# Three-dimensional open architecture enabling salt-rejection solar evaporators with boosted water production efficiency

### Evaporation structure design and fabrication

For conventional salt-rejection solar evaporation systems, water evaporation is confined to the solar absorber surface, and the salt backflow is accompanied by an undesired heat dissipation from the solar absorber to bulk water, thus resulting in a low evaporation rate. This limitation can be solved to a considerable extent by our 3D evaporator. As illustrated in Fig. 1a, the top surface of our evaporator is a solar absorber layer used for light-to-heat conversion to generate vapour. Beneath the solar absorber are a number of vertically aligned MTBs connecting the saline water to the solar absorber. MTBs have hydrophilic microchannels that can pump saline water to the solar absorber via a capillary force. Furthermore, excessive salt can flow back into the bulk water through these brine-filled microchannels via diffusion and convection (Fig. 1b-1). The adequate mass transfer via a high density of MTBs ensures a continuous water supply and an efficient salt backflow, thus enabling a unique salt rejection capability. Unlike conventional salt-rejection systems, where the heat conducted from the solar absorber to the bulk water is simply dissipated and considered “wasted,” the MTBs can efficiently recover this conductive heat to generate additional vapour from the brine flowing through their microchannels (Fig. 1b-2). The microchannels within the MTBs and macrochannels between the spaced MTBs together form a highly open structure that allows the generated vapour to be easily released from the MTB surfaces in all directions. We envision that by optimizing the MTB height, conductive heat can be largely confined in them for vapour generation, thereby significantly improving the water evaporation efficiency.

We achieved the designed structure by fabricating the top solar absorber layer by loading carbon nanotubes (CNTs) with a diameter of about one hundred nanometres on a glass fibre membrane (GFM). The solar absorption of wet CNT-coated GFM can reach ~96% (Fig. 1c) because of the porous fibrous light-trapping structure (Fig. 1d) and the inherent black property of the CNT27. Considering their abundant hydrophilic microchannels formed by intertwined glass fibres (Fig. 1e), the GFMs were also selected for use as MTBs. A GFM can immediately absorb a water droplet upon touching it because of its high affinity to water (Fig. 1f). Moreover, vertically aligned GFMs (i.e., MTBs) can pump water to 25 cm height in 60 min, demonstrating its strong capillary force for water transfer (Fig. 1g). A complete evaporation system was fabricated by assembling a number of MTBs and the solar absorber in a plastic frame (Fig. 1h, 1i and Fig. S1).

### Salt rejection capability

To avoid salt crystallization, excess salt must be efficiently transported back to maintain the top surface salinity below the saturation point. In this system, salt can be rejected via diffusion and convection through brine-filled microchannels under the driving force of the concentration gradient (osmosis) and gravity25. Its mass flow rate ((J)) can be described by the diffusion–convection equation as follows28,29:

$$J={J}_{{diff}}+{J}_{{conv}}={nA}varepsilon ({k}_{d}({C}_{{evp}}-{C}_{0})/l+{k}_{c}({rho }_{{evp}}-{rho }_{0}))$$

(1)

where ({J}_{{diff}}) and ({J}_{{conv}}) are the mass flow rate caused by diffusion and convection, respectively; (n) is the number of MTBs; (A), (varepsilon), and (l) are the cross-section area, porosity, and height of the MTBs, respectively; ({k}_{d}) and ({k}_{c}) are the diffusion and average convective coefficients of salt, respectively; ({C}_{{evp}}) and ({C}_{0}) are the salt concentrations on the evaporation surface and in the bulk saline water, respectively; and ({rho }_{{evp}}) and ({rho }_{0}) are the salt solution densities on the evaporation surface and in the bulk saline water, respectively.

In Eq. (1), the mass transport rate is proportional to the bridge number (n). We validated this relation by fabricating MTB structures with different bridge numbers ranging from 2 to 32 [Fig. 2a, cross-section area ((A)): ~0.135 cm2; height ((l)): 3 cm; porosity ((varepsilon)): ~65%] and evaluating their evaporation performance using high-salinity water (10 wt.% NaCl). The evaluation was performed under 1 sun illumination for 12 h. Figure 2b shows that salt crystals massively accumulated on the 2-bridge evaporator surface because of insufficient mass transfer. This salt accumulation was mitigated with increase in the bridge number. For the evaporator containing 32 MTBs, no salt crystals were observed on the surface after the 12 h operation (Fig. 2b). At an insufficient number of MTBs (≤16), the evaporation rate gradually decreased as the vapour generation progressed because of the increased evaporation surface salinity (Fig. 2c, see the corresponding mass change curves in Fig. S2). In contrast, with sufficient MTBs (e.g., 32 bridges), the excess salt can be efficiently rejected to maintain the evaporation surface at a relatively low salinity. Remarkably, the evaporation rate of the 32-bridge evaporator was ~1.44 kg/m2/h without degradation during the 12 h operation.

Subsequently, we performed a complementary experiment to more intuitively demonstrate the salt backflow introduced by the 32-bridge evaporator. In this experiment, the evaporator was placed in a high-concentration saline water (10 wt.% NaCl solution) and exposed to 1 sun illumination, and 1 g of NaCl salt was added on its surface (upper panel, Fig. 2d). It was seen that during vapour generation, the added salt was gradually dissolved and completely removed in 11 h (lower panel, Fig. 2d; more details in Fig. S3). This experiment demonstrated that the salt backflow rate of the 32-bridge evaporator in the 10 wt.% NaCl solution was higher than the salt generation rate, thus confirming the salt rejection feature of the proposed MTB architecture. We further increased the brine salinity to test the maximum applicable salt concentration of this evaporator. Because the effects of diffusion and convection backflow decreased as the salinity (i.e., ({C}_{0}) and ({rho }_{0})) increased, salt started to crystallize at the edges of the solar absorber after 12 h operation when 14 wt.% NaCl solution was used for the test (Fig. S4). Based on the corresponding evaporation rate, the salt backflow along the MTBs was calculated as ~1.1 g/cm2/h. Interestingly, this unique mass transport feature is intertwined with its heat transport feature, as demonstrated in the subsequent section.

### Heat management

We fabricated 32-bridge evaporators with different bridge heights (Fig. 3a) and evaluated their evaporation performance. Under dark conditions, the evaporator without MTBs (i.e., bridge height: 0 cm) exhibited a natural evaporation rate of 0.15 kg/m2/h, which became more pronounced with the incorporation of MTBs due to the increased surface area (Fig. S5). Specifically, it linearly increased by ~0.04 kg/m2/h for every 1 cm increase in the MTB height. Under 1 sun illumination, the evaporation rate of the evaporator without MTBs was only 0.99 kg/m2/h because of the massive conductive heat dissipation to the bulk water (Fig. 3b, see the mass change curves in Fig. S6). The MTB usage considerably promoted solar evaporation. The evaporation rate increased to 1.58–1.73 kg/m2/h when the bridge height reached 2–5 cm (Fig. 3b). These values are even higher than the theoretical upper limit for solar evaporation (~1.44 kg/m2/h, Supplementary Note 1 and Fig. S7), which can be attributed to the natural evaporation contribution (Fig. S8). When the MTB height exceeded 3 cm, the evaporation rate increased by ~0.04 kg/m2/h for every 1 cm increase in MTB height (Fig. 3b), which was consistent with the result obtained under dark conditions. This consistency suggests that the 3 cm height is sufficient for the MTB structure to maximize solar evaporation (note that additional increase in MTB height only increases natural evaporation). To reveal the mechanism of this observation, we analyzed the heat transport in this unique architecture.

The energy loss channels for this evaporation system primarily include conductive heat loss into the bulk water (({P}_{{cond}.})), radiative heat loss (({P}_{r{ad}.})), and convective heat loss to the environment (({P}_{{convec}.})). Therefore, the power flux available for evaporation (({P}_{{evp}})) can be described as follows16:

$${P}_{{evp}}={P}_{{solar}}-{P}_{{cond}.}-{P}_{{rad}.}-{P}_{{convec}.}$$

(2)

where the solar energy input ({P}_{{solar}}={{{{{rm{alpha }}}}}}{C}_{{opt}}{q}_{i}); ({{{{{rm{alpha }}}}}}) is the light absorption coefficient; ({C}_{{opt}}) is the optical concentration; and ({q}_{i}) is the direct solar illumination. The conductive heat flux ({P}_{{cond}.}=k({T}_{{sa}}-{T}_{{bw}})/l), where (k) is the thermal conductivity; ({T}_{{sa}}) and ({T}_{{bw}}) are the temperatures of the solar absorber and the bulk water, respectively; and (l) is the heat conduction path referring to the MTB height in our model. The radiative heat flux ({P}_{{rad}.}=varepsilon {{{{{rm{sigma }}}}}}({{T}_{1}}^{4}-{{T}_{2}}^{4})), while the convective heat flux ({P}_{{convec}.}=hleft({T}_{1}-{T}_{2}right),) (varepsilon) is the optical emission, ({{{{{rm{sigma }}}}}}) is the Stefan–Boltzmann constant, (h) is the convection heat transfer coefficient, and ({T}_{1}) and ({T}_{2}) are the temperatures of the evaporator and environment, respectively.

The energy loss caused by the heat transfer from the top surface to the bulk water (i.e., ({P}_{{cond}.})) can be minimized by increasing the MTB height (i.e., (l)) to confine the conductive heat within the MTB structure. This effect was visualized using infrared imaging to display the temperature gradients along MTBs with different heights. The results showed that the temperature at the bottom of the evaporator was similar to the ambient temperature when the MTB height reached or exceeded 3 cm (Fig. S9). This temperature distribution agreed well with the simulation modelled by COMSOL (Fig. S10). To obtain more insights into the heat transport in the architecture, we carefully recorded the internal temperature variation at different distances to the solar absorber under solar illumination. The results showed that the temperature stabilized after 60 min, when the internal temperature at 3 cm to the solar absorber was similar to that of the surrounding environment (Fig. 3c), indicating that the conductive heat was completely confined in the top 3 cm of the MTB structure. This confinement effect was also demonstrated by the temperature change of the bulk water (Fig. 3d): for the evaporator without MTBs, the bulk water temperature increased from ~21 to ~26.2 °C after a 3 h operation due to the continuous heat input (left panel); for the evaporator with 3-cm MTBs, however, the bulk water temperature was maintained at room temperature (~21.3 °C) (right panel), thus confirming the suppression of heat dissipation into the bulk water.

Importantly, the confined heat energy can be exploited to generate additional vapour from the MTB surfaces, which can be efficiently released via the highly open interbridge spaces. To reveal this additional vapour generation from the vertical surfaces of MTBs, we used an evaporator having 32 MTBs (3 cm high) to perform a control experiment. In this experiment, the evaporator body was enclosed with an airtight polypropylene film, thus leaving only the upper surface exposed to the open space for vapour release (Fig. S11). After a 3 h operation, many water droplets condensed on the inner film surface, thus confirming that the MTBs released vapour (Fig. 3e). Compared to the completely open evaporator, the evaporation rate of the partially enclosed system decreased by ~31% (Fig. S12), demonstrating the importance of the open-channel design for enhanced interfacial evaporation.

Furthermore, we performed a cycling experiment to evaluate the evaporator stability. In each cycle, the evaporator ran for 12 h under 1 sun illumination and in a dark environment for another 12 h to simulate day and night alternation. Figure 3f shows that during this long-term test (with 10 wt.% NaCl solution), the mass change of the NaCl solution in each cycle linearly evolved and the evaporation rate stabilized at ~1.44 kg/m2/h. No performance degradation was observed after a seven-day cycling experiment.

Compared with the previously reported salt-rejection evaporators (evaporation rate: from 1.24 to 1.28 kg/m2/h for 10 wt.% NaCl solution)9,26,30, our evaporator demonstrated a higher evaporation rate under similar conditions due to the heat confinement effect and the natural evaporation contribution. However, high evaporation efficiency alone is not sufficient for water production applications. If the evaporated moisture is not collected, it can only be considered as a pollutant to the environment considering that it has the greatest greenhouse effect among various components in the atmosphere31. Water collection that is equally important as vapour generation has been largely ignored in many previous studies on salt-rejection evaporators.

Therefore, we enclosed the evaporator with a transparent cover made of polymethyl methacrylate (PMMA) plates, creating a system that can produce water by condensing the evaporated moisture, and investigated the effects of bridge number and bridge height on the water production capacity of this system (Fig. S13a). When the bridge height was fixed at 3 cm, the amount of collected water increased with the number of bridges (Fig. S13b), which is consistent with the observation in the open system, confirming that the enhanced salt backflow facilitates water evaporation. When the bridge number was fixed at 32, the amount of collected water increased with the bridge height and reached the maximum at 3 cm, while further increasing the bridge height did not produce more water (Fig. S13c). This result is consistent with the conclusion above that 3 cm is sufficient to confine the conductive heat while further increasing bridge height only increases natural evaporation that does not contribute to water production. According to the three-hour test results, the water production rate of the enclosed evaporator in the optimal configuration (32 bridges; 3 cm high) is calculated to 0.68 kg/m2/h (Fig. S13).

We also investigated the water generation performance of the enclosed system under different salinity conditions using NaCl solutions (3.5−20 wt.%). The results showed that the water production efficiency monotonically decreased from ~0.73 kg/m2/h for 3.5 wt.% NaCl solution to ~0.63 kg/m2/h for 20 wt.% NaCl solution (Fig. S14a). The relatively low water production efficiency associated with the high-salinity brines is mainly due to their low saturated vapour pressure, partly due to the decreased photothermic conversion efficiency caused by salt precipitation. For instance, when using brine containing 20 wt.% NaCl, salt precipitation emerged at the periphery of the evaporator after three hours of testing (Fig. S14b).

### Field tests

As per the recently announced “best practice for solar water production”32, the daily water yield is an important evaluation criterion that deserves additional consideration in practical implementations. Therefore, we prepared closed system based on the MTB structure and measured their water generation capacity under practical outdoor conditions.

#### Rooftop experiment

The fabricated solar-driven water generation system has a 15 × 26 cm2 evaporator area (see Fig. S15). We first tested the system on the rooftop in KAUST, Thuwal, Saudi Arabia (Fig. S16). In this experiment, we employed the discharged water from an RO system of the KAUST Seawater Desalination Plant as the source water (salinity: ~8.7%). Our daily evaluation started at 8:00 and ended at 17:00. As shown in Fig. 4a, the evaporator surface was heated by solar light to a temperature 4–15 °C higher than the environment. However, the temperature at the bridge bottom was almost the same as the environment temperature, indicating that the conductive heat was confined, with only a small amount transferred to the bulk water. Consequently, saline water can be efficiently evaporated and condensed at the cover surface for the water collection. Figure 4b and Supplementary Movie 1 illustrate the relevant details. The total collected water was ~175 ml, of which ~110 ml flowed in the graduated cylinder, and ~65 ml was retained in the PMMA cover. Based on the evaporator area (390 cm2), the daily water productivity was calculated as ~5.0 L/m2. We measured the ion contentions of our water samples to evaluate the water quality. Compared with the discharged water from the RO plant, the ion concentration of condensed water was reduced by at least four orders of magnitude, thus fully meeting the WHO drinking water requirements (Fig. 4c).

We calculated its practical solar–water collection efficiency of the system, ({eta }_{{prac}}), using Eq. (3):

$${eta }_{{prac}}={m}_{{cond}}{h}_{{lv}}/left({A}_{{evp}}int {q}_{{solar}}left(tright){dt}right)$$

(3)

where ({m}_{{cond}}) is the daily water collection amount; ({h}_{{lv}}) is the total enthalpy of the liquid–vapour phase transition; ({A}_{{evp}}) is the evaporator area; and ({q}_{{solar}}) is the time-dependent solar flux. Benefiting from the highly efficient vapour generation, the overall solar–water collection efficiency of our system reached ~41.6%, representing a considerable improvement compared to the previously reported salt-rejection solar evaporation systems (e.g., maximum efficiency of a rooftop system: ~24%25). We performed a continuous test from Apr. 7 to Apr. 11, 2022 to evaluate the performance stability (Fig. 4d). The daily water collection rate fluctuated in the range of 4.7–5.2 L/m2 depending on the specific solar insolation of the day. The corresponding solar–water collection efficiency was 39%–42%. Remarkably, no salt accumulation was observed during this five-day outdoor operation (Fig. 4e). These results demonstrate the potential of the fabricated evaporator to extract freshwater from the wastewater discharged by RO plants.

#### Floating test

After the 5-day rooftop experiment, the same MTB-based evaporation system was tested in a floating configuration in the Red Sea (salt content: ~4.3%) to demonstrate its potential for practical seawater desalination (Fig. 4f, g). The test started and ended at 8:00 and 17:00, respectively, each day and lasted for five days from Apr. 17 to Apr. 21, 2022. As shown in Fig. 4h, the daily freshwater productivity ranged from 5.0 to 5.8 L/m2 with a stable solar–water collection efficiency of 42%–45%, which was consistent with the rooftop test. This freshwater productivity was approximately two times higher than the previous record of the salt-rejection solar evaporator (~2.5 L/m2 per day)25. The field test demonstrated a high-performance solar evaporator that will help in disaster relief or strengthen the resilience of individuals living on boats and coastal areas.

Source: Resources - nature.com