Applied Clay Science 80–81 (2013) 358–365 Contents lists available at ScienceDirect Applied Clay Science j ourna l homepage: www.e lsev ie r .com/ locate /c lay Research paper Cryo-FIB–SEM and MIP study of porosity and pore size distribution of bentonite and kaolin at different moisture contents B. Lubelli a,b,⁎, D.A.M. de Winter c, J.A. Post c, R.P.J. van Hees a,b, M.R. Drury c a Faculty of Architecture, Delft University of Technology, Julianalaan 134, 2628BL Delft, The Netherlands b TNO, Netherlands Organization for Applied Scientific Research, P.O. Box 2600AA, Delft, The Netherlands c Department of Earth Sciences, Faculty of Geosciences, Utrecht University, P.O. Box 80021, 3508TA Utrecht, The Netherlands ⁎ Corresponding author at: Faculty of Architecture, D Julianalaan 134, 2628BL Delft, The Netherlands. Tel.: + E-mail address:
[email protected] (B. Lubelli). 0169-1317/$ – see front matter © 2013 Elsevier B.V. All http://dx.doi.org/10.1016/j.clay.2013.06.032 a b s t r a c t a r t i c l e i n f o Article history: Received 2 August 2011 Received in revised form 22 March 2013 Accepted 30 June 2013 Available online 20 July 2013 Keywords: Pore size distribution Mercury Intrusion Porosimetry Cryo-FIB–SEM Kaolin Bentonite Desalination poultices Clays often constitute the main component of poultices used for salt extraction from porous materials in con- servation intervention. Knowledge of the evolution in porosity and pore size of clay based poultices, due to shrinkage during drying, is of crucial importance for the selection of the most suitable poultice. We have studied the porosity and pore size distribution of kaolin and bentonite based poultices at different moisture contents. Both Mercury Intrusion Porosimetry (MIP) measurements on freeze-dried samples and cryo-FIB–SEM observations on wet samples are employed. The results show that these complementary techniques provide complete information on the porosity, pore size and pore structure of clay materials at different moisture contents. Both kaolin and bentonite poultices show a change of their total porosity and pore size distribution during drying: the changes are moderate in the case of kaolin, whereas the changes are very significant in the case of bentonite. These findings underline the necessity, when selecting a desalination poultice, of taking into account possible changes in its pore size distribution during drying, since these changes may affect the effectiveness of the salt extraction. Our results indicate that the good desalination efficiency of kaolin on substrate of pore size between 1 and 10 μm ob- served in practice is related to the presence in the poultice of pores that are very effective in capillary trans- port (0.2–2 μm) and to the relatively constant pore size distribution of the poultice during drying. © 2013 Elsevier B.V. All rights reserved. 1. Introduction Salt extraction by poulticing is a more and more widely used prac- tice in the field of conservation of cultural heritage (Lopez-Arce et al., 2009; Verges-Belmin and Siedel, 2005). Recent research has shown that the efficiency of drying poulticing can be improved by adapting the pore size of the poultice to that of the substrate on which it is applied (Lubelli and van Hees, 2010a; Pel et al., 2010; Sawdy et al., 2010): the poultice must have smaller pores than the substrate in order to enhance capillary transport. Clays constitute the main component of desalination poultices. In particular the clay type (kaolin, attapulgite, sepiolite, bentonite, etc.) plays an important role controlling not only the workability and the adhesion (Bourges and Verges-Belmin, 2011), but also the pore size of the poultices and thus their salt extraction effectiveness (Lubelli and van Hees, 2010a, 2010b). Due to shrinkage during the drying process, pore sizes of the poultice may change and hence the behavior of the poul- tice may vary. The importance of these changes depends on the clay type. Therefore, to understand the behavior of a clay based elft University of Technology, 31 15 2781004. rights reserved. desalination poultice during drying, it is of primary importance to conduct accurate measurements of the pore size distribution of the poultice at the initial stage of application and at the different stages of drying during the desalination process. This information is neces- sary to select the desalination poultice with the best suitable pore size throughout the process. Measuring the porosity and pore size distribution of clays at different water contents is a difficult task which requires the use of advanced in- vestigation techniques. For dry poultices, Mercury Intrusion Porosimetry (MIP) can be employed to measure the pore size distribution. For the case of wet poultices, samples are freeze-dried and subsequently MIP can be applied (Auras, 2008). However, when the cooling rate of the sample is not high enough, water crystallizes which may increase the water volume and in turn affect the measured pore size distribution (Desbois et al., 2008). Due to the required size for MIP, the cooling rate is insufficient to prevent crystallization. An alternative technique for the study of the pore size distribution of wet poultices is Nuclear Magnetic Resonance, which can be used to measure both the volume and the size of water filled pores in a wet poultice (empty pores are not measured) (Sawdy et al., 2010). However, both NMR and MIP provide only an indirect measure- ment of the pore size and give no complete information on the actual pore shape. http://dx.doi.org/10.1016/j.clay.2013.06.032 mailto:
[email protected] http://dx.doi.org/10.1016/j.clay.2013.06.032 http://www.sciencedirect.com/science/journal/01691317 http://crossmark.crossref.org/dialog/?doi=10.1016/j.clay.2013.06.032&domain=pdf 359B. Lubelli et al. / Applied Clay Science 80–81 (2013) 358–365 Porosity and pore size distribution of a wet poultice can also be studied by the use of a Focused Ion Beam–Scanning Electron Micro- scope (FIB–SEM) with a cold stage. With this technique a small sam- ple can be rapidly frozen and subsequently sectioned with the ion beam to reveal the microstructure. (Desbois et al., 2008; Hayles et al., 2007; Heymann et al., 2006; Holzer et al., 2010; Keller et al., 2011). The cryo-FIB–SEM approach has the major advantage to visu- alize the pores directly. Therefore detailed information is not only obtained on the total porosity and pore sizes, but also on the pore shapes, which is possible neither by MIP nor by NMR. Here we present a study of the effect of drying on the pore size of kaolin and bentonite poultices applying MIP on freeze-dried samples and by cryo-FIB–SEM observations of wet samples. 2. Experimental 2.1. Materials Two types of clay, kaolin and bentonite, were used: • Kaolin Polwhite C (by IMERYS), a medium particle size kaolin (0.5– 20 μm particle size) • Bentonite Ceratosil ® R (by SUD CHEMIE) a natural calcium benton- ite with 60–70% of montmorillonite. XRD analyses show the presence of quartz, albite (calcian) and muscovite. For each of the clay types, poultices were prepared by adding demineralized water to the clay. An amount of water sufficient to ob- tain a workable paste was used: • 0.5 weight water/weight dry clay for the kaolin poultice • 0.6 weight water/weight dry clay for the bentonite poultice. The workability of a desalination poultice gives a measure of the easiness with which the poultice can be prepared and spread out on the wall during application. The workability of a desalination poultice can be measured according to an adapted version of existing standard flow test procedure for fresh mortars (Bourges and Verges-Belmin, 2011; EN459-2:2001, 2002). Due to the limited workability of the clay paste, it was not possible to perform any flow tests on the kaolin and bentonite poultices to measure their workability; the water amount was determined to be sufficient to obtain a homogeneous paste which could be modeled without showing cracks. The relatively small difference in water content used for obtaining kaolin and bentonite poultices of comparable consistency can be at- tributed to the fact that the bentonite used in this research is rich in calcium: calcium-rich bentonite is known to absorb much less water than sodium-rich bentonite. The clay pastes to be studied at their initial water content were closed in parafilm immediately after preparation to prevent evapora- tion and allow hydration of the clay, until the moment of freeze dry- ing or observation by FIB–SEM. The clay samples to be studied after partial drying were obtained by applying the fresh clay paste on a brick substrate (to simulate the practice situation) in a 15 mm thick layer. The brick/clay assem- blage was stored at 20 °C 50% RH until the clay dried up to 50% of its initial moisture content (the time necessary to obtain this moisture content was determined in a separate, preliminary measurement). At this time, clay was sampled and freeze dried. In the period between sampling and freeze drying (maximal 24 h) the clay was sealed in parafilm. In order to limit the pore size change due to sample prepara- tion, freeze-drying was carried out according to the following procedure: samples (in a vial) were frozen by immersion in liquid nitrogen (−196 °C for about 3 min) and subsequently transferred into a vacuum desiccator. The desiccator was connected to the vac- uum pump and placed in the cryostat set at −20 °C. The evacua- tion is prolonged until the minimum pressure is reached after about 4–6 weeks. After this period the samples were ready for MIP measurements. The clay pastes to be studied after complete drying were applied on a brick substrate as previously described, stored for 2 weeks at 20 °C 50% RH, removed from the brick and further dried in an oven at 40 °C until the weight did not decrease any further. At this point small samples were taken for MIP measurements. In all cases attention was paid to include the whole layer thickness in the sample for MIP measurements, eliminating the variation which might originate from differences at the interface between the poultice surface and the brick due to water suction by the underlying brick. Poultices of only clay, without the addition of other components as sand or cellulose, are generally not used for practical conservation applications because of their high shrinkage. However, since the main aim of this study was to investigate the evolution of pore sizes be- tween clay particles, it was decided to prepare poultices of clay only. The results can be easily applied to the interstitial porosity be- tween clay particles present in poultices containing cellulose and/or sand (Lubelli and van Hees, 2010b). 2.2. Methods Two investigation techniques were used to study the porosity and pore size distribution (PSD) of the poultices and to measure which properties change due to drying: - Mercury Intrusion Porosimetry (MIP) - Cryo-FIB–SEM. MIP was used to study poultices at the initial moisture content, at 50% of the initial moisture content, (both prepared by freeze-drying) and at dry conditions. Cryo-FIB–SEM observations were performed on wet poultices at the initial moisture content. 2.2.1. Mercury Intrusion Porosimetry (MIP) measurements Mercury Intrusion Porosimeter (MIP) is an indirect method for the determination of the pore size distribution of porous materials, first proposed by Washburn (1921) and applied by Ritter and Drake (Drake and Ritter, 1945; Ritter and Drake, 1945) in 1945. Since then the method has become commonly used for the determination of pore size distributions of a large rangeofmaterials.MIP is a technique fre- quently applied in the field of conservation of buildingmaterials (among others Brakel et al., 1981; Mallidi, 1996; Rubner and Hoffmann, 2006). The technique is also used to investigate the effect of deterioration processes of frost and salt crystallization on the pore structure of the materials. The main advantages of this technique are the wide range of pore sizes that can be measured (from 0.003 to 360 μm for the instrument used in this case), the limited size of the sample required and the fact that it is a fairly rapid test (few hours) with a high reproducibility. The main limit of the method derives from the assumption that the pores are cylindrical, whereas real pore systems seldom consist of cylindrical pores. The method measures the pore entrance of the pore instead of the actual pore size. This means that in the presence of ink-bottle-like pores, MIP results do diverge from the real pore size distribution of the material (e.g. Hildebrand and Urai, 2003; Wardlaw and Mc Kellar, 1981). In addition, for some materials, the method of sampling, preparation and drying of the samples affects the results (e.g. Gallé, 2001). Porosity data presented in this paper have been obtained using Autopore IV9500 (Micromeritics). A contact angle of 141° was as- sumed between mercury and clay (Delage et al., 2006; Griffiths and Joshi, 1989; Hildebrand and Urai, 2003). An equilibration time of 30 s has been used between each pressure increase step and mea- surement of the intruded volume. The pressure used varied between 0.0037 MPa and 210 MPa, corresponding to pore diameter sizes of Fig. 1. Original image with indication of the contours. The accuracy of the obtained bi- nary images is shown by the good overlapping (insert). 360 B. Lubelli et al. / Applied Clay Science 80–81 (2013) 358–365 about 400 and 0.007 μm respectively. Measurements were carried out in twice on samples of about 1 cm3. The PSD of clay samples of centrimetric size was measured at dif- ferent stages of drying: • initial stage, i.e. moisture content of 0.5 and 0.6 by weight for kaolin and bentonite respectively (freeze dried) • 50% of the initial moisture content (freeze dried) • dry state (oven dried). At the end of the MIP measurements the samples showed no vis- ible damage (cracks, powdering etc), which would have occurred due to the pressure used to intrude the material. The absence of dam- age is confirmed by the absence of peaks at the highest pressures (smaller pores) in the MIP curves (see Sections 3.1 and 3.2). However, minor damage to the structure and compression of the clay sample at high pressures cannot be excluded Hildenbrand and Urai (2003). 2.2.2. Cryo-FIB–SEM observations Cryo-FIB–SEM observations were made with a Nova Nanolab 600 (FEI), equipped with a cryo-transfer chamber and cold stage (Quorum). FIB–SEM operations take place under vacuum conditions (typically 10−6–10−7 mbar), which requires hydrated samples to be either dehydrated or frozen. To freeze the samples, a small volume was deposited in between two rivets and mounted on a sledge. This construction was plunge frozen into liquid nitrogen and transferred under vacuum conditions to the transfer chamber (Desbois et al., 2008). In the preparation chamber the top rivet was broken off with the knife to reveal a clean freeze fracture surface. Subsequently the sample was sputter coated in situ for 60 s with Platinum–Palladium to ensure electrical conduction. Before the transfer into the SEM cham- ber was made, the temperature of the stage was raised to −150 °C to balance the temperature with the SEM stage. The temperature of the cold trap mounted around the pole piece of the electron column was maintained at−170 °C to take care of anywater vapor left in themicro- scope chamber. Before milling with the FIB, a thick (~1 μm) local layer of platinum- carbon was deposited (Hayles et al., 2007). The deposition smoothens the surface, which improves the quality of the final cross section, preventing vertical striations called curtaining. Then a cross section was milled at 30 kV acceleration voltage and a current of 20 nA. The final polish was made with 0.3 nA. SEM images of the cross sections were taken in Back Scattered Electron mode (BSE) (at 3.5–5 keV accel- eration voltage) and SE mode (at 2–7 keV acceleration voltage). In BSE mode contrast is generated by differences in atomic number. The heavi- er clay particles will be brighter with respect to the ice. An additional advantage of using BSE mode is less susceptibility to electrical charging of the sample. 2.2.3. Image processing Image processing is needed to extract quantitative information on pore volume and size of the pores from the acquired SEM images. In this case segmentation of the images (i.e. distinction of the clay frac- tion and water phase) was done manually using the computer pro- gram Photoshop by Adobe. By the use of the “Magic Wand” tool and thresholding of gray values, binary images were obtained, which were carefully compared by visual inspection with the original im- ages to check that the pore space was accurately segmented. This check was necessary because of the differences in gray intensity across different parts of the images and because of some curtaining left after milling. An example showing the accuracy of the obtained binary images is given in Fig. 1. From the obtained binary images the total porosity and pore size distribution have been quantified by the application developed by Munch and Holzer (2008) which is available at ftp://ftp.empa.ch/ pub/empa/outgoing/BeatsRamsch/PSDsoftware/. This application an- alyzes the pores using a discrete or a continuous PSD model. A continuous PSD model regards the pore structure as one single con- tinuous network whereas in the discrete PSD model pores are con- sidered as discrete objects. In our research we have calculated the PSD based on the continuous PSD model because of the following arguments: - The presence of very irregular and elongated pores: The assump- tion for the pores to be only cylindrical (as in the case of MIP) or spherical (as in the discrete PDS) is clearly unrealistic (e.g. Fig. 1). - For the present study we are mainly interested in the PSD in rela- tion to the study of salt solution transport through the clay poul- tices, process which occurs by differences in (capillary) pressures through the continuous pore network. A continuous PSD will better resemble the transport processes in a continuous pore network. In order to define the optimal radial step increase (dr) when calculating the pore size distribution with the PSD program, some preliminary tests have been carried out using: dr = pixel size (and thus varying from image to image, depending on the image magnifi- cation), dr = 10 nm and dr = 30 nm. The result from dr = 10 nm appeared to be the most appropriate (smaller dr resulted in several peaks in the incremental curve and larger dr in a too low precision). Therefore dr = 10 nm was used for the calculations of the total pore size and the pore size distribution of both kaolin and bentonite images (Section 3). The continuous PSD method was applied on the binary images taking into account the necessary correction of the pixel size in the y direction due to the observation angle in the cryo-SEM: the pixel size in y direction is equal to the horizontal pixel size multiplied by 1/cos 38° (≈1.27). For kaolin four SEM images were considered for image analysis: one at 17,500× magnification (K49) and three at 50,000× magnifica- tion (K50, K51, K52); in the case of bentonite three images were used (B47, B58 and B59), all at 20,000× magnification. 3. Results 3.1. Kaolin poultice The total open porosity and the pore size of kaolin clay poultices at different moisture contents as measured by means of MIP are report- ed in Fig. 2. The kaolin poultice has, at all moisture contents, a unimodal pore size distribution with a peak which varies slightly with the moisture content of the poultice. The pore size of the kaolin ftp://ftp.empa.ch/pub/empa/outgoing/BeatsRamsch/PSDsoftware/ ftp://ftp.empa.ch/pub/empa/outgoing/BeatsRamsch/PSDsoftware/ Fig. 2. Total porosity and pore size distribution of kaolin poultices at the initial wet state, at 50% of the initial moisture content and at the dry state as measured by MIP. Fig. 5. SEMmicrophotographs of a section of the kaolin sample (BSE mode): agglomer- ates of parallel oriented particles can be distinguished; these particle agglomerations are randomly oriented. The presence of a grainy structure is visible. 361B. Lubelli et al. / Applied Clay Science 80–81 (2013) 358–365 poultice changes slightly during drying: at the wet state, a main pore diameter of 0.4 μm is measured, with 2/3rd of the pore volume be- tween 0.2 and 2 μm. The presence of these pores, which are very effective in capillary transport, explains the good salt extraction effi- ciency observed for kaolin poultices on substrates with pores in the range of 1–10 μm (Lubelli and van Hees, 2010a; Sawdy et al., 2010). At 50% of the initial moisture content, the main pore diameter Fig. 3. Cryo-SEM microphotograph of kaolin platelets on the freeze fracture surface of a sample after coating with platinum (SE mode). Fig. 4. Different stages of the process of milling of a cross section in the kaolin sample (SE (le cross section, which complicates fully automated post-processing based on gray levels. measured is 0.2 μm with almost 90% of the pore volume between 0.1 and 0.5 μm. When the moisture content is reduced further, no ad- ditional changes in the pore size occur. The total porosity of the kaolin poultice decreases from 57.93 ± 0.02 vol.% to 44.5 ± 0.5 vol.% when the moisture content of the poul- tice is reduced to half of the initial moisture content. This is related to the shrinkage of the poultice. When the moisture content is further reduced no additional significant change occurs. Cryo-FIB–SEM observations were carried out only on samples of the kaolin poultice at the initial moisture content. Initially a freshly freeze fracture surface of the kaolin poultice was observed. In these images kaolin platelets of the size of a few micrometers can be distin- guished (Fig. 3). A cross section, necessary for the study of the porosity and pore size and shape, was milled (Fig. 4). On this section ice filled pores (dark gray) can be easily distinguished from kaolin (light gray). Kao- lin particles agglomerates to form packages of parallel platelets; the orientation of these agglomerations of platelets is random. The pores are mainly inter-aggregate pores between aggregates of clay particles with some interparticle pores inside the aggregate. Planar interlayer spaces between elementary layers, inside the clay particles can be measured neither by MIP nor observed by Cryo-FIB–SEM (Delage et al., 2006). A closer observation of the pores shows a phase separation (Fig. 5), with the presence of a grainy structure similar to the one re- ported in Desbois et al. (2008). This structure remains present after sublimation of the water in the SEM chamber (Fig. 6). ft) and BSE (right) mode). Despite planarizing the surface, some curtaining is left at the image of Fig.€2 image of Fig.€3 image of Fig.€4 image of Fig.€5 Fig. 6. SEM microphotographs of a section of the kaolin sample (SE mode) after freeze drying of the sample: the grainy structure in the pores can still be observed. 362 B. Lubelli et al. / Applied Clay Science 80–81 (2013) 358–365 A quantification of the total porosity and pore size distribution was carried out on the collected cryo-SEM microphotographs as de- scribed in Section 2.2.3. Images at different magnifications were used. Fig. 7 shows one original image of kaolin and the corresponding binary image. The incremental and cumulative pore diameter distribution obtained using the continuous PSD method on the same image is given in Fig. 8. The results obtained show the prevalence of pore radii in the range of 100 nm. Fig. 7. Original SEM microphotographs of a section of the kaolin sample (BSE mod Fig. 8. Incremental and cumulative pore area calculat Fig. 9 reports the total pore fraction area and mean pore radius obtained by calculation on images at different magnifications using the continuous PSD method. The total pore fraction varies between 26% and 44% and the mean pore radius between 110 nm and 160 nm. The variability measured is partially related to the very small imaged area in each of the cryo-SEM microphotographs. It is possible to observe that the pore size obtained by image anal- ysis is similar to that measured by MIP, whereas the total pore area is significantly smaller than the total pore volume measured by MIP. This might be due, next to the obvious differences between the two measurement methods, to the absence of pores diameter larger than 1.2 μm in the studied cryo-SEM images. If we consider only pore diameters smaller than 1.2 μm the difference between MIP and of the cryo-SEM results becomes much smaller. 3.2. Bentonite poultice The total porosity and pore size distribution of the bentonite poul- tice sample from MIP measurements are reported in Fig. 10. The MIP total porosity significantly decreaseswith drying. At the initialmoisture content a total porosity of 47 ± 2 vol.% ismeasured,which decreases to 33 ± 1 vol.% at 50% of the initial moisture content, and to 18.35 ± 0.4 vol.% at the dry state. An equally important change in the MIP pore size is observed: at the initial moisture content, the bentonite poultice has mainly pores in 1–30 μm range. The pore size decreases during drying: at 50% of the initial moisture content most of the pores e, 17,419× magnification) (left) and the corresponding binary image (right). ed from Fig. 7 using the continuous PSD method. image of Fig.€6 image of Fig.€7 image of Fig.€8 Fig. 9. Total pore fraction (% of area constituted by pores) and mean pore radius of the kaolin poultice resulting from image analysis using the continuous PSD method on SEM images at different magnifications (K49: 17,500×; K50, K51 and K52: 50,000× magnification). 363B. Lubelli et al. / Applied Clay Science 80–81 (2013) 358–365 are in the 0.2–2 μm range; the dry poultice showsmainly pores smaller than 0.05 μm. The reduction of the mean pore size and of the total po- rosity is the result of a considerable shrinkage of the bentonite poultice, which can be clearly observed at the macroscale. Fig. 10. Total porosity and pore size distribution of bentonite poultices at the initial wet state, at 50% of the initial moisture content and at the dry state measured by MIP, using freeze drying for the wet samples. Fig. 11. SEM microphotographs of a free fracture surface of a bentonite poultice sample (BS guished. The arrows indicate shrinkage cracks. Cryo-SEM observations of a freeze fracture surface of the benton- ite poultice sample at the initial moisture content show the presence of agglomerates of bentonite particles oriented parallel to each other. Thin shrinkage cracks are visible at the surface (Fig. 11). For the study of the porosity and pore size, a cross section was milled by the use of the FIB. Fig. 12 left shows an area of the obtained cross section: ice filling the pores (dark gray) and clay grains (light gray) can be easily distinguished. Differently than observed for the kaolin poultice, the particles in the bentonite poultice are not clearly discernible, but they appear as compact solid areas. The pores in the bentonite poultice form a well-connected network. The shape of the pores is elongated, their section is not cylindrical. Pore diameters larger than 5 μm, measured by MIP, could not be observed in the SEM images. Thin clay layers, (partially) intersecting the pores, are observed. A quantification of the total porosity and pore size distribution was carried out on the collected cryo-SEM microphotographs as de- scribed in Section 2.2.3. Fig. 13 reports the cumulative and incremental pore radius distri- bution as calculated from Fig. 12 right (about 20,000×magnification), obtained with the continuous PSD method. Fig. 14 reports the total pore fraction area and the mean pore radi- us obtained by calculations on a number of images using the contin- uous PSD method. The total pore fraction of the bentonite poultice at the initial mois- ture content measured on binary images varies between 45% and 50% of the area (measured on pictures at 20,000× magnification). The mean pore radius varies between 360 nm and 410 nm. The total pore area calculated by image analysis is comparable to the total pore volume measured by MIP; however, the main pore size obtained by the continuous PSD method (radius about 0.3– 0.4 μm) is much smaller than that measured by MIP (radius between 1 and 10 μm). Considering that the total porosity measured by both methods is similar, it is assumed that the thin layers of bentonite which visibly separate the pores in the cryo-SEM images (see Fig. 12 left), might have been broken during freeze-drying during the preparation for mercury intrusion, connecting adjacent pores and resulting into larger pores but similar total porosity compared to image analysis. 4. Discussion and conclusions The combination of techniques used in this research, MIP on freeze-dried samples and cryo-FIB–SEM was proven to be suitable E mode): agglomerations of bentonite particles with parallel orientation can be distin- image of Fig.€9 image of Fig.€10 image of Fig.€11 Fig. 12. Original SEM microphotograph of a section of the kaolin sample (BSE mode, 17,419× magnification) (left) and the corresponding binary image (right). In the left image on the left the arrows indicate some of the thin bentonite layers intersecting the pores. In the image on the right, clay and ice filled pores are given respectively as gray and black areas. Fig. 13. Incremental and cumulative pore area calculated from Fig. 12 using the continuous PSD method (dr = 10 nm). 364 B. Lubelli et al. / Applied Clay Science 80–81 (2013) 358–365 for obtaining new insights on the actual porosity and pore size of clay poultices at different moisture contents. The differences observed between the porosity and the pore size values obtained by the two methods are related to the dissimilarities Fig. 14. Total pore fraction (% of area constituted by pores) and mean pore radius of the bentonite poultice resulting from image analysis on different SEM images (all images at 20,000× magnification) using the continuous PSD method. between the direct evaluation of a plunge frozen sample (cryo-FIB– SEM images) and the indirect measurement of freeze dried samples (MIP) (Munch and Holzer, 2008). More specifically, limitations are also related to:- - The presence of very irregularly shaped and elongated pores, which invalidates the assumption of cylindrical pores in MIP measurements. - The presence of ink bottle shaped pores which lead to an overestimation of the small pore volume fraction when measured by MIP. - The limited size of the sample studied by cryo-FIB–SEM (few mi- crometer), which excludes larger inhomogeneities in structure, respect to the large size of the sample used for the MIP measure- ments, should be taken into account. Because of this reason a very large number of SEM images would be necessary to guaran- tee statistical significance to the image analysis. Beside differences inherent to the MIP and the image analysis methods, differences in porosity and pore size might also result from the sample preparation. The possible occurrence of damage dur- ing plunge freezing prior to MIP measurements might have altered the microstructure of the sample. The same risk would not be present in the case of sample preparation for the cryo-SEM, thanks to the smaller size of the sample. Holzer et al. (2010) found that plunge freezing followed by freeze drying of bentonite resulted in freezing image of Fig.€12 image of Fig.€13 image of Fig.€14 365B. Lubelli et al. / Applied Clay Science 80–81 (2013) 358–365 artifacts. They suggested that the growth of large ice crystals during plunge freezing formed large pores with sizes in the range of 5 to 20 μm. In fact, pores of such large size were not observed in the cryo-FIB–SEM observations on bentonite; our plunge frozen benton- ite samples show no eutectic microstructure and they look similar to those of the high pressure frozen samples of Holzer et al. (2010). In particular, the cryo-SEM images of bentonite show the presence of very thin bentonite layers (partially) intersecting the pores: the breaking of these intersection due to the growth of ice crystals during plunge freezing in the preparation of the samples for the MIP, togeth- er with eventual collapsing of the smaller pores, can explain the difference between the pore size of bentonite observed in the cryo-FIB–SEM images and the MIP measurement, in spite of the sim- ilar total porosity measured by both methods. In the kaolin sample the presence of two phases has been detected in the pores by cryo-SEM at high magnification. The presence of this grainy structure might be due to the presence of impurities in the fluid (0.15% soluble salt are present in the kaolin, according to the IMERYS information sheet). This structure remains in fact present after sublimation of the water in the SEM chamber. The MIP measure- ment on kaolin does not show a significant number of pores in the range of 50–100 nm, which is the size of the grainy microstructure observed in the cryo-SEM images. This excludes the possibility that the MIP measurements were significantly influenced by the grainy microstructure. It might be that the small grainy microstructure col- lapsed either during freeze-drying or under mercury pressure during intrusion, or that the grainy structure wasn't formed at all due to dif- ferent freezing conditions in the larger MIP samples. This research shows an innovative application of the cryo-FIB– SEM technique: the possibility of imaging the actual size and shape of fluid filled pores in clay materials. Despite the limits of comparingMIPmeasurements with cryo-FIB– SEM images of wet clays, the combination of these techniques has shown to provide complete information on the porosity, pore size and structure of hydrated clay materials. Three dimensional re- constructions of cryo-FIB–SEM images of subsequent sections (de Winter et al., 2009; Holzer et al., 2010) in combination with pore size distribution calculation as done in e.g. Munch and Holzer (2008) and Yang et al. (2009) might lead to a better comparison between MIP and cryo-SEM data. Regarding the consequences of the presented results for the choice of a suitable desalination poultice, it can be concluded that both kaolin and bentonite poultices modify their total porosity and pore size during drying. These changes are quite limited in the case of kaolin, whereas they are very significant in the case of bentonite. It is established that when selecting an adequate desalination poultice on the basis of its pore size distribution, for kaolin based poultices it can be considered sufficient to measure the pore size after drying of the poultice, whereas for bentonite based poultices measurements of the pore size at the initial wet state are necessary. Moreover, the research has confirmed that the better effectiveness of kaolin-based poultices with respect to bentonite-based poultices, as observed in practice, is related to the pore size of the kaolin. The presence of a large amount of pores of diameter 0.2–2 μm, which are very effective in capillary transport, and the relatively small changes in pore size of the poultice during drying explain the high salt extraction efficiency observed for kaolin poultices on substrates with pores in the range of 1–10 μm. Acknowledgments NWO, Utrecht University and FEI Company funded the FIB–SEM system in Utrecht. References Auras, A., 2008. Poultices and mortars for salt contaminated masonry and stone ob- jects. In: Ottosen, L.M., et al. (Ed.), Salt Weathering on Buildings and Stone Sculp- tures, Proceedings of the Int. Conf. Copenhagen, October 22–24. Technical University of Denmark, Copenhagen, pp. 197–217. Bourges, A., Verges-Belmin, V., 2011. Application of fresh mortar tests to poultices used for the desalination of historical masonry. Mater. Struct. 44 (7), 1233–1240. Brakel, J.V., Modry, S., Svata, M., 1981. Mercury porosimetry: state of the art. Powder Technol. 29, 1–12. De Winter, D.A.M., et al., 2009. Tomography of insulating biological and geological ma- terials using focused ion beam (FIB) sectioning and low-kV BSE imaging. J. Microsc. 233 (3), 372–383. Delage, P., Marcial, D., Cui, Y.J., Ruiz, X., 2006. Ageing effects in a compacted bentonite: a microstructure approach. Geotechnique 56 (5), 291–304. Desbois, G., Urai, J.L., Burkhardt, C., Drury, M.R., Hayles, M., Humbel, B., 2008. Cryogenic vitrification and 3D serial sectioning using high resolution cryo-FIB SEM technolo- gy for brine-filled grain boundaries in halite: first results. Geofluides 8 (1), 60–72. Drake, L.C., Ritter, H.L., 1945. Macropore-size distributions in some typical porous sub- stances. Ind. Eng. Chem. Anal. Ed. 17 (12), 787–791. EN459-2:2001, 2002. Building Lime. Test Methods. Gallé, C., 2001. Effect of drying on cement-based materials pore structure as identified by mercury intrusion porosimetry. A comparative study between oven-, vacuum-, and freeze-drying. Cement Concrete Res. 31, 1467–1477. Griffiths, F.J., Joshi, R.C., 1989. Change in pore size distribution due to consolidation of clays. Geotechnique 39 (1), 159–167. Hayles, M.F., Stokes, D.J., Phifer, D., Findlay, K.C., 2007. A technique for improved fo- cused ion beam milling of cryo-prepared life science specimens. J. Microsc. 226 (3), 263–269. Heymann, J.A.W., Hayles, M., Gestmann, I., Giannuzzi, L.A., Lich, B., Subramaniam, S., 2006. Site-specific 3D imaging of cells and tissues with a dual beam microscope. J. Struct. Biol. 155 (1), 63–73. Hildenbrand, A., Urai, J.L., 2003. Investigation of the morphology of pore space in mudstones — first results. Mar. Petrol. Geol. 20, 1185–1200. Holzer, L., Münch, B., Rizzi, M., Wepf, R., Marschall, P., Graule, T., 2010. 3D-microstructure analysis of hydrated bentonite with cryo-stabilized pore water. Appl. Clay Sci. 47, 330–342. Keller, L.M., Holzner, L., Wepf, R., Gasser, P., 2011. 3D geometry and topology of pore path- ways in Opalinus clay: implications for mass transport. Appl. Clay Sci. 52, 85–95. Lopez-Arce, P., Doehne, E., Greenshields, J., Benavente, D., Young, D., 2009. Treatment of rising damp and salt decay: the historic masonry buildings of Adelaide, South Australia. Mater. Struct. 42, 827–848. Lubelli, B., van Hees, R.P.J., 2010a. Desalination of valuable brick masonry by poulticing, laboratory and in-situ study. Restor. Build. Monum. 16 (1), 27–38. Lubelli, B., van Hees, R.P.J., 2010b. Desalination of masonry structures: fine tuning of pore size distribution of poultices to substrate properties. J. Cult. Herit. 11, 10–18. Mallidi, S.R., 1996. Application of mercury intrusion porosimetry on clay bricks to as- sess freeze–thaw durability — a bibliography with abstracts. Constr. Build. Mater. 10 (6), 461–465. Munch, B., Holzer, L., 2008. Contradicting geometrical concepts in pore size analysis attained with electron microscopy and mercury intrusion. J. Am. Ceram. Soc. 91 (12), 4059–4067. Pel, L., Sawdy, A., Voronina, V., 2010. Physical principles and efficiency of salt extraction by poulticing. J. Cult. Herit. 11, 59–67. Ritter, H.L., Drake, L.C., 1945. Pressure porosimeter and determination of complete macropore-size distributions. Ind. Eng. Chem. Anal. Ed. 17 (12), 782–786. Rubner, K., Hoffmann, D., 2006. Characterization of mineral building materials by mer- cury intrusion porosimetry. Part Syst. Charct. 23, 20–28. Sawdy, A., Lubelli, B., Voronina, V., Pel, L., 2010. Optimizing the extraction of soluble salts from porous materials by poultices. Stud. Conserv. 55, 26–40. Verges-Belmin, V., Siedel, H., 2005. Desalination of masonries and monumental sculp- tures by poulticing: a review. Restor. Build. Monum. 11 (6), 1–18. Wardlaw, N.C., Mc Kellar, M., 1981. Mercury porosimetry and the interpretation of pore geometry in sedimentary rocks and artificial models. Powder Technol. 29, 127–143. Washburn, E.W., 1921. Note on a method of determining the distribution of pore sizes in a porous material. Proc. Natl. Acad. Sci. U. S. A. 7, 115–116. Yang, Zhen, Peng, Xiao-Feng, Lee, Duu-Jong, Chen, Ming-Yuan, 2009. An image-based method for obtaining pore-size distribution of porous media. Environ. Sci. Technol. 43, 3248–3253. http://refhub.elsevier.com/S0169-1317(13)00202-0/rf0135 http://refhub.elsevier.com/S0169-1317(13)00202-0/rf0135 http://refhub.elsevier.com/S0169-1317(13)00202-0/rf0135 http://refhub.elsevier.com/S0169-1317(13)00202-0/rf0135 http://refhub.elsevier.com/S0169-1317(13)00202-0/rf0005 http://refhub.elsevier.com/S0169-1317(13)00202-0/rf0005 http://refhub.elsevier.com/S0169-1317(13)00202-0/rf0010 http://refhub.elsevier.com/S0169-1317(13)00202-0/rf0010 http://refhub.elsevier.com/S0169-1317(13)00202-0/rf0025 http://refhub.elsevier.com/S0169-1317(13)00202-0/rf0025 http://refhub.elsevier.com/S0169-1317(13)00202-0/rf0025 http://refhub.elsevier.com/S0169-1317(13)00202-0/rf0015 http://refhub.elsevier.com/S0169-1317(13)00202-0/rf0015 http://refhub.elsevier.com/S0169-1317(13)00202-0/rf0020 http://refhub.elsevier.com/S0169-1317(13)00202-0/rf0020 http://refhub.elsevier.com/S0169-1317(13)00202-0/rf0020 http://refhub.elsevier.com/S0169-1317(13)00202-0/rf0030 http://refhub.elsevier.com/S0169-1317(13)00202-0/rf0030 http://refhub.elsevier.com/S0169-1317(13)00202-0/rf0140 http://refhub.elsevier.com/S0169-1317(13)00202-0/rf0035 http://refhub.elsevier.com/S0169-1317(13)00202-0/rf0035 http://refhub.elsevier.com/S0169-1317(13)00202-0/rf0035 http://refhub.elsevier.com/S0169-1317(13)00202-0/rf0040 http://refhub.elsevier.com/S0169-1317(13)00202-0/rf0040 http://refhub.elsevier.com/S0169-1317(13)00202-0/rf0045 http://refhub.elsevier.com/S0169-1317(13)00202-0/rf0045 http://refhub.elsevier.com/S0169-1317(13)00202-0/rf0045 http://refhub.elsevier.com/S0169-1317(13)00202-0/rf0050 http://refhub.elsevier.com/S0169-1317(13)00202-0/rf0050 http://refhub.elsevier.com/S0169-1317(13)00202-0/rf0055 http://refhub.elsevier.com/S0169-1317(13)00202-0/rf0055 http://refhub.elsevier.com/S0169-1317(13)00202-0/rf0060 http://refhub.elsevier.com/S0169-1317(13)00202-0/rf0060 http://refhub.elsevier.com/S0169-1317(13)00202-0/rf0060 http://refhub.elsevier.com/S0169-1317(13)00202-0/rf0065 http://refhub.elsevier.com/S0169-1317(13)00202-0/rf0065 http://refhub.elsevier.com/S0169-1317(13)00202-0/rf0070 http://refhub.elsevier.com/S0169-1317(13)00202-0/rf0070 http://refhub.elsevier.com/S0169-1317(13)00202-0/rf0070 http://refhub.elsevier.com/S0169-1317(13)00202-0/rf0075 http://refhub.elsevier.com/S0169-1317(13)00202-0/rf0075 http://refhub.elsevier.com/S0169-1317(13)00202-0/rf0080 http://refhub.elsevier.com/S0169-1317(13)00202-0/rf0080 http://refhub.elsevier.com/S0169-1317(13)00202-0/rf0085 http://refhub.elsevier.com/S0169-1317(13)00202-0/rf0085 http://refhub.elsevier.com/S0169-1317(13)00202-0/rf0085 http://refhub.elsevier.com/S0169-1317(13)00202-0/rf0090 http://refhub.elsevier.com/S0169-1317(13)00202-0/rf0090 http://refhub.elsevier.com/S0169-1317(13)00202-0/rf0090 http://refhub.elsevier.com/S0169-1317(13)00202-0/rf0095 http://refhub.elsevier.com/S0169-1317(13)00202-0/rf0095 http://refhub.elsevier.com/S0169-1317(13)00202-0/rf0145 http://refhub.elsevier.com/S0169-1317(13)00202-0/rf0145 http://refhub.elsevier.com/S0169-1317(13)00202-0/rf0105 http://refhub.elsevier.com/S0169-1317(13)00202-0/rf0105 http://refhub.elsevier.com/S0169-1317(13)00202-0/rf0110 http://refhub.elsevier.com/S0169-1317(13)00202-0/rf0110 http://refhub.elsevier.com/S0169-1317(13)00202-0/rf0115 http://refhub.elsevier.com/S0169-1317(13)00202-0/rf0115 http://refhub.elsevier.com/S0169-1317(13)00202-0/rf0120 http://refhub.elsevier.com/S0169-1317(13)00202-0/rf0120 http://refhub.elsevier.com/S0169-1317(13)00202-0/rf0125 http://refhub.elsevier.com/S0169-1317(13)00202-0/rf0125 http://refhub.elsevier.com/S0169-1317(13)00202-0/rf0130 http://refhub.elsevier.com/S0169-1317(13)00202-0/rf0130 http://refhub.elsevier.com/S0169-1317(13)00202-0/rf0130 Cryo-„FIB–SEM and MIP study of porosity and pore size distribution of bentonite and kaolin at different moisture contents 1. Introduction 2. Experimental 2.1. Materials 2.2. Methods 2.2.1. Mercury Intrusion Porosimetry (MIP) measurements 2.2.2. Cryo-FIB–SEM observations 2.2.3. Image processing 3. Results 3.1. Kaolin poultice 3.2. Bentonite poultice 4. Discussion and conclusions Acknowledgments References