This is an open access article published under an ACS AuthorChoice License, which permits copying and redistribution of the article or any adaptations for non-commercial purposes. Research Article Cite This: ACS Appl. Mater. Interfaces 2019, 11, 11763−11771 www.acsami.org Solvent-Assisted Self-Assembly of Gold Nanorods into Hierarchically Organized Plasmonic Mesostructures Christoph Hanske,*,† Eric H. Hill,†,∥ David Vila-Liarte,† Guillermo Gonzaĺ ez-Rubio,† Cristiano Matricardi,‡ Agustín Mihi,‡ and Luis M. Liz-Marzań *,†,§ † CIC biomaGUNE and Ciber-BBN, Paseo de Miramón 182, 20014 Donostia−San Sebastián, Spain ‡ Institut de Ciència de Materials de Barcelona (ICMAB-CSIC), Campus de la UAB, 08193 Bellaterra, Catalonia, Spain § Ikerbasque, Basque Foundation of Science, 48013 Bilbao, Spain * S Supporting Information ABSTRACT: Plasmonic supercrystals and periodically struc- tured arrays comprise a class of materials with unique optical properties that result from the interplay of plasmon resonances, as well as near- and far-field coupling. Controlled synthesis of such hierarchical structures remains a fundamental challenge, as it demands strict control over the assembly morphology, array size, lateral spacing, and macroscale homogeneity. Current fabrication approaches involve compli- cated multistep procedures lacking scalability and reproduci- bility, which has hindered the practical application of plasmonic supercrystal arrays. Herein, these challenges are addressed by adding an organic solvent to achieve kinetic control over the template-assisted colloidal assembly of nanoparticles from aqueous dispersion. This method yields highly regular periodic arrays, with feature sizes ranging from less than 200 nm up to tens of microns. A combined experimental/ computational approach reveals that the underlying mechanism is a combination of the removal of interfacial surfactant micelles from the particle interface and altered capillary flows. Assessing the efficacy of such square arrays for surface-enhanced Raman scattering spectroscopy, we find that a decrease of the lattice periodicity from 750 nm down to 400 nm boosts the signal by more than an order of magnitude, thereby enabling sensitive detection of analytes, such as the bacterial quorum sensing molecule pyocyanin, even in complex biological media. KEYWORDS: template-assisted self-assembly, gold nanorods, SERS, lattice plasmon, molecular dynamics simulation ■ INTRODUCTION Due to their strong interaction with electromagnetic radiation, stability against oxidation grants easy fabrication of clean surfaces that can get into direct contact with analyte molecules. microarrays of plasmonic nanoparticles show great promise as For biosensing in particular, gold displays further advantages components for high-performance optical sensors, light due to its low cytotoxicity, which can be problematic when managing layers in thin-film solar cells, novel color displays, using other plasmonic nanomaterials such as silver or copper, and other applications.1−6 In this context, tight control over for measurements in complex media or even in living systems. the array dimensions, internal structure, and overall macro- Among the many accessible shapes, anisotropic nanoparticles scopic homogeneity are crucial. The fabrication of plasmonic such as gold nanorods (AuNRs) excel due to the mesostructures (i.e., structures with feature sizes ranging from straightforward tunability of their localized surface plasmon 100 nm to microns) with controlled properties still poses a resonances via chemical adjustment of the nanoparticles’ significant challenge, as it requires bridging vastly different aspect ratio.11,12 This geometrical control allows precise length scales, typically ranging from few nanometers up to tailoring of the desired excitation wavelength and even several millimeters, for technological feasibility.7 achieving optical monodispersity.13 Further signal enhance- Arrangements of noble-metal nanoparticles are the most ment, needed for chemical ultradetection or even single common class of plasmonic nanomaterials for the fabrication molecule detection, requires optimization of near-field and far- of optical sensors, employed, for example, in surface-enhanced field coupling effects to boost light−matter interactions. A Raman scattering (SERS) or surface-enhanced infrared strong near-field coupling relies on interparticle distances of absorption (SEIRA) spectroscopy.3,8 These applications rely heavily on strong light−matter interactions in the resonant Received: January 7, 2019 noble-metal structure carrying the analyte to be detected.3,9,10 Accepted: March 7, 2019 Gold nanoparticles are often the material of choice, as their Published: March 7, 2019 © 2019 American Chemical Society 11763 DOI: 10.1021/acsami.9b00334 ACS Appl. Mater. Interfaces 2019, 11, 11763−11771 ACS Applied Materials & Interfaces Research Article 1−2 nm,14−18 which in practice rules out purely top-down between 1 and 50 μm for assembly from aqueous approaches such as e-beam lithography or focused ion beam dispersions.27 In this work, much smaller, hierarchically milling, which are just about to reach gap dimensions below 5 structured plasmonic arrays with feature sizes ranging from nm.19 Far-field coupling, on the other hand, depends on 200 nm up to 10 μm and periodicities from 400 nm to 20 μm diffraction and requires periodic arrays with close-to-perfect are realized by a solvent-assisted approach. Through systematic registration over extended areas.20 As a result, template- optimization of solvent composition and particle and surfactant assisted self-assembly of noble-metal colloids is the currently concentrations, as well as a detailed investigation of the role of preferred fabrication strategy, which efficiently combines both substrate application timing, mesostructured AuNR arrays with near-field and far-field coupling, to take full advantage of unprecedented quality are achieved over areas as large as 0.5 optical enhancement.21,22 cm2. All-atom molecular dynamic (MD) simulations are used For biomolecule detection, the ability to identify specific to elucidate the role of the volatile solvent (ethanol) in substances within complex mixtures (often at submicromolar improving the nanoparticle self-assembly, which is experimen- concentrations) is of key importance and has fueled research in tally confirmed by ζ-potential measurements. Ethanol is shown the field of SERS. Recent highlights include the detection of to strip away excess cetyltrimethylammonium bromide prostate-specific antigens in serum, multiplexed cancer type (CTAB) micelles from the MUHEG/water interface, thereby discrimination in cell cultures, and even SERS-based in vivo increasing the hydrophobicity of the particles and triggering imaging of tumors.23 Self-assembled AuNR structures in reversible agglomeration. We thus propose an assembly particular have been used for the detection of scrambled mechanism that takes into account the preorganization of prions or pyocyanin (a bacterial quorum-sensing molecule nanoparticles at the liquid/gas interface, which is further expressed by the human pathogen Pseudomonas aeruginosa) via confirmed by a kinetic study via photomicroscopy of drying SERS.24,25 The supercrystals used in these examples exhibit droplets and correlated atomic force microscopy (AFM)/ strong plasmonic near-field coupling due to the nanometer- scanning electron microscopy (SEM) characterization of the sized interparticle distances, thereby allowing molecular resulting substrates. Finally, application of the obtained detection in vitro in complex biological media, or even in plasmonic materials as SERS sensors for biomolecule detection situ in live bacterial cell cultures. However, in these examples, is demonstrated both in water and in complex biological the nanoparticles were assembled into isolated supercrystals or media. ■ micrometer-sized structures with array periodicities around 10 μm, and thus did not take advantage of far-field coupling. As RESULTS AND DISCUSSION shown by Reinhard et al. for micrometer-sized arrays, near-field coupled colloidal clusters can deliver significantly amplified Template-assisted nanoparticle deposition has recently become signal outputs when arranged in a periodic lattice.26 The far- a promising nanofabrication method that combines bottom-up field modes of the array must therefore be in resonance with and top-down approaches. The scalability of this method has the exciting light source, usually demanding submicrometer been largely facilitated by the synthesis of colloidal nano- lattice periodicities. These so-called lattice plasmons comprise particles in various shapes, with high monodispersity,33−35 an attractive route toward SERS substrates that offer high along with simple and inexpensive production of micro- sensitivity with low noble metal consumption and a reduced structured stamps by soft lithography.16,36,37 However, the laser power.22 quality of the nanomaterial is often compromised because of Despite recent reports on the colloidal assembly of AuNRs the well-known coffee ring effect: inhomogeneous material into micrometer-sized supercrystals, the production of high- distribution (in the worst case, accumulation outside of the quality AuNR arrays with submicrometer feature sizes has structured area) is observed when the dispersed nanoparticles proven to be an extraordinarily elusive task.27 Most of the are driven toward the rim of the confined droplet during drying current strategies are complicated, multistep procedures based due to convective forces (Figure 1a).38 We recently reported on pre-assembly by spin or dip coating, followed by transfer that template-assisted assembly of sterically stabilized (i.e., printing onto a target substrate.28,29 In fact, there is little poly(ethylene glycol)-capped) gold nanospheres (PEG− understanding of the factors influencing macroscopic array AuNSs) improved when the amount of residual surfactant homogeneity, which include particle surface functionalization, cetyltrimethylammonium chloride (CTAC) in the dispersion residual surfactant, and wetting of both the stamp and the was kept far below its critical micellar concentration (cmc of target substrate. Thus, the scaling and reproducibility of such CTAC: 1.6 mM at 25 °C).39 The most homogeneous assembly methods remain problematic. Recently, regular structures were achieved with a surfactant concentration of AuNR arrangements have been realized by combining around 50 μM, at which the nanospheres would slowly topographically structured templates with controlled con- agglomerate in the dispersion, forming extended golden films vective assembly or electrophoretic deposition techni- on the hydrophobic walls of plastic storage vessels.22 For the ques.30−32 However, to the best of our knowledge, no fast, MUHEG-coated rods used in this study, we could neither scalable method is available that yields homogeneous arrays of replicate the described effects nor sufficiently suppress the regular AuNR mesostructures over macroscopic surface areas, dominant coffee ring effect by simply decreasing the with no need for sophisticated equipment. concentration of free surfactant, which was tested down to We present herein an optimized approach for the self- 30 μM residual surfactant (see Figures 1b and S2). To improve assembly of gold nanorods (length: 55 ± 5 nm, diameter: 16.5 the wetting of the polydimethylsiloxane (PDMS) stamp,40 ± 1.5 nm) coated with (11-mercaptoundecyl)hexa(ethylene ethanol (EtOH) was added to the dispersions, thereby glycol) (MUHEG) into highly regular structures, using enabling a nearly complete suppression of the coffee ring ethanol/water mixtures as the dispersion media. Functionaliza- formation, which had strongly hindered the fabrication of well- tion with MUHEG has been shown to foster side-by-side defined AuNR arrays over areas larger than a few hundred organization of nanorods into structures with feature sizes μm2.27 Unexpectedly, the presence of EtOH also changed the 11764 DOI: 10.1021/acsami.9b00334 ACS Appl. Mater. Interfaces 2019, 11, 11763−11771 ACS Applied Materials & Interfaces Research Article Information (SI)), the drying droplets changed their appearance from dark brown to golden within seconds to minutes when AuNRs initiated pre-assembly at the air/liquid interface, as illustrated in Figure 2a.41 This pre-assembly process, not observed in purely aqueous dispersions, was found to consistently start at the droplet rim, i.e., the area of highest curvature and drying speed, indicating that the process is driven by convection due to EtOH evaporation.42 The macroscopic appearance of a droplet (insets of Figure 2b−d) at selected stages of evaporation can be correlated with the resulting mesostructures, as revealed by subsequent SEM measurements exemplified in Figure 2b−d. A clear relationship between the aggregation state of AuNRs at the air/liquid interface at the moment of contacting the substrate and the final particle distribution after drying was established: at short Figure 1. Comparison of MUHEG−AuNR assembly from water vs pre-assembly times, most of the liquid was found to move EtOH/water mixtures. (a) Scheme of the patterning process, which toward the nonpatterned rim of the PDMS stamp/substrate comprised drying of a dispersion droplet confined between a sandwich, where the particles would accumulate to form the structured soft PDMS stamp and a flat target substrate. (b) SEM images of the best microarrays that could be obtained from MUHEG- characteristic coffee ring. The opposite trend was observed for coated AuNRs dispersed in water. (c) Representative SEM images long pre-assembly times, where extended particle aggregation from an arbitrary part of a typical substrate obtained with the same led to overfilling of the patterned central region. Slight particle particles but now dispersed in an EtOH/water mixture. Only with agglomeration turned out to be most advantageous, since ethanolic dispersions, large area pattern replication was possible, and medium-sized lamellar AuNR clusters possess a moderate highly regular arrays of lying rod lamellae were formed. hydrodynamic mobility and cannot easily migrate to the rim or accumulate at the center of the substrate/stamp sandwich orientation of the nanorods from standing rods into side-by- (Figure S4).43 In general, the quality of the local array side packed lamellae, as shown in Figures 1c and S3. structure and the overall macroscopic surface homogeneity Intriguingly, ethanol-containing AuNR dispersions yielding were optimal when the substrates were placed onto the stamp high-quality assemblies displayed a macroscopic appearance just before the surface of the droplets turned golden (see the reminiscent of previously described PEG−AuNS dispersions, SI: Materials and Methods and Figure S5). i.e., a bright metallic luster.22 For fixed ambient temperature A detailed kinetic study of the process for varying and humidity, this characteristic golden shimmer was composition of the dispersion medium is summarized in consistently observed after a composition-controlled induction Figure 3, which shows that an increase in the initial EtOH/ period, during which the alcohol/water mixture was allowed to H2O ratio has an effect similar to a reduced duration of the evaporate partially: depending on particle concentration and evaporation period. Four different morphologies can be EtOH content (see Materials and Methods in the Supporting distinguished, which are marked with frames of different Figure 2. Influence of pre-evaporation and correlation between droplet appearance and the resulting assembly structure. Partial evaporation of EtOH leads to the pre-assembly of the AuNRs at the liquid−air interface, causing a golden shimmer at the droplet surface due to interparticle coupling (a and inset droplet photos). For identical initial composition and volume of the dispersion, contact with the target substrate at different times leads to distinct substrate qualities (b−d). The degree of pre-assembly directly correlates with the amount of particles deposited on the structured substrate area, where the underfilled surface is marked with a red frame, the optimally prepared sample with a green frame, and a vastly overfilled array with a blue one. 11765 DOI: 10.1021/acsami.9b00334 ACS Appl. Mater. Interfaces 2019, 11, 11763−11771 ACS Applied Materials & Interfaces Research Article Figure 3. Influence of initial EtOH content and evaporation time prior to contact with the substrate (particle concentration constant @ 100 mM Au0). Four different substrate qualities can be distinguished, marked with colored frames in (a). EtOH concentrations below 33 vol % always yielded empty substrate centers. Dispersions with higher alcohol content produced very homogeneous arrays (green), with correct timing. To avoid underfilling (red), longer waiting times were required for higher initial EtOH concentrations, whereby excessive solvent evaporation leads to particle accumulation at the substrate center (blue) in all series. AFM height images in (b) for substrates from the series with an initial EtOH content of 66 vol % correspond to evaporation times of 10, 75, 120, and 180 s (from left to right). The cross sections in (c) were measured for the bottom row of the corresponding square arrays in the AFM images and plotted with an offset in the z-direction for the sake of clarity. colors: apart from the above discussed three scenarios, namely “underfilled” (red frame), “optimally filled” (green frame), and “overfilled” (blue frame), we found that for long evaporation periods or very low initial EtOH content, nearly empty substrate centers (black frame) were obtained. This emergence of a dominant coffee ring effect can be attributed to the redispersion of agglomerated particles after the evaporation of a major fraction of added EtOH (Figure S4). With this effect in mind, we investigated whether the optimum time for contact with the substrate was determined by the EtOH/H2O ratio at the moment of substrate contact (i.e., independent of the concentration of dispersed particles), or by the amount of pre-assembled particle clusters at the droplet surface (i.e., strongly dependent on particle concen- tration in the dispersion). As shown in Figure 4 for fixed waiting time and initial EtOH/H2O ratio, increasing the Figure 4. Influence of the particle concentration on array height and particle concentration also lead to particle accumulation at the homogeneity. (a−d) AFM height images of representative arrays substrate centers (see Figure S6 for macroscopic character- prepared with identical initial EtOH concentration, timing of ization). We therefore conclude that the outcome of the substrate application, and surfactant concentration. Particle concen- assembly process is mainly determined by the state of particle trations were varied within the range [Au0] = 50−500 mM. From (a) to (c), the height of the separated AuNR posts gradually increased, agglomeration at the liquid−air interface at the moment of until a film of connected posts was obtained at the highest contact with the substrate. concentration (d). To better understand the influence of EtOH on the assembly behavior of the AuNRs, the molecular interactions which is attributed to minimization of the free energy by of MUHEG-coated Au surfaces in contact with the solvent reducing the amount of interfacial water (Figure S7b). were simulated using all-atom molecular dynamics. These Following 200 ns of simulation in pure water, water molecules simulations reproduced the environment of the AuNRs in the were removed from the system and replaced with either pure real dispersions, where CTAB was present at concentrations EtOH (Figure S8) or 52% (v/v) EtOH in water (Figure 5). As ranging from 30 to 500 μM. The equilibrated system shows demonstrated by the timeline of the simulation trajectories, the that some CTA+ molecules reside within the MUHEG layer introduction of EtOH results in a quick dissociation of the with their quaternary ammonium headgroups oriented toward CTAB micelle from the MUHEG interface, even in a mixture the Au surface, whereas CTAB micelles form and reside at the with water (Figures 5a and S9). MUHEG−water interface (Figure S7). In pure water, the Regardless of the initial position of the CTAB micelle at the CTAB micelle remained bound at the interface of the MUHEG/solvent interface, dissociation of the micelle from MUHEG layer for the entire duration of the simulation, the interface could be clearly observed after 100 ns, both in 11766 DOI: 10.1021/acsami.9b00334 ACS Appl. Mater. Interfaces 2019, 11, 11763−11771 ACS Applied Materials & Interfaces Research Article Figure 5. All-atom molecular dynamics simulations of CTAB micelles interacting with a MUHEG-functionalized Au(100) surface (91 MUHEG and 71 CTAB molecules in 52% (v/v) EtOH) and experimental verification. Initial atom coordinates in (a) are derived from a separate 100 ns simulation in pure water. The originally bound surfactant micelle completely detaches from the MUHEG ligand layer within 125 ns after switching the solvent from pure water to an EtOH-containing mixture. (b) Addition of EtOH to a particle dispersion ([CTAB] = 500 μM) causes a strong decay of the ζ-potential above 25% (v/v) of alcohol. At 35% (v/v) EtOH, surface-bound surfactant micelles are completely detached, leaving the particle surface uncharged. (c) In UV−vis spectroscopy measurements, fast agglomeration of the AuNRs at intermediate EtOH concentrations of 30−45% (v/v) becomes obvious. Figure 6. Influence of CTAB surfactant concentration (20−500 μm). (a) Photographs of three substrates obtained at different [CTAB], revealing similar macroscopic homogeneity. (b) Cross sections measured from AFM height images (c), which show increasing average post height with surfactant concentration in the dispersion, and the presence of an elevated rim in all posts. (d) SEM micrographs for low and intermediate CTAB concentrations demonstrate that complete filling is only reached above a minimum surfactant concentration. pure EtOH and in the water/EtOH mixture (Figures 5a and ments of AuNR colloids for varying CTAB and EtOH S8). Visual observations are confirmed by the radial concentrations. In water, the particles exhibited a strongly distribution function of the quaternary ammonium headgroup positive ζ-potential, even far below the cmc of CTAB (e.g., of CTA+ interacting with the terminal −OH group of +35 mV above 100 μM; Figure S10). With EtOH contents MUHEG, which was significantly decreased by the end of above 25% (v/v), however, the surface charge dropped, simulations in both EtOH and the water/EtOH mixture reaching 0 mV at 35% (v/v) and −10 mV at 45% (v/v) EtOH (Figure S9b,d). The electron density of the system over the (Figure 5b). Whereas AuNRs were prepared and stored at simulation trajectory also exhibits the dissociation of the CTAB concentrations above the cmc (1 mM), ζ-potential CTAB micelle from the MUHEG/solvent interface (Figure measurements were conducted after dilution, well below the S9a,c). The removal of adsorbed CTAB micelles can explain cmc (30 μM ≤ [CTAB] ≤ 500 μM). These data support the both the experimentally observed clustering of AuNRs and hypothesis that, in water, CTAB micelles remain bound to the their increased affinity toward the air−liquid interface, as the AuNR−MUHEG interface far below the cmc ([CTAB] ≥ 30 result of a more hydrophobic character of the particle coating. μM) but are readily removed by EtOH. The resulting loss of To verify the mechanism supported by the MD simulations, colloidal stability is mirrored in the UV−vis spectra (Figure we conducted ζ-potential and UV−vis spectroscopy measure- 5c) showing peaks around 600 nm that are characteristic for 11767 DOI: 10.1021/acsami.9b00334 ACS Appl. Mater. Interfaces 2019, 11, 11763−11771 ACS Applied Materials & Interfaces Research Article side-to-side aggregation of AuNRs.44 Collectively, these results make a strong case for the idea that EtOH efficiently triggers the agglomeration and surface activity of AuNRs, thereby reducing particle mobility during drying and ultimately leading to a strongly altered self-assembly behavior. Similar trapping of nanoparticles at the liquid−air interface has been recognized as a method for reducing coffee stains and building highly ordered close-packed layers.41 Interestingly, the surfactant concentration itself influences the structure of self-assembled AuNR arrays in a counter- intuitive manner. Whereas for dispersions in pure water the quality of the assembly deteriorated when increasing the surfactant concentration, EtOH-containing dispersions were less influenced by this parameter, as CTAB concentrations up to 500 μM did not significantly alter the homogeneity and array dimensions. However, the concentration of free surfactant did influence the average height of the individual posts, leading to roughly 3-fold higher structures when CTAB concentration was increased from 20 up to 500 μM (Figure 6). As the volume occupied by CTAB is negligible compared to the volume of gold in all cases (around 100 μM CTAB vs 100 Figure 7. Feature size variation from 10 μm to 200 nm (lattice mM Au0), such height differences must be caused by varying parameters from (a−d): 20 μm, 600, 500, and 400 nm). SEM images amounts of particles being deposited outside the patterned show that tall, micron-sized structures (a) exhibit the best array stamp area. quality when prepared with high particle concentrations (up to 850 Potential explanations for this behavior are changes of the mM Au0) and very short induction periods (below 15 s), whereas stamp wetting and meniscus shape during late drying stages, or submicrometer structures (b−d) exhibit good filling for much lower the interplay of solute- and surfactant-driven Marangoni flows particle concentrations (around 100 mM Au0) when combined with that influence particle repartition.43,45−47 Although this intermediate pre-assembly times (around 45 s). For all samples, a behavior beckons a separate in-depth investigation, we further side-by-side arrangement of AuNRs could be confirmed by high- resolution SEM images (insets). For large structures in (a), parallel explore the efficacy of substrates produced by this assembly lamellae of lying rods spanning over several micrometers were found. approach for molecular ultradetection. As shown in Figure 7, this kinetically controlled assembly technique paves the way to highly regular AuNR arrays with feature sizes ranging from 200 excitation source because it targets both the first transparency nm up to tens of microns. window of biological systems and the excitation wavelength of In a recent work, we demonstrated that arrangement of gold the analyte molecule. In fact, this laser is the most widespread nanoparticles into periodic microarrays can significantly boost light source for Raman and other optical measurements in SERS sensitivity via tailored lattice plasmon resonances.22 Such biological media due to its relatively low tendency to cause lattice plasmon resonances typically occur near a Rayleigh photoluminescence emission overlaying the signals.22 With the anomaly, i.e., a strong change of reflectivity due to diffraction excitation at 785 nm, one would expect the highest signal at the metal grating.48 For square arrays of gold nanoparticles, enhancement for a lattice parameter L = 500 nm, for which a the lattice resonance wavelength can be approximated as λmax = resonance around 770 nm is predicted by the formula for λmax Lnsubs, where the lattice parameter L corresponds to the center- above. to-center distance between neighboring posts and nsubs to the Compared to arrays with micrometer-sized lattices, the refractive index of the substrate.49,50 Shown in Figure 7 are arrangements with submicrometer lattice parameters indeed AuNR arrays with L ranging from 20 μm down to 400 nm. The give rise to significantly enhanced SERS: in Figure 8a, the corresponding extinction spectra (Figure S11) show that for SERS spectrum of pyocyanin in water recorded on a 1.66 μm each lattice parameter, distinct resonance peaks arise near the lattice shows the dominant Raman modes (1352 and ∼1600 predicted position of the Rayleigh anomaly. Exploitation of cm−1), with a meager signal of ∼1000 counts/s, whereas the these lattice resonances in SERS spectroscopy demands precise miniaturized arrays with lattice parameters of 400 and 500 nm matching with the wavelength of the excitation laser and have close to 25-fold intensity. Plotting the signal intensity of therefore requires highly regular arrays with submicrometer the dominant mode at 1600 cm−1 (from phenyl ring periodicities for the typically employed UV−vis/NIR Raman deformation and C−C stretching) against the lattice parameter setups.2,51 (Figure 8b) clearly reveals an inverse correlation of feature size To assess these substrates in a practical sensing application, and SERS intensity. As all the probed substrates have a very we chose the bacterial pigment/toxin pyocyanin as a probe similar percentage of effective area covered with nanoparticles molecule. Pyocyanin is a Raman-active quorum-sensing (25−27%), the observed trends must arise from the different molecule expressed by P. aeruginosa, which can be detected lattice parameters and increasing surface-to-volume ratios with by SERS using NIR excitation.25 The SERS spectra of decreasing feature size. Figure 8c displays the dependence of pyocyanin in water and Luria−Bertrani (LB) bacterial growth the SERS signal at 1600 cm−1 on the pyocyanin concentration medium were recorded using patterned substrates with lattice in water, for the best substrate (shown in Figure 7d), parameters between 400 nm and 1.66 μm (corresponding to suggesting a detection limit of approximately 10−12 M. feature dimensions between 200 nm and 1 μm; see SI Table For SERS-based clinical diagnostics, measuring in complex S1), as shown in Figure 8. A 785 nm laser was chosen as the biofluids represents a significant challenge due to fouling of the 11768 DOI: 10.1021/acsami.9b00334 ACS Appl. Mater. Interfaces 2019, 11, 11763−11771 ACS Applied Materials & Interfaces Research Article Figure 8. SERS spectroscopy of pyocyanin using hierarchically assembled gold nanoparticle arrays and 785 nm excitation. (a) SERS spectra of 1 μm pyocyanin in water, recorded on periodic square arrays with varying lattice parameter. Strongly improved signals are observed for submicrometer lattice arrays. The corresponding SERS intensity of the 1600 cm−1 peak rises steeply with decreasing lattice parameter below a critical value of L = 740 nm (b). (c) The normalized SERS intensity of 1 μM to 1 pM pyocyanin on the champion substrate displays a sigmoidal shape, where the SERS signal undergoes a 100-fold increase between 0.1 and 10 nM pyocyanin. (d) SERS spectra from pyocyanin in LB broth prove efficient detection in complex media. The spectra are the average from multiple measurements (n = 50−150) and measured with a laser power of 1.5 mW. The integration time was 1 s for all spectra in (a) and 5 s for the data shown in (d). plasmonic metal surface with biomolecules, which restricts the effect was studied, supporting the idea that the removal of access of the analyte to the surface and thus reduces signal residual surfactant micelles from the interface of the MUHEG- enhancement. In a recent study, we showed that micron-scale coating on the particles plays a crucial role in the self-assembly arrays of gold nanoparticles allowed ultrasensitive in situ behavior. The resulting colloidal destabilization leads to detection of pyocyanin in cultures of P. aeruginosa, when enhanced surface activity and partial agglomeration of the coated with mesoporous silica. The mesoporous layer acted as gold nanoparticles, which turned out to be advantageous for a sieve to prevent larger proteins and biomolecules from creating homogeneous AuNR arrays over mm2 areas. This fouling the gold nanoparticle interface while allowing the method enables the routine fabrication of high-quality periodic percolation and detection of small molecules down to arrays with submicrometer features and was found to be easier, femtomolar concentrations.25 In another recent study, a faster, and more reliable than common multistep methods nacre-like film impregnated with AuNPs exhibited a similar increase in signal due to filtration by the porous polymer based on spin or dip coating and subsequent transfer printing. network compared to a film of bare AuNPs.52 Intriguingly, we The resulting structures successfully combine plasmonic and observe that although the lack of such a filtration layer leads to photonic resonances to deliver significant improvements in a moderate signal background from the adsorption of SERS sensor fabrication. Consequently, they allow sensitive biomolecules like peptides and yeast tryptone, the submi- biomolecule detection with substantially reduced material costs crometer patterned substrates introduced in this work still and preparative effort, as demonstrated with the bacterial yield very clear SERS spectra in biological media (Figure 8d). pathogen pyocyanin both in water and Luria−Bertrani ■ medium. The capability of these substrates for detecting CONCLUSIONS bacterial quorum-sensing molecules in bacterial growth medium, even without an additional filtration layer, should A kinetically controlled approach for the template-assisted facilitate in situ studies of bacterial growth/quorum sensing via assembly of plasmonic nanoparticles, which relies on the SERS and could be particularly useful for studying the effects addition of alcohol to aqueous AuNP dispersions, is reported herein. This approach largely improves the large-scale of surface topography on bacterial attachment and biofilm reproducibility and overall quality of the resulting arrays. As formation.53 Finally, we propose that the particle assembly the crucial experimental parameters in this method, we mechanism discussed herein is of a universal nature and could identified the dispersion medium composition, solvent have important implications for the field of template-assisted evaporation time, particle concentration, and surfactant colloidal assembly. We therefore expect this proof of concept concentration. A strong correlation was found between particle to trigger deeper investigations utilizing nanoparticles of varied pre-assembly at the droplet−air interface and quality of the materials composition and surface functionalization, as well as resulting final substrate. Through a combination of experi- the exploration of a plethora of pattern designs and target ments and simulations, the underlying mechanism for this substrates. 11769 DOI: 10.1021/acsami.9b00334 ACS Appl. Mater. Interfaces 2019, 11, 11763−11771 ACS Applied Materials & Interfaces Research Article ■ * ASSOCIATED CONTENT S Supporting Information (4) Karg, M.; König, T. A. F.; Retsch, M.; Stelling, C.; Reichstein, P. M.; Honold, T.; Thelakkat, M.; Fery, A. Colloidal Self-Assembly Concepts for Light Management in Photovoltaics. Mater. Today 2015, The Supporting Information is available free of charge on the 18, 185−205. ACS Publications website at DOI: 10.1021/acsami.9b00334. (5) Chen, T.; Reinhard, B. M. Assembling Color on the Nanoscale: Experimental part; characterization of nanoparticles and Multichromatic Switchable Pixels From Plasmonic Atoms and large area substrates by electron microscopy; micro- Molecules. Adv. Mater. 2016, 28, 3522−3527. scopic investigation of an evaporating AuNR dispersion (6) Pastoriza-Santos, I.; Kinnear, C.; Perez-Juste, J.; Mulvaney, P.; droplet; study of the impact of evaporation time and Liz-Marzan, L. M. Plasmonic Polymer Nanocomposites. Nat. Rev. particle concentration on macroscopic substrate homo- Mater. 2018, 3, 375−391. geneity; additional simulations of the interaction of (7) Kraus, T.; Brodoceanu, D.; Pazos-Pérez, N.; Fery, A. Colloidal Surface Assemblies: Nanotechnology Meets Bioinspiration. Adv. CTAB with MUHEG-coated AuNRs in water and Funct. Mater. 2013, 23, 4529−4541. EtOH; UV−vis and ζ-potential data of MUHEG− (8) Schlücker, S. Surface-Enhanced Raman Spectroscopy: Concepts AuNRs with varying CTAB concentration; optical and Chemical Applications. Angew. Chem., Int. Ed. 2014, 53, 4756− characterization of the substrates by UV−vis−NIR 4795. spectroscopy (PDF) (9) Liu, N.; Mesch, M.; Weiss, T.; Hentschel, M.; Giessen, H. ■ Infrared Perfect Absorber and Its Application as Plasmonic Sensor. AUTHOR INFORMATION Nano Lett. 2010, 10, 2342−2348. (10) Petefish, J. W.; Hillier, A. C. Angle-Tunable Enhanced Infrared Corresponding Authors Reflection Absorption Spectroscopy via Grating-Coupled Surface *E-mail: christoph.hanske@googlemail.com (C.H.). Plasmon Resonance. Anal. Chem. 2014, 86, 2610−2617. *E-mail: llizmarzan@cicbiomagune.es (L.M.L.-M.). (11) Ye, X.; Zheng, C.; Chen, J.; Gao, Y.; Murray, C. B. Using Binary ORCID Surfactant Mixtures to Simultaneously Improve the Dimensional Agustín Mihi: 0000-0003-3821-7881 Tunability and Monodispersity in the Seeded Growth of Gold Luis M. Liz-Marzán: 0000-0002-6647-1353 Nanorods. Nano Lett. 2013, 13, 765−771. (12) Link, S.; El-Sayed, M. A. Shape and Size Dependence of Present Address Radiative, Non-Radiative and Photothermal Properties of Gold ∥ Institute of Advanced Ceramics, Hamburg University of Nanocrystals. Int. Rev. Phys. Chem. 2000, 19, 409−453. Technology, 21073 Hamburg, Germany (E.H.H.). (13) González-Rubio, G.; Díaz-Núñez, P.; Rivera, A.; Prada, A.; Notes Tardajos, G.; González-Izquierdo, J.; Bañares, L.; Llombart, P.; The authors declare no competing financial interest. Macdowell, L. G.; Alcolea Palafox, M.; Liz-Marzán, L. M.; Peña- ■ Rodríguez, O.; Guerrero-Martínez, A. Femtosecond Laser Reshaping ACKNOWLEDGMENTS Yields Gold Nanorods with Ultranarrow Surface Plasmon Resonances. Science 2017, 358, 640−644. This work has been funded by the Spanish MINECO (Grant (14) Camden, J. P.; Dieringer, J. A.; Wang, Y.; Masiello, D. J.; Marks, No. MAT2017-86659-R). C.H. acknowledges the Alexander L. D.; Schatz, G. C.; Van Duyne, R. P. Probing the Structure of Single- von Humboldt Foundation for funding in the framework of a Molecule Surface-Enhanced Raman Scattering Hot Spots. J. Am. Feodor Lynen fellowship. E.H.H. acknowledges the receipt of a Chem. Soc. 2008, 130, 12616−12617. Juan de la Cierva-fellowship (FJCI-2014-22598). A.M. is (15) Slaughter, L. S.; Wang, L.-Y.; Willingham, B. A.; Olson, J. M.; grateful to the funding from the European Research Council Swanglap, P.; Dominguez-Medina, S.; Link, S. Plasmonic Polymers (ERC) under the European Union’s Horizon 2020 research Unraveled Through Single Particle Spectroscopy. Nanoscale 2014, 6, and innovation program (grant agreement No. 637116, 11451−11461. ENLIGHTMENT) and funding by the Spanish Ministerio (16) Hanske, C.; Tebbe, M.; Kuttner, C.; Bieber, V.; Tsukruk, V. V.; de Economia,́ Industria y Competitividad (MINECO) (Grant Chanana, M.; König, T. A. F.; Fery, A. Strongly Coupled Plasmonic Nos. SEV-2015-0496 and MAT2016-79053-P). The authors Modes on Macroscopic Areas via Template-Assisted Colloidal Self- Assembly. Nano Lett. 2014, 14, 6863−6871. further thank Mathias Charconnet for fruitful discussions and (17) Tian, C.; Deng, Y.; Zhao, D.; Fang, J. Plasmonic Silver Vished Kumar for TEM imaging. This work was performed Supercrystals with Ultrasmall Nanogaps for Ultrasensitive SERS- under the Maria de Maeztu Units of Excellence Program from Based Molecule Detection. Adv. Opt. Mater. 2015, 3, 404−411. the Spanish State Research Agency (Grant No. MDM-2017- (18) Kang, M.; Park, S.-G.; Jeong, K.-H. Repeated Solid-state 0720). We acknowledge support of the publication fee by the Dewetting of Thin Gold Films for Nanogap-Rich Plasmonic CSIC Open Access Publication Support Initiative through its Nanoislands. Sci. Rep. 2015, 5, No. 14790. Unit of Information Resources for Research (URICI). (19) Bruchhaus, L.; Mazarov, P.; Bischoff, L.; Gierak, J.; Wieck, A. ■ REFERENCES (1) Knight, M. W.; Sobhani, H.; Nordlander, P.; Halas, N. J. D.; Hövel, H. Comparison of Technologies for Nano Device Prototyping with a Special Focus on Ion Beams: a Review. Appl. Phys. Rev. 2017, 4, No. 011302. Photodetection with Active Optical Antennas. Science 2011, 332, (20) Wang, D.; Yang, A.; Wang, W.; Hua, Y.; Schaller, R. D.; Schatz, 702−704. G. C.; Odom, T. W. Band-Edge Engineering for Controlled Multi- (2) Yan, B.; Boriskina, S. V.; Reinhard, B. M. Design and Modal Nanolasing in Plasmonic Superlattices. Nat. Nanotechnol. Implementation of Noble Metal Nanoparticle Cluster Arrays for 2017, 12, 889−894. Plasmon Enhanced Biosensing. J. Phys. Chem. C 2011, 115, 24437− (21) Cui, Y.; Björk, M. T.; Liddle, J. A.; Sönnichsen, C.; Boussert, B.; 24453. Alivisatos, A. P. Integration of Colloidal Nanocrystals Into Litho- (3) Bagheri, S.; Weber, K.; Gissibl, T.; Weiss, T.; Neubrech, F.; graphically Patterned Devices. Nano Lett. 2004, 4, 1093−1098. Giessen, H. Fabrication of Square-Centimeter Plasmonic Nano- (22) Matricardi, C.; Hanske, C.; Garcia-Pomar, J. L.; Langer, J.; antenna Arrays by Femtosecond Direct Laser Writing Lithography: Mihi, A.; Liz-Marzan, L. M. Gold Nanoparticle Plasmonic Super- Effects of Collective Excitations on SEIRA Enhancement. ACS lattices as Surface-Enhanced Raman Spectroscopy Substrates. ACS Photonics 2015, 2, 779−786. Nano 2018, 12, 8531−8539. 11770 DOI: 10.1021/acsami.9b00334 ACS Appl. Mater. Interfaces 2019, 11, 11763−11771 ACS Applied Materials & Interfaces Research Article (23) Hanske, C.; Sanz-Ortiz, M. N.; Liz-Marzan, L. M. Silica-Coated (41) Bigioni, T. P.; Lin, X. M.; Nguyen, T. T.; Corwin, E. I.; Witten, Plasmonic Metal Nanoparticles in Action. Adv. Mater. 2018, 30, T. A.; Jaeger, H. M. Kinetically Driven Self Assembly of Highly No. 1707003. Ordered Nanoparticle Monolayers. Nat. Mater. 2006, 5, 265−270. (24) Alvarez-Puebla, R. A.; Agarwal, A.; Manna, P.; Khanal, B. P.; (42) Deegan, R. D.; Bakajin, O.; Dupont, T. F.; Huber, G.; Nagel, S. Aldeanueva-Potel, P.; Carbó-Argibay, E.; Pazos-Pérez, N.; Vigderman, R.; Witten, T. A. Contact Line Deposits in an Evaporating Drop. Phys. L.; Zubarev, E. R.; Kotov, N. A.; Liz-Marzán, L. M. Gold Nanorods Rev. E 2000, 62, 756−765. 3D-Supercrystals as Surface Enhanced Raman Scattering Spectrosco- (43) Kim, H.; Boulogne, F.; Um, E.; Jacobi, I.; Button, E.; Stone, H. py Substrates for the Rapid Detection of Scrambled Prions. Proc. Natl. A. Controlled Uniform Coating From the Interplay of Marangoni Acad. Sci. U.S.A. 2011, 108, 8157−8161. Flows and Surface-Adsorbed Macromolecules. Phys. Rev. Lett. 2016, (25) Bodelón, G.; Montes-García, V.; López-Puente, V.; Hill, E. H.; 116, No. 124501. Hamon, C.; Sanz-Ortiz, M. N.; Rodal-Cedeira, S.; Costas, C.; (44) Funston, A. M.; Novo, C.; Davis, T. J.; Mulvaney, P. Plasmon Coupling of Gold Nanorods at Short Distances and in Different Celiksoy, S.; Perez-Juste, J.; Pastoriza-Santos, I.; Liz-Marzán, L. M.; Geometries. Nano Lett. 2009, 9, 1651−1658. et al. Detection and Imaging of Quorum Sensing in Pseudomonas (45) Karpitschka, S.; Riegler, H. Noncoalescence of Sessile Drops aeruginosa Biofilm Communities by Surface-Enhanced Resonance From Different but Miscible Liquids: Hydrodynamic Analysis of the Raman Scattering. Nat. Mater. 2016, 15, 1203−1211. Twin Drop Contour as a Self-Stabilizing Traveling Wave. Phys. Rev. (26) Yan, B.; Thubagere, A.; Premasiri, W. R.; Ziegler, L. D.; Dal Lett. 2012, 109, No. 066103. Negro, L.; Reinhard, B. M. Engineered SERS Substrates with (46) Li, P.; Li, Y.; Zhou, Z.-K.; Tang, S.; Yu, X.-F.; Xiao, S.; Wu, Z.; Multiscale Signal Enhancement: Nanoparticle Cluster Arrays. ACS Xiao, Q.; Zhao, Y.; Wang, H.; Chu, P. K. Evaporative Self-Assembly of Nano 2009, 3, 1190−1202. Gold Nanorods Into Macroscopic 3D Plasmonic Superlattice Arrays. (27) Hamon, C.; Novikov, S.; Scarabelli, L.; Basabe-Desmonts, L.; Adv. Mater. 2016, 28, 2511−2517. Liz-Marzán, L. M. Hierarchical Self-Assembly of Gold Nanoparticles (47) Sempels, W.; De Dier, R.; Mizuno, H.; Hofkens, J.; Vermant, J. Into Patterned Plasmonic Nanostructures. ACS Nano 2014, 8, Auto-Production of Biosurfactants Reverses the Coffee Ring Effect in 10694−10703. a Bacterial System. Nat. Commun. 2013, 4, No. 1757. (28) Horn, A.; Schoberth, H. G.; Hiltl, S.; Chiche, A.; Wang, Q.; (48) Khlopin, D.; Laux, F.; Wardley, W. P.; Martin, J.; Wurtz, G. A.; Schweikart, A.; Fery, A.; Böker, A. Nanostructured Wrinkled Surfaces Plain, J.; Bonod, N.; Zayats, A. V.; Dickson, W.; Gérard, D. Lattice for Templating Bionanoparticles-Controlling and Quantifying the Modes and Plasmonic Linewidth Engineering in Gold and Aluminum Degree of Order. Faraday Discuss. 2009, 143, 143−150. Nanoparticle Arrays. J. Opt. Soc. Am. B 2017, 34, 691−700. (29) Tebbe, M.; Mayer, M.; Glatz, B. A.; Hanske, C.; Probst, P. T.; (49) Guo, R.; Hakala, T. K.; Törmä, P. Geometry Dependence of Müller, M. B.; Karg, M.; Chanana, M.; König, T. A. F.; Kuttner, C.; Surface Lattice Resonances in Plasmonic Nanoparticle Arrays. Phys. Fery, A. Optically Anisotropic Substrates via Wrinkle-Assisted Rev. B 2017, 95, No. 155423. Convective Assembly of Gold Nanorods on Macroscopic Areas. (50) Auguié, B.; Bendaña, X. M.; Barnes, W. L.; García de Abajo, F. Faraday Discuss. 2015, 181, 243−260. J. Diffractive Arrays of Gold Nanoparticles Near an Interface: Critical (30) Kuemin, C.; Nowack, L.; Bozano, L.; Spencer, N. D.; Wolf, H. Role of the Substrate. Phys. Rev. B 2010, 82, No. 155447. Oriented Assembly of Gold Nanorods on the Single-Particle Level. (51) Huang, Y.; Zhang, X.; Ringe, E.; Hou, M.; Ma, L.; Zhang, Z. Adv. Funct. Mater. 2012, 22, 702−708. Tunable Lattice Coupling of Multipole Plasmon Modes and Near- (31) Flauraud, V.; Mastrangeli, M.; Bernasconi, G. D.; Butet, J.; Field Enhancement in Closely Spaced Gold Nanorod Arrays. Sci. Rep. 2016, No. 23159. Alexander, D. T. L.; Shahrabi, E.; Martin, O. J. F.; Brugger, J. (52) Hill, E. H.; Hanske, C.; Johnson, A.; Yate, L.; Jelitto, H.; Nanoscale Topographical Control of Capillary Assembly of Nano- Schneider, G. A.; Liz-Marzán, L. M. Metal Nanoparticle Growth particles. Nat. Nanotechnol. 2017, 12, 73−80. Within Clay−Polymer Nacre-Inspired Materials for Improved (32) Zhang, H.; Cadusch, J.; Kinnear, C.; James, T.; Roberts, A.; Catalysis and Plasmonic Detection in Complex Biofluids. Langmuir Mulvaney, P. Direct Assembly of Large Area Nanoparticle Arrays. 2017, 33, 8774−8783. ACS Nano 2018, 12, 7529−7537. (53) Hill, E. H.; Liz-Marzán, L. M. Toward Plasmonic Monitoring of (33) Zheng, Y.; Zhong, X.; Li, Z.; Xia, Y. Successive, Seed-Mediated Surface Effects on Bacterial Quorum-Sensing. Curr. Opin. Colloid Growth for the Synthesis of Single-Crystal Gold Nanospheres with Interface Sci. 2017, 32, 1−10. Uniform Diameters Controlled in the Range of 5−150 Nm. Part. Part. Syst. Charact. 2014, 31, 266−273. (34) O’Brien, M. N.; Jones, M. R.; Brown, K. A.; Mirkin, C. A. Universal Noble Metal Nanoparticle Seeds Realized Through Iterative Reductive Growth and Oxidative Dissolution Reactions. J. Am. Chem. Soc. 2014, 136, 7603−7606. (35) Scarabelli, L.; Sánchez-Iglesias, A.; Perez-Juste, J.; Liz-Marzán, L. M. A “Tips and Tricks” Practical Guide to the Synthesis of Gold Nanorods. J. Phys. Chem. Lett. 2015, 6, 4270−4279. (36) Qin, D.; Xia, Y.; Whitesides, G. M. Soft Lithography for Micro- and Nanoscale Patterning. Nat. Protoc. 2010, 5, 491−502. (37) Kraus, T.; Malaquin, L.; Schmid, H.; Riess, W.; Spencer, N. D.; Wolf, H. Nanoparticle Printing with Single-Particle Resolution. Nat. Nanotechnol. 2007, 2, 570−576. (38) Deegan, R. D.; Bakajin, O.; Dupont, T. F.; Huber, G.; Nagel, S. R.; Witten, T. A. Capillary Flow as the Cause of Ring Stains From Dried Liquid Drops. Nature 1997, 389, 827−829. (39) Nomura, T.; Asai, Y.; Murahashi, N.; Iwamoto, K. Formation of Spherical Micelles by the Novel Platelet Activating Factor Receptor Antagonist, E5880. Chem. Pharm. Bull. 2000, 48, 947−950. (40) Vazquez, G.; Alvarez, E.; Navaza, J. M. Surface Tension of Alcohol Water + Water From 20 to 50 °C. J. Chem. Eng. Data 1995, 40, 611−614. 11771 DOI: 10.1021/acsami.9b00334 ACS Appl. Mater. Interfaces 2019, 11, 11763−11771