Ultra-high freshwater production in multistage solar membrane distillation via waste heat injection to condenser | Nature Communications

Nature Communications volume 15, Article number: 7890 (2024) Cite this article

2232 Accesses

11 Altmetric

Metrics details

Passive solar membrane distillation (MD) is an emerging technology to alleviate water scarcity. Recently, its performance has been enhanced by multistage design, though the gains are marginal due to constrained temperature and vapor pressure gradients across the device. This makes condenser cooling enhancement a questionable choice. We argue that condenser heating could suppress the marginal effect of multistage solar MD by unlocking the moisture transport limit in all distillation stages. Here, we propose a stage temperature boosting (STB) concept that directs low-temperature heat to the condensers in the last stages, enhancing moisture transport across all stages. Through STB in the last two stages with a heat flux of 250 W m−2, a stage-averaged distillation flux of 1.13 L m−2 h−1 S−1 was demonstrated using an 8-stage MD device under one-sun illumination. This represents an 88% enhancement over the state-of-the-art 10-stage solar MD devices. More notably, our analysis indicates that 16-stage STB-MD devices driven by solar energy and waste heat can effectively compete with existing photovoltaic reverse osmosis (PV-RO) systems, potentially elevating freshwater production with low-temperature heat sources.

The Earth’s water cycle, a crucial source of freshwater, is inherently complex and fragile. Its vulnerability is heightened by the ongoing impacts of climate change1 and anthropogenic effects, such as deforestation2, urbanization3, and water withdrawal4 caused by agricultural and industrial activities. Driven by an ongoing mismatch between water supply and demand, nearly half the global urban population will face water scarcity by 20505. Continuous extraction of ubiquitous atmospheric moisture6,7,8 in arid9 and semi-arid10,11 regions with ample solar energy offers a viable solution for freshwater generation on a kilogram scale, especially in emergencies and highly remote locations12 without existing waterbody. However, the atmospheric water harvesting (AWH) technology using renewable energy sources is still in the early stage of development and is generally limited to a low daily water yield13,14 up to 10 L day−1 with a freshwater distillate flux of ~0.17 L h−1 m−2 12, rendering it less suitable for deployment in small-scale communities. Therefore, freshwater supply that meets a personal water demand of up to 50 L day−1 requires terrestrial waterbodies as a source of freshwater, such as seawater15,16, hypersaline brine17, and wastewater18,19. Thermally driven (TD) distillation processes, particularly membrane distillation (MD), have grown in popularity due to the possibility of utilizing low-grade thermal energy, either in the form of solar heat or waste heat from industrial and other processes utilization20. Compared to AWH, a 10-stage solar MD device provides an order of magnitude higher fluxes at ~4.7 L h−1 m−2 17, ensuring high specific water production per device’s overall volume and material surface area.

A core part of thermally driven membrane distillation (TDMD) is a microporous membrane with water vapor permeability and waterproofing characteristics21,22 (Fig. 1a). The membrane separates warm saline feed from the permeate, allowing water vapor to diffuse through its air-filled pores and condense on the cooler side due to vapor pressure differences. Since evaporation and condensation are energy-intensive processes involving phase change (hfg ≈ 2400 kJ kg−1), recent advances in solar MD have shown a viable path for latent heat utilization for both passive16,17,23 and active systems24,25. The main idea is to reuse the condensation heat, further driving the evaporation and condensation processes in subsequent stages. Since the inception of the vaporization enthalpy recycling concept26,27, the rapid progress of solar-to-water efficiencies and overall distillate fluxes was observed in the last years (Fig. 1c, upper part), and the practical adoption of multistage solar MD could thus be expected. Considering the techno-economic analysis, producing more water with less material consumption is crucial for solar desalination, which means the distillate per stage becomes the most important parameter. However, recent pursuits for higher solar-to-water efficiencies actually result in a decrease of distillate flux per stage16,17,23,26,27,28,29,30,31,32,33,34 (Fig. 1c, lower part, Supplementary Fig. 2), which is more obvious for studies with eight or ten stages. The marginal benefit of adding additional stages on overall distillate flux significantly lowers the long-term thermo-economic attractiveness of solar MD technology35,36. In this context, it is essential to increase the distillate flux per stage for multistage solar MD while maintaining a high overall distillate flux.

a Membrane distillation working principle, showing a conventional provision of heat into the system from the top stage using either solar energy using a solar absorber, a photothermal membrane, or a non-solar heater. b An innovative hybrid design that leverages an alternative possibility for ultra-low-temperature heat injection into the lower stages along with the solar energy utilization from the top stage. The hybrid heat provision encompassing the entire device allows temperature gradient manipulation (T+ΔT) across all stages, allowing increased saturation vapor pressure (pv+Δpv) and ultimately leading to elevated distillate fluxes mw+Δmw. c Yearly progress16,17,23,26,27,28,29,30,32,33,34 of solar-to-water efficiencies and distillate fluxes per entire device (upper part) and distillate fluxes per stages (lower part). d A graphical representation of saturation vapor pressure relative to dry bulb temperature across eight stages in a conventional MD device, and e a stage-temperature boosted MD device, with a total 25% boosted temperature primarily in 7th and 8th stages, yielding 62% larger saturation vapor pressure across the entire device. Source data are provided as a Source Data file.

Distillate flux enhancement is intrinsically challenging due to the low heat flux across the stages, accumulative heat leakage and heat and mass transfer resistance in each stage, building up the temperature and vapor pressure gradient across the entire device (Fig. 1a)37. More importantly, the distillation relies on moisture transport from the high-temperature feed to low-temperature permeate sides in the air gap35, which is constrained by the maximum moisture content in the air. To provide a larger temperature gradient throughout the device, effective heat sinks were adopted in previous studies16,17,23,26,27,28,29,30,32,33,34. However, according to psychrometrics, low dry bulb temperature could result in a low moisture content in the air and limit the distillation. Taking the same moist air temperature difference of 10 °C in distillation as an example, the vapor pressure difference under evaporation temperature of 85 °C is approximately five times greater than that of 45 °C (~19 kPa vs. ~4 kPa, inset of Fig. 1d), due to the distinct exponential relationship between saturation vapor pressure and temperature. This reveals that lower air temperature restricts the moisture content and significantly limits the distillate flux. More notably, this explains the sharply reduced distillate flux in low-temperature stages of multistage MD and suggests a reversed design in the condenser, i.e., ineffective cooling or even heating of the condenser might be able to increase the moisture content in lower stages and suppress the marginal benefit of multistage design. Such reverse design could be feasible, especially when solar distillation with constant heat flux but not constant heating temperature is considered.

Based on this understanding, we propose the stage temperature boosting (STB) design for multistage MD. As shown in Fig. 1b, each condenser in a multistage MD device also acts as an evaporator of the next stage, providing additional freedom for heat input into the system. Our STB approach is the opposite of the conventional method; rather than cooling the condenser, we apply additional heat to it. This involves injecting the heat into the lower stages in addition to the conventional heat input at the top stage. Considering the low working temperature of the last stages, the heat input could be from the free waste heat of many facilities, such as gas, coal, biomass, and nuclear power plants, engines, air conditioners or even PV panels. Such innovative hybrid heat provision design creates an alternative approach for manipulating temperature gradients within the system (Fig. 1b). As shown in Fig. 1d, a calculation assuming a 5 °C difference in evaporation and condensation temperatures in an 8-stage device yields a total vapor pressure difference of 22 kPa in a conventional MD device. However, by locally increasing the temperature difference to 10 °C only in the 7th and 8th stages through STB, the total vapor pressure difference increases by 62% from 22 kPa to 35.6 kPa with only a 25% (40 to 50 °C) larger temperature difference across the device (Fig. 1e), which could significantly increase distillation fluxes in each stage.

Several studies on different types of MD, including vacuum MD38 and direct contact MD (DCMD)39,40,41,42, have demonstrated possible advancements in distillate flux augmentation when MD device operation shifts to higher operating temperatures. In the case of DCMD, increasing the feed temperature from 40 °C to 80 °C resulted in a 4.6-fold increase in distillate flux43. Experimentally derived data showed a 4.6-fold enhancement in distillate flux of DCMD when the inlet feed temperature was increased from 80 °C to 95 °C44. By applying elevated pressure on the feed side, the distillation process could withstand temperatures up to 128 °C44 and 118 °C45, yielding 1.9-fold and 1.7-fold larger distillate flux values than the baseline configuration at 110 °C and 108 °C, respectively. Further numerical investigations revealed that the performance of DCMD could be further improved by raising the feed temperature from 80 °C to 180 °C, leading to a 9.4-fold increase in distillate flux46, while simultaneously decreasing the specific energy consumption (SEC) by 2.9 times. The positive correlation between SEC improvement and feed temperature has been further investigated47. Another study on spiral-wound air gap membrane distillation (AGMD) experimentally showed that increasing the feed temperature from 65 °C to 75 °C nearly doubled the distillate flux48.

While the literature clearly indicates that augmenting feed water temperature can significantly increase distillate fluxes, no such attempt has been made with solar-driven multistage MD devices. Additionally, all the aforementioned studies dealt with inefficient condenser cooling, whereas our study is the first to demonstrate that heating in the last stages shifts the operating conditions of a multistage MD device to higher temperatures. This realization led to the development of the world’s first hybrid solar and waste heat multistage MD device that can utilize low-grade thermal energy. This approach shows a potential path for advancements in the multi- and transdisciplinary fields of solar-driven multistage MD and waste heat utilization.

To show the effectiveness of condenser heating and our STB design, we carried out comprehensive theoretical and experimental studies, shedding light on the ways in which the STB’s enhancement mechanisms work. With an 8-stage solar MD device and STB in the last two stages, a record-high stage-averaged distillation flux of over 1.13 liter per square meter per hour per stage (L m−2 h−1 S−1) was demonstrated. Note that this high stage-averaged distillation flux is achieved when nearly 1/3 of the heat input has almost zero latent heat recovery potential. Considering only the effect of increased evaporation and condensation temperatures in the first six stages, the stage-averaged distillation flux could be further enhanced. When compared to the state-of-the-art 10-stage solar MD device23 with similar solar-to-water efficiency of 400% and stage-averaged distillation flux of 0.6 L m−2 h−1 S−1, our device achieves an impressive 88% increase in distillation flux; thus significantly lower water production costs are achievable. From a techno-economic perspective, our 16-stage STB-MD device could reduce water production costs by up to 75% compared to the top 10-stage solar MD systems. Additionally, exergy analysis suggests that devices with over 16 stages could rival PV-RO systems in efficiency under specific conditions. For the first time, our work demonstrates that heating the condenser with low-temperature waste heat in multistage MD devices significantly increases water production and offers an innovative pathway for involving waste heat in low-carbon water production. The core novelty of our work is based on the psychrometric fact that air can hold larger amounts of moisture at higher temperatures due to the exponential relationship between saturation vapor pressure and temperature. Through advanced system design, we successfully demonstrate a viable method for upshifting the operational temperatures of a multistage MD device, leading to a significantly enhanced total moisture flux.

The introduction of the STB concept is expected to catalyze further research in MD, particularly focusing on system-level innovations and scalable applications. Furthermore, our research offers innovative perspectives on enhancing freshwater production and reducing the carbon footprint of future membrane distillation systems.

Thermally driven MD devices function by generating vapor in a closed system20,35, where evaporation at the membrane-feed water interface is separated from the condenser by an air gap. Let us consider a single-stage MD device comprising two distinct temperatures (Fig. 2a): the lower bottom temperature (Tcond) associated with a cold plate where the vapor condenses and the warmer feed water temperature (Tevap), at which the evaporation occurs. For simplicity, we assume a constant temperature across the feed water and the thin selectively permeable membrane. This assumption is reasonable given that the typical membrane thickness (~10−1 mm) is an order of magnitude smaller than the feed water layer thickness ~100 mm. Manipulating the temperature gradient across the single-stage device can be achieved in two ways. First, by varying the heat input from the top through optical solar concentration or increasing the heat flux per evaporative surface area using a non-solar heater while maintaining a cold plate temperature. Second, the cold plate temperature is varied using an external system, concurrently with adjustments to the heat flux input at the top of a stage. The vapor transport and the generation of liquid water in systems with a small air gap (<10 mm) is driven by the saturation vapor pressure gradient between the evaporation and the condensation parts35. Under the 1D assumption, Fick’s law describes the generalized distillate flux for any given stage as follows: \({m}_{{{{\rm{w}}}},{{{{\rm{S}}}}}_{{{{\rm{n}}}}}}=({p}_{{{{\rm{w}}}},{{{\rm{sat}}}}}({T}_{{{{\rm{e}}}},{{{{\rm{S}}}}}_{{{{\rm{n}}}}}})-{p}_{{{{\rm{w}}}},{{{\rm{sat}}}}}({T}_{{{{\rm{c}}}},{{{{\rm{S}}}}}_{{{{\rm{n}}}}}})) \cdot {{R}_{{{{\rm{tot}}}}}}^{-1}\) where \({p}_{{{{\rm{w}}}},{{{\rm{sat}}}}}\) is the water saturation pressure at evaporation (\({T}_{{{{\rm{e}}}},{{{{\rm{S}}}}}_{{{{\rm{n}}}}}}\)) and condensation (\({T}_{{{{\rm{c}}}},{{{{\rm{S}}}}}_{{{{\rm{n}}}}}}\)) temperatures at each stage Sn (where n ≥ 1). \({R}_{{{{\rm{tot}}}}}\) is the total diffusional resistance, which is directly proportional to the molar mass of water, Mw, and mass diffusivity of water vapor in air49 Da, while inversely proportional to universal gas constant, R, water temperature at the saturation pressure during evaporation, \({T}_{{{{\rm{e}}}},{{{{\rm{S}}}}}_{{{{\rm{n}}}}}}\), and the air gap thickness \({t}_{{{{\rm{AG}}}}}\). The saturation pressure of water vapor is described by an exponential function featured in Antoine equation50 log10(pw,sat) = A – B·(T + C)−1, where T is expressed in degrees Celsius and pw,sat is in bars. After rearranging, the final form of the equation to predict the distillate flux with the unit kg m−2 s−1 due to vapor diffusion across an air gap with thickness tAG, m, is given by:

where β is a constant with a value of 103, /, Mw and Mair refer to the molar mass of water and air, kg kmol−1, respectively, R refers to universal gas constant, J kmol−1 K−1, \({t}_{{{{\rm{AG}}}}}\) refers to air gap thickness, m, pa refers to absolute pressure in the system, Pa, Vw and Vair refer to atomic diffusion volumes of water and air, m3 mol−1, according to Fuller51, respectively, while A, B, C, /, refer to constants of an Antoine equation50.

a A schematic showing the geometry and temperature boundary conditions for the evaporative mass flux potential assessment. tAG denotes air gap thickness, Tevap and Tcond denote evaporation and condensation temperatures, respectively. b Evaporative mass flux mapping through a 4 mm air gap under various evaporation and condensation temperature differences with fixed condensation temperatures. c Variation of evaporative mass flux at evaporation temperatures from 20 to 95 °C and condensation temperatures from 20 to 90 °C. d Variation in evaporative mass flux with air gap thicknesses of 2, 3, 4, 6, and 12 mm at evaporation temperatures from 20 to 95 °C and fixed condensation temperatures of 20, 50, and 80 °C. Inset figures in (c) and (d) show evaporative mass flux mapping under solar illumination of one-sun intensity (1 kW m−2), with a theoretical limit based on a vaporization enthalpy of 2400 kJ/kg. For figures (c) and (d), the distillate flux associated with specific condensation temperatures and air gap thicknesses is represented by different colors. Source data are provided as a Source Data file.

The temperature’s impact on distillate flux is notably non-linear, especially due to the saturation vapor pressure’s exponential representation. This relationship intensifies as temperatures approach the upper limit of 95 °C, evidenced by the steep increase in evaporative mass flux depicted in Fig. 2c. Such a steep curve at higher temperatures suggests a critical area for larger gains in distillate fluxes. In contrast, between 20 and 45 °C, the flux increase is more gradual. However, it is important to note that the temperature difference between the evaporation (feed) and condensation (permeate) sides significantly affects the magnitude of the distillate flux, as calculated by Eq. 1, and this effect becomes more pronounced at higher evaporation (dry-bulb) temperatures (Supplementary Fig. 3).

By looking at Fig. 2b, d, two key observations emerge. First, Fig. 2b illustrates that larger temperature difference coupled with rising condensation temperatures significantly boost evaporative mass fluxes, as already mentioned (Supplementary Fig. 3). Additionally, the region of highest fluxes does not extend to the maximum temperature difference, suggesting diminishing returns or a limiting factor in the system’s design or operation at low temperatures. Second, Fig. 2d reveals that while air gap thickness tAG impacts evaporative flux linearly, its effect diminishes at higher temperatures due to the dominating exponential vapor pressure difference with higher evaporation temperatures. This suggests that systems with wider air gaps still benefit substantially from elevated temperatures at all stages, maintaining high fluxes below the one-sun threshold.

In this analysis, we chose air gap thicknesses between 2 and 12 mm, guided by studies23,35,37 that pinpoint 0.4–0.6 mm as the optimum for balancing thermal and mass transfer resistances. Such air gaps are associated with peak solar-to-vapor efficiency. The analysis also demonstrates the effectiveness of air as a working fluid in enhancing diffusional water transport across these gaps. Although the dotted lines in Fig. 2b–d suggest a theoretical maximum performance of 1.5 kg m−2 h−1 under standard one-sun illumination, our data shows that systems designed to handle high heat loads are not limited by increased heat and corresponding mass fluxes. In addition, air, serving both as a working fluid and a vapor transport medium, does not create a bottleneck in the system. Hence, integrating additional heating strategies, such as stage temperature boosting, could substantially enhance MD device performance. Leveraging low-grade waste heat or solar thermal energy, potentially from photovoltaic electricity generators, and simultaneously harnessing the vaporization enthalpy could give MD an advantage over alternatives like PV+RO systems. A thorough discussion comparing both approaches, using distillate production per unit of exergy—defined as available work in heat—is presented in the final sub-section before the discussion. Notably, the utilization of vaporization enthalpy and STB enables significantly higher distillate fluxes, surpassing the one-sun limit. Detailed examination of these phenomena led to the assembly of three distinct STB-MD devices, which are discussed in the following sub-section.

To demonstrate the effectiveness of our proposed stage temperature-boosting concept in multistage membrane distillation devices, we initially assembled two proof-of-concept devices: one with five stages and another with eight stages (refer to Fig. 3a, showing the left and middle devices). Furthermore, we developed a variation of the eight-stage device, which incorporates a photothermal membrane (PTM) in the top stage and features two immersed electrical heaters in stages 7 and 8, as depicted in Fig. 3a (right device). The internal dimension of each device with respect to the membrane was 98 × 98 mm. To enhance thermal efficiency, all devices were thermally insulated with 2 cm thick insulation foam, having a thermal conductivity of 0.04 W m−1 K−1.

a 3D renderings of a five-stage device (left), an eight-stage device featuring solar absorber beneath the acrylic glass (middle). Both incorporated submerged heaters inside feed water channels at each stage. Eight-stage device with a photothermal membrane installed on top (right), with supplemental electric heaters in stages 7, 8. b Photothermal membrane fabrication preparation process. c UV-Vis-NIR spectroscopy for solar absorptivity characterization of the raw, CNT-coated, and CNT+FPTS-treated membranes, alongside the solar absorber. d SEM pictures depicting microstructural differences, e contact angle characterization of hydrophobicity, and f surface morphology and thickness characterization using a confocal microscope of raw, CNT, and CNT+FPTS membranes. g Topographical maps illustrating surface roughness and topology of the raw and CNT+FPTS membranes obtained by atomic force microscopy. Source data are provided as a Source Data file.

The initial two devices were employed to explore the impact of stage temperature boosting on enhancing water production in MD systems. These STB-enhanced MD devices comprised several key components: a 2 mm thick acrylic glass top, a water gap, a polytetrafluoroethylene (PTFE) microporous membrane supported by a robust polyethylene terephthalate (PET) layer, an air gap, and a 1 mm thick aluminum base plate. In the five and eight-stage devices, we strategically inserted 50 × 50 mm thin electrical heaters into each water gap to assess their influence on system efficiency and productivity. The water layers in the first and second stages were maintained at thicknesses of 5 mm and 4 mm, respectively, while in subsequent stages, this thickness was reduced to 3 mm. The air gap ranged between 4 and 5 mm in thickness. Additionally, the eight-stage system (Fig. 3a, middle device) featured a solar absorber positioned above the water layer for enhanced solar energy utilization. The third device (Fig. 3a, right device), equipped with a photothermal membrane, largely replicated the core design of the first two conventional devices, maintaining water gaps of 2–3 mm and air gaps of 3–4 mm. Each membrane was supported by 2 mm high silicone gaskets, a design choice to prevent membrane contact with the 1 mm thick aluminum plate while ensuring an adequate air gap. Further details on the feed water provision and distillate collection in these STB-MD devices are elucidated in the subsequent two sub-sections.

The photothermal membrane, acting as a solar interfacial evaporator52,53,54, was fabricated using a raw membrane as the substrate, which was spray-coated with multiwalled carbon nanotubes (CNTs). These CNTs were first ultrasonically dispersed in isopropyl alcohol (IPA) and then treated with 1H,1H,2H,2H-Perfluorooctyltriethoxysilane (FPTS) to augment its hydrophobic properties, as depicted in Fig. 3b. The CNT coating substantially improved the membrane’s photothermal properties, achieving a solar absorbance above 99%—a significant enhancement over the uncoated membrane at 30%. This was verified by the UV-Vis-NIR spectral analysis shown in Fig. 3c. In comparison, the solar absorber demonstrated lower absorptivity, at approximately 93% within the solar spectrum (0.3–2.5 μm), which suggests its solar-to-thermal energy conversion efficiency being intrinsically lower. The carbon nanotube-coated membrane has higher solar absorptivity compared to the classical solar absorber due to several factors. Larger surface roughness contributes to less ideal reflectance, as increased roughness creates numerous peaks and valleys on the membrane surface, disrupting the path of reflected solar rays. These rays are more likely to be reabsorbed by the surface, increasing the overall amount of absorbed solar radiation through the optical trapping effects55. Additionally, carbon nanotubes inherently have excellent light-trapping properties due to their unique porous structure, resulting in a low refractive index56, and a high aspect ratio57, further enhancing their solar absorptivity.

The Scanning Electron Microscopy (SEM) images (Fig. 3d) show a large contrast between the raw PTFE membrane and its carbon nanotube and CNT+FPTS-coated counterparts. In terms of the membrane’s selectivity20 for the separation of liquid feed water from gaseous water molecules—which subsequently diffuse to the colder condenser—the carbon nanotube coating preserves the intrinsic microstructure of the membrane58. The raw membrane’s SEM image reveals a fibrous structure with clear, distinct PTFE fibers59, indicative of a porous network. Upon the application of CNTs, the surface appears more uniform with the nanotubes creating a dense network, enhancing the photothermal properties while maintaining the membrane’s intrinsic microstructure. Notably, the CNT coating does not seem to impede the pores between the fibers, which is crucial for preserving the permeability characteristics for vapor transport. However, insights collected from the literature suggest that the CNT layer may marginally diminish the membrane’s long-term hydrophobicity. To mitigate this effect, FPTS was applied to reinstate the hydrophobic properties, as illustrated in Fig. 3e. The CNT+FPTS-coated membrane displays an even more homogenous surface, suggesting that the FPTS treatment fills in the interstices between the CNTs, promoting a smoother and more uniform surface topology that contributes to increased hydrophobicity60. This treatment ensures that the performance of the CNT-coated membrane remains uncompromised, maintaining optimal solar absorption capabilities alongside its hydrophobic microstructure.

The pore size, thickness, and surface roughness of the membrane are critical factors affecting its heat transfer efficiency and water vapor permeability in thermally driven membrane distillation61. Specifically, the pore size must be optimal to prevent liquid water from passing through while allowing vapor to transfer; smaller pores yield higher capillary pressures, contributing to the membrane’s hydrophobicity. The membrane’s thickness is directly proportional to the thermal resistance, thus affecting the temperature gradient across the membrane. Additionally, surface roughness62 influences the wettability and scaling propensity of the membrane. The membrane has a pore size of 1 μm (as demonstrated by the SEM images) and a thickness ranging from 126 to 150 μm, according to the confocal microscopy pictures (Fig. 3f). The raw membrane (Fig. 3f, left) exhibits significant surface roughness with a value of 0.772 µm, as indicated by the varied coloration and evident peaks and valleys in the topographical map. This level of roughness suggests a more textured surface, which can influence wettability and potentially contribute to the trapping of air63. The cross-sectional view shows a fairly uniform thickness without significant surface features, which is typical for raw, untreated membranes. The application of CNTs (Fig. 3f, middle) reduced the surface roughness to 0.278 µm. Such enhanced smoothness can be attributed to the filling of gaps and valleys by the CNTs, creating a more homogenized topology. The thickness of the membrane has increased slightly to 130 µm, suggesting that the CNT coating adds a minimal thickness while modifying the surface. Further treatment with FPTS (Fig. 3f, right) has reduced the roughness even more to 0.177 µm. This indicates that FPTS is filling in the interstitial spaces between the CNTs and creating a more uniform coating, which enhances hydrophobicity and decreases wettability64. The increase in thickness to 150 µm reflects the addition of the FPTS layer atop the CNT coating. This additional layer seems to be contributing to the smoothness observed in the topographical image. The color distribution in the CNT+FPTS image is more uniform compared to that of the CNT-only image, suggesting a smoothing effect imparted by the FPTS treatment. This observation aligns with the SEM image at a larger scale (20 µm) of the CNT+FPTS sample (Fig. 3d), which displays a consistent coating across a wider area.

Atomic Force Microscopy (AFM) analysis further explains the topographical modifications imparted by the CNT+FPTS coating on the raw membrane (Fig. 3g). The AFM image of the raw membrane reveals a highly textured surface with prominent peaks and troughs. By contrast, the CNT+FPTS-coated membrane displays a markedly different topology, where the surface roughness is significantly reduced to 0.146 µm. This suggests a denser and more uniform top layer, where the FPTS seem to have effectively filled the microscale irregularities.

Beyond the characterization of materials and design, the assembled devices and the materials employed were thoroughly investigated in providing a detailed mechanistic and performance overview of STB within MD devices in the subsequent two sub-sections.

Thermally driven membrane distillation employs a phase change mechanism to separate water from non-volatile pollutants, relying on substantial energy to break the hydrogen bonds between water molecules during evaporation. In MD systems, the enthalpy of vaporization, which is crucial for this process, shows a slight variation; it reduces by approximately 8% when the evaporation temperature increases from 30 °C to 100 °C. While the energy required for distillate production is significant, overall energy consumption remains a primary challenge and a focus area in MD process research. This study primarily addresses the need for a detailed mechanistic understanding of various factors in MD systems. Key aspects under investigation include stage temperatures, heat flux per stage, device orientation, and the type and number of stages having an influence on the overall distillate fluxes and solar-to-water efficiencies. To assess the opportunities and limitations of STB-MD devices through comprehensive mechanistic insight, we have initially constructed two experimental setups.

The setup for characterizing the role of STB in each of these two systems is depicted in Fig. 4a. Both the 5-stage (Fig. 3a, left) and 8-stage (Fig. 3a, middle) STB-MD devices were supplied feed water from a water tank positioned above them, leveraging siphon flow. Each of the three devices operated with a dead-end feed water inflow. To optimize the collection of distillate water and ensure adequate solar flux to the solar absorber of the 8-stage STB-MD device, we inclined both devices at a 30° angle from the horizontal plane. The heat sink of each device was partially submerged in a large water reservoir within a pool, maintaining a stable condensation temperature in the final stage. Silicone tubes connected the distillate outlets from each stage to separate water containers. After a typical duration of one hour, we measured the collected distillate water using a precision scale.

a An experimental setup comprising a large water pool, a water tank, solar simulator (for 8-stage STB-MD device), strategically positioned thermocouples, and a precision scale for measuring distillate weight. b Performance evaluation of 5- and 8-stage devices under standard operating conditions with the immersed heaters switched off. c An investigation into the differences between inclined and horizontal devices featuring numerical simulations of the air and water gaps with presented temperature and velocity fields. d Performance evaluation of 5- and 8-stage devices with enabled stage temperature boosting. e Distillate fluxes of 5- and f 8-stage STB-MD devices with the corresponding g temperature differences across two successive stages for 5- and h 8-stage devices. Note: the error bars in (b) and (d) represent standard deviation (SD). SD was calculated based on up to three performed measurements. Source data are provided as a Source Data file.

Figure 4b displays the distillate fluxes and solar-to-water efficiencies of both 5- and 8-stage STB-MD devices with heat fluxes supplied to the first stage either by solar energy or immersed heaters. Energy-to-water efficiency was quantified using the following equation:

Where \({m}_{{{{\rm{w}}}},{{{{\rm{S}}}}}_{{{{\rm{n}}}}}}\) is distillate flux per stage Sn, kg m−2 s−1, \({h}_{{{{\rm{fg}}}}}\) is vaporization enthalpy of feed water, J kg−1, \(\sum {q}^{{\prime} {\prime} }\) denotes the sum of all heat flux inputs into the device, including both solar and electric heaters W m−2, A denotes solar-absorbing area and/or effective membrane area for distillation (9.8 × 9.8 cm), m2.

Initially, these devices functioned without supplementary electrical heaters, thus bypassing the typical STB mode of operation. At a heat flux of 1000 W m−2, both devices demonstrated performance metrics comparable to current state-of-the-art technologies. Specifically, the 8-stage device achieved a distillate flux of 4.4 L m−2 h−1 with a solar-to-water efficiency of 302%, moderately outperforming the multistage MD device reported by Wang et al.34 (3.6 L m−2 h−1 and 246%). Conversely, our 5-stage electrically-driven MD device showed marginally lower performance (3.1 L m−2 h−1 and 207% efficiency) compared to a 5-stage photovoltaic-MD device from existing literature29 (3.25 L m−2 h−1 and 221%). This suggests that our devices are on par with leading multistage MD devices. An important aspect affecting performance in STB-MD devices, particularly in cases with higher heat flux input, is the inclination of the devices. To understand this, we conducted COMSOL numerical simulations comparing horizontal and inclined MD devices equipped with a solar absorber. The simulations, shown in Fig. 4c, revealed significant temperature non-uniformity around the solar absorber, especially pronounced near the edges of the inclined device. This uneven temperature distribution leads to noticeable velocity gradients within the device, most prominently in the water layer due to the steep temperature gradients at the interface between the glass, solar absorber, and water layer. Furthermore, we examined moisture transport within the air gap of these two configurations, with the findings illustrated in Supplementary Fig. 4b, d, f, h. Moisture transport in the air gap is controlled by two processes: diffusion and, in the presence of a developed velocity gradient, advection. Diffusion, which causes moisture to move from areas of higher to lower vapor concentration due to random molecular motion, was observed in both x and y directions in the inclined air gap setup (Supplementary Fig. 4b, f). The movement in the y-direction is particularly noticeable due to the temperature gradient and resultant vertical vapor flow (Supplementary Fig. 4f). In contrast, a non-zero value in the x-direction (Supplementary Fig. 4b) indicates the presence of enhanced advection, driven by the buoyancy effect created by temperature and corresponding density gradients of moist air. Conversely, when the air gap is positioned horizontally, there is minimal diffusive moisture movement in the x-direction, with magnitudes on the order of 10−9 (Supplementary Fig. 4d). The majority of the diffusive moisture flux thus occurs in the y-direction, with a flux that is five orders of magnitude larger—10−4 (Supplementary Fig. 4h). These results highlight a new layer of complexity in multistage MD systems. The findings illustrate how the orientation of water and air gaps can markedly affect moisture transport, oscillating between scenarios of sole diffusion or diffusion complemented by advection. Such variability becomes especially pertinent in the context of multistage MD systems with gaps of varying dimensions and inclinations. Shifting focus from device orientation, our subsequent investigation delved into how different amounts and locations of heat injection influence the performance of the 5- and 8-stage STB-MD devices (Supplementary Tables 2 and 3, respectively).

Figure 4d provides a comprehensive performance analysis (left part) of both devices under varying heat flux levels in each corresponding stage (right part). Initially, a clear pattern emerges: when heat injection occurs closer to the first stage, there is a noticeable increase in distillate flux, except for the 5-stage device with a total heat flux of 1000 W m−2. Additionally, partitioning larger heat fluxes into smaller values across multiple stages demonstrates a moderate improvement in both performance metrics. This is observed in cases 1–4 of a 5-stage MD and cases 1–3 of an 8-stage MD, both with a total heat flux of 1500 W m−2. Subsequently, we evaluated the performance of the devices with enabled STB relative to the baseline scenarios, considering heat flux levels ranging from 800 W m−2 to 2000 W m−2 in the first stage for both devices. However, in order to further elucidate the impact of temperature variations at each stage, particularly within the water gap on distillate fluxes, we integrated our observations on increased distillate fluxes with temperature differences between successive stages (Sn − Sn−1, as shown in Fig. 4e–h) and normalized temperature gradients to the baseline scenarios across all stages (Supplementary Fig. 5).

The implementation of a five-stage STB-MD device operating with a total heat flux of 1500 W m−2, as depicted in Fig. 4d (cases 1–5), results in a significant boost in distillate fluxes, registering an increase of 16-62% over the standard 1000 W m−2 setup showcased in Fig. 4b. While efficiency levels fluctuate from a decrease of −23% to an increase of +8%, the temperature gradient across all stages exhibits an impressive elevation of approximately 10 to 60% when utilizing STB-MD devices, relative to the baseline configuration seen in Supplementary Fig. 5a (second diagram). The data reveals a distinct correlation between the specific site of heat introduction into a stage—where stage temperature boosting is executed—and the consequent rise in distillate fluxes. For example, in the STB5 500 W m−2 scenario (with ‘5’ indicating the stage receiving a 500 W m−2 heat input), there is a marked increase in the temperature difference between stage 5 and the aluminum back plate (S5-bp), peaking at nearly 18 °C as shown in Fig. 4g (second diagram). This elevation in stage 5 temperature, which exceeds the baseline by over 60% (refer to Supplementary Fig. 5a, second diagram), translates into a notably higher distillate flux in this stage, surpassing even the traditionally dominant first stage—a rare feat in existing TD distillation systems. This pattern is also noticeable in the case of STB4,5 500 W m−2. However, the STB2,3 250 W m−2 setup presents an exception. Despite the fact that the temperature differences between S2-S3 and S3-S4 are not highly pronounced (as per Fig. 4g, second diagram), the temperature in stages S2 and S3 skyrockets relative to S1 and S4, respectively. This leads to an extraordinary distillate flux, with stage S3 delivering more than 1.2 L m−2 h−1 S−1 compared to the roughly 0.7 L m−2 h−1 S−1 of stage S1, defying typical expectations.

In our subsequent analysis, we compared the distillate fluxes of 1500 W m−2 STB-MD devices with those of a baseline 1500 W m−2 setup in the initial stage. Here, we observed a slight decline in performance, with fluxes and efficiency ranging from −24 to +6%. Notably, the STB2,3 250 W m−2 configuration was the standout, with stages S2 to S5 surpassing the baseline variant’s water production, with the sole exception of S1, as illustrated in Fig. 4e (second diagram). Temperature increases across stages were modest, approximately 0–20%, as shown in Supplementary Fig. 5a (third diagram). At a higher total heat flux of 2000 W m−2, we noticed a −9% reduction in both water production and temperature increase compared to the standard setup. This outcome is primarily due to the strategic placement of heat injections at stages 2 and 3 with 500 W m−2, which results in a substantial temperature rise from stage S3 onward when contrasted with the baseline, as evidenced in Supplementary Fig. 5a (fourth diagram). This alteration also slightly increases the temperature difference, as recorded in Fig. 4g (third diagram). Furthermore, we incorporated experimental data for a total heat flux of 1000 W m−2 against a baseline scenario of 800 W m−2, as depicted in Fig. 4e, g (first diagrams) and Supplementary Fig. 5a (first diagram). The observations and conclusions drawn in this section are equally applicable to these conditions.

The investigation, utilizing an 8-stage device, centered on two main elements. First, it considered a scenario that closely mirrors actual conditions, where heat is photothermally induced at the top stage due to the intrinsic spectral characteristics of the solar absorber. Second, it assessed the effect of integrating three extra stages, which equate to a 60% increase in the system’s capacity to recycle vaporization enthalpy.

Figure 4f, right, illustrates the water production per stage for a baseline configuration under one sun and six enhanced STB-MD configurations, each harnessing an additional 500 W m−2. Immediately, configurations such as STB1 500 W m−2 and STB2,3 250 W m−2 stand out, showcasing significantly higher fluxes. Notably, STB1 achieves an unprecedented flux in the first stage, exceeding 1.6 L m−2 h−1 S−1, an upsurge primarily attributed to the extension of stages. This setup also records the largest temperature difference seen at S8-bp, nearly 13 °C (as indicated in Fig. 4h, right), and a more than 70% higher temperature in the eighth stage (Supplementary Fig. 5b, right). Yet, it presents the smallest increase in distillate flux in the last stage compared to the baseline under one sun. This highlights the importance of implementing STB in the latter stages—specifically stages 7 and 8 (STB7,8 250 W m−2 and STB8 500 W m−2), where a substantial enhancement, averaging double the flux of the last stage than the baseline configuration, was observed. Despite these individual differences, a consistent finding emerges across all STB-MD devices. Heat injection near the first stage markedly boosts distillate fluxes, while distributing larger heat fluxes into smaller increments across several stages moderately enhances performance metrics. In sum, distillate fluxes saw an uplift from 4% to 85%, and efficiencies spanned from −31% to +23%, with STB2,3 250 W m−2 and STB1 500 W m−2 attaining efficiency increases of 11% and 23%, respectively. This reinforces the conclusion drawn from the 5-stage STB analysis: strategic heat management is crucial for optimizing the STB-MD device performance.

We further explored a real-world scenario with diminished solar intensity at 800 W m−2, assessing the influence of applying STB in the 4th and 8th stages with an additional 200 W m−2. The initial notable distinction between the two STB-MD setups is the temperature difference in the last stage relative to a baseline of 0.8 sun, as shown in Supplementary Fig. 5b, left. With STB8 200 W m−2, there was a −15% difference, while STB4 200 W m−2 exhibited a + 50% increase. Despite this, both configurations achieved significant enhancements in distillate fluxes at the eighth stage—approximately 50% for STB4 and an impressive 100% for STB8, as per Fig. 4f, left. This improvement was recorded even though the temperature difference for STB8 was actually 1 °C lower than the baseline, however, the primary cause of the nearly doubled distillate fluxes was the temperature increase within the stage. Overall, distillate fluxes increased by 6% and 20% for each configuration, although efficiencies dropped by −15% and −4%, respectively, in comparison to the baseline configuration. The results related to STB-MD devices utilizing ultra-low temperatures in the lower stages (specifically stages 7, 8, and in the case of the 8-stage device) have directed our focus towards a performance assessment of the photothermally-driven 8-stage STB7,8-MD device.

Figure 5a depicts the experimental setup of an 8-stage STB-MD device, incorporating a photothermal membrane in the top stage. In this arrangement, the device is supplied with feed water from an overhead tank through siphon flow and is operated with a dead-end feed water inflow. This setup differs from previous ones in that it utilizes a constant thermostatic bath with a heat sink attached to the back plate, ensuring a stable condensation temperature. The setpoint of 25 °C meant that the actual measured temperature at the backplate was approximately 26 °C. With a fixed bottom plate temperature, the additional variable related to increasing pool water temperature is thus removed, allowing for a more precise assessment of the temperature gradient across all stages. For distillate extraction, a syringe was used. Post a 1-hour experiment, each air gap was manually vacuumed, the distillate was collected into water bottles, and its weight was measured using a precision scale. Overall, five experimental runs were performed (as shown in Supplementary Fig. 6), where the baseline experiment under standard 1 sun illumination was the first (Supplementary Fig. 6, highlighted as 1st). Experiments with STB7,8 at 250 W m−2 and 500 W m−2 were conducted as the 3rd and 4th trials, respectively. The 2nd and 5th experiments involved baseline measurements with a bubble foam cover on the top side facing the solar simulator, under both baseline and STB7,8 250 W m−2 conditions.

a An experimental setup comprising a constant thermostatic bath with a heat sink attached to the back plate, solar simulator, thermocouples in each water gap, and a precision scale for measuring the weight of the distillate, which was removed from the gap using a syringe. b Temperature variation in the water gaps and the back plate after the immersed heaters were turned on, delivering 250 W m−2 and 500 W m−2. c Increase in feed water temperature for the cases with 1 sun + STB7,8 250 W m−2 and 1 sun + STB7,8 500 W m−2 compared to the baseline scenario (1 sun). d Temperature differences across two successive stages for baseline, STB7,8 250 W m−2 and STB7,8 500 W m−2. e Temperature field across all domains and vapor concentration field in the air gap domain of an MD, shown without STB (left) and with STB (right). f Distillate fluxes and solar-to-water efficiencies encompassing the entire device and g distillate flux per stage for baseline, STB7,8 250 W m−2 and STB7,8 500 W m−2 configurations. Source data are provided as a Source Data file.

In the 3rd and 4th experiments, dynamic variations in the water gap temperatures were observed when the heaters were turned on (Fig. 5b). This led to a significant rise in temperatures at all stages, particularly noticeable during the 500- to 600-min transition. Compared to the baseline, the most substantial temperature increase, about 25% and 14% higher for STB 500 W m−2 and 250 W m−2, respectively, occurred in stage 7 (Fig. 5c). Detailed comparisons between STB and non-STB temperature profiles are illustrated in Supplementary Figs. 6–8 for stages S1-2, S4-5, and S7-8, respectively. Although temperature increases were evident across all stages, the rate of increase varied. This variation was most pronounced with the introduction of higher heat flux in stages 7 and 8 (Fig. 5d), which acted as temperature-boosting zones with significantly higher temperature difference. In contrast, stages 2–7 experienced a temperature compression zone, with temperature differences being smaller than those in the baseline. However, when analyzing distillate fluxes at each stage (Fig. 5g), a consistent increase was observed in all stages for both STB cases. This could be attributed to the overall temperature rise in the device, enhancing evaporation and condensation processes. The elevated temperatures likely facilitated a more efficient transfer of gaseous water within the gap.

We verified these observations using 2D full-scale numerical simulations with COMSOL Multiphysics software (Fig. 5e), where heat and moisture transport were simultaneously solved (see Methods section for details). The numerically and experimentally obtained distillate fluxes are in close agreement, with a variation of only ±10% (Supplementary Fig. 11). By applying a heat flux of 250 W m−2 in stages 7 and 8, we observed not only a temperature increase but also a rise in vapor concentration (mol m−3). This increase is primarily attributed to the elevated saturation vapor pressures at the evaporation and condensation temperatures, denoted as \({p}_{{{{\rm{w}}}},{{{\rm{sat}}}}}\left({T}_{{{{\rm{e}}}},{{{{\rm{S}}}}}_{{{{\rm{n}}}}}}\right)\) and \({p}_{{{{\rm{w}}}},{{{\rm{sat}}}}}\left({T}_{{{{\rm{c}}}},{{{{\rm{S}}}}}_{{{{\rm{n}}}}}}\right)\), respectively. This relationship is due to the exponential correlation between saturation vapor pressure and temperature, as illustrated in Fig. 1c, d.

The performance metrics of a baseline configuration demonstrated a notable improvement compared to a device with the same number of stages using a solar absorber, showing an increase in water production from 4.4 L m−2 h−1 to 6.5 L m−2 h−1. The latter value aligned the performance of this passive 8-stage device with that of a state-of-the-art active solar-driven vacuum-multi-effect-membrane-distillation module, which achieved 7 L m−2 h−1,65. The elevated water production was accompanied by an increase in solar-to-water efficiency from 302% to a record-high 439%. The 48% higher fluxes and 45% increase in efficiency can be attributed to two factors. First, the device equipped with a photothermal membrane exhibited a more efficient conversion of solar energy to vapor, as the assembly with the photothermal membrane reached peak temperatures at the feed water–membrane–air gap interface (Supplementary Fig. 10c). This is in contrast to the device with the solar absorber (Supplementary Fig. 10d), which exhibited the peak temperature positioned above the water gap. However, temperature polarization in the case of interfacial solar heating is greatly reduced66, further enhancing the solar-to-water efficiency. Second, the inclination angle significantly influences performance. Both the photothermal (Supplementary Fig. 10a) and solar absorber (Supplementary Fig. 10b) devices displayed temperature non-uniformity and noticeable velocity gradients. However, the device with the photothermal membrane showed peak velocities of up to 1.1 m/s in the air gap (Supplementary Fig. 10e), while the velocity in the air gap of a device with a solar absorber was significantly reduced, experiencing more than a 50% decrease (Supplementary Fig. 10f). Notably, the x-component of the diffusive moisture flux was affected (Supplementary Fig. 4) by inclination-induced moist air vortices, unlike in the horizontal direction. In conclusion, the horizontal positioning of a photothermal membrane-based MD suppresses velocity development in both gaps. Simultaneously, the highest temperature being closer to the evaporation side significantly enhances the performance of the photothermal membrane device in the first stage (~1.2 L m−2 h−1 S−1, Fig. 5g), yielding a 50% larger value compared to the device utilizing a solar absorber (~0.8 L m−2 h−1 S−1, Fig. 4f).

At a total heat flux of 1500 W m−2, the device reached a record-high distillate flux of 9.0 L m−2 h−1, accompanied by a solar-to-water efficiency of 407% (Fig. 4f). However, increasing the heat flux further to 2000 W m−2 led to a slight decrease in performance, with water production fluxes dropping to 8.4 L m−2 h−1 and a reduction in efficiency to 283%.

To contextualize these STB results (Figs. 4d, 5f) within the existing state of the art for multistage TD distillation devices16,17,23,26,27,28,29,30,31,32,33,34, we plotted diagrams showing the relationship between distillate fluxes and energy-to-water efficiency (Fig. 6a). A comprehensive overview of the data for these TD multistage distillation devices is available in Supplementary Table 4. The data revealed that all existing non-STB devices followed a linear trend, with incremental increases in water production fluxes corresponding to elevated energy-to-water efficiencies. In stark contrast, our STB-equipped devices demonstrated a substantial leap in distillate fluxes, with efficiencies reaching up to 400%, a milestone only recently achieved23. Furthermore, when scaling these metrics per stage (Fig. 6b), the performance jump becomes even more pronounced. The 8-stage photothermal membrane and 5-stage MD devices represent a major breakthrough, offering efficient and ultra-high fluxes for compact multistage membrane distillation units. For comparison, we also included results from a 3-sun thermally concentrated 6-stage distiller31. While this distiller achieved improved distillate fluxes, it was at the considerable expense of solar energy efficiency, leading to significantly lower overall efficiency.

a Distillate flux in relation to energy-to-water efficiency and b both metrics scaled per stage S. The ‘No STB’ region is delineated by the placement of existing devices on the diagram, whereas ‘With STB’ refers to a significant increase in fluxes, concurrently maintaining state-of-the-art efficiencies. PTM photothermal membrane, SA solar absorber, EH electric heater. Additional information regarding the data presented in the figure can be found in Supplementary Table 4.

Significant stage-averaged distillation fluxes were observed from a comprehensive set of experimental data from both photothermal MD and solar absorber 8-stage devices. This occurred when nearly one-third of the heat input (STB7,8, each at 250 W m−2) offered minimal latent heat recovery potential. However, the performance of early stages relative to the total distillate flux shows that in the case of STB-enabled devices, the first three stages contribute 43.5% to the overall flux. By contrast, the first three stages of the device without the STB contribute 51% (Supplementary Fig. 12). This indicates that the STB further activates the lowest stages of the multistage design, as observed in experimental results from both 8-stage MD devices using photothermal membrane and solar absorber. This observation highlights the implications of the STB by shifting the operational temperatures to higher regions and directly injecting heat into the lower stages.

The thermally driven MD process hinges on the synergy between the evaporation process and energy input. Due to the intermittent nature of solar energy, waste heat, acknowledged as a renewable heat source under EU Directive 2018/200167, provides an untapped thermal energy reservoir suitable for MD systems. In the context of emphasizing waste heat’s significance in high-level EU strategies and policies as an energy efficiency measure, the EU expels more waste heat than the entire building stock necessitates68 (8.4 PWh69). In China, 42% of the total primary energy input manifests as waste heat, equating to a remarkable 16.2 PWh70. In the U.S. industrial sector, on the other hand, up to 3.8 PWh per year of energy input is transformed into waste heat71. Given that the global primary energy consumption was 179 PWh in 202272, the aforementioned waste heat quantities represent a staggering 16% of global primary energy use. One essential characteristic of waste heat is its temperature level, which affects its work potential based on Carnot efficiency (see Supplementary Note 1). Globally, 63% of waste heat is available at ultra-low temperatures below 100 °C73, aligning with the typical thermal energy grade required for simple MD systems. Low-grade heat, with temperatures ranging from 35 to 65 °C, is often discharged by refrigeration and air-conditioning equipment74,75. Nevertheless, water-stressed regions, such as the Middle East and North Africa, in addition to South Asia—where 83% and 74% of the population, respectively, face severe high water scarcity76—typically exhibit substantial cooling needs. Harnessing and repurposing waste heat generated during various processes in industrial, power generation facilities, and the tertiary sector represents a carbon-neutral energy source. This approach ensures a viable pathway towards a steady supply of stage-temperature-boosted distillate fluxes, regardless of climate conditions and the fluctuations in solar energy throughout the day.

Electrical energy, which owes its commercialization and practical application largely to Thomas Edison, has become a fundamental cornerstone of modern civilization. Energy conversion systems, including gas, coal, biomass, photovoltaic and nuclear power plants, convert various sources of energy—namely chemical, solar and nuclear—into electricity. However, during this conversion process, approximately 55 to 85% of the energy is lost as thermal energy to the environment (Fig. 7a). Additionally, numerous industrial processes generate substantial heat, which often remains unutilized. The second law of thermodynamics dictates that work and heat are not equivalent (see Supplementary Note 1). Various forms of energy—such as heat, electrical, and solar—possess differing amounts of available work, termed exergy. Figure 7b illustrates the available work in energy flux. Since electrical energy is purely exergy, its work-to-energy flux ratio is 1. Solar energy contains a slightly lower amount of exergy, as defined theoretically by Petela77. In contrast, electrical energy generated by photovoltaic systems exhibits a significantly higher exergy level than heat at temperatures ranging from 35-65 °C. Photovoltaic (PV) conversion efficiencies, reported at 39.5% for the highest 1 sun efficiency solar cell by the National Renewable Energy Laboratory (NREL)78 and 20% for commercially available panels, further illustrate this disparity. Moreover, the typical real-world efficiency of a Stirling engine79, which converts heat to mechanical work, ranges from 5 to 10%. This is due to factors like mechanical losses, heat transfer inefficiencies, and other practical limitations, which further diminish the work potential of ultra-low-temperature heat sources.

a An overview of electricity and waste heat generation by assessing conversion efficiencies in gas, coal, biomass, and nuclear power plants. b Analysis of available work in energy flux encompassing pure electrical energy, the theoretical upper bound of solar energy, electrical energy produced by a solar cell with the highest reported efficiency of 39.5%78 under 1 sun conditions, commercially available PV panels with a conversion efficiency of 20%, and ultra-low waste heat with temperature levels 35–65 °C. c Comparison of STB-MD with existing state-of-the-art TD distillation devices16,17,23,26,27,28,29,30,31,32,33,34 in terms of distillate fluxes, focusing on the amount of water produced (L) per unit of exergy used (kWh−1) across the entire device and d per MD stages. e Enhancement in water production per unit of exergy use, as predicted by numerical simulations for a 16-stage STB-MD device equipped with a photothermal membrane, with an additional comparison to a photovoltaic + reverse osmosis system80 and experimentally tested 8-stage STB-MD with photothermal membrane. f Comparative assessment of state-of-the-art16,17,23,26,27,28,29,30,31,32,33,34, experimentally tested 5- and 8-stage STB-MD, predicted 16-stage STB-MD device, and various PV+RO systems81,82,83,84,85,86,87,88,89,90. PTM photothermal membrane, SA solar absorber, EH electric heater. Source data are provided as a Source Data file.

To assess the potential of STB-MD devices, we calculated water production relative to exergy use, employing Eq. 4 for solar energy and Eq. S4 for thermal energy. Figure 7c illustrates the distillate fluxes in relation to water production per exergy use (L kWh−1). It demonstrates that the PTM 8S and SA 8S, representing 8-stage STB-MD devices with photothermal membranes and solar absorbers, respectively, achieve significant advancements in terms of water production per exergy use. Most notably, PTM 8S surpasses 9.3 L kWh−1. In other words, ultra-low thermal energy, typically characterized by low available work potential, greatly enhances the performance of such systems. Furthermore, for scenarios where compactness and a reduced number of stages are key factors, Fig. 7d reveals that a 5-stage STB-MD device, utilizing immersed electrical heaters to simulate waste heat injection, exhibits remarkable efficiency when considering water production per exergy use per stage almost reaching 1.2 L kWh−1 S−1.

Using our developed numerical simulation model, we predicted the performance of a 16-stage PTM device under various configurations: STB15,16 at 250 W m−2, STB7,8 at 500 W m−2, STB2,3,7,8 at 250 W m−2, and STB2,3 at 500 W m−2 (Fig. 7e). We then compared these results with a photovoltaic + reverse osmosis system (PV+RO). While an 8-stage PTM using STB-MD does not match the performance of this specific PV+RO system80, increasing the number of stages to 16—thereby enhancing vaporization enthalpy recycling capability—results in a significant performance improvement over the 8-stage PTM device. Furthermore, we plotted existing state-of-the-art TD distillation devices16,17,23,26,27,28,29,30,31,32,33,34, experimentally tested STB-MD devices, and the predicted performance of the 16-stage PTM device against various PV+RO systems81,82,83,84,85,86,87,88,89,90 (Supplementary Table 5) in terms of water production per exergy use (Fig. 7f). Interestingly, while 8-stage SA and PTM STB-MD devices are competitive with PV+RO systems in the fourth quartile of PV+RO system performance, a 16-stage PTM device activated by waste heat excels in the second and third quartiles. A primary reason for this performance uplift is the introduction of additional air gap layers, which act as extra thermal resistance. This significantly raises the temperatures in the higher stages, approaching the upper limit of 95 °C. Such an increase in temperature is accompanied by a steep rise in evaporative mass flux, as depicted in Fig. 2.

This finding highlights a substantial opportunity for next-generation MD devices. Hybrid heat provision from the top and within the lower stages opens up numerous possibilities for utilizing ultra-low-temperature waste heat. This approach can significantly augment distillate fluxes, paving the way for highly efficient water production in MD systems. However, the potential fluctuating nature of waste heat in terms of quantity and quality—temperature—means that the STB-MD device can achieve higher distillate flux with waste heat injection while still maintaining high performance even at lower waste heat temperatures or waste heat flux levels. This flexibility in waste heat injection does not compromise the performance of the original multistage design. Furthermore, policymakers and spatial planners need to carefully address the challenge of possible instabilities in waste heat provision. In such cases, experts need to carefully consider the location of the future distillation plant to maximize the economic benefit of using waste heat for enhanced distillate flux production. For instance, waste heat from cooling towers, which is abundant year-round, can be effectively utilized. Additional risk mitigation measures to tackle unstable waste heat sources include implementing thermal energy storage systems and flexible operational strategies.

Thermally driven multistage membrane distillation devices offer tremendous potential for freshwater provision, especially in landlocked regions with terrestrial water bodies. Our study resulted in the development of a multistage MD device incorporating the concept of stage temperature boosting. Our approach, supported by both theoretical analysis of water diffusion in an air gap and nearly 50 experimental investigations, focused on the effects of heat amount and injection location in the lower stages. The comprehensive analysis culminated in the assembly of three STB-MD devices. We demonstrated that injecting heat strategically into stages other than the top one, specifically by heating the condenser rather than cooling it, significantly improved distillate fluxes, achieving up to 9.0 L m−2 h−1 with a solar-to-water efficiency of 407%. This signifies a 50% increase in distillate flux in an 8-stage system compared to conventional 10-stage systems. Remarkably, we also report a record-high per-stage distillate flux of 1.13 L m−2 h−1 S−1, surpassing the state-of-the-art by ~88%23. Such elevated distillate fluxes also contribute to reducing the costs of produced water by up to 75% (0.078¢ L−1, Supplementary Note 2, Supplementary Table 1) compared to the state-of-the-art 10-stage MD device (0.31¢ L−1)17. Furthermore, the study highlighted the efficient use of ultra-low thermal energy. By employing 2nd law efficiency analysis, we showed that systems with more than 16 stages could actively compete with existing PV+RO systems in certain scenarios. This finding underscores the need for policy updates in PV system electricity production and its usage to enhance global energy efficiency and optimize energy utilization in membrane distillation systems. We anticipate that our STB concept will spur systematic studies into the future development of compact, multistage MD systems. Furthermore, the STB concept offers a cost-effective alternative to enhance distillate fluxes, particularly compared to solutions dependent on costly optical solar concentration and thermally concentrated MD systems. Furthermore, our study demonstrates the potential to transform freshwater production and significantly reduce the carbon footprint of future membrane distillation systems through waste heat utilization. These innovations are expected to significantly enhance distillate fluxes using ultra-low-temperature heat sources, aligning closely with several of the UN’s Sustainable Development Goals (SDGs), particularly SDG 6, 7, 11, and 13, which focus on clean water and sanitation, affordable and clean energy, sustainable cities and communities, and climate action.

Carboxylated multiwalled carbon nanotubes (CNTs, Energy Chemical, Anhui Zesheng Technology Co., Ltd., LOT#DMERRRDZ) were first ultrasonically dispersed (SCIENTZ-IID, Ningbo Scientz Biotechnology Co., Ltd.) in isopropyl alcohol (Aladdin Biochemical Technology Co., Ltd., ≥ 99.5%) for 3 h to achieve a concentration of 1 g L−1. Following this, we utilized the dispersed emulsion in a spray-coating process. Using a spray coating pistol, we evenly coated the raw membrane, holding the pistol at an angle of 30–40° relative to the horizontal plane and maintaining a distance of 15–20 cm from the membrane. During this process, the membrane was securely positioned on a hot plate set at a temperature of 160 °C to facilitate immediate drying.

Once the membrane was dry, it was then immersed in a solution for further treatment. Specifically, we soaked the CNT-coated membrane in a 1 wt% 1H,1H,2H,2H-perfluorooctyltriethoxysilane (Aladdin Biochemical Technology Co., Ltd., ≥ 96%) solution, using n-hexane (Aladdin Biochemical Technology Co., Ltd., ≥ 98%) as a solvent for 60 s. This step was crucial for enhancing the hydrophobic properties of the membrane, thereby improving its efficiency in the distillation process.

A 5-stage STB-MD device with all-electric heaters and an 8-stage STB-MD device featuring a solar absorber in the top stage and electric heaters in the other stages were assembled using 3D printed frames nylon 7500 (WeNext Technology). The Teflon tapes, which enclosed the PTFE membranes (bought from Alibaba.com) within the nylon frame for sealing, were used to enhance the structural integrity of the devices. In the first stage, an assembly consisted of a glass cover (10 × 10 cm² with 2 mm thickness) along with the solar absorber (B-SX/T-L/Z-Z-1.88, Linuo Paradigma, 9.8 × 9.8 cm² solar-absorbing area with 0.2 mm thickness) was positioned in the slot. Condensation heat dissipation to the ambient was achieved using an aluminum heat sink adhered to the bottom of the device, which was partially immersed in a pool. The devices were inclined at 30° relative to the horizontal plane.

An 8-stage STB-MD device with a photothermal membrane was assembled using the following configuration for each stage: a silicone frame for the feed water gap (4 mm), a membrane (~150 μm), a silicone frame for the air gap (4 mm), and a 1 mm thick aluminum plate. The first stage had the following setup: 2 mm thick acrylic glass, a silicone frame for the feed water gap (4 mm), the photothermal membrane, a silicone frame for the air gap (4 mm), and a 1 mm thick aluminum plate. Each aluminum plate was equipped with 3-4 silicone spacers measuring 2–3 mm in height, 6 cm in length, and 0.5 cm in width. These spacers were attached to the aluminum plate to prevent sagging of the membrane when filled with feed water. To ensure reliable sealing and a flexible modular design, the frames were designed with twelve screw holes for connecting adjacent stages together.

For all the experiments using all three STB-MD devices, we used a microporous membrane with a nominal pore size of 1.0 µm.

All other important details concerning the feed water and air gap thickness, as well as dimensions of the device, can be found in the sub-section “Design and material characterization of an STB-MD device”, while the description of the experimental setups is found in sub-sections “Mechanistic impact of STB on water production enhancement” and “Performance of eight-stage STB-MD device with photothermal membrane”.

Temperature measurements were conducted using K-type thermocouples, connected with the Keithley 2700 Integra series multimeter data acquisition system (DAQ). This system records the voltage from the thermocouples and then computes the temperature using a 9th-degree polynomial. The associated error from converting voltage to temperature through this polynomial method ranges between 0.04 °C and −0.05 °C. For these measurements, we utilized 1 mm outer diameter K-type bare-wire thermocouples, which have a measurement uncertainty of ±0.1 °C.

Membrane distillation experiments were conducted under one sun illumination provided by a solar simulator (Non-uniformity A <2%, spectral match A, temporal stability A+ <1%; Sciencetech, Nmerry Technology Co., Ltd.). Collected liquid water was measured using a digital balance (accuracy 0.001 g; for distillate measurements, it indicates less than ±0.1%, ME503TE, Mettler Toledo). Solar irradiance uniformity was checked by the optical power meter (accuracy 7–14 μV W−1 m−2; TBQP5T24SW, Wuhan Hanqin System Science&Technology Co., Ltd.).

Membrane distillation experiments with 5-stage and 8-stage STB-MD devices were conducted using immersed electric heaters (~0.5 mm thickness) with a nominal resistance of 20Ω. The resistance was measured using a multimeter (accuracy 0.01 mV and 0.1 mA; B35T+, Owon).

For the experimental assessment of an 8-stage MD STB device equipped with a photothermal membrane, we utilized a constant thermostatic bath (accuracy 0.01 °C; Hengping Instruments, Shanghai, DC-0506) as the source of cooling water.

The average standard deviation of all the data concerning 5- and 8-stage STB-MD devices (Fig. 4) was determined to be 2.6%. We evaluated an additional source of measurement uncertainty related to the distillate weighing. However, given the digital balance accuracy of 1 mg, as specified by the manufacturer, the measurement accuracy after a 1-hour experiment is less than 0.1% since the amount of extracted water in any stage is always greater than 1 g.

We carried out specific measurements to analyze the spectral properties of CNT membrane, CNT+FPTS membrane (hereafter referred to as photothermal membrane (PTM)), solar absorber and raw membrane, focusing on absorptivity across the UV-Vis-NIR spectrum (0.3–2.5 μm). For this purpose, a PerkinElmer LAMBDA 950 spectrometer was employed to measure the absorptivity in the UV-Vis-NIR range. These measurements were all taken at normal incidence. To calculate the integrated values, we applied a weighted integration method within this spectral range. This integration process utilized the standard solar spectrum (AM1.5 Global–ASTMG173) as a reference for the 0.3–2.5 μm spectrum range.

The morphology of membranes was examined by SEM (Nova NanoSEM 230). The contact angle of water was measured by a surface contact angle meter (Data physics OCA20) at ambient temperature (~24 °C) using a 5 μL water droplet as the indicator.

AFM measurements were performed on a FastScan Bio atomic force microscope (Bruker, Germany) with tapping mode, and the used probe was Ti/Pt coated Si (OMCL-AC240TS).

Confocal images of surface morphology were collected on a VK-X3000 laser scanning microscope (Keyence, Japan) under ambient atmosphere.

Exergy of thermal radiation processes, including solar radiation, has been studied by several authors. According to Petela77 and Press91, the exergy factor of solar radiation is defined as:

Therefore, the exergy flux flow from solar radiation to the solar-absorbing surface is:

where \({T}_{0}\) is ambient temperature, K, T corresponds to the temperature of the sun (5780 K), \({I}_{{{{\rm{s}}}}}\) denotes solar intensity, W m−2.

Theoretical simulation models were performed for two separated two-dimensional (2D) models representing 1. an 8-stage STB-MD device (Fig. 5e) with a photothermal membrane and 2. the top stage of the STB-MD devices with a solar absorber and photothermal membrane (Fig. 4c, Supplementary Figs. 4, 10). Supplementary Fig. 13 shows the numerical simulation framework for both simulation cases to help the reader better understand the physical model and boundary conditions.

The theoretical model, developed in a COMSOL Multiphysics simulation environment (version 6.0, release year 2021), allowed us to simultaneously solve a coupled heat transfer (HT) in solid and fluid domains, laminar nonisothermal fluid flow in the air gap domains (not applicable to 8-stage STB-MD device with photothermal membrane), surface-to-surface radiative HT between the photothermal membrane, acrylic glass and ambient (not applicable for the top stage analysis), moisture flow and diffusion (advection, diffusion) for the top stage analysis, and diffusion for 8-stage STB-MD device with photothermal membrane. The study allowed us to obtain the temperature, velocity, diffusive moisture flux, and vapor concentration fields, and surface radiosity field (where surface-to-surface radiation boundary condition is prescribed). During the simulation analysis, we employed the following assumptions:

The thermophysical properties of all solids are assumed to be constant and do not change with temperature.

The validity of Kirchhoff’s law of thermal radiation for the radiative heat exchange between the photothermal membrane, acrylic glass and ambient is assumed (ε = α at a given wavelength and temperature).

The spectral properties of the carbonous-based photothermal membrane and acrylic glass are assumed constant across all simulated wavelengths.

The photothermal membrane is assumed to be opaque surface, while the feed water layer in the first stage and acrylic glass are treated as a transparent medium.

We disregarded contact resistances inside the entire solid domain.

The 8-stage STB-MD device equipped with a photothermal membrane is assumed to be positioned perfectly horizontally.

We disregarded the formation of liquid water on the condenser plate’s surface, an aluminum plate, and its impact on the condensation process.

The evaporation from the membrane-like evaporation surface is assumed to be uniform across the entire surface.

Governing equations that describe the moist air concerning continuity, momentum (both Navier-Stokes equations), and energy equation, considered in conservation form, are solved simultaneously for the moist air:

where \({\rho }_{{ma}}\) is the density of moist air, kg m−3, \({{{{\bf{u}}}}}_{{ma}}\) is the velocity vector of moist air, m s−1; \({T}_{{ma}}\) is the temperature of moist air, K; \({{{\bf{g}}}}\) is the gravity vector, 9.81 m s−2 in the negative y-direction; \({p}_{{ma}}\) is the pressure inside the air gap of the device, Pa; \({k}_{{ma}}\) is the thermal conductivity of moist air, W m−1 K−1; \({\mu }_{{ma}}\) is the dynamic viscosity of the moist air, Pa s; \({C}_{p,{ma}}\) is the specific heat at constant pressure of moist air, J kg−1 K−1, \(Q\) is a heat source, W m−3.

The moist airflow velocity field (u), pressure (p), and temperature (T) are solved by addressing the Navier-Stokes equations alongside the energy equation. Moist air within the air gap domain, presumed to exhibit Newtonian properties, was modeled as a weakly compressible flow incorporating the acceleration due to gravity. Additionally, a no-slip boundary condition was applied to the walls of the air gap domains.

The moisture distribution was determined by solving an additional transport equation, which accounts for the advection and binary diffusion of vapor in moist air:

Where \({C}_{{{{\rm{v}}}}}\) is the concentration of the vapor in the moist air, mol m−3; \({D}_{{{{\rm{v}}}}}\) is the water vapor molecular diffusivity in moist air, m2 s−1, \({M}_{{{{\rm{v}}}}}\) is the molecular (molar) mass of water vapor, kg mol−1; G is the moisture flux, kg m−2 h−1. \({C}_{{{{\rm{v}}}}}\) calculation is made by determining the relative humidity and saturation concentration of vapor, as described in the following equation:

where \(\varphi\) is relative humidity, /; \({C}_{{{{\rm{v}}}},{{{\rm{sat}}}}}\) is saturation concentration of water vapor, mol m−3. Relative humidity is defined as the ratio of the water vapor’s partial pressure in the air (\({p}_{{{{\rm{v}}}}}\)) to its saturation pressure (\({p}_{{{{\rm{v}}}},{{{\rm{sat}}}}}\)) at a specific moist air temperature \(\left({T}_{{{{\rm{ma}}}}}\right)\):

The set of equations presented above (Eqs. 5–8) has been employed for numerically simulating the top stage of the STB-MD devices equipped with both solar absorber and photothermal membrane. Based on the assessment of the inclined versus horizontally placed air and water gap domains (Fig. 4c, Supplementary Figs. 4, 10), we disregarded the laminar nonisothermal fluid flow in the air gap domains. Thus, only the diffusional moisture transport across the air gaps was considered, which considerably simplified the calculation time. Such a simplification was employed for both 8-stage STB-MD device (Fig. 5e) with a photothermal membrane and the top stage of the STB-MD devices.

The following boundary conditions were prescribed for simulating the top stage of the STB-MD devices with solar absorber and photothermal membrane (Fig. 4c, Supplementary Figs. 4, 10):

Constant temperature boundary conditions were applied to photothermal membrane and solar absorbers:

and to aluminum plate:

Both temperatures have been obtained from actual experimental data.

A constant moisture flux boundary condition was applied to the photothermal membrane (the top stage with photothermal membrane) and raw membrane (the top stage with solar absorber):

This value has been obtained from actual experimental data.

Heat transfer between the acrylic glass above the feed water gap and the ambient environment was modeled as a convective heat flux boundary condition. This was calculated using external natural convection formulas for inclined and horizontal walls.

Water vapor condensation was modeled with the moist surface boundary condition applied to the condenser’s inner surfaces. The condensate mass flux, kg m−2 s−1, was determined using the equation below:

where K is the evaporation/condensation rate factor, m s−1.

The evaporative and condensation heat fluxes were prescribed with constant heat flux boundary conditions. The absorbed evaporation and released condensation heat fluxes, W m−2, owing to the phase change of water was calculated using the latent heat of vaporization/condensation:

where \({h}_{{{{\rm{fg}}}}}\) is the vaporization/condensation enthalpy, J kg−1.

To simulate the 8-stage STB-MD device with a photothermal membrane (Fig. 5), besides the already presented boundary conditions 3., 4., and 5. the following boundary conditions were prescribed:

Constant temperature boundary condition was applied to the bottom plate of an 8-stage and 16-stage STB-MD device using photothermal membrane:

The temperature has been obtained from actual experimental data.

The heat flux associated with the photothermal membrane was calculated using the following equation:

Where \({I}_{s}\) denotes solar irradiance, W m−2. The stage boosters, which simulate immersed electric heaters, were assigned a heat source boundary condition, W m−3, using the following equation:

The diffusivity of water vapor in air, m2 s−1, for each air gap across all stages, was calculated using the following equation51 by employing the user-defined function (UDF) capabilities:

Where \({T}_{{{{\rm{ma}}}},{{{{\rm{S}}}}}_{{{{\rm{n}}}}}}\) is temperature of moist air, K, and other variables are defined in the section where Eq. 1 is introduced.

The moisture flux boundary condition for each membrane has been calculated using Eq. 1, \({m}_{{{{\rm{w}}}},{{{{\rm{S}}}}}_{{{{\rm{n}}}}}}\) by leveraging the UDF.

The surface-to-surface radiation of a diffuse surface considered radiative heat transfer between the photothermal membrane, acrylic glass and the ambient. The net radiative heat flux that is absorbed/emitted by the surface is calculated using the following equation:

where \(\varepsilon\) is surface emissivity, /; \({I}_{{{{\rm{bb}}}}}\) is the blackbody hemispherical total emissive power, W m−2 at the surface temperature \(\left(T\right)\); \(G\) is surface irradiance, W m−2.

The ambient emissivity was approximated using the idealized assumption of a blackbody emitter at ambient temperature.

All the data needed to evaluate the conclusions are present in the paper and/or the Supplementary Information. Source data are provided with this paper.

Denissen, J. M. C. et al. Widespread shift from ecosystem energy to water limitation with climate change. Nat. Clim. Chang. 12, 677–684 (2022).

Article ADS Google Scholar

Mapulanga, A. M. & Naito, H. Effect of deforestation on access to clean drinking water. Proc. Natl Acad. Sci. USA 116, 8249–8254 (2019).

Article ADS PubMed CAS Google Scholar

Savelli, E., Mazzoleni, M., Di Baldassarre, G., Cloke, H. & Rusca, M. Urban water crises driven by elites’ unsustainable consumption. Nat. Sustain. 6, 929–940 (2023).

Article Google Scholar

Huang, Z., Yuan, X. & Liu, X. The key drivers for the changes in global water scarcity: water withdrawal versus water availability. J. Hydrol. 601, 126658 (2021).

Article Google Scholar

He, C. et al. Future global urban water scarcity and potential solutions. Nat. Commun. 12, 4667 (2021).

Article ADS PubMed CAS Google Scholar

Poredoš, P., Shan, H., Wang, C., Deng, F. & Wang, R. Sustainable water generation: grand challenges in continuous atmospheric water harvesting. Energy Environ. Sci. 15, 3223–3235 (2022).

Article Google Scholar

Xu, J. et al. Ultrahigh solar-driven atmospheric water production enabled by scalable rapid-cycling water harvester with vertically aligned nanocomposite sorbent. Energy Environ. Sci. 14, 5979 (2021).

Article CAS Google Scholar

Shan, H., Poredoš, P., Zou, H., Lv, H. & Wang, R. Perspectives for urban microenvironment sustainability enabled by decentralized water-energy-food harvesting. Energy 282, 129009 (2023).

Article Google Scholar

Shao, Z. et al. High-performance solar-driven MOF AWH device with ultra-dense integrated modular design and reflux synthesis of Ni2Cl2(BTDD). Device 1, 100058 (2023).

Article Google Scholar

Shan, H. et al. High-yield solar-driven atmospheric water harvesting with ultra-high salt content composites encapsulated in porous membrane. Cell Rep. Phys. Sci. 2, 100664 (2021).

Article CAS Google Scholar

Deng, F., Wang, C., Xiang, C. & Wang, R. Bioinspired topological design of super hygroscopic complex for cost-effective atmospheric water harvesting. Nano Energy 90, 106642 (2021).

Article CAS Google Scholar

Shan, H. et al. Exceptional water production yield enabled by batch-processed portable water harvester in semi-arid climate. Nat. Commun. 13, 5406 (2022).

Article ADS PubMed CAS Google Scholar

Humphrey, J. H. et al. The potential for atmospheric water harvesting to accelerate household access to safe water. Lancet Planet. Health 4, e91–e92 (2020).

Article PubMed Google Scholar

Shan, H. et al. Harvesting thermal energy and freshwater from air through sorption thermal battery enabled by polyzwitterionic gel. ACS Energy Lett. 8, 5184–5191 (2023).

Article ADS CAS Google Scholar

Elimelech, M. & Phillip, W. A. The future of seawater desalination: energy, technology, and the environment. Science 333, 712–717 (2011).

Article ADS PubMed CAS Google Scholar

Xu, Z. et al. Ultrahigh-efficiency desalination: via a thermally-localized multistage solar still. Energy Environ. Sci. 13, 830 (2020).

Article CAS Google Scholar

Gao, J. et al. Extreme salt-resisting multistage solar distillation with thermohaline convection. Joule 7, 2274–2290 (2023).

Li, J. et al. A tandem radiative/evaporative cooler for weather-insensitive and high-performance daytime passive cooling. Sci. Adv. 8, eabq0411 (2023).

Article ADS Google Scholar

van Loosdrecht, M. C. M. & Brdjanovic, D. Anticipating the next century of wastewater treatment. Science 344, 1452–1453 (2014).

Article ADS PubMed Google Scholar

Deshmukh, A. et al. Membrane distillation at the water-energy nexus: limits, opportunities, and challenges. Energy Environ. Sci. 11, 1177–1196 (2018).

Article CAS Google Scholar

Guo, Q. et al. PTFE porous membrane technology: a comprehensive review. J. Memb. Sci. 664, 121115 (2022).

Article CAS Google Scholar

Shiohara, A., Prieto-Simon, B. & Voelcker, N. H. Porous polymeric membranes: fabrication techniques and biomedical applications. J. Mater. Chem. B 9, 2129–2154 (2021).

Article PubMed CAS Google Scholar

Zhu, Z., Zheng, H., Kong, H., Ma, X. & Xiong, J. Passive solar desalination towards high efficiency and salt rejection via a reverse-evaporating water layer of millimetre-scale thickness. Nat. Water 1, 790–799 (2023).

Article Google Scholar

Omar, A. et al. Experimental and numerical investigation of a new hollow fiber-based multi-effect vacuum membrane distillation design. Desalination 501, 114908 (2021).

Article CAS Google Scholar

Li, Q. et al. A novel concentrated solar membrane-distillation for water purification in a building integrated design. Desalination 535, 115828 (2022).

Article CAS Google Scholar

Chiavazzo, E., Morciano, M., Viglino, F., Fasano, M. & Asinari, P. Passive solar high-yield seawater desalination by modular and low-cost distillation. Nat. Sustain. 1, 763–772 (2018).

Article Google Scholar

Xue, G. et al. Highly efficient water harvesting with optimized solar thermal membrane distillation device. Glob. Chall. 2, 1800001 (2018).

Article PubMed Google Scholar

Morciano, M., Fasano, M., Boriskina, S. V., Chiavazzo, E. & Asinari, P. Solar passive distiller with high productivity and Marangoni effect-driven salt rejection. Energy Environ. Sci. 13, 3646–3655 (2020).

Article CAS Google Scholar

Wang, W. et al. Simultaneous production of fresh water and electricity via multistage solar photovoltaic membrane distillation. Nat. Commun. 10, 3012 (2019).

Article ADS PubMed Google Scholar

Ghim, D., Wu, X., Suazo, M. & Jun, Y.-S. Achieving maximum recovery of latent heat in photothermally driven multi-layer stacked membrane distillation. Nano Energy 80, 105444 (2021).

Article CAS Google Scholar

Huang, L. et al. Enhanced water yield of solar desalination by thermal concentrated multistage distiller. Desalination 477, 114260 (2020).

Article CAS Google Scholar

Wang, W. et al. Integrated solar-driven PV cooling and seawater desalination with zero liquid discharge. Joule 5, 1873–1887 (2021).

Article Google Scholar

Dong, Y., Poredoš, P., Ma, Q. & Wang, R. High-yielding and stable desalination via photothermal membrane distillation with free-flow evaporation channel. Desalination 543, 116103 (2022).

Article CAS Google Scholar

Wang, W. et al. Solar seawater distillation by flexible and fully passive multistage membrane distillation. Nano Lett. 21, 5068–5074 (2021).

Article ADS PubMed CAS Google Scholar

Zhang, L. et al. Passive, high-efficiency thermally-localized solar desalination. Energy Environ. Sci. 14, 1771 (2021).

Article CAS Google Scholar

Choi, J. et al. Performance and economic analysis of a solar membrane distillation pilot plant under various operating conditions. Energy Convers. Manag. 268, 115991 (2022).

Article CAS Google Scholar

Zhang, L. et al. Modeling and performance analysis of high-efficiency thermally-localized multistage solar stills. Appl. Energy 266, 114864 (2020).

Article Google Scholar

Zhou, J., Wang, F., Noor, N. & Zhang, X. An experimental study on liquid regeneration process of a liquid desiccant air conditioning system (LDACs) based on vacuum membrane distillation. Energy 194, 116891 (2020).

Article CAS Google Scholar

Alwatban, A. M., Alshwairekh, A. M., Alqsair, U. F., Alghafis, A. A. & Oztekin, A. Effect of membrane properties and operational parameters on systems for seawater desalination using computational fluid dynamics simulations. Desalin. Water Treat. 161, 92–107 (2019).

Twibi, M. F. et al. Hydrophobic mullite ceramic hollow fibre membrane (Hy-MHFM) for seawater desalination via direct contact membrane distillation (DCMD). J. Eur. Ceram. Soc. 41, 6578–6585 (2021).

Article CAS Google Scholar

Zewdie, T. M., Habtu, N. G., Dutta, A. & Van der Bruggen, B. Flat sheet metakaolin ceramic membrane for water desalination via direct contact membrane distillation. Water Reuse 12, 131–156 (2022).

CAS Google Scholar

Silva, R. et al. Understanding the effects of operational conditions on the membrane distillation process applied to the recovery of water from textile effluents. Process Saf. Environ. Prot. 145, 285–292 (2021).

Article CAS Google Scholar

Alklaibi, A. M. & Lior, N. Comparative study of direct-contact and air-gap membrane distillation processes. Ind. Eng. Chem. Res. 46, 584–590 (2007).

Article CAS Google Scholar

Singh, D. & Sirkar, K. K. Desalination of brine and produced water by direct contact membrane distillation at high temperatures and pressures. J. Memb. Sci. 389, 380–388 (2012).

Article CAS Google Scholar

Singh, D. & Sirkar, K. K. High temperature direct contact membrane distillation based desalination using PTFE hollow fibers. Chem. Eng. Sci. 116, 824–833 (2014).

Article CAS Google Scholar

Luo, A. & Lior, N. Study of advancement to higher temperature membrane distillation. Desalination 419, 88–100 (2017).

Article CAS Google Scholar

Elmarghany, M. R., El-Shazly, A. H., Salem, M. S., Sabry, M. N. & Nady, N. Thermal analysis evaluation of direct contact membrane distillation system. Case Stud. Therm. Eng. 13, 100377 (2019).

Article Google Scholar

Lee, C.-K., Park, C., Woo, Y. C., Choi, J.-S. & Kim, J.-O. A pilot study of spiral-wound air gap membrane distillation process and its energy efficiency analysis. Chemosphere 239, 124696 (2020).

Article PubMed CAS Google Scholar

Poredoš, P. et al. Condensation of water vapor from humid air inside vertical channels formed by flat plates. iScience 25, 103565 (2022).

Article ADS PubMed Google Scholar

Poling, B. E., Prausnitz, J. M. & O’Connell, J. P. Properties of Gases and Liquids (McGraw-Hill Education, 2001).

Fuller, E. N., Schettler, P. D. & Giddings, J. C. A new method for prediction of binary gas-phase diffusion coefficients. Ind. Eng. Chem. 58, 18–27 (1966).

Shan, H., Ye, Z., Yu, J., Wang, R. & Xu, Z. Improving solar water harvesting via airflow restructuring using 3D vapor generator. Device 1, 100065 (2023).

Xu, S. et al. Microalgae film-derived water evaporation-induced electricity generator with negative carbon emission. Adv. Sci. 11, 2400856 (2024).

Article CAS Google Scholar

Yu, Z. et al. Micro–nano water film enabled high-performance interfacial solar evaporation. Nanomicro Lett. 15, 214 (2023).

ADS PubMed CAS Google Scholar

Ge, C.-X., Wu, Z.-S., Bai, J. & Gong, L. Effect of nanoscale roughness on optical trapping properties of surface plasmon polaritons exerted on nanoparticle. J. Quant. Spectrosc. Radiat. Transf. 219, 339–349 (2018).

Article ADS CAS Google Scholar

Yang, Z.-P., Ci, L., Bur, J. A., Lin, S.-Y. & Ajayan, P. M. Experimental observation of an extremely dark material made by a low-density nanotube array. Nano Lett. 8, 446–451 (2008).

Article ADS PubMed CAS Google Scholar

Wu, D., Wu, L., Zhou, W., Sun, Y. & Zhang, M. Relations between the aspect ratio of carbon nanotubes and the formation of percolation networks in biodegradable polylactide/carbon nanotube composites. J. Polym. Sci. B Polym. Phys. 48, 479–489 (2010).

Article ADS CAS Google Scholar

Rashed, A. O. et al. Carbon nanotube membranes—strategies and challenges towards scalable manufacturing and practical separation applications. Sep. Purif. Technol. 257, 117929 (2021).

Article CAS Google Scholar

Li, K. et al. Optimizing stretching conditions in fabrication of PTFE hollow fiber membrane for performance improvement in membrane distillation. J. Memb. Sci. 550, 126–135 (2018).

Article ADS CAS Google Scholar

Mofokeng, M. et al. Perflurooctyltriethoxy silane and carbon nanotubes-modified PVDF superoleophilic nanofibre membrane for oil-in-water adsorption and recovery. J. Environ. Chem. Eng. 8, 104497 (2020).

Article CAS Google Scholar

Boo, C., Lee, J. & Elimelech, M. Engineering surface energy and nanostructure of microporous films for expanded membrane distillation applications. Environ. Sci. Technol. 50, 8112–8119 (2016).

Article ADS PubMed CAS Google Scholar

Johnson, D. & Hilal, N. Polymer membranes—fractal characteristics and determination of roughness scaling exponents. J. Memb. Sci. 570–571, 9–22 (2019).

Article Google Scholar

Nguyen, D. T., Lee, S., Lopez, K. P., Lee, J. & Straub, A. P. Pressure-driven distillation using air-trapping membranes for fast and selective water purification. Sci. Adv. 9, eadg6638 (2024).

Article Google Scholar

Sianipar, M., Kim, S. H., Khoiruddin, Iskandar, F. & Wenten, I. G. Functionalized carbon nanotube (CNT) membrane: progress and challenges. RSC Adv. 7, 51175–51198 (2017).

Article ADS CAS Google Scholar

Zhao, K. et al. Experimental study of the memsys vacuum-multi-effect-membrane-distillation (V-MEMD) module. Desalination 323, 150–160 (2013).

Article CAS Google Scholar

Anvari, A., Azimi Yancheshme, A., Kekre, K. M. & Ronen, A. State-of-the-art methods for overcoming temperature polarization in membrane distillation process: a review. J. Memb. Sci. 616, 118413 (2020).

Article CAS Google Scholar

Directive (EU) 2018/2001 of the European Parliament and of the Council of 11 December 2018 on the promotion of the use of energy from renewable sources. Off. J. Eur. Union 5, 1–77 (2018).

European Commission. An EU Strategy on Heating and Cooling, 1–13 https://eur-lex.europa.eu/legal-content/EN/TXT/?qid=1575551754568&uri=CELEX:52016DC0051 (2016).

Pezzutto, S., Haas, F., Exner, D. & Zambotti, S. Europe’s building stock and its energy demand: a comparison between Austria and Italy in Smart and Sustainable Planning for Cities and Regions (eds Bisello, A. et al.) 35–47 (Springer, 2018).

Lin, Y., Chong, C. H., Ma, L., Li, Z. & Ni, W. Quantification of waste heat potential in China: a top-down Societal Waste Heat Accounting Model. Energy 261, 125194 (2022).

Article Google Scholar

U.S. Department of Energy. Waste Heat Recovery: Technology and Opportunities in U.S. Industry (U.S. Department of Energy, 2008).

Ritchie, H., Rosado, P. & Roser, M. Energy production and consumption. Our World Data https://ourworldindata.org/energy-production-consumption (2020).

Forman, C., Muritala, I. K., Pardemann, R. & Meyer, B. Estimating the global waste heat potential. Renew. Sustain. Energy Rev. 57, 1568–1579 (2016).

Article Google Scholar

Ragoowansi, E. A., Garimella, S. & Goyal, A. Realistic utilization of emerging thermal energy recovery and storage technologies for buildings. Cell Rep. Phys. Sci. 4, 101393 (2023).

Article Google Scholar

Poredoš, P. & Wang, R. Sustainable cooling with water generation. Science 380, 458–459 (2023).

Article ADS PubMed Google Scholar

Kuzma, S., Saccoccia, L. & Chertock, M. 25 Countries, Housing One-quarter of the Population, Face Extremely High Water Stress. https://www.wri.org/insights/highest-water-stressed-countries?itid=lk_inline_enhanced-template (2023).

Petela, R. Exergy of heat radiation. J. Heat. Transf. 86, 187–192 (1964).

Article Google Scholar

France, R. M. et al. Triple-junction solar cells with 39.5% terrestrial and 34.2% space efficiency enabled by thick quantum well superlattices. Joule 6, 1121–1135 (2022).

Article CAS Google Scholar

Al-Nimr, M., Khashan, S. A. & Al-Oqla, H. Novel techniques to enhance the performance of Stirling engines integrated with solar systems. Renew. Energy 202, 894–906 (2023).

Article Google Scholar

Manolakos, D., Mohamed, E. S., Karagiannis, I. & Papadakis, G. Technical and economic comparison between PV-RO system and RO-Solar Rankine system. Case study: Thirasia Island. Desalination 221, 37–46 (2008).

Article CAS Google Scholar

Aybar, H., Akhatov, J. S., Avezova, N. R. & Halimov, A. S. Solar powered RO desalination: investigations on pilot project of PV powered RO desalination system. Appl. Sol. Energy 46, 275–284 (2010).

Article Google Scholar

Shawky, H. A., Abdel Fatah, A. A., Abo ElFadl, M. M. S. & El-Aassar, A. H. M. Design of a small mobile PV-driven RO water desalination plant to be deployed at the northwest coast of Egypt. Desalin. Water Treat. 55, 3755–3766 (2015).

Article CAS Google Scholar

Dallas, S., Sumiyoshi, N., Kirk, J., Mathew, K. & Wilmot, N. Efficiency analysis of the Solarflow—an innovative solar-powered desalination unit for treating brackish water. Renew. Energy 34, 397–400 (2009).

Article CAS Google Scholar

Soric, A., Cesaro, R., Perez, P., Guiol, E. & Moulin, P. Eausmose project desalination by reverse osmosis and batteryless solar energy: design for a 1m3 per day delivery. Desalination 301, 67–74 (2012).

Article CAS Google Scholar

Bilton, A. M., Kelley, L. C. & Dubowsky, S. Photovoltaic reverse osmosis—feasibility and a pathway to develop technology. Desalin. Water Treat. 31, 24–34 (2011).

Article Google Scholar

Qiblawey, H., Banat, F. & Al-Nasser, Q. Laboratory setup for water purification using household PV-driven reverse osmosis unit. Desalin. Water Treat. 7, 53–59 (2009).

Article CAS Google Scholar

Mohamed, E. S., Papadakis, G., Mathioulakis, E. & Belessiotis, V. A direct coupled photovoltaic seawater reverse osmosis desalination system toward battery based systems—a technical and economical experimental comparative study. Desalination 221, 17–22 (2008).

Article CAS Google Scholar

Khayet, M., Essalhi, M., Armenta-Déu, C., Cojocaru, C. & Hilal, N. Optimization of solar-powered reverse osmosis desalination pilot plant using response surface methodology. Desalination 261, 284–292 (2010).

Article CAS Google Scholar

Clarke, D. P., Al-Abdeli, Y. M. & Kothapalli, G. The effects of including intricacies in the modelling of a small-scale solar-PV reverse osmosis desalination system. Desalination 311, 127–136 (2013).

Article CAS Google Scholar

Banat, F., Qiblawey, H. & Al-Nasser, Q. Design and operation of small-scale photovoltaic-driven reverse osmosis (PV-RO) desalination plant for water supply in rural areas. Comput. Water Energy Environ. Eng. 1, 31–36 (2012).

Press, W. H. Theoretical maximum for energy from direct and diffuse sunlight. Nature 264, 734–735 (1976).

Article ADS CAS Google Scholar

Download references

This research work was funded by the Research Fund for International Young Scientists of the National Natural Science Foundation of China [No. 52150410421, P.P.], the National Natural Science Foundation of China [No. 52376200, Z.X.], and the Fundamental Research Funds for the Central Universities (Shanghai Jiao Tong University), R.W. The authors thank the ITEWA (Innovative Team on Energy-Water-Air Nexus) research group members for their encouragement and support. The authors acknowledge the financial support from the state budget by the Slovenian Research Agency [Program/Project Nr. P2-0223].

These authors contributed equally: Primož Poredoš, Jintong Gao, He Shan.

Institute of Refrigeration and Cryogenics, Shanghai Jiao Tong University, 200240, Shanghai, China

Primož Poredoš, Jintong Gao, He Shan, Jie Yu, Zhao Shao, Zhenyuan Xu & Ruzhu Wang

Engineering Research Center of Solar Power & Refrigeration, MOE China, 200240, Shanghai, China

Primož Poredoš, Jintong Gao, He Shan, Jie Yu, Zhao Shao, Zhenyuan Xu & Ruzhu Wang

Laboratory for Sustainable Technologies in Buildings, University of Ljubljana, Faculty of Mechanical Engineering, Ljubljana, Slovenia

Primož Poredoš

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

Conceptualization: P.P., Z.X., and R.W. Methodology: P.P., J.G., and Z.X. Investigation: P.P., J.G., J.Y. and Z.S. Validation: P.P., J.G., H.S. and Z.X. Formal analysis: P.P., J.G. and Z.X. Visualization: P.P., J.G. and H.S. Data curation: P.P., J.G., and H.S. Funding acquisition: P.P., Z.X., and R.W. Project administration: P.P., Z.X. Resources: P.P., J.G., and H.S. Software: P.P. and J.G. Supervision: Z.X., and R.W. Writing—original draft: P.P., Z.X. Writing—review & editing: P.P., Z.X., and R.W.

Correspondence to Zhenyuan Xu or Ruzhu Wang.

The authors declare no competing interests.

Nature Communications thanks Amr Omar and the other anonymous reviewers for their contribution to the peer review of this work. A peer review file is available.

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions

Poredoš, P., Gao, J., Shan, H. et al. Ultra-high freshwater production in multistage solar membrane distillation via waste heat injection to condenser. Nat Commun 15, 7890 (2024). https://doi.org/10.1038/s41467-024-51880-y

Download citation

Received: 26 May 2024

Accepted: 21 August 2024

Published: 10 September 2024

DOI: https://doi.org/10.1038/s41467-024-51880-y

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative