Noble metal nanoparticles (NMNPs) composed of metals, such as Au, Ag, Pt, and Pd, that range in size from several to dozens of nanometers are effective heterogeneous catalysts for various chemical reactions, including degradation of organic molecules ^{[1, 2, 3, 4, 5]}, oxidation ^{[6, 7]}, hydrogenation ^[8], and C-C coupling ^{[9, 10]}. Because bare NMNPs tend to aggregate, which decreases their catalytic activity, supportive platforms are usually required to load, stabilize, and separate these nanocatalysts. Recently, carbon materials like activated carbon, carbon nanotubes, and graphene have been frequently used as supports because of their high specific surface area, mechanical strength, and chemical stability ^{[11, 12, 13, 14]}. The exfoliation and grain growth of loaded NMNPs, and the poor control of their dispersity are urgent issues that need to be resolved. Mesoporous silica (mSiO₂) materials can also host NMNPs inside their periodic, size-fixed channels. For example, NMNPs have been generated in situ within the pores of MCM-41 and SBA-15 by strategies such as chemical vapor deposition ^[15], supercritical fluids ^[16], impregnation ^[17], and ion exchange ^[18]. Although the exfoliation and grain growth of NMNPs can be substantially confined in mSiO₂, other crucial parameters including size, morphology, and loading of nanocatalyst are still difficult to control.

Bottom-up strategies developed in recent years may help to overcome these limitations ^{[19, 20, 21, 22]}. For example, the surface of polymer latex and silica colloidal nanospheres can be decorated with certain functional moieties like amino or thiol groups. As a result, NMNPs capped with suitable modifiers may attach onto these core templates through coordination or electrostatic interactions. Coverage of such complex by an mSiO₂ shell using sol-gel methodology would produce sandwich-like “core-NMNP-shell” structures. The advantages of such design are obvious: (1) NMNPs with determined composition, size, and morphology can be produced; (2) the loading of the nanocatalyst can be tuned simply by changing the concentration of their corresponding dispersion; (3) the agglomeration, exfoliation, and grain growth of NMNPs can be largely restricted because they are sandwiched between the core and shell; (4) reactants are easily transported across the mSiO₂ shell to react with the internal NMNP catalysts.

Compared with a spherical mSiO₂ shell, mSiO₂ nanotubes with open ends and a one-dimensional passage are thought to allow more efficient reactant transportation. Unfortunately, because of the lack of readily dispersible nanowire templates, the bottom-up fabrication of mSiO₂ nanotubes loaded with NMNPs is still challenging. Recently, the synthesis of aniline oligomer-based organic nanowires (ONWs) by self-assembly was reported ^[23]. Their unique water dispersity and surface charge means that these ONWs may be used as templates for the facile production of NMNP-mSiO₂ composite nanotubes. Herein, gold (Au) nanoparticles (NPs) are anchored to the inner wall of mSiO₂ nanotubes. The catalytic performance of these Au-mSiO₂ composite nanotubes is also examined by their ability to reduce the representative organic dye 4-nitrophenol (4-NP).

Poly(diallyldimethylammonium chloride) (PDDA, 20% aqueous solution) was purchased from Sigma-Aldrich. Poly(vinylpyrrolidone) (PVP, molecular mass = 58000 g/mol), ammonium persulfate (APS, 98.5%), and trimethoxy(octadecyl)silane (C₁₈TMS, 90%) were obtained from Aladdin Reagent Co. Aniline, xylene, hydrogen tetrachloroaurate (HAuCl₄), trisodium citrate (Na₃Cit), NaBH₄, tetraethyl orthosilicate (TEOS), ammonia (28%), and 4-NP were purchased from Sinopharm Chemical Reagent Co. Deionized water was used throughout the experiments.

Water-dispersible ONWs were synthesized according to the literature ^[23]. In a typical experiment, PVP (0.5 g) and APS (0.1 g) were dissolved in deionized water (50 g). The solution was covered with xylene (20 g) to construct an oil-water interface. Aniline (0.1 g) was added dropwise into the upper xylene phase, and then the system was left undisturbed at room temperature for 24 h. ONWs formed in the bottom aqueous phase. The ONWs were separated by centrifugation and rinsed with deionized water.

The ONWs were modified with PDDA as follows. ONW powder (90 mg) and PDDA solution (1.0 g) were added to deionized water (20 g) and stirred at room temperature for 10 min. The PDDA-modified ONWs were separated by centrifugation and rinsed with deionized water three times before further usage.

Au NPs (diameter = 12 nm) were synthesized by a reported method ^[24]. An aqueous solution of HAuCl₄ (5 mL, 3 mmol/L) was mixed with deionized water (10 mL) and heated to boiling point. Na₃Cit solution (10 mL, 5 mmol/L) was added, and then the mixture was reacted for 30 min to form a violet-red solution. Au NPs (diameter = 5 nm) were synthesized using a similar procedure ^[25]. An aqueous solution of HAuCl₄ (5 mL, 3 mmol/L), deionized water (25 mL), and Na₃Cit solution (10 mL, 5 mmol/L) was mixed at room temperature. NaBH₄ solution (10 mL, 2 mmol/L) was added dropwise under vigorous stirring, giving a pale red solution after 30 min.

To synthesize ONW-Au composites, a solution of Au NPs (12 nm, 1 mL) was slowly injected into a solution of the PDDA-modified ONWs (90 mg) in water (20 g). The mixture was stirred for 20 min before centrifugation and rinsing with water. Au NPs with a diameter of 5 nm were attached onto the PDDA-modified ONWs in the same manner. The loading of Au NPs can be tuned by varying the volume of their solution mixed with ONWs.

ONW-Au composite (90 mg) was added to deionized water (20 g), ammonia (0.3 g), and ethanol (2.0 g). TEOS (0.3 g) and C₁₈TMS (0.1 g) were added, and the mixture was stirred at room temperature for 12 h. The product was isolated by centrifugation, rinsed with deionized water and then ethanol, and dried at 80 °C to give an ONW-Au-SiO₂ composite as a pale yellow powder.

ONW-Au-SiO₂ composite was heated at a rate of 2 °C/min to 550 °C and then maintained at this temperature for 2 h. During heating, both the ONWs and pore-directing agents (the alkyl chain of C₁₈TMS) within the SiO₂ shell were removed to generate Au-mSiO₂ composite nanotubes as a pink powder. The Au-mSiO₂ composite nanotubes contained approximately 4% Au NPs when 1 mL of the 12-nm Au NP solution was used, as measured by inductively coupled plasma (ICP) spectroscopy. For comparison of catalytic performance, an analogue of the Au-mSiO₂ composite nanotubes was also prepared without using the pore-directing agent C₁₈TMS. This analogue is denoted as Au-SiO₂ composite nanotubes.

The reduction of 4-NP by NaBH₄ using Au-mSiO₂ composite nanotubes as a catalyst was performed in a quartz cuvette at room temperature. Typically, 4-NP aqueous solution (0.1 mL, 5 mmol/L) was mixed with an excess amount of fresh NaBH₄ solution (1 mL, 0.2 mol/L), and then diluted to 4 mL. Au-mSiO₂ composite nanotubes (0.1 mL, 1 mg/mL) were added. The mixture was immediately subjected to ultraviolet-visible (UV-Vis) measurements as the first run (denoted as the starting point t = 0). The absorption of the dispersion was tested at intervals of 3 min until the dye was completely reduced. As control experiments, Au-SiO₂ composite nanotubes and bare Au NPs were also used as catalysts in the reduction of 4-NP.

The zeta potential of the samples was measured by a zeta potential analyzer (Nano-ZS90, Malvern, UK). The morphology of samples was observed by a transmission electron microscope (TEM, Tecnai G2 F30, FEI) operating at 120 kV, as well as a scanning electron microscope (SEM, QUANTA 200 FEG, the Netherlands) equipped with a cold-field emission gun operating at 20 kV. Fourier-transform infrared (FTIR) measurements were carried out on an FTIR spectrophotometer (TENSOR27, Bruker) in the range of 4000-400 cm^-1 with powder-pressed KBr pellets. Thermogravimetric analysis (TGA, STA449F3, Netzsch, Germany) was performed from room temperature to 750 °C at a heating rate of 10 °C/min and N₂ flow rate of 20 mL/min. UV-Vis spectra were collected in the 200-600 nm range using a spectrophotometer (UV-2550, Shimadzu). Powder X-ray diffraction (XRD) patterns were measured on a diffractometer (RINT D/Max 2500, Rigaku) using Cu K_α radiation (λ = 0.15432 nm). The nitrogen physisorption experiment was performed on a gas sorption system (QUADRASORB SI, Quantachrome Instruments) using a dried powder sample that was degassed at 120 °C under vacuum. Pore size distribution curves were calculated from the desorption isotherm curves using the Barrett-Joyner-Halenda (BJH) method. The Au loading of the catalysts was determined by ICP on a spectrometer (ICPS-8100, Shimadzu) after dissolution in dilute HF/HCl solution.

The procedure used to produce Au-mSiO₂ composite nanotubes is illustrated in Scheme 1. First, negatively charged ONWs with a zeta potential of -10 mV were adsorbed by cationic polyelectrolyte PDDA to acquire a positively charged surface with a zeta potential of +60 mV. As a result, when the solution of citrate-stabilized Au NPs with a zeta potential of -30 mV was added, Au NPs spontaneously attached onto the PDDA-modified ONWs through electrostatic interactions to form the intermediate termed ONW-Au composite. The dispersion of ONW-Au composite was adjusted to pH = 9-10 before suitable amounts of TEOS and pore-directing agent C₁₈TMS were added to undergo a sol-gel reaction. Positively charged PDDA adsorbed on the surface of the ONWs interacted with the negatively charged silica sol-gel through electrostatic interactions, producing a SiO₂ shell on the surface of the ONW-Au composite to generate the ONW-Au-SiO₂ composite. The thickness of the SiO₂ coating could be controlled by varying the amount of precursors added. Because the Au NPs were sandwiched between the inner ONW template and outer SiO₂ shell, their agglomeration, exfoliation, and grain growth should be restricted during calcination post-treatment. Calcination removed both the ONW template and pore-directing agents within the SiO₂ shell to form Au-mSiO₂ composite nanotubes, in which Au NPs were anchored to the inner wall of an mSiO₂ shell.

Scheme 1. Synthetic procedure to form Au-mSiO2 composite nanotubes.

Fig. 1 shows TEM and SEM images of the intermediates and final product. The PDDA-modified ONWs were about 250 nm in diameter with a smooth surface (Fig. 1(a)). Well-dispersed Au NPs with a size of 12 nm (Fig. 1(b)) were then uniformly adsorbed onto the oppositely charged ONW templates (Fig. 1(c)). After coating with a SiO₂ shell (Fig. 1(d)), Au NPs were sandwiched between the inner ONW template and outer SiO₂ shell. The diameter of the ONW-Au-SiO₂ composite exceeded 300 nm, indicating that the SiO₂ shell was dozens of nanometers thick. After calcination, Au-mSiO₂ composite nanotubes were obtained (Fig. 1(e)). No free Au NPs were found in the images. In addition, the Au NPs anchored to the mSiO₂ shell were not washed away, confirming the strong interaction between the Au NPs and mSiO₂ shell. The open end of the Au-mSiO₂ composite nanotubes could also be observed (Fig. 1(f)). The uniform size and distribution of the loaded Au NPs were not obviously affected during the synthetic process (Fig. 1(g)), and the composite nanotubes did not aggregate after calcination (Fig. 1(h)).

Fig. 1. TEM images of (a) PDDA-modified ONWs, (b) Au NPs with a size of 12 nm, (c) ONW-Au composite, (d) ONW-Au-SiO2 composite, and (e) Au-mSiO2 composite nanotubes. (f) SEM image of open-ended Au-mSiO2 composite nanotubes, (g) size distribution of Au NPs in Au-mSiO2 composite nanotubes, and (h) SEM image of well-dispersed Au-mSiO2 composite nanotubes.

Unlike the ONWs and ONW-Au-SiO₂ composite, the Au-mSiO₂ composite nanotubes (Fig. 2(a)) did not exhibit peaks at 3263, 3242, 1642, 1580, 1510, 1288, 741, and 694 cm^-1 in their FTIR spectrum ^[26]. The disappearance of these characteristic peaks of the ONWs confirms the complete removal of this template by calcination. TGA measurements (Fig. 2(b)) were also consistent with this result because the Au-mSiO₂ composite nanotubes showed negligible mass loss of organic components.

Fig. 2. (a) FTIR spectra and (b) TGA curves for ONWs (1), ONW-Au-SiO2 composite (2), and Au-mSiO2 composite nanotubes (3).

Fig. 3(a) depicts the UV-Vis spectra recorded for the intermediates and final product dispersed in water. The absorptions at 430 and 523 nm correspond to ONWs and Au NPs with a size of 12 nm, respectively. Compared with the ONW-Au-SiO₂ composite, the Au-mSiO₂ composite nanotubes showed negligible shift of the 523-nm peak, indicating that the size of the Au NPs barely changed during calcination ^[22]. This is in accordance with the TEM observations. In the XRD patterns of the samples (Fig. 3(b)), the broad peak at 20°-25° can be attributed to the amorphous silica shell, and characteristic diffraction peaks for cubic Au (JCPDS No. 04-0784) are observed for the Au-mSiO₂ composite nanotubes.

Fig. 3. (a) UV-Vis spectra and (b) XRD patterns obtained for ONWs (1), ONW-Au-SiO2 composite (2), and Au-mSiO2 composite nanotubes (3).

The N₂ adsorption-desorption isotherm of the Au-mSiO₂ composite nanotubes is a classical type-IV curve (Fig. 4), indicating that the nanotubes are mesoporous. The most common pore size is around 3.7 nm (inset of Fig. 4), which is large enough for the transportation of molecular reactants. The Brunauer-Emmett-Teller surface area of the Au-mSiO₂ composite nanotubes was 469 m²/g.

Fig. 4. N2 adsorption-desorption isotherm and pore size distribution curve (inset) of the Au-mSiO2 composite nanotubes.

When the amount of ONW template was fixed, the loading of Au NPs could be controlled by varying the volume of Au NP solution added to the ONWs, as shown in Fig. 5. The uniform dispersity and distribution of Au NPs were well preserved even for high-density loading (Fig. 5(c)). In addition, the size of the Au NPs could also be tuned by using Au NPs of different size (Fig. 6).

Fig. 5. TEM images of ONW-Au composites with different Au NP loading. The volume of Au NP (12 nm) solution added to a fixed amount of ONWs was: (a) 0.2 mL (~1% Au content), (b) 1 mL (~4% Au content), and (c) 3 mL (~8% Au content). In each case, Au NP solution was injected very slowly into a solution of the PDDA-modified ONWs (90 mg in 20 g of water) and stirred for 20 min.

Fig. 6. TEM images of ONW-Au composites containing Au NPs of different size. (a) 5 nm (~2% Au content); (b) 12 nm (~4% Au content). The volume of Au NP solution added to the ONWs in both cases was 1 mL. Au NP solutions were injected very slowly into solutions of the PDDA-modified ONWs (90 mg in 20 g of water) and stirred for 20 min.

The reduction of 4-NP by NaBH₄ was chosen as a representative reaction to examine the catalytic performance of the Au-mSiO₂ composite nanotubes with a Au content of ~4%. The original 4-NP solution exhibited an absorption peak at 317 nm (Fig. 7(a)). After NaBH₄ was added, this absorption band shifted to 400 nm because of the formation of 4-nitrophenolate. Before adding the catalyst, the color of the organic dye did not fade, even after 24 h; therefore, the addition of catalyst was necessary for the degradation of 4-NP. After Au-mSiO₂ composite nanotubes were added, the mixture was immediately subjected to UV-Vis measurement, which was labeled as the starting point (t = 0). UV-Vis measurements of the dispersion were then recorded at intervals of 3 min. The intensity of the absorption peak at 400 nm gradually decreased as the reaction proceeded (Fig. 7(b)) and a new peak emerged at 295 nm, revealing the conversion of light yellowish 4-NP into colorless 4-aminophenol (4-AP) (inset of Fig. 7(b)). Conversion into 4-AP exceeding 95% was achieved within 33 min. As previously reported ^[16], when the concentration of NaBH₄ is far higher than that of 4-NP, the reaction rate would be first order with respect to the concentration of 4-NP. As a result, linear plots of ln(c/c₀) against time were obtained (Fig. 7(c)). The calculated rate constant k was 0.04 and 0.08 min^-1 using Au-SiO₂ composite nanotubes (without mesoporous shell, Au content of ~4%) and Au-mSiO₂ composite nanotubes as catalyst, respectively. These values indicate that the mesoporous shell provides a larger surface area and more channels for reactants to diffuse through to the Au NPs, thus increasing catalytic efficiency. To compare the reusability of the Au-mSiO₂ composite nanotubes with unprotected Au NPs, both catalysts were separated and used in repeated catalytic cycles (Fig. 7(d)). The Au-mSiO₂ composite nanotubes exhibited similar conversion (>95%) within the same time period for five cycles. In contrast, the catalytic efficiency of bare Au NPs quickly decreased as the cycles proceed because of their agglomeration and loss of catalyst particles. Moreover, the size and content of the Au NPs in the Au-mSiO₂ composite nanotubes did not change obviously after five cycles (Fig. 7(e)), excluding leaching and agglomeration of catalyst during reaction.

Fig. 7. (a) UV-Vis spectra of 4-NP and 4-nitrophenolate; (b) UV-Vis spectra of the reduction of 4-NP catalyzed by Au-mSiO2 composite nanotubes; (c) Plots of ln(c/c0) versus time using Au-SiO2 composite nanotubes and Au-mSiO2 composite nanotubes as the catalyst; (d) The reusability of Au-mSiO2 composite nanotubes and unprotected Au NPs as catalysts in the reduction of 4-NP; (e) TEM image of a Au-mSiO2 composite nanotube after five cycles of catalysis.

The activity of previously reported composite catalysts prepared by bottom-up approaches in the reduction of 4-NP is summarized in Table 1. Overall, the Au-mSiO₂ composite nanotubes showed comparable activity to other spherical-shaped composite catalysts. However, unlike spherical supports, the tubular supports contain open ends and a one-dimensional passage. We believe that optimization of these structural features will allow more efficient catalysis of chemical reactions.

Table 1 Activity of composite catalysts prepared by bottom-up approaches in the reduction of 4-NP. Composite catalyst BET surface area (m2/g) Loaded Au NPs Rate constant k (min-1) Reusability (cycle) Ref. Size (nm) Content (%) Au@SiO2 yolk/shell particles — 43-104 — 0.10-0.20 — [19] Fe3O4/SiO2/Au/porous-SiO2 composite colloids 137 15 — 0.39-0.82 6 [20] Fe3O4@SiO2-Au@mSiO2 microspheres 236 12 ~2.0 0.20-0.35 8 [21] Au@SiO2 hollow mesoporous nanocomposite spheres 432 12 ~4.5 0.10-0.15 8 [22] Au-SiO2 composite nanotubes 136 12 ~4.0 0.04 — this work Au-mSiO2 composite nanotubes 469 12 ~4.0 0.08 5 this work

Table 1 Activity of composite catalysts prepared by bottom-up approaches in the reduction of 4-NP.

Au-mSiO₂ composite nanotubes were prepared by a bottom-up approach using water-dispersible ONWs as a template. During the synthetic process, a sandwich-like structure of core-Au NP-shell was formed, in which Au NPs were located between the inner ONW template and outer SiO₂ shell. As a result, the agglomeration, exfoliation, and grain growth of Au NPs were restricted. After calcination, Au-mSiO₂ composite nanotubes with Au NPs anchored to the inner wall were formed. The loading and size of Au NPs could be easily tuned by manipulating the concentration and size of Au NPs in the dispersion mixed with the ONWs. When the Au-mSiO₂ composite nanotubes were used as catalysts in the reduction of 4-NP, the mesopores, open ends, and one-dimensional passage of the mSiO₂ nanotubes jointly facilitated the diffusion of reactants and thus increased catalytic efficiency. These composite catalysts also exhibited good reusability, and their overall activity was comparable to that of previously reported spherical composite catalysts.