Introduction

The polymer industry has become one of the most developing industries in the world in the last decades, producing continuous changes in everyday life. Special attention is paid to synthetic polymers, where the structure of the materials can be modified in multiple ways to make them suitable for specific applications. One of the strategies used to obtain materials with the desired combination of properties is polymer blending. Since the first patent in 18461, polymer blending has become widespread, as the resulting materials find interesting applications in numerous fields due to their advanced properties. Polymer blending is becoming popular in advanced and conventional manufacturing techniques, such as Additive Manufacturing (AM) and Injection Moulding (IM). For example, the use of polymer blending in Fused Filament Fabrication (FFF), one of the most widely used AM extrusion methods, holds the promise to significantly improve the mechanical properties and the printability of high performance polymers (such as poly-ether-ether-ketone (PEEK) or polyetherimide (PEI)), which usually present processing challenges related to their low viscosity (see a Review in2) and printability. In IM, however, virtually all identified polymer blend types can be processed. In general, while IM offers advantages in terms of precision, high-volume production rates, and part mechanical properties, FFF excels in flexibility, low initial cost, and the ability to produce complex and customized geometries.

In order to understand and optimize the mechanical and functional properties of polymer blends as well as the effect of their synthesis conditions, information on the structural characteristics of the materials is essential. There is a variety of characterization techniques that can be used to determine the nature of the polymer blends3,4. For example, Fourier-transform infrared spectroscopy offers information on chemical interactions in the material, including hydrogen bonding and information regarding possible phase separation5. X-ray diffraction has been used to study the crystalline properties of PCL/PLA polymer blends, reporting evidences of the immiscibility of the polymers6. The melting point depression obtained by differential scanning calorimetry (DSC) has been used to establish miscibility and the polymer–polymer interaction parameter in biobased polyamide blends7. Other techniques include dynamic mechanical analysis, useful for the determination of the crosslinking density between the two components of a blend8 or mass spectroscopy, providing information on the surface composition distribution, for example in blends of cyclic and linear polystyrene9. However, most of these techniques offer indirect information of the structure of the blends, and often it is useful to obtain direct information on the components distribution at micro/nano-scale.

In recent years, the FIB instrument has become a powerful analysis tool. It has the unrivalled ability to micro/nanomanipulate a wide variety of materials site-specifically with nanometer precision, while allowing the analysis of the machined material at different stages with a variety of detectors, including secondary electrons (SE) detectors, electron backscattered diffraction (EBSD) detectors or energy dispersive x-ray spectroscopy (EDX) detectors. Because of this, it has found multiple applications in nanotechnology (see a review in10). Within the versatility of FIB applications, methods for imaging materials in three dimensions (3D) are attracting increasing interest. Here, the ion beam of the FIB is used to mill slices of material sequentially, where each of the milled faces is imaged with the chosen detector (this technique is known as slice and view, FIB tomography or serial sectioning, more details can be found in11). The reconstruction of the obtained data with specialized algorithms allows obtaining 3D information of the studied material. Care should be taken in the selection of the appropriate detector in these 3D analyses. Although electron beam induced SE (known as SE microscopy, -M) are often used in FIB tomography12,13, in general this SEM signal is best suited for imaging topography and, since a FIB polished face does not contain topography per se, sometimes contrast is insufficient for the analyses. This is often the case of polymer materials, where FIB-based characterization reports are not numerous. FIB-SEM tomography has been used to create high fidelity 3D reconstructions of diblock copolymers14, as in this case the contrast between the components resulted very large, and also for the study of porous polymers15. In other cases, topography contrast resulting from a differential sputtering rate of the two polymers under the ion beam has been found useful for the analysis of polymer multicomponent systems, as in bicomponent combinations of low density polyethylene (PE) and nylon 616. Additional strategies have also been reported to optimize the contrast between the components, for example, heavy metal staining has been used to reveal the 3D structure of thermoplastic polyolefin blends17,18. On the other hand, the use of ion beam induced SE (iSE) images have demonstrated useful for the analysis of some materials, for example for the study of the crack morphology in silicon nitride19 and for polymeric scaffolds for the generation of 3D tissues20. iSE imaging advantages include high contrast sensitivity to surface topography and crystalline grains. Ions, however, are heavier than electrons and, because of this, they always produce a certain degree of sample sputtering, limiting the probing dwell time21.

X-ray scattering techniques, such as small and wide-angle X-ray scattering (SAXS), have been previously used to investigate the orientation of chains in thermoplastic polymers during operando FFF22. Although these techniques provide information about the average structure and orientation of the polymer chains, they lack the spatial resolution to directly visualize individual features. On the contrary, by providing real-space information, iSE tomography allows for the direct observation of morphological features, which can significantly impact the materials properties. By combining these two techniques, a more comprehensive understanding of the structure-property relationships in 3D printed polymers can be achieved. However, for this iSE tomography of polymer blends need to be optimized.

In this work, we evaluate the applicability of iSE tomography for the analysis in 3D of the components distribution in polymer blends of ASA-PC obtained by FFF and IM, with applications in the automotive industry23. This is the first time than iSE tomography is demonstrated for the analysis in 3D of the phase distribution in polymer blends. Our results show that the use of iSE images is a good choice for these analyses, and that the results obtained with this technique can be of great use to shed light at the effect of the fabrication parameters on the blends microstructure, which is determinant to understand their technological properties.

Experimental methods

Pellets of ASA and of PC were purchased at Biesterfeld Iberica (Spain) (in particular, materials with ref. ASA: LI-912 and PC: TRIREX 3025N2 are considered). All the materials were dried before processing to remove any residual moisture in a DPA30 compressed air dryer (Piovan Group, Italy) for at least 24 h at 90 °C. Then, 250 g of ASA and 250 g of PC were processed in a twin-screw SCAMEX extruder (SCAMEX, France, L/D 508/10,8 mm, 90 rpm) to obtain a ASA 50 wt% - PC 50 wt% blend (ASA50 / PC50), using a temperature profile of five heaters of 245-250-250-250-245 °C (from hopper to nozzle). A continuous filament with a diameter of 1.75 mm was obtained. With this filament, a solid piece (4 × 10 × 70 mm) was manufactured using a FFF 3D-printer Raise 3D Pro 2 (Impresoras 3D, Spain). The slicer software used was ideaMaker, using a linear infill of 100% with an overlap of 10% between beads to generate all the g-codes and a nozzle of Ø 0.4 mm, printing at 250 °C and 50 mm/s. Neat ASA and neat PC were also extruded and printed, to be used as reference materials in the compositional analysis. On the other hand, the ASA/PC filament was cut into pellets (2–3 mm length) in a pelletizer (SCAMEX, France) at 71 rpm, and these were used in IM to produce injected parts with same size and geometry as the printed objects in a BabyPlast 10/12 (Cronoplast, Spain).

To prepare samples for FIB analyses, solid specimens were cryogenically sectioned using a scalpel at liquid nitrogen temperature. Prior to FIB analysis, samples were sputter-coated with a conductive Au/Pt layer using a SCD 004 Sputter Coater (BAL-TEC, Balzers, Liechtenstein) to mitigate charging effects. Structural analyses were conducted using a Scios 2 DualBeam FIB-SEM system (Thermo Fisher Scientific, Waltham, MA, USA) operating at 30 kV ion acceleration voltage. Tomographic datasets were acquired using iSE images collected with an in-lens SE detector. The final pixel dimensions were 2010 × 220 and 979 × 1482 for FFF and IM samples, respectively. The imaging pixel sizes were approximately 3.5 × 5.7 nm² for the FFF sample and 7.5 × 9.5 nm² (x/y) for the IM sample. FIB milling was performed at 100 nm intervals per slice using a beam current of 50 pA, with 100 slices for the FFF sample and 70 slices for the IM sample. The iSE images were acquired using an accelerating voltage of 30 kV and a Ga+ ion current of 50 pA. The total sampled volumes were 7 × 12.5 × 9.2 μm³ and 7.3 × 14.1 × 7 μm³ for FFF and IM samples, respectively. Image processing included application of a Non-local Means filter using a custom MATLAB script. Image stack alignment was performed using the sum of squared differences algorithm without rotation in Dragonfly software (Object Research Systems, Canada). Edge shadowing artefacts were corrected using a radial basis function. Segmentation was achieved using a Deep Learning Tool. A U-Net binary model was trained using an input dataset of eight manually segmented images, with the Adadelta algorithm for optimization. Simulations of electron and Ga+ ion interactions with matter were conducted using the Casino software (monte CArlo Simulation of electroN trajectory in sOlids)24,25 and the SRIM (Stopping and Range of Ions in Matter) software26, where parameters such as initial beam energy, target material type, target dimensions, and incidence angle are maintained as fixed boundary conditions. 10,000 trajectories for each material were simulated for ions and electrons. The sputtering rate under the ion beam has been quantified using a corrected ion current, although nominal currents are reported.

Results and discussion

Specimen preparation for slice and view experiments

FFF printed objects are formed by the deposition of melted filaments layer-by layer following a predefined design. For the analysis of the ASA/PC distribution in the polymer blend of the FFF printed object, it is of interest to select the core of one of the inner beads as the region to be studied, and to carry out the analysis in the direction parallel to this bead. Although differences in the structural characteristics maybe expected from the core to the edge of the beads, the core is a representative region that can be used to compare the microstructure for different fabrication parameters and methods. In order to gain access to one of the inner beads of the object without modifying the microstructure, the printed object needs to be fractured in a fragile manner. As the polymers that compose the blend are ductile to some extent, the material has been cooled in liquid nitrogen to obtain a fragile fracture when cut with a scalpel. Figure 1a shows the fracture surface of one of the inner beads of the printed object, which presents a nicely flat surface resulting from the fragile fracture. It should be mentioned that the sample has been placed in the FIB holder with the longitudinal axis of the deposited beads parallel to the holder surface. Because of this, for SEM imaging of the lengthwise section of the bead, the holder is tilted to 52°, resulting 38° between the vector normal to the surface of interest and the electron beam; when imaging with the ion beam, the holder is at 0°, resulting 38° between the ion beam and the vector normal to the surface of interest (see an schematic in Supporting Information (SI), Fig. S1). This clarification is of interest in relation to the simulations shown in Fig. 3. In order to be able to perform slice and view experiments in a bulk material, the volume of interest needs to be prepared for the process. For this, initially the volume of interest needs to be protected of unintentional surface milling, in our case by a layer of Pt with a thickness of 1.5 μm, obtained by ion beam deposition at a current of 0.3 nA using a Pt-precursor gas of a gas injection system (GIS). In Fig. 1a, this square Pt deposition can be observed in the centre of the bead. After this, the volume of interest needs to be partially separated from the bulk material to avoid both redeposition of the material milled during the serial sectioning, and possible shading of the signal used for imaging. Although a complex block lift-out method has been reported as a substantial improvement for automated 3D serial sectioning experiments in a FIB27, that method is time consuming and, in our study, it is not necessary when the SE detector is used for imaging. In our case, a U-pattern has been milled around the region of interest as it can be observed in Fig. 1b and c, using decreasing currents of 7 nA, 3 nA and 1 nA to optimize the milling time while obtaining a precise milling in the region closer to the Pt deposition. On the other hand, it is worth mentioning that the reconstruction of the structural features of the material using a 2D images stack requires the previous correct alignment of each of the 2D images and, for that, fiducial markers are of great help. Two fiducial markers have been introduced close to the region of analysis, one of them with an orientation optimized for imaging with the electron beam and the second one for imaging with the ion beam. For this, a layer of Pt has been deposited, and marks (a cross and a circle) have been milled on them with the ion beam, as shown in Fig. 1d. After this, the material is ready to start the slice and view experiment. The same process has been followed to analyse the IM object, selecting as the region of interest the core of the object.

Fig. 1
figure 1

(a) SEM image of a fractured bead of the printed object in cross-section, where the square Pt deposition can be observed; (b) and (c) SEM images showing different steps of the milling of the U-pattern around the region of interest; (d) SEM image showing the fiducial markers milled next to the region of interest. The surface of interest is delineated by a red dashed line.

Full size image
Fig. 2
figure 2

2 keV SEM images of the neat ASA (a) and the neat PC samples (b); 30 keV iSE images of the neat ASA (c) and neat PC (d) samples, obtained in the same regions as (a) and (b), respectively.

Full size image

Comparison between SEM and iSE images of ASA, PC and the ASA-PC blend

Before analysing the ASA-PC blend and in order to obtain information for the correct design of the tomography experiments, the neat polymer constituents ASA and PC have been studied in 2D in the FIB. Fig. 2a shows a 2 keV SEM image of the neat ASA sample. As it can be observed, round contrasts can be distinguished in the milled flat surface. ASA is a terpolymer that consists of three distinct monomers: acrylonitrile, styrene and acrylate. The slightly crosslinked acrylate rubber (functioning as an impact modifier) form spherical particles, which are chemically grafted with styrene-acrylonitrile copolymer chains, and embedded in a styrene-acrylonitrile matrix28. According to this structure, the round contrasts observed in the SEM image can be associated to the acrylate rubber particles. Fig. 2b shows a SEM image of the PC sample. As expected, the round features observed in ASA are not present. However, some pores are found in the polymer surface, likely resulting from the extrusion process. For comparison, Fig. 2c and d show 30 keV iSE images of the neat ASA and neat PC samples, obtained from the same regions as Fig. 2a and b, respectively. As it can be observed, the features found by SEM are more clearly distinguished in the iSE images than in the SEM ones. In particular, the contrasts associated to the acrylate particles are noticeably sharper. This is due to the different interaction of ions and electrons with matter, as it will be discussed next.

Fig. 3
figure 3

Monte Carlo simulated electron and Ga+ ion trajectories for incidence angle of 38° in Si and in PC. (a) Projected range of ions (in red) and of electrons (in blue) for Si (continuous line) and PC (dashed line). Ions (in red) and electrons (in blue) trajectories in (b) Si and (c) PC.

Full size image

In order to shed some light at the differences observed between SEM and iSE images, Monte Carlo simulations of the interaction of a 2 keV electron beam and a 30 keV ion beam with PC (the database of the simulation software used does not contain information regarding ASA) have been carried out, and results are shown in Fig. 3. For comparative purposes, results obtained for Si as a reference material are also included. As explained earlier, the incidence angle of both ions and electrons with the vector normal to the surface when acquiring the images included in Fig. 2 is of 38°, therefore this angle has been used for the simulations. Figure 3a shows the projected range of ions (in red) and of electrons (in blue) for both materials. As it can be observed, the projected range of Ga+ ions in Si is smaller than that of electrons, approx. of 21.9 nm vs. 40.1 nm, respectively; the lateral spread of the projectile ions trajectories (Fig. 3b) is also smaller than that for electrons (17.4 nm vs. 42.3 nm, respectively). Due to the high concentration of energy deposition of ions compared to electrons, the SE generated by ions are extracted from a significantly smaller volume than those generated by primary electron beams. This localized interaction results in higher spatial resolution in iSE imaging compared to conventional SEM. In this sense, analyses reported for the perpendicular incidence of 30 keV He ions on Si have shown projected ranges with values in between those corresponding to Ga+ ions and to electrons, due to their intermediate mass29. In PC, the same tendency as in Si is observed by comparing the projected range (Fig. 3a) and the lateral spread (Fig. 3c) between incident electrons and Ga+ ions. Here, it is worth mentioning that larger values of projected range (64.3 nm for incident electrons and of 33.4 nm for Ga+ ions) and of lateral spread (61.7 nm for incident electrons and of 26.2 nm for Ga+ ions) than those for Si are found. This dependence of the projectile electrons/ions trajectories with the atomic number have also been reported previously, in that case for metallic materials, in particular by comparison of Au and Al as target materials30, where the differences found were larger than in our work due to the larger difference between the Z number of the metals considered. The authors show that for low Z metals, the three components of the SE (those derived from collision of projectile ions, those caused by the recoiled target atoms and those related to cascade electrons) contribute equally to the SE yield, whereas the electron excitation by the projectile ions dominates in high Z number target materials21. Our results obtained from the simulation of the electrons/ions trajectories show that the spatial resolution in SEM/iSE images of polymer materials is expected to be somewhat worse than in Si due to the lower Z number, but it is also evident that iSE images would show better spatial resolution than SEM images, as corroborated experimentally in Fig. 2c and d.

Figure 4a and b show a SEM and an iSE micrograph of the ASA/PC printed object, prepared for the FIB analysis following the methodology explained above. As it can be observed, in both images the contrast related to the spherical acrylate particles are observed, appearing more clearly in the iSE image, analogously to Fig. 2. Although the regions with these round contrasts can be inferred to be ASA, at the sight of the SEM image it is not easy to distinguish the interface between the two polymer phases ASA and PC. On the contrary, in the iSE image, dark and bright regions with a sharp interface are clearly observed, which are not visible in the SEM image. A detailed observation of the image shows that the spherical acrylate particles are present in the dark phase of the blend, indicating that this dark phase corresponds to ASA, whereas the brighter one corresponds to PC. EDX analysis have been used to corroborate this identification. For this, initially, EDX analysis of the pure ASA and pure PC samples have been carried out, and spectra are included in Fig. 4c. As it can be observed, together with the C and O signals, the N signal that stems from the acrylonitrile monomers is clearly detectable in the EDX spectra of ASA, whereas N is absent in PC, as expected (see chemical formulas of these polymers in Fig. S2). This difference in the chemical composition is of great help to distinguish the two phases in the analysis of the polymer blend. Figure 4d shows an iSE image of the blend, together with the N, O and C EDX compositional maps. As it can be observed, the dark regions in the iSE image are N rich, whereas the brighter regions are O rich, which is in good agreement with the first one corresponding to ASA, and the second one to PC. Quantitative values of the elements composition in each material are included in SI, in Table S1.

Fig. 4
figure 4

(a) SEM and (b) iSE images of the ASA/PC blend obtained by FFF; (c) EDX spectra obtained from the pure ASA and pure PC samples; (d) iSE image of the ASA/ PC blend obtained by FFF, and N, O and C EDX compositional maps obtained in the same region.

Full size image

Differences in contrast in SEM and iSE images obtained from the same region of metallic specimens have been reported before29,31, and this has led to investigations of the image formation mechanisms in both imaging modes. The interaction of energetic ions or electrons with solids is complex as they could suffer a number of scattering events including elastic collisions with the nuclei of the specimen atoms, which modify their travel direction, and inelastic collisions, leading to generation of SE, Auger electrons, photons, electron hole pairs, and others. The interpretation of both SEM and iSE images requires knowledge of the SE yields for the material of interest, as a function of the incident beam energy. However, iSE yield data is only tabulated for a few elements and a very few compounds, what limits the interpretability of the experimental images. For fixed acquisition parameters, this SE yield (and, thus, the contrast) in iSE images depends, mainly, on: (a) the atomic number of the atoms in the sample, (b) the angle of incidence of the ion beam relative to the sample surface normal (i.e., topography contrast) and (c) the channelling of the incident ions between lattice planes of the material (useful for the identification of grains in polycrystals)32. In our case, it is the first factor, the difference in SE yield associated to the atomic number of the material, the most likely cause of the difference in contrast between the phases observed in the iSE images, as the imaged milled face of the specimen has a flat topography and the existence of crystalline grains is not expected in the polymeric blend. However, there is some controversy on the relation of the iSE yield and the atomic number of the studied material. Monte Carlo simulations of ion-induced kinetic electron emission have predicted that the iSE yield of different metals would decrease with an increase in the atomic number of the target material for 30 keV Ga+ ion bombardment33,34. On the contrary, experimental analyses have shown a non-monotonic behaviour of the iSE contrast with Z35. Additionally, in a different report also using Monte Carlo simulations of ion–solid interactions, it has been also shown that the 30 keV Ga+ iSE signal is a non-monotonic function with respect to the atomic number. This is due to the influence of both the stopping power (which is defined as the energy loss per unit distance of any charged particle within a target) and the sputter yield36. Those analyses are related to pure metals, with a specific atomic number associated to each element. In the case of polymer blends, although they are based on C, they also contain other elements, such as H, O and/or N. Additionally, the distribution of atoms in a volume necessary to calculate an average Z number depends of the arrangement of the polymer chains, including the crystallinity degree. Although polymers for FFF are mostly amorphous, in the case of polymer blends the density of the components seems to be a more reasonable parameter to be considered in order to understand their contrast. As it can be observed in the iSE image in Fig. 4b, there is a clear contrast between the PC phase and the styrene-acrylonitrile matrix of ASA, being the second one darker. The density of PC is 1.2 g/cm3, whereas the average density of the styrene (d = 0.909 g/cm3)-acrylonitrile (d = 0.806 g/cm3) matrix is smaller. Our results suggest that this relatively small difference in density could be the reason for the different iSE yield, demonstrating a large sensitivity of iSE imaging for the analysis of polymer blends. In the literature, low-voltage (5 kV) transmission electron microscopy with a novel construction combining visual-light and electron microscopical techniques has been reported useful for the analysis in 2D of polymer blends, including PC / poly(styrene-co-acrylonitrile) (SAN)37. In this case, the difference in density between the individual components of polymer blends is also suggested to be the reason for the obtained image contrast (indicating that differences less than 0.04 g/cm3 can be traced with this technique), although the authors claim that the mechanisms of the imaging contrast are not fully understood.

iSE tomography of ASA-PC blends

At the sight of the results above, it seems of interest to carry out the tomography experiments using the iSE images instead of the classical SEM images. In order to obtain further information on the behaviour of the two polymer components of the blend under the Ga+ ion beam to set adequate parameters for the tomography study, the sputtering rate under the ion beam has been quantified for both neat ASA and neat PC. For this, 5 μm x 5 μm craters have been sputtered in ASA and in PC by scanning the Ga+ beam at normal incidence using the apertures corresponding to currents of 0.05 nA, 0.1nA and 1 nA, and times of 100 s, 200 s and 300 s. Figure 5a shows a SEM image of an ASA polymer surface coated with a conductive coating (to prevent charging effects) after creating a crater with ion beam milling (additional images of some of these craters are included in Fig. S3 in SI). In the central area of the image, the cubic-shaped crater can be observed, and around the crater, the conductive coating has partially been eliminated. This effect is due to FIB lateral beam spreading. When the ion beam is focused on a specific area, the ions also spread around this area, expanding the zone affected by the ions. This effect depends on various parameters, including the beam current - higher currents result in larger laterally affected areas38. The removal rate has been calculated by dividing the crater volume by the Ga+ dose required to remove that volume, and the results are plotted in Fig. 5b. As it can be observed, the sputtering rate is quite similar for both polymers. This is very beneficial in order to carry out a successful iSE tomography, as it ensures that the milled faces will be flat during the experiment and, thus, that the analysis will allow a true reconstruction of the 3D distribution of the polymers in the blend. For automatized iSE slice and view series, it is desirable that the same current is used both for imaging and for milling, to avoid the need of multiple re-alignments, which would extend the duration of the experiment. Because of this, a compromise needs to be reached in the ion current used to reduce the damage during the imaging step while allowing an effective slice milling. For the polymer blend considered in the present study, we have selected a compromise value of 0.05 nA for the tomography experiment. Fig. S4 shows iSE images taken with this current, at dwell times of 0.2 µs, 3 µs and 20 µs. As it can be observed, features are clearer for larger dwell times, as the signal to noise ratio is expected to be larger. However, damage due to the ion beam is also larger for larger dwell times. For dwell times of 0.2 µs, 3 µs and 20 µs, in images with resolution of 2809 × 3391 pixels, and considering the sputtering rates shown in Fig. 5b, depths of approx. 1 nm, 16 nm and more than 100 nm, respectively, are removed per iSE image acquisition. In order to find a balance between the quality of the images and the damage produced, and at the sight of the size of the features observed in the iSE image in Fig. 4b, 3 µs seems to be a reasonable dwell time for the slice and view experiment, as this surface damage seems not to have a strong deleterious effect on the results of the 3D polymers distribution obtained.

Fig. 5
figure 5

(a) SEM image of a crater milled in ASA with the aperture corresponding to a current of 0.1 nA; (b) removal rate (µm3/nC) vs. time for different ion currents, for ASA and for PC.

Full size image

Using the parameters determined above (0.05 nA and 3 µs) we have carried out slice and view experiments to compare the polymers distribution in the pieces manufactured by FFF and by IM. Figure 6 shows representative 2D iSE images of the ASA50 / PC50 blends obtained during the slice and view experiments. In Fig. 6a, corresponding to the ASA50 / PC50 manufactured by FFF, a surface with diagonally oriented contrasts is observed. As previously mentioned and corroborated by EDX, the clear and homogeneous contrasts correspond to PC, while the dark contrasts with dispersed round features correspond to ASA. Some small pores (black spots) are observed in the acrylate rubber of ASA likely due to the effect of ions, as these areas may be more sensitive to Ga+ ions bombardment. As it can be observed in Fig. 6a, the polymers are not completely mixed at the molecular level, i.e., a miscible state has not been reached as the two separate phases can be clearly distinguished. The visualization in 2D of the orientation of these phases suggests a directional structure and good dispersion, where the width of the domains is around 1 micrometer. Figure 6b shows a representative iSE image of the ASA/PC blend manufactured by IM, showing a clear separation between two distinct regions. A large clear and homogeneous area corresponding to PC occupies most of the image, while ASA appears as dark circular contrasts. In this case, the polymer individual domains are much larger than in the previous case, and their phase distribution is more heterogeneous, with large and small regions of ASA close to each other. However, care should be taken in the interpretation of individual 2D cross sections of 3D features, as sometimes results can be misleading. For example, the small round regions of ASA observed in Fig. 6a could be part of a much larger region, as this 2D section does not provide accurate information on the size of the domains. Because of this, a carefully analysis of the 3D data obtained in the slice and view experiments has been carried out.

Fig. 6
figure 6

iSE images of ASA50 / PC50 of the pieces fabricated by (a) FFF and (b) IM.

Full size image

Figure 7a and b show the tomographic reconstructions of the two blends analysed. For clarity, only ASA is represented, and different colours are used to indicate individual (unconnected) zones. Regarding the FFF object (Fig. 7a, the 3D analysis carried out corroborates the large directionality in the distribution of the polymers observed in 2D in Fig. 6a. The distribution of the phases follows the longitudinal axis of the analysed bead (y axis in Fig. 7a), with a small deviation likely due to the imperfect cut of the material piece prior to the FIB analysis, which was intended to be perpendicular to the bead longitudinal axes. Additionally, the 3D analysis carried out shows that the blend presents a co-continuous morphology, meaning that each phase forms a 3D spatially continuous network throughout the material. By observing a single continuous phase of ASA, for example the one represented with blue colour, a large degree of co-continuity (defined as the fraction of material that belongs to a continuous network)39 can be inferred. This polymers distribution is quite different to that found in the IM blend, shown in Fig. 7b. The morphology, in this case, exhibits a droplet-matrix structure, i.e., PC forms a continuous matrix, where ASA forms dispersed droplets in that matrix. This morphology has also been observed in polymer blends of PC / SAN, polystyrene/polypropylene (PS/PP), and PE / PP by 2D low voltage TEM37. The size distribution of the ASA droplets found in Fig. 7b is included in the histogram in Fig. 7c, showing that most of them present Mean Feret diameters < 1 μm. Thus, in both the FFF and the IM blends a good degree of mixing is obtained, with domains sizes of the order of the micrometer. Our results also show a good chemical interaction between ASA and PC in the two processes considered, as a tight interface between them without any gaps is clearly observed, making the ASA/PC a promising polymer blend for these manufacturing processes.

Fig. 7
figure 7

3D volume representation of the ASA regions in the ASA/PC blends fabricated by (a) FFF and (b) IM. The colors show the independent volumes of ASA surrounded by PC. (c) Histogram of size distribution of the ASA regions in the object fabricated by IM.

Full size image

At the sight of the results in Fig. 7, the manufacturing process seems to have a significant influence on the polymer distribution of the resulting polymer blend. In FFF, bead deposition produces ASA and PC regions homogeneously distributed and aligned in a preferred direction, which is the direction through which the polymers flow in FFF fabrication. This preferred alignment has been reported to occur in fibres in polymer/fibre composites fabricated by extrusion processes40,41,42, where this alignment of fibres usually produces an increase of stiffness and strength in the direction of the deposited material43,44. However, it also increases the anisotropy of the object, both in mechanical properties41,45,46 and in thermal expansion behaviour47. In the IM case, the droplets might represent a grain growth evolution of the ASA portion in PC, characteristic of a spinodal decomposition in a co-continuous structure48,49. This evolution in meta-stable polymers blends can be attributed to some extent to the melting time and volumetric melt rate of the manufacturing process, which allows (or not) the structure evolution. In FFF the deposition and cooling of the material from melting to below Tg is carried out in a few seconds, while in IM, the melting time can take several minutes (including the time spent melting the material, the time in the plasticizing chamber, the time spent filling the mould and cooling the injected part), so that this distribution can proliferate in agglomerates of ASA in PC. It is well known that, to prevent the evolution of the internal structure of materials, it is usual to incorporate compatibilizers between the polymer phases, which increase the interfacial adhesion between the polymers50,51. In the ASA/PC blend analysed in the present work, a tight interface without gaps between the polymers have been observed, therefore the use of compatibilizers may not be necessary.

On the other hand, it is worth mentioning that a different geometry in the distribution of polymers in a blend is expected to have an influence in the mechanical properties of solid pieces. In this sense, it has been reported that changes in morphology are manifest in improvement in the impact strength without loss of stiffness in iPP/EP blends52. It has also been reported that the morphology of the constituent phases is important in determining the viscoelastic properties of binary thermoset and thermoplastic polymer blends, affecting the magnitude of the mechanical coupling effects between the phases53. In relation to the results shown in Fig. 7 and analogous to the fibre/polymer composites explained above, it is worth mentioning that a distribution of polymers with large directionality is more anisotropic than the droplet-matrix structure observed, therefore if extended through an object it could affect the isotropy of its mechanical behaviour. Also, the relevance of the interfacial area between the components should be highlighted. The interfaces between two different polymers in a blend are expected to be regions of reduced strength because the chemical interaction between the constituents is normally weaker than that within the bulk of the component phases. In Fig. S5, the interfaces between ASA and PC are represented for both the printed and the injected samples. We have calculated the total interfacial area between ASA and PC over the analysed volume for the two materials studied, using the method described by Lindblad54 based on surface area estimation of digitized 3D objects through weighted local configurations. The calculated interfacial area between ASA and PC phases was 3200 μm² for the FFF sample and 1400 μm² for the IM sample, that it is approximately double in the piece fabricated by FFF than in the one manufactured by IM (for this calculation, a correction factor has been introduced to account for the deviation from the expected 50%ASA-50%PC composition found in the specific volumes analysed). A larger interfacial area is likely to have an effect in the mechanical strength of solid pieces. The geometry of these interfaces could also have an effect in the ease/difficulty of delamination. Thus, for the same interface size, somehow flat interfaces similar to those observed for the FFF object and located perpendicularly to the direction of application of a mechanical stress are likely weaker than the spherical ones found in the injected material. In general, the comparison of the mechanical properties of pieces obtained by different manufacturing processes is complex, as a number of factors need to be taken into account. Thus, FFF pieces usually show lower mechanical strength than IM ones because of the poor layer-layer adhesion and intrinsic porosity55. Our results have evidenced that the different morphology of the polymer blends obtained in the different fabrication methods should also be considered for understanding the mechanical behaviour of the pieces. A more detailed analysis considering different compositions as well as the thermal properties (DSC, thermogravimetry), the mechanical testing, and the morphology characterization by FIB is in progress to shed light at the properties of these and other interesting polymer blends.

Finally, we would like to highlight that the method shown in this communication for the analysis of the polymers distribution in blends could be also useful in other research fields, as the morphology of the phases and the size and shape of interfaces are determining parameters in the application of many polymers. For example, these characteristics have been reported to play a crucial role in semiconductor polymers blends, determining exciton-dissociation efficiency and charge-collection efficiency56. The capability of the FIB equipment for distinguishing polymers with similar composition using iSE images and for offering 3D information on the polymers distribution will allow understanding the effect of the fabrication parameters as well as shedding light at the functional properties to progress in the development of novel polymer blends.

Conclusions

In this work, we demonstrate the capability of iSE tomography for the analysis of the polymer distribution in ASA50 / PC50 FFF and IM polymer blends. Our results evidence that iSE images provide larger contrast between the phases than the commonly used SEM images, allowing the slice and view experiments. Monte Carlo simulations are used to obtain information of the trajectories of 2 keV electron and 30 keV Ga+ ion beams in polymer materials, showing that the lower projected range of ions (64.3 nm for incident electrons vs. 33.4 nm for Ga+ ions) could be responsible for the sharper contrasts observed. The sputtering rates produced by the Ga+ ion beam in ASA and in PC have been quantified experimentally, in order to optimize the parameters used in the 3D experiment, which resulted in a current of 0.05 nA and a dwell time of 3 µs. The 3D analysis carried out have shown that the morphology of the ASA/PC blends are highly dependent on the manufacturing method used. Pieces obtained by IM show a droplet-matrix structure, PC forming a continuous matrix where ASA is dispersed. On the contrary, blends obtained by FFF exhibit strong directionality and a co-continuous morphology, showing larger interface area between the two polymer constituents. Our findings evidence that FIB tomography using iSE images is well-suited for determining the 3D morphology in polymer blends, enabling accurate and detailed analysis of phase distribution and interface characteristics. Additionally, our results show a good interaction between ASA and PC in the two processes considered, as a tight interface between them without any gaps is clearly observed, and a good degree of mixing, with domains sizes of the order of micrometres, is found, making the AS/PC a promising polymer blend for FFF or injection moulding manufacturing processes.