Molybdenum Disulfide and Tungsten Disulfide as Novel …
199
on salt permeation. The fifth mechanism can be viewed as the possibility of mixing
different ions and observing different salts rejection. At last, entropic differences
can be expressed as a mix of nanopore morphology, charge distribution, and so on
as a result of the possible configurational restrictions, which in turn affects the free
energy barrier through the membrane.
When it comes to cleaning aqueous solutions using membranes, we need to face
a range of ions with different valences. Fortunately, MD simulations on solutions of
water and three cations with different valences (Na
+ , Zn
2+, and Fe
3+ ) indicate that
the higher the valence, the greater the ion rejection through 2D nanoporous MoS 2
and graphene [13]. The desalination rates were confirmed even under a wide range
of test conditions, including high pressures and different nanopore sizes.
The cation charge dependence leads to another question: what if we add more
cations with higher valence to the solution? It would seem very inappropriate at
first, but surprisingly, simulations have shown that this procedure can lead to higher
desalination rates [25]. The possible reason is that the whole membrane-mediated
desalination process is based on the assumption that size matters. Furthermore, the
combination of larger ions and smaller nanopores (large enough to let water pass)
is crucial for efficient selectivity. When high valence ions are added, they aggregate
with counterions to form clusters that in time will be rejected by the membrane.
There are physical and chemical aspects that can be decisive either to improve or
to hinder 2D membrane-based desalination. Unfortunately, very few experimental
studies of ionic conductance through MoS 2 and WS 2 nanoporous membranes with
diameters lower than 2.0 nm have been reported. Almost all of our knowledge is
based on computational simulations and theoretical models. For instance, one important parameter to control the flow through the nanopore is ionic conductance. Perez
et al. [26] developed a continuum model of ionic conductivity for a KCl electrolyte
through a sub-5-nm single-layer MoS 2 nanopore using all-atom MD simulations.
They showed that electrolyte behavior deviates by 50% from bulk properties for
diameters below 2.0 nm: ion pore conductivity is about half of the bulk value for
2.0 nm and only a third when the diameter approaches 1.0 nm. Their results corroborate the idea that the nanopore’s size plays a fundamental role in the desalination
process.
Usually, layer-stacked membranes made of 2D MoS 2 and WS 2 are synthesized
with hundreds to thousands of nanosheets. But this thickness is unlikely to be modeled
in a traditional MD simulation due to computational limitations. Alternatively, we
can take advantage of the fact that there is a relation between measured flux and
membrane thickness. Indeed, if we fit the water flux versus the membrane thickness,
we can get a parabolic dependence in a way that we can estimate the experimental
value. For instance, Wang et al. [27] noticed that the predicted flux decreases as
membrane thickness increases, consistent with their experimental data. They found
that as their model membrane thickness increased to ~500 nm, water flux would
decrease to around 50 L·m
−2 ·h
−1 ·bar
−1 , matching the experimental value.
Whether measuring mass transport properties or the membrane’s mechanical
strength, computer simulations have constantly contributed to the prediction and
confirmation of 2D membranes’ use for desalination. There are many aspects which
199
on salt permeation. The fifth mechanism can be viewed as the possibility of mixing
different ions and observing different salts rejection. At last, entropic differences
can be expressed as a mix of nanopore morphology, charge distribution, and so on
as a result of the possible configurational restrictions, which in turn affects the free
energy barrier through the membrane.
When it comes to cleaning aqueous solutions using membranes, we need to face
a range of ions with different valences. Fortunately, MD simulations on solutions of
water and three cations with different valences (Na
+ , Zn
2+, and Fe
3+ ) indicate that
the higher the valence, the greater the ion rejection through 2D nanoporous MoS 2
and graphene [13]. The desalination rates were confirmed even under a wide range
of test conditions, including high pressures and different nanopore sizes.
The cation charge dependence leads to another question: what if we add more
cations with higher valence to the solution? It would seem very inappropriate at
first, but surprisingly, simulations have shown that this procedure can lead to higher
desalination rates [25]. The possible reason is that the whole membrane-mediated
desalination process is based on the assumption that size matters. Furthermore, the
combination of larger ions and smaller nanopores (large enough to let water pass)
is crucial for efficient selectivity. When high valence ions are added, they aggregate
with counterions to form clusters that in time will be rejected by the membrane.
There are physical and chemical aspects that can be decisive either to improve or
to hinder 2D membrane-based desalination. Unfortunately, very few experimental
studies of ionic conductance through MoS 2 and WS 2 nanoporous membranes with
diameters lower than 2.0 nm have been reported. Almost all of our knowledge is
based on computational simulations and theoretical models. For instance, one important parameter to control the flow through the nanopore is ionic conductance. Perez
et al. [26] developed a continuum model of ionic conductivity for a KCl electrolyte
through a sub-5-nm single-layer MoS 2 nanopore using all-atom MD simulations.
They showed that electrolyte behavior deviates by 50% from bulk properties for
diameters below 2.0 nm: ion pore conductivity is about half of the bulk value for
2.0 nm and only a third when the diameter approaches 1.0 nm. Their results corroborate the idea that the nanopore’s size plays a fundamental role in the desalination
process.
Usually, layer-stacked membranes made of 2D MoS 2 and WS 2 are synthesized
with hundreds to thousands of nanosheets. But this thickness is unlikely to be modeled
in a traditional MD simulation due to computational limitations. Alternatively, we
can take advantage of the fact that there is a relation between measured flux and
membrane thickness. Indeed, if we fit the water flux versus the membrane thickness,
we can get a parabolic dependence in a way that we can estimate the experimental
value. For instance, Wang et al. [27] noticed that the predicted flux decreases as
membrane thickness increases, consistent with their experimental data. They found
that as their model membrane thickness increased to ~500 nm, water flux would
decrease to around 50 L·m
−2 ·h
−1 ·bar
−1 , matching the experimental value.
Whether measuring mass transport properties or the membrane’s mechanical
strength, computer simulations have constantly contributed to the prediction and
confirmation of 2D membranes’ use for desalination. There are many aspects which