Continuous and complete conversion of high concentration
 p
 ‐nitrophenol in a flow‐through membrane reactor

Here, we report on a green and effective method for the continuous and complete conversion of high concentrations of p‐nitrophenol (PNP) using a flow‐through membrane reactor and less NaBH4. The catalytic membrane was successfully fabricated by loading Pd nanoparticles onto the surface of a branched TiO2 nanorod‐functionalized ceramic membrane. The modification with branched TiO2 nanorods can significantly improve the loading amount of Pd nanoparticles onto ceramic membranes, resulting in enhanced catalytic performance. With 6 mg of Pd, 93 L m−2 hr−1 of flux density and 8.04 cm2 of membrane surface area in the flow‐through membrane reactor, PNP at a concentration of 4,000 ppm can be converted to high‐value p‐aminophenol using less NaBH4 (using a molar ratio of NaBH4:PNP of 9.6) within 24 s at 30°C. More importantly, the conversion can be continuously and stably performed for 240 min.


Introduction
p-Nitrophenol (PNP) is a common environmental pollutant in water and is therefore a great public concern. 1 It is usually used as a precursor or a synthetic intermediate in the industrial manufacturing of analgesics, pharmaceuticals, insecticides, dyes and other chemicals. 2-4 PNP may bring about significant health hazards because of its carcinogenic toxicity. Short-term inhalation or ingestion in humans can cause headaches, drowsiness, nausea and cyanosis, even at low concentrations. 5,6 The concentration of PNP in industrial wastewater is usually much higher than 500 mg/L. 7 Therefore, the complete elimination of this compound from industrial effluents is a matter of concern for environmental protection.
Various methods have been proposed for the treatment of PNP-contaminated wastewater. 8 The conventional physical methods (sedimentation, filtration, adsorption, etc.) transfer the contaminants into other forms and cannot solve the waste disposal problem. 9 Biological methods may require long treatment time, and complete degradation may be impossible, This article is protected by copyright. All rights reserved. especially for effluents containing PNP at a high concentration. [10][11][12] Thus, the development of effective chemical methods to convert PNP to high-value products or to achieve complete degradation is urgently needed.
Catalytic reduction of PNP to p-aminophenol, an important chemical intermediate, is a feasible method for turning waste into a renewable resource. 13,14 In addition, aromatic amines are less toxic and considerably easier to mineralize than their corresponding nitroaromatics. 15,16 Conventional metal-acid reduction methods employ reagents such as iron-acids or tin-acids. The major disadvantage of such reduction processes is the generation of large amounts of metal oxide sludge that is associated with severe pollution problems. 17 One-step hydrogenation of PNP in the presence of metal catalysts or supported metal catalysts is considered to be the most promising process because of its high efficiency and environmentally friendly properties. 18 As a strong reducing agent, NaBH 4 can effectively reduce PNP to p-aminophenol under mild operation conditions (room temperature and atmospheric pressure). 19 Thus, the direct reduction of PNP with NaBH 4 as a reducing agent is considered to be an efficient and greener catalytic route for the conversion of PNP.
In the practical treatment of industrial wastewater, fixed-bed reactors and slurry reactors inevitably develop problems. For a fixed-bed reactor, fine catalysts cannot be directly used, and their inner surface cannot be fully utilized. In addition, the regeneration and replacement of catalysts are inconvenient. In a slurry reactor, metal nanoparticles that are present on the This article is protected by copyright. All rights reserved. powdered catalysts easily aggregate or leak, and it is difficult to separate them from the reaction system. 20 Catalytic membranes exhibit good particle distribution, and no additional separation is required. 21,22 Moreover, the porous membrane structure and flow-through mode can enhance the catalytic efficiency by increasing mass transfer. 23 These advantages allow flow-through catalytic membrane reactors to efficiently treat industrial wastewaters that contain high concentrations of PNP. Wang et al. 24 developed a novel poly (vinylidene fluoride) membrane with Pd/poly (methacrylic acid) microspheres immobilized inside the membrane pores. Its use in the catalytic reduction of PNP indicated that a conversion of 99.8% could be achieved in a cross-flow model. Domènech et al. 25 reported on the synthesis of Pd nanoparticles in sulfonated polyethersulfone membranes. The catalytic performance was evaluated by following the reduction of PNP in the presence of NaBH 4 . Greater than 90% of the PNP was reduced within 4 hours using a single reaction step, and deactivation was observed after consecutive catalytic cycles. Our group 26 successfully demonstrated that the use of TiO 2 nanorod-functionalized ceramic membranes is an effective approach for enhancing the loading amount of Pd and the corresponding catalytic activity. However, the catalytic membranes obtained when used directly in a batch reactor could not achieve continuous conversion of PNP. Furthermore, although complete conversion of PNP can be achieved, a high molar ratio of NaBH 4 to PNP like 100 is required. 27,28 An excess of the reductant NaBH 4 will increase the treatment cost significantly. Therefore, the efficient This article is protected by copyright. All rights reserved.
transformation of PNP remains a great challenge.
The objective of this work was to accomplish the continuous and complete conversion of high concentrations of PNP by using less NaBH 4 . We designed a flow-through catalytic membrane reactor for use in a continuous mode. Furthermore, branched rutile TiO 2 nanorod arrays were synthesized on an alumina ceramic membrane using a simple hydrothermal grafting method. Pd nanoparticles were then immobilized on the surface of the modified ceramic membranes by a sol impregnation method (denoted as Pd/BTiO 2 -CM). For comparison, Pd nanoparticles loaded onto bare ceramic membranes (Pd/CM), and TiO 2 nanorod array-modified ceramic membranes (Pd/TiO 2 -CM) were also fabricated using the same conditions. The crystallinity, morphology and structures of the prepared catalytic membranes were investigated to understand the dependence of catalytic performance on their microstructure.

Synthesis of TiO 2 nanorods and branched TiO 2 nanorods on ceramic membranes
TiO 2 nanorod arrays on sheet Al 2 O 3 ceramic membranes (circular, 3.2 cm in diameter, 1.6 mm in thickness and 30% in porosity) were obtained via the following two-step hydrothermal method. 26 Twenty mL of concentrated hydrochloric acid (Analytical Reagent, This article is protected by copyright. All rights reserved. mass fraction 36.5-38%) were mixed with 20 mL of deionized water. After stirring at 30 °C for 10 min, 955 μL of titanium butoxide (Chemically Pure, 98.0%) were added to the mixture.
The feedstock prepared above was injected into a Teflon-lined stainless steel autoclave in which a piece of ceramic membrane was placed vertically. Hydrothermal synthesis was initiated when the autoclave was placed in an electric oven and maintained at 150°C for 5 hours. In the second hydrothermal synthesis process, similarly to the first step, 3 mL of a saturated NaCl aqueous solution were added to the same mixture as described above and the hydrothermal synthesis was conducted at 150 °C for 20 hours. Chloride ions can accelerate the growth of TiO 2 in the direction of the top (001) crystal facet. 29 Thus, TiO 2 nanorods having a greater length to diameter ratio can be synthesized by the introduction of NaCl.
Branched TiO 2 nanorod arrays were grown by immersing the TiO 2 nanorod-modified ceramic membranes prepared above in an autoclave filled with 40 mL of deionized water, 0.1 mL of hydrochloric acid and 0.8 mL of a titanium chloride solution (Analytical Reagent, mass fraction 15-20%). The autoclave was then sealed and maintained at a constant temperature of 80 °C for 16 hours. Finally, to eliminate the influence of the chloride impurity, the obtained sample was rinsed extensively with deionized water until it was neutral, and dried at 60 °C for 4 hours.
The branched TiO 2 nanorod-modified ceramic membranes were further modified with N-(β-aminoethyl)-γ-aminopropyl trimethoxy silane (AAPTS) by immersing them in 50 mL of This article is protected by copyright. All rights reserved.

Synthesis of catalytic membranes
Three catalytic membranes (Pd/CM, Pd/TiO 2 -CM and Pd/BTiO 2 -CM) were prepared by a sol impregnation method. 30 Typically, 0.56 g of palladium acetate (mass fraction of Pd is approximately 47%) and 5.55 g of polyvinylpyrrolidone (PVP, K30, Guaranteed Reagent) were dissolved in 50 mL of ethanol (Analytical Reagent, 99.7%), and the mixture was then stirred at 60 °C for 2 hours to prepare the Pd nanoparticle colloid. The bare ceramic membranes, the TiO 2 nanorod-modified ceramic membranes and the branched TiO 2 nanorod-modified ceramic membranes were immersed in the Pd nanoparticle colloid (50 mL, 0.05 mol/L) at 40 °C for 12 hours. The samples were washed with ethanol to remove the Pd nanoparticles that were not firmly loaded onto the ceramic membranes and/or TiO 2 nanorods, and then dried at ambient temperature.

Structure and morphology of catalytic membranes
The crystal structure of the catalytic membranes was determined by X-ray diffraction (XRD, Miniflex-600) analysis at a scan rate of 10° per min in the range of 5-70°. The surface morphology of the catalytic membranes was characterized using a field emission scanning electron microscope (FESEM, Hitachi S-4800II). The Brunauer-Emmett-Teller (BET) surface This article is protected by copyright. All rights reserved.
area of the catalytic membranes was measured by N 2 physisorption using a physical adsorption apparatus (Micromeritics, ASAP 2020). The pore size of the catalytic membranes was determined using a mercury intrusion porosimeter (Poremaster GT-60).

Pd nanoparticles in catalytic membranes
The distribution of elemental Pd in the catalytic membranes was determined by energy-dispersive X-ray spectroscopy (EDS, Horiba Emax). Inductively In this study, the prepared catalytic membrane was a circular flat membrane with a diameter of 3.2 cm. For TEM tests, 0.05 g of powders were scraped off the surface of the catalytic membrane and dispersed in 10 mL of ethanol. To obtain reproducible results, the particle size of the Pd was evaluated by counting more than 100 particles and samples were tested in triplicate.

Determination of pure water flux
This article is protected by copyright. All rights reserved.
Pure water permeation by the membranes was investigated using a dead-end permeation apparatus at a transmembrane pressure of 0.2 bar and a temperature of 18 o C. The effective diameter of the membrane in contact with the solution was 2.6 cm. The pure water flux through the catalytic membrane was calculated using the following equation: where F is the pure water flux through the catalytic membrane (L·m -2 ·h -1 ·bar -1 ), Q is the volume of penetration liquid (L), A is the effective membrane area (m 2 ), t is the permeation time (h) and P is the transmembrane pressure (bar).

Evaluation of Catalytic Performance
Reduction of PNP was performed in a flow-through membrane reactor system in two

Structure and morphology of catalytic membranes
As illustrated in Figure  This article is protected by copyright. All rights reserved.
The TiO 2 nanorods are fully covered by short needle-shaped branches which provide greater surface area (Table 1).

Pd nanoparticles in catalytic membranes
EDS mapping was performed on the three catalytic membranes to investigate the distribution of elemental Pd. In Figure 4 highest Pd content, which is nearly three times greater than that found in the Pd/CM, and the trend is in good agreement with the EDS results ( Figure 4). As expected, the total quantity of Pd in the reactor consisting of the Pd/BTiO 2 -CM is the highest (3 mg). Therefore, we can deduce that modification with branched TiO 2 nanorods can effectively increase the surface area, resulting in increased loading amounts of Pd, which is one of the reasons for the improvement of catalytic activity.  Table 1). This article is protected by copyright. All rights reserved.

Determination of pure water flux
L·m -2 ·h -1 ·bar -1 , respectively. The water flux drops sharply after synthesis of TiO 2 nanorods and branched TiO 2 nanorods on the ceramic membranes. Generally, the decrease in the membrane pore size will make the filtration resistance increase, thereby reducing the membrane flux. 35 In this work, the synthesis of TiO 2 nanorods and branched TiO 2 nanorods on the ceramic membranes makes the membrane pore size decrease (Table 1), leading to the reduced water flux. However, the water flux is still larger than the reported values. For example, the pure water flux reported by Mahdavi et al. 36 or Emin et al. 37 were 2-10 L·m -2 ·h -1 ·bar -1 and less than 1600 L·m -2 ·h -1 ·bar -1 , respectively. The larger membrane pore size (approximately 3 µm) is responsible for the high water flux.

Evaluation of catalytic performance in a cycle mode
The PNP reduction process was evaluated in a cycle mode (Figure 6a Figure 6b shows that the initial reaction rate during the first 5 minutes of the reaction using Pd/BTiO 2 -CM is 1.6-fold higher than that using Pd/CM. The initial reaction rate in the system comprising Pd/TiO 2 -CM is 1.1 times higher than that This article is protected by copyright. All rights reserved. The stability of catalytic membranes was evaluated by comparing the conversion of PNP in every recycling experiment. Once the reaction process was completed (60 minutes for each run), the catalytic membrane was removed and immersed in ethanol for 1 hour, then applied to the next reaction. As indicated in Figure 7, after 5 cycles of catalytic reduction of PNP using Pd/BTiO 2 -CM, the conversion efficiency declined slightly but remained above 91.6%.
For Pd/CM, after 5 cycles, the conversion efficiency of PNP was only 74.0%, which is significantly lower than that in the first run (87.7%). These results indicate that the branched TiO 2 nanorods are favorable for enhancing the catalytic stability of Pd nanoparticles. This may be because the TiO 2 branches can play a role in fixing Pd nanoparticles to prevent them from falling off during the recycling experiments ( Figure 5). To assess this assumption, the This article is protected by copyright. All rights reserved. This article is protected by copyright. All rights reserved.
or -NH is a strong electron donor and has a strong chelating ability for transition metals, thus the Pd can react with them to form a chelated complex and be bound to the ceramic membrane. 39  These results indicate the good stability of Pd/BTiO 2 -CM-AAPTS.

Evaluation of catalytic performance in a continuous mode
To investigate the feasibility of continuous conversion of PNP, a continuous operation mode was developed and the prepared Pd/BTiO 2 -CM-AAPTS was evaluated.
To achieve the complete conversion of PNP, the dependence of PNP conversion on the residence time was first investigated. Figure 9 shows the degree of conversion when a 4.176 g/L solution of PNP flowed through the catalytic membrane in a continuous mode. The total amount of Pd in the two catalytic membranes in the reactor was 6 mg. A 99.4% conversion was achieved at 4 seconds of residence time (flux density 560 L·m -2 ·h -1 ). In accordance with the features of conventional flow-through reactors, 40 it is clear that an increasing residence time through the membrane results in an increase in the conversion of PNP, and the degree of conversion can reach 100% at 24 seconds of residence time (flux density 93 L·m -2 ·h -1 ). In This article is protected by copyright. All rights reserved. was not significantly different from the fresh particles and remained at 4.2 ± 0.2 nm ( Figure   12). Based on the ICP analysis, the Pd content in the recovered Pd/BTiO 2 -CM-AAPTS was 0.36 mg/cm 2 , which is within the measurement error relative to the fresh one (0.37 mg/cm 2 ).
These results further confirm the good stability of the flow-through membrane reactor for continuous conversion of PNP.

Insights into the enhancement mechanism of catalytic performance
The concentration of PNP used in our reaction is dozens or hundreds times higher than those reported in the literature, and the molar ratio of NaBH 4 to PNP is much smaller than literature values (Table 2). More importantly, Pd/BTiO 2 -CM-AAPTS can achieve continuous complete conversion of PNP for more than 240 minutes, which has not been reported previously, especially at such a high concentration. In this work, the quantity of Pd used in the reactor is significantly higher than those reported in the literature (Table 2), and could account for the complete conversion of high concentrations of PNP with less consumption of NaBH 4 . However, the NaBH 4 is still in a large excess (Table 2), far from the stoichiometric amounts required. NaBH 4 is easily hydrolyzed during the reaction, therefore excess NaBH 4 is often needed. 41 In future work, we will attempt to further reduce the amount of NaBH 4 , enhance the catalytic efficiency of Pd and the stability of the catalytic membrane, and improve the economic efficiency of the catalytic membrane reactor. Notably, the productivity per g of catalyst in this work is almost the same as that in the work of Emin et al. 28 This article is protected by copyright. All rights reserved.
irrespective of the different Pd contents, which may be caused by the flow-through mode.
The excellent catalytic activity of the catalytic membrane in this work may be attributed to the small size of the Pd nanoparticles, the higher quantity of Pd in the reactor, and the rapid mass transfer that is ascribed to the porous membrane structure and the flow-through mode.
For example, in terms of particle size, the average size of Pd nanoparticles in the report of Wang et al. 24 was greater than 15 nm. A larger average diameter (approximately 26 nm) was found for the Pd nanoparticles in the PVDF composite membranes prepared by Xu et al. 42 However, the Pd nanoparticles in our research are only 4.2 ± 0.2 nm in size (Figure 12), which is much smaller than those reported in the literature and can provide more active sites for the conversion of PNP. Regarding the quantity of Pd used, the total amount of Pd in our reactor is 6 mg, which is significantly higher than those reported in the literature (Table 2).
Furthermore, the flow-through catalytic membrane reactor is constructed with a porous membrane that is loaded with active Pd nanoparticles, and is operated in a dead-end mode to force the feed stream to flow over the active sites as shown in Figure 13. These features indicate that the reaction solution can more easily pass through the membrane and contact the active sites in the pores in our research, resulting in enhanced catalytic efficiency 23 .

Conclusions
In summary, a novel flow-through catalytic membrane reactor system was designed and constructed for the continuous complete conversion of high concentration p-nitrophenol. The This article is protected by copyright. All rights reserved. modification of ceramic membranes with branched TiO 2 nanorods can enhance the Pd content but renders parts of the Pd nanoparticles unusable. High concentrations of p-nitrophenol can be completely converted with less consumption of NaBH 4 in a short time in the flow-through catalytic membrane reactor, which can be attributed to the small Pd nanoparticles, a higher Pd content and rapid mass transfer. The conversion efficiency remained at 100% for 240 minutes, and no byproducts were detected. The catalytic membrane reactor developed here provides a promising prospect for practical applications of industrial wastewater treatment.

Acknowledgments
The financial supports from the National Natural Science Foundation (21776127)  This article is protected by copyright. All rights reserved. This article is protected by copyright. All rights reserved. This article is protected by copyright. All rights reserved.        This article is protected by copyright. All rights reserved.   This article is protected by copyright. All rights reserved.  This article is protected by copyright. All rights reserved.  This article is protected by copyright. All rights reserved.   This article is protected by copyright. All rights reserved.  This article is protected by copyright. All rights reserved.  This article is protected by copyright. All rights reserved.