CHAPTER 2.1.2
Characterization of nanomaterials by transmission electron microscopy: Measurement procedures Jan Masta, Eveline Verleysena, Vasile-Dan Hodoroabab, Ralf Kaegic a Sciensano, Brussels, Belgium Bundesanstalt f€ ur Materialforschung und -pr€ ufung (BAM), Berlin, Germany c Swiss Federal Institute of Aquatic Science and Technology (Eawag), D€ ubendorf, Switzerland b
Introduction Transmission electron microscopy (TEM) is a versatile technique to analyse the size, morphology, crystallographic structure, and chemical composition of a wide range of nanomaterials (NM). It can be considered as a golden standard for the characterization of NM for several reasons [1]: (i) TEM analysis is one of the few methods that can provide a spatial resolution reliably covering the complete nanometre size range from 1 to 100 nm. (ii) TEM readily produces projected 2D images of NM. (iii) The combination of TEM imaging with image analysis allows determining the physical properties (size, shape, and surface morphology) of individual nano-objects quantitatively, based on the characteristics of their 2D projections. Multiple properties can be measured simultaneously for each individual particle, from which descriptive statistics and corresponding number-based distributions can be determined, as requested in regulations and guidelines [2–4]. (iv) TEM allows assessing the agglomeration and/or aggregation state of a material, and to some extent constituent, primary particles can be identified in agglomerates/aggregates. (v) Spectroscopic methods (EDS and EELS) can be incorporated in the TEM for elemental analysis of nano-objects and examination of chemical bonding allowing characterization of subpopulations of nano-objects in mixtures and nano-objects in the context of a complex matrix. (vi) Selected area electron diffraction (SAED) allows studying the crystallographic structure of nano-objects. Despite the wealth of information that can be obtained, the applicability of TEM to characterize nano-objects is currently limited by the following shortcomings: (i) The reliability of TEM analysis depends on the transfer of a representative fraction of the sample containing a sufficiently large amount of particles to the specimen carrier (TEM grid). This is influenced by the sample preparation, including purification and concentration
Characterization of Nanoparticles https://doi.org/10.1016/B978-0-12-814182-3.00004-3
© 2020 Elsevier Inc. All rights reserved.
29
30
Characterization of nanoparticles
steps, and by the TEM-specimen preparation. (ii) The spatial resolution of a transmission electron microscope depends on the acceleration voltage of the primary electrons. Consequently, conventional TEM typically applies acceleration voltages of 100–300 kV to obtain high spatial resolution. Although compatible with metallic or robust inorganic materials (so-called hard matter), the beam damage limits the analyses of beam-sensitive ‘soft matter’, such as certain zeolites, porous materials, polymers, hybrid materials, and selected carbon-based nanomaterials. Ke et al. [5] and Borrnert et al. [6] reviewed the advances of instruments applying aberration correction to overcome the aberrations due to the imperfection of the electromagnetic lenses in the electron microscope operated at low acceleration voltage, improving thus high-resolution imaging at low acceleration voltages. (iii) Conventional TEM applications are further limited by the artifacts resulting from the high vacuum conditions. Technological advances, such as liquid cell electron microscopy (EM) [7], environmental EM [8], and cryoEM-based applications [9], can help to overcome this constraint at least to some extent. (iv) TEM methods typically estimate the two dimensions of particles perpendicular to the electron beam but do not assess the dimension parallel to the beam. As a result, measurements of particles that have a preferential orientation on the EM grid, such as platelets, can be biased. Operation modes such as energy-filtered TEM (thickness mapping) and transmission electron tomography can be used to obtain information about the third dimension of the particles [10–16]. Clearly, the recent technological innovations can overcome the limitations of conventional TEM and are becoming essential tools in understanding the relation between the physical properties, structure, and composition of nanomaterials down to the atomic scale. Nevertheless, as many of the earlier-indicated advanced methodologies remain very cost and labour intensive, dedicated infrastructure remains limited to highly specialized research institutions. Initiatives for standardization and validation of these methodologies are only starting. In this regard, this chapter describes approaches for TEM characterization of nanomaterials using widely available equipment. The proposed measurement procedures aim to be applicable for NM that are presently produced and marketed in important volumes. The methodology requires that the fraction of nano-objects deposited on the EM grid is representative for the entire material and that individual nano-objects can be detected. To become suitable for TEM analysis, the samples and EM specimens need to be carefully prepared, for example, following to the guidelines described in Ref. [17]. The focus of this chapter lies on the description and measurement of the physicochemical properties accessible by TEM, which are essential in a legislatory and regulatory context to define the material as a NM [2], or to assess its safety and toxicological potential [3,4,18–20].
Characterization of nanomaterials by TEM
Measuring principle of transmission electron microscopy Conventional bright-field TEM A transmission electron microscope uses an electron source to generate a primary electron beam, which is accelerated by an electric potential and projected onto a thin specimen through a set of lenses and apertures. Fig. 1 illustrates how the electron beam passes through an electron microscope in TEM mode, under the influence of the lens and aperture settings. Part of the electron beam can pass through the specimen without interacting with it. Other electrons will be scattered by the atoms in the specimen. The information contained in the electron waves that exit from the specimen is used to create an image by means of the objective lens system. By the placement of an aperture in its back focal plane (i.e. the objective aperture), individual parts of the sample can be selected or excluded. In imaging mode (i.e. when an image is projected onto the viewing screen), a so-called bright-field image is formed when the aperture is positioned such that only the transmitted electrons can pass.
Fig. 1 Illustration of how an electron beam passes through the electron microscope in TEM mode (imaging and diffraction) and in STEM mode, under the influence of the lens and aperture settings.
31
32
Characterization of nanoparticles
A dark-field image is formed when the aperture selects only diffracted electrons. The image plane of the objective lens is projected onto a viewing screen, by means of a series of additional lenses that also ensure further magnification and correct positioning of the electron wave distribution. Usually, the viewing screen is a fluorescent screen, and a CCD camera is applied to record the image. At low magnifications, TEM image contrast originates from the absorption and scattering of electrons in the specimen, due to the thickness and composition of the material (i.e. mass–thickness contrast), and from the crystal orientation (i.e. diffraction contrast). By combining bright-field TEM imaging with image analysis based on grey values of large amounts of nano-objects, number-based distributions of the size and shape properties of the projected nano-objects can be obtained (Cfr. ‘Descriptive (qualitative) TEM analysis of nano-objects’ section). At high magnification (HR-TEM mode), the specimen can be modelled as an object that modifies the phase of the incoming waves. In a crystalline nano-object, elastic coherent scattering leads to diffraction of the incident electron beam on the regularly arranged atoms. The diffracted electron waves interfere constructively or destructively with the undiffracted transmitted wave depending on the defocus and phase shift of the waves in the specimen. The interference of transmitted and diffracted electrons gives rise to so-called phase contrast. As a consequence, the electrostatic potentials of the atom columns can be visualized and related to the crystal structure of the nano-object, although they are not a direct projection of the atom positions. For HR-TEM images to provide information on atomic arrangements, a comparison of the images with computer simulations based on atomic models is usually required. Besides giving information on the crystal structure, HR-TEM can be applied to study defects and to measure the thickness of a nano-object based on a focal series [21,22]. Diffraction from a 3D periodic structure, such as a crystal, is a consequence of interference between waves reflecting from different crystal planes. For a crystalline material, the electron beam undergoes Bragg diffraction. This can be used to obtain information about the crystal structure of crystalline nano-objects [23–27]. In the TEM, one can easily switch to electron diffraction mode by adjusting the lens settings such that the back focal plane of the objective lens rather than the imaging plane is projected on the viewing screen (see Fig. 1). Electron diffraction patterns of larger regions containing many crystalline nanoparticles consist of ring patterns analogous to those from X-ray powder diffraction. Such a ring pattern originates from the different orientations of the particles (which can be mono- or polycrystalline) with respect to the electron beam and can be used to identify texture besides obtaining crystallographic information. The recorded diffraction patterns should be indexed using a diffraction database (e.g. AtomWork [28]). An aperture can be inserted in the image plane of the objective lens (see selected area aperture in Fig. 1) to obtain a diffraction pattern solely from this selected area of the specimen (SAED).
Characterization of nanomaterials by TEM
More complex behaviour in the diffraction plane can also be observed. For example, in convergent beam electron diffraction (CBED), the interaction of the convergent beam with the specimen can provide information beyond structural data such as determination of the space group or determination of specimen thickness. A relatively new application specifically useful for studying NM is precession electron diffraction (PED) where the electron beam is tilted and precessed on a cone surface having a common axis with the TEM optical axis [29,30]. PED allows to solve ab initio crystal nanostructures, to perform artifact-free 3D BF image reconstruction, and to create nanometre resolution EBSD-TEM like orientation/phase maps.
Scanning transmission electron microscopy Similar to a SEM, a TEM can be operated in scanning mode (STEM), where the electron beam is focused into a (sub)nanometre-sized probe and scans over the specimen. The spatial resolution that can be obtained is mainly controlled by the nanosized illuminating electron probe, down to 0.1 nm in high-resolution STEM mode, when a probecorrected TEM is used. Fig. 1 illustrates how an electron beam passes through the electron microscope in STEM mode, under the influence of the lens and aperture settings. Various signals produced by the scattering of the electrons can be detected and displayed as a function of the probe position. The interactions with the specimen give accurate localized physical and chemical information. Often, images are created using either small on-axis detectors (bright field) or large annular detectors with a fixed size (dark field). Similar to HR-TEM, BF-STEM can be used to visualize atomic columns. If an annular detector is used and the inner collection semiangle is set up to collect only diffracted beams, a diffraction contrast dark-field image is obtained. This is typically used for a specimen composed only of light elements. High-angle annular dark-field (HAADF) imaging refers to the use of a geometrically large ring-shaped detector, placed in the optical far field beyond the specimen. Because of the detector geometry, only electrons that are scattered at high angle are detected. While electrons scattered under low angles are predominantly coherent (and as such conventional bright-field and dark-field images can give contrast reversals with changes in specimen thickness, orientation, or defocus), electrons scattered to high angles are predominantly incoherent. The incoherence arises mainly from the geometry of the detector and from contributions of thermal diffuse scattering [31–34]. Since high-angle incoherent scattering is associated with scattering from the atomic nuclei, it gives an intensity that is more or less proportional with the squared atomic number (I Z1.6–2), as expected on the basis of Rutherford scattering. This imaging mode is therefore also called Z-contrast imaging. One of the advantages of HAADF-STEM imaging is the possibility to visually distinguish materials with a different chemical composition. Furthermore, if the elements contained in the specimen are known, an
33
34
Characterization of nanoparticles
HAADF-STEM image gives directly a qualitative 2D distribution of the different materials in the specimen. Since the annular detector is used to exclude Bragg scattering and eliminate the phase problem, contrast reversion of HAADF-STEM images by the change of defocus and thickness is largely suppressed.
Analytical TEM The elemental composition of the sample can be obtained by analysing either the energy of the emitted X-rays [energy dispersive X-ray analysis (EDS)] or the energy loss of the transmitted electrons [electron energy loss spectroscopy (EELS)]. Both microanalysis techniques can be incorporated in the TEM system, and EDS can be incorporated in most SEM systems as well. Inelastic scattering occurs when there is an exchange of energy between a primary electron from the electron beam and the specimen. If sufficient energy is transferred, an electron from the inner atomic shells of the specimen material is ejected, and a vacancy is formed, leaving the atom in an excited state. To stabilize the atom, an electron in one of the outer atomic shells (higher energy level) will subsequently fill the vacancy in the inner atomic shell (lower energy level). The energy difference between the two energy levels will be released during the transition in the form of an X-ray or an Auger electron. EDS measures the energy of the X-rays that are emitted from the specimen. Because the X-ray energy is characteristic for the electronic structure of the atom, it can be used for both qualitative and qualitative analysis. Elements with atomic numbers ranging from that of beryllium to uranium can be detected. The minimum detection limits vary from approximately 0.1 to a few atomic percent, depending on not only the element specimen but also the type of EDS detector used. In addition, quantitative results can be obtained by extracting the intensities of the characteristic peaks of the constituting elements after correcting for signal intensities related to the background (Bremsstrahlung), based on a calculated or experimentally determined sensitivity factor. In a TEM, EDS is usually performed in STEM mode to obtain high spatial resolution. EDS elemental mapping combines the spatial information obtained in STEM mode with the X-ray signals to construct elemental distributions of each detected element in a predefined region. More recent high-throughput EDS systems record the entire EDS spectrum at every location of the primary electron beam allowing to extract virtually any elemental distribution maps from the recoded data cube. The elemental distribution maps can be overlaid with the STEM image. For more details and examples on EDS operating with a SEM, see Chapter 4.4. EELS (see Fig. 2) measures the energy that is lost by the inelastically scattered primary electrons to obtain information on the scattering atoms in the specimen. The energy required to remove an electron from one of the inner shells of an atom in the specimen is element specific. Consequently, the energy that the primary electrons lose by exciting
Characterization of nanomaterials by TEM
Fig. 2 Illustration of how the electron beam passes through the electron microscope in STEM mode and where the EDS and EELS detectors are positioned.
one of the atoms in the specimen is also element specific and can be used to obtain both qualitative and quantitative information about the elemental composition of the specimen [35,36]. Quantitative analysis in EELS typically uses calculated cross-sections after background removal and deconvolution to exclude multiple scattering effects. EELS and EDS are therefore both related to the same phenomena (ejection of an electron from the inner shell of an atom) but use different ways of detecting this process. EELS is generally considered to be complementary to EDS. While EDS excels in identifying the elemental composition of a material, is quite easy to use, and is particularly sensitive to heavier elements, EELS tends to work better for elements with relatively low atomic number, where the excitation (core-loss) edges are sharper and well defined. Elemental distribution maps based on EELS can be obtained by either sequentially acquiring an EELS spectrum from each position of the (focused) primary electron beam (STEM EELS) when scanning the electron beam over the sample or by collecting a series of images (parallel illumination) acquired at increasing energy loss [energy filtering (EF) TEM] around the excitation edge of the element of interest.
35
36
Characterization of nanoparticles
Due to the higher energy resolution of EELS compared with EDS (<1 eV for EELS and 140 eV for EDS), an advantage of EELS is the ability to ‘fingerprint’ different forms (speciation) of the same element (e.g. graphite vs diamond) by comparing the fine structure of core-loss edges. This fine structure is called energy loss near edge spectrum (ELNES) and contains information about the electronic structure of the specimen. In addition, EELS provides a methodology for thickness determination, termed the logratio method [36].
Specimen preparation for TEM analyses The described methods for specimen preparation are suitable for nanoparticles present in liquid dispersions. In the case of dry powders, several reports have provided guidelines for developing and reporting appropriate dispersion protocols [37–44]. Furthermore, several dispersion protocols for different materials have been developed in different EU projects (e.g. NanoDefine, Nanolyse, and NanoREG 1 and 2), and the protocols can be found in the respective public reports. To goal of the specimen preparation procedure is to achieve an even distribution of the particles of interest on a suitable carrier, in this case a TEM grid. A sufficiently high number of particles have to be deposited ideally as single particles on the TEM grid, but at the same time, care has to be taken not to overload the grid with particles because overlapping particles will limit the applicability of automated image analysis procedures to extract accurately the measurands of interest from recorded TEM images. As a rule of thumb, a coverage with nanoparticles of a few percent of the TEM grid surface is well suited for later (automated) image analyses.
Choice of sample carriers for TEM analyses For TEM analyses, the particles have to be deposited representatively on a dedicated TEM grid. These are standardized grids, mostly 3.05 mm in diameter, with different meshed sizes, made of various metals. Cu is the most commonly used material for this purpose, but gold, nickel, and molybdenum are also commercially available. These grids are not only coated with electron transparent foils, mostly carbon and/or formvar, but also grids coated with silicon oxide are available. In addition, foils of Si3N4 of different sizes and thickness are available, which can be beneficial for the analysis of carbon-based materials. As modern scanning electron microscopes (SEM) are often equipped with the transmission operation mode, STEM in SEM or TSEM [45,46] (see Chapter 2.1.1), we generally recommend using TEM grids as specimen carriers for nanomaterials. In this way, one keeps the flexibility of investigating the specimen on both TEM and SEM.
Characterization of nanomaterials by TEM
Tailoring TEM grids Particles in dispersion carry a surface charge, commonly referred to as zeta potential. For an efficient and reliable deposition of the particles on the TEM grid, the surface of the TEM grid has to be opposite of the charge of the particles. In many cases, the charge of the particles in the respective dispersion is known. If this is not the case, the charge can be assessed by measuring the electrophoretic mobility of the particles in dispersion, for example, using phase analysis light scattering. Freshly prepared carbon-coated TEM grids are generally hydrophobic and tend to have a slight negative charge. Over time, the surface of the TEM grids gradually becomes more hydrophilic. Most commonly, nanoparticles will be dispersed in aqueous (polar) media, and the TEM grids therefore should be hydrophilic, which can be achieved by a glow discharge procedure. If particles in the dispersion carry a positive charge, no further treatment of the grid is required. However, if particles carry a negative charge, the TEM grids should additionally be treated with, for example, Alcian blue [47] or poly-L-lysine [48] to render the TEM grids positively charged.
Sample preparation Drop deposition One of the simplest methods to bring particles from liquid dispersion onto the TEM grid includes the drop-on-grid or grid-on-drop technique. After TEM grid functionalization, for example, glow discharge followed by Alcian blue treatment, the grid is floated on a drop of nanoparticle dispersion (grid on drop), or a drop of dispersion is placed on the TEM grid (drop on grid). After a selected time, ranging from a few seconds to a few minutes, TEM grids are removed from the liquid and washed two to three times in a drop of deionized water. This washing step prevents the formation of drying artifacts, for example, formation of solid precipitates of salt particles, which may mask the particles of interest on the TEM grid. Detailed description of these specimen preparation procedures can, for example, be found in Refs. [41,49]. The drop-on-grid or grid-on-drop technique can be applied without any special equipment and is thus easily being accommodated in any standard laboratory. As only a small fraction of the nanoparticles from the dispersion is deposited on the TEM grid, to ensure sufficient measurement statistics, the method requires a rather high particle concentration in the dispersion, up to grammes/ litre under the assumption that no significant agglomeration takes place at deposition. Furthermore, an accurate quantification of the absolute particle number concentration in the dispersion is not possible on a routine basis. On-grid (ultra)centrifugation The alternative approach to deposit particles from liquid suspension on the TEM grid includes (ultra)centrifugation of the particles directly on the TEM grids. This can be
37
38
Characterization of nanoparticles
achieved by either using a centrifuge with dedicated rotors for TEM grids (e.g. Airfuge, Beckman Coulter, USA) as described in Ref. [50] or a standard centrifuge in combination with a swing-out rotor and custom made flat-bottom centrifugation vials [51,52]. The latter approach has been refined by, for example, fabricating reusable aluminium cones that fit into standard (1.5 mL) Eppendorf tubes making the implementation of this approach also accessible for common laboratories. A YouTube video explaining the samples preparation in detail can be found at https://www.youtube.com/watch? v¼PplBlJ7zCCA. The centrifugation method results in a deposition of all particles from the water column above the TEM grid. Thus, in addition to the particle size distribution, also the absolute number concentration of the particles in the dispersion can be estimated based on the particles detected on the TEM grid. Based on an estimated particle size and density on the grid, the optimal particle concentration in the liquid dispersion, the centrifugation speed (g force), and time can be calculated. For this purpose, several online tools are available, for example, at https://www.eawag.ch/en/department/eng/ main-focus/particle-laboratory/. Also with the centrifugation technique, the grids have to be washed in deionized water to avoid the precipitation of salt particles on the TEM grid. ESI deposition of suspensions Most recently, an interesting approach using an electrospray to aerosolize the dispersions followed by an electrostatic deposition of the particles onto electrically conductive TEM grids has been presented [53]. In this technique, small droplets are generated by an electrospray device. A thin jet of suspension is ejected from a conically shaped meniscus (Taylor cone) as end part of the liquid suspension in a capillary tip and accelerated to the TEM grid as the collector. Due to electrostatic repulsion, the droplets explode, leading to drying and finally direct deposition of particles (without solvent) on the TEM grid. Initial results confirmed that the electrospray is rather tolerant to various liquid dispersions, and good particle coverage for a range of different materials was obtained. It could be demonstrated that there are no significant loses of particles during electrospraying on substrate. Thus, this approach would enable determination (at least roughly) of particle number concentration in liquid suspension. Although this is a very promising method, it requires sophisticated instrumentation, which unfortunately is currently not commercially available.
Descriptive (qualitative) TEM analysis of nano-objects Before more advanced analyses are initiated, a detailed description of the material present on the EM grid is a first and essential step to assess the quality of the sample and specimen preparation and to determine the basic properties of the examined nano-objects. Based on a series of representative electron micrographs recorded at magnifications ranging from 100 to 100,000 times and covering the entire specimen, an overview of
Characterization of nanomaterials by TEM
the specimen and properties of the nano-objects of interest such as size, shape, surface and inner morphology, and crystallographic structure can be visualized in the electron microscope. A descriptive TEM analysis includes, at least, representative and calibrated micrographs, a description of (i) the roughly estimated size (distribution) of the primary and aggregated/agglomerated particles, (ii) the agglomeration and aggregation status, (iii) the general morphology, (iv) the surface topology, (vi) the structure (crystalline, amorphous, etc.), and (vii) the presence of contaminants and aberrant particles. In addition, selected micrographs can highlight unusual or rare features, such as impurities, large agglomerates, and crystal defects. A descriptive TEM analysis should further report the specimen quality describing possible impurities and artifacts, the amount of particles on the EM grid, their contrast with the background, and how even they are distributed on the TEM grid. Supporting on these parameters, the relevance and suitability of a quantitative EM analysis based on the examined specimen can be assessed. The proposed methodology complies with the EFSA guidance documents [18] that foresee application of TEM (or SEM). It describes several key parameters important to assess the nanomaterial safety as specified [18, 54–56]. A report of a descriptive TEM analysis also includes the characteristics of the nano-objects of interest such as aggregates and agglomerates and primary particles, as specified in Table 1. Formal guidelines for the unambiguous and detailed qualitative description of a nanomaterial are currently lacking. Krumbein and Sloss [57], Munoz-Marmol [58], and Lopeze-de-Uralde [59] propose systems to categorize and describe the surface structure, the particle shape, and the shape of aggregates and agglomerates, respectively. A scheme for systematic and uniform reporting, as exemplified in Table 1, facilitates efficient reporting of descriptive EM results of a large number of samples. It can increase throughput of qualitative EM analyses and assure uniformity in reporting, reducing misunderstandings by suggesting a set of possible descriptions.
Quantitative TEM By combining TEM imaging with image analysis, individual nano-objects of interest can be characterized quantitatively, and properties like the size, shape, and surface texture from their two-dimensional projections areas can be estimated.
TEM imaging TEM imaging for quantitative TEM analysis aims to record a set of calibrated electron micrographs showing objects that are representative for the objects deposited on the EM grid. The imaging conditions, magnification of the micrographs, and the number of particles have to be chosen such that they are suitable for subsequent quantitative image analyses.
39
Table 1 Example of a scheme facilitating systematic reporting of descriptive EM analyses Level
Property
Description
Specimen
Impurities
• Presence
Quantitative EM analysis
Aggregation/ agglomeration
Primary particles
State
• •
• • • •
Size
•
Shape
• •
Polydispersity (size)
•
Size
•
Crystal structure
• •
Two-dimensional shape Three-dimensional shape Surface topology
• • •
The sample is pure; no impurities are found Occasionally, an impurity is observed Nanoparticles are embedded in a matrix Between the impurities, the NP are visible Micrographs only contain impurities; no nanoparticles are observed Impurities consist of … (general description) A quantitative TEM analysis is feasible The EM specimen is representative for the sample The particles are evenly distributed over the grid The particles can be distinguished from the background and matrix A quantitative analysis is not feasible The particles are individual particles The particles are agglomerated and agglomerated The particles contain X to Y particles per agglomerate and on average X particles Estimated size Approximately XX nm Smaller than XX nm Ranging from XX nm to XX nm The minimal aggregate/agglomerate size is … Described according to Lo´pez-de-Uralde [59] Spheroidal Ellipsoidal Linear Branched/dendritic Type of polydispersity Monomodal distribution Bimodal distribution Trimodal distribution Polymodal or broad distribution Estimation of size Approximately XX nm Smaller than XX nm Ranging from XX nm to XX nm XX XX nm for N ¼ XX The minimal primary particle size is … The crystal structure Particles show diffraction contrast suggesting a crystalline structure Particles show no diffraction contrast suggesting an amorphous structure Described according to the taxonomic hierarchy of 2D shapes of Munoz-marmol et al. [58] Described according to the taxonomic hierarchy of 3D shapes of Munoz-marmol et al. [58] Described according to Krumbein and Sloss [57]
Characterization of nanomaterials by TEM
Aligning the microscope and calibration As most electron microscopes are computerized, there are a growing number of users that rely entirely on the (semi)automated, software-assisted alignment tools that can set most parameters automatically in minutes, depending on the type and make of the TEM. For most purposes, doing what the computer and the instruction manual tell you to do is sufficient to get reasonable results [60]. Suitable alignment of the microscope remains however a prerequisite to reaching optimal lens conditions for TEM imaging and asks for daily evaluation, particularly if high resolution is ambitioned [61]. The papers of Rodenburg et al. provide elementary tutorials in the main elements of beam alignment for TEM [60] and STEM [62]. Saxton et al. [63] and Pennycook and Williams [35] provide a theoretical and experimental basis of the different approaches to the alignment, stigmating, and focusing of electron microscopes and make recommendations to the best procedures to adopt in various circumstances. To assure the precision and accuracy of TEM measurements and to relate them to the international system of units (SI), calibration of the TEM is critical. The guidance document ISO 29301 [64] provides specific guidance for magnification calibration of the images over the applied magnification range and describes methods for calibration of the image magnification in STEM, respectively, using a reference material with periodic structures. The reference materials used for calibration possess a periodic structure, such as a diffraction grating replica; a superlattice structure or an analysing crystal for X-ray analysis; and a crystal lattice image of carbon, gold, or silicon. An advantage of these materials is that they are suitable for automated calibration using specialized software. A disadvantage is that they are not directly SI traceable. Therefore, a complementary calibration based on measurement of the size of a certified reference materials with an SI-traceable size is advisable. Imaging conditions To assure that the recorded micrographs are representative for the specimen, selectivity during imaging should be avoided. For specimens with particles that are evenly distributed over the EM grid, selectivity of the imaged particles can be avoided by taking the micrographs using a random and systematic sampling scheme. Verleysen et al. [49] and De Temmerman et al. [41] recorded, for example, images at several positions, predefined by the microscope stage, that were distributed over the entire grid, reducing subjectivity introduced by the operator. When the field of view at a specific position was not suitable, for example, because it was obscured by a grid bar or contained an artifact, the stage was moved sideways to the nearest suitable field of view. Preferential detection of particles is avoided by setting the measurement frame as described in ISO 13322-1:2014 [65]. Biased particle measurements are avoided by eliminating particles near the borders of the micrograph.
41
42
Characterization of nanoparticles
Because the contrast of particles obtained using minimal contrast focusing (Gaussian) can be so low that particle detection using image analysis software is difficult, it can be useful to record the images slightly under focus. This facilitates particle detection because it increases contrast. Magnification selection and limits of quantification Approaches to select the proper magnification aim to make an objective compromise between examining enough of the grid surface and thus measuring enough particles and having the resolution to detect and measure them accurately. Typically, the magnification is chosen as such that the size of the smallest particle that needs to be measured and that is estimated during a descriptive TEM analysis (Cfr. ‘Descriptive (qualitative) TEM analysis of nano-objects’ section) is larger than the lower limit of quantification. For a given magnification and corresponding pixel size, the lower quantification limit can be calculated based on the work of Merkus [66], who showed that large systematic deviations in size measurements can be avoided if the particle area of equiaxial particles consists of at least 100 pixels. To measure the dimensions of nonequiaxial particles, as a rule of thumb, the minimal dimension that can be measured accurately by the software should consist of at least 10 pixels. The method quantification limit is higher than the limit of detection. The detection limit for quantitative TEM can be seen as the number of pixels of a particle (size) that yields enough contrast with the background to allow detection of the particles by the image analysis software. This is determined by the mass–thickness contrast and the diffraction contrast of the particles depending on the thickness, the atomic number, and the orientation of the particles. Obviously, these limits are larger than the instrument detection limit, which is the resolution of the TEM. For TEM, line resolutions in the subnanometre range are typically claimed and confirmed by visualization of a specific lattice spaces of crystalline ˚ and thus can quanmaterials. High-resolution TEM and STEM obtain resolutions of 1 A tify, applying the criterion of Merkus, particles of 1-nm size. The upper size quantification limit is restricted by the number of pixels in the field of view of the applied detector/camera. To avoid bias of the measurement of the large particles, it usually is set to the size corresponding to the number of pixels making up onetenth of the image size in one dimension, as proposed in ISO 13322-1 [65]. The useful working range is defined by the lower and upper size quantification limits. Table 2 presents an example for a 4K CCD camera (40,960 by 4096 pixels), with a camera pixel size of 15 15 μm2.
Particle detection and identification Individual nano-objects of interest, visualized in TEM images can be detected and identified based on criteria such as mass–thickness contrast, elemental composition, size, morphology, and crystallographic structure.
Characterization of nanomaterials by TEM
Table 2 Limits of quantification for a given microscope and CCD camera configuration at indicated magnificationsa
Magnification Pixel size (nm) CCD field of view (mm2) Sample field of view (nm2) Lower size quantification limit (nm) Upper size quantification limit (nm) Useful quantification range (nm) a
Condition 1
Condition 2
18,500 0.81 61.4 61.4 3318 3318 8.1 332 8.1–332
68,000 0.22 61.4 61.4 901 901 2.2 90 2.2–90
4K CCD camera (4096 by 4096 pixels), with a camera pixel size of 15 15 μm2.
In conventional transmission electron micrographs, particles are typically distinguished from the background based on their mass–thickness and diffraction dependent grey value. In second instance, the detected particles are identified based on their size and shape properties. Structures with clearly deviating size and shape are considered as contaminant materials and may be omitted from analysis. Detection of particles from STEM images has the advantage that particles with different elemental compositions can be distinguished based on their Z-contrast. Manual identification and measurement of particles have been successfully applied in many cases and remain necessary to validate new, automated approaches [67] and in complex samples [68]. Because manual detection and measurements are prone for operatordependent bias, relatively time-consuming, and labour intensive, and because only a limited number and type of measurands can be measured, (semi)automatic approaches are developed and implemented. These can measure particle properties more efficiently and accurately [69–71]. Automated image analyses have the advantage that they can be integrated, with automated imaging and reporting in completely automated EM-based measurement of particle properties. ISO 13322 [65] provides a standardized description of the technique used to analyse particles, including nano-objects, in static images, including electron micrographs. A typical image analysis consists of several consecutive steps. In first instance, images are prepared for particle detection facilitating the separation of particles of interest from the background. Preparative steps include the correction of uneven image illumination, the reduction of the noise of the images, and the optimization of the contrast. For this purpose, a (combination of ) series of image filters, reviewed in Ref. [72], have been applied, ranging from simple median filters to complex waveletbased algorithms [73]. These filters aim to preserve signal details and simultaneously significantly suppress the noise level. Secondly, a threshold value is set and adjusted such that the separation of the particles from the background reflects the real particle boundaries. When the image preparation
43
44
Characterization of nanoparticles
Fig. 3 Illustration of the approach of the Particle Sizer software that allows applying different image analysis algorithms depending on the type of particle and type of overlap.
step results in a uniform background and a clear contrast of the particles, a global cut-off value can be efficient to identify all particles of the entire image. In case the background remains irregular, as frequently is observed for particles in complex matrices, an approach based on local thresholding or by edge detection filters is useful. Depending on the type of particles and their overlap, the preparative steps and the thresholding strategies require adaptation and optimization. This is illustrated in Fig. 3. The so-called particle sizer, an image analysis software developed to implement the EC NM recommendation for a definition, allows applying different image analysis algorithms depending on the type of particle and type of overlap (none, touching, slightly overlapping, and high or completely overlapping) [74]. Automated image analysis procedures for particulate materials imaged by electron microscopy allow determining automatically from EM images the distributions of the characteristic properties of aggregates and agglomerates and of primary particles in aggregates and agglomerates or present as single particles. A range of size and shape measurands (descriptors) can be assessed. Different size and shape measurands can be selected in function of the purpose of the analyses according to ISO 9276-6:2008.
Number of particles to be analysed The number of particles that needs to be analysed for the estimation of the mean value of a given measurand with a certain statistical error can be determined numerically following the method of Masuda [75], elaborated in ISO 13322-1:2014 [65]. This method assumes a log-normal distribution of the measurand and gives general analytical solutions for the distribution of sample-mean diameter and the number of particles required.
Characterization of nanomaterials by TEM
Alternatively, the relation between the number of measured particles and the measurement uncertainty can be determined for each relevant measurand (and in the case of a multimodal material, for each subpopulation), via, for example, a validation study. The number of particles required to estimate a specific measurand with a given uncertainty is estimated from this relation [76–78].
Data analysis and representation Detailed guidelines for further representation of results are provided in ISO standards. The results of particle size analysis can be represented according to ISO 9276-2:2014 and ISO 9276-5:2005. The particle shape and morphology can be described and quantitatively represented according to ISO 9276-6:2008. A function can be fitted to the distributions as in ISO 9276-3:2008. Particles can be classified according to ISO 9276-4:2001. Currently, an ISO standard (ISO/DIS 21363) dedicated to measurement of nanoparticle size and shape distributions including many instructive case studies is in its final development phase.
References [1] T.P.J. Linsinger, et al., Requirements on Measurements for the Implementation of the European Commission Definition of the Term ‘Nanomaterial’, (EUR 25404 EN). 2012. [2] EC, Commission recommendation on the definition of nanomaterial, in: Commission Recommendation, European Commission, Brussels, 2011. [3] OECD, Guidance Manual for the Testing of Manufactured Nanomaterials: OECD Sponsorship Programme: First Revision (ENV/JM/MONO(2009)20/REV), Organisation for Economic Co-Operation and Development, Paris, 2010. [4] OECD, Guidance on Sample Preparation and Dosimetry for the Safety Testing of Manufactured Nanomaterials (ENV/JM/MONO(2012)40), Organisation for Economic Co-Operation and Development, Paris, 2012. [5] X. Ke, C. Bittencourt, G. Van Tendeloo, Possibilities and limitations of advanced transmission electron microscopy for carbon-based nanomaterials, Beilstein J. Nanotechnol. 6 (2015) 1541–1557. [6] F. Borrnert, et al., A flexible multi-stimuli in situ (S)TEM: concept, optical performance, and outlook, Ultramicroscopy 151 (2015) 31–36. [7] F.M. Ross, Opportunities and challenges in liquid cell electron microscopy, Science 350 (6267) (2015) aaa9886. [8] T. Walther, What environmental transmission electron microscopy measures and how this links to diffusivity: thermodynamics versus kinetics, J. Microsc. 257 (2) (2015) 87–91. [9] P.L. Stewart, Cryo-electron microscopy and cryo-electron tomography of nanoparticles, Wiley Interdiscip. Rev. Nanomed. Nanobiotechnol. 9 (2) (2017). [10] G. Mobus, R.C. Doole, B.J. Inkson, Spectroscopic electron tomography, Ultramicroscopy 96 (3–4) (2003) 433–451. [11] P.A. Midgley, M. Weyland, 3D electron microscopy in the physical sciences: the development of Z-contrast and EFTEM tomography, Ultramicroscopy 96 (3–4) (2003) 413–431. [12] S. Van Aert, et al., Three-dimensional atomic imaging of crystalline nanoparticles, Nature 470 (7334) (2011) 374–377. [13] S. Sueda, K. Yoshida, N. Tanaka, Quantification of metallic nanoparticle morphology on TiO2 using HAADF-STEM tomography, Ultramicroscopy 110 (9) (2010) 1120–1127.
45
46
Characterization of nanoparticles
[14] E.M. Pouget, et al., The initial stages of template-controlled CaCO3 formation revealed by cryo-TEM, Science 323 (5920) (2009) 1455–1458. [15] E. Van Doren, et al., Determination of the volume-specific surface area by using transmission electron tomography for characterization and definition of nanomaterials, J. Nanobiotechnol. 9 (1) (2011) 17. [16] T. Malis, S. Cheng, R. Egerton, EELS log-ratio technique for specimen-thickness measurement in the TEM, J. Electron Microsc. Tech. 8 (2) (1988) 193–200. [17] J. Mast, E. Verleysen, P.-J. De Temmerman, L.D. Francis, A. Mayoral, R. Arenal (Eds.), Physical characterization of nanomaterials in dispersion by transmission electron microsco-py in a regulatory framework, in: Electron Microscopy of Materials, Springer International Publishing AG, Cham, Switzerland, 2014. [18] EFSA, Scientific opinion: guidance on the risk assessment of the application of nanoscience and nanotechnologies in the food and feed chain, EFSA J. 9 (5) (2011) 2140. [19] SCENIHR, Risk Assessment of Products of Nanotechnologies, Scientific Committee on Emerging and Newly Identified Health Risks, 2009. [20] E.A.J. Bleeker, et al., Considerations on the EU definition of a nanomaterial: science to support policy making, Regul. Toxicol. Pharmacol. 65 (1) (2013) 119–125. [21] L. Audinet, et al., Structural properties of coated nanoparticles: the CdS/ZnS nanostructure, Philos. Mag. A 79 (10) (1999) 2379–2396. [22] M.A. O’Keefe, E.C. Nelson, L.F. Allard, Focal-series reconstruction of nanoparticle exit-surface electron wave, Microsc. Microanal. 9 (S02) (2003) 278–279. [23] R. Jin, et al., Photoinduced conversion of silver nanospheres to nanoprisms, Science 294 (5548) (2001) 1901–1903. [24] S. Sun, et al., Monodisperse FePt nanoparticles and ferromagnetic FePt nanocrystal superlattices, Science 287 (5460) (2000) 1989–1992. [25] M. Li, H. Schnablegger, S. Mann, Coupled synthesis and self-assembly of nanoparticles to give structures with controlled organization, Nature 402 (6760) (1999) 393–395. [26] I. Srnova´-Sˇloufova´, et al., Core-shell (Ag) Au bimetallic nanoparticles: analysis of transmission electron microscopy images, Langmuir 16 (25) (2000) 9928–9935. [27] T. Klaus, et al., Silver-based crystalline nanoparticles, microbially fabricated, Proc. Natl. Acad. Sci. U. S. A. 96 (24) (1999) 13611–13614. [28] Y. Xu, M. Yamazaki, P. Villars, Inorganic materials database for exploring the nature of material, Jpn. J. Appl. Phys. 50 (11S) (2011) 11RH02. [29] R. Vincent, P. Midgley, Double conical beam-rocking system for measurement of integrated electron diffraction intensities, Ultramicroscopy 53 (3) (1994) 271–282. [30] S. Nicolopoulos, D. Bultreys, Precession Electron Diffraction and TEM Applications, NanoMEGAS SPRL, http://www.nanomegas.com/Documents/Precession%20Applications.pdf. [31] P. Nellist, S. Pennycook, Incoherent imaging using dynamically scattered coherent electrons, Ultramicroscopy 78 (1–4) (1999) 111–124. [32] M. Shiojiri, T. Yamazaki, Atomic resolved HAADF-STEM for composition analysis, Jeol News 38 (2003) 54. [33] S. Pennycook, D. Jesson, High-resolution Z-contrast imaging of crystals, Ultramicroscopy 37 (1–4) (1991) 14–38. [34] P. Hartel, H. Rose, C. Dinges, Conditions and reasons for incoherent imaging in STEM, Ultramicroscopy 63 (2) (1996) 93–114. [35] S.J. Pennycook, B. David, C.B. Williams, Transmission electron microscopy: a textbook for materials science, Microsc. Microanal. 16 (1) (2010) 111. [36] R.F. Egerton, Electron Energy-Loss Spectroscopy in the Electron Microscope, Springer Science & Business Media, 2011. [37] ISO, Sample Preparation—Dispersion Procedures for Powders in Liquids (ISO 14887:2000), International Organization for Standardization, Geneva, 2000. [38] J.S. Taurozzi, V.A. Hackley, M.R. Wiesner, Ultrasonic dispersion of nanoparticles for environmental, health and safety assessment—issues and recommendations, Nanotoxicology 5 (4) (2011) 711–729.
Characterization of nanomaterials by TEM
[39] B. Michen, et al., Avoiding drying-artifacts in transmission electron microscopy: characterizing the size and colloidal state of nanoparticles, Sci. Rep. 5 (2015) 9793. [40] C. Guiot, O. Spalla, Stabilization of TiO2 nanoparticles in complex medium through a pH adjustment protocol, Environ. Sci. Technol. 47 (2) (2013) 1057–1064. [41] P.-J. De Temmerman, et al., Quantitative characterization of agglomerates and aggregates of pyrogenic and precipitated amorphous silica nanomaterials by transmission electron microscopy, J. Nanobiotechnol. 10 (2012) 24. [42] P. Bihari, et al., Optimized dispersion of nanoparticles for biological in vitro and in vivo studies, Part. Fibre Toxicol. 5 (1) (2008) 14. [43] N.B. Hartmann, et al., Techniques and protocols for dispersing nanoparticle powders in aqueous media—is there a rationale for harmonization? J. Toxicol. Environ. Health, Part B 18 (6) (2015) 299–326. [44] I. Kaur, et al., Dispersion of nanomaterials in aqueous media: towards protocol optimization, J. Vis. Exp. (130) (2017). [45] E. Buhr, et al., Characterization of nanoparticles by scanning electron microscopy in transmission mode, Meas. Sci. Technol. 20 (8) (2009) 084025. [46] V.-D. Hodoroaba, et al., Performance of high-resolution SEM/EDX systems equipped with transmission mode (TSEM) for imaging and measurement of size and size distribution of spherical nanoparticles, Microsc. Microanal. 20 (2) (2014) 602–612. [47] J. Mast, L. Demeestere, Electron tomography of negatively stained complex viruses: application in their diagnosis, Diagn. Pathol. 4 (2009) 5. [48] A. Prasad, J. Lead, M. Baalousha, An electron microscopy based method for the detection and quantification of nanomaterial number concentration in environmentally relevant media, Sci. Total Environ. 537 (2015) 479–486. [49] E. Verleysen, et al., Quantitative characterization of aggregated and agglomerated titanium oxide nanomaterials by transmission electron microscopy, Powder Technol. 258 (2014) 180–188. [50] G.W. Hammond, et al., Improved detection of viruses by electron microscopy after direct ultracentrifuge preparation of specimens, J. Clin. Microbiol. 14 (2) (1981) 210–221. [51] T. Nomizu, A. Mizuike, Electron microscopy of submicron particles in natural waters—specimen preparation by centrifugation, Microchim. Acta 88 (1–2) (1986) 65–72. [52] D. Mavrocordatos, Non-artifacted specimen preparation for transmission electron microscopy of submicron soil particles, Commun. Soil Sci. Plant Anal. 26 (1995) 2593–2602. [53] J. Mielke, et al., Evaluation of electrospray as a sample preparation tool for electron microscopic investigations: toward quantitative evaluation of nanoparticles, Microsc. Microanal. 23 (1) (2017) 163–172. [54] SCENIHR, Scientific Committee on Emerging and Newly Identified Health Risks, 2009. [55] OECD, Preliminary Guidance Notes on Sample Preparation and Dosimetry for the Safety Testing of Manufactured Nanomaterials (ENV/JM/MONO(2010)25), Organisation for Economic Co-Operation and Development, Paris, 2010. [56] OECD, Guidance Manual for the Testing of Manufactured Nanomaterials: OECD Sponsorship Programme: First Revision, OECD, Paris, 2010. [57] W.C. Krumbein, L.L. Sloss, Stratigraphy and Sedimentation, W.H.F.A. Company, San Francisco, 1963, p. 660. [58] M. Munoz-Marmol, et al., Towards the taxonomic categorization and recognition of nanoparticle shapes, Nanomedicine 11 (2) (2015) 457–465. [59] J. Lopez-de-Uralde, et al., Automatic morphological categorisation of carbon black nano-aggregates, in: Proceedings of the 21st International Conference on Database and Expert Systems Applications: Part II, Springer-Verlag, Bilbao, Spain, 2010, , pp. 185–193. [60] J.M. Rodenburg, Understanding transmission electron microcsope alignment: a tutorial, Microsc. Anal. 18 (3) (2004) 3. [61] M. Haider, et al., Electron microscopy image enhanced, Nature 392 (1998) 768. [62] J.M. Rodenburg, E.B. Macak, Optimising the resolution of TEM/STEM with the electron ronchigram, Microsc. Anal. 90 (2002) 3.
47
48
Characterization of nanoparticles
[63] W.O. Saxton, D. Smith, S.J. Erasmus, Procedures for focusing, stigmating and alignment in high resolution electron microscopy, J. Microsc. 130 (2) (1983) 14. [64] ISO29301, Microbeam Analysis—Analytical Electron Microscopy—Methods for Calibrating Image Magnification by Using Reference Materials With Periodic Structures, 2017. [65] ISO 13322-1, Particle Size Analysis—Image Analysis Methods—Part 1: Static Image Analysis Methods, International Organization for Standardization, Geneva, 2014. [66] H.G. Merkus, Particle Size Measurements: Fundamentals, Practice, Quality, Springer, Pijnacker, 2009, p. 533. [67] P.-J. De Temmerman, et al., Semi-automatic size measurement of primary particles in aggregated nanomaterials by transmission electron microscopy, Powder Technol. 2061 (2014) 191–200. [68] K.S. Hougaard, et al., Effects of lung exposure to carbon nanotubes on female fertility and pregnancy. A study in mice, Reprod. Toxicol. 41 (2013) 86–97. [69] R. Fisker, et al., Estimation of nanoparticle size distributions by image analysis, J. Nanopart. Res. 2 (3) (2000) 267–277. [70] A.-L. Persson, Image analysis of shape and size of fine aggregates, Eng. Geol. 50 (1–2) (1998) 177–186. [71] M.N. Pons, et al., Particle morphology: from visualisation to measurement, Powder Technol. 103 (1) (1999) 44–57. [72] W. Meiniel, J.C. Olivo-Marin, E.D. Angelini, Denoising of microscopy images: a review of the stateof-the-art, and a new sparsity-based method, IEEE Trans. Image Process. 27 (8) (2018) 3842–3856. [73] M. Unser, Wavelets: on the virtues and applications of the mathematical microscope, J. Microsc. 255 (3) (2014) 123–127. [74] T. Wagner, https://imagej.net/ParticleSizer and https://zenodo.org/record/820296#.XAZ9XttKi00, 2016. [75] H. Masuda, K. Gotoh, Study on the sample size required for the estimation of mean particle diameter, Adv. Powder Technol. 10 (2) (1999) 159–173. [76] E. Verleysen, et al., TEM and SP-ICP-MS analysis of the release of silver nanoparticles from decoration of pastry, J. Agric. Food Chem. 63 (13) (2015) 3570–3578. [77] P.-J. De Temmerman, et al., Size measurement uncertainties of near-monodisperse, near-spherical nanoparticles using transmission electron microscopy and particle-tracking analysis, J. Nanopart. Res. 16 (10) (2014) 1–17. [78] P.-J. De Temmerman, et al., Measurement uncertainties of size, shape, and surface measurements using transmission electron microscopy of near-monodisperse, near-spherical nanoparticles, J. Nanopart. Res. 16 (1) (2013) 1–22.