Progress in building materials analysis (Part 2)

In the following some examples are described relating to the manifold microscopic and analytical capabilities of the FEI Nova NanoSEM in combination with high-pressure cryo-preparation techniques. These methods are especially useful to answer fundamental questions on cement chemistry.

The C-S-H phases formed during the hydration process at ordinary temperatures and for realistic water/cement ratios of C₂S, C₃S and OPC always appear in the same needle-like morphology when ESEM-FEG and NanoSEM imaging techniques are used (Fig. 15 a). By means of these techniques uncoated samples in their native hydration state can be imaged. This repeating habit gives some indication that there must be a certain long-range order principle. Furthermore, we found that C-S-H phases with the same habit show an oriented and an accelerated growth on calcite crystal surfaces caused by heterogeneous nucleation. In this case the calcite surface acts as a nucleation site. This is a further indication for the crystallinity of the above-mentioned C-S-H phases [7].

By means of XRD investigations these C-S-H phases show, however, that even after a very long hydration time (completely hydrated C₃S due to repeated milling) only rudimentary peaks with strong reflex broadening occur. For these C-S-H phases the term “X-ray amorphous” is used. It is important to note that the name “amorphous” only makes sense when the meas-uring method is indicated at the same time.

The wavelength of the electrons is around two powers of ten smaller when using electron diffraction in the TEM. The de Broglie wavelength in a 200 kV TEM is about 0.027 Å (1 Å equals 0.1 nm) whereas the wavelength of Cu-Ka radiation usually used in XRD measurements is 1.54 Å. Furthermore, due to the strong interaction of the electrons with the potential field of the atomic nuclei the intensity of diffraction is 10⁸ greater than in the case of X-rays. For this reason fine-crystalline substances can still be identified by means of electron diffraction whereas the XRD diffraction method fails. However, in the past, electron diffraction experiments on C-S-H phases as a hydration product of OPC or C₃S only show amorphous TEM diffraction patterns [8–9].

Comparative ESEM studies have shown that particularly under high-vacuum conditions this type of C-S-H phases is very ­sensitive when exposed to electron radiation at relatively high-charge densities (Fig. 3 and Fig. 15 b). From these investigations it was deduced that through the electron diffraction method using a too high charge density the crystallinity of the sample is destroyed, i.  e. “amorphization“ of the C-S-H phases takes place. The same phenomenon that can be seen in Figure 15 b occurs in the TEM, namely the typical formation of bubbles in the inner structure of the C-S-H phases.

Electron diffraction experiments show that through the irradiation of C-S-H phases when deploying the usual charge densities of more than 2 x 10^-1 C/cm² (charge density 1.5 x 10² ­e/Ų) at ambient temperature the characteristic morphology and the lattice structure of the nanocrystals deteriorate within seconds.

By means of electron diffraction pattern experiments in the cryo-TEM the crystalline nature of the needle-like C-S-H phases could be proved beyond a doubt [10]. When the sample is cooled down by means of liquid nitrogen (approx. –175 °C in the cryo-TEM), the stability of the C-S-H phases is significantly improved. A charge density of up to 2.4 x 10³ e/Ų (i.  e. one power more) can be used which is not possible with the uncooled sample. The cryo-TEM investigations made it possible for the first time to gain diffraction patterns of individual fibrous C-S-H phases proving their crystallinity. The electron diffraction images show lattice points and the diffraction patterns of continuous lines. These continuous lines which could also be observed on individual tobermorite needles are due to stacking faults within the crystal structure [11]

ESEM investigations on potassium sulfate rich OPC (i.  e. cement with a high portion of arcanite) in the early hydration stage show a temporary syngenite formation in the microstructure. The syngenite crystals longer than 10 µm have a prismatic habit with a crystal thickness of approximately 50 nm. In the past the XRD Rietveld analysis was not so well ­developed. That is why syngenite could not be determined since it is ­often near the detection limit of the method. Therefore it has been ­discussed that the formation of syngenite during the ­early ­
hydration process could be an artifact caused by the local ­super-saturation of the highly concentrated pore solution in the ESEM investigations.

By using the cryo-immobilization preparation method in the high-pressure freezer it becomes possible to freeze the samples without the formation of noticeable ice crystals (vitrification: cooling rates of up to 10 000 K/s can be achieved). The ­formation of amorphous and/or crystalline precipitates (e. g. syngenite), which can appear in the ESEM-WET mode as an artifact, does not occur using this preparation method.

High-pressure (2100 bar) frozen and freeze fractured samples allow the direct view into the microstructure. A comparison between cryo-SEM and ESEM images shows that the same long ­prismatic crystals also occur when cryo preparation is used. Those crystals can be identified as syngenite by means of EDS [12].

The cryo-SEM visualization of the double salt syngenite can therefore be taken as a direct proof for the actual occurrence of this phase. Figure 16 a shows the freeze fracture surface of a potassium sulfate rich OPC after a hydration time of 90 minutes. The following phases in the deep-frozen microstructure (–130 °C) are clearly visible: amorphous ice, ettringite (short prismatic habit) and syngenite (long prismatic habit). Even at low magnification, the random distribution of syngenite crystals which was observed in the ESEM could be confirmed by means of cryo-SEM.

In contrast the microstructure of an OPC with a low content of water soluble potassium sulfate shows no syngenite formation even after a longer hydration time of 40 min (Fig. 16 b). No syngenite is detectable during the whole hydration process. Only short prismatic ettringite crystals and individual foil-like C-A-H phases are visible.

By means of cryo-SEM investigations it was revealed that syngenite is formed during the early hydration of potassium sulfate rich OPC. Thus, it can be ruled out that syngenite observed by means of ESEM is an artifact. Furthermore, temporary syngenite formation was detected by XRD Rietveld investigations and computation of saturation index in the aqueous phase of cement paste. In addition it was shown that already minor contents of syngenite decrease the fluidity of cement pastes and concretes significantly [13].

A new possibility to image non-conductive samples without coating under high-vacuum conditions at very low landing energies of the primary electron beam is the use of the beam deceleration mode in combination with a very sensitive low-voltage, high-contrast detector (vCD) in the NanoSEM. A negative voltage applied to the stage reduces the landing voltage of the electron beam, acting like an additional electrostatic lens on top of the magnetic immersion field. This technique enhances image contrast by accelerating secondary (SE) and backscattered electrons (BSE) back into the detectors. The vCD is a solid state detector perfect for low accelerating voltage applications as low as 200 V. The diode has an active area of approx. 100 mm² and offers BSE atomic number contrast and maximum detection efficiency. The BSE have much higher energies in comparison to SE (max. 50 eV), so the image is less affected by unpredictable specimen charging. Therefore the contrast information is easier to understand and sometimes object details can be imaged more clearly.

Figure 17 shows a comparison between different imaging possibilities of uncoated UHPC. On the left the Helix detector image (mainly SE) in LV mode at a V₀ of 4 kV and on the right the vCD BSE-image in high vacuum with a landing energy of the primary electrons of 1.5 keV. The UHPC sample was heat treated in a regime of 90 °C maximum temperature for 48 h and afterwards it was stored under standard conditions of 20 °C and 65  % relative humidity. The detection of the pozzolanic reaction of spherical shaped silica fume particles in the UHPC microstructure was of special interest in these investigations. Both the above- mentioned imaging techniques are capable of showing the dissolution structures as well as the reaction products of the pozzolanic reaction but a more detailed image is achieved by means of the beam deceleration method in combination with the vCD.

Another technology for charge control of non-conductive samples without coating in a high- vacuum environment is the use of the low voltage SEM mode by means of the through-the-lens detector (TLD). In this imaging mode the pozzolanic reaction of silica fume can be observed in great detail.

In Figure 18 one can see a direct comparison of a progressing pozzolanic reaction: A silica fume particle was half embedded in a portlandite aggregate during the early hydration process. Through this encapsulation the pozzolanic reaction cannot take place since the third reaction partner (water) is missing. Thus, the silica fume particle still shows its original even surface structure. However, the part which is not embedded shows clear signs of dissolution. Consequently, cavities are formed which in this case are smaller than 10 nm.

It could be shown that a device consisting of an EDS-EBSD system connected to an SEM is the best combination. This configuration allows the simultaneous collection of chemical and crystallographic data. For a comprehensive analysis of certain samples it is important not only to have morphological and chemical information but also to carry out a phase identification in the same sample area.

The traditional phase identification method outside the micro­scope by means of XRD (Debye-Scherrer method) has the ­disadvantage that the localization gets lost. The integral meas­urement causes low limits of detectability. Furthermore, the danger arises that the samples are changed by the necessary preparation procedures such as milling, stopping of the hydration process, drying or through the carbonatization during the preparation process.

Furthermore, with the EBSD-system it is possible under low­vacuum conditions to get diffraction patterns rich in contrast using a pressure of up to 0.7 mbar (0.5 Torr). This brings the following advantages compared to investigations in the highvacuum mode:

­– It becomes possible to obtain diffraction patterns of massive samples in an environment that suppresses charge build-up on non-conductive material. That means that no signal interference by the sample coating occurs.

­– The samples do not need to be high-vacuum stable in the low-vacuum mode but they can be analyzed in “close to native” condition. The destruction of the crystalline structure (of water containing minerals) through the high vacuum can thus be avoided (such as ettringite).

­– The surface contamination which has also negative influence on the pattern quality is strongly reduced.

The use of an EBSD-system is especially helpful when a sample has the same chemical composition but exits in various crystal modifications. This is not only the case with frequently occurring SiO₂ but also with CaCO₃ which are formed during the carbonatization processes of lime mortars on the surface of concrete or in the microstructure of damaged concrete. The thermodynamic stable form of calcium carbonate is calcite which crystallizes in the trigonal crystal system. In contrast, the rhombic modification aragonite and the hexagonal vaterite are metastabile CaCO₃ modifications.

Calcite exists in very many variations in form and shape so that it is difficult to tell which CaCO₃-modification we are seeing. Only by means of combining information concerning morphology, chemistry and crystallography does it become possible to get an unambiguous answer. An unambiguous differentiation between the calciumcarbonate modifications aragonite and calcite (cf. Fig. 10) can be made by combining the measured diffraction patterns and the Kikuchi bands calculated from the atom data which were used for the indexation [7].

A further wide field of application for the EBSD system is the differentiation of phases having a similar chemical composition. Such phases cannot be identified by means of a microanalysis using EDS alone. One example is the differentiation of water-free calcium sulfate (anhydrite), semihydrate (bassanite) and gypsum with two molecules of water. These phases only differ as to their chemical composition in the crystal water content. In comparison to anhydrite, bassanite has an additional water content of 6.21 wt.-%. In order to identify such phases unambiguously it is necessary to index the crystal lattice.

Another application field for the EBSD method is the identification of nano-sized phases and small particles which cannot be analyzed exactly by means of EDS, because under ­typical microscope operating conditions the interaction volume in cement based material analyzed by the electron probe has micro­meter scale (and thus the interaction volume is too big). Since the signal of an EBSD method comes from a depth of approx. 100 atomic positions it is possible to identify phases having a size of approximately 10 nm. The EBSD method can also be used in case of line overlapping caused by the insufficient energy resolution of the EDS method.However, all tests have failed so far to gain diffraction patterns by means of EBSD of the needle-like C-S-H phases formed during the hydration of C₃S or OPC in such a quality which would allow an identification of the crystal structure. Only by means of electron diffraction pattern experiments in the cryo-TEM could it be proved that the phases which were thought before to be “X-ray-amorphous” in fact have a crystal lattice.

When information on the crystal orientation and the phase distribution are needed, a so-called EBSD-orientation imaging microscopy (OIM) can be carried out. In doing so, one can make statements on the grain size, the crystal orientation, the texture, the grain boundary type and the residual plastic strain. In conjunction with the simultaneously acquired chemical data from EDS, phase identification can also be performed. So-called strained quartz is shown here as example of an OIM application. This is one of the first attempts to use this method for the characterization of aggregates. It is possible for example to determine the sub-grain crystal size of strained quartz which is a significant factor for the solubility of the aggregate and thus it determines the risk of damage to the concrete by the mechanism of alkali silica reaction (ASR). The grain size distribution gained by the OIM (Fig. 19) shows that approx. 20  % of the strained quartz have a grain size smaller than 10 µm (however, the analysis was not extended to the nanometer range). These subgrains are not even distributed but they occur more frequently on the grain boundaries of large quartz crystals and thus form the solubility path from the surface into the grain. By means of crystal orientation mappings it becomes possible to find out how vulnerable aggregates in the concrete are to ASR.

A considerable improvement of the amount of information available in the field of building material(s) analysis of cementitious binders could be achieved by combining high- pressure freezing cryo-preparation, ultra-high resolution electron-optical and analytical techniques.

By optimizing the system Peltier cooling stage/sample holder high resolution imaging to characterize “wet” samples in their “close to native” state is now possible in the ESEM-WET mode at high accelerating voltages.

The deployment of the Nova NanoSEM 230 with a Helix detector permits a detailed and high-contrast imaging of very dense microstructures e. g. ultra-high performance concrete (UHPC). This microscope enables high-resolution imaging capabilities in a low-vacuum water vapor atmosphere at low acceleration voltages useful for charging and water containing material. A further new capability of the Nova NanoSEM is the beam deceleration mode. In combination with a very sensitive low-voltage high contrast BSE detector it becomes possible to image uncoated non-conductive samples under high-vacuum conditions. Another way of secondary electron imaging of such samples is the LV-SEM mode in a high-vacuum environment. All these methods offer the possibility to image the hydration progress in very dense microstructures (such as UHPC) in great detail but microanalytical measurements on native samples are only applicable by using a low-vacuum environment (Helix detector) or cryo-SEM investigations.

By means of cryo-SEM investigations it could be clearly proved that the temporary syngenite formation during the early hydration of OPC with a high content of potassium sulfates is not an ESEM artifact.

From the ESEM investigations it can be deduced that the C-S-H phases occurring during the hydration process of OPC or C₃S in the outer hydration rim always show the same needle-like habit. These phases are very sensitive to electron radiation. Thus, the lattice structure, as well as the morphology, is destroyed when the charge density is too high. That means that phase “amorphisation” could takes place through the measuring process itself. Only by means of electron diffraction pattern experiments in the cryo-TEM at very low electron doses could the crystalline nature of these individual fibrous C-S-H phases be proved without doubt[10].

An analytical EDS – EBSD device connected to the NanoSEM allows the building material characterization by the simulta­neous collection of images, chemical as well as crystallographic data. An unambiguous phase identification between the cal­cium carbonate modifications vaterite, aragonite and calcite can be carried out.

Furthermore, the OIM analysis can be applied for the subgrain crystal size determination in the aggregate microstructure (e.  g. strained quartz). By means of OIM it becomes possible to evaluate the inclination of the aggregates in the concrete as regards ASR.