Distributed Bragg reflectors from colloidal trilayer flake solutions

Distributed Bragg reflectors (DBRs) are one-dimensional photonic crystals with a periodic alternation of high and low refractive index materials, typically in a thin film geometry.^1,2 The coherent light scattering through such a multilayer can lead to highly reflective structures with optical properties determined by the refractive index and the thickness of the layers.³ Optical multilayered structures are widely encountered in the natural world^4,5 and are, for example, responsible for the brightly colored displays of several species of butterflies^6,7 and beetles,^8–10 and even in the fruits of some plant species.^11,12 In technological applications, DBRs are widely employed as anti-reflective coatings, narrowband filters,² and light emission and detection technologies.¹³ In general, the manufacture of such multilayered structures is done through the sequential coating of thin solid films using vacuum deposition techniques.

Over the past decades, often inspired by the structures found in nature,¹⁴ alternative, solution-based, processing techniques involving polymeric and inorganic materials have seen important progress, enabling low-cost fabrication methods. A particular focus was on the exploration of stimuli-responsive DBRs in a variety of sensing platforms, enabled by the capability of manufacturing refractive-index changing porous layers or using volume-changing materials upon changes in the surrounding environment.^15,16 Several innovative approaches to produce such stimuli-responsive DBRs have been reported. The combination of polymer layers exploiting their swelling response upon permeation of vapor molecules has been demonstrated both in planar and fiber-based geometries.^17,18 Periodic porous multilayers have been demonstrated through sequential, layer-by-layer spin-coating of colloidal suspensions of nanoparticles, where the size of the nanoparticles directly determines the pore size.¹⁹ Sol–gel chemistries based on metal-oxides have also been widely used,²⁰ either in combination with a nanoparticle layer approach²¹ or by incorporating self-assembling block copolymers (BCPs) into the metal-oxide chemistry.^22,23 Despite their high versatility for large area coverage, such solution based methods are not necessarily suited for small-area applications where site-selective integration through printing or lithographic patterning steps is required. In addition, sol–gel processes require high temperatures, which poses limitations on the type of substrates onto which they can be processed and also challenges in achieving defect-free stacks when a large number of layers (i.e., annealing steps) are required.^23,24

Here, we report the rapid and on-demand fabrication of DBRs from a colloidal solution of composite trilayer flakes obtained by combining a sol–gel fabrication approach with an etch-release step. The fabrication concept is schematically summarized in Fig. 1(a). As a first step, a sacrificial layer is patterned on a substrate [Fig. 1(a-i)], followed by the deposition and annealing of a seed trilayer composed of high and low refractive index materials [Fig. 1(a-ii)]. An etch-release step is then carried out to selectively remove the sacrificial layer [Fig. 1(a-iii)], and finally, the colloidal flake solution is collected and transferred, by means of drop-casting, onto an arbitrary substrate [Fig. 1(a-iv)], where a DBR is readily formed by the stacking of the trilayers upon drying of the solvent.

A similar flake release concept was previously reported with a focus on solar cell applications.²⁵ The method presented here differs significantly as it only requires the deposition of a seed trilayer, thereby minimizing layer defects, while at the same time providing a better control of flake geometry by prepatterning of the sacrificial layer. Furthermore, this approach significantly reduces the time required to build a DBR with a large number of layers. Both compact and mesoporous DBRs are demonstrated, with the latter showing sensitivity to environmental changes through spectral band shifts as a function of the surrounding refractive index medium. Such spectral shifts are comparable to inorganic DBRs manufactured by the conventional multi-layer spin-coating method.^22,23 This approach should be fully compatible with a variety of existing assembly techniques, i.e., droplet printing or microfluidic applications, where a DBR structure can be deterministically integrated at specific locations of a microfluidic chip.^26,27 The on-demand assembly of highly reflective DBRs could also be easily integrated in fiber-end sensors²⁸ or together with colloidal based light-emitters for enhanced light–matter interaction applications.^29,30

In the first fabrication step, sacrificial germanium (Ge) patterns were defined onto a silicon (Si) substrate by photolithography (Heidelberg, MLA 150) and lift-off of a physical vapor deposited 20 nm Ge layer (Leybold Optics, LAB600H). Figure 1(b) shows an optical microscopy image of the hexagonal patterns used throughout this work with a diameter of 200 μm. Next, a sol–gel derived compact trilayer (TiO₂–SiO₂–TiO₂) was deposited onto the Ge patterned substrate. For this step, a stock titania precursor solution was prepared by adding 0.371 ml of titanium(IV) isopropoxide (TTIP, 97%, Sigma Aldrich) under strong stirring to a solution containing 0.120 ml of hydrochloric acid (HCl, ACS reagent, 37%, Sigma Aldrich) in 1.5 ml of ethanol (98%, Sigma-Aldrich). The mixture was stirred for one hour before use. A silica precursor solution was prepared by adding 0.372 ml of tetraethyl orthosilicate (TEOS, reagent grade, 98%, Sigma Aldrich) to a mixture of 0.160 ml of HCl and 1 ml of ethanol. Thin films of controlled thickness of both materials were obtained by spin-coating of the solution onto the patterned substrate with varying spin-coating speeds. After each deposition, the samples were annealed following a protocol established previously,²³ followed by the deposition of the next layer, yielding compact dielectric layers with refractive index values of 2.09 and 1.43 for TiO₂ and SiO₂, respectively, as measured by optical ellipsometry (alpha-SE, J. A. Woollam) on reference samples with single layers on Si. Figure 1(c) shows an optical image after the final TiO₂ deposition.

For the deposition of mesoporous layers (mp-TiO₂ and mp-SiO₂), we relied on the self-assembly of a polyisoprene-block-poly(ethylene oxide) (PI-b-PEO) BCP, added to the sol–gel solutions with a mixing ratio of 1:4:1 (TTIP/TEOS:HCl:BCP). After the spin-coating and annealing steps, a final calcination step at 600 °C crystallizes the inorganic matter and burns out all the organic components, leaving behind a mesoporous layer.²³ The refractive indices of the mesoporous layers were 1.48 and 1.17 for the mp-TiO₂ and mp-SiO₂, respectively.

During annealing [step ii of Fig. 1(a)], the sacrificial Ge layer oxidizes and thus enables a convenient etch-release step as GeO_x dissolves in water.³¹ The patterned substrates with the coated trilayers were therefore immersed in water to etch-release the trilayers. This etch-release step can be followed under an optical microscope [Fig. 1(d)]. Thin film interference is observed at the hexagons as water permeates from the edges and starts dissolving the underlying GeO_x layer (yellow circle). Once the etch is complete, the residual stress in the trilayers is sufficient to release the flakes from the substrate and disperse them in the water solution (red circle). The released flakes can then be collected, thus obtaining a colloidal suspension of the trilayered flakes. Drop casting the colloidal flake suspension onto a different substrate and subsequently allowing them to dry result in an assembly of the flakes into DBR stacks of a varying number of layers, as seen in the optical image in Fig. 1(e). As can be observed, the areas with DBR assemblies are limited to the flake overlap regions, which alludes to one of the main challenges of this technique (i.e., obtaining fully overlapping flakes). The latter could be further optimized through the interplay between the flake concentration, droplet size, and solvent evaporation rate. To confirm the stacking quality, cross sections of the assembled DBRs were imaged using a dual-beam focused ion-beam SEM (FEI, Scios 2). Figure 1(f) shows a cross section of a DBR region with five stacked trilayers. The layers with a darker contrast correspond to the core SiO₂ layer, while the lighter contrast corresponds to the outer TiO₂ layers.

To optically characterize the DBRs assembled from the colloidal flake solutions, the reflectance of stacks with an increasing number of flakes was measured using an adapted bright field microscope (Zeiss, Axio Scope.A1) with a 20× objective (Zeiss, EC Epiplan-Apochromat 20×/NA 0.6). The reflected light was confocally collected in the image plane with an optical fiber (QP50-2-UV-BX, 50 μm core) connected to a diode spectrometer (Ocean Optics, Maya-2000-LSL). Figure 2 shows the reflectance of stacked compact TiO₂–SiO₂–TiO₂ trilayer flakes, with TiO₂ and SiO₂ thicknesses of 49 ± 2 nm and 88 ± 2 nm, respectively, assembled on a glass substrate. The spectra show that the reflectance of the DBR increases with increasing stack thickness. A stack with four flakes (purple line) already surpasses a reflectance of 90%, and near-unity values are reached for six stacked flakes (light-blue line), within a ∼190 nm wide reflectance band centered at 680 nm.

The process can be easily adjusted for the fabrication of DBRs of different color. For a fixed choice of materials (i.e., fixed refractive indices), spectral tuning in a DBR structure is achieved by changing the thickness of the layers. To demonstrate this, we fabricated samples where the protocol for the outer compact TiO₂ layers was kept fixed (thickness of 40 ± 5 nm), while the thickness of the compact SiO₂ core was varied by adjusting the spin-coating speed. This resulted in three different flake samples with SiO₂ core thicknesses of 96 ± 3 nm, 111 ± 3 nm, and 136 ± 2 nm, respectively, as determined from FIB-SEM cross sections. Figure 3 shows the reflectance spectra for four-trilayer stacks of each flake type. As the thickness of the SiO₂ core increases, a clear red-shift of the main reflectance peak is observed. All reflectance spectra surpass 80% reflectance at the peak wavelength. Note that the three types of DBR stacks were assembled on the same glass substrate, which illustrates the versatility of this approach to fabricate, on demand, an arbitrary number and types of small-scale DBR sensors on a chip, a feature that could be compatible with multiplexed optical sensor applications.³² 

Porous photonic materials are ideal gas and vapor/liquid sensors, as changes in the refractive index occur through pore permeation resulting in a measurable spectral band shift.³³ The employed sol–gel approach provides flexibility to fabricate either compact or mesoporous layers. By incorporating BCPs into the synthesis protocol, both porous TiO₂ and SiO₂ can be obtained. Following the same etch-release approach, we obtained a colloidal solution of mesoporous trilayer flakes, where the thickness of the outer mp-TiO₂ was 52 ± 2 nm, and the mp-SiO₂ core layer had a thickness of 115 ± 2 nm.

To demonstrate the refractive index sensing ability of DBRs fabricated from porous trilayer flakes deposited on a glass substrate, the reflectance spectra of the porous stacks were recorded as a function of the surrounding medium (air and water). Figure 4 shows the reflectance in both media for varying numbers of stacked flakes. In air [Fig. 4(a)], the main reflection band is centered at ≈564 nm. As the sample is immersed into water [Fig. 4(b)], the reflectance of the stacked flakes is reduced and red-shifted by ≈96 nm for the case of the four-trilayer stack [dark red line in Fig. 4(b)]. This spectral shift corresponds to a sensitivity of ≈291 nm per refractive index unit (RIU), which is comparable to previously reported values for mesoporous DBRs fabricated in a layer-by-layer fashion.^22,23

Finally, transfer matrix simulations were performed to elucidate the large spectral shift caused by the immersion of the mesoporous stacks in water. Supplementary material, Fig. 1, shows that the wavelength shift is caused by the complete imbibition of water into the pores. The supplementary material, Fig. 2, shows similar predictions for alcohols and toluene.

This study demonstrates a method for the assembly of highly reflective inorganic DBRs from colloidal trilayer flake solutions. Our approach involves the solution processed deposition of trilayer seed flakes (i.e., TiO₂–SiO₂–TiO₂) onto a pre-patterned sacrificial layer followed by etch-release into water to obtain a colloidal flake solution. From this solution, the flakes are then transferred onto arbitrary substrates. This approach significantly shortens the manufacturing time of small-scale DBRs²³ to the time required to produce three layers, remove them from the substrate, and redeposit them at a substrate of choice.

To highlight the versatility of this approach, several DBR assemblies using both compact and mesoporous sol–gel processed thin films are demonstrated, using a trilayer design consisting of TiO₂ as the outer high refractive index material and SiO₂ as the core material with a low refractive index. Stacked DBR assemblies reached near unity reflectance values in the visible spectrum, and the peak reflectance could be easily tuned by varying the thickness of the SiO₂ core (Figs. 2 and 3). The possibility to assemble stimuli-responsive layers is demonstrated by introducing porosity into the trilayer flakes. Using mesoporous materials, a refractive index sensitivity of 291 nm/RIU was achieved, comparable to previous reports using a more time consuming layer-by-layer process.^22,23

This approach is not restricted to sol–gel processed films, and trilayer fabrication can be generalized to standard vacuum coating techniques and other spin- and spray-coating-based approaches. Upscaling the production of the colloidal suspensions should be possible, given its compatibility with similar commercially produced structural “effect pigments,” which employ thick dielectric flakes (typically mica or silica) with a coating of a high refractive index material, such as TiO₂ or Fe₂O₃.^34,35

A further development of this approach could include strategies to create self-assembled DBR stacks already in solution, for example, by using water soluble polyelectrolytes, such as positively charged poly(diallyldimethylammonium chloride) and negatively charged poly(styrenesulfonate). The mixing of opposite charged particles should lead to their self-assembly in the suspension media before deposition onto a substrate.³⁶ Alternatively, by adding much smaller colloidal particles to the suspension, depletion interactions could be employed to achieve the stacking of neutral flakes in solution.³⁷ In a further approach, flake-stacking can be controlled by limiting the areas in which the flake suspensions are deposited, for example, by combining inkjet printing and surface functionalization techniques.^38,39

Despite the current limitation to small areas on substrates, the ability to deposit DBRs from a colloidal flake solution makes this approach compatible with spray coating and printing techniques.⁴⁰ The latter provides an opportunity to form DBRs on substrates that are not suitable for standard layer-by-layer deposition methods, for example, non-planar surfaces. We envision that this approach could pave the way for the on-demand integration of stimuli-responsive DBRs onto fiber-end sensors²⁸ and microfluidic sensing platforms, where multiplexing may be possible by integrating different DBRs on the same chip, each with its own spectral properties or sensing affinities.³² The on-demand assembly of highly reflective DBRs may also be useful when integrated with colloid-based light-emitters for enhanced light–matter interactions.^29,30

Supplementary material, Figs. 1 and 2: simulations on the sensing response of mesoporous multilayer stacks.

M.M. and E.B.-U. contributed equally to this work.

This research was supported by the Swiss National Science Foundation through Grant No. 168223 (B.D.W.) and the National Center of Competence in Research Bio-Inspired Materials as well as the Adolphe Merkle Foundation. This project has received funding from the European Union’s Horizon 2020 research and innovation program under the Marie Skłodowska-Curie grant, Agreement No. 741855 (E.B.-U). We thank the Center for MicroNanotechnology (CMi) of the Ècole Polytechnique Federal Lausanne (EPFL) for technical support and access to their cleanroom facility.

The data that support the findings of this study are available from the corresponding author upon reasonable request.