3.1 Introduction

Soil sampling is the first activity of most research performed in soil science. It is self-evident that badly designed soil sampling (e.g., location, size, soil structure disturbance) or subsampling can lead to erroneous results spoiling any subsequent analytical techniques. This is even more crucial considering the fragility of the soil pores network, which is a key property on which soil microtomography focuses. Thus, both soil sampling and soil preparation have to be considered key fundamental activities preliminary for any study in soil science.

This chapter is divided into four sections which provide an overview of: (a) the main steps to be followed for optimum sampling, (b) the definition of representative elementary volumes, (c) criteria for geospatial sampling, and (d) preparation of undisturbed samples before X-ray CT analysis.

Sampling is a critical part of all micromorphological studies and, thus, also of soil microtomography. Any mistake made in the sampling phase could affect subsequent interpretations (i.e., what is seen, processed, or analysed in the 3-D domain), so causing potentially erroneous conclusions about the soil material being studied. It is important that where possible the collected samples represent the soil material as it exists either in the field or in a laboratory (lab) experiment.

For lab experiments, sampling may generally involve the entire soil volume employed for a specific trial (e.g., small soil cylinders subjected to a specific lab treatment such as measurement of hydraulic conductivity or assessment of wetting and drying cycles), thus collecting representative soil material may not be an issue. This is the case when subsampling of a large soil column (e.g., for a given specific laboratory assessment, such as infiltration) and then a subsample is collected from this. This subsampling involves addressing the sample size issue (see section below) and the avoidance of sampling artefacts as they could occur at the outer margin of the sample.

In the case of field sampling (such as from experimental field trials), collecting representative samples is much more difficult than in laboratory experiments because soils in the field have an intrinsic heterogeneity and spatial variability.

Thus, special precautions must be taken to ensure collection of representative materials. The following questions must be carefully addressed before sample collection is undertaken: (i) What is the aim of the sampling? (ii) where to sample? (iii) what size dimensions should the sample have? (iv) what orientation of samples is needed? (v) how many samples are needed? (vi) when is the best time to sample? (vii) how to sample? and (viii) how to document, transport, and store samples?

3.2 The Main Steps for Successful Sampling

3.2.1 What is the Aim of the Sampling

The objective of the investigation is a consideration for soil sampling orientation in all microtomography studies. Samples usually consist of 3-D blocks, generally comprised of undisturbed soil aggregates. It is important to stress that there is no single technique that can be applied to all investigations. Therefore, the most suitable sampling technique must be selected to meet the requirements of a specific study and, in some cases, new sampling techniques are necessary.

Based on the aim of the study, sample volume, number of samples, and the choice of what to sample need to be defined. It must be decided in advance whether the target of the planned analyses is quantitative or descriptive, in fact X-ray tomography (which indeed produces quantitative X-ray attenuation data) could also be simply employed to quickly identify processes (e.g., in contaminated sites the mixing of heavy metal sludge into soils). Descriptive approaches require fewer sample replicates than quantitative studies to achieve their goal.

3.2.2 Where to Sample

Selection of both location and selection of the portion of the soil profile to be sampled is strictly connected with the objectives of the research. For the sake of clarity, here we separate the two sampling systems defined by the research objectives: (i) geospatial sampling design (e.g., how to sample topsoils for bulk density analysis considering the spatial variability such as that occurring in two areas (a & b) of Fig. 3.1 and (ii) how to sample a soil profile (e.g., sampling following soil horizons in the soil profile in Fig. 3.1).

Fig. 3.1
A 3-dimensional structural diagram has 3 columns, a kubiena box, soil aggregates, and cylinders on one face. Multiple replicate structures on top and lateral face. Other labels are A P 1, A P 2, B w, C, a area, and b area.

General view about “where to sample” with respect to soil horizons (Ap, Bw, C). Abbr. Rep. = replicate

Full size image

Geospatial Sampling Design

Over recent decades, geospatial sampling design has received a great deal of research interest (e.g., pedometrics, agronomy) and also in the case of sampling for the assessment of soil structure (e.g., Leopizzi et al., 2018) and other microtomographic (e.g., Carducci et al., 2017) analysis. Additional information will be provided in the following section on geospatial sampling.

Profile Sampling Design

When a profile sampling design is applied, the soil should first be divided into soil horizons and the profile description must be as detailed and complete as possible. Samples are then taken from each horizon. If soil horizons are not well developed and expressed, samples may be spaced at regular depth intervals, while horizon boundaries and special or unique features might be sampled separately. Examples of recommended sample locations for various investigations are shown in Fig. 3.1.

For both geospatial and profile sampling design, it is good practice and thus strongly recommended that researchers always collect bulk samples for any extra laboratory analyses (e.g., particle size, clay mineralogy) that may be required to better understand tomographic results. Indeed, soil profile descriptions and lab analysis are extremely useful in order to support final evaluations of 3-D analysis.

3.2.3 Determining Sample Size

Here we define the “sample size” as the actual physical dimensions (e.g., cm3) of the soil sample and not the number of individual samples (most often named “sample size” in statistics applied to soil studies). The required size of the sample depends on:

  • the size of the specific feature of interest (e.g., pores, soil organic carbon, minerals) and their distribution. This is covered in more detail in the following section on “Representative Elementary Volume”. Nevertheless, we stress that examining planar pores that occur between large prismatic peds/aggregates will require larger samples than examining pores within a crumb microstructure would.

  • the sample stage of the available CT scanner is often a limiting factor. The sample size that can be accommodated might range from a few mm3 for high-resolution acquisition systems to over 1 m3 for industrial grade systems (e.g., GE V Tomex L).

  • the required trade-off between 3-D sample size and preferred spatial resolution (e.g., du Plessis et al., 2017) which represents a powerful limiting factor in all CT studies.

  • the size of the required image stack and associated image processing needs, and the need for digital subsampling for statistical analysis (e.g., digital pseudo replicate samples).

As additional observation, we wish to emphasize that larger 3-D samples (cm’s rather than mm’s) are often preferred to smaller ones because they support the observation of a wider range of features and how these features are interrelated. In some cases, large samples may be collected in the field and subdivided in the laboratory later in order to fit/be accommodated with the CT system being used. This is especially advisable when large aggregates are present and can be easily separated into smaller peds/aggregates.

3.2.4 How to Orientate the Sample

Although a soil block can be rotated in any direction using 3-D image processing software, it is self-evident that the orientation of the sample may be critical in some investigations, especially when sampling with cylinders is necessary (e.g., water saturation producing horizontal Fe segregation at a specific depth) or when considering specific depths within a soil profile. Therefore, it is usually important that the orientation of a sample is known. Generally, samples are collected either in the horizontal or vertical plane, while inclined samples may be required for specific purposes (e.g., studying pore–matrix interaction along a slickenside which is a sliding surface produced after wetting and drying cycles in a Vertisol). Vertically orientated samples are the most common in soil sampling procedures employed in soil microtomographic studies; they are especially useful to examine topsoil and specific soil horizons of soil profiles, bedding planes of sediments or vertical pore networks such as those created by roots or earthworms.

3.2.5 Number of Samples

The number of samples to be collected is a well-known issue in soil studies as it is very important and affects the reliability of the results obtained. Here we can consider two rather broad categories: (i) the required number of replicates in an experimental setup (e.g., soil structure studies after physical simulation of processes such as wetting and drying cycles) and (ii) the number of samples needed for a geospatial sampling design (e.g., soil structure studies of different land use mapping units). The latter of these two sampling designs will be considered in the following section on geospatial sampling. Substantial scientific literature already exists for both these considerations (e.g., Diel et al., 2019; Pöhlitz et al., 2019), but here we shall limit our investigation to a brief general overview of the subject.

With respect to the number of replicates to be acquired for a given experimental setup, the preliminary estimation of the number of replicates before sampling is necessary to ensure a given level of precision in the results. Indeed, the quantity of replicates is defined by considering the variability of the feature of interest in the soil. Therefore, it is recommended to perform preliminary tests before planning any specific experiment; these tests must enable the final decision about the required number of replicates for that specific study to be taken (e.g., Gargiulo, 2008; Gargiulo et al., 2015). However, such preliminary tests are generally not reported in scientific literature. Indeed, a short overview of the experiments involving analysis of pore networks by X-ray CT highlights the following variability in the number of replicates (although it is likely that scientists perform “look-see” scans in advance and do not report this in the literature):

  • 2 replicates: (e.g., Valdez et al., 2019)

  • 3 replicates: (e.g., Scotson et al., 2021; Menon et al., 2020; Ferreira et al., 2018; Gargiulo et al., 2016; Müller et al., 2018; Zhang et al., 2018)

  • 4 replicates: (e.g., Singh et al., 2020)

  • 5 replicates: (e.g., Diel et al., 2019; Pöhlitz et al., 2019).

However, it is fair to note that the replication used in soil CT studies has increased over time, most likely due to increased access and availability of the instruments. There are many studies focused on soil processes (sometimes performed as physical simulations in laboratories) in which it is not evident whether replicates were present or not (e.g., Arai et al., 2019; Pires et al., 2020; Liu et al., 2021), although this is not recommended even for pilot studies. In general terms, replicate samples should be collected whenever possible because a single sample is unlikely to be representative of the variation in soil structure. Replicate samples must be collected in all studies requiring a statistical analysis of data.

From a pragmatic standpoint, the number of replicate samples is frequently connected to the total number of 3-D acquisitions that a researcher can perform given access to X-ray CT instrumentation. This number can often be limited. In this respect, Gregory et al. (2009) highlighted the difficulties in dealing with a large number of samples often required in soil science. They suggest that by CT it is usually only possible to examine less than 10 samples in 1 day in normal conditions, with a consequent reduction of the potential for replicated trials. Although the speed and availability of CT equipment has increased considerably with time, that number is still a typical number of daily samples imaged in many laboratories around the world. Since imaging has certainly become faster, higher quality images still take time and this is often prioritized rather than scanning more samples, because lower quality images usually lead to longer processing time. There is a wide literature focused on the variability of soil porosity (e.g., Nunan et al., 2006) and correlated properties (e.g., bulk density), while few studies have been conducted to explore the variability of other micromorphological features (as highlighted by Vanden Bygaart & Protz, 1999).

3.2.6 When to Sample

Many soil features, such as pore space, can change dynamically over the course of a year. Indeed, depending on the seasonal soil water status, large planar pores (e.g., cracks) open or close, soluble minerals crystallize or dissolve and organic substances can decompose. Accordingly, potential seasonal changes in the features of interest must be considered when the most appropriate time for sampling is selected. Another potential problem that can impact on sampling is the extent of soil dryness, since dry fine textured soils such as clays are difficult to dig and then artefacts, such as cracks, may form as the sampling container is forced into the soil. In such a situation, sampling should be postponed until the soil becomes wet enough to avoid the occurrence of cracks during sampling, although in some cases introducing a small amount of water to the surface in the field can facilitate sampling. Evaluating when a soil is at an optimal moisture content for sampling is subjective, requires some testing in the field and it is fundamental to achieve good sampling especially in clayey soils. Quantitative evaluation of the optimal moisture content to avoid the above disturbance would usually require a preliminary triaxial test (a standard test in soil mechanics/engineering studies but rarely performed in this context by soil scientists).

3.2.7 How to Sample

The majority of CT applications require the use of undisturbed soil samples. The most common sampling methods are those using soil cores, metal or cardboard boxes or single aggregates (see Fig. 3.1 right, left and central part of the soil profile). Cores are used in the vast majority of sampling methods employed in experimental setups particularly when topsoil has to be sampled but single aggregates can also be sampled. In addition, other methodologies are used because of the specific consistency of the soil. Methods for the sampling of friable soils, cemented and cohesive materials, loose materials are discussed below.

Friable Soils

Most of these soils can be sampled by using plastic (e.g., PVC) cylinders with a cutting-edge side. These cylinders can be made at very low cost, can be manufactured easily to many different sizes (e.g., by cutting plastic PVC tubes), and allow reduced sampling disturbance in comparison with the traditional micromorphological boxes (e.g., Kubiëna tins). Cylinders are also the most common type of sample holder in soil hydrology and soil physics studies.

However, truly undisturbed soil and sediment sampling do not exist. Using X-ray CT, Carr et al. (2020) demonstrated artefacts in the pore system in all core samples due to pushing, cutting and hammering actions during sampling procedures. Any combination of hammer, rotation, percussive, and continuous push led to deformations, although the advanced trimming methods (Hvorslev, 1949) resulted in the least disturbance. This latter method refers to the slow insertion of a cylinder core vertically into the soil with simultaneous removal of the surrounding soil to minimize mechanical stress (Kemp, 1985; Carr, 2004), as should be done with a Kubiëna box (FitzPatrick, 1984; Stoops, 2009). Indeed, this demonstrates the value of the standard sampling approach, based on excavation of surrounding material as is widely used in soil micromorphological studies (FitzPatrick, 1993). Samples should ideally have a length and width that are suited to the X-ray CT scanner sample stage. The depth of the sample should be at least one-half its width to prevent its bending when filled (FitzPatrick, 1984; Murphy, 1986).

Another method, which is especially suitable for structured soils, is the sampling of soil aggregates (Fig. 3.1). This method (e.g., Peth et al., 2014) ensures that undisturbed samples are used when the element of research interest falls within the single aggregate (e.g., intra-aggregate pore space such as in Peth et al., 2008). After their collection, soil aggregates should be wrapped in protective material such as paper or plastic masking tape to minimize any disturbance damage.

Samples should be labelled with the i) orientation, ii) depth, iii) replicate number, and iv) profile number. Some researchers label the boxes on the lid first and photograph them in place before removing them from the soil, so a complete record of sample location and orientation is made (Murphy, 1986). Prior to sealing samples for transport, a few drops of fungicide or formalin can be added to each sample to prevent soil fauna burrowing within the sample during storage. All samples should then be wrapped in aluminium foil or placed in plastic bags to prevent them from either drying out or absorbing water while stored in the laboratory which can be a problem for air-dried materials, even from arid regions.

Cemented and Cohesive Materials

Undisturbed specimens of cemented materials (e.g., Bkm horizons) can be obtained without sample containers by cutting or carving pieces out of the soil. However, these should then be placed in boxes to prevent breakage or separation of the structural units. All such samples, as in the case of soil aggregates, should be wrapped in protective material such as foil or plastic and labelled as described previously.

Loose Materials

Some materials may be too loosely structured to be sampled without disturbance during removal from the soil. Particularly sensitive samples include sands, gravelly soils, self-mulching soil, some organic soils and recently tilled soils. These loose materials can be partially hardened in place prior to their being removed from the soil. Murphy (1986) gives step-by-step instructions for such procedures. Basically, the sample is first surrounded by a frame while it is still in the soil. The sample is then covered and partially impregnated with a cementing agent such as epoxy resin, cellulose acetate, plaster of Paris, sodium silicate, polyester resin, or a similar material. After the cementing material has hardened, the sample can be removed, labelled, and wrapped as previously described. This allows the sample to be dried and, if required, to be impregnated in the laboratory prior to scanning. Epoxy resin and plaster of Paris can be used to impregnate materials containing numerous large pores such as tilled Ap horizons (FitzPatrick 1984).

3.2.8 How to Document, Transport, and Store Samples

Prior to removal from a sampling site, all samples should be checked to make sure they are appropriately labelled and bagged. The profile description should be reviewed for any omissions. Site data should be recorded including standard GPS coordinates, landscape morphology, land use, etc., then photos (with additional sketches) of the profile or site can be made, and where possible they must show sample location (e.g., reference to specific horizons).

Before transport, samples should be placed in boxes and packed with suitable material to prevent them from coming into contact with each other and being damaged. Vibration of the samples should also be minimized using packing foam to reduce the risk of disturbance especially for friable samples. After arriving in the laboratory, samples should be stored in a cool, dry place but not frozen (as this may cause cracks). A temperature of about 4 °C or lower may be needed to retard the growth of plants and organisms. This is also especially important if reducing the disturbance of organic materials is required.

3.3 Representative Elementary Volume

The concept of representative elementary volume (REV) was introduced by Bear (1972) to describe flow in porous media. The approach deals with the definition of the minimum size of a sample necessary to present its characteristics of interest. In other words, the size at which the measured parameter (or property) becomes independent of the sample size. Its use in the quantification of soil structure, as REA (Representative Elementary Area), started with Vanden Bygaart and Protz (1999). The analysis of REV is commonly carried out by selecting consecutive soil volumes around a central point in the sample image (see Fig. 3.2). This is indeed feasible in the case of X-ray CT where adjacent constructions within the same 3-D image, but centred at different points, can be used.

Fig. 3.2
A 3-dimensional cuboidal structure has multiple internal layers. It has a small cuboidal structure inside.

Visual Representative Elementary Volume (REV) concept. A set of consecutive volumes where to measure a soil parameter to establish REV

Full size image

The representative size is then defined as the one corresponding to the transition from the detailed scale where the “microscopic effects domain” prevails (red signal in Fig. 3.3) to the more stable “porous media domain” (green line in Fig. 3.3) (Baveye et al., 2002; Borges & Pires, 2012). Thus, the REV size established for a specific property (Costanza-Robinson et al., 2011) corresponds to the smallest employed soil volume (the point named “V min” with the smallest volume on the green line in Fig. 3.3) to obtain representative measurements of the selected soil property.

Fig. 3.3
A graph of property versus volume has 3 regions. Region 1 for detailed scale has a wave-like curve. 2 for R E V domain has a horizontal line. 3 for aggregated scale has a small inclined line.

REV conceptualization (inspired after Bear 1972)

Full size image

Of course, REV varies according to the specific property under investigation, but also on the basis of the type of soil material. Indeed, even limiting our analysis just to the soil pores, it is evident that each porous media has its own characteristics (e.g., continuous pore space, channels, discrete pore within aggregates); therefore, the REV of a pore space parameter calculated for a specific soil may differ for the same parameter when a different soil type is analysed.

Limited research is available on the use of a REV in soils (Baveye et al., 2002; Vanden Bygaart & Protz, 1999) and, due to mild fluctuations, it is sometimes difficult to understand when it has reached the green line region (Fig. 3.3). In 1999, Vanden Bygaart and Protz (1999) highlighted a lack of suitable techniques for the study of representative size of undisturbed soil samples. But more recently, Borges and Pires (2012) analysed, respectively, the bulk density REV (50-100 cm3) and soil porosity distribution in a representative element area (REA) (8.8 cm2) for Brazilian soils, while Baveye et al. (2002), by analysing volumetric water content, volumetric air content, gravimetric water content, and dry bulk density of Ck horizons, demonstrated that the sizes of sampling volumes up to 60 × 60 × 30 mm influence the measured values of soil parameters significantly. They report that dry bulk density did not reach the green REV region and may move towards the red line of the aggregate scale as reported in Fig. 3.3.

In this scenario, it is not feasible to have “magic” REV numbers for a specific soil feature, while it is most appropriate to suggest some preliminary tests to researchers in order to evaluate the required REA or REV for the specific feature under investigation. Again, these preliminary tests are very highly recommended, but not difficult to perform. For instance, in Fig. 3.4, an example is reported in which the REV of an Ap horizon was defined before the experiment was performed. In this case, it is evident that 50 cm3 referred to the REV to be employed.

Fig. 3.4
Two images. Left image has a 3-dimensional cuboidal structure with areas marked by colored squares. Right is a scattered plot for total porosity versus volume with a vertical dotted line for R E V at 50. The higher concentration of dots is between 0 and 10.

Experimental setting to establish REV after X-ray microtomography. Here increasing soil volumes (x axis) are analysed and plotted against their corresponding total porosity. The vertical broken line identifies the experimental obtained REV (around 50 cm3)

Full size image

3.4 Geospatial Sampling

Geospatial sampling can be an important concern when samples for X-ray CT analysis are collected from the field. In general terms, we can treat this item in the following cases (moving from very specific to general cases):

  1. (i)

    A degraded contaminated site that requires remediation. In many cases contaminated soils show also signs of soil physical degradation. This may be related to soil compression and reworking by heavy machines. In these cases, it is important to combine the standard geochemical soil characterization with analysis of soil porosity. In fact, a detailed analysis of the porous system (e.g., pore size distribution and shape of pores) may help in addressing the most sustainable approach to soil remediation for each specific case. More details are reported below.

  2. (ii)

    Experimental field trials. This is the standard case when X-ray CT contributes to other agronomic measures (e.g., manure application, tillage practices, etc.), including cases when measurements of soil-pore parameters are influenced by crop rotations and cover crops (e.g., Singh et al., 2020).

  3. (iii)

    Standard landscape analysis. This is the case when soil survey may benefit from X-ray CT analysis to evaluate specific issues such as soil compaction.

  1. (i)

    contaminated site showing also signs of physical degradation,

    In contaminated sites, soil spatial variability is a key concern since it can combine to create a twofold spatial variability due to the two main sources of the contamination, natural and anthropogenic. In further detail, the spatial distribution of anthropogenic soil contamination is almost always unknown and ex-ante information is generally lacking or completely missing. In these cases, the use of proximal soil sensing devices can assist. In fact, the most widespread proximal sensing devices such as Electro Magnetic Induction (EMI) devices or Electrical Resistivity Tomography (ERT) devices are strongly affected by the soil bulk density (in addition to other soil properties such as water content, salt content, etc.). Therefore, the incoming geophysical mapping results (generally obtained by a quick scan of the site) can be profitably employed as a powerful means to better address soil sampling for the analysis of the soil porous system.

    Two interesting cases, that can be profitably applied to X-ray CT, are reported by Langella et al. (2018) and Vingiani et al. (2022). These authors using different geophysical EMI and ERT sensors (Profiler, DUAL-EM and ARP) and γ-ray spectrometry recognized spatial structures and zones with different levels of anomalies, in agricultural and industrial sites. High consistency was found among the maps obtained by the different sensors. More specifically data was merged to report all the information on a regular grid at very high-resolution (50 cm at the field scale). This fusion of the data collected from the different sensors was transformed to the same raster-based geometry. This procedure produced a map of the areas in which to collect soil samples and account for the spatial variability of the target variable and highlight possible hot spots. The EMI, ERT, and γ-rays sensors approach was validated by XRF measurements of soil elemental composition by the use of a field portable technique.

    In contaminated sites, soil sampling may require policy compliance: this refers to those cases where sampling must follow specific regulations and some specific sampling rules are prescribed or strongly suggested. This, for instance, is the case for soil contamination in some countries where a specific number of samples/ha are required (e.g., Italian D.M. 471/99) on the basis of the global spatial extent of the site (e.g., at least 5 sampling points for areas <1 ha).

  2. (ii)

    Experimental field trials. This is the standard approach employed when X-ray CT contributes to an agronomic assessment. A comprehensive description is given in Gilbert (1987), de Gruijter, (2002) and Pennock (2004). Although many types of sampling designs exist, only two main types (random and systematic) are commonly used: i) Simple Random and Stratified Random Sampling, where in Stratified Sampling points refer to predefined strata and a simple random sample chosen from each stratum and ii) Systematic Sampling which refers to either transects or most commonly to grids. Where constant spacing is applied, the major caution has to be that the soil property to be sampled must not be arranged in an orderly manner which might correspond to the spacing of the grid. Bonfante et al. (2017) highlight that the implicit assumption of soil homogeneity of the agronomic design experiments may lead to erroneous results especially when subsoil condition matter. In this case, special additional care is required in identifying the location of the different soil types before performing appropriate sampling.

  3. (iii)

    Standard landscape analysis. This can be performed by:

    1. a.

      Sampling using deterministic/judgement rules. This is the classical approach applied for pedology, geomorphology, geology, etc. surveys. In the case of soils, the approach relies on the evidence that soils (S) are determined by specific soil forming factors (Jenny, 1941) following the classic Jenny eq. S = f (climate, organism, relief, parent material, time), thus data on these factors may lead to soil knowledge. In view of the factual soil complexity in real landscapes, this approach has exhibited evident limitation. Therefore, in the last three decades extensive research has developed an entirely new branch known as Digital Soil Mapping. Despite the above, currently, judgement-based rule approaches are very important. In this case, researchers use their judgement to locate sampling points in landscape positions where the soil representative of a specific land is most likely found. For instance, this type of sampling is widely used by private-sector environmental consultants and the specific objective may range from an initial evaluation of the extent of the problem to be addressed (e.g., contamination) to the final stage of problem solving through a mapping exercise. Laslett (1997) stated that consultants who undertake these surveys almost always employ judgement sampling and place their sampling points where their experience and prior knowledge of the site history suggest the contamination might be located.

    2. b.

      Sampling using Geostatistical, Spectral, and Wavelet Analysis. These types of sampling have become a “must” in sound based scientific sampling over the last few decades. Details are reported in Brus and de Gruijter (1997), de Gruijter (2002), Webster and Oliver (1990), McBratney et al. (2002), Mulla and McBratney (2000). All these approaches—which fall under the broad umbrella of Digital Soil Mapping approaches—address the spatial dependence in soil properties between locations. Thus, the location of each sample point in space is critical information.

3.5 Sample Preparation

3.5.1 Preliminary Laboratory Analysis

Different types (e.g., size and shape) of samples can arrive in the laboratory for X-ray imaging such as cylinders, Kubiëna boxes, aggregates, and bulk samples (Fig. 3.1). In all cases, the start point of any investigation should be a visual assessment. This can be done by direct stereo-microscopic analysis of the soil aggregates or, when this is not feasible (e.g., cylinders and Kubiëna boxes), by analysing the corresponding bulk samples. In this case, the soil has to be observed directly with oblique incident light to identify microstructure, macropores, Fe-Mn features (such as concretions, segregation, pyrite), and CaCO3 features; indeed, all these features may influence X-ray absorbance and can usually give some important insights into the sample. At this point, samples (Fig. 3.5) would usually either/both go directly for X-ray scanning without any treatment or/and undergo specific treatments.

Fig. 3.5
A diagram has 2 types of field scale, soil profile, cylinder, x-ray microtomography, and 3 treatment methods which are Air-drying, scanning treatment, and Impregnation with resins.

Synoptic figure showing (i) location of sampling (either free survey in the landscape or experimental fields), soil profile sampling, different soil samples treatments before X-ray CT analysis

Full size image

No treatment: It is worth noting that one important advantage of X-ray CT is the opportunity to acquire 3-D soil blocks directly without any prior preparation (including impregnation) which accelerates analysis of the soil-pore system.

Treatment: even considering the advantages of no treatment, there are two major cases where pre-treatments are required:

  1. (i)

    Staining specific features to separate different soil features better during (and after) 3-D scan acquisition. Even though the soil-pore system analysis is often the main focus of X-ray CT imaging, there is a considerable interest in analysing the pore system in combination with other soil features such as water and organic matter. Iltis et al. (2011) employed potassium iodide (KI) as a contrast-enhancing agent enabling a better segmentation of structure and porous media. Furthermore, under specific conditions, many microtomography studies also showed the ability to separate soil, air, and water (an overview is given in Tracy et al. (2015) and Mao et al. (2016)) while the situation is more complex for other soil features such as minerals (Guntoro et al., 2019) and soil organic matter (SOM) (Mao et al., 2016). The main problem lies with the evidence that, although the physical and chemical properties of the SOM are very different from those of the bulk soil (Kasteel et al., 2013), it can be very difficult to differentiate between these properties in the soil complexity due to the similarity of their X-ray attenuation coefficients (Quinton et al., 2009). The huge interest in separating the SOM phase from the bulk soil has pushed soil scientists to invest a great effort to achieve such separation by using either or both pre-treatments and image processing techniques (see Chap. 10). X-ray CT can be used, under specific conditions, to visualize large organic fragments (e.g., Particulate Organic Matter—POM) and plant roots (Kravchenko et al., 2015; Van Loo et al., 2014), but the visualization of organic matter that is well dispersed in the soil matrix (e.g., nonparticulate SOM) is very difficult (e.g., Mueller et al., 2013) because of the well-known difficulties in segmenting an image where SOM and the rest of the soils have similar ranges of varying linear attenuation coefficients. An advantageous approach to simplify image segmentation is to treat the sample with heavy elements. Indeed, soil organic matter has a marked affinity for heavy elements (with Z > 30)—due to their charge, ionic radius, and ionic potential and, in turn, heavy elements have an X-ray attenuation that increases with their atomic number and, thus, improves imaging contrast (Van Loo et al., 2014). This is covered in greater detail in Chap. 10.

    The most well-known heavy elements for soil “staining” (here we use the word staining even though it originally referred to optical microscopy) are the following:

    • Silver (Ag): (Van Loo et al., 2014) which is the most well-known staining system.

    • Lead (Pb): Kettridge and Binley (2008) successfully used lead (II) nitrate solution to increase the linear attenuation of peat by flushing it over a sample before X-ray scanning. However, we should also emphasize the highly carcinogenic nature of lead (II) nitrate.

    • Osmium (Os) tetroxide (Rawlins et al., 2016). In 2014, Peth et al. (2014) proposed a new method of in situ SOM visualization which could be implemented on intact soil samples up to a few centimetres in size. This method is based on the ability of osmium tetroxide, OsO4, to react with organic substances, in particular, lipids. An air-dry soil is subjected to OsO4 vapours, which, upon diffusion into the soil, bind with organic materials and stain them. Mao et al. (2016) found that the method of Os (OsO4 vapours) staining (analysed via X-ray Dual-Energy Tomography) was effective in staining organic materials of root origin and the organics associated with fine soil particles, but not biochar. Arai et al. (2019) demonstrated effective SOM localization within large macroaggregate based on synchrotron X-ray micro-computed tomography (μCT) coupled with a vapour-phase, osmium (Os)-staining pre-treatment. Unfortunately, osmium (Os) tetroxide is also toxic (Van Loo et al., 2014) and special care when handling is required.

    • Iodine (I) (Boyde et al., 2014). This approach is largely used in combination with hydrological testing (Scotson et al., 2021). See Chap. 7 for more details.

    • Eosin (Br based) (Lammel et al., 2019).

    Several other heavy elements such as Cu, Co, Fe, Mn, Mo, and Zn may be of interest in binding SOM, but, since they already occur naturally in soil, their use must be evaluated in each specific soil case. Lammel et al., (2019) tested many different contrast enhancing agents by using synchrotron radiation microcomputerized tomography (SR-μCT) and demonstrated that I2 was the most efficient method as it was able to improve the image contrast, so providing a powerful tool to determine the spatial location of SOM.

  2. (ii)

    Water removal: In this context the aim is to achieve standardization of the 3-D acquisition and an improved soil-pore segmentation. Moist and wet samples can affect the quality of X-ray acquisition images because air, water, and soil phases have different linear X-ray attenuation coefficients and, therefore, soils with heterogeneous humidity may produce images that are more difficult to segment due to these variations. Water removal can be carried out by (a) air- or oven- drying of samples, and (b) alternative dehydration methods from soil micromorphology (e.g., acetone dehydration). The method used depends on the purpose of the survey and on the characteristics of the soil sample. Air-drying is a rather simple method which may take from 5 to 8 weeks (depending on the soil properties) and may still require the use of an oven. The most popular method is to oven-dry samples at 40 °C for just a few days. However, in soils characterized by the presence of expandable minerals (e.g., smectites) and/or by high organic matter content, the treatment could modify the soil structure. In such a case, standard soil micromorphology approaches, including dehydration by saturating the sample with an acetone-water mixture, become advantageous. This involves the stepwise substitution of ever-increasing concentrations of acetone (Murphy, 1986; FitzPatrick, 1993), in order to remove the water with the help of zeolite.

  3. (iii)

    Water removal and resin impregnation: This treatment permits to submit the same sample to both X-ray CT scanning and thin section analysis. Once the water has been removed, the sample can be impregnated with different mixtures of unsaturated polyester synthetic resins or epoxies depending on how the sample was dried. For further details, see FitzPatrick (1984), Jongerius and Heitzberger (1975), and Murphy (1986). This approach is rather unusual when dealing with microtomography, but it can be very profitable both as a support for 3-D scan interpretation and analysis. Indeed, in this case, after the analysis of a 3-D scan, which provides grey levels on the basis of the X-ray attenuation coefficient, the researcher, through a standard optical microscopy analysis or through chemical microanalysis (e.g., EDS-SEM) of a soil thin section, can access the power of a much larger spectral content (e.g., visible, UV, X-ray chemical mapping) than provided by the 3-D CT scan. This combination of approaches can be very rewarding as demonstrated by Hapca et al. (2011 and 2015) and Bendle et al. (2015).

3.6 Conclusions

In this chapter, the crucial value of soil sampling and sample preparation in X-ray CT studies has been addressed. Indeed, badly designed soil sampling or poor sample preparation can lead to erroneous results spoiling any subsequent analytical techniques. Actually, the more complex and advanced applied X-ray techniques (e.g., synchrotron based) may result in very small samples which are even more affected by the above problems. Therefore, also the concepts of representative elementary area (REA) and representative elementary volume (REV) are vital, due to the importance of knowing the size at which the measured parameters (or properties) becomes independent from the sample size, and then when representative measurements are really obtained.

However, nowadays in which there is an increasing possibility to access a wide range of equipment, we also highlight the potential to integrate X-ray microtomography analysis with standard 2-D analysis (including standard optical microscopy via thin sections) and visual soil sample analysis in order to obtain a deeper knowledge on any specific soil sample, to then achieve stronger basis for upscaling the results.