Article Cite This: Macromolecules XXXX, XXX, XXX−XXX
pubs.acs.org/Macromolecules
Coupling HAADF-STEM Tomography and Image Reconstruction for the Precise Characterization of Particle Morphology of Composite Polymer Latexes Noushin Rajabalinia,† Shaghayegh Hamzehlou,† Evgeny Modin,‡ Andrey Chuvilin,‡,§ Jose R. Leiza,† and José M. Asua*,†
Downloaded via NOTTINGHAM TRENT UNIV on July 22, 2019 at 21:15:46 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.
†
POLYMAT, Kimika Aplikatu saila, Kimika Fakultatea, University of the Basque Country UPV/EHU, Avda Tolosa 72, 20018 Donostia-San Sebastián, Spain ‡ CIC Nanogune, Avda Tolosa 76, 20018 Donostia-San Sebastián, Spain § Ikerbasque, Basque Foundation for Science, 48013 Bilbao, Spain S Supporting Information *
ABSTRACT: The performance of waterborne polymer− polymer composite materials strongly depends on the particle morphology, but the precise quantitative characterization of complex morphologies is a long-standing problem. In this work, high angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) has been used to obtain a 3D quantitative characterization of the morphology of polymer−polymer particles. The potential of this technique was demonstrated in the study of the effect of process variables (Tg of the seed, type of initiator, and instantaneous conversion of the monomer) on particle morphology during the semicontinuous emulsion copolymerization of styrene and butyl acrylate on methyl methacrylate-rich seeds. This analysis unveiled new mechanisms of morphology development.
1. INTRODUCTION
Interestingly, both the experimentally driven approach and the model-based one assume that the particle morphology can be unambiguously defined and determined. Whereas this can be relatively easy for the equilibrium morphologies (core− shell, inverted core−shell, and hemispherical),26 the nonequilibrium morphologies are more difficult to define as all of the particles are different. This is the case of the particles in Figure 1 that at first sight can be defined as multilobbed, but this only describes the general shape without giving any quantitative detail about the morphology (size, number, and position of the lobes). It has been recently proposed that the morphology of the whole composite latex can be described by the size distribution of clusters at different positions in the particle.24 Practically, to have these distributions, a precise characterization of the particle morphology is needed, which is difficult to obtain with the techniques currently used as discussed below. A wide range of techniques have been used to determine the particle morphology. Information about surface composition can be obtained by surfactant titration.27 However, this technique requires an accurate determination of the surface area of the particles, which is not always available for polydispersed latexes and for nonspherical particles as shown
Structured polymer−polymer composite particles provide properties that cannot be attained by conventional homogeneous particles and allow targeting new applications.1 These waterborne products are mainly synthesized by seeded semibatch emulsion polymerization although the use of miniemulsion polymerization may be advantageous when the particles should contain step-growth polymers or polymers derived from highly water-insoluble monomers.2 The production of particles with well-defined morphology is of great interest, as application properties strongly depend on the morphology of the synthesized structured latex.3−5 This is a challenging goal because the morphology is created during the polymerization as a result of complex interactions between kinetics and thermodynamics involved in the polymerization.6−17 Currently, the production strategies used commercially are largely based on extensive experimental work guided by a rich literature on the effect of the operation variables on the final particle morphology.3,4,8,18−23 However, it is open to discussion whether this approach would be sufficient in a search for a globally optimal strategy. Thus, by use of a mathematical model for the process,24 it has been recently demonstrated in silico that the optimal strategies may involve complex profiles of temperature and monomer feed rate,25 which are unlikely to be obtained by the currently used approach. © XXXX American Chemical Society
Received: April 17, 2019 Revised: June 6, 2019
A
DOI: 10.1021/acs.macromol.9b00787 Macromolecules XXXX, XXX, XXX−XXX
Article
Macromolecules
damage the polymer affecting resolution. On the other hand, reliable interpenetration of low-loss EELS can only be based on characteristic spectral features associated with specific polymer structures. Therefore, in practice, the technique is limited to polymers containing either Si or aromatic rings. Combinations of EELS and imaging based on spectroscopic contrast with either scanning transmission electron microscopy (STEM) or energy-filtering (EFTEM) techniques have been used to study several aspects of multiphase morphology in polymer−polymer and polymer−inorganic systems.44−52 Thus, Libera and co-workers53−55 have used EELS in cryo-STEM to study the composition of the interface between PDMS and poly((meth)acrylates) in structured particles synthesized by seeded emulsion polymerization. However, they only analyzed a relatively simple Janus-like morphology. The main limitation of above-mentioned imaging techniques is a generic 2D nature of obtained data for an object that is actually 3-dimensional. This creates an ambiguity in interpretation and statistical analysis. Electron tomography provides 3D information from a tilt series of 2D projections obtained via TEM or STEM imaging.56 TEM tomography was used to analyze the morphology of organic/inorganic nanocomposites in the bulk state57,58 and hybrid polymer/inorganic particle latexes.42,59 High angle annular dark-field (HAADF)-STEM tomography, which was formerly used to determine the 3D structure of inorganic specimens,60 has started to find its application in the characterization of complex polymer systems. Thus, it has been used to determine the distributions of nanoparticles in inhomogeneous matrices,61 to study the spatial organization of thin film of various polymer systems including rubber blend and semicrystalline polyethylenes,62 and to image networks of nanoparticles within polymer− nanoparticle blends in photovoltaic devices.63 Despite all efforts devoted to the development of characterization techniques for particle morphology, the precise determination of the morphology of the waterborne composite polymer−polymer particles remains elusive, especially when dealing with complex morphologies. In this work, the particle morphologies of structured polymer−polymer latex particles were quantitatively determined in 3D by means of HAADF-STEM tomography of selectively stained latexes. Fiducial-less tilt-series alignment and tomographic reconstructions with weighted back-projection (WBP) and simultaneous iterative reconstruction (SIRT) techniques64 were implemented by using in-house-developed software. The characterization technique was used to analyze the effect of different reaction parameters such as glass transition temperature (Tg) of the seed, type of initiator, and the monomer concentration during the semicontinuous emulsion copolymerization of styrene and butyl acrylate on methyl methacrylate-rich seeds. The precise characterization provided unexpected insights into the mechanisms involved in the process.
Figure 1. TEM of sample case A1 stained with RuO4.
in Figure 1. Minimum film formation temperature (MFFT) and environmental scanning electron microscopy (ESEM) can also provide some idea about the surface composition because the temperature at which the particles start to coalescence gives an indication of the glass transition of the polymer at the surface.28 However, MFFT and ESEM have only a limited value for particles having different phases at the surface, e.g., for hemispherical particles. Differential scanning calorimetry (DSC) has been used to quantify the amount of interfacial material between two polymeric phases in blends and latex films.29,30 This information is contained in the plot of dCp/dT vs temperature obtained from the first heating cycle in the DSC experiment. In a fully separated two-phase polymer−polymer system the DSC curve is formed by narrow peaks with no signal in the region between the peaks. The peaks in this plot are associated with the Tgs of the pure polymeric phases. Phase intermixing results in broader peaks that often are closer than those of the pure phases and in the presence of positive values in the region between the peaks.31 The microstructure of the particles is often characterized by direct observation of the particles using transmission electron microscopy (TEM). The internal morphology of the particles can be observed by the cross section of the particles embedded in a resin.32 Scanning transmission electron microscopy (STEM) deposits less electron dose on the sample compared to TEM and thus is more suitable for analyzing the soft polymer phases which are beam-sensitive.33 The contrast of the polymer phases often is not strong enough in polymer− polymer systems for a good distinction between the phases. This problem is often addressed by selective staining34−37 and also by using advanced optical techniques such as defocusing, holography, and Zernike phase plate methods that increase the contrast between phases.37,38 This problem is much less pronounced for polymer−inorganic systems due to the relatively higher electronic density of the inorganic material that enhances the contrast between the polymer phase and the inorganic material.39−43 Other techniques are also available. Electron energy-loss spectroscopy (EELS) uses the inelastic interaction between the energetic electrons and materials to determine their relative compositions. Core loss and low loss EELS might in principle be used. However, core loss is a high-dose technique that can
2. SYNTHESIS OF POLYMER−POLYMER COMPOSITE LATEXES AND CHARACTERIZATION OF THE PARTICLE MORPHOLOGY 2.1. Materials. Styrene (S), methyl methacrylate (MMA), and butyl acrylate (BA) from Quimidroga as well as acrylic acid (AA) and acrylamide (AM) from Aldrich were used as received. Sodium persulfate (NaPS, Fluka) and azobisisobutyronitrile (AIBN, Aldrich) were used as thermal initiators, and tert-butyl hydroperoxide (TBHP, Aldrich) and sodium acetone B
DOI: 10.1021/acs.macromol.9b00787 Macromolecules XXXX, XXX, XXX−XXX
Article
Macromolecules Table 1. Summary of the Experiments Performed To Synthesize Composite Latex Particles cases group A
group B
A1 A2 A3 A4 B1 B2
seed monomer
Tg (°C)
second stage monomer
Tg (°C)
Treaction (°C)
initiator
instantaneous conversion (%)
MMA/BA/AA/AM MMA/BA/AA/AM MMA/BA/AA/AM MMA/BA/AA/AM MMA/BA/AA/AM MMA/BA/AA/AM
86 63 46 86 86 63
S/BA/AA/AM S/BA/AA/AM S/BA/AA/AM S/BA/AA/AM S/BA/AA/AM S/BA/AA/AM
49 45 46 48 48 48
80 80 80 80 80 65
TBHP+ACBS TBHP+ACBS TBHP+ACBS AIBN TBHP+ACBS TBHP+ACBS
>90 >90 >90 >90 70a 70a
a
Based on second stage monomer.
lands) operated at acceleration voltage of 300 kV. The HAADF-STEM imaging mode provides the contrast that is strongly dependent on the atomic number (∼Z2), and thus the stained polymer phase looks much brighter at HAADF-STEM images. Tilt series were acquired automatically at angles between −74° and +74° at 2° tilt step. To reduce beam-damage effects, reasonably low-dose conditions were set up, and some sensitive samples were imaged at cryogenic temperature by using a cryotomographic sample holder (Gatan, model 914) cooled by liquid nitrogen. Images were taken with a FEI Tomography 4.0 software in automatic mode; the dwell time for acquisition was set to 20 μs for the images of 1024 × 1024 pixels. The fiducial-less tilt-series alignment and tomographic reconstructions with weighted back-projection (WBP) and simultaneous iterative reconstruction (SIRT) techniques were done using in-house DigitalMicrograph (Gatan, USA) scripts (see the Supporting Information for details). Reconstructed volumes had a voxel size of ∼2 × 2 × 2 nm3. For the stained phase separation, the intensity-based segmentation with a local threshold criterion and manual supervision was used. Depending on the intensity of pixels, they were assumed as belonging to the feature of interest which is stained phase (bright) or belonging to the matrix (dark). An example is presented in Figure S1 of the Supporting Information. Segmentation of different phases in the particles, subsequent 3-D rendering, and statistical calculations were done by using FEI Avizo 8.1 software.
bisulfite (ACBS, BASF) were used as a redox pair initiator. A mixture of ionic (sodium lauryl sulfate (SLS) from Aldrich) and nonionic (Emulan OG from BASF) surfactants was used in the formulation. Deionized water was used in the emulsions, and hydroquinone (Aldrich) was used for stopping the reaction in the samples withdrawn from the reactor. EPoFix resin (Struers) was used as cold mounting system in sample preparation for cross-section characterization by TEM. 2.2. Composite Latexes Used. In this work, six composite latexes presented in Table 1 were used. All of them were synthesized in a seeded semibatch emulsion polymerization process. The seeds were synthesized in a glass reactor using the formulation given in the Supporting Information. Latexes A1, A2, and A3 were synthesized with seeds of different glass transition temperature (Tg,A1 = 86 °C, Tg,A2 = 63 °C, and Tg,A3 = 46 °C) achieved by varying the ratio of MMA and BA. A monomer ratio S/BA = 67/31 w/w was used in the second stage of the process. Details of the synthesis can be found in the Supporting Information and in ref 65, where particle morphologies determined by conventional TEM were used to validate a mathematical model. Latex A4 was similar to latex A1, but an oil-soluble initiator (AIBN) was used to modify the radical concentration profile in the particles. The latexes of group A were synthesized under starved conditions. The cases of group B were synthesized at a lower constant instantaneous conversion (70%) based on the total polymer and monomer of the second stage. For this, the reactions were performed in a calorimeter reactor (RTCal, Mettler Toledo) that allows online monitoring and control of the instantaneous conversion based on the heat released (see the Supporting Information for more details). For both the seeds and the second stage polymers, 1 wt % of functional monomers (acrylic acid and acrylamide) was used to improve the colloidal stability. The seeds were more hydrophilic than the second stage polymers; therefore, the equilibrium morphology was inverted core−shell in all cases. 2.3. Characterization of the Particle Morphology. Transmission electron microscopy (TEM) was performed with a Tecnai G2 electron microscope (FEI Company, Netherlands) at 200 kV. One droplet of diluted latex with deionized water with 0.05 wt % solids content was placed on a carboncoated copper grid and dried at ambient temperature. TEM samples were stained with RuO4 vapor for 1 h to increase the contrast of the styrene-containing component. To obtain the cross sections, the films cast from latexes were embedded in epoxy resin and were cured for 12 h at room temperature. The crosscut slices with 70 nm thickness prepared by microtome were collected on carbon-coated copper grids and stained with vapor of RuO4 for 1 h. Three-dimensional particle morphology was characterized by the HAADF-STEM electron tomography technique using a Titan 60-300 electron microscope (FEI Company, Nether-
3. RESULTS AND DISCUSSION Figure 2 presents a series of slices of the reconstructed composite polymer particle for case A1. Note that in HAADF STEM the styrene-rich phase appears brighter than the methyl methacrylate-rich phase as discussed earlier. It can be clearly seen that the polymer particle presents styrene-rich lobes close to the surface of the particle. In addition, smaller clusters present in the interior zone of the polymer particles.
Figure 2. Tomographic cross section of case A1 from upper surface of the samples to the bottom obtained from HAADF-STEM. C
DOI: 10.1021/acs.macromol.9b00787 Macromolecules XXXX, XXX, XXX−XXX
Article
Macromolecules
conversion of 94% during the monomer feeding time in the second stage of polymerization). In cases A2 and A3, the Tg of the seed was reduced to 63 and 46 °C, respectively, in an attempt to facilitate the movement of the styrene-rich clusters toward the center of the particle, namely to the equilibrium morphology. It is worth mentioning that in these experiments the average instantaneous conversions during the monomer feeding time in the second stage of polymerization were 98% and 94%, respectively. Figure 5 shows the comparison of 3D reconstructed volumes for cases A1−A3 as well as conventional TEM images of the
Figure 3 shows the reconstructed 3D image of the polymer particle, the matrix, and the clusters. Each PS-rich cluster is
Figure 3. Reconstructed 3D image of polymer particle of case A1. Segmentation of the clusters is made based on local threshold criteria.
shown with a different color to simplify their visual recognition. Segmentation confirms that the latex particle consists of big polystyrene lobes at the surface of the particle and a number of small clusters in the interior region of the particle. Figure 3b shows the clusters at the surface and in the interior region as separate volumes. For the sake of comparison, the morphologies determined by TEM for case A1 are presented in Figure 4. It can be seen
Figure 5. Reconstructed 3D image of polymer particle of cases A1− A3 by tomographic analysis of samples by HAADF-STEM and the corresponding TEM images.
stained samples. It can be seen that by decreasing the glass transition temperature of the matrix, from 86 °C to 63 °C and 46 °C, the whole particle became more spherical and the clusters penetrated more toward the center. This change in the particle morphology was due to the lower internal viscosity of the matrix in the softer systems, which led to an easier movement of the clusters toward the equilibrium morphology. In addition, a less sharp profile of radicals is expected in softer matrices because the diffusion of the growing chains was less hindered. The detailed particle morphologies in Figure 5 were further analyzed by 3D statistics, and independent size distributions of the inner clusters and outside lobes were obtained. Figure 6 presents the cluster size distributions for cases A1−A3 for internal and external clusters. Figure 6 shows that as the hardness of the seed polymer decreased, the size of the external clusters increased because cluster aggregation became easier. A close look at the morphologies of cases A2 and A3 reveals an unexpected result. According to the existing views, the morphology of the softer system (case A3) is expected to be closer to the equilibrium morphology (inverted core−shell with the styrene-rich polymer in the core of the particle) than that of harder case A2. However, this is not the case, and there is no visible difference in clusters distribution in the latex particle volume, besides more pronounced lobes on a harder seed polymer in case A2. To verify the difference numerically, the distribution of the amount of the second stage polymer along the radius r of the particles was calculated as
Figure 4. TEM analysis of case A1: (a) TEM of sample stained with RuO4; (b) RuO4 stained crosscut image of a film.
that although they clearly show the presence of clusters near the surface of the particle, the observed morphology is just the 2D projection of the actual morphology in which the location of these clusters inside the particles is difficult to determine, and the statistical evaluation is hardly possible even with the cross sections of the particles (Figure 4b). The results in Figure 3 show that the equilibrium “inverted core−shell” morphology is not attained. The most likely reason for the large fraction of styrene-rich clusters near the surface of the particle is the formation of a profile of radicals in the particle.65 The redox initiator used in the second stage polymerization was TBHP-ACBS which is known to form hydrophobic tert-butoxyl radicals in the aqueous phase. These radicals can directly enter into the polymer particles, where they rapidly react with the monomer present there, forming growing polymer chains that cannot further diffuse toward the center of the particle due to the high viscosity of the matrix. The latter was the result of the combination of the high Tg of the seed polymer (Tg = 86 °C), high molar mass of the forming chains, and low concentration of the monomer in the particles (starved process with an average instantaneous D
DOI: 10.1021/acs.macromol.9b00787 Macromolecules XXXX, XXX, XXX−XXX
Article
Macromolecules
Figure 6. Cluster size distributions for latexes of cases A1−A3: (a) sizes of external clusters calculated as diameters of equivalent sphere for the experimentally measured volumes; (b) sizes of internal clusters, calculated in the same way; (c) volume-weighted average size of external clusters vs sample cases; the softness of the matrix increases in a row A1 < A2 < A3.
F (r ) =
∑r ≤ ∑0 ≤
x 2 + y 2 + z 2 < r +Δr x 2 + y 2 + z 2
