On the Structure and Properties of Clays1

William G. Bragg

In lectures I delivered in 1925–1926, I had occasion to speak about those changes in the old crafts that had been brought about by new knowledge.2 This knowledge was concerned in part with a deepening of our understanding of the materials used in the crafts. Thus, for example, the smith acquired new knowledge about the composition of his alloys, and the textile worker—about the details of the structure of his fibers. But an exception fell to the lot of the potter. At that time there was still no simple answer to the question: “What is clay?” No one could then explain either the plasticity of clay or its ability to absorb water, i.e., those properties that must depend on the arrangement of the atoms and molecules composing clay. New X-ray methods showed that clay is a crystalline substance and that in its structure it differs both from aluminum oxide and from quartz, although its composition is such that precisely these two substances seem, as it were, to be its components. At that time the X-ray methods were not sufficiently developed to be applied to the investigation of such obscure substances. Now the situation has changed. With the aid of X-ray methods it has proved possible to discover the principal features in the structure of clays, although very many enigmatic details still await explanation. It now seems possible to give some explanations of the properties of clays for those who use them for the most diverse purposes.

These applications are numerous, and the properties of clays are very curious. The potter finds that his clay, when it has been carefully prepared, can easily flow into his molds, and that a very small deviation from the treatment accepted in such cases ultimately leads to the destruction of his product. The agro-

W. H. BRAGG

the chemist finds that clay can change its behavior as a soil material depending on the ease with which it proves capable of exchanging its bases for others, for example, potassium for calcium. He is also interested in the ability of clay to absorb such large quantities of water that from a solid body it turns into a completely plastic mass. The foundryman requires that his clay be plastic, but that at the same time it not wet his fingers. The geologist, however, has to deal with the differences among various clays, since only in this way is he able to draw a picture showing when each of the clays was formed. Finally, the crystalline structure itself, as well as the physical and chemical conditions of existence of clay crystals, are extremely important from the scientific point of view.

Modern determinations of the structure of clays are based to a considerable extent on a whole series of investigations carried out in Manchester by my son W. L. Bragg1 and his collaborators. Several years ago, in studying the structures of silicates, he succeeded in discovering their principal features and in establishing certain details of their construction. This work is described in Proc. Roy. Soc. and other journals. The next important step in this direction was made by Pauling2, who showed that talc, pyrophyllite, micas, and other minerals of the group to which clays belong are “layer crystals,” in which layers of atoms are arranged one above another. He made measurements and described the chief features of their structure. Other workers, to whom I must now refer, continued to move forward along the path indicated by Pauling.

It may be recalled here that silicates form a considerable part of the earth’s surface. Two elements—oxygen and silicon—account for about three quarters of the earth’s crust, and therefore it is not surprising that compounds of these elements should be quite common. The relative abundance of some of the elements is shown in Table 1, which is borrowed from Goldschmidt’s work[^3].

TABLE 1

Quantities of elements contained in the earth’s crust, expressed in grams per ton

In silicates the silicon atom always proves to be surrounded by four oxygen atoms, which are arranged at the vertices of a tetrahedron, at the center of which the silicon atom itself is located. This constitutes the most important feature of the structure of silica. The length of the edge of this tetrahedron is always 2.55 Å.

This bond of the silicon atom with four oxygen atoms is in agreement with the rules according to which molecules are formed from atoms. According to these same rules the structure of clays is also built up, which is what I wish to show here; moreover, I shall confine myself here only to that part of these rules which is of interest for the present purpose. I hope that my chemist friends will not resent these indications, since the present report is intended also for those who are not fully acquainted with chemical laws.

According to the now accepted, extremely useful conception—which, to be sure, seems somewhat crude in the light of modern knowledge—electrons, which have a negative charge, revolve around the positive nuclei of each atom. Normally the number of electrons is sufficient to compensate the positive charge, but it can also change. If it increases, the atom acquires a negative charge, and conversely. These changes occur mainly owing to the tendency toward certain preferred groupings of electrons. In particular, a very favorable grouping is one in which eight electrons, situated at the vertices of a cube, form the outer shell of the atom. This is especially applicable to the case of those atoms which form clays.

An oxygen atom normally possesses eight electrons, and this number of them is just sufficient to compensate the charge of the nucleus. However, two of them are firmly held in the inner shell, the one closest to the nucleus. The remaining six form the outer shell, and therefore the oxygen atom tends, when an opportunity presents itself, to accept two more electrons from other atoms, despite the fact that thereby it becomes electrically charged. On the other hand, metal atoms have one or more electrons in excess of those required for a complete cubic arrangement. Thus, for example, a sodium atom has only eleven electrons; two of them form the inner shell, eight fill the vertices of a cube, and one remains in the outer shell. Although the latter, too, is together with the other electrons under the action of a positively charged nucleus eleven times stronger, it can easily be removed, leaving the atom as a whole with a unit positive charge, sufficient to compensate the negative charge of one electron.

If oxygen comes into combination with two sodium atoms, it can take to itself two electrons which are held more or less freely by the sodium atoms. In this way the negative “ion” of oxygen will attract two positive “ions” of sodium, and all three ions together will form what is called a molecule of sodium oxide, to which a certain ...

stability. The elements usually contained in clays are listed in Table 2 together with the number of deficient or excess electrons.

TABLE 2

From this table it is clear that a certain molecule may be formed from one oxygen atom and one calcium atom; this will be a molecule of lime. One oxygen atom and one magnesium atom give magnesium oxide. One oxygen, one sodium, and one hydrogen give a molecule of caustic soda. In structures of various kinds, oxygen is so often encountered in combination with one hydrogen that this pair has received the special name “hydroxyl.” The hydrogen nucleus in this case is concealed in the electron shell of the oxygen atom, and the hydroxyl therefore proves to be an approximately spherical formation. The hydroxyl as a whole is negatively charged; like many molecules, it is polar because the opposite charges are not concentric. The combination of one oxygen atom and two hydrogen atoms gives, of course, a molecule of water, which has no need to attract further electrons to itself and at the same time does not tend to give them back.

Let us now imagine a group consisting of a silicon atom surrounded, at the vertices of a tetrahedron, by four oxygen atoms. The silicon atom gives up the four electrons that can be borrowed from it, and each of the four oxygen atoms captures one of these electrons, needing no further quantity of them. If one hydrogen atom is added to each oxygen atom, the group becomes saturated. The oxygen atoms become hydroxyl groups. The group obtained in this way is the well-known molecule of silicic acid, $\mathrm{Si(OH)_4}$. In this way more or less stable molecules are obtained. The reason for such stability is, first of all, that the giving up and capturing of electrons among the constituent atoms satisfy the rules governing the arrangement of electrons and, secondly, that the silicon atom thereby becomes positively charged and firmly holds the four negatively charged hydroxyl groups that surround it. Any atoms capable of giving up one electron, such as, for example, sodium or potassium, do this as easily as hydrogen.

Let us now consider further combinations of silicon with oxygen. Let us imagine a combination of two tetrahedra having one common vertex and, consequently, one common oxygen atom (Fig. 1). To each of the six other oxygen atoms there must be added...

one electron each by adding hydrogen atoms or atoms equivalent to them. In this case the molecule $\mathrm{Si_2O_7H_6}$ will be obtained. Let us trace the next stage of such a combination. Imagine a chain of tetrahedra, as in Fig. 2, where each tetrahedron has one oxygen atom in common with each of its neighbors. Then for each silicon atom there will be one oxygen atom of the bonded type and two oxygen atoms that are not bonded. The latter two require the addition of electrons, which can be done by adding to them an atom of hydrogen, sodium, or another monovalent atom. Let this be a sodium atom. As a result there will be obtained a chain of indefinite length, consisting of links in each of which the compound $\mathrm{Na_2SiO_3}$ is repeated. Of course, the ends of the chain, for their completion, also require the addition of one monovalent atom to each of them.

Fig. 1. Arrangement of two tetrahedra having one common vertex. They are shown by solid lines. One of them is turned with respect to the first, but this feature of the arrangement is not essential. The vertices of each tetrahedron may be considered as centers of oxygen atoms that touch one another, while two cavities of the tetrahedra may be occupied by silicon atoms, the centers of which lie at two places of intersection of the dotted lines.

This compound is known as sodium silicate. It has a strong affinity for water, as does clay; it is highly probable that in both cases this property is due to a common structural feature. For example, when surrounded by water molecules, it forms a viscous medium known as water glass, the viscosity being due to the entanglement of long chains.

Now, after repeating the tetrahedra in one direction to construct a chain, we shall extend these chains sideways so that layers are obtained. Such an arrangement is shown in Fig. 31.

Fig. 2. Chain of tetrahedra in which successive links have one common vertex each. One face of each tetrahedron lies in a common plane, shown in the drawing as a straight board, and all the tetrahedra shown lie on one side of this plane; but this particular arrangement is not essential.

Solid circles on the solid line denote hydroxyl groups,

beneath each of which lies a silicon atom. This latter, in turn, rests on three oxygen atoms. The silicon and oxygen atoms are shown by dotted lines. The bonding rules are satisfied here, since each silicon atom gives four electrons to the three oxygen atoms and to the one hydroxyl group with which it is in contact, while each oxygen atom accepts one electron from each of the silicon atoms with which it is in contact. Each silicon atom has one independent hydroxyl group and, in addition, halves of three oxygen atoms, so that the composition in this case will be \( \mathrm{SiH}\left(\mathrm{O} + \frac{3}{2}\mathrm{O}\right) \), or, more simply, \(\mathrm{Si}_2\mathrm{O}_5\mathrm{H}_2\). A photograph of such an arrangement is shown in Fig. 4. Of course, such a layer has a boundary, at which the composition will be somewhat different; but when the layer is very broad and contains many thousands of atoms, this border does not play a significant role and does not alter the arguments given above.

Fig. 3. In this drawing the hydroxyl groups are represented by solid circles, each of which rests on three oxygen atoms, whose dimensions are smaller by a factor of \(2:\sqrt{3}\). Between all four of them is a cavity in which there is a silicon atom. Each tetrahedron has three oxygen atoms in common with neighboring tetrahedra. All tetrahedra lie on one side of the plane in which one face of each tetrahedron is located. Cf. also Fig. 4.

Such a layer is called a layer of silicic-acid hydrate; in what follows, for brevity, we shall call it simply a “silicic-acid layer.”

It has not been possible to observe it in isolated form, but it very often occurs in compounds, to the consideration of which we shall now proceed.

From such a two-dimensional layer we may pass to a three-dimensional solid body. Thus, for example, in quartz, tridymite, and cristobalite each oxygen atom belongs to two tetrahedra, while hydrogen does not enter into the compound at all. The structures thereby obtained

Fig. 4. Photograph showing the arrangement of atoms in a layer of silica hydrate. The oxygen atoms that form the lowest layer are arranged as in Fig. 9 A. The hydroxyl groups, denoted by the dark paint, rest on the oxygen atoms and form the arrangement shown in Fig. 9 B. Silicon occupies cavities in the tetrahedra; two cavities have been exposed for illustration.

Fig. 5. The black dots represent the centers of hydroxyl atoms. The white circles represent the second layer, lying above the first. In both layers each hydroxyl group is in contact with six similar groups in its own layer and with three in the other layer. The crosses mark the centers of magnesium atoms, each of which is situated in the cavity of an octahedron formed by six hydroxyl groups, three from each layer. See also Fig. 6.

are solid and strong. However, three-dimensional structures of this kind lie outside the scope of our discussion, and we shall return again to layers, for they are the most important constituent part of clays.

Attention should be paid to the compounds of oxygen with aluminum and magnesium. Let, in Fig. 5, the centers of hydroxyl groups lying in one plane and arranged in such a way that each such group, which may be regarded as spherical in shape, is in contact with six similar groups, be denoted by points. Let, further, the magnesium atoms be in the positions denoted by small crosses, each of these atoms corresponding to a triangular cavity between three adjacent hydroxyl groups. It is evident that there are the same number of crosses as points, since each of them lies exactly under the corresponding point. If a second layer, likewise consisting of hydroxyl groups, is placed over the first layer, then their centers will lie in one plane in the positions indicated by small circles. Thus each magnesium atom is surrounded by six hydroxyl groups touching it, three above and three below. The six hydroxyl groups lie at the vertices of a regular octahedron. At first glance this does not seem obvious, but one must agree with it after considering Fig. 6, in which the octahedra are shown from two points of view. Of course,

Fig. 6. Two representations of six spheres lying at the vertices of an octahedron; each of the spheres touches four others. The left drawing shows the octahedral arrangement better; the right drawing shows the arrangement of three of them in relation to the three others. The seventh sphere is in this octahedral cavity. The six large spheres may be hydroxyl groups, and in the cavity there may be a spherical magnesium atom. Constructing such a model from spheres is very useful for understanding.

the radius of the magnesium atom must be much smaller than the radius of the hydroxyl group, since only by this means can it be located in the cavity; the ratio of these two radii must be less than 0.41.

In this double layer of hydroxyl groups, including a single layer of magnesium atoms, each of these atoms corresponds to two hydroxyls. Since each hydroxyl accepts one electron, and each magnesium atom must give up two electrons, the rule of combination is satisfied. The aggregate of such complex layers forms a crystalline substance called “bruci-

… Mg(OH)\(_2\). The crystals belong to the hexagonal system, as indeed should have been expected on the basis of the arrangement of their constituent atoms.

If, in such an arrangement, the magnesium atoms are replaced by aluminum atoms, a new compound is obtained, called “gibbsite,” or “hydrargillite.” Instead of two electrons, aluminum gives up three electrons, so that the number of aluminum atoms in the indicated compound will amount to only two thirds, in comparison with the number of magnesium atoms in brucite. The corresponding arrangement is shown in Fig. 7. If the arrangement of the atoms in Fig. 7 is compared

Fig. 7. Shown here is the same arrangement of hydroxyl groups as in brucite, but there are fewer aluminum atoms here than magnesium atoms, in the ratio \(2:3\). The crystal is not perfectly hexagonal.

with the arrangement in Fig. 5 (the crosses in Fig. 5 and the black dots in Fig. 7), and horizontal lines are taken as the object of comparison, it will turn out that every third atom of brucite corresponds to an empty site in gibbsite.

The clay known by the name “halloysite” consists of alternately recurring layers of gibbsite and silica. The union of each layer with the neighboring layer must take place according to a definite plan; with regard to this we shall present here several considerations.

In the layer of silica hydrate, the atoms are bound to one another by means of electrical forces that arise through the transfer of electrons from atom to atom. The same kind of bond also exists within the given layer of gibbsite. But between the layers of silica

and the layers of gibbsite such an exchange of electrons does not occur—each layer as a whole is neutral. Consequently, for binding two layers of different types to one another there are no binding forces of the same kind as those that hold the atoms together within the given layer. In what way, then, are these layers attracted to one another?

Silicate and gibbsite layers are neutral, since each of them contains an equal number of positive and negative charges. Nevertheless, lines of force emerge from their surfaces in order to return back at another place. If, near some layer, one draws a geometrical surface in such a way that all the charges are located on one side of it, then this surface will be pierced in one direction by the same number of lines of force as in the opposite direction. The surface will be equivalent to a lattice consisting of electric charges. Roughly, we can imagine such a state of affairs by picturing a system of magnets placed on a board, as indicated in Fig. 8: half of the magnets are oriented in one direction, and the other half in another.

Fig. 8. Magnets placed on two boards \(A\) and \(B\), as shown in the figure. Board \(B\) is fixed, while board \(A\) is suspended on long threads. The magnets alternate according to their polarity. However, the correctness of this figure is not essential; it is only necessary that both systems of magnets be identical. Whatever displacement may occur between the two boards, forces must act between them, and board \(A\) must assume one or another of the possible positions of equilibrium.

If, above this first row, one places some second system of magnets, then a displacement of the lines of force from these two fields will take place. If the bringing together of the two systems is carried out in a certain manner, attraction will occur, but with a different mutual arrangement it will turn into repulsion. If the system denoted by \(A\) is suspended on threads so that it can move only from side to side, then it will occupy one or another of the possible equilibrium positions, to which it will return if it is displaced slightly to the side. If it were possible to carry out an experiment in which systems \(A\) and \(B\) were silicate layers or gibbsite layers or other such

identical or different layers, then the same effect would have to occur. However, in all such cases there must be a certain limitation, which we shall now consider.

In each crystalline layer there exists a certain two-dimensional unit of structure. Thus, for example, in the silica layer it is shown in Fig. 9a by the rectangle \(ABCD\), and in the gibbsite layer in Fig. 9b by the rectangle \(A'B'C'D'\). In the first case the entire layer can be divided into the indicated small rectangles, each of which has one and the same composition and one and the same structure. The same division can also be carried out in the second case. It may happen that \(ABCD\) and \(A'B'C'D'\) prove to be very close to one another in all respects; for our purposes we may assume that they are exactly identical. Suppose now that, in our imaginary experiment, the mutual arrangement of these two layers—the silica layer and the gibbsite layer—is such that the parts within the areas \(ABCD\) of one layer and \(A'B'C'D'\) of the other layer, when they are placed one over the other, are

Fig. 9. These two drawings show: a the arrangement of oxygen atoms in the silica layer, and b the arrangement of hydroxyl groups in the gibbsite layer. The rectangular elementary plane cell is shown by dotted lines and has the same dimensions in both cases. In each case the cell contains six atoms, counting them as whole atoms or in parts.

in a state of equilibrium, or, in other words, upon a lateral displacement of one layer relative to the other they tend to return to the initial position. Then such reasoning may also be applied to the layers as a whole; the effect will be cumulative. Both layers will be in a state of stable equilibrium.

However, if there is some difference in the dimensions of the two lattice units, then in some parts of the interface there must exist a tendency to reverse the displacement that occurs, and in other parts—to strengthen it. The forces will adapt themselves to one another, and, on the whole, the two layers will exert no influence at all

one another. In the model (Fig. 8), the upper layer \(A\) cannot approach the layer \(B\) directly and can be displaced only to the sides. Therefore the analogy given with the actual case would be imperfect. But we can easily imagine what would happen if two molecular layers had completely free motion with respect to each other. They would have to join together and occupy such a position (it would be one of many possible positions) in which the electric fields are adjusted to one another in the best way, or, in other words, their mutual arrangement would possess a minimum of potential energy.

Thus the entire construction depends on the equality of two rectangular lattice cells. These two crystalline layers must form a perfect combination, even if they differ slightly from one another in isolation. It is not difficult to imagine that such a difference can quite well be overcome, and a strong bond will occur, for example, when one of the two layers is laid down exactly, i.e. atom by atom, upon the other, or if both of them grow together. Evidence that this actually can occur will now be presented.

The side lengths of the rectangle in each case are almost exactly \(5.1\) and \(8.8\) Å. The ratio of these quantities is \(1:\sqrt{3}\), as is also evident from Fig. 9. The short side is equal in length to twice the diameter of an oxygen atom, while the long side is three times greater than the diameter of the hydroxyl group; i.e. these diameters are respectively \(2.55\) and \(2.93\) Å. In the case of gibbsite these quantities are precisely determined by means of X-rays\(^{4}\). In the case of silicic acid hydrate layers they have not been observed directly, but they can be inferred partly on the basis of the dimensions of quartz structures, and partly from X-ray determinations of the structures of clays themselves, micas, talc, and other crystals in which silica layers are found.

The structure of galuazite crystals was determined by Memlem\(^{5}\). The dimensions of the unit cell here are: \(a = 5.20\) Å, \(b = 8.92\) Å, \(c = 10.25\) Å, \(\beta = 100^\circ\). The quantities \(a\) and \(b\) correspond almost exactly to the separate dimensions of the gibbsite layer. The magnitude \(c\) has the order that could have been expected on the assumption that four oxygen or hydroxyl layers are arranged one above another, two of which belong to the silica and two to the gibbsite. X-ray measurements show that the thickness of the latter is \(4.86\); consequently, the thickness of the former must be approximately the same. The complete data, with an enumeration of the component parts of the unit cells, may be obtained from consideration of Table 3. The lowest layer in the unit cell contains six oxygen atoms within the rectangle \(a \times b\), as is seen from Fig. 9a. In the next layer there are four silicon atoms for each cell, followed by a layer of four hydroxyl groups. Above the latter lie six

TABLE 3

Arrangement of layers of atoms in clays and related substances. The elementary cell of the crystal consists of layers of atoms stacked one upon another; each layer in such a cell has the form of a rectangle measuring 5.2·8.8 Å. The height of the cell depends chiefly on the number of layers of oxygen atoms and hydroxyl groups. The numbers assigned to the layers in the table indicate the number of atoms or hydroxyl groups in each layer. In addition, approximate values are given for the thickness of each group of layers, i.e., from the beginning of one crystal cell to the beginning of the next.

hydroxyl groups, four aluminum atoms and, finally, again six hydroxyl groups. The composition of the elementary cell may be described by the formula Al$_4$Si$_4$O$_{22}$H$_{10}$. Otherwise this composition may also be denoted by the formula 2Al$_2$O$_3$4SiO$_2$8H$_2$O, but it should not be forgotten that although by this latter method the content is very conveniently described with the aid of familiar molecular forms, these latter nevertheless lose their independent existence in the crystalline structure.

When halloysite is slightly heated, its two middle layers of

hydroxyl groups merge into a single layer consisting of four oxygen atoms and two hydroxyl groups. Four oxygen atoms and eight hydrogen atoms are thereby released, so that we may describe this process as a loss of water, although before this no definite molecules of water existed in halloysite. In the new form of the crystal, the so-called metahalloysite, the oxygen atoms in the middle layer lie above the silicon atoms, while the hydroxyl groups in the same layer lie in the cavities above the lower layers of silicon and oxygen. It is easy to see that the electronic relations are thereby satisfied.

The structure of metahalloysite is of interest in that it differs only very little from the well-known structure of kaolinite, or china clay. X-ray methods show that the composition and arrangement of the layers are the same, but on the X-ray photographs some slight differences are also observed,

Nakrite

Dickite

Kaolinite

Halloysite

Montmorillonite

Fig. 10. X-ray photographs of several types of clays, obtained by Nagelschmidt

differences which may be due to some displacement of the layers relative to one another, or to a slight displacement of the atoms within each layer.

Nakrite and dickite are two other forms of clay, very closely related to kaolinite and differing from it, in all probability, in the same way as metahalloysite. A series of X-ray photographs of dif-

ON THE STRUCTURE AND PROPERTIES OF CLAYS

of various clays is shown in Fig. 10. Many features of similarity between them are quite obvious. There are, however, certain differences: in the relative intensities of individual lines, which still require their explanation.

Let us now turn to the second and only other group of clays. Chemical analysis shows that, generally speaking, the ratios of the amount of silicon to aluminum in this second group are considerably higher than in the first. There are, however, also considerable differences in the composition of the clays of the second group, and so great that their differentiation and the distinction of individual members of the group constitute a difficult problem. This problem has been solved only in that part of it which concerns the main features of the structure and properties of the clays.

A typical feature of the clays of the second group is the fact that in them there are two layers of silica for each layer of the other kind—brucite or gibbsite, or their modifications; for every two layers of tetrahedra of oxygen atoms or hydroxyl groups there is one layer of octahedra of the same components. It is assumed that the latter layer lies between two layers of the first type,^2,6 so that this arrangement is symmetrical and nonpolar. Evidence on this point is currently being assembled and is already stringent. The arrangement of the layers is shown in Table 3. Pyrophyllite may be regarded as a typical representative of this group, to which montmorillonite, beidellite, fuller’s earth, and other varieties of clays also belong. Micas, talc, clintonites, and other crystals of this kind, although in their properties they differ from clays, represent extreme varieties of this second group.

Since the individual layers here are the same as in the first group, and the difference between the two groups consists only in the order and in the relative proportions of these layers, in their combination with one another, it is natural to expect that the X-ray diagrams should be identical in many respects. This is also evident from Fig. 10. All lines arising from planes containing the \(c\)-axis of the crystal, i.e., planes which depend on the shape and dimensions of the lattice cell in the layer, must be the same everywhere. The main difference lies in the inequality of the interplanar distances along the \(c\)-axis, which directly depends on the number of layers; they are approximately proportional to the number of layers of oxygen atoms or hydroxyl groups. It is precisely for this reason that the length of the \(c\)-axis is used to calculate the number of layers.

The differences between the members of the pyrophyllite group consist in their different metal contents. It would be more convenient to study them in connection with the important phenomenon of base exchange, and therefore we shall dwell here on one variation which is of immediate interest. When gibbsite is replaced by brucite, we have talc. It would be natural to ask why there is no such parallelism in the first group as well. Why, then, are there no variations of kaolinite in which such a replacement would take place. Paulin-

This was given an interesting explanation. The unit rectangle of the magnesian crystal of brucite differs considerably from the unit rectangle of the silica layer, having dimensions approximately 8% greater than the corresponding dimensions in the gibbsite crystal. If the silica layers lie on both sides of the brucite, as in talc, then the bond may prove to be very strong; but this cannot occur in a structure of the kaolinite type, where there is only one layer of each type. Such a conclusion is in agreement with the hypothesis that the tetrahedral and octahedral layers in pyrophyllite are symmetrical in their arrangement.

Now we must consider whether, on the basis of these structural data, it is possible to give an explanation of the well-known properties of clays. Of course, the details of these structures as a whole have not yet been fully elucidated, and our knowledge in this respect is still far from perfection. But the general features of the structure are nevertheless so clear that attempts to answer the questions posed would be quite appropriate, insofar as this is, of course, possible.

Let us consider first of all the well-known property consisting in the fact that clay particles are in many respects similar to many other colloids, namely in that they possess electric charges. In the case of clays these charges are almost always negative. This fact can easily be made the subject of a lecture demonstration. Two electrodes made of platinum foil are immersed in a narrow test tube placed before a source of scattered light, and when they are viewed on a screen they appear as two thin lines. The test tube is filled with a suspension of very finely divided kaolin particles, and a certain potential is applied to the electrodes. After 1–2 minutes it becomes perfectly obvious that the space near the negative electrode is being cleared of suspension particles.

A clay particle is normally neutral. But, as I have already indicated, its surface, or each layer drawn near it in view of the enveloping surface, is twice intersected by all the lines of force. The lines enter, bend, and come out again, and the points of their intersection with the surface will act as electric charges. This picture must change when other fields approach, producing as a result a superposition of both fields. The atoms of the crystal, owing to which this field arises, may change their position, but on the whole they form a connected system, so that such movements are very small (if these displacements are so large that the atoms are all displaced together, then we have “chemical action”) and have only a secondary influence. The matter proceeds as when one magnet is brought near another, when a new complex system of lines arises from the two systems of lines; but if the magnets are separated again, their individual fields are restored in their former form.

Such a crystal surface exerts many local attractions and repulsions on charges approaching it. If ions approach, i.e. charged atoms or molecules

in the surrounding liquid, then positive ions will be attracted to the negative places on the surface, and vice versa. If ions of both kinds are attracted, both positive and negative, and if their quantities are equal, then the surface remains unchanged. But such equality is not to be expected, since positive and negative ions differ both in the form of their fields and in their mobility. One of them adapts itself better to the surface field and has greater chances of doing so than the other. It is evident that the oxygen surface of a clay particle retains negative hydroxyl groups more firmly than equivalent positive ions, and thus becomes negatively charged. The positive ions remain lying freely in the surrounding liquid. The electric field of the hydroxyl group must differ greatly from the field of a positive ion of the monatomic type, since the former is dipolar and therefore can adapt itself especially well to the oxygen surface.

From this point of view the charge of a colloidal particle is determined chiefly by structural conditions. It may be either positive or negative, and its sign in each individual case depends partly on the form of the surface field, and partly on the properties of the ions which, under the action of Brownian motion, constantly wash over the surface of the crystal. Thus a clay particle, laden with negatively charged hydroxyl groups, repels other similar particles, so that their suspension in the surrounding liquid remains permanent. We know, however, that the repulsion can be eliminated by adding known ions to the liquid. Thus, for example, a small amount of hydrochloric acid entails immediate coagulation of the plastic mass used by the potter. The negative charges of the ions located on the surface of the clay particles are compensated by positively charged hydrogen ions, which are deposited upon them, and the particles no longer repel one another. A comparatively small amount of acid is required because the number of atoms deposited is small in comparison with the number of atoms in the particle itself, just as the weight of paint required to cover a house is very small in comparison with the weight of the house.

One of the most remarkable properties of clay is the ease with which water can be added to it or removed from it, with the changes taking place within the clay itself. Of course, water can be added and removed in the suspension of clay particles itself as well, but the addition and removal of water considered here are of a structural character. This property, especially well manifested in montmorillonite, was studied specially by Nagelschmidt[^7], Hofmann, and other authors. As indicated in Table 3, pyrophyllite contains four hydrogen atoms in each elementary cell; they form part of the principal structure and cannot be removed without destroying it. It may be said that the cell

contains two molecules of water, but again it should be remembered that these four hydrogen atoms and the two corresponding oxygen atoms are not bound in the way they would be in separate water molecules.

The structure of pyrophyllite is the same as the structure of montmorillonite from which all the water has been expelled, except, of course, for those hydrogen and oxygen atoms which form water only when the structure is destroyed. The length of the \(c\) axis is \(9.4\) Å, a value which, as we have already seen, corresponds to a column of four oxygen or hydroxyl layers arranged one above another and centered by atoms of silicon and aluminum.

When water is added to this structure, the \(c\) axis increases in length. Nagelschmidt’s measurements, made in 1936, were later repeated by Bradley, Grim, and Clark\(^8\), who found that there exists a whole series of hydrates in which the water content in the cell increases successively to 8, 14, 20, and 26 molecules, and that the \(c\) axis grows in length correspondingly from \(9.6\) to \(12.4\), \(15.4\), \(18.4\), and \(21.4\) Å. This means that layers of water molecules are added to the layers in each crystalline cell. The mean increase is in agreement with what might be expected from this hypothesis, since the additional volume of one layer is \(3 \cdot 5.2 \cdot 8.9 = 138.5\) Å, and the additional weight is \(6 \cdot 18 \cdot 1.66 = 179.5\) corresponding units. Hence for the density of the layer of added water one obtains approximately \(\frac{179.5}{138.5} = 1.3\). The size of the water molecule is known from many measurements in crystals. From these data it can be found that such molecules, in closest packing, should give a density one and a half times greater than that of ordinary water. The question of the structure of water apparently remains open for the time being. According to Bernal and Fowler\(^9\), it tends to assume a quartz-like character; groups of molecules acquire a temporarily ordered arrangement and then tend to break down again. It must therefore be considered that the density of the layer is in agreement with the idea of an arrangement of water molecules packed more closely than in water itself, but not to the degree that might have been possible.

If, therefore, the water molecules are bound to the clay particle, they cease to belong wholly to the liquid, even in the case where the oxygen layer on which they are situated forms the outer surface of the crystal. The clay adsorbs or absorbs them, depending on whether the attachment occurs on the surface of the clay particle or in the spaces between the layers of oxygen and silica. The loose character of their bond is the reason for the ease with which clay particles slide relative to one another. But clay cannot become wet until the amount of water retained by it reaches a certain limit. It then becomes plastic, without thereby reaching a muddy state. The relative looseness of the bonds is also the reason for the easy removal of water by means of

evaporation heat. The amount of hydration water is in equilibrium with the moisture content in the surrounding atmosphere.

Thus, the ability of clay to absorb and release comparatively large amounts of water is a consequence of its special structure. Layers of water molecules can successively penetrate into the layered structure of clays. Thanks to them, an increase in the length of the \(c\)-axis takes place, but no changes occur in the length of the other axes. This feature is also the cause of plasticity.

We must now consider one of the most remarkable properties of certain clay substances, consisting in the “exchange of bases.” It is of enormous interest to agrochemists. A whole series of other crystalline substances, for example zeolites, also possess this property. These crystals contain, in definite proportions, metallic ions of sodium, potassium, calcium, and others, whose presence is connected with the special features of the structure of these crystals, which we shall now examine. Since the named atoms are positively charged, they may be regarded as “bases” bound to the negative residue of the crystal. In certain cases they can replace one another, whence the name of this process arises.

Zeolitic crystals are well formed and easily accessible to treatment, owing to which their structures can be studied by means of X-ray methods. The structure of the first member of this group—analcite—was studied in considerable detail by Taylor\(^{10}\), and thanks to this the process of exchange of bases has now become clear. The structures of clays do not lend themselves to such easy study, but the process, of course, proceeds in approximately the same way.

A special feature in the structure of these crystals, on which the process of exchange of bases depends, consists in the replacement of silicon atoms in the oxygen tetrahedra by aluminum atoms; other substitutions may also occur, but they are all less important. An aluminum atom can give up only three electrons, whereas the silicon atom that replaces it gives up four. Consequently, when such a replacement takes place, one electron must come from somewhere. It may be borrowed from an external atom of sodium or potassium, or from calcium. Such an external atom need not necessarily enter into the main structure of the mutually connected tetrahedra. Its removal may make the structure negatively charged, but the general character of the structure will not thereby be changed. In some cases it is also very easy to replace compensating atoms of one kind by others.

However, not all structures possessing such compensating atoms exhibit the process of exchange of bases. It appears necessary that, for this, there should also exist in the structure certain channels through which the ions under consideration could penetrate. In the case of analcite, as was shown by Taylor, these channels pass through the crystal, through which, therefore, the journey of ions can take place. In these channels there was found a certain

a certain number of water molecules, because of which the substance may contain them or be deprived of them, as in the case of clays.

In the class of silicates to which clays belong, the indicated replacement of aluminum by silicon also occurs. Here, too, compensation is brought about by additions of sodium, potassium, or other atoms capable of giving up electrons. For example, in the structure of mica such a compensating element is potassium, as is shown in Table 3. In this case one should expect that the potassium will be so strongly bound to the rest of the structure that it will be difficult to remove. Potassium atoms cement the layers together and make of them a strong crystal. Yet they are not so strong as to prevent the easy cleavage that is characteristic of mica. When the potassium atoms are replaced by calcium, as in the “clintonites,” the bonds become stronger, and such easy cleavage becomes more impossible.

In montmorillonites, however, exchange of bases is very easily effected. We may suppose that the half-liquid layers of water present in montmorillonite allow the compensating atoms to possess a certain freedom of movement. They perform a function analogous to that characteristic of the channels of analcime. It is precisely in the montmorillonite group that exchange of bases is quite usual and at the same time has such great significance.

To summarize, it may be said that the structures which have now become known thanks to X-ray investigations provide a plausible explanation of the properties of clays. True, many details still await clarification, since it is still difficult to obtain completely pure material for investigation, and since such material, even if homogeneous in its composition, may prove to be very complex as a result of the substitutions of which we have just spoken and which almost always occur in it. Nevertheless, the general character of the solution of this problem has already become clear, and fuller knowledge will, of course, now begin to come rapidly from the many laboratories in which the problem of clay is being examined.

LITERATURE

1. V. L. Bragg, The Crystalline State, Vol. I, ONTI, p. 129 ff.
2. L. Pauling, Proc. Nat. Acad. Sci., 16, 123, 453, 578, 1930.
3. V. M. Goldschmidt, J. Chem. Soc., 5, 656, 1937.
4. Megaw, Z. Krist., 87, 185, 1934.
5. M. Mehmel, Z. Krist., 90, 35, 1935.
6. J. W. Gruner, Z. Krist., 88, 412, 1934.
7. G. Nagelschmidt, Z. Krist., 93, 481, 1936.
8. G. L. Clark, R. E. Grim and W. F. Bradley, Z. Krist., 97, 216, 1937.
9. Bernal and Fowler, J. Chem-Phys., 1, 515, 1933.
10. W. H. Taylor, Z. Krist., 74, 1, 1930.

1. Clay, W. Bragg, Proc. Roy. Inst. Gt. Brit., Nov. 19th, 1937. Translated by N. A. Shishakov. ↩↩↩

2. These lectures were published as a separate book: W. H. Bragg, Old Trades and New Knowledge, G. Bell, London, 1926. The book was translated into Russian and published in the series “Nature and Culture” (Gosizdat). ↩↩