Desalination xxx (xxxx) xxx–xxx

Contents lists available at ScienceDirect

Desalination
journal homepage: www.elsevier.com/locate/desal

Overcoming temperature polarization in membrane distillation by

thermoplasmonic effects activated by Ag nanofillers in polymeric
membranes
Antonio Politanoa, Gianluca Di Profiob, Enrica Fontananovab, Vanna Sannac, Anna Cupolillod,
⁎
Efrem Curciob,e,
a
Fondazione Istituto Italiano di Tecnologia, Graphene Labs, via Morego, 30, 16163 Genova, Italy
b
Institute on Membrane Technology, National Research Council of Italy ITM-CNR, Italy - Via P. Bucci Cubo 17C, 87036 Rende, CS, Italy
c
Nanomater S.r.l., Porto Conte Ricerche, Loc. Tramariglio, 07041 Alghero, SS, Italy
d
Department of Physics, University of Calabria, Via P. Bucci Cubo 31C, 87036 Rende, CS, Italy
e
Department of Environmental and Chemical Engineering DIATIC, University of Calabria, Via P. Bucci Cubo 45A, 87036 Rende, CS, Italy

A R T I C LE I N FO A B S T R A C T

Keywords: In recent years, Membrane Distillation (MD) has emerged either as a promising alternative or as a complement to
Membrane distillation Reverse Osmosis (RO) in seawater desalination. However, the performance of MD is significantly offset by
Thermoplasmonics temperature polarization, a phenomenon intrinsically related to water evaporation that causes the decrease of
Photothermal effects the solution temperature at the membrane surface and, ultimately, the loss of driving force. In this work, we
Surface plasmon resonance
show that temperature polarization can be withdrawn by exploiting thermal collective effects activated by the
Seawater desalination
excitation of plasmonic modes in UV-irradiated Mixed Matrix Membranes composed of silver nanoparticles (Ag
NPs) incorporated in polyvinylidene fluoride (PVDF) microporous films. For experimental tests carried out under
vacuum (VMD) and with an absorbed radiant heating power of 2.3 · 104 W/m2, best results were obtained using
PVDF membranes loaded with 25%Ag NPs. In this case, measured transmembrane fluxes to pure water and 0.5 M
NaCl solution were 32.2 and 25.7 L/m2 h, respectively, i.e. about 11- and 9- fold higher than the corresponding
values for unloaded membranes. Remarkably, energy analysis revealed that the heat generation of Ag NPs under
plasmonic resonance was able to withdrawn temperature polarization, resulting in estimated temperature po-
larization factor (TPF) values of 106.5% for 0.5 M NaCl solution.

1. Introduction Low temperature gradients are usually adequate to obtain a trans-

membrane flux of 1–10 L/m2 h with feed temperatures spanning be-
In recent years, Membrane Distillation (MD) has emerged as a tween 50 and 70 °C, thus permitting the effective recovery of low-grade
promising alternative or as a complementary unit to Seawater Reverse or waste heat streams, along with the exploitation of alternative energy
Osmosis (SWRO). MD is a thermally-driven membrane operation based sources (solar, wind or geothermal) [6].
on the use of microporous hydrophobic membranes [1]. The hydro- Presently, the main drawback of MD is represented by “temperature
repellent nature of the membrane prevents the permeation of liquids, polarization”, a phenomenon inherently related to the removal of the
while sustaining a vapor-liquid interface at the entrance of each pore, latent heat associated to water evaporation, which causes the pro-
where water evaporation takes place. Under a temperature gradient, a gressive decline of the feed temperature profile in proximity to the
net diffusive flux of vapor is observed from the hot feed compartment to membrane surface. As a result, the temperature at the membrane in-
the opposite side of the membrane (distillate compartment) [2–4]. terface is lower than the bulk temperature and, consequently, the net
In desalination applications, MD is able to produce desalted water at driving force to the mass transport falls down.
high purity since, theoretically, non-volatile compounds (ions, colloids, Some technological solutions have been proposed so far in order to
and macromolecules) are completely retained by the membrane. In alleviate the negative impact of temperature polarization. Since polar-
addition, MD is unlimited by polarization phenomena and, therefore, ization phenomena are localized within the micrometric boundary layer
recovery factors > 85% can be achieved [5]. adjacent to the membrane, improved hydrodynamic conditions

⁎
Corresponding author.
E-mail address: [email protected] (E. Curcio).

https://doi.org/10.1016/j.desal.2018.03.006
Received 2 October 2017; Received in revised form 18 February 2018; Accepted 6 March 2018
0011-9164/ © 2018 Elsevier B.V. All rights reserved.

Please cite this article as: Politano, A., Desalination (2018), https://doi.org/10.1016/j.desal.2018.03.006
A. Politano et al. Desalination xxx (xxxx) xxx–xxx

(increase of Reynolds number, applying turbulence promoters like 0.688RePr(d/ L)

Nu = 3.66 +
spacers) are somehow beneficial for the simultaneous mass and heat 1 + 0.44[RePr(d/L)]2/3 (3)
transfer in MD [7,8]. Hengl et al. [9] have manufactured and tested flat
where Nu, Re and Pr are the Nusselt, Reynolds and Prandtl di-
hydrophobic metallic membranes for water evaporation in an opera-
mensionless numbers, respectively, and (d/L) is the tube diameter-to-
tional configuration approaching sweep gas MD; here, the membrane
length ratio.
acts as an electric resistance that brings energy directly to pores by
The thermoplasmonic heat flux is experimentally determined by a
Joule effect [9].
macroscopic energy balance over the whole system, considering that
However, due to the negative impact represented by temperature
the sum of qplasm and of the heat flux associated to the radiation of the
polarization on the overall thermal efficiency, the effective technolo-
UV lamp qrad (the latter is measured in a VMD cell equipped with virgin
gical implementation of MD on large industrial scale is missing yet.
PVDF membrane, i.e. qplasm = 0) balances the sum of the heat flux as-
As a result of a multidisciplinary cooperation with solid-state phy-
sociated to water evaporation (obtained by multiplying the measured
sicists active in the field of plasmonics, here we propose an innovative
transmembrane flux Jw and the latent heat of water λ) and of the heat
technological approach to reduce or withdraw temperature polariza-
flux through the liquid feed stream (obtained by multiplying the feed
tion. In thermoplasmonics [10–14], the photoexcitation of plasmons in
flowrate ṁ f , the specific heat coefficient cp and the measured tem-
metal nanoparticles (NPs) [15–17] is exploited to obtain a local in-
perature increase from inlet to outlet Tf, outb − Tf, inb). Hence:
crease of the temperature [18]. The presence of nanoscale thermal
hotspots has been demonstrated to be beneficial for different applica- ̇ p (T fb, out − T fb, in ) − qrad
qplasm = Jw⋅λ + mc (4)
tions in membrane technology [19], including MD [20]. However, the
physical mechanism ruling thermoplasmonic MD have not been clar-
ified yet. As a matter of fact, different experimentalists [21,22] have 3. Materials and methods
shown that, upon illuminating an extensive quantity of Au NPs dis-
persed in solution, the temperature profile is not localized around each 3.1. Membrane preparation and characterization
NP, as a consequence of the occurrence of thermal collective effects
[23]. The distribution of the temperature becomes nearly constant Ag NPs were synthesized by the chemical reduction method using
throughout the NP assembly, regardless the nanoscopic size of heat sodium citrate as metal precursor [24]. In a usual batch, AgNO3 salt
sources [23]. While applications in nanotechnology could be inhibited (Sigma-Aldrich, US) was first dissolved into water (100 mL of aqueous
by such collective thermal effects, actually they are even advantageous solution at concentration of 1 mM) and successively heated up. Once
in membrane technology, where nano-heaters have the potential to reached the boiling point, an aqueous solution of trisodium citrate
provide a temperature increase in a large area, leading to a significant (10 mL, 1 wt%) was added, in order to reduce Ag+ ions to produce Ag
enhancement of the mass transfer. NPs. Consistently, the colloidal solution assumed a light yellow tint.
In this work, a quantitative evaluation of temperature polarization The obtained NPs were instantly lyophilized and stored at room tem-
and energy efficiency is provided for a vacuum MD (VMD) system op- perature. Size and shape of Ag NPs were characterized by transmission
erated with microporous Mixed Matrix Membranes (MMM) made in electron microscopy (TEM, FEI Tecnai G12, US). Spherical NPs with
PVDF and incorporating Ag NPs. UV-irradiation of MMMs allowed ex- average diameter of 31 ± 4 nm with rather uniform size distribution
ploiting thermal collective effects activated by the excitation of plas- are shown in Fig. 1. Ag NPs were dispersed in Dimethylformamide
mons. Process performance in terms of transmembrane flux and (DMF) bought from Sigma-Aldrich (US) at 30% w/w. UV-VIS analysis
permeate quality is also investigated. performed on a solution sample diluted 200 times evidenced a max-
imum absorbance intensity at ~420 nm, corresponding to the wave-
length of the plasmonic resonance in Ag NPs [25].
2. Theoretical background Flat microporous membranes with Ag nanofillers were prepared by
nonsolvent-induced phase inversion process. The mix of Ag NPs in DMF
VMD configuration implies a suppression of the heat flux at the was firstly sonicated for 15 min (M 428-BD, MPM Instruments Srl) at
vacuum/permeate interface [4]. Consistently, temperature gradient is room temperature, and successively mixed with PVDF homopolymer
negligible along the membrane thickness. Therefore, the heat required powder Solef®6010 (Solvay, Italy). Proper amount of pure DMF was
for the evaporation of water molecules at the feed interface is supplied added in order to modulate the final concentration of Ag NPs in the
by the heat flux through the liquid film adjacent to the membrane at the range of 15–25% w/w. After 24 h of magnetic stirring at 30 °C, the
feed side [18]. However, in our innovative system, the presence of Ag polymeric solution was cast on a glass plate by automated machine
nanofillers in the polymeric membrane leads to an additional term as- (Elcometer, UK) using a knife with gap of 150 μm, and then immersed
sociated to the thermoplasmonic heat flux (qplasm), which considers the for 2 h at room temperature in a coagulation bath of distilled water to
increase of membrane temperature at the feed interface under UV ir- activate demixing. Residual traces of solvents were removed by another
radiation: immersion in distilled water (for one day), followed by a washing with
ethanol. Membranes were dehydrated in oven at 60 °C under vacuum
hf (T fb − T fm) + qplasm = Jw λ (1) for one day. The morphology of membranes was studied by a Quanta
200 FEI (Philips, Netherlands) scanning electron microscope.
where superscript m refers to membrane interface, Tf is the average feed Experiments for evaluating the contact angle were performed by CAM
temperature (arithmetic average between inlet and outlet temperature), 200 contact angle meter (KSV Instruments LTD, Finland). Pore size
hf is the heat transfer coefficient at the feed side, Jw is the transmem- distribution was determined by Capillary Flow Porometer (CFP 1500
brane flux of vapor and λ is the latent heat of vaporization. From Eq. AEXL, PMI Inc., US).
(1), the average value of temperature at the membrane interface T fm is
evaluated to be: 3.2. VMD setup and testing

qplasm − Jw λ The VMD setup schematized in Fig. 2 consists of a dead-end cell

T fm = T fb +
hf (2) with an active membrane area of 21.2 cm2. A quartz window was fixed
and taped up at the middle of the upper cap. Above the quartz window,
The heat transfer coefficient at the feed side can be estimated by the in order to irradiate the surface of the membrane, we put a high-pres-
Hausen empirical correlation valid for Re < 2300 and Pr ≥ 5: sure UV mercury lamp (Helios Ital Quartz srl, Italy) with a wavelength

2
A. Politano et al. Desalination xxx (xxxx) xxx–xxx

ΔW
Jw =
A⋅Δt (5)

Permeate conductivity was evaluated by conductivity meter at 20 °C

(YSI Model 3200). Dynamic Light Scattering/DLS (90Plus Particle Size
Analyzer, Brookhaven Instruments Corporation, US) and High-
Resolution Continuum Source Atomic Absorption Spectrometer/HR-CS
AAS (ContrAA 700, Analytik Jena AG, Germany) were used to de-
monstrate the absence in the permeate of solid and dissolved (in the
form of Ag+ ions) NPs, respectively.
In addition to pure water and 0.5 M NaCl solution artificially mi-
micking seawater, VMD tests were also carried out on real seawater.
Raw samples from Tyrrenian sea (Amantea, Italy) were collected and
stored at 4 °C before use. Ion Chromatography (Metrohm 861 Advanced
Compact Ion Chromatograph) was employed to quantify the con-
centration of ions in both feed and distillate streams; 3.2 mM
40 Na2CO3 + 1 mM NaHCO3 was used as eluent for anion column
Metrosep A Supp 5–250/4.0, and 2 mM nitric acid + 0.25 mM oxalic
acid was used as eluent for cation column Metrosep C4–250/4.0. The
organic content in real seawater and permeate solutions of CO2-free
solutions by a Total Organic Carbon (TOC) analyzer (Shimadzu VCSN).
30 Conductivity of solutions was measured by YSI model 3200
Density function (%)

Conductivity Instrument (US). We measured pH by HI 223 Calibration-

Check Microprocessor pH Meter (HANNA Instruments, US).

20 4. Results and discussion

4.1. Performance of membranes

Table 1 summarizes the morphological and physicochemical prop-

10
erties of NP-loaded PVDF membranes. SEM micrographs proved that Ag
NPs, entrapped in the lab-made membranes during the non-solvent
phase inversion process, were well dispersed throughout the PVDF
microporous matrix. Specifically, the cross-section morphology ap-
0 peared asymmetric, with a outermost layer displaying macrovoids
0 10 20 30 40 50 generally protracted over 40% c.a. of the membrane thickness, and with
NPs diameter (nm) a steady microporous structure at the bottom layer. While thickness was
almost uniform (spanning from 60 to 72 μm), the porosity of unloaded
Fig. 1. (top panel) TEM image and (bottom panel) distribution function of the particle
PVDF membranes progressively increased with the amount of in-
size of synthesized Ag NPs.
corporated NPs (up to 20%w/w), as a result of their hetero-nucleant
effect, which accelerates the kinetics of solid phase formation. How-
of 366 nm, a watching angle of 90° and with a GR.E. 500 W portable ever, for PVDF-25% Ag NPs, the porosity reached the lowest value
power generator. The recirculation of the feed solution (0.5 M NaCl) (22%), probably due to a partial collapse of the membrane structure. As
was carried out at 20 L/h by digital peristaltic pump Masterflex L/S evaporation tests were carried out under vacuum and with non-pres-
(Cole Parmer, US). Temperature monitoring at inlet and outlet of the surized feed stream, no significant variations of the pore size were
module was performed by means of K-type thermocouples, character- detected after 6 months of experimental activities.
ized by a response time of 15 s and a resolution of ± 0.1 °C. The op- With the aim to validate the stability of membranes against NPs
erative pressure of the distillate side was 20 mbar; the initial feed dissolution with subsequent Ag+ release, samples of PVDF-Ag NPs
temperature was set to 30 °C. polymeric films (surface area of 2 cm2) were immersed (i) in pure water
The transmembrane flux of water vapor Jw was evaluated by com- and (ii) in 0.5 M NaCl for one week at 20 °C. Ion concentration was
puting the weight decrease (ΔW) of the feed solution in a fixed interval checked by HR-CS AAS.
of time (Δt) and normalizing it over the membrane area A: According to dissolution data reported in Table 2, the highest Ag+

Fig. 2. Lab-scale experimental setup. (1) VMD unit; (2) feed inlet stream; (3) feed outlet stream; (4) distillate outlet stream; (5) glass window for UV irradiation.

3
A. Politano et al. Desalination xxx (xxxx) xxx–xxx

Table 1
Morphological and physicochemical properties of investigated PVDF membranes. The scale bar in SEM images represents 10 μm.

Membrane Morphology (SEM micrographs) Thickness (μm) Porosity (%) Pore size (μm) Contact angle (°)

Virgin PVDF 64 ± 11 27 0.45 ± 0.05† 86.3 ± 2.5†

0.45 ± 0.06‡ 86.5 ± 3.4‡

PVDF-15%Ag NPs 72 ± 14 32 0.22 ± 0.03† 83.6 ± 3.9†

0.21 ± 0.05‡ 84.1 ± 2.8‡

PVDF-20%Ag NPs 65 ± 12 36 0.38 ± 0.04† 81.4 ± 0.7†

0.37 ± 0.02‡ 82.0 ± 2.5‡

PVDF-25%Ag NPs 60 ± 15 22 0.40 ± 0.05† 80.3 ± 3.6†

0.38 ± 0.04‡ 80.6 ± 3.1‡

†
Before VMD tests.
‡
After 6 months.

Table 2 Qi = σabs n s ε0 c |E ext|2 /2 (6)

Dissolution tests of Ag NPs in MMM.
where ns is the refraction index of the neighbouring medium, c is the
Membrane Pure water 0.5 M NaCl
speed of light, and Eext is the electric field of the external electro-
Dissolved Ag+ (ppb)† Dissolved Ag+ (ppb)† magnetic radiation. Thus, the choice of the material for plasmonic na-
nofillers is orientated toward systems that exhibit high values of the
PVDF-15%Ag NPs 3.4 ± 0.08 69.8 ± 2.41 absorption cross section σabs under resonance conditions for plasmonic
PVDF-20% Ag NPs 9.9 ± 0.27 126.0 ± 5.95
excitations. Based on the state of the art, the materials showing the best
PVDF-25% Ag NPs 14.0 ± 0.68 138.0 ± 6.50
performances concerning photothermal effects are Ag and Au [14,28].
†
After 7-days immersion at 20 °C. Average values on 3 samples, standard devia- However, shape-controlled Ag NPs [29], under plasmon resonance
tion < 5%. conditions, can generate a heat flux about ten times more than Au NPs
[28]. In addition, by appropriately selecting the synthesis procedure
concentration in water - detected for PVDF-25% Ag NPs membrane - is [29], it is possible to control the size and the shape of Ag NPs. Ac-
about three orders of magnitude lower than AgOH solubility. In gen- cordingly, our synthesis method allowed us (see Section 3.1) to obtain
eral, these values are lower than those reported in literature and related NPs with diameter of ~30 nm (Fig. 1) in order to maximize photo-
to the solubilization of Ag NPs dispersed in water, as a result of the thermal effects.
shielding effect of the hydrophobic polymeric matrix [26]. The incre- Ultraviolet-visible (UV–Vis) spectra of manufactured membranes
ment of dissolution rate of Ag NPs in 0.5 M NaCl solution is in agree- (Fig. 3) demonstrated that the maximum absorbance intensity is
ment with previous studies [26,27]. slightly affected by the polymeric matrix, ranging between 420 and
Incorporation of citrate-capped Ag NPs into PVDF matrix reduced 440 nm, coherently with the wavelength of the plasmonic resonance of
the contact angle to water: as the percentage of Ag NPs increased from 0 Ag NPs [30].
to 25%, an overall reduction of the contact angle by ~7% (down to Fig. 4 illustrates the difference between outlet and inlet feed tem-
80.3 ± 3.6°) was observed. Remarkably, only a very small increase of perature for 0.5 M NaCl solution, as resulting from the heat flux asso-
contact angle was noticed after 6 months, presumably due to the ciated to the radiation of the UV lamp and the thermoplasmonic effects
aforementioned slight dissolution of Ag NPs. of irradiated Ag NPs. The temperature difference gradually increased
The heat power Qi carried by the single i-th NP is directly related to with time, getting a stationary behavior in about 30 min for the VMD
its absorption cross section σabs by the following relation [14]: setup used. When the cell was operated with unloaded PVDF mem-
branes (qplasm = 0), the energy associated to the UV radiation

4
A. Politano et al. Desalination xxx (xxxx) xxx–xxx

The analysis of structural characteristics shown in Table 1 clearly

indicates that the extent of morphological variations associated to the
addition of nanofillers in the membranes (primarily in terms of porosity
and pore size) is not enough to explain the noteworthy improvement of
the transmembrane flux, which conversely is determined by thermo-
25%wt AgNPs
Absorbance (A.U.)

plasmonic effects.

20%wt AgNPs 4.2. Energy analysis

The study of temperature changes at the membrane interface, due to

15%wt AgNPs polarization phenomenon, as a function of the NP content can shed
light on the influence of photothermal plasmonic excitation on the
overall MD performance. A temperature polarization factor (TPF) is
defined as:
AgNPs in DMF
T fm
Pristine PVDF TPF = × 100
T fb (7)
350 400 450 500 550
with TPF < 100 in conventional VMD.
Wavelength (nm)
While bulk temperature was experimentally measured, interfacial
Fig. 3. UV–Vis characterization of mixed matrix PVDF membranes incorporating dif- temperature was estimated by Eq. (2). The TPF, predicted for unloaded
ferent amounts of Ag NPs. The color gradation of membrane surface, from white to PVDF membrane (qp = 0) under the operating conditions adopted in
brown, as a function of NP content is shown on the left side. (For interpretation of the
this study, was 99.35% for pure water and 98.25% for 0.5 M NaCl so-
references to color in this figure legend, the reader is referred to the web version of this
article.)
lution. For salt solution, the lower TPF reflects the lower heat transfer
coefficient (660 W/m2 h) with respect to pure water (1010 W/m2 h).
The thermal energy produced by plasmonic nanofillers reversed the
5 polarization effect, leading (i) to an interface feed temperature higher
that in the bulk and (ii) to a TPF > 100%. Fig. 6 displays the behavior
of the TPF (squares) as a function of the content of Ag NPs for VMD tests
Bulk temperature increase (°C)

on 0.5 M NaCl solution. It is evident that, by increasing the amount of

4
25%wt AgNPs Ag NPs, the TPF increases up to values higher than 106; likewise, the
interfacial temperature (circles) increases with the amount of Ag NPs.
20%wt AgNPs To understand experimental results shown in Figs. 4–6, one has to
3 15%wt AgNPs
consider that - when illuminating a large amount of metal NPs dis-
persed in a medium - the temperature profile throughout the system
may no longer be localized around each NP as nanometric source of
heat, due to thermal collective effects [21,22]. Baffou et al. [23] built a
Pristine PVDF
2 theoretical model for reproducing the temperature distribution
throughout large assemblies of metal NPs. In details, they considered
two well-distinct regimes: (i) the “temperature confinement regime”
with a temperature distribution essentially confined at the vicinity of
1 each NP and (ii) a “delocalization regime” characterized by a smooth
temperature profile throughout the surrounding medium [23]. There-
fore, the total temperature increase ΔT experienced by the system is
0 given by the sum of a term ΔTs associated to the self-contribution of a
0 20 40 60 80 single nano-heater, and of a term ΔText, due to the heat delivered by the
Time (min) other NPs located in proximity of the NP. Specifically:
2⋅σabs⋅I
Fig. 4. Time-dependent difference between outlet and inlet feed temperature (30 °C) for ΔT0s =
0.5 M NaCl solution. 4πkd (8)

and, for uniform and circular illumination of an infinite array:

determined a temperature difference by ~2.5 °C. For PVDF membranes
loaded with Ag NPs, the illumination of metal NPs with wavelength σabs⋅I 1 D ⎛ 2 A⎞
ΔT0ext ≈ ⎜1 − ⎟

close to their plasmonic resonance turned them into local nano-heaters, k 4 A⎝ πD⎠ (9)
thus determining an enhancement of the outlet feed temperature. This
where subscript “0” identifies the center of an ensemble of plasmonic
thermoplasmonic effect increased with Ag content: the highest tem-
NPs.
perature increase, i.e. ~4 °C, was detected for PVDF-25% Ag NPs
By considering a NP diameter d of 30 nm, an approximate inter-
membrane.
particle distance p of 2 · 10−7 m for PVDF-25% Ag NPs, a unit cell area
Transient transmembrane flux profiles measured over an operation
A for a NP square lattice of p2 = 4 · 10−14 m2, an irradiance I of
time of 1 h are illustrated in Fig. 5 for both pure water (panel a) and
2.3 · 104 W/m2, a diameter of the irradiated area D of 5.2 · 10−2 m, an
0.5 M NaCl solution (panel b). Vapor fluxes through unloaded PVDF are
average thermal conductivity k of 0.37 W/mK (estimated by the ar-
noticeably lesser than through MMM embedding NP fillers. Specifically,
ithmetic average between the thermal conductivity of PVDF membrane
fluxes through PVDF-25% Ag NPs, evaluated at one hour, are about
with porosity of 0.22 [3] and water), and a NP absorption cross section
nine-fold (feed: 0.5 M NaCl solution) and eleven-fold (feed: H2O)
σabs of 2 · 10−15 m2 [31], the total ΔT is estimated to be ~42 K. Under
greater with respect to the corresponding values for virgin PVDF
these assumptions, the self-contribution term ΔT0s is largely negligible
membranes.
with respect to collective effects produced by the lattice of NPs. Since

5
A. Politano et al. Desalination xxx (xxxx) xxx–xxx

40 40

(b)

Transmembrane vapour flux (l/m2h)

(a)

Transmembrane vapour flux (l/m2h) 30 30

20 20
PVDF-25%AgNPs
PVDF-20%AgNPs
PVDF-15%AgNPs
10 PVDF-25%AgNPs 10
PVDF-20%AgNPs
PVDF-15%AgNPs
PVDF-REF PVDF-REF

0 0
0 20 40 60 0 20 40 60
Time (min) Time (min)
Fig. 5. Time-dependent profile of the transmembrane flux through PVDF membranes loaded with different amounts of Ag nanofillers. PVDF-REF denotes the unloaded PVDF membrane
assumed as a reference. The feed is: (a) pure H2O; (b) 0.5 M NaCl.

Fig. 6. Steady-state behavior of the interfacial temperature (circles) and TPF (squares) as
a function of the content of Ag NPs entrapped in PVDF membranes. Feed: 0.5 M NaCl
solution.

Eqs. (8) and (9) do not take into account the cooling effect associated to
water evaporation at the membrane interface, the predicted value of ΔT
is obviously higher than the increase of bulk temperature estimated by
energy balance in the VMD unit as per Eq. (2), i.e. 23.3 K for PVDF-25%
Ag NPs operated on 0.5 M NaCl solution (see Fig. 6). However, this
theoretical approach provides a useful tool for outlining the potential-
ities of plasmonics in thermal processes occurring at solid-liquid in-
terfaces.
Fig. 7 quantitatively illustrates the different terms of the energy
balance expressed by Eq. (4). Interestingly, for tests with pure water,
data show that the plasmonic heat flux overcomes by about 1.7-fold the Fig. 7. Heat fluxes in thermoplasmonic VMD. Feed: (a) pure H2O; (b) 0.5 M NaCl solution.
heat flux associated to the evaporation through MMM, doubling for
0.5 M NaCl solution. where A is the membrane area. EE is enhanced at increasing con-
Energy efficiency (EE) provides a quantitative estimation of the centration of NPs within the membrane matrix, reaching a maximum of
percentage of the heat stream (absorbed UV radiant flux and plasmonic 32.1% and 28.1% when using pure water and 0.5 M NaCl as feed
heat flux) that is effectively used to promote water evaporation through stream, respectively (Table 3). The exceeding energy merely increases
the membrane. EE is thus defined as: the temperature of the feed outlet stream and might be recovered by
Evaporation Heat Flux Jw λA conventional heat exchanger systems or innovative salinity gradient
EE = = power technologies [32].
Total Heat Flux ̇ p (T fb, out − T fb, in ) + Jw λA
mc (10)

6
A. Politano et al. Desalination xxx (xxxx) xxx–xxx

Table 3 4.4. Tests on real seawater

Energy analysis for thermoplasmonic VMD.
The chemical composition of real seawater samples collected from
Feed Membrane EE (%)
Tyrrenian sea is reported in Table 4. VMD tests were carried out using
Pure water PVDF-REF 8.10 PVDF-25% Ag NPs membrane, i.e. the one exhibiting the highest per-
PVDF-15% Ag NPs 20.6 formance.
PVDF-20% Ag NPs 30.0
With respect to 0.5 M NaCl solution, the average vapor flux in a
PVDF-25% Ag NPs 32.1
0.5 M NaCl PVDF-REF 6.06 VMD experimental test cycled twice was 6% lower (Fig. 8). The slightly
PVDF-15% Ag NPs 14.9 reduced water activity coefficient of real seawater (0.97), estimated by
PVDF-20% Ag NPs 28.1 using PHREEQC speciation software [33], gives a partial explanation
PVDF-25% Ag NPs 28.4 this result.
Analysis of the distillate revealed only traces of Na+ (1.0 ppm),
Mg2+ (0.12 ppm), Cl− (1.8 ppm) and SO42− (0.11 ppm). The low
Table 4
Chemical composition of feed seawater (Tyrrenian sea, location: electrical conductivity (5.9 μS/cm), the negligible organic content
Amantea, Italy). (TOC = 0.013 mg/l) and the absence of dispersed colloidal NPs in the
condensed permeate confirmed the good stability of the membrane
pH 8.0 against wetting.
Temperature (°C) 25
Density (kg/l) 1.02
Ca2+ (ppm) 492.5 5. Conclusions
Mg2+ (ppm) 2120.7
Na+ (ppm) 11,941 In this work, we have applied the innovative concept of thermo-
K+ (ppm) 348
plasmonics in order to prove that temperature polarization phenom-
Cl− (ppm) 20,975
Br− (ppm) 116.5 enon in MD can be withdrawn by using UV-irradiated Ag NPs, acting as
HCO3− (ppm) 140 nano-heaters resonating at the Ag plasma frequency, so as to produce
SO42− (ppm) 2192 collective thermal effects. In principle, this preliminary application to
Conductivity (μS/cm) 68,000 seawater desalination opens the way to the exciting perspective of an
TOC (mg/l) 4.1
eco-friendly and economically feasible membrane technology for solar
Silt density index (SDI15)† 2.4
Water activity 0.97 steam generation.
By exploiting the inherent dependence of mass transfer on tem-
†
SDI15 measurement procedure from ASTM standard test perature gradient, we made a remarkable step-change in the fields of
method D 4189-82. thermoplasmonics and MD, showing that membranes manufactured
with hydrophobic polymers and filled with size-selected plasmonic NPs
0
UV lamp ON UV lamp ON can provide remarkable improvement in the system performance, both
in terms of transmembrane flux and thermal efficiency.
Transmembrane flux (L/m2h)

30 Pure water
The stability of NPs-charged membranes is a crucial issue in water
0.5M NaCl treatment operations. In our innovative system, the release of silver
Seawater
ions and NPs in the permeate was prevented by the shielding effect of
20 the hydrophobic polymeric matrix and the peculiar VMD configuration,
in which there is not a liquid distillate contacting the membrane. Note
that the use of thermoplasmonic membranes could be also extended to a
10
wide spectrum of applications where diffusion-driven separation me-
chanisms depends on temperature, beyond seawater desalination (see
0 as an example, the reviews in Ref. [14,19]).
0 100 200 300 400 500
Time (min)
Acknowledgments
Fig. 8. Transmembrane flux of vapor through a 25% w/w Ag-NPs loaded PVDF mem-
brane. Feed typology: pure water, 0.5 M NaCl and real seawater. We thank Guillaume Baffou for helpful discussions.

4.3. Permeate quality References

Analytical tests were carried out on distillate samples in order to [1] E. Drioli, A. Ali, F. Macedonio, Membrane distillation: recent developments and
check the quality of desalted water. Incorporation of citrate-capped Ag perspectives, Desalination 356 (2015) 56.
[2] E. Drioli, A. Criscuoli, E. Curcio, Membrane contactors and catalytic membrane
NPs into PVDF matrix, although decreasing moderately the value of reactors in process intensification, Chem. Eng. Technol. 26 (2003) 975.
water contact angle (Table 1), did not cause pore wetting: in experi- [3] E. Curcio, E. Drioli, Membrane distillation and related operations — a review, Sep.
mental tests carried out on both pure water and 0.5 M NaCl solution, Purif. Rev. 34 (2005) 35.
[4] M.A.E.-R. Abu-Zeid, Y. Zhang, H. Dong, L. Zhang, H.-L. Chen, L. Hou, A compre-
protracted over 7 days/10 h of operation per day, the permeate con- hensive review of vacuum membrane distillation technique, Desalination 356
ductivity ranged within 0.9–2.1 μS/cm, thus providing evidence of the (2015) 1.
negligibility of salt leakage. Additionally, concentration of Ag NPs and [5] P. Wang, T.-S. Chung, Recent advances in membrane distillation processes: mem-
brane development, configuration design and application exploring, J. Membr. Sci.
Ag+ ions were below the detectability threshold of DLS and HR-CS AAS 474 (2015) 39.
instruments, as a result of both the specific evaporative transport me- [6] S. Al-Obaidani, E. Curcio, F. Macedonio, G. Di Profio, H. Al-Hinai, E. Drioli,
chanism in MD (non-volatile Ag+ ions dissolved in the feed stream are Potential of membrane distillation in seawater desalination: thermal efficiency,
sensitivity study and cost estimation, J. Membr. Sci. 323 (2008) 85.
retained), and the peculiar VMD configuration, where condensed dis- [7] L. Martı́nez-Dı́ez, M.I. Vazquez-Gonzalez, Temperature and concentration polar-
tillate is not directly in contact with the membrane. ization in membrane distillation of aqueous salt solutions, J. Membr. Sci. 156
No significant changes (< 5%) in terms of transmembrane flux were (1999) 265.
[8] A. Ali, F. Macedonio, E. Drioli, S. Aljlil, O. Alharbi, Experimental and theoretical
observed during 10-h experiments repeated over 6 months.
evaluation of temperature polarization phenomenon in direct contact membrane

7
A. Politano et al. Desalination xxx (xxxx) xxx–xxx

distillation, Chem. Eng. Res. Des. 91 (2013) 1966. [22] H.H. Richardson, M.T. Carlson, P.J. Tandler, P. Hernandez, A.O. Govorov,
[9] N. Hengl, A. Mourgues, M. Belleville, D. Paolucci-Jeanjean, J. Sanchez, Membrane Experimental and theoretical studies of light-to-heat conversion and collective
contactor with hydrophobic metallic membranes: 2. Study of operating parameters heating effects in metal nanoparticle solutions, Nano Lett. 9 (2009) 1139.
in membrane evaporation, J. Membr. Sci. 355 (2010) 126. [23] G. Baffou, P. Berto, E. Bermúdez Ureña, R. Quidant, S. Monneret, J. Polleux,
[10] J.B. Herzog, M.W. Knight, D. Natelson, Thermoplasmonics: quantifying plasmonic H. Rigneault, Photoinduced heating of nanoparticle arrays, ACS Nano 7 (2013)
heating in single nanowires, Nano Lett. 14 (2014) 499. 6478.
[11] M. Essone Mezeme, C. Brosseau, Engineering nanostructures with enhanced ther- [24] P.C. Lee, D. Meisel, Adsorption and surface-enhanced Raman of dyes on silver and
moplasmonic properties for biosensing and selective targeting applications, Phys. gold sols, J. Phys. Chem. 86 (1982) 3391.
Rev. E 87 (2013) 012722. [25] H. Baida, P. Billaud, S. Marhaba, D. Christofilos, E. Cottancin, A. Crut, J. Lermé,
[12] R. Rodríguez-Oliveros, J.A. Sánchez-Gil, Gold nanostars as thermoplasmonic na- P. Maioli, M. Pellarin, M. Broyer, Quantitative determination of the size dependence
noparticles for optical heating, Opt. Express 20 (2012) 621. of surface plasmon resonance damping in single Ag@SiO2 nanoparticles, Nano Lett.
[13] M. Zhu, G. Baffou, N. Meyerbröker, J. Polleux, Micropatterning thermoplasmonic 9 (2009) 3463.
gold nanoarrays to manipulate cell adhesion, ACS Nano 6 (2012) 7227. [26] X. Li, J.J. Lenhart, Aggregation and dissolution of silver nanoparticles in natural
[14] G. Baffou, R. Quidant, Thermo-plasmonics: using metallic nanostructures as nano- surface water, Environ. Sci. Technol. 46 (2012) 5378.
sources of heat, Laser Photonics Rev. 7 (2013) 171. [27] K.A. Huynh, K.L. Chen, Aggregation kinetics of citrate and polyvinylpyrrolidone
[15] Z. Liu, X. Liu, S. Huang, P. Pan, J. Chen, G. Liu, G. Gu, Automatically acquired coated silver nanoparticles in monovalent and divalent electrolyte solutions,
broadband plasmonic-metamaterial black absorber during the metallic film-for- Environ. Sci. Technol. 45 (2011) 5564.
mation, ACS Appl. Mater. Interfaces 7 (2015) 4962. [28] A. Lalisse, G. Tessier, J.r. Plain, G. Baffou, Quantifying the efficiency of plasmonic
[16] Z. Liu, P. Zhan, J. Chen, C. Tang, Z. Yan, Z. Chen, Z. Wang, Dual broadband near- materials for near-field enhancement and photothermal conversion, J. Phys. Chem.
infrared perfect absorber based on a hybrid plasmonic-photonic microstructure, C 119 (2015) 25518.
Opt. Express 21 (2013) 3021. [29] Y. Sun, Y. Xia, Shape-controlled synthesis of gold and silver nanoparticles, Science
[17] X. Chen, Y. Chen, M. Yan, M. Qiu, Nanosecond photothermal effects in plasmonic 298 (2002) 2176.
nanostructures, ACS Nano 6 (2012) 2550. [30] V. Amendola, O.M. Bakr, F. Stellacci, A study of the surface plasmon resonance of
[18] Z.J. Coppens, W. Li, D.G. Walker, J.G. Valentine, Probing and controlling photo- silver nanoparticles by the discrete dipole approximation method: effect of shape,
thermal heat generation in plasmonic nanostructures, Nano Lett. 13 (2013) 1023. size, structure, and assembly, Plasmonics 5 (2010) 85.
[19] A. Politano, A. Cupolillo, G. Di Profio, H. Arafat, G. Chiarello, E. Curcio, When [31] D.D. Evanoff, G. Chumanov, Synthesis and optical properties of silver nanoparticles
plasmonics meets membrane technology, J. Phys. Condens. Matter 28 (2016) and arrays, ChemPhysChem 6 (2005) 1221.
363003. [32] R.A. Tufa, E. Curcio, W. Van Baak, J. Veerman, S. Grasman, E. Fontananova, G. Di
[20] A. Politano, P. Argurio, G. Di Profio, V. Sanna, A. Cupolillo, S. Chakaraborthy, Profio, Potential of brackish water and brine for energy generation by salinity
H.A. Arafat, E. Curcio, Photothermal membrane distillation for seawater desalina- gradient power-reverse electrodialysis (SGP-RE), RSC Adv. 4 (2014) 42617.
tion, Adv. Mater. 29 (2017) 1603504. [33] B.J. Merkel, B. Planer-Friedrich, D. Nordstrom, Groundwater Geochemistry, A
[21] A.O. Govorov, W. Zhang, T. Skeini, H. Richardson, J. Lee, N.A. Kotov, Gold nano- Practical Guide to Modeling of Natural and Contaminated Aquatic Systems 2,
particle ensembles as heaters and actuators: melting and collective plasmon re- (2005).
sonances, Nanoscale Res. Lett. 1 (2006) 84.