Soil Fabric and Its Measurement
5.7 SAMPLE ACQUISITION AND PREPARATION FOR FABRIC ANALYSIS
Obtaining representative soil samples with minimal disturbance is essential if reliable measurements of en-gineering properties are to be made. The same consid-erations apply in the selection and preparation of samples for the study of fabric. Accordingly, the sam-pling and preparation phases of fabric study are criti-cal, and special methods are many times needed.
Proven methods for reliable determination of fabric can also be used for evaluation of the effects of dif-ferent sampling procedures used in engineering prac-tice, although there does not appear to be much record of this having been done.
Both direct and indirect methods are used to study the fabric and fabric features of soils, as listed in Table 5.3. An illustrative schematic diagram prepared by R. N. Yong that summarizes methods for analysis of soil composition and fabric using various parts of the electromagnetic spectrum is shown in Fig. 5.18. In in-terpreting the results from any of these methods, judg-ment is required to be sure that the conclusions pertain to properties and behavior of interest. For example, discontinuities, fractures, and anisotropy on a macro-scale can override the influences of microfabric details.
Of the methods listed in Table 5.3, optical and elec-tron microscopy, X-ray diffraction, and pore size dis-tribution offer the advantage of providing direct (usually) unambiguous information about specific fab-ric features, provided the samples are representative and the sample preparation procedures have not de-stroyed the original fabric. On the other hand, these techniques are limited to small samples, and they are destructive of the samples studied. The other tech-niques are nondestructive, at least in principle, and can be used for the study of soil fabric in situ and for the study of changes in fabric that accompany compres-sion, shear, and fluid flow. However, with most of these methods interpretation is seldom straightforward or un-ambiguous. The use of several methods of fabric
anal-ysis may be appropriate in some cases in order to obtain information of more than one type or level of detail.
Sample Preparation for Fabric Analysis
Acoustical, dielectric, thermal, and magnetic measure-ments can be made directly on samples in their undis-turbed, wet state. Optical and electron microscopy, X-ray diffraction, and porosimetry require that the pore fluid be removed, replaced, or frozen. To do this with-out disturbance of the original fabric is difficult, and in most cases there is no way to determine how much disturbance there may have been.
Pore Fluid Removal Air drying without significant disruption of the natural fabric may be possible for soils that do not undergo much shrinkage. For soft samples at high water content, oven drying may cause less fabric change than air drying, evidently because the longer time required for air drying allows for greater particle rearrangement (Tovey and Wong, 1973). On the other hand, the stresses induced during oven drying may result in some particle breakage.
Water removal by drying at the critical point has also been used. If the temperature and pressure of the sam-ple are raised above the critical values, which for water are 374C and 22.5 MPa, respectively, the liquid and vapor phases are indistinguishable. The pore water can then be distilled off without the presence of air–water interfaces that can lead to shrinkage. The high tem-perature and pressure may change the clay particles, however. To avoid this, replacement by carbon dioxide has been used. The critical temperature and pressure of carbon dioxide are 31.1C and 7.19 MPa, respec-tively. The procedure requires prior impregnation of the sample with acetone, which may cause swelling in partly saturated and expansive soils (Tovey and Wong, 1973). Both critical point and freeze-drying cause less sample disturbance and shrinkage than do air or oven drying, but they are more difficult and time consuming.
Freeze-drying can be used for removal of water.
Sublimation of the ice in a soil that has been rapidly frozen avoids the problem of air–water interfaces and shrinkage that accompany water removal by drying.
Sample size must be small, usually thinner than about 3 mm, if disruption due to nonuniform freezing is to be avoided. Quick freezing is best done in a liquid that has been cooled to its melting point in liquid nitrogen, such as isopentane at⫺160C or Freon 22 at⫺145C.
This avoids gaseous bubbling caused by direct immer-sion in liquid nitrogen at⫺196C (Delage et al., 1982).
The freezing temperature should be less than⫺130C to avoid formation of crystalline ice. Sublimation of
125
Polarized Light Micrograph
Replica Transmission Electron Micrograph or Diffraction Pattern
Scanning Electron Micrograph
Figure 5.18 Methods for examining mineralogy, fabric, and structure of soils using parts of the electromagnetic spectrum (prepared by R. N. Yong, McGill University Soil Mechanics Laboratory).
Table 5.3 Techniques for Study of Soil Fabric
Method Basis
Scale of Observations and Features Discernable
Optical microscope (polarizing)
Direct observation of fracture surfaces of thin sections
Individual particles of silt size and larger, clay particle groups, preferred orientation of clay, homogeneity on a millimeter scale or larger, large pores, shear zones
Useful upper limit of magnification about 300⫻
Electron microscope Direct observation of particles or fracture surfaces through soil sample (SEM)
observation of surface replicas (TEM)
Resolution to about 100 A˚ ; large depth of field with SEM; direct observation of particles; particle groups and pore space; details of microfabric; environmental SEM can be used to observe
specimens containing water and gas
X-ray diffraction Groups of parallel clay plates produce stronger diffraction than randomly oriented plates
Orientation in zones several square millimeters in area and several micrometers thick; best in single mineral clays
Pore size distribution (1) Forced intrusion of a nonwetting fluid (usually mercury)
(1) Pores in range from⫺0.01 to
⬎10m
(2) Capillary condensation (2) 0.1m maximum Wave propagation Particle arrangement, density,
and stress influences wave velocity
Anisotropy; measures fabric averaged over a volume equal to sample size and properties; measures fabric averaged over a volume equal to sample size
Thermal conductivity Particle orientations and density influence thermal conductivity
Anisotropy; measures fabric averaged over a volume equal to sample size
Magnetic susceptibility
Variation in magnetic
susceptibility with change of sample orientation relative to magnetic field
Anisotropy; measures fabric averaged over a volume equal to sample size
Properties reflect influences of fabric; see Chapter 11
Fabric averaged over a volume equal to sample size; anisotropy;
macrofabric features in some cases
METHODS FOR FABRIC STUDY 127 the water is then done at temperatures between ⫺50
and⫺100C rather than at the initial freezing temper-ature to increase the rate of water vapor removal. At temperatures less than ⫺100C the vapor pressure of the ice, about 10⫺5torr, may be less than the capability of the vacuum system.
The freezing process may produce fabric changes in very high water content systems such as a 10 percent by weight slurry of bentonite in water (Kumai, 1979).
However, with more typical saturated clays at consis-tencies likely to be encountered in geotechnical inves-tigations, the effects of freeze-drying on the fabric are small. Additional considerations in sample preparation by freeze-drying are given by Tovey and Wong (1973) and Gillott (1976).
Pore Fluid Replacement If thin sections are re-quired, as for optical microscopy or when drying shrinkage must be minimized, but the presence of a material in pore spaces is not objectionable, replace-ment of the pore water may be necessary. Various res-ins and plastics have been used for this purpose.
High-molecular-weight ethylene glycol such as Car-bowax 6000 is miscible with water in all proportions and has been used for many studies. Carbowax 6000 melts at 55C but is solid at lower temperatures.
Impregnated samples are prepared by immersing an undisturbed cube sample, 10 to 20 mm on a side, in melted Carbowax at 60 to 65C. The top surface of the specimen should be left exposed to vapor for the first day of immersion to allow escape of trapped gases and prevent specimen rupture. The wax should be changed after 2 or 3 days to ensure water-free wax in the sample pores. Replacement of all water by the Carbowax is usually complete in a few days. After removal from the liquid wax and cooling, the sample is ready for sectioning.
Thin sections are prepared by grinding using emery cloth or abrasive powders and standard thin-section techniques. However, heat, water, or other water-soluble liquids cannot be used at any stage of the grinding or section mounting process. Measurements by X-ray diffraction have shown that Carbowax re-placement of water has essentially no effect on the fab-ric of wet kaolinite (Martin, 1966).
Gelatins or water-soluble resins may be used in lieu of Carbowax, or the sample may be impregnated with methanol or acetone before replacement with resins or plastics. Further details on resin impregnation are given by Smart and Tovey (1982) and Jang et al.
(1999).
Preparation of Surfaces for Study
Surfaces chosen for study should reflect the original fabric of the soil and not the preparation method.
Grinding or cutting air-dried and Carbowax-treated samples may result in substantial particle rearrange-ment at the surface, thus making them of little value for study by the electron microscope. To overcome this problem, successive peels from the surface of a dried specimen using adhesive tape can be used to expose the original fabric. Alternatively, the surface may be coated with a resin solution that partly penetrates the sample. After hardening, the resin is peeled away re-vealing an undisturbed fabric. A comparison of sur-faces before and after this procedure is shown in Fig.
5.19.
The disturbed zone at the surface of Carbowax-treated samples extends to a maximum depth of about 1 m in kaolinite (Barden and Sides, 1971). As thin sections used for polarizing microscope study are of the order of 30m thick, this disturbed zone is of little consequence. It is also insignificant for X-ray diffrac-tion studies.
Fracture surfaces in dried specimens are sometimes taken as representative of the undisturbed fabric. Ad-ditional preparation, such as gentle blowing of the sur-face or peeling is needed following fracture because (1) there may be loose particles on the surface, and (2) a fracture surface may be more representative of a plane of weakness than of the material as a whole. An alternative approach to avoid these problems is to frac-ture a frozen wet specimen as described by Delage et al. (1982).
The method of sample preparation should be se-lected after consideration of scale of fabric features of interest, method of observation to be used, and the soil type and state as regards water content, strength, dis-turbance, and so forth. With these factors in mind, the probable effects of the preparation methods on the fab-ric can be assessed.