Micron 44 (2013) Contents lists available at SciVerse ScienceDirect. Micron. journal homepage:

Transcription

1 Micron 44 (2013) Contents lists available at SciVerse ScienceDirect Micron journal homepage: Deciphering the three-dimensional morphology of free-standing block copolymer thin films by transmission electron microscopy Frances I. Allen a,b,, Peter Ercius a, Miguel A. Modestino c,d, Rachel A. Segalman c,d, Nitash P. Balsara c,d, Andrew M. Minor a,b a National Center for Electron Microscopy, Lawrence Berkeley National Laboratory, Berkeley, CA , USA b Department of Materials Science and Engineering, University of California, Berkeley, CA , USA c Materials Sciences Division, Lawrence Berkeley National Laboratory, Berkeley, CA , USA d Department of Chemical and Biomolecular Engineering, University of California, Berkeley, CA , USA article info abstract Article history: Received 27 June 2012 Received in revised form 26 September 2012 Accepted 30 September 2012 Keywords: Electron tomography Energy-filtered TEM Block copolymer morphology PS-b-PEO thin films Salt doping Casting solvent Block copolymer thin films with distinct morphologies are prepared by spin casting a nominally lamellar assay of poly(styrene-block-ethylene oxide) from a variety of solvents with and without salt doping. The 3-D morphologies of free-standing thin-film regions, which are obtained by casting directly onto holey substrates, are investigated in detail using various energy-filtering transmission electron microscopy techniques and by electron tomography. Surface characterization is achieved by atomic force microscopy. Our results demonstrate that in order to fully characterize the unique 3-D morphologies of the block copolymer thin films, a multi-method approach is required. When casting from a binary solvent, an unexpected layered honeycomb-type morphology is revealed, which likely results from an expansion of the poly(ethylene oxide) phase. A dramatic effect of selective cation coordination on the morphology of the as-cast block copolymer films is also directly observed. Published by Elsevier Ltd. 1. Introduction The self-assembly of block copolymers into ordered nanostructures is an area of intense research in which thin-film geometries feature widely. Applications of block copolymer thin films include porous membranes for the filtration of biomolecules (Yang et al., 2010), flexible membranes for organic photovoltaics and light-emitting devices (Segalman et al., 2009), and electrolyte membranes for batteries and fuel cells (Orilall and Wiesner, 2011). Targeted selection of the constituent blocks of the copolymer enables the synthesis of new materials with unique properties. Coupled with the rich array of morphologies forming upon phase separation, seemingly inexhaustible technological possibilities result (Bates and Fredrickson, 1999). Since the desired behavior of the block copolymer is intimately linked to its morphology, a deep understanding of the structures forming upon phase separation is required. Consequently, effective techniques for the structural characterization of block copolymers are of crucial importance. Block copolymer phase separation is driven by competing enthalpic and entropic effects, which must be balanced in order for Corresponding author. address: FIAllen@lbl.gov (F.I. Allen). thermodynamic equilibrium to be reached. Since the blocks of the copolymer molecule are joined by a covalent bond, phase separation proceeds on nanometer levels (Bates and Fredrickson, 1990; Bates, 1991). The particular morphology formed in the bulk material is determined by the relative volume fractions of the polymer blocks and by the segregation strength between them, quantified by the Flory Huggins interaction parameter. In the bulk, self-assembly generally manifests as grains of ordered domains arranged randomly throughout the material. In contrast, in thin films interfacial energy effects result in ordering over much longer length scales and can be used to control domain orientations (Marencic and Register, 2010; Fasolka and Mayes, 2001; Albert and Epps, 2010). The domain spacings of the block copolymer can range from several hundred to just a few nanometers in size, hence techniques for the structural characterization of these materials must also be sensitive at the nanoscale. Investigations based on X-ray and neutron scattering effects are routinely implemented, as are several direct imaging techniques. Direct imaging of sectioned bulk specimens and thin films can be achieved by transmission electron microscopy (TEM), with recent advances in instrumentation having resulted in a surge in the use of energy-filtering methods for imaging this class of soft matter (Libera and Egerton, 2010). Energy-filtering enables the generation of image contrast without the need for heavy-element stains, which are often of low selec /$ see front matter. Published by Elsevier Ltd.

2 F.I. Allen et al. / Micron 44 (2013) tivity and are known to produce artifacts. For the direct imaging of block copolymer surfaces, atomic force microscopy (AFM) and scanning electron microscopy (SEM) are further invaluable tools. In this study, poly(styrene-block-ethylene oxide) (PS-b-PEO) thin films (<200 nm in thickness) with distinct morphologies are prepared by spin casting from various solvents with and without salt doping. PS-b-PEO is a favorable candidate for the dry polymer electrolyte in solid-state rechargeable lithium-ion batteries, since it combines a soft phase for ion conduction (PEO) with a hard phase for mechanical strength (PS) (Singh et al., 2007). In current lithium polymer batteries the polymer electrolyte membrane is around 15 m in thickness, yet in order to probe the limits of microbattery fabrication, investigations of much thinner films are also important. The lithium ions in the block copolymer are coordinated by the ether groups of the PEO block and ion transport is mediated by segmental motion of the PEO chains (Druger et al., 1983). It is observed that the phase behavior of the block copolymer is strongly influenced by salt concentration, resulting from a combination of effects including suppression of crystallinity in the PEO phase and modification of the Flory Huggins interaction parameter due to selective cation co-ordination (Epps et al., 2002, 2003; Young and Epps, 2009; Metwalli et al., 2011). Various effects of block copolymer morphology on the ion transport properties of the electrolyte have also emerged (Singh et al., 2007; Mullin et al., 2011). The morphologies of the various PS-b-PEO films prepared here are characterized in detail by TEM. By casting directly onto holey substrates, stable free-standing thin regions ideal for TEM analysis are obtained. This relatively straightforward sample preparation process avoids the need for more challenging preparation techniques such as the transfer of fragile thin films from continuous substrates, or embedding of thin films in resin and subsequent sectioning to obtain electron-transparent segments (Radzilowski et al., 1996). It should be noted, however, that the morphologies of the free-standing films can differ from those of their substratesupported counterparts due to interfacial energy effects (Fasolka and Mayes, 2001). We make extensive use of energy-filtering TEM (EFTEM) to obtain high-contrast elastic images, thickness maps, and chemical maps (Libera and Egerton, 2010; Egerton, 2011; Du Chesne, 1999) of the free-standing thin-film specimens. In addition, tilt series of energy-filtered elastic images are acquired for tomographic reconstructions (Midgley and Weyland, 2003; Jinnae and Spontak, 2009). The results obtained using the various TEM techniques described are then combined to precisely determine the 3-D morphologies of the PS-b-PEO films. To augment the TEM results, surface characterization of the thin films is performed by atomic force microscopy (AFM). 2. Experimental 2.1. Thin-film preparation The films were prepared using a PS-b-PEO diblock copolymer with number molecular weights for the PS and PEO blocks of 74 and 98 kg/mol, respectively, and a polydispersity index of The polymer was synthesized via anionic polymerization and forms a lamellar morphology in the bulk with a domain spacing of 100 nm (Singh et al., 2007). Polymer electrolyte solutions for spin casting were prepared using acetonitrile, N-methyl-2-pyrrolidone (NMP), tetrahydrofuran (THF), and benzene solvents. The salts used were lithium/sodium hexafluorophospate (PF 6 ) and trifluoromethanesulfonate (Tf ), and lithium bis(trifluoromethanesulfonyl)imide (LiTFSI). Acetonitrile is a selective solvent for PS-b-PEO, solvating only the PEO block. In contrast, NMP, THF and benzene are good solvents for both blocks. The ionic salts are solvated by the polar solvents acetonitrile, NMP and THF, although the relative solubilities vary. For the purposes of this study, thin films were cast from (1) acetonitrile, (2) NMP, and (3) benzene THF. The salts and PS-b-PEO were dried before use, and all samples were prepared in an argon glovebox. For the acetonitrile and NMP solutions, a given salt was first dissolved in the solvent and PS-b-PEO then added to achieve a polymer concentration in the solution of 1% by weight. Salt concentrations, r, of 1/60, 1/12, and 1/6 cations per ethylene oxide moiety were targeted. For the binary benzene THF combination, separate solutions consisting of 10% by weight salt in THF and 1% by weight PS-b-PEO in benzene were first prepared. Then a few drops of the salt THF solution were added to the PS-b-PEO in benzene to give a salt concentration r of 1/12. The resulting ratio of benzene to THF was 97 to 3% by weight. Solutions of PS-b-PEO in the absence of salt were also prepared using each of the three solvent systems discussed. In the case of the binary solvent combination, the same ratio of benzene to THF was used as before. Free-standing thin films were obtained by spin casting onto holey silicon nitride TEM substrates purchased from Ted Pella. Each substrate was mounted onto a metal disc, using small tabs of removable tape, in order to enable placement onto the vacuum chuck of the spin coater. For each sample, 5 L of polymer solution were pipetted onto the substrate, which was then spun at 2800 rpm for 60 s. All TEM samples were left to dry in the glovebox for at least 24 h and were then transported in a desiccator Thin-film characterization Characterization of the free-standing, unstained films by TEM was performed at the National Center for Electron Microscopy at Lawrence Berkeley National Laboratory (LBNL), using a Zeiss Libra 200MC operated at 200 kv and the TEAM 0.5 microscope operated at 80 kv, and at the Molecular Foundry, LBNL, using a Zeiss Libra 120 PLUS operated at 120 kv. These electron microscopes are equipped with energy filters enabling various EFTEM-based mapping techniques and the implementation of zero-loss filtering to improve contrast in bright-field images by reducing the image blurring caused by chromatic aberrations and inelastically scattered electrons. Local thickness variations in the thin films were investigated by recording zero-loss filtered and unfiltered images of the same region and applying the standard log-ratio method to generate a relative thickness map (Egerton, 2011). Chemical mapping was performed by implementing EFTEM spectrum-imaging of the plasmon resonance in the low electron energy-loss range, which serves as a chemical fingerprint for different polymer types. The EFTEM spectrum-image datasets were typically implemented using a slit width of 5 ev to record a series of filtered images across the energy-loss range of 5 60 ev in 1 ev steps. The acquisition time for each image was 0.5 s and the dose rate was of the order of 10 5 e /nm 2 /s. By applying principal component analysis to the low-loss spectrum-image datasets, as described in a previous work (Allen et al., 2011), even subtle differences in the plasmon peak shapes can be detected and hence used to distinguish chemical phases. Tilt series of zero-loss filtered images for tomographic reconstructions were recorded manually at the 200MC Zeiss Libra. In a typical series, images were recorded with an acquisition time of 0.5 s from -60 to +60 in 2 intervals. The dose rate was 10 5 e /nm 2 /s. Compensations for sample drift and image focus were implemented between successive image acquisitions as required. Three-dimensional reconstructions were then obtained from the tilt series by applying an algorithm for weighted backprojection (Midgley and Weyland, 2003). Given the electron dose rates applied, it is expected that the total dose received by a polymer sample during the acquisition of a given EFTEM or tilt series

3 444 F.I. Allen et al. / Micron 44 (2013) Fig. 1. (a) Bright-field zero-loss filtered TEM image of a free-standing PS-b-PEO:LiTFSI film (r = 1/12) cast from acetonitrile. Chemical map in inset obtained by low-loss EFTEM spectrum-imaging showing PS in red and PEO in green. (b) Schematic diagram of a spherical micelle. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.) will have resulted in radiation damage to the material. However, repeated measurements of the same sample region gave the same chemical distributions and significant mass-loss was not observed. Hence at the nanometer length scales investigated here, we do not expect radiation damage to have affected the chemical distributions or 3-D morphologies observed. Surface characterization of the thin films was achieved by AFM using a Digital Instruments Multimode SPM in tapping mode, in the repulsive regime. Both height and phase maps were recorded. certain degree of preferential indentation of the specimen in the softer PEO regions may contribute to the local height differences observed. In the phase map of Fig. 2(b), the brighter contrast corresponds to regions with a higher elastic modulus. Hence the 3. Results and discussion 3.1. Films cast from acetonitrile In selective solvents, block copolymers form micelles which become kinetically trapped when the solvent is removed. Hence when casting PS-b-PEO from acetonitrile, a thin film composed of micelles in a random arrangement results. The micelle cores comprise the non-solubilized polymer block (PS), each surrounded by a corona of the solubilized block (PEO). Fig. 1(a) shows a bright-field TEM image of a free-standing portion of the micellar PS-b-PEO film obtained when casting from acetonitrile and doping with LiTFSI (r = 1/12). It is representative of the results obtained when doping with other salts, as well as of those films cast from acetonitrile without salt doping. The film is not stained. The spherical, darker structures in the image correspond to the aggregated PS chains forming the micelle cores, and the brighter phase corresponds to the less dense coronae of PEO. In order to confirm the chemical identity of the two polymer phases, low-loss EFTEM spectrum-imaging (Allen et al., 2011) was performed. The inset in Fig. 1(a) shows the chemical map obtained, with red and green corresponding to the PS and PEO phases, respectively. A schematic diagram illustrating the basic arrangement of PS-b-PEO chains forming a single spherical micelle is shown in Fig. 1(b). Surface characterization of the micellar films by AFM reveals good wetting of the holey silicon nitride substrate and no obvious change in morphology between substrate-supported and freestanding regions. Examples of the AFM height and phase maps for the PS-b-PEO:LiTFSI (r = 1/12) sample are given in Fig. 2(a) and (b), respectively. In the height map, the circular depression in the film surface corresponds to the free-standing region spanning one of the holes in the substrate. The height map also reveals local height differences which could correspond to partially embedded micelle cores at the thin-film surface. However, it should be noted that a Fig. 2. (a) AFM height map and (b) phase map for a PS-b-PEO:LiTFSI film (r = 1/12) cast from acetonitrile. The circular depression in (a) corresponds to the free-standing region of the film across the hole in the substrate.

4 F.I. Allen et al. / Micron 44 (2013) bright spheres in the image correspond to the micelle cores made of the hard PS block of the copolymer. The micellar morphology observed for PS-b-PEO films cast from acetonitrile is no surprise considering the selective nature of this solvent towards the respective polymer blocks. However, in the case of PS-b-PEO films cast from NMP and benzene THF, which are good solvents for both blocks, the particular morphologies observed were not always as predicted. In order to interpret the 445 microstructures observed in those films, more rigorous 3-D analysis was required, as detailed in the following Films cast from NMP A selection of bright-field TEM images of free-standing regions of the PS-b-PEO films cast from NMP without salt-doping, and doped with LiPF6, NaPF6, LiTf, NaTf, and LiTFSI salts (r = 1/6) are shown in Fig. 3(a) (f), respectively. Similar morphologies are observed for all of the salt-doped films. Upon first inspection, these Fig. 3. Bright-field zero-loss filtered TEM images of free-standing PS-b-PEO films cast from NMP (a) undoped and (b) doped with LiPF6, (c) NaPF6, (d) LiTf, (e) NaTf, and (f) LiTFSI (r = 1/6). The dark phase corresponds to PS and the bright phase corresponds to PEO.

5 446 F.I. Allen et al. / Micron 44 (2013) Fig. 4. Relative thickness map and extracted relative thickness profile obtained by EFTEM of a free-standing PS-b-PEO:NaPF 6 film (r = 1/6) cast from NMP. Thicker regions (PS) give rise to brighter contrast in the thickness map and thinner regions (PEO) give darker contrast. morphologies could be interpreted to be lamellar or cylindrical in nature. The actual morphology is elucidated below. We note that the spacings between the microstructures observed in the various films vary strongly, indicating that thermodynamic equilibrium has not been reached in these as-cast films. In the case of the undoped PS-b-PEO film shown in Fig. 3(a), a different spherical morphology in a hexagonal arrangement is observed, demonstrating the dramatic effect of selective salt solvation on the morphology of the block copolymer film. In addition, needle-like structures are clearly visible in the undoped PEO phase. These structures could correspond to discrete crystallites of PEO. When salt is introduced, expanded chain conformations are adopted to enable coordination of the salt cations (Hamley, 1999), explaining the absence of crystallites in the salt-doped films. A more detailed comparison of the phase behavior in the salt-doped versus undoped films is complicated by the fact that one is dealing with non-equilibrated systems. Instead, we focus on deciphering the precise 3-D morphology of the salt-doped films shown in Fig. 3(b) (f). Upon closer consideration of the TEM images in Fig. 3, it is significant to observe such strong contrast, since the films are not stained, and the chemical compositions and nominal densities of PS and PEO are so similar. The origin of this contrast becomes clear from EFTEM measurements of relative sample thickness. As an example, Fig. 4 shows the relative thickness map obtained for a region of free-standing PS-b-PEO:NaPF 6 (r = 1/6). Beneath the map, the extracted relative thickness profile is plotted in standard units of absolute thickness, t, over the inelastic mean free path of electrons in the specimen,. Clearly there is a marked difference in relative thickness between the two polymer phases, which cannot be entirely due to subtle differences between their corresponding inelastic mean free path lengths. Hence the dominating contrast mechanism in the bright-field TEM images of Fig. 3 appears to be thickness contrast, with the darker phase corresponding to thicker regions of the polymer film. Using previous chemical mapping results (Allen et al., 2011), as well as those presented later on in this section, we assign the thicker phase to be PS and the thinner phase to be PEO. In terms of the cross-sectional profile of the film, the smoothly modulated thickness profile shown in Fig. 4 is suggestive of cylindrical PS structures surrounded by a wetting layer of PEO. Measurements of the surface topography of the thin film by AFM support this hypothesis, as shown in the height map presented in Fig. 5(a). The corresponding phase map is given in Fig. 5(b). The AFM Fig. 5. (a) AFM height map and (b) phase map for a PS-b-PEO:NaPF 6 film (r = 1/6) cast from NMP. Significant dewetting from the substrate has occurred, leaving a circular free-standing film across the hole. results also reveal that when casting from NMP, extreme dewetting of the PS-b-PEO electrolyte from the silicon nitride surface occurs. NMP has a low vapor pressure and thus dries slowly. During this process, the polymer solution apparently drains to stable areas leaving a circular free-standing film across a given hole in the substrate surrounded by a featureless substrate surface. Unequivocal determination of the 3-D shape of the PS phase in the films cast from NMP has been achieved by TEM tomography. A visualization of the results obtained for the PS-b-PEO:NaPF 6 film (r = 1/6) is presented in Fig. 6. Two orthogonal slices through the reconstruction are shown, one in the plane of the thin film near the Fig. 6. Orthogonal slices through the tomographic reconstruction of a free-standing PS-b-PEO:NaPF 6 film (r = 1/6) cast from NMP. The dark phase corresponds to PS and the bright phase corresponds to PEO.

6 F.I. Allen et al. / Micron 44 (2013) Fig. 8. Schematic cross-section showing cylindrical morphology of the PS-b-PEO electrolyte films cast from NMP. Fig. 7. Chemical mapping of a free-standing PS-b-PEO:NaPF 6 film (r = 1/6) cast from NMP, achieved by low-loss EFTEM spectrum-imaging. PS, PEO, and salt phases in red, green, and blue, respectively Composite RGB image shown on the right. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.) lower surface, and the other showing the circular cross sections of the PS cylinders (dark contrast) surrounded by the bright PEO phase. The diameter of the PS cylinders is 40 nm. Artifacts from the tomographic reconstruction method and low contrast from the thin PEO phase make it difficult to determine the location of the PEO-vacuum interface. As a result, detection of the PEO wetting layer above and below the PS cylinders is a challenge when using the tomography result alone. Evidence for the wetting layer has instead been collected from chemical and elemental maps. Fig. 7 shows an example of the chemical maps obtained for the PS-b-PEO:NaPF 6 film (r = 1/6) by applying lowloss EFTEM spectrum-imaging (Allen et al., 2011). The images in red, green, and blue show the individual maps for the PS, PEO, and salt phases, respectively, with the composite RGB image shown on the right. Additive red and green color mixing in the RGB image has resulted in yellow patches in the PS-rich regions, indicating a layering of PS and PEO. This is particularly evident for slightly thicker films, which presumably also have a thicker PEO wetting layer around the PS cylinders. As observed in chemical mapping performed previously, the salt segregates to the PS-PEO interface (Allen et al., 2011). Oxygen mapping of the same specimen, achie- ved using a recently developed advanced acquisition scheme for EFTEM spectrum-imaging, provides further evidence for the existence of PEO around the PS phase (Watanabe and Allen, 2012). Summarizing the results detailed above, Fig. 8 shows a schematic cross section of the cylindrical thin-film morphology obtained when casting the PS-b-PEO electrolytes from NMP. Cylindrical morphologies were also observed for PS-b-PEO electrolyte films prepared using salt concentrations r of 1/12 and 1/60. The deviation from the lamellar morphology observed in the bulk could be due to the confinement of the system into a thin-film geometry of thickness equaling the natural width of the bulk PS lamellae ( 40 nm), with cylinders of PS enabling a minimization of the interfacial energies. However, it is also possible that by selectively doping the PS-b-PEO with salt, the neutral behavior of NMP towards PS and PEO shifts towards preferential solvation of one of the copolymer blocks. As a polar aprotic solvent, NMP might preferentially solvate the PEO:salt phase, hence one could be observing cylindrical micelles with cores made of PS, which have become kinetically trapped in the dried film. Given that a clear indication of the formation of micelles would have been a clouding of the casting solution, which in this case was not observed, we conclude that if micellization occurred, this would have been during the film casting process itself Films cast from benzene THF Bright-field TEM images of free-standing PS-b-PEO films cast from benzene(97%) THF(3%) without salt doping and then doped using LiTFSI (r = 1/12) are shown in Fig. 9(a) and (b), respectively. The undoped film exhibits strong microphase separation, yet the morphology is not well-defined. In contrast, the salt-doped film presents a well-defined morphology suggestive of a layered honeycomb-type structure. Similar results are obtained for PS-b- PEO films cast using the same solvent ratio but doped using various other salts. The dark specks in the bright-field image in Fig. 9(b) are 10 nm gold nanoparticles deposited onto the surface of the film to Fig. 9. Bright-field zero-loss filtered TEM images of free-standing PS-b-PEO films cast from benzene THF (a) undoped and (b) doped with LiTFSI (r = 1/12). Chemical map in inset of (b) obtained by low-loss EFTEM spectrum-imaging showing PS in red and PEO in green. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

7 448 F.I. Allen et al. / Micron 44 (2013) Fig. 10. Relative thickness map and extracted relative thickness profile obtained by EFTEM of a free-standing PS-b-PEO:LiTFSI film (r = 1/12) cast from benzene THF. The brighter contrast corresponds to the PS honeycomb and the darker contrast corresponds to PEO. be used as alignment markers for TEM tomography. The chemical map in the inset has been obtained by low-loss EFTEM spectrumimaging (Allen et al., 2011) and shows PS in red and PEO in green. The reason for the dramatic effect of salt-doping on the morphology of the PS-b-PEO film is not clear, although it is possible that the degree of segregation, which is enhanced by selective salt solvation, plays a role. An EFTEM relative thickness map of a portion of the freestanding PS-b-PEO:LiTFSI film from Fig. 9(b) is presented in Fig. 10, with the extracted thickness profile plotted directly beneath. While the mean relative thickness of the cylindrical PS-b-PEO electrolyte film cast from NMP was 0.18 (see Fig. 4), the relative thickness of this film averages at The transformation from a cylindrical to a layered honeycomb-type morphology in the PS-b-PEO electrolyte films could thus be driven by a thickness effect. Measurements of the surface topography of the PS-b-PEO:LiTFSI film by AFM show undulations corresponding to an extended honeycomb network on the film top surface. An example of these results is shown in the height map of Fig. 11(a). The AFM data also show that the film evenly wets the holey silicon nitride substrate. The circular depression in the film surface corresponds to the freestanding region of the film spanning one of the holes in the silicon nitride. The corresponding AFM phase map is given in Fig. 11(b). Since the AFM data were acquired in the repulsive regime, the brighter regions in the phase map must represent the glassy PS domains. The results from EFTEM chemical mapping, shown in the inset of Fig. 9(b), support this assignment. Hence we can conclude that the honeycomb network is made of PS. Recently, the morphologies of non-equilibrated PS-b-PEO:LiTFSI thin films spin cast from benzene THF solutions have also been reported elsewhere (Metwalli et al., 2011). AFM characterization of those films revealed surface structures resembling the honeycomb structures observed in our study. The layered nature of the PS-b-PEO:LiTFSI films, which is already suggested in the 2-D TEM images discussed above, is directly confirmed by TEM tomography. Fig. 12 presents a 3-D reconstruction of the free-standing film revealing its trilayer structure. A slice through the plane of the film shows the extended PS honeycomb network on the lower surface, and the orthogonal slice shows a cross-section through the PS honeycombs on top and bottom with the PEO layer in between. The thickness of each PS honeycomb layer is 40 nm, while the thickness of the central PEO layer is 80 nm. The maximum film thickness is thus estimated at 160 nm. Combining the results presented above, a schematic diagram showing Fig. 11. (a) AFM height map and (b) phase map for a PS-b-PEO:LiTFSI film (r = 1/12) cast from benzene THF. The circular depression in (a) corresponds to the freestanding region of the film. a cross-section through the PS-b-PEO:LiTFSI thin film is given in Fig. 13. The volume fraction occupied by each polymer block in the thin film can be estimated from the 3-D reconstruction. Using thresholding for the PS honeycombs and the full thickness of the central PEO layer, we obtain volume fractions for the PS and PEO blocks of 0.3 and 0.7, respectively. Although the missing wedge of information is obviously rather large and will affect the accuracy of this result, even a qualitative inspection of the 3-D reconstruction already reveals that the PEO volume fraction is considerably greater in the thin film compared to the bulk material, where it has a Fig. 12. Orthogonal slices through the tomographic reconstruction of a freestanding PS-b-PEO:LiTFSI film (r = 1/12) cast from benzene THF. In this visualization the bright-field contrast has been inversed, hence the brighter phase corresponds to PS and the darker phase corresponds to PEO.

8 F.I. Allen et al. / Micron 44 (2013) Table 1 Thin-film morphologies determined for free-standing PS-b-PEO electrolyte films cast from various solvents. Solvent Acetonitrile NMP Benzene THF Thin-film morphology PS spheres; PEO/salt matrix PS cylinders; PEO/salt wetting layer PS honeycombs; PEO/salt central layer Fig. 13. Schematic cross-section showing hypothesized morphology of PS-b- PEO:LiTFSI films cast from benzene THF. PS honeycombs on top and bottom with PEO layer in between. nominal value of 0.55 (Singh et al., 2007). Hence we conclude that in the thin film significant swelling of the PEO phase has occurred. Since thin-film morphologies are very sensitive to film thickness (Fasolka and Mayes, 2001), an expansion of the PEO phase could explain the transformation away from the single-layer cylindrical morphology observed when casting from NMP, into the trilayer structure obtained when casting from benzene THF. It is also conceivable that the PS honeycomb networks on the upper and lower film surfaces resulted from the aggregation of PS cylinders during the drying process. Furthermore, following the previous discussion of the potential formation of micelles when casting salt-doped PS-b-PEO from NMP, one should consider the possibility that micelles are formed when casting from the binary solvent combination benzene THF. Since the polymer solutions again showed no evidence of cloudiness before casting, any micellization will have likely only occurred during the film casting process. A potential cause for the selective expansion of the PEO phase driving the formation of the trilayer films could be the watersoluble nature of PEO. For example, in work by Bang et al. (2007), exposure of PEO-containing diblock and triblock copolymers to humid air was found to dramatically affect the thin-film morphologies that developed. In our work we endeavored to minimize exposure to water, with glovebox integrity maintained throughout the film casting and drying process. However, upon transferring samples to the microscope brief exposure to humid air will have occurred, hence water-induced swelling of the PEO phase might have happened during this stage. In any event, our results indicate that the expansion observed is certainly linked to the presence of LiTFSI salt in the PEO phase, since salt-doped PS-b-PEO films cast from benzene THF are typically about 50 % thicker than their undoped counterparts, with the latter not developing a layered honeycomb structure (see Fig. 9(a)). Closer inspection of the layered honeycomb films by TEM reveals a multitude of pores in the PEO phase, which could have also contributed to the expansion observed. The pores are a few nanometers in diameter and are particularly evident in the EFTEM thickness map shown in Fig. 10. Such pores could have formed as a result of trapped pockets of solvent localized around the polar LiTFSI in the PEO phase. Since the lithium salt is highly hygroscopic, the solvent responsible could have been any water absorbed during transfer to the microscope and which was then rapidly removed under vacuum. Alternatively, the solvent pockets could have contained the THF used to solvate the salt, and which has a higher volatility than its co-solvent, benzene. It should be noted, however, that pores could also have developed in the microscope during imaging due to localized mass loss resulting from radiolysis reactions induced by exposure to the electron beam (Grubb, 1974). A summary of the various thin-film morphologies determined for the PS-b-PEO electrolyte films investigated in this work is presented in Table Conclusions Definitive characterization of various block copolymer thin-film morphologies in 3-D has been demonstrated using a combination of EFTEM techniques and electron tomography. By casting directly onto holey substrates, free-standing thin films ideal for extensive investigation by TEM were able to be prepared in a convenient and reproducible manner. This sample preparation technique combined with the multi-method TEM approach offers the opportunity to expand upon the hitherto rare insights into the 3-D morphologies of block copolymer thin films prepared under the symmetric boundary condition of two free interfaces. In the case of unexpected morphologies resulting from uncertainties in the casting process, detailed 3-D characterization is essential, since the results from basic TEM imaging or AFM surface characterization alone may be misleading. For example, when casting the nominally lamellar PS-b-PEO electrolyte from benzene THF, the layered honeycomb-type morphology adopted in the as-cast film was only entirely revealed once the electron tomographic reconstruction was obtained. Further analysis of the reconstruction enables one to deduce a swelling of the central PEO layer sandwiched between the two PS honeycombs, which likely drove the formation of the structure observed. We also find that the structure of the layered honeycomb-type films is relatively stable under the electron beam as compared to the films with the cylindrical morphology. This could be attributed to the stabilizing nature of honeycomb networks (Gibson and Ashby, 1999). Such structural stability would be a favorable attribute in applications requiring robust thin films. Acknowledgements Financial support for this work was provided through the Electron Microscopy of Soft Matter Program supported by the Director, Office of Science, Office of Basic Energy Sciences, Materials Science and Engineering Division, of the U.S. Department of Energy under Contract No. DE-AC02-05CH The TEM experiments were performed as user projects at the National Center for Electron Microscopy and the Molecular Foundry, Lawrence Berkeley National Laboratory, which are supported by the Office of Science, Office of Basic Energy Sciences, Scientific User Facilities Division, of the U.S. Department of Energy under Contract No. DE-AC02-05CH We thank Dr. Mohit Singh for the synthesis of the block copolymer. F.I.A would also like to thank Professor Enrique Gomez, Dr. Sergey Yakovlev and and Dr. Daniel Hallinan for helpful discussions. References Albert, J.N.L., Epps I., T.H., Self-assembly of block copolymer thin films. Materials Today 13, 24. Allen, F.I., Watanabe, M., Lee, Z., Balsara, N.P., Minor, A.M., Chemical mapping of a block copolymer electrolyte by low-loss EFTEM spectrum-imaging and principal component analysis. Ultramicroscopy 111, 239. Bang, J., Kim, B.J., Stein, G.E., Russell, T.P., Li, X., Wang, J., Kramer, E.J., Hawker, C.J., Effect of humidity on the ordering of PEO-based copolymer thin films. Macromolecules 40, Bates, F.S., Polymer-polymer phase behavior. Science 251, 898.

9 450 F.I. Allen et al. / Micron 44 (2013) Bates, F.S., Fredrickson, G.H., Block copolymer thermodynamics: theory and experiment. Annual Review of Physical Chemistry 41, 525. Bates, F.S., Fredrickson, G.H., Block copolymers designer soft materials. Physics Today 52, 32. Druger, S.D., Nitzan, A., Ratner, M.A., Polymeric solid electrolytes: dynamic bond percolation and free volume models for diffusion. Solid State Ionics 9 10, Du Chesne, A., Energy filtering transmission electron microscopy of polymers benefit and limitations of the method. Macromolecular Chemistry and Physics 200, Egerton, R.F., Electron Energy-Loss Spectroscopy in the Electron Microscope, Third edition. Springer, New York. Epps, T.H., Bailey, T.S., Pham, H.D., Bates, F.S., Phase behavior of lithium perchlorate-doped poly(styrene-b-isoprene-b-ethylene oxide) triblock copolymers. Chemistry of Materials 14, Epps, T.H., Bailey, T.S., Waletzko, R., Bates, F.S., Phase behavior and block sequence effects in lithium perchlorate-doped poly(isoprene-b-styreneb-ethylene oxide) and poly(styrene-b-isoprene-b-ethylene oxide) triblock copolymers. Macromolecules 36, Fasolka, M.J., Mayes, A.M., Block copolymer thin films: physics and applications. Annual Review of Materials Research 31, 323. Gibson, L.J., Ashby, M.F., Cellular Solids: Structure and Properties, Second edition. Cambridge University Press, Cambridge. Grubb, D., Radiation damage and electron microscopy of organic polymers. Journal of Materials Science 9, Hamley, I.W., Crystallization in block copolymers. Advances in Polymer Science 148, 113. Jinnae, H., Spontak, R.J., Transmission electron microtomography in polymer research. Polymer 50, Libera, M.R., Egerton, R.G., Advances in the transmission electron microscopy of polymers. Polymer Reviews 50, 321. Marencic, A.P., Register, R.A., Controlling order in block copolymer thin films for nanopatterning applications. Annual Review of Chemical and Biomolecular Engineering 1, 277. Metwalli, E., Nie, M., Körstgens, V., Perlich, J., Roth, S.V., Müller-Buschbaum, P., Morphology of lithium-containing diblock copolymer thin films. Macromolecular Chemistry and Physics 212, Midgley, P., Weyland, M., D electron microscopy in the physical sciences: the development of Z-contrast and EFTEM tomography. Ultramicroscopy 96, 413. Mullin, S.A., Stone, G.M., Panday, A., Balsara, N.P., Salt diffusion coefficients in block copolymer electrolytes. Journal of the Electrochemical Society 158, A619. Orilall, M.C., Wiesner, U., Block copolymer based composition and morphology control in nanostructured hybrid materials for energy conversion and storage: solar cells, batteries, and fuel cells. Chemical Society Reviews 40, 520. Radzilowski, L.H., Carvahlo, B.L., Thomas, E.L., Structure of minimum thickness and terraced free-standing films of block copolymers. Journal of Polymer Science Part B 34, Segalman, R.A., McCulloch, B., Kirmayer, S., Urban, J.J., Block copolymers for organic optoelectronics. Macromolecules 42, Singh, M., Odusanya, O., Wilmes, G.M., Eitouni, H.B., Gomez, E.D., Patel, A.J., Chen, V.L., Park, M.J., Fragouli, P., Iatrou, H., Hadjichristidis, N., Cookson, D., Balsara, N.P., Effect of molecular weight on the mechanical and electrical properties of block copolymer electrolytes. Macromolecules 40, Watanabe, M., Allen, F.I., The smarteftem-si method: development of a new spectrum-imaging acquisition scheme for quantitative mapping by energyfiltering transmission electron microscopy. Ultramicroscopy 113, 106. Yang, S.Y., Yang, J.A., Kim, E.S., Jeon, G., Oh, E.J., Choi, K.Y., Hahn, S.K., Kim, J.K., Single-file diffusion of protein drugs through cylindrical nanochannels. ACS Nano 4, Young, W.S., Epps, T.H., Salt doping in PEO-containing block copolymers: counterion and concentration effects. Macromolecules 42, 2672.