Discarded PE separators were provided by Zhongxing Innovative Material Technologies Co. Ltd. [Supplementary Figure 1]. Mg(NO₃)₂·6H₂O and Ni(NO₃)₂·6H₂O were supplied by Shanghai Aladdin Biochemical Technology Co. Ltd. Poly(ethylene glycol) (PEG-200) and (NH₄)₆Mo₇O₂₄·4H₂O were purchased from Sinopharm Chemical Reagent Co. Ltd. Gelatin and glutaraldehyde were supplied by Shanghai Macklin Biochemical Co. Ltd. The non-woven cotton was procured from EAXAY.

Ni_xMo_yMg_z catalysts (where x, y, and z refer to the mole ratio of Ni, Mo, and Mg) were synthesized by the combustion method. Firstly, 30.00 g Ni(NO₃)₂·6H₂O, 2.64 g Mg(NO₃)₂·6H₂O, 0.36 g (NH₄)₆Mo₇O₂₄·4H₂O, and 10.00 g PEG-200 were mixed and milled thoroughly and then calcined at 350 °C for 15 min in a muffle furnace to prepare Ni₅Mo_0.1Mg_0.5 catalysts. Other catalysts with different proportions were prepared by fixing the molar amount of Ni salt added and adjusting the amounts of Mo salt and Mg salt. The detailed amounts of metal salts are listed in Supplementary Table 1.

Discarded PE separators (5.00 g) and Ni_xMo_yMg_z catalyst (0.50 g) were mixed in a crucible (30 mL) and subsequently carbonized in a muffle furnace to afford CNT_x-y-z (850 °C, 4 min). Then, CNT_x-y-z was refluxed in concentrated HNO₃ at 60 °C to produce oxidized CNTs (denoted as OCNT_x-y-z).

Firstly, OCNT_x-y-z (20.5 mg) was added to the gelatin solution (2 wt%, 460 μL) and stirred overnight at room temperature (ca. 25 ^oC, 12 h, 400 rpm). Then, the mixture was spread onto cotton cloth (6.15 cm²) and dried in air. OCNT_x-y-z evaporators used in the indoor water evaporation experiments were prepared after soaking and crosslinking with aqueous glutaraldehyde solution (5 wt%). For comparison, a modified cotton evaporator without OCNT_x-y-z was constructed using the same process. To meet the requirements of various scenarios, evaporators with different scales were fabricated using the same preparation process. The specific material quantities were listed in Supplementary Table 2.

An experimental system was constructed to simulate solar interfacial evaporation, including a sunlight simulator (Perfectlight, PLS-SEX300, China) and an analytical balance (Sunny Hengping Instrument, JA1003L, China) to track water mass loss. The surface temperature of evaporators was monitored using an infrared camera (Dongmei, DMI220, China). The computational methods of calculating evaporation rate and photothermal efficiency were described in Supplementary Note 1.

A polyethylene terephthalate (PET) substrate (15 × 5 cm²) was ultrasonically washed in deionized water and ethanol. Subsequently, an OCNT_x-y-z evaporator (15 cm²) was placed on the PET substrate and fixed with conductive adhesive as electrodes. The power generation electrodes (two L-shaped pieces) were fabricated and coated with epoxy resin. After drying at room temperature for 15 min, the encapsulated evaporation device was obtained. The generated voltage and current were monitored using a Source Measure Unit.

A high-resolution transmission electron microscope (HRTEM, FEI, Tecnai G2 F30, Netherlands) was used to investigate the microstructure of the materials. Other characterizations were described in Supplementary Note 2.

Ni_xMo_yMg_z catalysts were produced through the combustion method [Supplementary Figure 2]. As shown in Figure 1B, the NiO diffraction peaks at 2θ = 37.3°, 43.3°, 62.9°, 75.2°, and 79.6° are assigned to the (111), (200), (220), (311), and (222) planes, respectively. Peaks attributable to Ni appear at 2θ = 44.7° (111), 51.8° (200), and 76.3° (220). Interestingly, the diffraction intensities of Ni and NiO are affected by the addition of Mo and Mg [Supplementary Figure 3]. Compared with Ni, the diffraction intensity of NiO increases with an increase in the Mo/Ni mole ratio from 0 to 0.1/5; correspondingly, the yield of CNT_5-y-0.5 rises from 11.7 to 57.4 wt% [Supplementary Figure 4A]. Mo can effectively prevent sintering and improve the dispersion of Ni particles at high temperatures, thereby making more catalytically active sites accessible^[38]. As the Mo/Ni mole ration increases to 1/5, the yield of CNT_5-y-0.5 decreases sharply to 27.4 wt%. The strong interfacial interaction between metal and support promotes the generation of an inert phase that covers active sites^[39], resulting in the decrease of catalyst activity. Similarly, the yield of CNT_5-0.1-z increases first from 53.4 to 57.4 wt% (with the Mg/Ni mole ratio increasing from 0.1/5 to 0.5/5) and then decreases to 48.6 wt% (with the Mg/Ni mole ratio increasing from 0.5/5 to 2/5, Supplementary Figure 4B). However, no CNTs are produced when the Mg/Ni mole ratio is 0/5. During the carbonization process, catalyst activity is readily reduced because active sites are covered by deposited carbon. As an alkaline oxide, MgO in the composite catalyst prevents carbon deposition on the catalyst surface and improves the yield of CNT^[40]. As a result, the highest carbon yield reaches 57.4 wt% when the Ni/Mo/Mg mole ratio is adjusted to 5/0.1/0.5. Ni₅Mo_0.1Mg_0.5 possesses higher catalytic activity than other catalysts, which is inseparable from the synergistic effect of Mo and Mg. Raman spectroscopy reveals Mo-O, Ni-O, Ni-O-Mo, and Mo=O bonds within Ni₅Mo_0.1Mg_0.5 [Figure 1C]. However, the first-order vibration peak of MgO is absent in the Raman spectrum due to its own lattice symmetry^[41]. The morphology of Ni₅Mo_0.1Mg_0.5 consists of irregular clusters composed of nanoparticles with a size of 10-30 nm [Figure 1D, Supplementary Figures 5 and 6]. NiO crystal particles in Ni₅Mo_0.1Mg_0.5 are wrapped by amorphous MgO [Figure 1E]. The mass loss of Ni₅Mo_0.1Mg_0.5 at 800 °C is ca. 1.5% [Supplementary Figure 7], indicating its high thermal stability.

The effects of Mo and Mg on the overall performance of catalysts were studied using H₂-TPR and CO₂-TPD. The two reduction peaks of the H₂-TPR profile are observed at 200-300 and 300-500 °C [Figure 1F]. After the addition of Mo to the catalysts, the reduction peaks of Ni₅Mo_0.1Mg₀ and Ni₅Mo_0.1Mg_0.5 catalysts within 200-300 °C shift to the low-temperature region, indicating that their reducibility is stronger than that of Ni₅Mo₀Mg_0.5. Mo promotes the formation of smaller NiO clusters, thereby exposing more active sites on the catalyst^[42]. Furthermore, the reduction peaks at 300-500 °C shift to higher temperatures, implying a robust interaction between Mo and Ni. A wider temperature range was observed in the Ni₅Mo_0.1Mg_0.5 profile because mixed-metal clusters with stronger interactions formed after the Mg addition. The characteristic peaks at low CO₂ desorption temperatures represent the weakly alkaline adsorption sites on the Ni-O-support, while the peaks at high temperatures correspond to the strongly alkaline sites. Different from Ni₅Mo_0.1Mg₀, high-temperature CO₂ desorption peaks appear at 700-800 °C for Ni₅Mo₀Mg_0.5 and Ni₅Mo_0.1Mg_0.5 [Figure 1G], which confirms the presence of the stronger alkaline Mg-O site and the generation of strongly monodentate carbonate^[43]. Previous work showed that strong adsorption is beneficial for the efficient elimination of carbon deposition by strong basic sites (O^2-)^[44].

The morphology of carbon products is significantly affected by the mole ratio of Ni/Mo/Mg [Figure 2A]. CNT_x-y-z was oxidized using concentrated HNO₃ to prepare OCNT_x-y-z with enhanced hydrophilicity. By using Ni₅Mo_0.1Mg₀ as a catalyst, the carbon product (OCNT_5-0.1-0) is composed of agglomerated spherical carbon with a size of 50-200 nm [Figure 2B, Supplementary Figure 8]. By contrast, the addition of Mg to the catalyst results in CNT formation, implying that MgO in the catalyst plays a crucial role in suppressing carbon deposition and promoting CNT growth. When the Mg/Ni mole ratio increases to 2/5, the diameter of CNT becomes significantly wider, and the yield of CNT decreases [Supplementary Figure 9]. By using Ni₅Mo₀Mg_0.5 as a catalyst, the OCNT_5-0-0.5 consists of amorphous carbon and a few CNTs with a length of ca. 1-5 μm [Figure 2C, Supplementary Figure 10]. However, the yield of CNT_5-0-0.5 is only 11.7 wt%, reflecting the poor catalytic activity of Ni₅Mo₀Mg_0.5. When the Mo/Ni mole ratio increases to 0.1/5, OCNT_5-0.1-0.5 features a diameter of 20-30 nm and a length of 10-30 μm [Figure 2D and E, Supplementary Figures 11 and 12], and the yield of CNT_5-0.1-0.5 is the highest. The interlayer spacing of OCNT_5-0.1-0.5 is 0.34 nm, consistent with the interlayer spacing of graphite^[45]. The diameter of OCNT_5-0.1-0.5 is smaller than that of OCNT_5-0-0.5, which is attributed to the addition of the Mo element. In addition to enhancing catalyst activity, Mo effectively regulates the morphology of OCNT. As the Mo/Ni mole ratio increases to 1/5, OCNT_5-1-0.5 shows a smaller diameter and longer length than OCNT_5-0.1-0.5, but a large number of nano-carbon particles are formed [Supplementary Figure 13]. The above results prove that the appropriate addition of Mo and Mg synergistically enhances the yield and quality of CNT.

The phase structure of OCNT_5-0.1-0.5 was analyzed by X-ray diffraction (XRD) and Raman spectroscopy. A sharp graphite peak (002) in the XRD patterns appears at 2θ=26.5° [Figure 3A, Supplementary Figure 14A], proving the highly graphitized structure. After purification, the diffraction peak of Ni disappears, and the degree of graphitization is slightly enhanced. By fitting and calculating the peak area ratio between the D and G peaks, the I_G/I_D value of OCNT_5-0.1-0.5 is determined to be 0.60 [Figure 3B, Supplementary Figure 14B], indicating the presence of edge-unsaturated carbons, asymmetric carbon species, and/or sidewall lattice defects^[46]. The thermal stability of OCNT_5-0.1-0.5 is studied by Thermal gravimetric analysis (TGA) and derivative thermogravimetry (DTG) curves. The maximum weight-loss rate occurs at 625 °C, and the weight loss after complete oxidation decomposition is 98% [Figure 3C], indicating good thermal stability and high purity. The specific surface area of OCNT_5-0.1-0.5 is 190.1 m² g^-1 [Figure 3D and E], higher than that of OCNT_5-0.1-0 (95.6 m² g^-1) or OCNT_5-0-0.5 (141.2 m² g^-1). OCNT_5-1-0.5 exhibits a specific surface area of 167.18 m² g^-1, whereas that of OCNT_5-0.1-2 reaches 191.0 m² g^-1 [Supplementary Figure 15]. Furthermore, the surface elements and functional groups of OCNT_5-0.1-0.5 were analyzed using Fourier transform infrared spectroscopy (FT-IR) and X-ray photoelectron spectroscopy (XPS). As illustrated in Figure 3F, the peaks near 1,723, 1,582, and 1,383 cm^-1 are attributed to C=O, C=C-C (aryl group), and C-O, respectively. Figure 3G reveals that carbon and oxygen are detected on the OCNT_5-0.1-0.5 surface. The curve-fitting of the high-resolution C 1s XPS spectrum reveals four peaks at 284.7, 285.6, 288.5, and 290.5 eV, corresponding to C-C/C=C, C-O, C=O, and π-π* transitions, respectively [Figure 3H]. The high-resolution O 1s XPS spectrum is decomposed into two peaks located at 531.8 and 533.2 eV [Figure 3I], attributable to C=O and C-O, respectively.

A variety of methods have been reported for converting waste polyolefins into CNTs, for instance, chemical vapor deposition^[47], Joule flash heating method^[11], microwave-assisted carbonization^[9,12,13], and combined catalyst carbonization^[14]. Compared with these methods, the “all-in-one” catalytic strategy in this work possesses the following advantages. Firstly, the catalytic performance of Ni is synergistically enhanced by the addition of Mo and Mg, thereby enabling the controlled growth of CNTs from a single catalyst without additional degradation catalysts. Secondly, CNTs are directly prepared in a few minutes under ambient air without any protective gas. Finally, the catalytic method shows potential for large-scale CNT preparation.

The OCNT_5-0.1-0.5 evaporator is constructed by coating the dispersion of OCNT_5-0.1-0.5 and gelatin on cotton cloth, followed by crosslinking with glutaraldehyde [Figure 4A]. OCNT_5-0.1-0.5 is evenly distributed on the surface of the evaporator [Figure 4B and C, Supplementary Figure 16]. Similarly, OCNT_5-0.1-0 and OCNT_5-0-0.5 evaporators are prepared [Supplementary Figures 17 and 18]. The optical absorption performance of the evaporators was characterized by a UV-Vis-NIR absorption spectrum. The OCNT_5-0.1-0.5 evaporator exhibits an average light absorption of 98% within a broadband spectral range of 300-2500 nm [Figure 4D], greater than that of the cotton evaporator (ca. 48%). The absorptions of OCNT_5-0-0.5 and OCNT_5-0.1-0 evaporators reached 97% and 95%, respectively [Supplementary Figure 19A]. Due to high light absorption, the surface temperature of the OCNT_5-0.1-0.5 evaporator increased rapidly to 94.2 °C after irradiation of 1 Sun for 2 min [Figure 4E and F], reflecting dramatic photothermal conversion ability. Under 1 Sun illumination for 15 min, the surface temperature of the cotton evaporator reaches only 44 °C. For OCNT_5-0.1-0 and OCNT_5-0-0.5 evaporators, the temperature was stable at ca. 94 and 96 °C under irradiation of 1 Sun, respectively [Supplementary Figures 19B and 20]. The hydrophilicity of evaporators is evaluated by water contact angle. It takes 84 s for a water droplet to completely wet the cotton evaporator. The contact angle of the OCNT_5-0.1-0.5 evaporator decreases rapidly to 0° within 0.03 s [Supplementary Video 1], while OCNT_5-0.1-0 and OCNT_5-0-0.5 evaporators require 3 and 0.03 s, respectively [Figure 4G, Supplementary Figure 21]. The hydrophilic property is inseparable from rich oxygen-containing functional groups and large amounts of pore channels, which improve water transport capacity during evaporation.

A self-built evaporation device was established for solar interfacial steam generation, as shown in Figure 5A and Supplementary Figure 22. A polystyrene (PS) foam serves as a supporting and thermal isolation material to prevent heat loss and transfer. Based on the good photothermal conversion ability and hydrophilicity of the OCNT_5-0.1-0.5 evaporator, water rapidly transfers to its surface to complete evaporation (from liquid to gas). For all evaporation systems, water mass loss increases linearly with irradiation time [Figure 5B]. The evaporation rate of the OCNT_5-0.1-0.5 evaporator is 2.73 kg m^-2 h^-1 [Figure 5C], higher than that of the OCNT_5-0.1-0 evaporator (2.39 kg m^-2 h^-1), OCNT_5-0-0.5 evaporator (2.48 kg m^-2 h^-1), or cotton evaporator (0.88 kg m^-2 h^-1). The OCNT_5-0.1-0.5 evaporator shows a high evaporation rate, attributed to high light absorption & photothermal conversion ability, rich pore channels & functional groups, and good thermal localization. Solar intensity is an important factor affecting the evaporation rate of evaporators. The water mass of the OCNT_5-0.1-0.5 evaporation system decreases linearly with irradiation time under different light intensities [Figure 5D]. As the light intensity is enhanced from 0.5 to 3.0 Sun, the evaporation rate increases from 1.51 to 6.99 kg m^-2 h^-1 [Figure 5E]. Over 10 cycles of water evaporation, the OCNT_5-0.1-0.5 evaporator exhibits an average evaporation rate of 2.71 kg m^-2 h^-1 [Figure 5F]. Besides, the photothermal conversion efficiency of the OCNT_5-0.1-0.5 evaporator is calculated to be 95.4%, which is superior to that of many previously reported evaporators [Figure 5G, Supplementary Table 3].

The evaporation performance is closely tied to the latent heat of water evaporation. Under dark conditions, the water mass losses are 426 mg for OCNT_5-0.1-0.5, 360 mg for OCNT_5-0.1-0, 375 mg for OCNT_5-0-0.5, and 306 mg for cotton, higher than that of pure water (295 mg) [Figure 5H]. Correspondingly, the evaporation enthalpies were calculated as 1.686, 1.995, 1.915, and 2.347 kJ g^-1 [Supplementary Note 3], which are similar to those of the DSC measurement [Supplementary Figure 23 and Supplementary Table 4]. The lowest evaporation enthalpy of the OCNT_5-0.1-0.5 evaporator is the fundamental reason for its higher evaporation rate than that of other evaporators^[48]. Compared with pure water, the evaporation enthalpy is reduced by 30.7% for the OCNT_5-0.1-0.5 evaporator. In addition, the thermal conductivity of the OCNT_5-0.1-0.5 evaporator is measured as 0.0752 W m^-1 K^-1. The heat distribution of the evaporation system is analyzed by finite element simulation [Supplementary Note 4]. The OCNT_5-0.1-0.5 evaporator exhibits a higher surface temperature than cotton due to the good photothermal conversion ability of OCNT_5-0.1-0.5 [Figure 5I, Supplementary Figure 24]. The localization of heat on the surface of the evaporator promotes the liquid water-to-vapor conversion. The heat loss of the OCNT_5-0.1-0.5 evaporator during water evaporation is calculated as 15.7% [Supplementary Note 5].

The water-evaporation power-generation device is shown in Figure 6A. The evaporator was cut into rectangles (10 × 1.5 cm²) to facilitate the generation of electrical energy [Supplementary Figures 25 and 26]. The electrodes at the ends of the evaporators were encapsulated with epoxy resin to prevent short circuits or oxidation, and the device was placed at 45°. The numerous oxygenated chemical groups and abundant pore channels of OCNT_5-0.1-0.5 facilitate water transport. Under the synergistic action of capillary forces and hydrophilic properties, water is continuously transported from the bottom of the evaporator to the surface. The abundant oxygen-containing functional groups of OCNT_5-0.1-0.5 dissociate and generate surface charge upon contact with water molecules. Under the electric double-layer effect at the solid-liquid interface, the counterions selectively migrate in the nanochannels of the evaporator. The continuous enrichment of counterions at the evaporation end leads to the formation of a potential difference [Supplementary Figure 27]. When the external circuit is connected, the potential difference can drive current, producing power output. OCNT_5-0.1-0.5, OCNT_5-0.1-0, and OCNT_5-0-0.5 devices yield the maximum open-circuit voltage of 0.221, 0.130, and 0.082 V, respectively, all exceeding that of cotton [Figure 6B]. Correspondingly, the generated currents were 0.21, 0.15, and 0.11 μA, respectively [Figure 6C]. Specifically, the generated voltage was relatively stable between 0.21 and 0.18 V over 10,000 s, while the current decreased from approximately 220 to 100 nA due to the gradual variation in the evaporation-driven ion transport process [Figure 6D]. The power-generation performance of OCNT_5-0.1-0.5 devices is better than that of OCNT_5-0.1-0 and OCNT_5-0-0.5 devices, due to abundant pores, high-quality CNT structure, and abundant functional groups. The Zeta potential generated by OCNT_5-0.1-0.5, OCNT_5-0.1-0, and OCNT_5-0-0.5 in aqueous solution is -9.3, -1.16, and -6.69 mV, respectively [Figure 6E]. Compared with many reported power generation devices, the OCNT_5-0.1-0.5 device exhibits better steam and electricity co-generation performance [Figure 6F, Supplementary Table 5].

Streaming potential is commonly accepted as the primary mechanism underlying electricity generation in hydrovoltaic processes^[49-52]. During spontaneous water evaporation, the aqueous electrolyte solution migrates from the bottom to the top and flows over the surface of the charged material. The counterions from the electrolyte solution are attracted by the charged ions in the nanopores of OCNT_5-0.1-0.5, resulting in the formation of an electric double layer. The potential difference between the upper and lower ends is gradually formed due to the continuous directional migration and accumulation of free charge. Actually, hydrovoltaic power generation is the process of converting environmental latent heat into electrical energy, which depends on the interaction between ions and OCNT_5-0.1-0.5 materials. Ion selectivity and charge separation in the nanochannel are usually governed by the ion-functional-group interactions, which are further investigated using MD simulation [Supplementary Figure 28 and Supplementary Note 6]. H₃O⁺ and OH^-, produced by the dissociation of water molecules, were added to the simulation system [Figure 6G]. To simplify the simulation, H⁺ was placed in the simulation system instead of H₃O⁺ in the real-world environment. The MD simulation was performed in an NPT ensemble (constant number of particles (N), pressure (P), and temperature (T)) with a simulation time of 20 ns. The relationship between the ionic charge and the surface functional groups of materials in the simulation system was analyzed using the radial distribution function (RDF). As shown in Figure 6H, the first peaks of radial distribution between H⁺ and oxygen-containing functional groups in OCNT_5-0.1-0.5 molecules appear at 0.25 nm (C-OH) and 0.238 nm (COOH), respectively. By comparison, the first peak of C-OH-H⁺ is stronger than that of COOH-H⁺. However, there is no obvious peak in the radial distribution between OH^- and oxygen-containing functional groups, indicating the absence of coordination. The diffusion coefficients of H⁺ and OH^- are 1.06 × 10^-6 and 9.05 × 10^-7 cm² s^-1, respectively [Figure 6I]. In the pure water system, the diffusion coefficients are 3.14 × 10^-7 and 3.36 × 10^-7 cm² s^-1, respectively. The faster diffusion of H⁺ after the addition of OCNT_5-0.1-0.5 confirms the strong attraction between H⁺ and OCNT_5-0.1-0.5. During water evaporation, H₃O⁺ selectively migrates in the nanochannels of the evaporator and accumulates at the evaporation section, resulting in the formation of a potential difference.

The practical freshwater-power co-generation experiment was carried out outdoors [Supplementary Figures 29 and 30]. The self-built outdoor device [Figure 7A] mainly included an evaporator, a container of bulk water, an evaporation chamber, a vapor condenser, and a freshwater collector. With continuous evaporation and bulk water consumption, water in the container can be supplied in time through inlets. The OCNT_5-0.1-0.5 evaporator (20 × 10 cm²) was connected by two wires and placed in the evaporation chamber. As the sun shines, liquid water is constantly evaporated and collected, while generating electricity [Figure 7B-E]. As the vapor condensed, the droplets slid under gravity and were collected at the outlet. The maximum temperature and radiation intensity are 42.3 ^oC and 0.64 kW m^-2, respectively [Figure 7F]. The maximum water collection rate is 0.66 kg m^-2 h^-1 [Figure 7G]. The water production rate is lower than that in the indoor experiment, mainly attributed to lower solar intensity (< 0.8 kW m^-2) and saturated vapor pressure induced by the closed system. The cumulative water production reaches 2.51 kg m^-2 during a 5-h outdoor test. The device yields an open-circuit voltage of 0.215-0.265 V [Figure 7H] and a short-circuit current of 1.8-2.0 μA [Figure 7I]. To further demonstrate the potential utilization of the generated energy, multiple power-generation devices are connected in series/parallel configurations to generate electricity that is subsequently collected through an external circuit and stored in capacitors [Supplementary Figure 31]. Finally, low-power electrical appliances, such as an LED light bulb, fan, and display screen, are powered by connecting capacitors in series [Figure 7J, Supplementary Video 2]. The above results indicate that the OCNT_5-0.1-0.5 device possesses practical application value for co-generating freshwater and power.