Anion‐Selective Layered Double Hydroxide Composites‐Based Osmotic Energy Conversion for Real‐Time Nutrient Solution Detection

Abstract Nanofluidic channels based on 2D nanomaterials are promising to harvest osmotic energy for their high ion selectivity and osmotic conductivity. However, anion‐selective nanofluidic channels are rare and chemical modification is necessary through fabrication. Here, a naturally anion‐selective composite membrane is reported, that is, NiAl‐Layered double hydroxide (LDH) coated anodic aluminum oxide (LDH@AAO), using a simple precipitant‐free in situ growth technique. Positively charged LDH plates growing in channels of AAO function as screening layers for anions. Both experiments and theoretical simulations are enforced to certify the vital role of LDH growth in ion distribution and salinity gradient energy conversion. The composite membrane achieves high output performance and long‐term stability. Furthermore, novel applications of nanofluidic channels are explored in hydroponic production and design a real‐time detecting system based on LDH@AAO composite membranes for nutrient solution. This work provides insights into naturally anion‐selective nanofluidic channels for osmotic energy harvesting and broadens the application in agricultural information sensing.


Introduction
As a type of blue energy, salinity gradient energy exists between two solutions with different salinities, which is common in seawater and river water. Considering an increasing number of miniaturized sensors in daily life demand sustainable and environmental-friendly power, exploiting salinity gradient energy is of great value. Pressure-retarded osmosis and reversed electrodialysis are two main traditional approaches to harvesting salinity gradient energy. However, the development of these two technologies is limited due to undesirable power density, high DOI: 10.1002/advs.202103696 energy costs, and intrinsic trade-off between power output and energy harvesting efficiency. [1] This limitation mainly results from the lack of appropriate membranes able to allow for a sufficient flow. [2] Inspired by biological ion channels embedded in lipid bilayers in living organisms, the occurrence of nanofluidic channels has large potential in realizing transmembrane ion flow efficiently. [3] Existing nanofluidic channels applied to osmotic energy harvesting consist of singlepore-and membrane-based channels. Although single-pore-based nanofluidic channels exhibit ultrahigh output power density (10 3 -10 6 W m −2 ), it is challenging to translate the high power density into real high power with porous membranes. [4] To gain the promoted output performance, high surface charge density, large surface area, and nanoscale channel sizes are necessary for membrane-based nanofluidic channels. [5] Many two dimensional (2D) materials including graphene, hexagonal boron nitride, MXene, graphene oxide, and molybdenum disulfide have been exploited in membrane-based nanofluidic channels for osmotic energy harvesting recently in view of their superiority in electrical and mechanical properties. [6][7][8][9][10][11][12][13][14][15] However, almost all of them are naturally cation-selective membranes due to hydroxide adsorption of most 2D materials. Considering anion-selective membranes and cation-selective membranes are of the same value as necessary components for reverse electrodialysis (RED), advances in anion-selective membranes are still required for large-scale application of RED in future. To obtain anion-selective membranes, surface modification with chemicals is a common choice, which adds complexity during the fabrication process. [12,16,17] Layered double hydroxides (LDHs) have attracted widespread attention in optical sensors, water splitting, supercapacitors, electrocatalytic oxygen evolution, lithium batteries, and biomedical applications recently. [18][19][20][21][22] Positively charged layers of LDHs equip them with the capability to act as functional layers in anion-selective nanofluidic channels.
Here, we report an anion-selective composite membrane, that is, NiAl-LDH coated anodic aluminum oxide (LDH@AAO), for harvesting salinity gradient energy (Figure 1a). On one hand, NiAl-LDH growing on surfaces of AAO provides sufficient positive charge density to screen cations. The distribution of NiAl-LDH in the inner surfaces of AAO also tunes the nanochannel LDH plates are distributed on the outer and inner surfaces of the AAO channels. The salinity gradient drives both anions and cations flow from a higher to a lower concentration. However, the positive surface charge brought by LDH contributes to a higher population of anions (blue balls) in nanofluidic channels and repels the entrance of cations (orange balls). The net movement of anions in the composite membrane results in the observation of a current. b,c) Cross-sectional SEM images of the AAO template and the LDH@AAO composite membrane (scale bar: 500 nm). Inset displays the outer surface (scale bar: 500 nm). d) XRD patterns of the AAO template and the LDH@AAO composite membrane. e) Nitrogen adsorption/desorption isotherms and f) corresponding pore size distribution of the LDH@AAO composite membrane. g) XPS spectrum of the AAO template and the LDH@AAO composite membrane. h,i) EDX mapping of Na and Cl on the cross-section of the LDH@AAO composite membrane (scale bar: 1 μm).
size to an appropriate range so that an excellent ion selectivity could be achieved. On the other hand, AAO participates in the formation of NiAl-LDH in Ni(NO 3 ) 2 6H 2 O aqueous solutions. It is also the backbone of the LDH@AAO composites. When mounted between artificial seawater (0.5 m NaCl) and artificial river water (0.01 m NaCl), the fabricated membrane can harvest salinity gradient energy up to 2.85 W m −2 .
Importantly, we first developed the application of nanofluidic channels in real-time detecting ion concentration of nutrient solutions in hydroponic production. Hydroponic production has been a popular choice in agriculture. [23] Crops are grown in nutrient solutions without soil in this way. Nutrient solutions typically consist of multiple fertilizer ions and provide crops with the necessary nutrients. It is vital to monitor the ion concentration in nutrient solutions because the yield will be limited when the concentration is too low or too high. [24] Electrical conductivity (EC) is commonly used to reflect the total amount of ions in nutrient solutions. However, long-term immersion in nutrient solution tends to pollute the electrode for conductivity measurement, which burdens users with the demand for regular calibration and replacement of electrodes. Also, high costs arise from high density of sensors per unit of area for real-time monitoring. [25] The capability of nanofluidic channels in converting salinity gradient energy into electricity makes it possible to employ electrical signals generated from nanofluidic channels as a measurement of ion concentration levels in nutrient solutions without an external power supply. We demonstrated a real-time detecting system based on LDH@AAO composite membranes for the nutrient solution. The measured current is positively correlated to the ion concentration of nutrient solutions. When the measured current is beyond the appropriate range for the growth of crops, the system will warn users to adjust the ion concentration in time.

Characterization of LDH@AAO Composite Membranes
The LDH@AAO composite membranes are fabricated in the method of simple and precipitant-free in situ growth. [26] In order to gain the appropriate channel size of AAO, we employed AAO templates with different channel sizes and fabricated composite membranes ( Figure S1, Supporting Information). When the channel size of AAO is smaller than 160 nm, the diffusion of Ni 2+ into the AAO pores is hindered, which is the reason for the failure of nucleation of NiAl-LDH and the formation of boehmite in channels. The channel size of AAO was then controlled in the range of 160 to 200 nm. Figure S2, Supporting Information, gives the photograph of the LDH@AAO composites, which is 1.2 cm in diameter and appears light green. Scanning electron microscopy (SEM) reveals the morphology and structure of the AAO template and NiAl-LDH@AAO composite membranes (Figure 1b,c). The outer and inner surfaces of the AAO template are both distributed with LDH plates. Interlaced LDH plates shield pores on AAO templates. Large and dense LDH plates are parallel to each other along the channels in AAO. The pH of the reaction solution and reaction time exert great influence on the nucleation and growth of LDH ( Figures S3 and S4, Supporting Information). The thickness of AAO is 43.3 μm while the thickness of the composite membrane is 45.8 μm ( Figure S5a,b, Supporting Information). The distribution of LDH layers on the top and bottom surface of AAO is nearly symmetrical ( Figure S5c,d, Supporting Information). The structure of AAO templates and LDH@AAO composite membranes were certified through X-ray photoelectron spectroscopy (XPS) and X-ray diffraction (XRD) (Figure 1d,g and Figure S6, Supporting Information). As for O 1s of AAO templates, two peaks at the binding energy (BE) of 531.0 and 532.0 eV corresponded to the characteristic of O 2− and OH − ions respectively, which is consistent with the existence of lattice oxygen and hydroxyl ions in the composition of AAO templates. In contrast, the peak at BE of 532.1 eV is attributed to the surface OH − ions of LDHs. The results are in agreement with previous research. [27,28] The diffraction peaks of LDH@AAO near 10°and 20°shown in the XRD pattern are assigned to the (003) and (006) planes of LDHs ( Figure 1d). The nitrogen adsorption/desorption isotherms of the LDH@AAO composites in Figure 1e which is a typical type IV curve indicate that the composite membrane has a large uniform mesoporous structure. As shown in Figure 1f, the mean size of nanochannels in composite membranes is calculated to be 17.8 nm using the Barrett-Joyner-Halenda (BJH) model. After immersing the NiAl-LDH@AAO composite membrane in 0.5 m NaCl for 24 h and removing unbound ions with deionized water, we conducted energy-dispersive X-ray spectroscopy (EDX) mapping on the cross-section of the composite membrane (Figure 1h,i). The result demonstrates the selectivity of anions induced by positive surface charges of LDH plates. The composite membrane also exhibits great hydrophilicity as the contact angle is almost 41°( Figure S7, Supporting Information).

Evaluation of Surface-Charge-Governed Ion Transport
We investigated the ion's transport properties of the LDH@AAO composites with an electrochemical cell ( Figure S8, Supporting Information). The ion conductivity of the composite membrane was measured as a function of NaCl concentration (Figure 2a). When the concentration is higher than 1 m, the conductivity of the composite membrane is mainly dependent on NaCl concentration, which is similar to the conductivity of bulk NaCl solution (dashed gray line). The further decrease of NaCl concentration increases the thickness of electric double layer, which is close to the width of the nanochannel. [29,30] As a result, the deviation of the ionic conductivity from the bulk value at low concentrations certifies the surface charge-governed ion transport property of the LDH@AAO composites. [7,31] Figure 2b presents current-voltage (I-V) curves for a series of NaCl concentrations. The symmetric structure of the composite membrane contributes to the linear ohmic behavior without ionic current rectification. We further explored the ionic transport behaviors under various concentration gradients. One side of NaCl solution was fixed at 10 −6 m, while the other side ranged from 10 −5 to 5 m. A pair of salt bridges was used to eliminate the imbalanced electrode potential. [12,40] The intercepts on the current and voltage axes of I-V curves are corresponding to the diffusion current (I diff ) and potential (E diff ) respectively. When the concentration gradient increases, the E diff and I diff both increase. The calculated energy conversion efficiency ( ) sharply decreases from 20.4% to 3.3% as the concentration gradient increases from 10 to 5 × 10 6 (Figure 2d, Note S1, Supporting Information).

Osmotic Energy Conversion of the Composite Membrane
To investigate the osmotic energy conversion capability of the LDH@AAO composites, we employed artificial seawater and river water to provide the salinity gradient. As shown in Figure 3a, we connected an external load resistance (R L ) to output the harvested energy. Figure 3b recorded the measured current which decreased when the external resistance increased. The electric power density, calculated as P R = I 2 × R L , reached a maximum of 2.85 W m −2 when the load resistance was about 23 kΩ. The I-V curve for a 50-fold salinity gradient was illustrated in Figure 3c. The intercepts on the current and voltage axes are corresponding to I diff and E diff which are 2.46 μA and 55.7 mV respectively (the imbalanced electrode potential has been eliminated through the use of a pair of salt bridge). The corresponding energy conversion efficiency of the composite membrane calculated from these parameters is 17.6%. As stated before, the pH of the reaction solution and reaction time during the fabrication of composite membranes exert great influence on the nucleation and growth of LDH. These two parameters further affect the osmotic energy conversion behavior of the composite membrane since LDH plates mainly function as screening layers in nanochannels.
To investigate the influence of pH, we fixed the reaction time at 12 h and recorded the corresponding power density for different pH in Figure 3d. The power density increases from 0.09 to 2.55 W m −2 when the pH increases from 3.5 to 5.5. The relatively high pH of the reaction solution is favorable for LDH growth in nanochannels, contributing to the increased surface charge den- sity and strengthened anion selectivity. The continuous increase of pH leads to dense distribution of NiAl-LDH on the surface of AAO, thus hindering the diffusion of Ni 2+ into nanochannels and preventing the growth of LDH plates along nanochannels. The large steric hindrance to the transmembrane ion transport and relatively low charge density decreases the power density. This mechanism is also adaptable for the effect of the reaction time. We fixed the pH at 5.5 and obtained a peak value of power density as 2.64 W m −2 when the reaction time was 16 h (Figure 3e). Figure 3f compares the osmotic energy conversion capability of our LDH@AAO composite membrane with existing nanofluidic membranes. These membranes are classified into three categories according to their ion selectivity. The development of cation-selective nanofluidic membranes is better than anion-selective membranes. Advances in anion-selective membranes are required for RED. As a novel anion-selective nanofluidic membrane, the LDH@AAO membrane possesses relatively high energy conversion efficiency and output power density. The simple precipitant-free in situ growth technique without chemical modification also makes the LDH@AAO membrane more appropriate for large-scale application in the future.

The Effect of LDH Growth on the Ion Concentration Distribution and Energy Conversion
In order to give further insight into the mechanism of power generation under the growth of LDH on AAO, numerical simulations were carried out based on Poisson and Nernst Planck (PNP) equations (Note S2, Supporting Information). Considering the growth of LDH in nanofluidic channels narrows the channel size (d), improves the surface charge density ( ), and extends the channel length (l) (Figure 4a), we then focused on the influence of these three parameters. The nanofluidic channel was simplified into a 2D model as shown in Figure S9, Supporting Information, to simulate the ion transport. Figure 4b displays the concentration distribution of the cation and anion in the nanofluidic channel (l = 500 nm, = 0.08 C m −2 ) with various channel sizes. For the channel with a smaller size, the selectivity of anions is better (Figure 4c-e). When d is 25 nm, the perm-selectivity of nanofluidic channels almost disappears since the concentration curves of anions and cations overlap. The reason behind this phenomenon is that the range of overlapped electrical double layer (EDL) is large for narrow channels. The range is defined as Debye length: where is the permittivity of water 0 is the permittivity of a vacuum, R is the universal gas constant, n bulk is the concentration of electrolyte solution, T is the absolute temperature, z is the valence number, and F is the Faraday's number. [7,38] When the range of EDL covers the physical dimensions of the channel, good selectivity is achieved. [1,12] High surface charge density also favors the improvement of ion selectivity ( Figure S10a, Supporting Information). The output performance of channels with small These membranes are classified into three categories including negatively charged membranes, positively charged membranes, and oppositely charged membranes. [6,7,12,16,29,[31][32][33][34][35][36][37] channel size and high surface charge density is outstanding correspondingly ( Figure S10b, Supporting Information). However, long channel length seems to weaken the energy conversion capability especially for narrow channels. As shown in Figure 4f, the output power of narrow channels (d = 5 nm) deteriorates with the increase of channel length. In theory, the maximal power is calculated as: where G osm and osm are the osmotic conductance and the osmotic potential, respectively. For LDH@AAO composite membranes, the bulk conductance dominates in osmotic conductance: [1] G osm ≈ G bulk = d 2 4l (3) where G bulk and are the bulk conductance and the bulk conductivity, respectively. So a long channel length will limit the output power because of the low osmotic conductance. Besides, when the channel size is smaller, the decline of G osm is larger with the increase of the channel length, which is the reason for the distinct deterioration of channels (d = 5 nm). It is worthy to mention that the output increases when the channel length is 550 nm for an-other two pore sizes. This may be attributed from the trade-off between ion selectivity and osmotic conductance. As a result, the growth of LDH is a two-edged sword for osmotic energy conversion and needs to be controlled at an appropriate degree.

The Effect of Environmental Factors
To explore the capability of composite membranes for practical application, we investigated the effect of environmental factors including the pH of electrolyte solutions, species of electrolyte, and working time. As the pH increases from 3 to 11, the power density decreases from 2.83 to 1.79 W m −2 (Figure 5a). The variation results from the influence of pH value on the space charge and surface charge created by NiAl-LDH nanoplates and AAO nanochannels. [7] The measured zeta potential of NiAl-LDH increased when the pH decreased (Figure 5b). The value of pH also exerts great influence on the amphoteric -OH groups on AAO channels. Thus, the positive surface charge density of these anion-selective composite membranes is larger in acidic environment, contributing to better performance. We also compared the output power density of the composite membranes in various sodium salts to examine the selectivity of various anions. As shown in Figure 5c, the composite membranes exhibit the  best output performance in NaI solutions (4.14 W m −2 ). We attribute this superiority to the large diffusion coefficient of I − (2.045 × 10 −9 m 2 s −1 ). Since the composite membranes are anionselective, the fast diffusion of anion contributes to the efficient charge separation. [39] Figure 5d recorded the change of the current density measured on the external circuit in a week without electrolyte replenishing. The slight attenuation (6.4%) demonstrates the stability of composite membranes in osmotic energy conversion. The ionic bonding between AAO and NiAl-LDH contributes to the durability of the composite membranes.

Application of the Composite Membrane in Real-Time Detection of Ion Concentration in Nutrient Solutions for Hydroponic Production
Considering the osmotic energy arises from variation in ion concentration, we explored the application of the LDH@AAO composite membrane in detecting ion concentration of nutrient solutions. A real-time nutrient solution detecting system based on LDH@AAO composite membranes was designed (Figure 6a).
The composite membrane was fixed on the side of a cuboid box using epoxy ( Figure S11, Supporting Information). The box acted as containers for nutrient solutions and was surrounded by water (Figures S12 and S13b, Supporting Information). The salinity gradient energy between nutrient solutions and water was converted into electricity. Two Ag/AgCl electrodes were dis-tributed on two sides of the box and connected to an electrometer data acquisition (DAQ) board which could acquire and process the signals from nanofluidic channels ( Figure S13a, Supporting Information). Figure 6c demonstrates the information interaction process. The electrometer DAQ board sends the current signal to a customized application (APP) on the smartphone via Bluetooth. The composite membrane was fixed between nutrient solutions with different concentrations and ionized water. Hoagland's nutrient solution is used in nutrient detection, which has been widely applied in hydroponic production. The concentration of standard nutrient solution is set as 1 strength (S). can estimate the concentration level according to the appropriate range which was set before and warn users to adjust the concentration of nutrient solutions in time when the level is low or high (Video S1, Supporting Information). For real application, a long-term monitoring of actual hydroponic production was car- ried out. Seedlings of romaine lettuce were hydroponically cultivated during a week ( Figure S15, Supporting Information). The I sc was measured per 12 h during 7 d and the cuboid box was immersed in the tank all the time. To ensure the overall volume of nutrient solution to be the same during the test, a certain volume of water was added to the nutrient solution daily. EC was also measured and used as a reference for the real concentration level of nutrient solution. Figure 6d exhibits the variation of the measured I sc and EC with the growth of lettuce, comparing the difference between measured nutrient concentration and the actual concentration. It is evident that the variation of measured I sc can reflect the level and changing trend of nutrient solution concentration. However, after immersion in nutrient solutions for 7 d, the composite membrane was partly fouled on the outer surface ( Figure S16, Supporting Information). It turns out that the output performance of composite membrane undergoes a deterioration (12.8%) after comparing I sc in standard nutrient solution (1 S) with the value obtained at 0 d. The antifouling property of LDH@AAO composite membrane remains to be improved for future application.

Conclusion
In conclusion, we adopted a precipitant-free and simple in situ growth method without chemical modification to fabricate LDH@AAO composite membranes for osmotic energy harvesting. Positive surface charges of LDH layers endow the composite membrane with excellent selectivity of anions. Both experiments and theoretical simulations certify the vital role of LDH growth in ion distribution and osmotic energy conversion. The fabricated membrane can achieve a power density of 2.85 W m −2 by mixing artificial seawater and river water. The output performance is distinctive among anion-selective nanofluidic membranes. Intriguingly, we designed a real-time detecting system based on LDH@AAO composite membranes for the nutrient solution. Current signals generated from variation in ion concentration of nutrient solutions were employed to measure concentration levels and warn users to adjust nutrient solutions in time. Our work provides insights into naturally anion-selective membranes for osmotic energy harvesting and explores novel applications of nanofluidic channels in hydroponic production. It can be envisioned that more precise detection of specific ion concentration is one of the promising directions for the application of osmotic energy harvesters.

Experimental Section
Chemicals: The AAO membranes were purchased from Hefei Pu-Yuan Nano Technology, Ltd., China. Hoagland's nutrient solution was purchased from Shenzhen Konong Horticulture Technology, Ltd., China. Other chemicals were all analytical grade and obtained from Sinopharm Chemical Reagent Co. Ltd., China. All solutions were prepared using Milli-Q water (18.2 MΩ cm).
Fabrication of LDH@AAO Composite Membranes: The composite membrane was fabricated based on an in situ growth technique. AAO plates were vertically fixed in beakers containing Ni(NO 3 ) 2 6H 2 O aqueous solutions (1 m). These beakers were then heated in water bath. The temperature was controlled at 80°C. The pH of Ni(NO 3 ) 2 6H 2 O aqueous solutions was adjusted by nitric acid or ammonia solution. The obtained composite membrane was dried at room temperature for 24 h after the rinse with deionized water.
Characterization: SEM images were captured using SEM (Gemini 300, ZEISS, Oberkochen, Germany). The XPS was performed in VG ESCALAB MARK II. XRD was performed in a diffractometer (D8 Advance, Bruker, Karlsruhe, Germany). Nitrogen sorption isotherms were obtained using TriStar II Plus (Micromeritics, Norcross, GA, USA). The pore distribution was acquired from the desorption branch based on the BJH model. Contact angles were measured in Theta Lite (Biolin, Stockholm, Sweden). To obtain the element distribution on the cross-section of LDH@AAO composite membrane, EDX spectroscopy was conducted. Zeta potential of NiAl-LDH was analyzed in a solution system using Zetasizer Nano ZS90 (Malvern, Worcestershire, UK).
Electrical Measurements: The composite membrane was fixed between a customized electrochemical cell. The ion transport properties and energy conversion capabilities were acquired with a Keithley 6487 semiconductor picoammeter (Keithley, Cleveland, OH, USA). A pair of Ag/AgCl electrodes were employed to apply the transmembrane electrical potential and measure the resistance and current. During the detection of I-V curves under various concentration gradients, the range of the sweeping voltage was −0.2 to 0.2 V, with a step voltage of 0.02 V. A pair of agar-saturated potassium chloride salt bridge was used to eliminate the imbalanced electrode potential. The effective working area of the membrane was set as 3 × 10 4 μm 2 . The conductance value of bulk NaCl solution was obtained by measuring the conductance of a neutral charged silicon membrane in the customized electrochemical cell. The pH of electrolyte solutions was adjusted by HCl and NaOH solutions. During the stability test, the composite membrane was fixed in the electrochemical cell and the electrolyte solutions were not refreshed.
Statistical Analysis: No filtering was applied in the experimental results presented in this work. For measuring conductance versus salt concentration, experimental results presented used one composite membrane (sample size n = 1). For measuring current-voltage curve, experimental results presented used n = 1. For measuring E diff and I diff under various concentration gradients, three samples were tested (n = 3). For measuring the current density as the function of load resistance, one composite membrane was used (n = 1). For measuring the power density corresponding to different growth pH and reaction times, the data were presented as the mean ±SD of three independent experiments (n = 3). For measuring power density under various environmental conditions, three samples were tested (n = 3). For measuring the stability of the composite membrane, one sample was used (n = 1). For nutrient solution detection, one cuboid box with LDH@AAO composite membrane was employed (n = 1).

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.