Sample Preparation Strategies for Characterisation of Materials

Sample Preparation Strategies for Characterisation of Materials


Dr James Wesley-Smith ()

National Centre for NanoStructured Materials, CSIR

Sample preparation is a critical aspect of any investigation. This is particularly true of microscopy, and the “make or break” stage which will either make available the informationrequired, or obscure it by distorting the native structure of the sample, and thus producing “artefactual” or “spurious” images.

The following is a working guide that aims to provide some insight into the rationale behind sample preparation approaches for various characterisation techniques. It is a “work in progress” that will be continually updated, so any feedback will be appreciated!

Transmission Electron Microscopy

Image formation and sample considerations

Image formation in the TEM relies on scattering of beam electrons by atoms in a sample. Atoms of relatively large molecular weights will be more electron-dense, and tend to scatter electrons by larger angles, and these are stopped by the microscope’s objective aperture. Because these electrons fail to reach the image plane, it leads to corresponding darkregions developing in the image. Scattered and un-scattered electrons collectively give rise to an image containing varying degrees of contrast (grey levels). Information from samples is seen as a 2-D projection image along the Z direction, therefore conventional TEM cannot allow you to determine if a feature is located at the top, middle or bottom of a sample as all this information with be imaged at the same time on a single plane.

Optimal results are obtained when samples do not exceed 100 nm thickness. In thicker samples, multiple collisions between illuminating electrons and sample atomsinevitably lead to heating damage (as energy is shared/transferred to the sample) as well as lowerimage quality because a) the information from too many stacked atomic ‘layers’ is seen superimposed, and b) because of these (inelastic) collisions emerging electrons have a large spread of electrons energies (chromatic aberration).

Another fundamental requirement for TEM (and also conventional SEM) is that samples must be completely dry before imaging. ‘Wet’ samples (either water or other solvents and oil) will degrade the high-vacuum environment inside the column, not only causing undesirable contamination of the microscope, but also leading to artefacts in the sample (contamination and changes in sample structure).

Sample Preparation TEM

TEM samples are supported by a 3 mm grid (typically made from Cu but Ni, Au, and even nylon are available for special applications) of various mesh sizes. Thin sections of material (see below) are sufficiently robust to spread out over the grid without additional support. However, other samples usually need a thin electron-transparent film (typically carbon) to be cast over the grid so as to support the particles during imaging.

Two main categories of TEM samples may be distinguished:

Powder samples/ solutions / carbon nanotubes (CNTs): These must be dispersed as thinly as possible (mono-dispersed being ideal) on the carbon film coating TEM grids. An extremely small amount of material is suspended in a solvent (just enough to obtain a slightly turbid solution). We routinely use ethanol, which dries relatively quickly. Water is also possible, but these should ideally be prepared in advance and allowed to dry overnight. We do not accommodate samples suspended in toluene into the TEM as this solvent invariably introduces excessive contamination to the column and low image quality. Brief (15-20 s) ultrasonication is sometimes used to disperse particles, after which carbon-coated grids are briefly immersed into the suspension and allowed to dry thoroughly before viewing. Just for the record, each carbon coated TEM grid is about R100!

Bulk / solid materialsneed to be sectioned to <100 nm, as mentioned above. Depending on the hardness of the material, two approaches are possible:

Soft materialsdo not have the necessary rigidity to allow sectioning to the ideal 70 nm using a microtome, and for this reason they need to be either:

Paper, biological samplesand some polymers: samples not exceeding 1-2 mm3 must be previously embedded in epoxy resin so tey can can be sectioned to 70-100 nm at room temperature using an ultramicrotome. Sections are place on regular TEM Cu grids and are ready for TEM imaging (unless staining is required)

Polymers.Most polymers are too soft to withstand sectioning at room temperature without distorting, even if supported by a surrounding layer of epoxy resin. For this reason polymers need to be hardened by cooling to sub-zero temperatures (anywhere between -40 to -150°C) and cryo-sectioned at that temperature. Sections are placed on regular TEM Cu grids for support during viewing.

Please note: Ultramicrotomyis a highly specialised technique, requiring sophisticated instrumentation and dedicated, skilled operators (i.e., not a technique that can be taught in one or two sessions!). It is both demanding and time consuming, requiring about 1 h per sample.

Hard materials need to be thinned through other methods, such as ion beam milling. At the NCNSM, this is achieved using a Focused Ion Beam (FIB) in our Zeiss AurigaSEM, where Ga ions are used to mill a thin (100 nm) lamella, which is then spot-welded onto a TEM grid by means of a micromanipulator, before being transferred to the TEM.

Please note: This is also a highly specialised technique, requiring sophisticated instrumentation and dedicated, skilled operators (i.e., not a technique that can be taught in one or two sessions). Depending on the properties of the material in question, it may take several hours (up to 8) to prepare a single lamella

Staining for additional contrast

Samples with low atomic weights (e.g. polymers, biological materials) often need to be stained positively(and selectively) with electron-dense stains (e.g. osmium tetroxide, ruthenium tetroxide, uranyl acetate, lead citrate) to generate sample-specific contrast. It is also possible to stain the background and not the sample itself, thus creating an image in reverse contrast (sample bright, background dark). This is known as negative staining(a non-selective method).

Scanning Electron Microscopy

Image Formation

In contrast to the TEM, a SEM is used to generate information from the surface of ‘bulk’ samples (i.e. not restricted to <100 nm) as the electron beam is scanned repeatedly in a raster pattern over a given area. Signals generated from a sample include secondary electrons (SE), backscattered electrons (BSE) which are useful to obtain topographical and atomic weight contrast from a sample (respectively), and X-rays which can be used to obtain elemental information from a sample using Energy Dispersive X-rays Spectroscopy (EDS). Image formation relies on the relative “visibility” of the SE and BSE signals generated from the sample by the detectors, with electrons in the line of sight of the detector giving rise to bright areas in the image, and vice versa. In this manner, SEM reveals surface information from samples, and this is especially true when using low (<2kV) accelerating voltages.

Sample Preparation

If the surface of materials needs to be imaged ‘as is’, then theyare mounted securely on a SEM stub by means of conductive carbon tape before transfer to the interior of the SEM.As with TEM, samples must be completely dry to avoid outgassing inside the chamber,which would lead to undesirable deterioration of the high vacuum as well as contamination. Also, samples should be kept to the smallest size practically manageable: large samples require longer pump-down before high vacuum is achieved (especially if porous), carry a bigger load of potential contaminants into the instrument, and also are more prone to charging if non-conducting (see below).

However, if internal information of a bulk sample is to be imaged (e.g. distribution of clays / pores or inclusions in a matrix, or to establish if core-shell structures exist among particles) some additional sample preparation prior to viewing will be necessary (see below).


Because SEM samples are relatively bulky, electrons can build-up within them asthey are scanned by the beam. Metallic samples are naturally conductive, so beam electrons entering the sample will areeasily grounded. In contrast, non-conductive samples (e.g. polymers, biological materials) or those making poor contact with the carbon tape (e.g. long fibres, ‘stacks’ of powder particles or CNTs) or will tend to accumulate beam electrons. These samples rapidly become negatively-charged, and this “charging” artefactis evident by excessively bright areas of the sample, especially along edges and raised pointed. The sudden flashes when excess electrons are suddenly released interferes with image formation by creating “charge lines” across the images obtained. It is for this reason that poorly or non-conductive samples are coated before viewing with a 4 nm layer of a conductive material (typically Au/Pd or carbon, but Pt, Ir and W are also possible) that will prevent charging by grounding beam electrons. There is an additional advantage in that the signal coming from a non-metallic sample (low Z) will improve after coating with gold, as this layer will increase the number of electrons generated at each point on the surface and thus the signal coming from the sample. However, coating can potentially hide the finer surface details under a “carpet” of gold or carbon coating, and one should attempt viewing samples uncoated if the information required is below 10 nm range. However, for most applications, coating is the routine approach taken.

Gold vs.carbon

For ordinary SEM imaging, the preferred approach is to sputter gold onto a sample. Again, this method relies on forming a conductive layer all over the sample, continuous with the carbon tape/ metallic stub. This can be challenging in particulate or fibrous samples unless they are in close contact with the carbon tape: more than one coat may be applied, but at the expense of laying a thicker “carpet” over the surface and without the guarantee that there will be sufficient conductivity from the top, loosely bound fibre / particles! This is oneinstance where “less is more”: a very sparsely distributed sample will be more likely to make good contact with the carbon tape and thus give best results than one simply piled-on.

Carbon is evaporated onto the surface of samples where analysis by Energy Dispersive Spectroscopy (EDS) is being considered. As a relatively low atomic weight element, carbon will not mask X-rays from heavier elements as gold will. Charging is minimised, and therefore SEM imaging is also possible, albeit with a lower signal being generated.

Accessing sub-surface (internal) information from a bulk sample

Fracturing under liquid Nitrogen (LN2).Quick and easy to do, but results often depend on the material and the skill of the user! To do this we place sample under LN2 and upon cooling we fracture it using a blade or scalpel and hammer. The fracture follows the path of least resistance, and often reveals the boundaries between two phases (e.g. two polymers) or weaknesses in a sample. In porous or heterogeneous samples the plane of fracture is often random and giving a somewhat jaggered appearance.

Atomic Force Microscopy (AFM)

Image formation

AFM relies on the deflection of a laser beam focused on the back of a cantilever bearing a finely-pointed SiN probe on its underside, as it is scanned in a raster pattern over a sample. A piezo scanner takes care of the accuracy of the movement in x, y and z directions during scanning. The probe can either be held in direct contact with a sample, or made to oscillate at a certain frequency over it in tapping mode. During the course of scanning the sample, the SiN probe is subject to deflection with changes in sample topography in contact mode, or changes in the amplitude of the oscillation as it experiences different attractive / repulsive forces from the sample relative to a set point. The laser beamreflecting on the back of the cantilever will be affected accordingly,and strike different quadrants of a detector. From this a number of signals can be extracted simultaneously on different channels. The most commonly used signals include height information (i.e. topography), surface roughness, amplitude and phase deviation of the tapping oscillation relative to a reference baseline and which can provide information of physical properties such as surface hardness, adhesion, viscosity.

The lateral resolution of the AFM technique is similar to that of a Field Emission Gun SEM (FEGSEM; c. 1 nm), which would possible be preferable to investigate large areas of a sample. However, AFM is ideally suited to explore variation in sample surface detail along the Zplane. Perhaps more importantly, AFM can provide information about the distribution (mapping) of physical properties of a sample (hardness, adhesion, roughness, magnetism) and even chemical information where functionalised tips are used to explore affinities with the sample surface.

Our Veeco Multimode AFM has a heating stage ( xto y °C) that allows melting / recrystallization studies of polymers to be conducted while imaging

There are limitations to the usefulness of AFM:

  1. It is a time-consuming technique: each scan can take (on average) about 5 minutes, and finding a suitable area can be difficult in a heterogeneous sample. It is not unrealistic to allocate 1 hour per sample.
  2. The area scanned is extremely small (≤ 100 µm2) raising questions about sampling frequency to capture representative images of a sample
  3. It is not suitable for ‘rough’ sample surfaces (steps > 8 µm in height) and only samples smaller than 1 cm diameter can be accommodated on our Multimode AFM.
  4. Samples must be firmly fixed to an atomically-smooth substrate: any movement will lead to drift artefacts in images.
  5. Probe geometry, and impurities/dirt on the probe itself can lead to artefacts in the image
  6. Interpretation of results can be challenging at times.

X-Ray Diffraction (XRD)

XRD is used to identify the phases present in crystalline materials. Each crystalline substance has its own ‘fingerprint’ pattern (spectrum) that allows its chemical characterisation, the relative amount of each phase in a composite sample, and whether or not there are preferred orientations of crystals.

The PanAlyticalX’pertPRO PW3040/60 X-ray diffractometerat the NCNSM has a Cu X-ray source Kα (=0.154 nm) monochromated radiation source. We characteristically run the XRD between 2 and 90 2θ angles, and can accommodate up to 15 samples at a time in each run.

Sample dimensions

Holders are discs with a cup-like hole machined in the centre. The diameter of the hole is 40 mm and the height 4 mm. Both finely ground and solid samples may be analysed, but theymust fit within these dimensions, althoughthey may be smaller. The surface to be analysed must be flat, as a rough sample will give spurious results. We do not have cutting / grinding facilities, so solid samples need to be cut to the right dimensions and polished before brought for analysis.

Samples must be delivered to the Centre at least the day before the analysis is booked to take place to ease the loading stage.