Article pubs.acs.org/Langmuir
Cobalt Nanoparticles Growth on a Block Copolymer Thin Film: A Time-Resolved GISAXS Study Ezzeldin Metwalli,*,† Volker Körstgens,† Kai Schlage,‡ Robert Meier,† Gunar Kaune,†,§ Adeline Buffet,‡ Sebastien Couet,‡,∥ Stephan V. Roth,‡ Ralf Röhlsberger,‡ and Peter Müller-Buschbaum† †
Physik-Department, Lehrstuhl für Funktionelle Materialien, Technische Universität München, James-Franck-Str. 1, 85748 Garching, Germany ‡ Deutsches Elektronen-Synchrotron (DESY), Notkestrasse 85, D-22607 Hamburg, Germany § Martin-Luther-Universität Halle-Wittenberg, von-Danckelmann-Platz 3, 06120 Halle, Germany ∥ Instituut voor Kern-en Stralingsfysica, KU Leuven, Celestijnenlaan 200D, B-3001 Leuven, Belgium ABSTRACT: Cobalt sputter deposition on a nanostructured polystyrene-block-poly(ethylene oxide), P(S-b-EO), template is followed in real time with grazing incidence small-angle Xray scattering (GISAXS). The polymer template consists of highly oriented parallel crystalline poly(ethylene oxide) (PEO) domains that are sandwiched between two polystyrene (PS) domains. In-situ GISAXS shows that cobalt atoms selectively decorate the PS domains of the microphase-separated polymer film and then aggregate to form surface metal nanopatterns. The polymer template is acting as a directing agent where cobalt metal nanowires are formed. At high metal load, the characteristic selectivity of the template is lost, and a uniform metal layer forms on the polymer surface. During the early stage of cobalt metal deposition, a highly asymmetric nanoparticles agglomeration is dominating structure formation. The cobalt nanoparticles mobility in combination with the high tendency of the nanoparticles to coalescence and to form immobile large-sized particles at the PS domains are discussed as mechanisms of structure formation. by an annealing step.19−23 The third method employs the dewetting of polymer films that were created from low concentrations of mixed solutions of polymer and polymergrafted nanoparticles to create metal nanostructures.4,24,25 The fourth approach is based on the selective deposition of metal nanoparticles on a nanostructure polymer template by wetting the soft nanostructured surface with nanoparticles solution subjected to a hydrodynamic flow.26 The fifth method employs the self-organization characteristic of evaporated or sputtered metal atoms on a microphase separation polymer film to create metal nanopatterns by selective adsorption of the metal atoms or clusters.27,28 These methods used for the metal patterning, along with the various mechanisms (i.e., adsorption, surface diffusion, nucleation, and agglomeration) involved in the dispersion process29−33 of the nanoparticles within the polymer film, make it difficult to understand the kinetics of nanoparticle formation and growth in a polymer matrix. For example, in the coevaporation method, both metal and organic particles impinge on the solid surface and codeposition occurs in
1. INTRODUCTION Microphase separation and self-assembly of block copolymers provide an exquisite and effective “bottom-up” approach for fabricating nanostructured polymer films.1−3 The resulting soft nanostructured polymer films can be used as a template for making patterns of hard inorganic materials.4−8 Generally, patterns of metal nanoclusters in a matrix of insulating polymer have unique physical properties and have been proposed for electronic, magnetic, and photonic applications.6,9−12 In particular, magnetic nanoparticles in a polymer matrix have recently attracted tremendous interest due to their ability to tune and optimize the magnetic response of the surface in a magnetic field.12−14 So far, typically the patterning of metal nanoparticles within polymer films has been achieved via five main routes: The first method is the vapor phase codeposition of polymers/ nanoparticles in a high-level vacuum prior to thermal annealing.15−18 Annealing of the polymer film above the glass transition temperature (Tg) of the polymer allows for the structural relaxation of the polymer matrix and was proven to be responsible for the dispersion of the metal nanoparticles within the polymer film. The second method is based on the deposition from a mixture of a block copolymer and organiccoated nanoparticles in solution onto a solid surface followed © XXXX American Chemical Society
Received: February 28, 2013 Revised: May 2, 2013
A
dx.doi.org/10.1021/la400741b | Langmuir XXXX, XXX, XXX−XXX
Langmuir
Article
inside an inverted beaker that fit a Petri dish containing 20 mL of benzene. The annealing setup was finally hosted in an unsealed chamber that was placed in a hood with a controlled humidity of 30% at room temperature for 48 h to create the polymer nanotemplate. Further details about the preparation of the polymer template are described elsewhere.46 2.2. Real-Space Surface Characterizations. Topographical and phase images of the polymer and metal−polymer hybrid films were obtained by using an AFM AutoProbe CP research instrument. Microfabricated V-shaped silicon cantilevers with a spring constant k = 3.2 N/m and a resonance frequency of ∼80 kHz with a silicon conical tip of a typical radius of 10 nm were used. Areas of 1 × 1 μm2 were scanned under constant applied force conditions (non-contact mode). All AFM images were collected in air at ambient conditions. 2.3. DC Sputtering System. An UHV sputtering system, including two gas inlets, three dc magnetron sputter sources, and a load/lock system, has been designed to fit into synchrotron radiation instruments for in-situ and real-time experiments. The sputtering chamber has two beryllium windows, making GISAXS measurements feasible. More details on the sputtering setup are provided elsewhere.36,47 The main chamber is pumped down by a two-stage turbomolecular pump that brings the chamber into a base pressure of 5 × 10−8 mbar. Prior to each deposition experiment, the target was cleaned by sputtering for a short time while its shutter was closed to prevent any deposition during this cleaning step. The deposition was performed at an argon pressure of 5 × 10−3 mbar and at a rate of 0.8 nm/min. All parts of the system, including the sputter guns, shutters, and gas inlets, were controlled by a computer that allowed for precise and reproducible control of the sputtering process and also provided a complete remote control outside the inaccessible X-ray hutch. 2.4. Grazing Incidence Small-Angle X-ray Scattering (GISAXS) Measurements. The GISAXS measurements were carried out at beamline BW4 of the DORIS III storage ring48 at DESY (Hamburg). In GISAXS, the incoming X-ray beam impinges on the sample surface at a small incidence angle αi, and the scattered signal is collected using a 2D detector. The selected wavelength was λ = 0.138 nm. The pathway of the X-ray beam was fully evacuated, and the beam was focused to the size of 45 × 25 μm2 (h × v) by using an assembly of beryllium lenses. The X-ray beam shape in and out of the plane of reflection was set by high-quality entrance cross-slits. A twodimensional (2D) detector (Pilatus 100k; 487 × 195 pixels) was placed at a distance of LSD = 2.01 m from the center of rotation of the sputtering chamber, where the sample was located. A point-like movable beam stop in front of the detector was used to block the highintensity geometrical X-ray beam specular reflection. The sample inside the sputtering chamber was placed horizontally (xy-plane) and at an incidence angle αi = 0.4° with respect to the incidence X-ray beam. This selected incidence angle is well above the critical angles of the polymer film, cobalt metal, and the substrate (αc(PS) = 0.138°, αc(PEO) = 0.144°, αc(Co) = 0.354°, αc(Si) = 0.202°). Therefore, the material characteristic Yoneda49 peaks of the surface components and the geometrical specular peak are well separated on the 2D detector. At this angle of incidence, both the surface and bulk nanostructures of the polymer film are accessible. Structural information is obtained with horizontal (qy direction) and vertical (qz direction) cuts of the 2D intensity distribution (q/pixel = 3.55 × 10−3 nm−1). The intensity was integrated over a width of 10 pixels for each line cut to improve statistics. The qy cut is oriented parallel to the sample surface at the qz position of maximum intensity, and the qz cut is oriented perpendicular to the surface plane at the qy position of the maximum interference peak position (at qy = 0.20 nm−1). The horizontal cut (qy) provides lateral structure information, including particle shape, as well as the particle size and spatial distributions.45,50 The qy cut is normally termed an out-of-plane GISAXS cut in reference to a cut that is normal to the plane of incidence beam. In the real-time sputtering experiment, 1 s time-scale GISAXS measurements were collected. After 30 min sputtering time we collected in total 1800 2D GISAXS images of the metal/polymer hybrid film while the Co atoms were simultaneously deposited at a rate of 0.8 nm/min. Because of the limited dimension (487 × 195 pixels)
conditions that are far from thermodynamic equilibrium.15 In wet chemical synthesis processes, challenges that are attributed to residual solvent may influence the nanoparticles’ aggregation behavior. To gain a better understanding of metal atom aggregation and particle formation/growth within polymer matrix, the physical vapor deposition (PVD) technique is the appropriate choice. The metal atoms in the gas phase deposit on the polymer surface, followed by particle formation and growth as the deposited metal concentration increases. Also, the PVD process practically yields the advantage of good conformity on a variety of complex topographies. Previous studies have shown that metal nanopatterns can be achieved by the preferential adsorption of metal atoms on a particular nanodomain in a diblock copolymer film.34−39 Given the complexity of the structures of metal−polymer nanocomposites in terms of dispersion, spatial organization of particles, particleinduced modification of polymer morphology, and possible chemical interactions, the rational design of metal nanostructures in a polymer matrix requires a broad, fundamental understanding of the mechanistic aspects of metal assembly in polymers. In this article, we investigate the deposition behavior of cobalt metal on a microphase-separated polystyrene-blockpoly(ethylene oxide) P(S-b-EO) diblock copolymer thin film in a real-time mode. We employ a portable, remotely controlled dc magnetron sputter deposition system that is mounted on a synchrotron radiation instrument. The latter instrument allows for grazing incidence small-angle X-ray scattering (GISAXS)40−44 measurements to probe cobalt nanoparticle formation and growth on the nanostructured polymer template. A 1 s time scale in-situ GISAXS measurements method is used as a primary tool to gain real-time kinetic information on the nanoparticles’ formation/growth in the bulk and on the surface of the polymer template. Also, the kinetics of cobalt metal film growth is investigated. As a complementary real space surface technique, atomic force microscopy (AFM) is used to probe structures of some selected hybrid cobalt−polymer nanocomposite films. This article has the following structure: After a description of the investigated samples and a brief introduction to the experimental methods that were employed, the established morphologies of the polymer template and the selectivity characteristics of the Co metal on the polymer template are presented and discussed. Next, the results of the real-time intensity evolution of diffuse X-ray scattering upon Co metal deposition on the polymer template as well as the corresponding simulated data are discussed. The article concludes with a summary of the results.
2. EXPERIMENTAL SECTION 2.1. Substrate Cleaning and Polymer Template Creation. Silicon substrates (Si 100, n-type, Silchem) were cleaned via sonication in dichloromethane at 35 °C for 15 min, water-rinsing for 5 min, and then soaking in the cleaning bath at 80 °C for 15 min. The cleaning solution was composed of 100 mL of 96% H2SO4, 35 mL of 35% H2O2, and 65 mL of deionized water.45 The cleaned substrates were further rinsed in deionized water for 10 min and finally spin-dried. A diblock copolymer polystyrene-block-poly(ethylene oxide), which is denoted P(S-b-EO), with a total number-average molecular weight of Mn = 26.5 kg/mol, a weight ratio of 75:25 (PS:PEO), and a polydispersity index (Mw/Mn) of 1.05 was purchased from Polymer Source Inc. The polymer was dissolved in benzene at 10 mg/mL. The precleaned silicon substrates were coated using the spin-coating method (2500 rpm, 30 s). The polymer films were solvent-annealed B
dx.doi.org/10.1021/la400741b | Langmuir XXXX, XXX, XXX−XXX
Langmuir
Article
of the Pilatus 100k detector, two identical measurements with different detector orientations were performed in order to cover a large q-range in both qz and qy directions. In the first measurement the detector’s long axis (487 pixels) was aligned parallel to the normal of the sample surface (qz direction), and in the second measurement, the detector was rotated by 90° to orient it parallel to the sample surface (qy direction). The individual 2D GISAXS data with 1 s acquisition time represent a compromise between sufficient counting statistics and high time resolution for the in-situ experiment. A second experiment on particlefree BC film was performed using another cobalt deposition mode called stop-sputter method to provide evidence for the existence of all low-intensity characteristic scattering peaks observed from the realtime (1 min) measurements. In this stop-sputter mode a short-time metal sputtering step is followed by a long-time (of about 30 min) GISAXS measurement. With such long counting time and very high statistics the reported low-intensity signals of the real-time measurements are confirmed. 2.5. Simulation. To extract more information regarding the structures of the growing metal particles and its correlation to the polymer film morphology, a simulation was performed using the software package IsGISAXS.51 Computation of the scattered intensity was based on the distorted wave Born approximation (DWBA), assuming nanostructured film composed of polymer domains positioned on the solid silicon support at a fixed periodic distance of 31 nm. Upon metal deposition, it is assumed that box-like-shaped metal particles are continuously growing with dimensions: width (2R), height (H/R= 1). The length of the particles are initially used at a very high asymmetric shape at W/R = 10 that exponentially changes with time to W/R = 1. Gaussian distributions were assumed for the variance of the particle radius ΔR/R, width (ΔW/W), and the height (ΔH/H). The interference function (structure factor S(q)) is the FT of the object position autocorrelation function. The particle growth (R) was assumed to be exponentially growing with metal deposition time (i.e., metal nominal thickness). Here, we have used the local monodisperse approximation (LMA) together with a one-dimensional paracrystal. A parameter of a cumulative disorder, ω, inducing a loss of long-range order, is used. The probability densities of the particle positions in the 1D paracrystal (1DDL) is calculated using a triangle function.51
Figure 1. Selected 2D GISAXS data (in qz vs qy) of cobalt-free polymer film (bottom left image) and cobalt−polymer nanocomposite films at different cobalt deposition times as indicated from 100 up to 1700 s. The detector long-axis is oriented parallel to the normal of the sample plane (along the qz direction). The intensity is shown on a logarithmic scale. The qy and qz directions of all 2D GISAXS data are indicated on the lower left corner image.
3. RESULTS 3.1. Morphology of the Polymer Template. Upon annealing the P(S-b-EO) diblock copolymer film in a benzene vapor flow, highly oriented parallel crystalline poly(ethylene oxide) (PEO) domains that are sandwiched between two polystyrene (PS) domains are obtained. Nanostructured areas (3−7 μm in size) are located within large random breakout crystallites within the polymer film. The full characterization of the polymer film morphology serving as the template in this investigation was reported elsewhere.46 As indicated in Figures 1 and 2, first-order diffraction peaks of the cobalt-free polymer template on both sides of the specular beam stop are observed in the 2D GISAXS data (lower left corner image). These intensity maxima at qy,max = 0.20 nm−1 correspond to a periodic distance (D) of 31 nm between the crystalline PEO cylinders in the diblock copolymer film. The value of the periodic distance D is calculated using the simple equation D = 2π/qy,max. In the GISAXS measurement, the footprint of the X-ray beam on the sample surface along the beam direction is ∼3.6 mm, which guarantees a nonlocal averaged value of the nanoscale periodic distances. Generally, the interaction of X-rays with matter depends on the complex refractive index n expressed by n = 1 − (λ2/2π)rel+i(λ/4π)μ, where rel is the scattering length density, re is the classical electron radius, l is the electron density, μ is the linear absorption coefficient, and λ is the wavelength. The value (λ2/2π)rel is the real part (δ) of the X-ray refractive index of the investigated matter.52 The small difference in the X-ray
scattering length density of 1.4 × 10−4 nm−2 between the microphase-separated PS block (9.60 × 10−4 nm−2) and PEO block (8.2 × 10−4 nm−2) of the polymer template is sufficient to produce an order of magnitude difference in the X-ray scattering amplitude. The single material scattering length density is calculated from rel( f1 + if 2).52 The X-ray scattering factors ( f1, f 2) are found in the Henke X-ray scattering factors database at the Lawrence Berkeley Laboratory Center for X-ray optics or in the tables given by Henke and co-workers.53 As a consequence, already after 1 s acquisition time in GISAXS the first-order Bragg peaks (located at qy = 0.20 nm−1) related to the polymer template structure are visible (low left corners images in Figures 1 and 2). The observation of these lowintensity characteristic scattering maxima is important to understand the polymer morphology correlation to the cobalt nanoparticles formation and growth. It is worth mentioning that GISAXS measurements with a significantly increased acquisition time (for example, up to 30 min) of the Co-free polymer template46 allow the detection of higher order Bragg peaks (qn/q1 = 2) which demonstrate the highly oriented polymer nanostructure (image not shown). However, such long data acquisition time is not matching with the needs of timeresolved in-situ investigations. C
dx.doi.org/10.1021/la400741b | Langmuir XXXX, XXX, XXX−XXX
Langmuir
Article
Figure 2. Selected 2D GISAXS data (in qz vs qy) of cobalt-free polymer film (bottom left image) and cobalt−polymer nanocomposite films at different cobalt deposition times as indicated from 100 up to 1700 s. The detector long-axis is oriented parallel to the sample plane (along the qy direction). The intensity is shown on a logarithmic scale. The two white arrows (in the top right corner image) indicate the two intensity maxima in the qy direction corresponding to first- and second-order Bragg peaks. The qy and qz directions of all 2D GISAXS data are indicated on the lower left corner image.
3.2. Cobalt Metal Layer Formation. The vertical line cuts (qz) at position of the maximum interference peak position (qy = 0.20 nm−1) of the 2D GISAXS data are shown in Figure 3a. Also, the time evolution of the scattering pattern that emphasize the metal deposition behavior on the soft nanostructured polymer film as a function of time is also represented in time mappings (Figure 4a,b). The time mapping of the qz cuts (at constant qy value of 0.20 nm−1) is shown in Figure 4a. At deposition times shorter than 400 s, the vertical cuts from the 2D GISAXS data (Figure 3a) and the time mapping of the qz profiles (Figure 4a) do not show oscillations (i.e., no Kiessig fringes). Such a result suggests that cobalt does not form a smooth metal layer on the polymer surface up to a nominal cobalt metal thickness of 5.3 nm. Upon further metal load, a thin quasi-uniform cobalt metal layer is formed, and the metal layer thickness increases monotonically as indicated by the formation of more regularly spaced intensity oscillations along the qz direction (Figure 3a). These Kiessig fringes are marked with arrows in Figure 4a. From the Lorentzian fits of these Kiessig fringes the effective metal layer thickness can be evaluated. The linear fit of the effective thickness data points and a reference dotted line of the nominal Co thickness are plotted in Figure 5a. From Figure 5a, a sputtering rate 0.8 nm/ min is consistent with the sputtering conditions employed in our experiment. However, the effective thickness is a line with a nonzero intercept. From Figure 5a, it is assumed that about 2.5
nm thick dispersed Co metal particles are buried inside the polymer film. In Figure 3a, the critical angles (Yoneda peaks) of both cobalt and polymer are marked with dashed lines. Relative qualitative information about the surface coverage are achieved by Lorentzian fitting the Yoneda peaks using the software package DPDAK.54 For simplicity, two Lorentzian peaks are used: one peak at an average peak position of the critical angle value of 0.160° (i.e., the average value for PS, PEO, and Si) and a second peak at the critical angle of the Co metal (0.354°). Fitting all individual Yoneda peaks for all surface materials is not possible due to lack of well-resolved separated peaks for all constituents. As well, due to broad flat nature of the materials’ characteristic peaks in the deposition time 500−1300 s, the fitting shows poor correlations; thus, the data points at 500− 1300 s are excluded in Figure 5b. From Figure 5b, at ∼400 s deposition time, the characteristic average critical angle of Co metal is overwhelming the surface as indicated from the area ratio of cobalt Yoneda peak (∼0.8). It is assumed that the area of the material characteristic peak (Yoneda peak) is related to the volume fraction of a particular component multiplied by its scattering power. Thus, taking into account the scattering power of the surface constituents, the metal surface coverage is about 10%. From the real-time mapping of the vertical cuts qz (Figure 4a) it is seen that the first intensity oscillation already developed, and further fringes are formed as the deposition D
dx.doi.org/10.1021/la400741b | Langmuir XXXX, XXX, XXX−XXX
Langmuir
Article
Figure 3. (a) Vertical cuts (qz) at a constant qy of 0.02 nm−1 and (b) horizontal cuts (qy) at constant qz value from the 2D GISAXS data for both cobalt-free and cobalt−polymer nanocomposite film at different cobalt sputtering deposition times from 100 up to 1800 s (from bottom to top). The dotted lines in the qz plot indicate both the average critical angle of the polymer film (qz,c = 0.45 nm−1) and that of cobalt metal (qz,c = 0.61 nm−1). The dashed line in the qy plot shows the lower resolution limit, while the dotted line represents the width of the diffuse scattering near the specular reflection plane. The arrows in the qy plot indicate the first- and second-order characteristic scattering peaks.
Figure 5. (a) Nominal metal thickness (dashed line) and the effective metal thickness (solid line). (b) Evolution of the area under Yoneda peaks assuming one peak for an average critical angle value of 0.160° (i.e., the average value for PS, PEO, and Si) and a second peak for the Co metal (0.354°). (c) GISAXS peak position in the qy direction. The dashed line is a guide line for the eye representing the behavior of the nanoparticles characteristic qy values as a function of deposition time. (d) Full width at half-maximum of the metal nanoparticle characteristic qy peaks as a function of deposition time. The inset shows a single qy profile after 300 s that emphasis the two Bragg peaks as well as the metal characteristic broad peak.
time increases (>400 s). These results conclude that, at a nominal film thickness 1.2 nm−1, Figure 4b) and moves toward lower qy values before it levels off at a constant qy (marked with white dashed-line arrow in Figure 4b). The latter broad peak is relevant to the formation of cobalt nanoparticles (see the inset in Figure 5d). The growth of these particles is indicated by the shift in the qy peak that finally reaches a constant qy value. During the metal deposition on bare silicon surfaces a similar behavior has been recently reported55 where the out-of-plane peak first emerges at a very large qy value (small distance) and then shifts toward smaller qy values (large distance) before it stays at some constant qy values. Generally, as the metal atoms arrive at the surface, the adatoms diffuse on the surface, and when reaching a short mutual distance, they can form stable nuclei. With increasing metal load on the surface, large numbers of nucleation sites are formed. The small metal islands are also mobile on the surface as long as their sizes are below a critical size. When the distance between the two neighboring islands is less than the diffusion length of the small islands, they coalesce, forming larger particles. These continuous coalescence between small aggregates increases the distance between the evolved growing nanoparticles. The large particles are capturing the new adatoms and as the particles are getting larger and larger; two particles meet and form cap-shaped clusters. With a neck between two big particles, a physical connection of the metal parts is created. The metal characteristic peak is weak in intensity but welldistinguished from the background intensity. Its presence is confirmed in additional experiments using the stop-sputter method (see the Experimental Section). In the presented in-
Figure 6. Spatial distance between the cobalt particles as a function of the nominal film thickness. The dashed line indicates a scaling law Dm ∼ d0.60.
interparticle distance of 7.3 nm is developed before the formation of a quasi-uniform top metal layer on the polymer film. The metal−metal correlation distance (Dm) is found to progressively vary with the metal upload according to the scaling law Dm ∼ dα. As indicated in Figure 6 a scaling factor α of 0.6 is obtained. The Dm value increases up to 7.3 nm via metal coalescence process, and then after 400 s it levels off where already formed metal nanoparticles acts as nuclei for further metal growth. The interparticle distance of 7.3 nm is comparable with the assumed percolation threshold of cobalt (7 ± 1 nm).28
4. DISCUSSION When metal atoms in their atomic state are deposited on the polymer surface, they may perform a random walk on the surface or diffuse inside the polymer bulk. An attractive metal− polymer interaction that is mediated by a sticking coefficient S determines the ability of the metal to wet the polymer surface. The sticking coefficient is the ratio of the number of adsorbed metal atoms to the total number of metal atoms that arrive on the surface.30 Quantitative values for metal-sticking coefficients on polymer surfaces have been previously30 obtained through the application of X-ray photoelectron spectroscopy (XPS) and F
dx.doi.org/10.1021/la400741b | Langmuir XXXX, XXX, XXX−XXX
Langmuir
Article
of the polymer film is favored over the possible growth of the particle inside the polymer bulk. To date, selective metal decoration on one block of a diblock copolymer has resulted in well-isolated nanoparticles instead of continuous arrays of metal nano-objects. Thus, the formation of large nanoparticles on the polymer surface and their progressive growth would only allow the formation of dense chains of closely spaced but separated nanoparticles.56 AFM images are collected for some cobalt metal−polymer hybrid films. The surface profile of the hybrid film at a deposition of a nominal cobalt thickness of 4 nm is shown in Figure 7. From Figure 7, cobalt decoration of the PS
the radiotracer method. Generally, the sticking coefficient of metal tends to follow the surface energy of the polymer surface.31 The lower the polymer surface energy, the higher the sticking coefficient. However, as indicated in previously published investigations, the difference in sticking coefficients alone is not enough to justify the formation of metal nanopatterns on microphase-separated nanostructured polymer surfaces. For instance, although the surface energies of both PS (40.7 mN/m) and PMMA (41.1 mN/m) are quite the same, the noble metal decoration of the polystyrene blocks in polystyrene-block-poly(methyl methacrylate) P(S-b-MMA) diblock copolymer thin films has been reported.56 Therefore, the sticking coefficients may not solely explain the high selectivity of the cobalt metal toward the PS block of our P(S-b-EO) diblock copolymer template. One important parameter that may control the selectivity and, hence, the aggregation of cobalt metal inside a certain polymer domain is the high mobility within the polymer matrix.7,29,57−59 The deposited cobalt atoms on the polymer surface are mobile and may diffuse inside the polymer matrix or on the polymer surface. On their diffusion paths, they coalesce at the polymer surface and in the polymer bulk to form metal nanoclusters. The evolved nanoclusters are also diffusive in the polymer matrix if their sizes do not exceed a critical dimension. In a recent study59 it has been approved that if the entanglement mesh length (dt) of the polymer melt is larger than the nanoparticle radius (R), then the particle diffuses ∼250 times faster than predicted by the viscosity-driven Stokes−Einstein (SE) equation. However, if the metal particle radius approaches the polymer radius of gyration (Rg), then the particle diffuses as predicted by the SE theory.59 It is also assumed that the metal nanoclusters with R < dt < Rg are not randomly distributed but may diffuse and reside in preferred sites in the polymer matrix. The reason why the metal nanoclusters reside at preferential sites may be attributed to the chemical affinity of the polymer chains toward the metal nanoclusters or because of an attractive local arrangement of the polymer chains.33 As a consequence, we attribute the cobalt metal selectivity toward the PS domains to the possible high mobility of cobalt metal atoms at the PS matrix. On their diffusion path in the PS domains, cobalt atoms’ aggregate and the nanoparticles formation are then favored. However, on the PEO block, the cobalt metal atoms have limited mobility because they are sequestered by the chelating oxygen atoms of the PEO chains as well as by the limited free volume within the crystalline PEO domains. The nanoparticle formation and growth within the PS domains are indicated by an increase in the GISAXS intensity of the characteristic polymer rod-like scattering maxima upon cobalt deposition (Figures 1−3). Though, our experimental method does not directly address the diffusion property of the metal particles or investigate mobility aspects, previous investigations have experimentally proved that the diffusion of metal nanoparticles in a glassy polymer matrix occurs even at room temperature.29,60 At higher metal coverage, the already formed metal nanoparticles on the polymer surface are effectively attracting the freshly deposited metal atoms. At this stage, the metal sticking coefficient on the surface is approaching unity, which is typical for metal deposition on metal surfaces.31 Therefore, the possible high mobility at the PS domains results in faster formation of stable metal nuclei that grows at faster rate (Figure 6) upon further sputtering due to the high metal−metal sticking coefficient. The growth of larger particles at the surface
Figure 7. AFM (a, c) topography and (b, d) phase images of the (a, b) cobalt-free polymer film and (c, d) cobalt−polymer nanocomposite films after cobalt sputtering of nominally 4 nm. The visible structure originates from the microphase separation of the P(S-b-EO) diblock with a weight ratio of 75:25 (PS:PEO). The nanocylinder (dark) of PEO is in a PS matrix (bright). After cobalt metal deposition (c, d) the phase image (d) clearly indicates a selective metal deposition at the PS domains.
domains is indicated by the formation of aggregates of relatively uniform nanoparticles along the PS domains. The formation of 1D nanoparticles aggregates may be attributed to direct interparticle van der Waals attractions that favor particle assembly along the PS domains. A simulation (Figure 8) is performed to get insights into the particle growth correlation with the morphology of the polymer template. Upon metal upload, the first stage corresponds to nucleation of islands, the second to island growth, and the last is the coalescence of neighboring islands. The current GISAXS conditions are probing the island (particle) sizes below about
