tary Figs. 3, 4a).
Fog harvesting properties of PVA/PPy gel membrane with
micro-tree array
. We conducted fog collection experiments to
quantify the water collection rate of microstructured PVA/PPy
gels (Supplementary Fig. 5). Our experiments demonstrate that
under a continuous fog
fl
ow generated by an ultrasonic humi-
di
fi
er, the micro-trees capture micro-sized water droplets that
quickly grow and coalesce with one another as they move towards
the cone base while new droplets continuously condense onto the
cones. As this process continues, droplets from different branches
merge together into a millimeter-size droplet, which is ultimately
collected into the beaker with the guide of the support layer
(Fig.
3
a and Supplementary Fig. 6, Supplementary Movie 1). This
cycle of fog droplets nucleation followed by their transport,
growth, and eventual drainage of the large water drops repeats
with an average period of ~20 s, which corresponds to a saturated
fog collection rate (
m
/
A
p
) of ~5.0 g cm
−
2
h
−
1
calculated using the
projected membrane area (Fig.
3
b). Note that our fully hydrated
hydrogel membranes can only collect water droplets in fog
through their surface. They are not able to condense or absorb
gaseous water in an environment with relative humidity from 50
to 90%.
We quanti
fi
ed the effect of conical geometries on water droplet
transport and fog collection rate by fabricating and testing similar
PVA/PPy gel membranes that contained equivalently spaced,
geometrically identical surface micro-topologies of cones and
cylinders, as well as
fl
at surfaces. Figure
3
c summarizes these
fi
ndings and reveals that the micro-tree array exhibits a 34%
higher fog collection rate than that of a
fl
at surface, the cone array
is 17% more ef
fi
cient, and the cylinder array is 29% lower, after
being normalized by total surface area. Since the directions of
cones were not a key factor in the directional movement of the
water drops
30
, the effect of conical geometries was further studied
by conducting systematic experiments on gel cone arrays with
different conicity (Fig.
3
d), which demonstrated that lower apex
angles resulted in faster water collection rates; for example, the
Fig. 1 Design of the bifunctional gel membranes and all-day water harvesting prototype. a
Conceptual representation of the PVA/PPy hydrogel
membrane with micro-topologies that is capable of 24-h fresh water harvesting.
b
,
c
Schematic illustration of nighttime (
b
) and daytime (
c
) modes of
fl
oating device for all-day water harvesting.
NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-23174-0
ARTICLE
NATURE COMMUNICATIONS
| (2021) 12:2797 | https://doi.org/10.1038/s41467-021-23174-0 | www.nature.com/naturecommunications
3
collection rate (normalized by total surface area) of cone array
increased by 14.7% when sin
α
decreased from 0.24 to 0.10
(Supplementary Fig. 7).
Figure
3
e compares fog collection rates for several representa-
tive polymer meshes, as well as a Cactus spine and reveals that the
areal ef
fi
ciency of the PVA/PPy gel micro-tree arrays is 115%
higher than that of double-layered Raschel mesh and 61% higher
than that of a Cactus stem. The PVA/PPy gel micro-tree arrays
also show the highest fog collection rates among all different
materials based on the mass of polymeric materials (Supplemen-
tary Fig. 8). We evaluated the long-term stability and durability of
micro-tree membranes by testing their structural integrity
(Supplementary Fig. 4b, c) and bi-functional water harvesting
properties for a twenty-month period in the lab. Figure
3
f shows
that the average fog collection rate, as well as solar vapor
generation rate of PVA/PPy gel micro-tree array was well
maintained after more than twenty-month storage.
Fog deposition and droplet transport are key processes that
determine the fog collection performance
37
,
41
. Nucleation of
water vapor and small water droplets is energetically more
favorable on hydrophilic surfaces than hydrophobic ones
30
,
42
,
43
.
An ideal fog collection structure should provide the enhanced
surface area with hydrophilic nature to maximize droplet
nucleation density
38
. Our design of micro-tree arrays is such
that its footprint area of 1 cm
2
corresponds to a total surface area
of ~3.5 cm
2
and increases the density of active sites for fog
capture and droplet nucleation by increasing surface area
(Supplementary Fig. 9
–
11). The contact angle of 65° revealed
the surface of PVA/PPy gel to be hydrophilic. As a comparison,
membranes with the same geometric features printed out of PR48
(a commercial photo-resin) were hydrophobic, with a contact
angle of 128°, and had a >65% lower fog collection rate
(Supplementary Fig. 12). Membranes of pure PVA showed a
similar contact angle to PVA/PPy gel and exhibited similar fog
collection behaviors (Supplementary Fig. 13).
The conical structure of the PVA/PPy gel micro-trees enables
ef
fi
cient directional transport of deposited droplets, thus
re-exposing the gel surface to incident vapor and accelerating
the collection cycle. We compared fog collection behaviors of gel
membranes with different surface topologies. Directional droplet
movement was observed on gel cones (Supplementary Fig. 14)
and the movement was faster as the apex angle decreased. On a
tilted
fl
at surface, initial water droplets randomly deposited and
then increased their size through capturing drops in fog or
coalescing with adjacent droplets but without obvious transfer of
mass center (Supplementary Fig. 15). On gel cylinders, the
droplet grew slowly while sticking on the cylinder until it fell
(Supplementary Fig. 16). Both of these geometries do not lend
themselves to quick regeneration of available droplet attachment
which reduces the collection rate. Assembled by cones with the
smallest apex angle (sin
α
=
0.10), our gel micro-trees array
achieves the most ef
fi
cient fog collection.
In addition, the hierarchical array provides a drag force
resisting fog
fl
ow by lowering their speed in the region between
the trees, thus increasing the possibility of droplets deposition on
gel surface (Supplementary Fig. 17)
36
. This is also indicated by
the varied time for different gel structures to reach their saturated
collection rates, as shown in Fig.
3
b. The
fl
at membrane reached
its maximum collection rate in the
fi
rst 15 min because its whole
surface was contacting with droplets right after it was exposed to
fog
fl
ow while the gel micro-tree array showed much longer ramp
time due to reduced
fl
ow speed and increased surface area. This
dragging effect also affects the drainage of collected water and
thus the size of the gel-tree array is tuned to facilitate the drainage
(Supplementary Fig. 18).
Solar steam generation by PVA/PPy gel membrane with micro-
tree array
. PVA/PPy gel has been reported to be a highly ef
fi
cient
interfacial solar steam generator because it ef
fi
ciently transports
water through porous gel matrix and reduces water evaporation
enthalpy
16
. We measured the solar steam generation properties of
PVA/PPy gel membranes with different surface microstructures
under 1 sun illumination (1 kW m
−
2
) by recording the overall
mass change over 1 h, which represents the amount of evaporated
a
cd
b
3D printed structures (PR48)
PDMS mold
PVA/PPy gel structures
e
Fig. 2 Fabrication and structure characterization of microstructured PVA/PPy gel membranes. a
Schematic illustration of the fabrication of
microstructured PVA/PPy gel membranes.
b
Images of representative fabricated PVA/PPy gel micro-tree array. Scale bar: 1 cm.
c
Images of an individual
representative tree micro-topology. Scale bar: 1 mm.
d
SEM image of one gel branch. Scale bar: 50
μ
m.
e
Porous structure of gel matrix. Scale bar: 5
μ
m.
ARTICLE
NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-23174-0
4
NATURE COMMUNICATIONS
| (2021) 12:2797 | https://doi.org/10.1038/s41467-021-23174-0 | www.nature.com/naturecommunications
water. The membrane was
fl
oated on water and placed under the
light beam. The mass of the water loss was measured every 10
min after the temperature of gel membranes achieved steady
status. The PVA/PPy gel membrane with micro-tree array
showed the best evaporation rate calculated per projected (illu-
minated) area,
A
p
, of 3.64 kg m
−
2
h
−
1
, which is 7.1 times higher
than that of free water and 14.1% higher than that of gel
fl
at
membrane (Fig.
4
a). We fabricated additional hybrid gel micro-
trees arrays with large areas (Supplementary Fig. 19) or with a
4 mm thick supporting layer and found that the water evapora-
tion rate remained similar.
Energy ef
fi
ciencies of different gel membranes can be
calculated using
16
:
η
¼
_
mh
V
=
C
opt
P
0
ð
2
Þ
where
ṁ
is the mass
fl
ux of evaporated water, h
V
is the
vaporization enthalpy of the water,
P
0
is the solar irradiation
power (1 kW m
−
2
), and C
opt
is the optical concentration on
b
c
a
d
e
2α
ΔP
m/A
p
m/A
s
+0 s
+3.0 s
+4.0 s
+6.0 s
+8.0 s
+15.0 s
+16.0 s
+17.0 s
f
=−
2
(
+
)
sin
Fig. 3 Fog collection properties of PVA/PPy gel membranes with micro-tree topologies. a
Snapshots of fog collection process for a single gel tree. Red
circles correspond to droplet formation events and arrows point to droplet motion trajectories. Scale bar: 0.5 mm.
b
Fog collection rates measured for
different gel membrane geometries as a function of time that demonstrates saturation at a particular time, unique to each geometry.
c
Fog collection rates
of different gel membranes at steady states normalized by the projected area,
A
p
, (left column) and by the total surface area,
A
s
, (right column).
d
Fog
collection rate and Laplace pressure difference as a function of apex angle.
e
Comparison to commercial meshes and a real Cactus stem.
f
PVA/PPy gel
micro-tree array maintains dual water harvesting functions after more than twenty-month storage.
NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-23174-0
ARTICLE
NATURE COMMUNICATIONS
| (2021) 12:2797 | https://doi.org/10.1038/s41467-021-23174-0 | www.nature.com/naturecommunications
5
absorber surface. Note that the water con
fi
ned in hydrogel
molecular mesh is evaporated to a state with a lower enthalpy
change than conventional latent heat
16
,
18
,
44
,
45
. We used Raman
spectra to con
fi
rm the existence of water molecules with different
bonding states in PVA hydrogel and conducted controlled
evaporation and differential scanning calorimetric (DSC) mea-
surements to measure the evaporation enthalpy (Supplementary
Figs. 20, 21). All gel membranes showed similar water evapora-
tion enthalpy, which was demonstrated to be unaffected by
micro-scale structures. The energy ef
fi
ciencies of different gel
structures are shown in Fig.
4
b together with their evaporation
rates, which conveys that PVA/PPy gel micro-tree array has the
highest energy ef
fi
ciency out of all tested geometries, and reaches
up to ~96%, a factor of 65% greater than that of a porous
plasmonic absorber
46
and 10% higher than that of carbon foam
12
.
The PVA/PPy gel micro-trees array also shows the ability to
effectively purify brines with different salt concentrations
(Supplementary Fig. 22) and it will not contaminate the collected
fog water (Supplementary Table 1).
The difference in water loss rates among gel membranes with
different micro-topologies indicates that surface features, i.e.,
surface area, speci
fi
c geometries, etc., affect solar steam genera-
tion. To understand the mechanisms, we examined the energy
fl
ow at steady-state by calculating the energy balance between
solar irradiation, convection, radiation loss, evaporation, and loss
to the water (Supplementary Fig. 23). We identi
fi
ed four
structure-related factors that most signi
fi
cantly in
fl
uence the
energy
fl
ow: (1) light absorption, (2) surface area, (3) surface
temperature, and (4) local humidity. We found that all gel
membranes exhibited light absorption above 90% (Fig.
4
c), with
the micro-tree array having the highest absorption from
wavelength of 250
–
2500 nm, possibly enhanced by increased
light scattering within the
“
forest
”
.
In an interfacial solar steam generator, the light-to-thermal
energy conversion and water evaporation processes are con
fi
ned
to the gel-air interface, which implies that a large surface area and
a high equilibrium surface temperature are bene
fi
cial for steam
generation. These two factors are found to be affected by surface
microstructures due to structural shadowing and changed light
incident angle
15
,
47
,
48
. We simulated and experimentally con-
fi
rmed the temperature distribution within the PVA/PPy gel
membranes subjected to normal incidence irradiation from the
light in the solar simulator (Supplementary Fig. 23). The contour
plots in Fig.
4
d and the temperature vs. time plots at four different
positions along the height of a representative tree shown in
Fig.
4
e, indicate that all gel microstructures have a lower average
surface temperature at steady states compared with a
fl
at surface.
It appears that the cone absorbs light along its entire surface, thus
reaching an average surface temperature of ~27.5 °C. The cylinder
absorbs light only at the top surface, which results in a ~1.0 °C
lower average surface temperature in the cylinder array and limits
its overall water evaporation rate despite having a larger surface
area compared with the cone array. By assembling the cones in a
branched way, gel micro-trees can directly absorb sunlight
through most of their surfaces, which enables maintaining a high
average surface temperature of ~28.0 °C under 1 sun irradiation
and results in energy ef
fi
ciency close to 100%. We also noted that
compared to
fl
at membrane, microstructured gels allowed more
heat consumption through increased gel-air interface and
minimized the energy dissipated to the gel underneath the
membrane surface, thus improving their energy ef
fi
ciencies.
Local humidity near the gel-air interface also in
fl
uences vapor
generation and can be affected by surface morphology
16
. An ideal
surface structure should facilitate easy escape of generated vapor
since accumulated vapor leads to increased local humidity and
hinders water evaporation. Assouline et al.
49
, reported that an
individual cone has a lower resistance to vapor
fl
ow compared
with cylinder or inverted cone structures due to its convergent
fl
ow lines towards the narrow opening. In an array, the vapor
fl
ow is also affected by the eddy currents from the adjacent
structures. In our experiments where the pitch among the features
was systematically varied while other parameters remained
fi
xed,
Fig. 4 Solar steam generation properties of microstructured PVA/PPy gel membranes. a
Water loss for different membrane shapes under 1 sun, with
free water as control.
b
Evaporation rate and energy ef
fi
ciency for different tested micro-topologies.
c
Light absorption spectra over wavelengths of
250
–
2500 nm of gel membranes with different micro-topologies. The small jump of the curves at wavelength ~900 nm is caused by the switch of
detectors.
d
Surface temperature contours (left) and illuminated pattern (right) for different micro-topologies under normal 1 sun illumination, simulated
using COMSOL.
e
Measured surface temperature as a function of time at four positions along the height of a typical gel micro-tree.
ARTICLE
NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-23174-0
6
NATURE COMMUNICATIONS
| (2021) 12:2797 | https://doi.org/10.1038/s41467-021-23174-0 | www.nature.com/naturecommunications
the gel cylinder array exhibited a performance drop of 15.5%
when the cylinder distance was reduced by two times; the cone
array maintained its performance as virtually unchanged over the
separation distances of 0.4
–
1.2 mm (Supplementary Fig. 24a). We
explain this behavior by the fact that the closely packed cylinders
trap vapor more effectively compared with cone array. We
introduce a shape factor as a
fi
gure of merit to qualitatively
account for this effect (Supplementary Fig. 24b). We further
compared two geometrical factors, shape factor and total surface
area, of gel micro-trees array to those of other micro-structure
arrays at different inter-distances (Supplementary Fig. 25). The
results showed that the cone-based gel micro-trees array enables a
larger evaporation area for a comparable shape factor, indicating
that the generated vapor could still ef
fi
ciently escape when the
evaporation area is greatly increased.
Ef
fi
ciency and performance of all-day water collection by
micro-topological PVA/PPy gel membranes
. We conducted a
rooftop test to evaluate the water collection ability of micro-
topological membranes in a natural environment by harvesting
fog over 12 h periods, from 20:00 pm to the next day
’
s 8:00 am
and desalinating brine water under sunlight from 8:00 am to
20:00 pm (Supplementary Figs. 26, 27 and Supplementary
Table 2). The gel samples with total membrane areas of
55
–
126 cm
2
were held by a supporting structure shown in Fig.
5
a
and placed in a prototype device for all-day water collection
(Fig.
5
b). Solar irradiation was carefully traced every hour using a
portable solar power radiation meter (Fig.
5
c). The results in
Fig.
5
d showed that on a typical sunny day in Pasadena, CA, with
an average solar heat
fl
ux of ~1 kW m
−
2
, the amount of collected
water during daytime was ~150 mL and ~35 mL during night-
time, which translates into ef
fi
ciencies of ~28 L m
−
2
and
~6 L m
−
2
, correspondingly, based on the area of gel membranes.
The water collection rates based on the total water surface area
were also calculated to show the overall ef
fi
ciency of our rooftop
prototype (Supplementary Fig. 28). The average energy ef
fi
ciency
of microstructured gel membranes in the system is around 50%.
Parameters including temperatures, wind speed, relative humidity
of outdoor system were also recorded and analyzed for two
daytime tests, which indicated that the restrained water vapor-
ization was mainly caused by lower sunlight input and saturated
internal humidity of the closed system (Supplementary Figs. 29,
30)
13
. It should be noted that the fog collection rates vary with the
weather conditions (Supplementary Fig. 31). On cloudy nights,
like those on days 1 and 3, around 10 L m
−
2
of fresh water can be
harvested. We expect this rate could be higher in a foggy location.
Sample holder
Fabric wick
Connection to storage
Foldable cover
a
b
e
d
9:00 AM
14:00 PM
19:00 PM
c
f
b
Fig. 5 All-day water harvesting by PVA/PPy gel micro-tree array outdoors. a
Gel samples are held by a supporting structure made of polyurethane foam.
b
Schematic illustration and photos of a rooftop prototype acting as a solar water desalination system during daytime.
c
Solar radiation recorded during
rooftop tests by portable solar power radiation meter.
d
Daily water collection per square meter of gel membrane during rooftop tests. Red: water collected
during daytime (8 am
–
8 pm); blue: water collected during nighttime (8 pm to next day
’
s 8 am).
e
Daytime and
f
nighttime modes of a
fl
oating water
harvesting prototype. Insets show the water collected during a day (~170 mL) and night (~70 mL).
NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-23174-0
ARTICLE
NATURE COMMUNICATIONS
| (2021) 12:2797 | https://doi.org/10.1038/s41467-021-23174-0 | www.nature.com/naturecommunications
7