A list of figures, tables and equations as well as the full protocol as pdf is included in the supplementary information.
· Because of the possible toxicity of the reagents and materials, and the possible presence of unknown contaminants, it is highly recommended to wear gloves during this procedure.
· All waste materials generated during this procedure should be disposed in the suitable container for chemical waste.
1 Preparation of Epon medium
· Prepare Spurr solution: Dissolve 13.3 g vinyl cyclohexene dioxide, 7 g DER736, 34.7 g 2-nonen-1- ylsuccinic anhydride and 0.53 g 2-dimethylamino ethanol in 50 ml of distilled water.
· Prepare Epon812 solution: mix 27.3 g Epon812 , 10.8 g 2-dodecen-1-ylsuccinic anhydride, 18.7 g methylnorbornene-2,3-dicarboxylic-acid and 0.9 g 2,4,6-Tris(dimethyl-aminomethyl)phenol with 50 ml distilled water.
· Mix equal amounts (usually 50 ml) of Spurr and Epon812 solutions.
2 Preparation of TEM specimen from mask
· Use a pair of scissors to cut a 1 cm x 1 cm square from the sample mask.
· Separate the different layers of textile.
· Cut a 1 mm x 5 mm piece of each layer.
· Transfer the layer samples into the silicone rubber embedding mould. Care should be taken to give the samples a correct orientation (Figure 1). A small piece of filter paper with the required references written on it in pencil is placed in each cup to label the samples. Optional: To obtain more of the sample on the grid, 2 pieces of textile can be brought in the same block of epon (on top of each other without spacing).
· Fill the mould with the Epon812-Spurr resin mixture and allow to polymerize for 48 hours at 70 °C in an embedding oven. Let cool down for 2 hours afterwards.
· After embedding, trim the specimen block with the EM TRIM2 trimming unit to obtain a cutting face of 0.5 x 1 mm2 to 1 x 2 mm2.
· Cut semi-thin sections with a thickness of 150 nm (interference colour in between gold and purple) using the Ultracut microtome.
· Bring the sections on carbon and pioloform-coated copper grids (150 mesh). If only TEM imaging is performed, the carbon layer on the grid is not necessary. For STEM-EDX analysis, the grid has to be coated with both carbon and pioloform to avoid specimen drift.
3 HAADF-STEM imaging of fibres and particles
For HAADF-STEM imaging and EDX analysis, a well-calibrated and well-aligned Talos transmission electron microscope operated at 200 kV is used.
· Load the grid into the single tilt/double tilt TEM holder and insert the holder into the TEM.
· Velox will open a window asking to specify the folder where the STEM images/EDX data cubes will be automatically saved and to define the concept for the file names.
· Load the appropriate FEG register or set the appropriate probe conditions for imaging beam sensitive materials where a large depth of field is needed: Smaller I (e.g. spotsize 9) and smaller α (e.g. CA 50 µm; α = 7.5 mrad).
· Adjust the z-height to focus the image and optimize the detector settings in Velox using the scope tool.
· Identify the fibres on the grid and record a series of at least 10 well-focused representative images of different fibres using the Velox software. Choose the magnification such that the fibre cross-section just fits on the image (Figure 2). These images will be used in section 5.3 for the determination of the fibre diameter.
· Record a set of images showing particles that are representative for the sample on the TEM grid and that are suitable for a subsequent quantitative analysis of the agglomerate and constituent particle size. Make sure to capture at least 100 particles (either aggregates/agglomerates at lower magnification or constituent particles at higher magnification) in line with FprCEN/TS1727314 to assure accurate median size measurements. For these masks, typically 10 images both at low and high magnifications is sufficient, although this amount depends on the particle density and the size of the aggregates/agglomerates. The specific imaging conditions for the imaging in STEM mode are summarized in Table 1 (see SI). The pixel size and the associated lower limit of quantification depend on the selected magnification and are determined based on the criterion of Merkus16. The upper size detection limit is limited to one tenth of the image size supporting on ISO 13322-1, 201417.
4 EDX spectral imaging (SI) of particles
· Keep the same recording settings as for the STEM imaging, i.e. dwell time 20 µm and image size 1024 x 1024 px2. The number of frames is typically set to 5. In case more signal is needed to get significant peaks in the EDX spectrum (which might be the case due to the high spot size), a higher value is advised. Alternatively, the option to stop the experiment manually can be selected.
· Record STEM + EDX SI images of the (nano-)objects of interest. Make sure not to overlook particles or aggregates/agglomerates of different elemental composition, particles inside and on the border of the fibres and (nano-)objects with special features. Choose an appropriate magnification to make sure the object of interest is well-centred in the image with still some background space around to allow for some drift between scans.
· By drawing a region on the STEM image using the rectangle tool, the EDX spectrum becomes visible. Choose to automatically match the peaks in the spectrum with the elements or add elements of interest (e.g. Ti, Ag,…) manually to the energy spectrum or the spectral maps by selecting them in the periodic table.
5 Image analysis
5.1 ParticleSizer for quantitative analysis
The physical properties of the particles of interest are measured based on the properties of their 2D projections. For constituent particles, the size distribution is determined based on manual measurement using ImageJ. For aggregates/agglomerates, the size distribution is determined using the NanoDefine ParticleSizer (https://imagej.net/plugins/particlesizer) plugin for ImageJ. This method allows determining the properties, such as the characteristic size and shape distributions, (semi-)automatically from electron micrographs.
· Open ImageJ and load the image series by File>Open>Image Sequence. Apply the image analysis settings summarized in Table 2 (see SI) on the pop-up window.
· Open the settings manager of the NanoDefine ParticleSizer plugin by Plugins>NanoDefine>SettingsManager. The settings summarized in Table 2 (see SI) are typically used for the analysis of agglomerates of TiO2 in fibres of face masks and can be applied as a starting point for further optimization.
· Run the ParticleSizer algorithm pressing Plugins>NanoDefine>ParticleSizer. An image with the identified particles overlaid in red, a table presenting the measured properties of the detected particles and a histogram of the distribution of the Fmin are generated as output. Save the table as csv file.
5.2 Calculation and visualization of the descriptive statistics
The results of the quantitative analysis of the detected nanoparticles were presented following the ISO 9276-1 (ISO, 1998) guidelines for representation of results of particle size analysis. A non-parametric approach is applied. For all samples, the normalized number-based distributions and the kernel density estimates (KDE) of the Fmin, Fmax, and AR of the constituent particles and of the aggregates/agglomerates are determined using the seaborn distplot function. The bin width is calculated using Freedman-Diaconis rule18. The mean and median values of the datasets are determined. For the KDE plot, producing a continuous density estimate, a Gaussian kernel is applied. The modes of the distributions are calculated from the KDE.
Python can be used to process the raw data resulting from the image analysis by ImageJ. Helpful packages that can be used to calculate the relevant descriptive statistics and plotting parameter histograms are numpy, scipy, pandas and seaborn. Generate a csv file summarizing the statistical parameters of the particle size distribution, including the Fmin, which will be used in the calculation in Section 6. Alternative software can be applied also.
5.3 Measuring the median fibre diameter
For each of the STEM images of the fibre cross sections, the fibre diameter is determined by manual measurement using ImageJ. Since for fibres which are not aligned parallel to the viewing direction in the TEM, the cross section appears as an ellipse rather than a circle, the diameter of the maximum inscribed circle is selected as the best measure for the fibre diameter.
· Use the polygon tool to draw a shape on the image representing the fibre cross section (Figure 2A).
· Click Plugins>BIOP>Image Analysis>Binary>Max Inscribed Circles.
· In the pop-up window, set the minimum disk diameter to 0 and check the box ‘Use selection instead of mask’ and click OK.
· The maximum inscribed circle appears on the image as annotation (Figure 2B). Find the diameter in pixel by using the circle tool and selecting the drawn circle.
· Covert to nm by looking up the pixel size of the image in Velox and multiplying it with the diameter in pixels.
Repeat this for all 10 images of fibres and determine the median fibre diameter (df).
6 Estimation of the amount of particles at the fibre surface
This estimation is valid on the assumption that the (agglomerated) particles are homogeneously distributed in the matrix of the fibres (this is usually the case for TiO2).
· Define F, the fraction of the particles at the surface of the fibres, as the ratio of an external ring-shaped surface of the cross-section of the fibres (Sr) and the total surface of the cross-section of the fibres (Scs).
F = Sr/Scs
· The thickness of the external ring-shaped surface, Sr, is determined by the median diameter (da) of the agglomerates (Figure 3). Then, approximate F, assuming circular cross-sections of fibres, as:
F = [(pi/4)*df2 - (pi/4)(df-da)2] / [(pi/4)*df2]
F = [df2 - (df-da)2] / df2
with df the median diameter of the fibres (see Section 5.3) and da the median minimum Feret diameter of the agglomerates (see Section 5.2).
· Now, calculate the mass of the particles at the surface of the fibres in a mask (Msf) as:
Msf = F * Mtot
with F the fraction of the particles at the surface of the fibres and Mtot the total mass of the particles in the mask. The latter quantity can be determined by inductively coupled plasma-optical emission spectroscopy (ICP-OES) as described in Verleysen et al.4
For a specific case of fibres, namely bi-component microfibres, the assumption of a homogeneous distribution of the agglomerated particles in the fibres is invalid. Bi-component microfibres are characterized by a larger surface area from which particles can be released. To account for this increased surface, Sr,mf, a correction factor should be introduced:
Sr,mf = Sr * [ SUM(wedges perimeter) / fibre diameter ]
With the sum of the perimeters of the wedge-shaped (nanoparticle containing) polyester parts of the fibre, and
the perimeter of the (near-)circular cross-section of the microfibre.