Concept and Operation of SVGMD
Unlike conventional photothermal MD where the light-absorbing nanomaterials (e.g. nanoflakes and nanoparticles in the form of dispersion) are deposited onto the surface of membrane or even incorporated into the membrane [11,12,13,14], the key feature of SVGMD is the use of a graphene array-based free-standing, multifunctional light absorber separated from the membrane (Fig. 1a).
Beyond its intrinsic role of harvesting solar energy, the light absorber in the SVGMD system has the following features:
- (1)
It is superhydrophilic and can pump water via graphene nanochannels from the bulk feed solution. A thin water layer is rapidly formed on the surface of the light absorber, and the thermal energy converted from the light illumination can be used to directionally heat the confined water layer, rather than heat the bulk solution (Fig. 1b). Such an excellent heat localization induced by self-guided water transport and engineered water pathways could significantly increase the solar-water energy efficiency.
- (2)
It is standalone (i.e. not attached to the membrane) and thus enables a ‘gap’ between the light absorber and the membrane (Fig. 1c). Salt/contaminant-free vapour (instead of saline/contaminated water in common devices) contacts with the membrane and condenses on the distillate side. It could largely prevent the deposition of the salt/contaminant and thus resolve the fouling problem without any complex chemical modification on the membrane.
- (3)
It is underwater superoleophobic and anti-salt-clogging, which could prevent the accumulation of the salt residues and weaken the adhesion of oil on its surface.
The operational principle of SVGMD is schematically shown in Fig. 1d. The bottom part of the light absorber is dipped into the bulk saline/contaminated feed. The feed water is lifted to the upper part of the light absorber while the oil is rejected underwater, which is attributed by the superhydrophilic and underwater superoleophobic nature of the light absorber. Under illumination, the light-absorbing materials of graphene arrays capture solar light and convert it into thermal energy. Through directional heat transfer, the thin layer of water surrounding the graphene arrays is heated. As a consequence, vapour is generated via liquid/vapour phase change and fills up the gap, leading to an increased vapour pressure. Driven by the pressure gradient between the two sides of membrane (serves as a physical barrier between the distillate flow and the water vapour), the hot vapour transports through the membrane and then condenses at the opposite side due to the cooling effect of distillate flow. Specifically, the membrane plays a key role in microorganism removal. For instance, bacteria could be rejected and absorbed by the membrane, as shown in Fig. S1. Meanwhile, the fouling problem of membrane caused by the direct contact with feed water in conventional MD systems [30] is rationally solved, dramatically improving the membrane lifespans and long-term stability. Detailed comparison between the current SVGMD and conventional photothermal MD systems in terms of system design, water flow, and heat transfer is given in Table S1.
Efficient Solar-Thermal Conversion of P–G–Nifoam
An ideal light absorber of the SVGMD should be free-standing and simultaneously be able to absorb solar light, pump water from feed solution, and reject salt/contaminant (e.g. oil). In the current proof-of-concept work, the light absorber is produced by directly growing graphene arrays on the surface of nickel foam through a one-step PECVD process. The upper-layer graphene array serves as light-absorbing material to capture solar energy when the supporting nickel foam works as the skeleton to ensure its entirely standalone form and mechanical robustness. The as-synthesized hierarchical structure is then coated with PEDOT-PSS, which converts the wettability of graphene array from naturally hydrophobic to superhydrophilic and minimizes the accumulation of salt.
Figure 2a shows a photograph of the light absorber, i.e. P–G–Nifoam, which exhibits a free-standing, flexible, and mechanically robust structure. SEM image of Fig. 2b shows its three-dimensional frameworks. Magnified SEM images in Fig. 2c, d show its upper-layer graphene array before and after PEDOT-PSS coating, respectively. A dense oriented plasma-made nanostructure is deposited on the nickel scaffold, and the span width of an individual graphene nanosheet is approximately 300 nm [31]. The pristine graphene array is naturally hydrophobic (water contact angle of 161.1°, see Fig. S2). After PEDOT-PSS coating, P–G–Nifoam becomes superhydrophilic with a water contact angle of 1.8° (inset of Fig. 2f) and thus can effectively pump water from feed solution via graphene nanochannels. Figure 2e shows fluidic transport across the P–G–Nifoam over 180 s. The height, width, and thickness of the sample are 4.5 cm, 2.0 cm, and 0.4 mm, respectively. The bottom region is dipped into the feed water. Due to its superhydrophilicity and the capillary forces induced between the neighbouring graphene nanosheets, the feed water is rapidly pumped and flows across the sample with a transport direction from the bottom to top. The flow rate within the first 15 s is 0.48 mm s−1, which is comparable with the performance of previously reported material with superior water transport behaviour [32]. Meanwhile, the P–G–Nifoam light absorber exhibits an underwater superoleophobicity (characterized by the oil contact angle of 151.8°, see Fig. S3), making it possible to repel oil from its surface underwater.
The solar-thermal conversion performance of P–G–Nifoam is characterized with an infrared (IR) camera. The sample is exposed to a constant illumination of 1 kW m−2 (i.e. 1 sun), and the IR images at different exposure times are recorded. As shown in Fig. 2g, the surface temperature of P–G–Nifoam rapidly increases from 23.0 to 35.7 °C within 1 s and eventually reaches a steady state (63.2 °C) at 20 s. Figure 2h shows the surface temperature of P–G–Nifoam (at central region) as a function of time at an illumination of 1 kW m−2. For comparison, the temperature increase at the surface of a 60-nm-thick Au film is measured. While Au is commonly considered as an excellent photothermal material [9], P–G–Nifoam presented faster temperature increase response and higher stable temperature. The effective solar-thermal conversion and fast self-heating of P–G–Nifoam could be mainly attributed to its superior photonic absorption. Figure 2i shows the absorption spectra of P–G–Nifoam with a broad wavelength ranging from 220 to 2600 nm. The average light absorption of P–G–Nifoam is calculated as 97.1%, which is much higher than that of the pristine nickel foam (43.1%). Meanwhile, its three-dimensional, interconnected skeleton with the decorated graphene open nanochannels also favours light absorption. The light reaching the surface of P–G–Nifoam will be trapped in the dense graphene arrays and then almost completely absorbed after multiple internal reflections in graphene nanochannels (schematically shown in Fig. S4), instead of reflection or transmission, resulting in the broadband and highly efficient solar light absorption [33].
High Solar-Water Energy Efficiency of SVGMD
The P–G–Nifoam light absorber is then applied to the SVGMD system for solar desalination and purification. The experimental system is schematically shown in Fig. 3a. The solar simulator with a xenon lamp is equipped with an optical filter to obtain the standard AM 1.5G solar spectrum. The gap width between P–G–Nifoam and a commercial polyvinylidene fluoride (PVDF) membrane is set as 1 mm, which is optimized through a series of tests (see Fig. S5). Natural seawater from Hainan Island (110.1°E, 20.0°N) is used as the feed solution. An illumination of 1 sun is employed to simulate the solar radiation. Thermocouples are used to measure the temperature evolutions of the hot vapour in the gap and the bulk liquid in the tank. As shown in Fig. 3b, the vapour temperature gradually increases from 22.4 to 36.1 °C. In contrast, the bulk water has a negligible temperature rise less than 1 °C. It suggests that the as-obtained thermal energy is effectively confined in the thin water layer only existing in graphene nanochannels, leading to a localized heat transfer and minimal heat dissipation.
Although materials with excellent solar harvesting ability (e.g. polydopamine, light absorption coefficient of 97.0%) [17] and good water permeability/wettability (e.g. mixture of polydopamine and reduced graphene oxide, water contact angle of 46.0°) [15] have been used, the solar-water energy efficiency of conventional photothermal MD devices still remains below 50%. Even at an elevated ambient temperature of 40 °C, the solar-water energy efficiency of conventional photothermal MD is still low (53.8%) [13]. At the present SVGMD, a high permeate flux of 1.10 kg m−2 h−1 is obtained at 1 sun. Correspondingly, the solar-water energy efficiency is calculated to be 73.4%, significantly outperforming the previous works (45.0–53.8%) [13, 15, 17], as shown in Fig. 3c. Additionally, the solar-water efficiencies can be further improved by increasing the solar density to 5 sun and 10 sun, whose details are available in Fig. S6.
The superior solar-water efficiency of the current SVGMD could be predominantly attributed to the following three factors. First, the free-standing P–G–Nifoam presents broadband and highly efficient photonic absorption (i.e. 97.1% at the wavelengths ranging from 220 to 2600 nm), exhibiting superior ability to harvest solar energy. Second, the localized heating achieved by engineered water pathways dramatically suppresses the heat loss to bulk liquid and leads to directional heat transfer from light absorber to the thin water layer. Third, the graphene nanosheets with dense, sharp edges can serve as nanoscale fin-like heat exchangers, leading to highly effective directional heat transfer and very low heat losses. Thermal analysis on the current SVGMD system is conducted (see details in Section S8), and the results are presented in Fig. 3d. The convection and radiation heat loss to the adjacent environment and the conductive heat loss to bulk water are calculated as 45, 0.08, and 55 W m−2, respectively, while the total heat loss is only ~ 10% of the incident solar energy (1000 W m−2).
Long-Term, Effective Desalination and Purification
To demonstrate the feasibility of SVGMD to treat diverse solutions, natural seawater, highly saline brine and oil-contaminated water have been used as the feed. Two main applications are demonstrated as follows.
The first application is to process natural seawater (3.25 wt%) and two types of salty brine (9.85 and 16.70 wt% of NaCl solution). Figure 4a shows the permeate flux of distilled water and salt rejection rate over 72 h of consecutive operation at 1 sun. For all the tests, the salt rejection rates (see Section S9 for calculation method) are close to 100%. Even for salinity as high as 16.70 wt%, the salt rejection rate is at a high level of > 99.6%. To evaluate the long-term operation stability, the permeate flux of distilled water is recorded every 1 h. For three types of saline brine (natural seawater, 9.85 and 16.70 wt% of NaCl solution), the average permeate flux over 72 h is 1.13 ± 0.05, 0.99 ± 0.07, and 0.96 ± 0.06 kg m−2 h−1, respectively. The very small fluctuation of both the salt rejection rate and the permeate flux of distilled water confirms the excellent long-term stability of the current SVGMD system. SEM and EDX analyses (see Section S10) show no noticeable salt deposition on the membrane after 72 h of SVGMD operation, which may dramatically expand the membrane’s lifespans.
The second application is to process mineral oil-contaminated water (i.e. 1 g L−1 mineral oil and 1 mM NaHCO3 in water). Figure 4b shows the permeate flux and total organic carbon (TOC) level of purified water after 72-h consecutive test at an illumination of 1 kW m−2. The TOC concentration in the distillate is at a low level of < 2 mg L−1, clearly better than the World Health Organization (WHO) standard for drinking water (5 mg L−1) [34, 35]. Meanwhile, stable permeate flux of 1.07 ± 0.09 kg m−2 h−1 is achieved. Liquids with low surface tension contaminants (e.g. surfactants and mineral oil) commonly cause detrimental membrane fouling, resulting in failure of conventional MD processes [21, 24, 25]. Control experiments using conventional MD are thus conducted (see “Experimental” section for details). As shown in Fig. 4c, d, oil-based contaminants are observed on the surface of the PVDF membrane after only 1-h operation of the conventional MD device. In contrast, as shown in Fig. 4e, the membrane used in the current SVGMD system remains clean after 72 h of operation. Wettability tests on the membrane before and after distillation are conducted, and the small change in the water contact angle further confirms that there is no obvious chemical/morphological change or oil-based contaminant residues on the membrane (see Fig. S9).
Different from the existing approaches based on complex modifications of commercial membranes [24, 25, 30, 36, 37], SVGMD can realize excellent anti-fouling performance using pristine membranes. The rationally designed gap effectively avoids direct contact between the saline/contaminated feed and the membrane, and thus, only clean vapour reaches the membrane. Instead, the free-standing P–G–Nifoam layer contacts with the saline/contaminated water which is guided and confined over the superhydrophilic solid–liquid interface [38, 39]. Meanwhile, its underwater superoleophobic surface repels oil from the P–G–Nifoam and its superhydrophilic nature avoids salt accumulation at the surface though ion diffusion and advection. Moreover, the PEDOT-PSS coating further prevents the adhesion of salt particles to the surface [40, 41]. Only a very few salt residues are observed on the surface of P–G–Nifoam after long-term, consecutive distillation processes (see Fig. S10).
Highly Efficient Water Collection
The performance of collecting clean water from the hot vapour is important for practical applications [14, 15]. In most of previous works, the solar-vapour efficiency is used to evaluate the solar desalination/purification performance, i.e. the proportion of a given quantity of solar energy used for vapour production at open environment, and is calculated as Eq. 2:
$$\eta_{\text{solar-vapour}} = \frac{{\dot{m} \left( { h_{\text{lv}} + Q } \right)}}{{C_{\text{opt}} q_{\text{i}} }}$$
(2)
where \(\dot{m}\) denotes the vapour flux (kg m−2 h−1), calculated based on the mass decrease in the feed water. hlv is the temperature-dependent latent heat of vaporization of water. Q is the sensible heat. Copt is the optical concentration and qi is the nominal direct solar irradiation of 1 kW m−2.
However, a key factor of water collection ratio is largely ignored. We reiterate that the water collection ratio, one of the key energy and matter efficiency indicators, has not recently attracted the attention. In typical solar desalination/purification systems, the vapour naturally condenses on a transparent plate/dome and then falls down into a container by gravitational force [10, 42]. Although high vapour generation rates and high solar-vapour efficiency up to 95% have previously been reported [5, 6, 8, 33, 42], the water collection ratio still remains at a low level (Only 40% of clean water could be collected from the hot vapour). As a result, the solar-water efficiency is only 22% [10], which means a large portion of the as-generated vapour needs to be further recycled, thus requiring additional energy input.
In addition, in conventional solar vapour systems, the incident light might be partially blocked (absorbed or reflected) by the vapour mist and the droplets forming on the plate/dome, as schematically shown in Fig. S11. The blocked light will also lead to an unwanted energy waste and diminish the vapour generation performance. In the current SVGMD system, the generated vapour diffuses to the gap at the side opposite to the incident light instead of the space between light source and absorber. Thus, the energy loss caused by vapour mist and condensed droplets is dramatically reduced. In addition, the cold distillate flow (6 mL min−1; 18.1 °C) at the opposite side of the membrane can facilitate the hot vapour to directionally transport through the membrane and condensed with the cold distillate flow.
The SVGMD thus inherits the advantage of MD in vapour condensation and water collection and meanwhile dramatically improves the energy utilizing efficiency. As shown in Fig. S11, a test to compare the performances of conventional solar vapour system with a dome and the current SVGMD is conducted. With the current P–G–Nifoam absorber, the evaporation rate and solar-vapour efficiency from open environment can reach 1.41 kg m−2 h−1 and 90%, respectively. However, only 0.75 kg m−2 h−1 of clean water is collected and a low solar-water efficiency of 48.4% is obtained from the dome. On the other hand, the clean water collection rate in the current SVGMD can reach 1.16 kg m−2 h−1, which means 82.3% of vapour is recycled. Consequently, the high solar-water efficiency of 74.5% is obtained, which is 3.5 times higher compared to the existing solar vapour systems [10] and also much higher than the previous photothermal MD systems (45.0–53.8%).
Scaled SVGMD Module for Clean Water Production
To further demonstrate the applicability of SVGMD in practical conditions, an outdoor test is conducted with a scaled module operated in a closed loop configuration (Fig. 5a, b). A light exposure window of 3 × 7 cm2 and a Fresnel lens with 12-fold concentration are assembled to collect and concentrate solar energy. A mixture composed of natural seawater (3.25 wt%), 1 g L−1 mineral oil, and 1 mM NaHCO3 is applied as the feed solution with a flow rate of 2 mL min−1 (see section S14 for the oil size distribution of the emulsion mixtures and the composition of natural seawater). The opposing cold permeate side is cooled by a water flow of 6 mL min−1. The feed and permeate flows are continuously circulated using peristaltic pumps, which used very low (5 W) electric power from the grid and can easily be replaced by renewable energy (e.g. photovoltaic panels) in the portable, off-grid module. The scaled SVGMD module was operated for 9 days with a duration of 8 h per day (from 9:00 am to 16:00 pm) at Hangzhou city (120.1°E, 30.3°N), where the solar intensity ranged from 0.6 to 1.2 kW m−2 (see Fig. S13). As shown in Fig. 5c, the daily yield of clean water is approximately 0.195 kg day−1. Given the sample area of 21 cm2, the daily yield of fresh water is approximately 92.8 kg m−2 day−1 under natural sunlight. The salinity and the TOC in the collected water were monitored. After 9 days of operation, the scaled SVGMD module still maintains a high salt rejection close to 100% (see red squares in Fig. 5d). The ion concentrations of Na+, Mg2+, Ca2+, K+, and B3+ are reduced from 7328, 434, 202, 112, and 9.5 mg L−1 to 9.70, 1.04, 3.61, 0.62, and 0.26 mg L−1, respectively (inset in Fig. 5c), much better than the baseline levels of drinking water defined by WHO [34, 35]. The low TOC level (< 2 mg L−1) in purified water suggests that oil is repelled during the distillation process (see blue hexagons in Fig. 5d). The outcomes of this work demonstrate the potential of SVGMD for long-term, effective water purification with excellent simultaneous salt and oil rejection and point to SVGMD as a promising solution for remote water treatment or desalination in a household- or community-scale application.