Mesoporous colloidal silica cubes with catalytically active cores

Preparation methods of cubic core-shell particles with specific functionality are limited. Here we demonstrate the possibility to transform cuprous oxide cubes coated with mesoporous silica into functional core-shell particles, while retaining their cubic shape. Cuprous oxide nanocubes are coated with mesoporous silica using cetyltrimethylammonium bromide as a template, after which the cuprous oxide core is transformed using liquid phase calcination and galvanic replacement. Nitrogen physisorption and electron microscopy confirm that mesoporous silica coatings are obtained with tuneable thickness. The successful transformation of cuprous oxide into gold and silver is assessed via UV–VIS spectroscopy and energy dispersive X-ray spectroscopy. Particles with a silver core and a cubic mesoporous silica shell are demonstrated to be catalytically active in the degradation of the dye Congo red.


Introduction
Colloidal cubes are able to form dense packings when assembled by gravity [1,2], the depletion interaction [3], or confinement in 2D [4] and 3D [5,6]. Cubic colloids can be prepared as noble metals [7][8][9], metal oxides [10,11], and perovskites [12], all of which have properties depending on their detailed size, shape and chemistry. For applications it is desirable to combine the catalytic [8] or plasmonic properties [13] of nanomaterials with a specific shape, arranged in a crystal lattice which is only obtainable by assembling particles with a different shape. The formation of bicontinuous Pickering emulsions illustrates this point, as specific particle shape and surface chemistry are beneficial, but the inclusion of shape decoupled functionality is not yet achieved. [14,15] A method commonly proposed to add functionality to particles, is the impregnation of porous materials with functional materials [16], sometimes referred to as the "ship-in-a-bottle approach" [17]. Although this approach allows to incorporate different functional materials in a carrier material, the product often has low specificity and a low fraction of filled particles [18]. Here we propose a different route towards the deposition of functional material into an anisotropic carrier material, using the morphology of silica on cuprous oxide cubic-core-shell particles [19]. By using a method that alters the core but does not affect the silica shell, it is possible to completely transform the cuprous-oxide to gold or silver, while maintaining the outer shape of the particles.
Core-shell particles, or hollow materials containing functional materials, have been the subject of various studies. Since these hierarchical and complex morphologies often have a large surface area, they are expected to find applications as catalysts, energy storage or drug carriers [20][21][22][23]. One method to obtain functional core-shell particles is the self-templating method [21], via which functional colloids with complex morphologies, including cubes [24] and plates [25], have been prepared. The self-templating method uses the nanoparticle as a template to obtain a core-shell particle and complete transformation often results in a different morphology of the particle [21]. Functional materials deposited in an inert carrier material, such as silica, allows for the complete transformation of the functional material while preserving the shape and stability provided by the silica shell. An example of transformation in a silica shell, is the etching and overgrowth procedure of gold nanorods coated with mesoporous silica [26,27]. Using the etching and overgrowth method, it is possible to obtain rods with a tuneable surface plasmon resonance. Similar procedures could be used to transform cubic core-shell particles. However, preparation methods of cubic particles coated with mesoporous silica are limited [28][29][30][31][32] and to the best of our knowledge no reports have been presented yet on the transformation of functional materials inside mesoporous silica nanocubes. A mesoporous silica coating on functional cores provides stability while allowing free transport of molecules to the core [33,34]. A common method to obtain mesoporous silica is by the deposition of silica in the presence of surfactants [35]. For instance, using cetyltrimethylammonium bromide (CTAB) micelles it is possible to obtain free standing mesoporous silica [36] or core-shell particles [37]. The CTAB template can afterwards be extracted by ion exchange [38] or calcination [36,39] resulting in silica with mesoscale pores. Characteristic for mesoporous silica obtained from CTAB micelles are pores with a diameter of 2 nm and surface areas over 1000 m 2 g −1 [35].
The aim of the present work is to explore the coating of cuprous oxide cubes with a mesoporous silica shell and the subsequent transformation of the cuprous oxide core into a functional material while conserving the cubic mesoporous silica. First, we attempt coating cuprous oxide nanocubes [19,40] with a mesoporous silica shell using the CTAB soft template. The formed shells are then characterized using TEM and nitrogen physisorption. Using a high-temperature boiling solvent, the silica shell can afterwards be calcined while simultaneously thermolyzing the cuprous oxide into metallic copper. Finally, using galvanic replacement, the copper core can be substituted by noble metals like silver or gold. The resulting particles are characterized with UV-VIS spectroscopy and energy dispersive X-ray spectroscopy. The functionality of the newly formed particles is assessed by the catalytic dye degradation of Congo red.

Cuprous oxide cube synthesis
The Cu 2 O cubic template was produced using a method developed in our group [19]. To a 500 mL 3-neck roundbottom flask equipped with reflux and dripping funnel, 223 g 1,5-pentanediol and 25.3 g PVP-55 were added. The reflux condenser and dripping funnel were connected to a N2/Vacuum Schlenk line. The mixture was vacuum flushed three times at room temperature, heated to 100 °C by oil bath and vacuum flushed three times. 5.2 g Cu(acac) 2 and 75 g 1,5-pentanediol were weighed in a 100 ml bottle and mixed by a combination of shaking and sonication. The Cu(acac) 2 mixture was transferred to the dripping funnel and flushed four times. The PVP-55 solution was heated to 200 °C and the Cu(acac) 2 mixture was added over 30 s aided by N 2 pressure. After 22 min, heating was discontinued, and the mixture was allowed to cool down to room temperature, yielding a brown mixture. The product was collected with 500 ml acetone. The resulting sol was separated by centrifugation (3000 g, 2.5 h) and washed twice with ethanol by centrifugation (12,000 g, 30 min). Finally, the sample was dispersed in 50 mL ethanol.

Mesoporous shell synthesis
Cubic cuprous oxide core, mesoporous silica shell particles (Cu 2 O@mSiO 2 ) were obtained by coating Cu 2 O cubic template particles with silica according to a modified procedure from Jiang et al. [41]. Cu 2 O particles (100 mg in 164 mL of ethanol) were added to a 500 mL round-bottom flask equipped with mechanical stirrer. Water (36 mL), TEOS (0.125 mL), and CTAB (0.2 g) were added to the nanocubes and the mixture was stirred and sonicated for 15 min. An aqueous sodium hydroxide solution (20 mM, 10 mL) was added over 10 min using a Gilson peristaltic pump. Sonication was continued for 2 h after which the reaction was stirred overnight. The particles were washed by centrifugation (12,000 g, 30 min) three times and finally dispersed in ethanol (5 mL).

Template removal
Template extraction was performed to obtain hollow mesoporous particles with a cubic void (mSiO 2 shells) by a method adapted from Deekamwong et al. [38]. To a 100 mL 2-neck round-bottom flask equipped with magnetic stirring and a reflux condenser, silica coated cubes (50 mg) and ethanol (35 mL) were added. Ammonium chloride (1.02 g) was added and the mixture was refluxed for 45 min. The mixture was allowed to cool down, yielding an aqua-coloured mixture. Water (33 mL) followed by a drop a nitric acid were added, resulting in a clear dispersion. The product was washed by centrifugation, first three times with water, followed by three times with ethanol (20,000 g, 40 min). Finally, the particles were dispersed in ethanol (5 mL).

Liquid phase calcination and cuprous oxide thermolysis
Liquid Phase Calcination (LPC) was performed in a high boiling point solvent, trioctylphosphine oxide (TOPO), under nitrogen atmosphere to obtain Cu@mSiO 2 particles.
[39] TOPO (7.1 g) was heated to 60 °C in a 100 mL roundbottom flask equipped with reflux condenser and stir bar. Cuprous oxide cubes coated with mesoporous silica (Cu 2 O@ mSiO 2 , 50 mg in 2.5 mL ethanol) were added to the liquified TOPO, followed by removal of the ethanol by heating under vacuum. The mixture was heated to 350 °C under nitrogen atmosphere, and allowed to react for 1 h, after which the mixture was cooled down to 100 °C. Water (25 mL) was added and the sample was washed by centrifugation with water, three times, and ethanol, three times. The mixture was finally dispersed in ethanol (50 mL).

Galvanic replacement
Galvanic replacement of copper into silver was done following a procedure adapted from Stewart et al. [42]. The Cu@mSiO 2 (0.4 mg) obtained from LPC was dispersed in water (0.5 mL) after which freshly prepared solution of ascorbic acid (0.5 mL, 1.16 M in water) was added. The mixture was vortexed for 10 s and directly washed by centrifugation (16,000 g, 10 min) with water twice, and then dispersed in water (0.5 mL). Silver nitrate (0.5 mL, 50 mM in water) was added, the mixture was vortexed for 10 s and then directly washed by centrifugation with water (twice) and ethanol (twice). Finally, the particles were dispersed in ethanol (5 mL). For the galvanic replacement of copper with gold, HAuCl 4 (0.5 mL, 25 mM in water) was added instead of the silver nitrate.

Catalytic dye degradation
Catalytic dye hydrogenation by Ag@mSiO 2 particles was demonstrated by degradation of Congo red. To water (9 mL) Ag@mSiO 2 particles were added (0.4 mL, 0.1 mg mL −1 ). To this dispersion an aqueous Congo red solution (70 µL, 5 mM) was added. Freshly prepared sodium borohydride (0.5 mL, 1 M) was added to initiate the dye hydrogenation, and changes in colour were monitored visually and with UV-VIS, with time intervals of approximately six minutes. During the reaction gas bubbles develop on the surface of the cuvet, a baseline correction was performed to account for scattering caused by these bubbles. Two control experiments were performed, one with mSiO 2 particles (0.01 mL, 0.4 mg mL −1 ) with NaBH 4 and Congo red, and one with NaBH 4 and Congo red in the absence of any particles.

Characterization
Transmission Electron Microscopy (TEM) was performed on a TecNAI 12 and TecNAI 20 electron microscope. Images were recorded on an SIS CCD Megaview II camera. Samples were prepared by diluting a drop of analyte in water and casting a drop of dilute analyte on a copper grid, followed by drying of the copper grid under a heating lamp. Energy-dispersive X-ray spectroscopy (EDX) maps were obtained from a FEI Talos F200X, with samples prepared on an aluminum grid. Nitrogen physisorption isotherms were obtained with a Micrometrics Tristar II plus. Prior to the measurements the samples were dried overnight at 250 °C.
The surface area was determined using the BET isotherm [43], and the micropore size distribution was determined using the BJH method assuming cylindrical pores. UV-VIS spectra were obtained using a PerkinElmer Lambda 35 UV-VIS spectrometer. Samples were measured in quartz cuvettes and the obtained spectra, ranging 250-700 nm, were normalized using Wolfram Mathematica. IR spectra were obtained using a PerkinElmer FT-IR/FIR Frontier spectrometer in transmission mode. 250 mg KBr with 1 mg sample was dried in a Memmert oven at 70 °C overnight. Copper core samples were dried under vacuum and nitrogen flow.

Template synthesis and mesoporous coating
For coating with mesoporous silica (MCM-41) cuprous oxide cubes were synthesized with edge lengths of 96 ± 10 nm. MCM-41 silica is commonly prepared with ammonia as a base catalyst. Ammonia, however, dissolves the cuprous oxide core [44], which required us to find an alternative method. Using tetramethyl ammonium hydroxide as base catalyst results in non-uniform coatings, and adding TEOS to Cu 2 O cubes with sodium hydroxide results in damaged cubes. By drop-wise addition of sodium hydroxide to the precursor mixture, it is possible to obtain cubes coated with mesoporous silica. Below, Cu 2 O nanocubes are visible before (Fig. 1ab) and after (Fig. 1cd) coating, showing a lower contrast shell around the cubic core after the coating step. Close up inspection of the obtained silica shells, Fig. 1c, reveals distinct features with sizes ranging between 2 and 4 nm. Similar to the microporous silica, the coating thickness can be controlled by increasing the amount of added TEOS (Fig. 2b). The main differences between Stöber silica coating previously studied in our group, [19] and the here explored mesoporous coating, are the thickness of the coating with respect to the TEOS/Cu 2 O ratio, and the thickness of the coating at low amounts of added TEOS, visible in Fig. 2. For a given TEOS/Cu 2 O ratio the mesoporous silica coatings are significantly thicker than for Stöber silica. This increased thickness is a direct result from the incorporated CTAB micelles in the coating, increasing the total volume of the silica shell. At low amounts of added TEOS, (below 1.20 µL/mg) the thickness drops down sharply, and suggests that adding less than 0.7 µL/mg TEOS would not result in a mesoporous coating. This is in contrast to the Stöber silica, where the data fit intersects the origin. [19] The reason for this difference can be attributed to the different reaction conditions. If instead of the layer thickness, the silica volume nm 3 is plotted, the trend is expected to be linear. This is the case above 1.5 µL/mg (TEOS/Cu 2 O), and is shown in SI.1.
After extraction of the cuprous oxide template, the porosity of cubic mesoporous mSiO 2 shells was assessed with nitrogen physisorption (Fig. 3). In Fig. 3b, the N 2 physisorption isotherm of mSiO 2 shells is presented, showing a step increase at p/p 0 ≈ 0.2, and a large hysteresis loop with sudden desorption at p/p 0 ≈ 0.5, both typical for mesoporous materials [43]. The pore size distribution of mSiO 2 shells is depicted in the inset of Fig. 3b. From Fig. 3b it is apparent that mesopores are present in the silica obtained with CTAB, showing a distinct peak in the pore size distribution around 2 nm. This peak is not visible in the analysis of the nitrogen physisorption isotherm of Stöber silica shells (SI.2). The peak around 4 nm in de desorption curve is an artefact caused by meniscus instabilities, often observed in mesoporous materials [45]. Additionally, the mSiO 2 shells show a small peak in the desorption curve around 60-90 nm, which roughly corresponds with the size of the cubic void present in the hollow particles. The BET surface area was determined to be 1156 ± 17 m 2 g −1 . The density of the mesoporous silica was found to be 0.8 g/cm 3 by comparing the relative weight increase to the relative volume increase. This is in agreement with the 0.8 to 0.9 g/cm 3 found in literature. [46] Further insight into this calculation is given in SI.3.

Cuprous oxide transformation and galvanic replacement
Liquid phase calcination (LPC) [39] was used to calcine the cubic CuO 2 @mSiO 2 particles while maintaining their colloidal stability. It was found that conventionally applied air calcination [36] resulted in aggregated particles that could not be dispersed in water or ethanol. During LPC of Cu 2 O@mSiO 2 it was found that the calcination transformed the cuprous oxide core in a metallic copper core. This was earlier observed by Gao et al. for Cu 2 O microcubes, who attributed the transformation to thermolysis [47]. After LPC, it is visible in Fig. 4a that the cubic cuprous oxide core evolves into a smaller rounded core, which is a result of the expulsion of oxygen. This transformation of the Cu 2 O into metallic copper is also visible on IR and UV-VIS spectra. In Fig. 4b a clear decrease of the Cu-O stretching vibration at 650 cm −1 is visible after LPC, indicating the absence of cuprous oxide. Additionally, a decrease in O-H stretching and bending vibrations at 3500 cm −1 and 1650 cm −1 , respectively, is observed, indicating a loss of hydroxyl groups present in the particles. Furthermore, the UV-VIS spectra in Fig. 4c show a clear transformation from a typical Cu 2 O spectrum (blue) [8] to the absorption spectrum of Cu (red) [48]. After LPC the particles could be dispersed in ethanol and regained stabilities comparable to before the calcination step, indicating the LPC method is a good way to calcine particles while retaining colloidal stability. In SI.4 TEM image are shown of samples taken during the LPC process, displaying the effects of the oxygen expulsion and subsequent surface melting. shells. The inset shows the dV/dD pore volume, with a characteristic pore size of 2-3 nm found in mesoporous silica prepared with CTAB Table 1 Half reaction standard potentials [49] Half reaction Standard Potential (V) Since the copper cores are present in all Cu@mSiO 2 cubes, and can be readily accessed by reagents, they can function as seeds for further modification. Using galvanic replacement, it is possible to transform the copper cores into silver or gold. This transformation is visible from the change in UV-VIS spectra after galvanic replacement, Fig. 4c showing distinct UV-VIS absorption of both silver (green) and gold (yellow). Another indication of the successful galvanic replacement of copper with silver and gold, is the change in volume. Since the galvanic replacement follows standard half-reaction stoichiometry, as listed below in table 1, every three copper atoms are replaced by either two gold atoms or six silver atoms. Fig. 4 a TEM images of Cu 2 O@mSiO 2 (blue), Cu@mSiO 2 (red), Ag@mSiO 2 (green) and Ag@mSiO 2 (yellow). Scale bar 100 nm b IR spectroscopy before (Cu 2 O@mSiO 2 ) and after (Cu@mSiO 2 ) liquid phase calcination demonstrating the removal of cuprous oide. c UV-Vis spectra displaying surface plasmon resonance peaks characteristic of metallic copper, silver and gold The result is an increase in core volume after the galvanic replacement with silver and a decrease in core volume after replacement with gold, both visible in Fig. 4a.
Since the changes observed by TEM and UV-VIS are not conclusive whether the formed material is indeed silver or gold, energy-dispersive X-ray spectroscopy (EDX) was employed to determine the nature of the formed core. In Fig. 5, high angle annular dark field (HAADF/DF) micrographs of the particles after galvanic replacement with gold are shown in combination with EDX maps. The EDX signals of gold overlap with the core visible in HAADF, indicating the high-contrast material inside the cubes is indeed gold. Furthermore, EDX indicates that some trace amounts of copper are still present. Also, the EDX maps of Si and O are shown, which also coincide with the HAADF image of the low contrast silica shell. The carbon signal observed on the particle is speculated to be a result of adsorbed organic contaminants. The core-shell structure becomes apparent when the gold and silicon maps are combined, as visualized in Fig. 5, showing a clear silicium shell around the silver core. In Fig. 6, similar EDX maps of the particles obtained by the galvanic replacement with silver are presented, showing the same general features as the Ag@mSiO 2 particles. In contrast to the golden core, the silver core leaves a central void and covers the inner silica confinement, these shapes are not unusual for galvanic replacement and are attributed to the Kirkendall effect [50]. The noble nature of the gold core is confirmed via the traditional "acid test" and is discussed in SI.5.

Catalytic dye degradation
The functionality of the silver cores was assessed by the catalytic degradation of Congo red (Fig. 7). Congo red is readily hydrogenated by NaBH 4 in the presence of silver nanoparticles to colourless biphenyl and 4-amino-1-naphtalene-sulfonate [51]. Because Congo red degrades into colourless reaction The degradation rate is further discussed in SI.7 products, the reaction is easily monitored by UV-VIS. After addition of NaBH 4 , the Ag@mSiO 2 particles quickly discoloured the Congo red solution, resulting in very little absorption after 47 min (Fig. 7b). In Fig. 7a the Congo red concentration, determined from a calibration curve, is plotted against time for the dye degradation with Ag@mSiO 2 , mSiO 2 shells, and without particles. It is clearly visible that the silver-containing nanocubes quickly degrade the dye, while the hollow mesoporous silica particles and the control (no particles) show only a slow degradation of Congo red. The spread in the data is likely caused by the production of gas bubbles in the cuvette by sodium borohydride, which interferes with the light-beam in the UV-VIS spectrophotometer.

Conclusions and outlook
Using a combination of liquid phase calcination and galvanic replacement it is possible to transform cuprous oxide -mesoporous silica core-shell nanocubes into nanocubes with functional cores. Cuprous oxide nanocubes are coated with mesoporous silica by using a micellar cetyltrimethylammonium bromide template. The thickness of the coating can be tuned between 10 and 70 nm by controlling the tetraethyl orthosilicate concentration in the reaction mixture. Nitrogen physisorption shows the colloidal cubic mesoporous silica shells to have a BET surface area of 1156 ± 17 m 2 g −1 and mesopores in the 2-3 nm range. With liquid phase calcination, the cubic cuprous oxide core is transformed in a spherical metallic copper core, which can be transformed into silver and gold using galvanic replacement reactions. Purely based on their standard potential, also other catalytically potent metals, such as platinum, palladium and rubidium, could potentially be used to replace the copper. [52] It is demonstrated that the silver core catalyzes the degradation of Congo red in the presence of sodium borohydride.
With a route to transform cubic core-shell particles into functional materials, it is possible to explore the deposition of different functional materials in the cubic void. Since galvanic replacement is limited to reductions with higher standard reaction potentials compared to the reduction of Cu 2+ , ion exchange might offer a possibility to deposit materials other than noble metals. Another possible functionalization is by using the copper core as catalyst for controlled radical polymerization, which might allow the selective deposition of functional polymers in the voids.