1
Supplementary
Information for
All
-
day Fresh Water Harvesting by Microstructured Hydrogel
Membrane
s
Ye Shi
1,
*
, Ognjen Ilic
1,2
, Harry A. Atwater
1
& Julia R. Greer
1,
*
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
*
yeshi119@utexas.edu
, jrgreer@caltech.edu
S1 Supplementary Methods
S1.1
CAD
design of membranes with micro
-
tree, cone, and cylinder array by
Solidworks
The micro
-
structured gel membranes are initially designed and drawn by Solidworks
software.
Supplementary Figure
1
a
shows the blueprints of membrane with micro
-
tree
array in 45 deg
ree top, side, and top views. The height of tree trunk is 5 mm and its
bottom diameter is 1 mm. 9 smaller cones with same conicity are distributed
uniformly as branches at 1/3, 1/2, and 2/3 height of trunk with 45 degree tilted. 100
2
these micro
-
trees are h
exagonally planted on a 1.5
-
mm thick supporting layer and the
distance (center to center) between two adjacent trees is 2.5 mm.
The gel thi
ckness
has been demonstrated to have no effects on s
aturated water content and water
transport rate of hybrid hydroge
l
s
1
.
The relatively thick supporting layer
used
is
to
maintain the structure integrity of hydrogels during fabrication and processing.
The cones have exactly same size with micro
-
tree
’
trunk, with height of 5 mm
and bottom diameter of 1 mm (
Supplementary F
igure
1
b
). 100 these micro
-
cones are
hexagonally distributed on the supporting layer with inter
-
distance (center to center)
of 2.5 mm.
The hexagonal arrangement is beneficial for fog collection because with
the staggered arrangement each cone is wholly
wrapped by the flow stream and water
drops can not only deposit on the windward side but also o
n the leeward side of the
cones
2
.
The cylinders have height of 5 mm and diameter of 1 mm (
Supplementary
Figure
1
c
). 100 these micro
-
cylinders are hexagonally dis
tributed on the supporting
layer with inter
-
distance (center to center) of 2.5 mm.
T
he projected area of supporting layer is designed to be slightly larger than that
of micro
-
structures array, thus facilitating mold assisted fabrication. The supporting
l
ayer is cut to fit the area of array for final PVA/PPy gel membranes during all tests. It
is also worthy to notice that PVA/PPy gels shrinks a little during cycles of freeze
-
thaw
processing. The size of final gel structures is ~80% of the size in CAD desig
n (in one
dimension). For examples, the height of PVA/PPy gel micro
-
tree is ~4 mm and its
bottom diameter is ~0.8 mm. The ratio
of
total surface areas
for
membranes with flat
surface, cone array, cylinder array and micro
-
tree array is 1:1.8:2.4:3.5, respec
tively.
3
S1.2
Set
-
up of f
og collection test
in lab
To test the fog collection ability of
microstructured
PVA/PPy gels, a hydrogel
membrane
sample with 4 cm
2
projected area is placed with a inclined angle
θ
to the
horizontal surface, meanwhile, a sustained fog flow generated by ultrasonic
humidifier
(LV600HH, Levoit)
with a velocity of about 1 m s
–
1
is kept blowing to the
surface with a tilted angle (15 degree
) to the tangent direction of the
membrane
at
room temperature
(
Supplementary Figure
5
a)
3
. The fog flow angle is carefully
selected to make sure the fog flow passes through the gel micro
-
structure array and
minimize the influence of supporting layer. If th
e angle is smaller, the fog flow may
directly hits the supporting layer and causes significant edge effect. When the angle is
larger, we observed that part of the fog flow bounced back from the substrate and flow
into the collection beaker directly.
The ou
tlet of fog is kept 3 cm from the bottom of gel membrane.
θ
is tuned from
15 degree
, 45 degree
to 85 degree. The fog flow is just blown to the structured region
and higher than the solid substrate, which helps avoid edge effect on supporting layer
3
.
A beak
er is placed under the gel sample to collect drained water and the amount of
collected water is measured every 15 minute
s
. Without further clarification, the fog
collection rate in this Supplementary Text is calculated based on the projected area of
gel me
mbranes. We studied the effect of inclined angle for the gravity assisted
drainage and found that there was no obvious difference in resulted fog collection rate
(
Supplementary Figure
5
b)
. This is mainly because the millimeter
-
size droplets
formed by coale
scence of smaller droplets from all branches have an initial speed
4
when they drop from the micro
-
trees, which facilitates their drainage. Thus
θ
is set as
45 degree for all fog collection tests. The room temperature for fog collection tests is
25 °c and th
e relative humidity in artificial fog flow is 100%.
To study the influence of fog flow velocity, fog flows with 0.5 m/s and 2 m/s
speed
were applied (
Supplementary Figure
5
c)
.
It can be seen
that as the fog flow
speed increases, the time for gel micro
-
tree
array to reach the saturated collection rate
decreases but the maximum fog collection rate remains almost same. This is because
it takes longer time for
slower
fog flow to pass through the micro
-
tree array and
to
fully wrap the whole surface of gel micro
-
trees. Once
the array is
saturated
by the fog
flow, the concentration of water droplets in the fog exceeds the fog collection capacity
of the gel surface. The whole surface of gel
membrane
continuously captures dr
oplets
and transports them for collection.
S1.3
Experimental measurement of equivalent vaporization enthalpy in
microstructured PVA/PPy hydrogels
To compare the water vaporization enthalpy in microstructured PVA/PPy hydrogels,
we designed a control exper
iment to measure the vaporization enthalpy
1
. As shown in
Supplementary Figure
21a, a container is set under room temperature (RT) and half of
the container is filled up with supersaturated potassium carbonate solution to enable
stabilized relative humidity
(RH) of ~45% in the closed space. Free water and gel
samples with same evaporation area are synchronously put in the closed container
above potassium carbonate solution. To keep the evaporation area same, an optical
5
profilometer was used to measure the su
rface area of hydrogel, which is ~2 m
2
/m
2
.
Then the total surface area of gel microstructures was estimated by combining the
parameter from CAD file used for 3D printing. The mass change of free water and gel
sample caused by water evaporation is measured
every hour and corresponding
equivalent evaporation enthalpy
(Δ
H
equ
) is calculated based on the following equation
using average evaporation amount in an hour:
푈
푖푛
=
∆
퐻
푣푎푝
푚
0
=
∆
퐻
푒푞푢
푚
푔
where U
in
is the power input which is identical for free water and gel samples;
Δ
H
vap
and m
0
are the vaporization enthalpy and average evaporation amount of free water;
m
g
is the average evaporation amount of gel samples.
To further prove the reduced evaporation en
thalpy of water in PVA/PPy hybrid
gel, differential scanning calorimetric (DSC) measurement is used for measuring the
vaporization energy of pure water and water in the gel
.
The gel sample was placed in
an open Al crucible and measured with a linear heatin
g rate of 5 K min
-
1
, under a
nitrogen flow (20 mL min
-
1
), in the temperature range from 20 to 180 °C. The
effective specific heat capacity was calculated by comparing the heat flow of
measured gels with that of the standard sapphire sample.
S1.4
Simulatio
n of s
urface temperature distribution
of PVA/PPy gel
microstructures
At steady state, the net temperature and evaporation rate is determined from the
energy balance between various terms: solar irradiation, convection, radiation loss,
6
evaporation, and loss
to the water underneath. This balance can be expressed as:
푄
푠표푙푎푟
+
푄
푐표푛푣
+
푄
푟푎푑
+
푄
푒푣푎푝
+
푄
푤푎푡푒푟
=
0
(S1)
In our system, there a number of surfaces that are not normal to the incident light
direction (z), so the energy flux due to irradiation can be expressed as
푄
푠표푙푎푟
=
훼
퐼
푠표푙푎푟
|
푛
̂
⋅
푧
̂
|
, where
훼
is the surface absorptivity (Fig. 4c in main text), an
d
퐼
푠표푙푎푟
is
the solar irradiance at Earth level. Generally, for a closed environment with controlled
ambient parameters (humidity, pressure, temperature), the evaporative flux can be
expressed as
푄
푒푣푎푝
=
퐻
휈
푘
(
퐶
푠푎푡
(
푇
)
−
퐶
푤푎
)
where
퐻
휈
,
푘
are
the heat of evaporation
and the mass transfer coefficient, respectively, and
퐶
푠푎푡
,
퐶
푤푎
are the concentration of
saturated vapor and the concentration of vapor in air, respectively
4
. The saturation
concentration relates to the saturation pressure
푝
푠
푎푡
as
퐶
푠푎푡
(
푇
)
=
푝
푠푎푡
(
푇
)
/
푅푇
. The
convective heat flux is expressed as
푄
푐표푛푣
=
ℎ
푐표푛푣
(
푇
−
푇
푎
)
, where
푇
푎
≈
23
°C
is
the ambient environment temperature, and
ℎ
푐표푛푣
is the convective heat transfer
coefficient. The radiation loss term is propo
rtional to emissivity of the material, the
local temperature, and the background environment temperature, i.e.
푄
푟푎푑
=
휖휎
(
푇
4
−
푇
푎
4
), where
휎
is the Stefan
-
Boltzmann constant; from Kirchoff
’
s law, we
assume the emissivity of the gel structure is equal
to its absorptivity, e.g.
휖
=
훼
. Last,
the energy flow to the underlying water is incorporated through the temperature
boundary condition where the water temperature is equal to the environment
temperature
푇
푎
(room temperature). We use this energy balanc
e model to simulate
and identify qualitative trends in surface temperature distribution for different
morphologies (cone, cylinder, tree). In COMSOL Multiphysics, cone and cylinder
7
case is analyzed as a two
-
dimensional axisymmetric model, while the tree ca
se is
analyzed as a three
-
dimensional model with an illumination source incident from the
top (
-
z direction). The energy balance of Equation (S1) is applied as the net heat flux
boundary condition at
“
top
”
interfaces that are exposed to illumination/evapor
ation.
For the edges of the boundary domain below the top interface (side walls and below
surface level), we assume insulating boundary conditions (i.e. no heat flux across the
boundary). In our model, we assume values for the incident solar intensity (1,0
00
W/m
2
), heat of evaporation (~1,000 kJ/kg), convective heat transfer coefficient (~10
W/m
2
K), and estimate
푘
~
2
.
2
⋅
10
−
5
m/s
4
.
S1.5
Calculation of shape factors for different gel microstructure arrays
To get a qualitative understanding for how the
inter
-
distance affects vapor escape
in arrays with different morphology, we draw a similarity to the concept of shape
factor in radiative heat transfer
5
. Shape factor is a geometrical function that depends
on the size, shape, separation distance, and orien
tation of participating elements. The
shape factor between two surfaces A and B, labelled
퐹
퐴
→
퐵
, relates the proportion of
radiation leaving surface A that is intercepted by surface B. We used shape factor as a
geometrical characteristic to qualitatively
describe the
“
packing density
”
of
participating elements in our gel micro
-
structure arrays by showing how closely these
micro
-
structures are packed together in one array and how much open space the array
could provide for the vapor to escape. As the inter
-
distance of gel micro
-
structures on
our membranes is at millimeter level, smaller shape factor indicates more open space
8
for vapor to escape and lower chance for generated vapor to be interfered by the
adjacent micro
-
structures.
We numerically evaluate t
he shape factor (COMSOL Multiphysics) between the
nearest
-
neighbor elements for cone, cylinder, and tree arrays.
Geometrical shape
factors are calculated using COMSOL Multiphysics Surface
-
to
-
Surface Radiation
interface. The computational domain consists of
two nearest
-
neighbor elements. The
boundary conditions applied to the surface of the elements treat them as diffuse
black
-
body radiators with emissivity near
-
unity emissivity. The shape factor, also
known as the view factor, between the two elements/surfa
ces A1 and A2 is defined as
the ratio between the diffuse energy leaving A1 and intercepted by A2 and the total
diffuse energy leaving A1, that is
퐹
퐴
1
−
퐴
2
=
∫
Υ
퐴
2
(
퐽
1
)
푑푠
∫
퐽
1
푑푠
퐴
1
, where
퐽
1
is the radiosity of
element 1, and
Υ
(
J
1
)
is the irradiation operator from the surface
-
to
-
surface radiation
interface in COMSOL Multiphysics.
S1.6
Design of floating prototype for all day water collection in natural
environments
We designed a floating prototype for all day water collection in n
atural environments,
such as sea, lake, or pool
.
The structure was constructed from polyester thin film,
cellulose
-
based fabric, polyurethane foam, metal wires and wood rods. The
condensation structure was constructed from lightweight and cheap polyester f
ilms.
The film was cut into several pieces and glued together on skeleton made by steel
wires. Droplet collection was facilitated by inclined polymer film and
9
super
-
absorbable fabric wicks (Zorbs)
6
. Collected water was transported to water
storage bag by f
abric wicks. The PVA/PPy gel samples were held by a supporting
structure which was made by polyurethane foam and nylon mesh. The wholesale
materials cost of the entire floating prototype is ~ $4.
The unique feature of our floating prototype is its foldabl
e condensation structure
which enables dual mode for all day water collection. Our design can be easily
replicated at home or modified and produced by factory. Smart or remote modulus can
be further added to the device to enable intelligent control of wate
r collection modes.
10
S2 Supplementary Figures and Tables
Supplementary Figure 1.
CAD blueprints of membranes with (
a
) micro
-
tree, (
b
)
cones and (
c
) cylinder array.
11
Supplementary Figure
2
.
SEM images of PVA/PPy hydrogels showing (
a
)
broad
internal gaps with diameters from 50 to 150
μ
m
, which
together with micro
-
pores
enable rapid water diffusion and capillary pumping
to
supporting a sustained high rate
vapor generation
1
.
(
b
) the wrinkled internal surface
,
which
indicat
es
shrinkage of the
polymeric skeleton (PVA network) during dehydration of the hydrogel
.
12
Supplementary Figure
3
.
FTIR spectra of PVA, PPy and PVA/PPy hybrid gels.
PVA
shows a characteristic peaks at 1093 cm
−
1
, which can be attributed to C
–
O stretching.
PPy shows absorption signals at 1552 cm
−
1
and 1045 cm
−
1
, which are corresponding
to the in
-
ring stretching of C=C bonds in the pyrrole rings and the in
-
plane
deformation of N
–
H bonds, respectively. All these characteristic peaks of PVA and
PPy can be found
in the FTIR spectra of PVA/PPy hybrid gel, which confirms the
presence of PPy in the PVA matrix. These peaks show no shifts, indicating that PPy
particles are physically mixed with PVA.
13
Supplementary Figure
4
.
(a)
T
he storage modulus (G
’
) and loss
modulus (G
’’
)
of
as
-
prepared PVA and PVA/PPy gels.
B
oth samples exhibit solid gel behavior. The
hybrid gel exhibits a ~40% lower G
’
than the pure PVA gel because it has fewer
crosslinking points caused by the introduction of PPy. The lower G
”
of the PVA/PP
y
hybrid gel indicates that the polymeric PVA chains are immobilized by the hard PPy
segments.
(b) T
he storage modulus (G
’
) and loss modulus (G
’’
)
of PVA/PPy gels after
~20 month
’
s storage.
T
he crosslinked network structure was well maintained after
long
-
t
erm storage
.
(c) Pictures of PVA/PPy gel micro
-
tree membrane being bent.
14
Supplementary Figure
5
.
(
a
) Schematic illustration of set
-
up of fog collection test in
lab. (
b
) Fog collection rates of PVA/PPy micro
-
tree array with different inclined
angels
in lab tests.
(c) Fog collection rates of PVA/PPy gel micro
-
tree array under
different fog flow speeds.
15
Supplementary Figure
6
.
Photos showing fog collection
behavior
of one PVA/PPy
gel micro
-
tree
.
16
Supplementary Figure
7
.
(
a
) Photos of PVA/PPy
gel cone arrays with different apex
angle. (
b
) Fog collection rates of PVA/PPy gel cone arrays with different apex angle
in fog harvesting tests.
After normalizing the fog collection rates by total surface area,
the fog collection ability of different cone
arrays improves as the apex angle decreases,
indicating faster directional movement of droplets on cones with smaller apex angle.
These results indicate that while the surface area is maintained, the fog collection
ability of conical gel structures
can be
improved
by decreasing their apex angle.
17
Supplementary Figure
8
.
(a)
Fog collection rates of different materials under
equivalent testing conditions.
All the tested materials are cut into same diamond shape
as PVA/PPy gel membrane and tested under
same experimental conditions. The fog
collection rates are calculated based on the projected area of these membranes. For
porous Raschel mesh (double layered, 35% shading) and hex mesh (
double layered,
50% shading), the fog collection rates are calculated
based on their effective area
(area of pores is excluded)
.
(b) Fog collection rates of different materials calculated
based on their mass
in wet and dry states
.
18
Supplementary Figure
9
.
(a) Fog collection rates of microstructured PVA/PPy gel
membranes
along with their
number of cones.
The droplets formed on cone tips sit for
much longer time than droplets at other locations.
T
he contribution from each cone
are different in three different structures
.
(b) Fog collection rates normalized by
number of
cones for different microstructured PVA/PPy gel membranes
, which
indicate that
t
he number of cones may not be a determining factor for fog collection
rate
.
19
Supplementary Figure
10
.
(a) Schematic illustration of cone arrays with changed
size (side view
).
The dimension (both height and bottom diameter) and inter
-
distance
of cones are 3/4 (medium cones) and 1/2 (small cones) of original ones
.
(b) Fog
collection rates of cone arrays with different sizes
during tests.
(c)
Fog collection rates
of cone arrays with
different
size and their number of cones.
All these evidence
indicate that the droplets deposited on the cone tips contribute little to the fog
collection ability of cone
-
based structures.
20
Supplementary Figure
11
.
(a)
Fog collection rates of microstructured PVA/PPy gel
membranes with different projected membrane area
s
.
T
heir fog collection rate during
steady states has a proportional relationship to the membrane area, as far as the entire
membrane is covered by
full fog flow
.
(b) Areal fog collection rate
remained constant
for each structure
.
The results demonstrate that
fog collection rates of different gel
structures
can be
normalize
d
by their total surface area and
thus
the effects of other
factors
can be
exam
ine
d
separately.
21
Supplementary Figure
1
2
.
Fog collection rates of (a) PVA/PPy gel membranes and
(b) cured PR48 membranes with different microstructures. The insets show the
contact angle tests of two materials.
The hydrophobicity of surface can
affect fog
collection behavior
7,8
.
S
tructured PR48 membranes show much worse fog collection
performance than PVA/PPy gel membranes with same microstructures
,
demonstrat
ing
that hydrophilic nature of PVA/PPy hydrogel can benefit fog collection through
facil
itating fog droplets deposition.
22
Supplementary Figure
13
.
(a)
Fog collection rates of (a) PVA/PPy gel membranes
and (b)
pure PVA hydrogel
membranes with different microstructures. The insets
show the contact angle tests of two materials.
T
he addition of PPy doesn
’
t affect the
fog collection behavior of micro
-
structured hydrogels. The reason
could
be that the
PPy particles are firstly synthesized and then added to PVA solution for gelation. They
are embedded in the PVA matrix, rather than on
the gel surface.
23
Supplementary Figure
14
.
Photos showing fog collection behavior of PVA/PPy gel
micro
-
cone.
After the fog flow is applied, tiny droplets deposit on the gel surface.
From the 4
th
second, one major droplet forms due to coalescence of small droplets and
it moves towards the base of cone. The droplet keeps growing by absorbing fog
droplets in air and new deposited droplets on gel surface during its directional
movement. After around
20 s, the droplet drains from the gel cone and
the whole
surface is refreshed.
24
Supplementary Figure
1
5
.
Photos showing fog collection behavior of PVA/PPy gel
flat surface
.
T
iny water droplets randomly deposit on the smooth gel surface after fog
flow
is applied. With continued deposition, the water drops increase their size through
directly capturing drops in fog or coalescing with other drops nearby but without
obvious transfer of mass center in either case. After 75 s, a large and heavy enough
water
droplet forms and drains off from the gel surface. The absence of quick
regeneration of the fresh deposition sites in the overall process counts against the fog
collection.
25
Supplementary Figure
1
6
.
Photos showing fog collection behavior of PVA/PPy gel
cylinders
.
After the initial drop forms on the gel cylinder, the size of droplet keeps
increasing with a much lower rate by absorbing water in fog flow. The droplet sticks
on the gel surface without ob
vious movement of its mass center. After more than 2.5
min, the droplet falls from the gel cylinder when it becomes too large for the structure
to support its weight. The sticking behavior leads to even worse fog collection
performance of gel cylinders tha
n that of flat surface.
26
Supplementary Figure
1
7
.
Simulated flow field for a configuration with a reduced
number of tree elements
.
The dynamics were modeled with the Reynolds
-
averaged
Navier
-
Stokes (RANS) equations, with automatic wall treatment and
default
COMSOL flow parameters.
Arrow corresponds to the inflow direction, with the
boundary velocity of 1 m/s at the inlet.
This qualitative result corroborates the
assessment that the PVA/PPy gel micro
-
tree array disrupts and slows down the fog
flow fiel
d.