Self-Assembly of Colloidal Cubes via Vertical Deposition - Langmuir

to enhance sensing signals. Piyush Sindhu Sharma , Marcin Dabrowski , Francis D'Souza , Wlodzimierz Kutner. TrAC Trends in Analytical Chemistry 20...
0 downloads 0 Views 3MB Size
Article pubs.acs.org/Langmuir

Self-Assembly of Colloidal Cubes via Vertical Deposition Janne-Mieke Meijer,*,† Fabian Hagemans,† Laura Rossi,† Dmytro V. Byelov,† Sonja I.R. Castillo,† Anatoly Snigirev,‡ Irina Snigireva,‡ Albert P. Philipse,† and Andrei V. Petukhov† †

Van ’t Hoff Laboratory for Physical and Colloid Chemistry, Debye Institute for Nanomaterials Science, Utrecht University, The Netherlands ‡ European Synchrotron Radiation Facility, Grenoble, France S Supporting Information *

ABSTRACT: The vertical deposition technique for creating crystalline microstructures is applied for the first time to nonspherical colloids in the form of hollow silica cubes. Controlled deposition of the cubes results in large crystalline films with variable symmetry. The microstructures are characterized in detail with scanning electron microscopy and small-angle X-ray scattering. In single layers of cubes, distorted square to hexagonal ordered arrays are formed. For multilayered crystals, the intralayer ordering is predominantly hexagonal with a hollow site stacking, similar to that of the face centered cubic lattice for spheres. Additionally, a distorted square arrangement in the layers is also found to form under certain conditions. These crystalline films are promising for various applications such as photonic materials.

1. INTRODUCTION The vertical deposition (VD) technique is well-known for the convenient preparation of crystalline microstructures of spherical colloids with well-defined shape and control of the crystal structure and thickness.1−4 The technique is applied for novel functional nanomaterials, such as photonic materials, with a bottom-up fabrication approach.2−5 However, to the best of our knowledge the technique has never been applied to nonspherical colloids. Recently, a new anisotropic colloidal system of hollow silica cubes was developed, the cubes being well-defined in size and shape.6 Particle shape is known to greatly influence the final crystal symmetry of close packed structures, as was also shown by recent simulations7 and therefore, it is of fundamental and practical interest to study close-packed structures of nonspherical particles.7 In this article, the self-assembly of the colloidal silica cubes into multilayered crystals via the VD technique is studied. The VD technique is based on the self-assembly of colloidal spheres on a substrate, driven by capillary forces and convex flow, induced by fast solvent evaporation at the meniscus.18 Improvements of the technique have allowed to gain control of the speed of the meniscus movement and the use of large (∼1 μm) colloids by preventing sedimentation.4,9 The resulting crystals show a tendency toward FCC stacking, which has also inspired detailed studies of the assembly process.10−13 Moreover, the need to prepare perfect defect-free crystals for photonic applications resulted in the identification of different defect structures and their causes14−17 and has even led to control over their incorporation.18−23 With the VD technique large close packed films of colloidal cubes could be produced in a controllable way. Microstructures © XXXX American Chemical Society

based on colloidal cubes are of specific interest because cubic building blocks allow the formation of essentially space filling structures24 and of rare crystal symmetries such as the simple cubic (SC) lattice.6 That cubic building blocks can indeed form different lattices was recently shown by Dish et al. for truncated maghemite nanocubes. These cubes formed the SC lattice in 3D but in addition also the body centered tetragonal (BCT) lattice was found.25 However, these cubes were very small (8.5 nm) and have a low particle yield, which makes them difficult to apply in large scale applications. In this respect the novel colloidal silica cubes are promising, as they are micrometer sized and can be prepared in large amounts. In addition, because of their nonuniform shape and larger size their crystal structures could also be potential photonic materials.1,2 Furthermore, the cubes can be used to create patterned surfaces as alternative for etching. Rossi et al. have shown with optical microscopy that, already with depletion-attraction driven self-assembly, these silica cubes can form a SC lattice in 3D.6 However, the preparation of large (1 cm) crystalline films of the particles is difficult to achieve with depletion forces without the help of an aligning external field. With the VD method, we prepared crystalline films of the hollow silica cubes employing various growth setups. Detailed characterization of the formed microstructures was provided by electron microscopy and microradian X-ray diffraction measurements. The different arrangement of cubes in single layers and Received: February 17, 2012 Revised: April 19, 2012

A

dx.doi.org/10.1021/la3007052 | Langmuir XXXX, XXX, XXX−XXX

Langmuir

Article

multilayered structures were analyzed in detail. Furthermore, the specific type of stacking and the long-range order of the microstructures were characterized.

2. EXPERIMENTAL SECTION Experimental System. Precursor hematite (α-Fe2O3) particles were prepared from a condensed ferric hydroxide gel.26 First, 54 g of FeCl3·6H2O (Sigma) was dissolved in 100 mL of deionized water from a Milli-Q water system (which was used for all experiments) in a 250 mL Pyrex bottle. Subsequently, 90 mL of 6.2 M NaOH solution were slowly added (one drop every 2−3 s) to the solution using a dropping funnel while stirring. Next, the bottle was placed undisturbed in an oven for eight days at 100 °C. Subsequently, the formed hematite cubes were washed three times by centrifugation and redispersed in water. Silica-coated cubes were prepared by an adaptation of the Stöber method.6,27 For the silica coating 10 g polyvinylpyrrolidone (PVP, Mw = 40.000, Aldrich) was dissolved in 100 mL water and added to the solution to improve particle stability during silica coating. The excess PVP was removed via centrifugation and redispersion in ethanol. In a 2 L round-bottom flask, 83 mL of 6.2 wt % PVP-coated hematite cubes dispersion, 782 mL ethanol, 80 mL water, and 7 mL 3 wt % tetramethylammonium-hydroxide (TMAH, purum, Fluka) were mixed under mechanical stirring and sonication. The silica deposition was started by slow addition (one drop every four seconds) of a mixture of 32 mL tetraethyl ortho silicate (TEOS, purum, Aldrich) and 16 mL ethanol using a peristaltic pump. When the reaction was completed, a solution of 2.55 g PVP in 100 mL ethanol was added to improve the stability of the cubes. After 16 h the dispersion was centrifuged to remove the excess of PVP and unreacted TEOS molecules and dispersed in ethanol. To obtain hollow silica cubes, the hematite cores were dissolved by using a 18.5 wt % hydrochloric acid solution.6 The particles were transferred to water via centrifugation and 50 mL 37 wt % hydrochloric acid was added to 50 mL 10 wt % silica-coated cubes in water. The dispersion was stirred for 16 h, during which a color change from red to yellow was observed. The particles were centrifuged and redispersed in water. The procedure was repeated until the water reached a pH value of 4 after which the now white dispersion was transferred to ethanol. The cubes were kept in ethanol for storage, as nanoporous silica is known to slowly dissolve in water.28 The template colloidal hematite, α-Fe2O3, cubes have slightly rounded corners and their shape can be best described by that of a socalled superball;29

x a

m

+

y a

m

+

z a

Figure 1. TEM image of hollow silica cubes (scale bar is 500 nm) and schematic illustration of the cube parameters.

Table 1. Characteristics of the Precursor Hematite Cubes and Hollow Silica Cubes hematite silica

⟨D⟩ (nm)

σD (nm)

⟨L⟩ (nm)

σL (nm)

m

558 774

36 35

653 858

51 46

3.7 2.9

m

≤1

where a is half of the cube face side, D, and m is the deformation parameter, that indicates for how much the particle shape has deformed from that of a sphere, for which m = 2, toward that of a perfect cube, for which m → ∞. The average cube face side, ⟨D⟩, the face diagonal, ⟨L⟩, and the standard deviation, σ, were determined from transmission electron microscopy images as shown in Figure 1. With ⟨D⟩ and ⟨L⟩, the mvalue of the cubes was determined, which clearly decreases after the silica coating, see Table 1. Vertical Deposition Conditions. The cubes consist of thin hollow silica shells, which makes determination of the volume fraction from the weight concentration of the stock suspensions difficult. Therefore, the stock suspensions were centrifuged at 680 g in thin 4 × 0.2 mm capillaries. Assuming that the sediment has a random packing density of 72 v% based on their shape,30 the volume fraction was estimated from the obtained sediment volume. The desired volume fractions were achieved by diluting the stock dispersions with ethanol and/or water. For pure water samples, a stock suspension in water was made prior to the experiments. Small 2.5 mL vials and thin glass microscope slides (10 × 35 × 0.15 mm) were etched for 24 h in a 5 wt % KOH (Merck) aqueous solution in ethanol. Subsequently, the slides were rinsed thoroughly with water and dried in air. The vials were filled with 2.5 mL of cube suspension

Figure 2. A) Schematic illustration of the vertical deposition process. Cubes are transported to the meniscus by solvent evaporation and form a crystal. B) Schematic representation of the two different growth setups. The cube dispersion of a certain volume percentage, v%, with a substrate is placed in a heated oven at constant temperature, or in an oil-bath creating a temperature gradient along the dispersion vial. and a clean substrate was placed into the colloidal sol with an angle of 30° as seen in part A of Figure 2.10 Two different vertical depositionsetups were used, as shown in part B of Figure 2. For the first setup, the vial was placed in an oven at 50−70 °C for at least 24 h to allow the solvent to evaporate. In the second setup, a temperature gradient of about ∼5 °C was produced across the height (∼2.5 cm) of the vial by placing the bottom in an oil bath of 60 °C while keeping the area around the vial at room temperature. The latter method prevented sedimentation of the particles but also induced slow solvent evaporation over 24 h. These setups will be referred to as the ovensetup and ΔT-setup, respectively. Instrumentation. Scanning electron microscopy (SEM) was performed with a Phenom FEI microscope. High-resolution SEM (HR-SEM) images were obtained using a FEI XL30S FEG. A thin layer of platinum (usually 6 nm thick) was sputtered onto the samples B

dx.doi.org/10.1021/la3007052 | Langmuir XXXX, XXX, XXX−XXX

Langmuir

Article

prior to imaging. To reveal the thickness, the samples were scratched with a razor blade. By tilting the sample 40°, it was possible to image the edge of the films. Small angle X-ray scattering (SAXS) measurements were performed at the European Synchrotron Radiation Facility (ESRF) micro optics test bench at the ID06 beamline. An X-ray energy of 12.2 keV was selected by a silicon double crystal monochromator. To get highresolution diffraction patterns we used a microradian X-ray diffraction setup with compound refractive lenses (CRL).31−36 The CRL consisted of 8 beryllium parabolic lenses with radius of parabola apex of 300 μm and one lens with radius of parabola apex of 1000 μm.37 The CRL focus the X-rays at the phosphor screen of the CCD X-ray detector (Photonic Science, 9 × 9 μm2 pixel size and dimensions of 4008 × 2671 pixels) located at ∼9 m from the lens. The samples were placed behind the CRLs at 7.54 m before the detector. The beam size on the sample was around 400 × 400 μm2.

from blue-green (bottom) to red (top), is due to the reflection angle-change toward the camera. The nonuniform intensity of the scattering indicates the presence of small regions with different crystal orientation. A variation in film thickness along the growth direction (arrow Figure 3) can be observed, and can be explained by fluctuations in the evaporation speed of the solvent due to uncontrolled variation of the humidity. Furthermore, a temporary absence of particles 4 mm above the bottom of the film can be seen (marked area). This can be explained by a change in the meniscus shape of the solvent due to the closer proximity of the container wall toward the bottom of the substrate. The wetting of both the wall and the substrate will change the shape of the meniscus and cause inconsistent movement. This movement is a stick−slip motion, as has been previously reported for the films of spheres containing stripelike patterns.8,38 However, the humidity and air flux were not specifically controlled and are also known to cause this kind of effect. 2D Structure Characterization. The crystalline structure and thickness of deposited films were studied with SEM. Part A of Figure 4 shows a typical SEM image of a single layer of

3. RESULTS Visual Sample Appearance. The vertical deposition experiments performed, with both the oven- and ΔT-setup, resulted in thin films of deposited cubes. However, only films made from aqueous suspensions displayed Bragg reflections of visible light. The films prepared with the oven-setup from water had a length of only 2−5 mm, with most of the substrate uncovered. This indicates that the cubes sediment faster than the time needed for evaporation of the water. As a result, no cubes are left in the suspension to be transferred to the interface. Raising the oven temperature to 70 °C, which increases the evaporation speed of water and therefore shortens the overall time needed for deposition, did not greatly improve the film length or quality. In contrast, the ΔT-setup resulted in multilayered, completely substrate covering films. Clearly, the use of a temperature gradient during deposition is an effective method to prevent sedimentation of the large colloidal particles and allows the formation of large uniform films. Figure 3 shows a film prepared from a 0.5 v% dispersion in water using the ΔT-setup over a period of 24 h. The film covers the width and length of the substrate, up to the point where the VD was stopped, at 13 mm from the top. It displays strong Bragg reflections under white light illumination indicating crystalline ordering. The color-change of the reflected light,

Figure 4. A) SEM image of a single layer of assembled cubes of a sample made with the oven-setup. Different arrangements of cubes occur, ranging from B) a square-like to C) hexagonal-like. Scale bars are A) 5 μm and B, C) 1 μm.

assembled particles. The square shape of all cubes can be observed indicating that the cubes are lying flat on the substrate. The orientation of the cubes is not fixed by the growth direction but is correlated with that of neighboring cubes. Varying square-like to hexagonal-like arrangements can be seen, as indicated with the two circles, which extend over typically 3−4 cube lengths. In parts B and C of Figure 4 larger magnification are shown where the orientation are indicated with a square and hexagon, respectively. Evidently, the monodisperse cubes can be self-assembled with the vertical deposition technique into monolayers of square to hexagonally coordinated arrays, where the cube orientation is predetermined by substrate and neighboring cubes.

Figure 3. Photograph of the colloidal crystal film made with the ΔTsetup with light illumination under an angle showing Bragg reflections. The arrow indicates the growth direction. C

dx.doi.org/10.1021/la3007052 | Langmuir XXXX, XXX, XXX−XXX

Langmuir

Article

Figure 5. A, B) SEM images taken near the top of the film in part A of Figure 3 showing two different regions, with large (top) and small (bottom) cracks, B) higher magnification of the interface showing connected cubes with matrix at the top and separated cubes at the bottom. C, D) HR-SEM images of a cracked region with more infiltration and a fracture cross section of infiltrated region, showing the hollow-inside of the particles. Scale bars are A) 100 μm, B) 10 μm, C,D) 2 μm.

Figure 6. SEM image of multiple layered crystalline cube arrays of; A) an infiltrated region and B) a noninfiltrated region. For both images their corresponding Fourier transforms are shown in the insets, arrows indicate the growth direction during VD. C) Enlargement of a side of the layers showing the location of cubes on top of niche formed by neighboring cubes in the layer below. D) Fracture cross section of infiltrated region, showing the layered structure, hollow-inside and positioning of the cubes. Scale bars are A,B) 10 μm C,D) 5 μm.

was observed at 3.5 mm from the bottom of the film, where deposition had stopped momentarily (Figure 3). Samples prepared with the oven-setup also showed some infiltration at the start of the deposition, be it only for short regions and mainly at the highest temperature, 70 °C. Apparently, this matrix only formed at the onset of deposition. The parts of the sample with a matrix will be further referred to as infiltrated regions. Figure 6 shows SEM images taken at the top (1 mm) and center (5 mm) of the crystalline ΔT-sample, of the ordered top layer of an infiltrated crystal (part A of Figure 6) and a fracture through multilayered crystals (part B of Figure 6). The Fourier transform of the images (inserts in parts A and B of Figure 6) reveal that the local cube positions are predominantly distorted hexagonal and have the same orientation. In part C of Figure 6, the stacking of the cubes can be seen more clearly. Although the image is taken without tilting the sample, cubes in all layers are visible. Evidently, the lateral positions of the cubes in a layer are slightly shifted. Furthermore, the cubes are located in the niche between two neighboring rows of cubes below, and on top of the niche between two cubes in these rows. In part D of Figure 6, a straight fracture through an infiltrated part of the sample is shown, revealing the square hollow-inside of the cubes. Interestingly, directly beside the crack the hollow-inside of the cubes is only visible in the top and third layer of the

3D Structure Characterization. The ΔT-setup was found to produce large substrate covering films with multiple layers that showed very distinct Bragg reflections. Therefore, the sample prepared with the ΔT-setup, shown in Figure 3, was studied in more detail with both SEM and SAXS. A complete SEM scan of the sample showed that the crystal contained two distinctly different regions. In part A of Figure 5 a SEM image 2 mm from the top of the film is shown. The top of the image shows a lighter region, with large cracks running through it, and a darker area that contains smaller cracks. The scratch on the left was made with a razor blade to reveal the inner structure. Higher magnification images (part B of Figure 5) show that the different crack-size arises from a difference in the intercube voids, the origin of which will be discussed later. In the top part, the voids are filled with a connecting matrix, while in the bottom part separate cubes can be observed. In parts C and D of Figure 5 SEM images at ∼1 mm from the top of the film are shown, where the cubes are hardly visible due to large amount of matrix material. Because of this matrix, the cubes are strongly interconnected and when fractured, the structures will break along the path of least resistance, which is no longer along the cube edge (part C of Figure 5). Consequently, the hollowinside of the cubes can even be seen (part D of Figure 5) but also edges of cubes sticking out. The same infiltration behavior D

dx.doi.org/10.1021/la3007052 | Langmuir XXXX, XXX, XXX−XXX

Langmuir

Article

infiltrated region with a silica matrix is present, whereas in the middle part air is present between the cubes. See Supporting Information for more details. The inner q-values of the Bragg peaks are q = 8.01 μm−1 in the horizontal direction and q = 9.17 μm−1 for the two vertical peaks, with a ratio of 1.12. This corresponds to a hexagonaltype crystal with two lattice dimensions of 784 and 685 nm. The radial width of the peaks indicates that the crystal domain size extends over about 10 particle diameters.39 The azimuthal width of the peaks shows that the crystal grain orientation fluctuates but the hexagonal order is directed along the growth direction. The presence of the second order Bragg peaks indicates that the crystal ordering is quite well established but decay of P(q) and positional correlation fluctuations decrease the intensity of the higher orders. This is even further influenced by the number of layers of the crystal, which is slightly higher in the separate region, nine layers compared to seven layers. Besides the dominant distorted hexagonal ordering, the square arrangement of the cubes was also found to persist in multilayered structures and hence forming a cubic lattice. In Figure 8, a diffraction pattern of the second infiltrated region

structure, in contrast to the second and fourth layer, where the edge of the cubes are visible. Clearly, the cubes are located in a niche formed by three cubes in the layer below, a so-called hollow-site stacking, which is characteristic for hexagonally close packed (HCP) and face centered cubic (FCC) structures of spheres. For the SAXS measurements, the substrate was positioned perpendicular to the beam and multiple scans were performed, covering the complete area of the film. In Figure 7, two typical

Figure 8. SAXS pattern of the infiltrated cubic region at the bottom of the film. (Inset) SEM image of cubic ordering in the infiltrated region where the hollow cube inside is visible. Scale bar is 5 μm.

(about 3.5 mm above the film-bottom) is shown. The inset is a SEM image of the same region, close to a scratch. Here, the structure has been broken along the crystal layer parallel to the substrate, showing the cubic ordering of the hollow cube insides. The pattern shows distinct peaks in the horizontal direction at q = 8.78 μm−1 and in the vertical direction q = 9.42 μm−1, corresponding to a cubic structure with lattice distances of 716 and 667 nm, respectively. Compared to the first infiltrated region (part A of Figure 7), it is clear that a decrease of 70 nm in cube size has occurred. Closer analysis of peaks reveals that the angle between the right horizontal peak and the vertical peak is 85° in contrast to the expected angle of 90° for a square structure. Also the location of the horizontal peaks has rotated 5° with respect to the hexagonal pattern. The absence of second order peaks is due to the hollow cube form factor but also because the cubic regime does not possess true long-range order. Apparently, this is not the most favorable structure and

Figure 7. SAXS patterns of the ΔT-sample taken at the A) top, with infiltration matrix, and B) middle, with separate particles. The vertical growth direction is aligned top to bottom. The coloring is on a log scale to emphasize the scattering intensity at high q-values.

2D-SAXS scattering patterns of the infiltrated top (part A of Figure 7) and noninfiltrated middle (part B of Figure 7) regions of the ΔT-sample are shown. The dark rounded bar visible at the top is the shadow of the beam stop, which protects the CCD-camera from the direct X-ray beam. Both patterns show Bragg peaks at low q-values with hexagonal order, which are slightly elongated along the growth direction. At higher q-values ring-like reflections are visible that are attributed to the form factor P(q). Despite their difference, both pattern originate from the same ordered structure but the Bragg peaks are emphasized differently by P(q) as a result of the different intercube matrix. At the top part of the ΔT-sample, an E

dx.doi.org/10.1021/la3007052 | Langmuir XXXX, XXX, XXX−XXX

Langmuir

Article

distorted square arrangement. Jiao et al. also showed that for deformation parameters m = 2.6, which is very close to the one in our study m = 2.9, the hexagonal and square lattices are equally favorable. The typical periodic distances of the hexagonal and square arrangement can be calculated (part A of Figure 9). For the hexagonal arrangement, three periodic distances occur; one in the horizontal direction of D, and two in the vertical direction of √(4/5)D, as indicated by the lines through the cube centers. For the square arrangement, touching cube corners form a denser packing and result in a square structure with a 85° rotation, resulting in two different periodic distances. These calculated distances do not correspond exactly to the lattice values found with the SAXS patterns (Figures 7 and 8). However, good agreement is found for the hexagonal lattice when the ratio between the reciprocal Bragg peaks is calculated 1.12 ≈ √(5/4), which converts to √(4/5) in real space. For the cubic lattice, good agreement is found in that the SAXS data shows indeed two different periodic distances with a rotation angle of 85°. The main part of the ΔT-sample with multilayered structures consist of stacked hexagonal layers. In this arrangement the cubes have a so-called hollow-site stacking. Part B of Figure 9 shows a schematic model of how the stacking is build up for a second and third layer of cubes. For the second layer, there is no preferential position. However, for the third layer, the first possible location is above a niche in both layers and the second is located above a particle in the first layer. From the sideward stacking observed in part C of Figure 4, it is clear that the third layer is positioned above the first niche and not a particle in the first layer. In parts C and D of Figure 9, a 3D representation of the resulting structure is given. The specific location of the particles could be a result of growth at a particular crystal plane during the VD, which is known to occur for thin crystals of spheres, inducing FCC stacking. However, the growth mechanism of the specific FCC structure during the deposition of thick (>5 layers) crystals of spheres is still under investigation.11,12 So, to understand the formation mechanism of multilayered cube structures, more investigation is needed. Crystal Defects. In the formed crystals, different defects were observed, such as cracks, vacancies and imperfect crystal structures, which affect the crystal quality. The small cracks in the crystals are common defects observed in crystals created with VD.8 Because of tensile stresses during the drying step, the distances between two particles decreases and cracks will form. In the SEM images, additional shrinkage is caused by the exposure to electrons and the high vacuum in the SEM.8 Vacancies in the crystalline structure will occur when the evaporation speed of the solvent is faster than the influx of cubes. The fast solvent evaporation also causes the imperfect crystal structures as the cubes do not have enough time to move to the optimal lattice position before they are pinned down to the substrate. In order to gain more control over the formed structures and possibly prevent defects, the VD conditions should be optimized and well controlled. Crystal Infiltration. The silica matrix observed between the cubes at the top of the film, can be explained by the etching of silica by aqueous solutions at elevated temperatures: solid silica, SiO2, partly dissolves in water as silicic acid, Si(OH)4,28 which will precipitate in between the cubes during deposition, creating a SiO2 matrix embedding the cubes (Figure 4). With the ΔTsetup the cube dispersion is heated to 55 °C, which promotes silica dissolution. In addition, the deposition does not start

might be caused by the difference in wetting angle and silica infiltration, that occurred in this particular part of the sample.

4. DISCUSSION The crystalline structures of the cubes can be explained in terms of formation mechanism and shape. First, we will explain the formation and interpret the formed structures of single layers and multilayered structures. Then, we briefly address the observed defects in the structure and discuss their formation mechanism and the presence of the intercube matrix is explained and its advantages and disadvantages discussed. Finally, the crystal quality and optimization possibilities of the microstructures are discussed. Crystal Structure and Formation. The crystal symmetry is determined by the shape of the building blocks, the cubes, as well as the VD preparation method. In the VD method, the deposition starts at the thin solvent layer that is formed on the substrate. When the solvent layer is of the same size as the cubes, strong capillary forces will act upon the cubes, so-called immersion forces.8 The immersion forces will cause the flat orientation of the cubes on the substrate and induce long-range attraction between the cubes, resulting in dense packings. For single layers of cubes, distorted square to hexagonally coordinated arrays, with the latter being more dominant, were observed. Interestingly, these arrangements can be simply explained by the rounded square shape of the flat lying cubes. Jiao et al. showed that for superdisks, the 2D analogue of superballs, two different types of dense lattice packings can be formed.24 The Λ0-lattice, with hexagonal arrangement for more sphere-like shapes and the Λ1-lattice, with distorted square arrangement for shapes close to a square, as shown in part A of Figure 9. For the Λ1-lattice a denser packing is achieved by closing the gap between four rounded squares giving a slightly

Figure 9. A) Schematic model of arranged rounded squares (superballs in 3D) in a hexagonal arrangement and square arrangement, (comparable to the Λ0-lattice and Λ1-lattice as in ref 22), with their periodic lattice distances. B) Possible locations for the second (dots) and third (asterisk) layer in a hexagonal arrangement and C) representation of the formed 3D lattice. F

dx.doi.org/10.1021/la3007052 | Langmuir XXXX, XXX, XXX−XXX

Langmuir

Article

immediately when using the ΔT-setup, because, in contrast to the oven-setup, evaporation is slower due to the cooling at the surface in the ΔT-setup. This additional time allows the concentration of soluble silica to increase over a longer time period and explains the presence of large infiltration regions observed in samples made with the ΔT-setup. The fact that the infiltration is only located at the top of the film indicates that at a certain point the Si(OH)4 concentration has decreased such that the interparticle voids cannot be completely filled anymore. Apparently, the concentration of soluble silica is not constant over time. The concentration decrease can be caused by the deposition in the form of SiO2 between the cubes during the deposition of the cubes, which also means a decrease in the source of the Si(OH)4. Dissolution of the particles is on the one hand nondesirable because when the cubes are etched they become smaller and eventually will lose their shape. However, a silica matrix also has advantages. First, it stabilizes the interparticle interactions by ‘glueing’ particles together preventing fracture of the formed structure. The presented mechanism has already been employed for the improvement of photonic crystals of spheres fabricated with the vertical deposition technique.19,21 Second, the silica matrix also seems to be the cause of the stabilization of the cubic phase. Third, the presence of the matrix resulted in different form factor contributions and made it easier to determine the structure factor. Crystal Quality. The presented research shows that for the specific superball shape, m = 2.9, it is hard to selectively get one of the two arrangements, hexagonal or square, over the other in a single layer. On the basis of the findings of Jiao et al., the square arrangement should be more dominant for m > 2.6.24 However, in the ΔT-sample with mainly multilayered structures the hexagonal stacking is found to dominate over the square stacking. This is undesirable for potential applications, such as photonics, where large single crystalline structures are needed. At the moment, it is not clear why the two different crystalline arrangement are found in different parts of the sample. However, it is expected to be the result of the VD, where the self-assembly does not occur under ideal equilibrium conditions. The self-assembly is instead influenced by many factors, that are all connected, such as evaporation speed and solvent temperature.14 Furthermore, it is unclear how the multilayered structure are exactly formed. Thus, to understand the complete self-assembly and to gain more control over the resulting arrangement, more research is needed. For spheres, a number of different approaches exist, which we will try to apply in future research.3,18−23 Another way to obtain a dominant square arrangement is to use cubes with a larger m-value.24 In addition, with a silica matrix, a completely substrate covering membrane, without voids other than the hollow particle inside, can be obtained.21 This is of importance as it allows the use of these slightly rounded cubes for the preparation of void- and defect-free crystals.

ments, as expected for dense packings of rounded squares. In multilayered stacking of hexagonal arrangement, hollow-site stacking was found, similar to that of FCC for spheres. Silica infiltration occurred at the start of deposition because silica partly dissolved in water and formed an intraparticle matrix upon drying, connecting the cubes in certain regions. The preparation of large crystalline films of colloidal cubes with VD opens up a new route for the fabrication of novel microstructures. Besides the cubic shape of the colloids, also the VD mechanism determines the final crystal structure in multilayered structures by predetermining the particle orientation and growth of the crystals at a specific crystal front. The large size of the cubes also allows optical microscopy studies of their formation mechanism in 3D, in contrast to structure formations of (magnetic) nanocubes.25,40,41 Furthermore, because of their nonuniform shape and micrometer size, the cubes are of potential interest as photonic materials. In future work, we will study the effect of the deformation parameter m on the obtained arrangements. We expect that larger deformation parameters will induce a more square-like lattice. Furthermore, we will investigate which conditions are needed for obtaining one lattice structure over the other, when m ∼ 2.6.



ASSOCIATED CONTENT

S Supporting Information *

Detailed explanation of the form factor effect in the diffraction patterns. This material is available free of charge via the Internet at http://pubs.acs.org.



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS



REFERENCES

We would like to thank the European Synchrotron Radiation Facility (ESRF) in Grenoble for the provided beam-time. C. Detlefs and T. Roth from ID06 are thanked for the support during the measurements. We would like to thank J. Hilhorst for many fruitful discussions. J.D. Meeldijk and C.T.W.M. Schneijdenberg are thanked for their assistance with the electron microscopes.

(1) Jiang, P.; Bertone, J. F.; Hwang, K. S.; Colvin, V. L. Single-crystal colloidal multilayers of controlled thickness. Chem. Mater. 1999, 11, 2132−2140. (2) Vlasov, Y.; Bo, X. Z.; Sturm, J. C.; Norris, D. J. On-chip natural assembly of silicon photonic bandgap crystals. Nature 2001, 414, 289− 293. (3) Zhou, Z.; Zhao, X. S. Flow-controlled vertical deposition method for the fabrication of photonic crystals. Langmuir 2004, 20, 1524−6. (4) Zhang, J.; Sun, Z.; Yang, B. Self-assembly of photonic crystals from polymer colloids. Curr. Opin. Colloid Interface Sci. 2009, 14, 103− 114. (5) Blanco, A. C.,E.; Grabtchak, S.; Ibisate, M.; John, S.; Leonard, S.; Lopez, C.; Meseguer, F.; Miguez, H.; Mondia, J.; Ozin, G.; Toader, O.; van Driel, H. M. Large-scale synthesis of a silicon photonic crystal with a complete three-dimensional bandgap near 1.5 micrometres. Nature 2000, 405, 437−40.

5. CONCLUSIONS Using the vertical deposition (VD) technique, large films of ordered monodisperse hollow colloidal cubes can be prepared in a controlled way. The thickness and areas of the films can be dictated by cube volume fraction, solvent flux, and temperature. During the deposition, capillary forces induce the self-assembly and cube orientation is predetermined by the flat substrate and neighboring cubes. The resulting monolayer structures are ordered arrays of cubes with square and hexagonal arrangeG

dx.doi.org/10.1021/la3007052 | Langmuir XXXX, XXX, XXX−XXX

Langmuir

Article

(6) Rossi, L.; Sacanna, S.; Irvine, W. T. M.; Chaikin, P. M.; Pine, D. J.; Philipse, A. P. Cubic crystals from cubic colloids. Soft Matter 2011, 7, 4139−4139. (7) Glotzer, S. C.; Solomon, M. J. Anisotropy of building blocks and their assembly into complex structures. Nat. Mater. 2007, 6, 557−562. (8) Denkov, N.; Velev, O.; Kralchevski, P.; Ivanov, I.; Yoshimura, H.; Nagayama, K. Mechanism of formation of two-dimensional crystals from latex particles on substrates. Langmuir 1992, 8, 3183−3190. (9) Zhou, Z.; Zhao, X. S. Opal and inverse opal fabricated with a flow-controlled vertical deposition method. Langmuir 2005, 21, 4717− 4723. (10) Im, S. H.; Kim, M. H.; Park, O. O. Thickness control of colloidal crystals with a substrate dipped at a tilted angle into a colloidal suspension. Chem. Mater. 2003, 15, 1797−1802. (11) Meng, L.; Wei, H.; Nagel, A.; Wiley, B. J.; Scriven, L. E.; Norris, D. J. The role of thickness transitions in convective assembly. Nano Lett. 2006, 6, 2249−2253. (12) Brewer, D. D.; Allen, J.; Miller, M. R.; Santos, J. M. D.; Kumar, S.; Norris, D. J.; Tsapatsis, M.; Scriven, L. E. Mechanistic principles of colloidal crystal growth by evaporation-induced convective steering. Langmuir 2008, 24, 13683−13693. (13) Cao, H.; Lan, D.; Wang, Y.; Volinsky, A.; Duan, L.; Jiang, H. Fracture of colloidal single-crystal films fabricated by controlled vertical drying deposition. Phys. Rev. E 2010, 82, 1−6. (14) Ye, Y.; LeBlanc, F.; Hache, A.; Truong, V. Self-assembling threedimensional colloidal photonic crystal structure with high crystalline quality. Appl. Phys. Lett. 2001, 78, 52−52. (15) Hartsuiker, A.; Vos, W. L. Structural properties of opals grown with vertical controlled drying. Langmuir 2008, 24, 4670−4675. (16) Hilhorst, J.; Abramova, V. V.; Sinitskii, A.; Sapoletova, N. A.; Napolskii, K. S.; Eliseev, A. A.; Byelov, D. V.; Grigoryeva, N. A.; Vasilieva, A. V.; Bouwman, W. G.; Kvashnina, K.; Snigirev, A.; Grigoriev, S. V.; Petukhov, A. V. Double stacking faults in convectively assembled crystals of colloidal spheres. Langmuir 2009, 25, 10408−12. (17) Born, P.; Blum, S.; Munoz, A.; Kraus, T. Role of the meniscus shape in large-area convective particle assembly. Langmuir 2011, 27, 8621−8633. (18) Kuai, S.; Hu, X.; Hache, a.; Truong, V. High-quality colloidal photonic crystals obtained by optimizing growth parameters in a vertical deposition technique. J. Cryst. Growth 2004, 267, 317−324. (19) Wang, L.; Zhao, X. S. Fabrication of crack-free colloidal crystals using a modified vertical deposition method. J. Phys. Chem. C 2007, 111, 8538−8542. (20) Khunsin, W.; Kocher, G.; Romanov, S. G.; Torres, C. M. S. Quantitative analysis of lattice ordering in thin film opal-based photonic crystals. Adv. Funct. Mater. 2008, 18, 2471−2479. (21) Hatton, B.; Mishchenko, L.; Davis, S.; Sandhage, K. H.; Aizenberg, J. Assembly of large-area, highly ordered, crack-free inverse opal films. Proc. Natl. Acad. Sci. U.S.A. 2010, 107, 10354−10359. (22) Napolskii, K. S.; Sapoletova, N. A.; Gorozhankin, D. F.; Eliseev, A. A.; Chernyshov, D. Y.; Byelov, D. V.; Grigoryeva, N. A.; Mistonov, A. A.; Bouwman, W. G.; Kvashnina, K. O.; Lukashin, A. V.; Snigirev, A. A.; Vassilieva, A. V.; Grigoriev, S. V.; Petukhov, A. V. Fabrication of artificial opals by electric-field-assisted vertical deposition. Langmuir 2010, 26, 2346−2351. (23) Ng, E. C. H.; Chin, K. M.; Wong, C. C. Controlling inplane orientation of a monolayer colloidal crystal by meniscus pinning. Langmuir 2011, 27, 2244−2249. (24) Jiao, Y.; Stillinger, F. H.; Torquato, S. Optimal packings of superdisks and the role of symmetry. Phys. Rev. Lett. 2008, 100, 245504. (25) Disch, S.; Wetterskog, E.; Hermann, P.; Salazar-Alvarez, G.; Busch, P.; Br, T.; Bergstr, L. Shape Induced Symmetry in SelfAssembled Mesocrystals of Iron Oxide Nanocubes. Nano Lett. 2011, 11, 1651−1656. (26) Sugimoto, T.; Khan, M. M.; Muramatsu, A. Preparation of monodisperse peanut-type alpha-Fe2O3 particles from condensed ferric hydroxide gel. Colloids Surf. Physicochem. Eng. Aspects 1993, 70, 167−169.

(27) Graf, C.; Vossen, D. L. J.; Imhof, A.; van Blaaderen, A. A general method to coat colloidal particles with silica. Langmuir 2003, 19, 6693−6700. (28) Alexander, G. B.; Heston, W. M.; Iler, R. K. The solubility of amorphous silica in water. J. Phys. Chem. 1954, 6, 453−455. (29) Jiao, Y.; Stillinger, F.; Torquato, S. Optimal packings of superballs. Phys. Rev. E 2009, 79, 1−12. (30) Jiao, Y.; Stillinger, F. H.; Torquato, S. Distinctive features arising in maximally random jammed packings of superballs. Phys. Rev. E 2010, 81, 1−8. (31) Snigirev, A.; Kohn, V.; Snigireva, I.; Lengeler, B. A compound refractive lens for focusing high-energy X-rays. Nature 1996, 384, 49− 51. (32) Kohn, V.; Snigireva, I.; Snigirev, A. Diffraction.theory of imaging with X-ray compound refractive lens. Opt. Commun. 2003, 216, 247− 260. (33) Drakopoulos, M.; Snigirev, A.; Snigireva, I.; Schilling, J. X-ray high-resolution diffraction using refractive lenses. Appl. Phys. Lett. 2005, 86, 014102. (34) Petukhov, A.; Thijssen, J.; t Hart, D.; Imhof, A.; van Blaaderen, A.; Dolbnya, I.; Snigirev, A.; Moussaid, A.; Snigireva, I. Microradian Xray diffraction in colloidal photonic crystals. J. Appl. Crystallogr. 2006, 39, 137−144. (35) Bosak, A.; Snigireva, I.; Napolskii, K. S.; Snigirev, A. Highresolution transmission x-ray microscopy: A new tool for mesoscopic materials. Adv. Mater. 2010, 22, 3256−3259. (36) Thijssen, J. H. J.; Petukhov, A. V.; ’t Hart, D. C.; Imhof, A.; van der Werf, C. H. M.; Schropp, R. E. I.; van Blaaderen, A. Characterization of photonic colloidal single crystals by microradian X-ray diffraction. Adv. Mater. 2006, 18, 1662−1666. (37) Lengeler, B.; Schroer, C.; Tummler, J.; Benner, B.; Richwin, M.; Snigirev, A.; Snigireva, I.; Drakopoulos, M. Imaging by parabolic refractive lenses in the hard X-ray range. J. Synchrotron Radiat. 1999, 6, 1153−1167. (38) Marqués-Hueso, J.; Schöpe, H. J. Regular horizontal patterning on colloidal crystals produced by vertical deposition. Prog. Colloid Polym. Sci. 2008, 48−56. (39) Guinier, A. In X-Ray Diffraction in Crystal, Imperfect Crystals, and Amorphous Bodies; Dover: New York; 1994. (40) Chen, M.; Kim, J.; Liu, J. P.; Fan, H. Y.; Sun, S. H. Synthesis of FePt nanocubes and their oriented self-assembly. J. Am. Chem. Soc. 2006, 128, 7132−7133. (41) Ahniyaz, A.; Sakamoto, Y.; Bergstrom, L. Magnetic field-induced assembly of oriented superlattices from maghemite nanocubes. Proc. Natl. Acad. Sci. U.S.A. 2007, 104, 17570−17574.

H

dx.doi.org/10.1021/la3007052 | Langmuir XXXX, XXX, XXX−XXX