ORIENTATION OF MOLECULES AT THE PHASE BOUNDARY*

G. Freundlich, Berlin

I. Introduction. II. Results of capillary-chemical investigations of layers of insoluble substances. III. Results of capillary-chemical investigations of layers of soluble substances. IV. Comparison of the results of investigations obtained by methods of capillary chemistry with other investigations. V. Experiments with thin layers obtained from colloidal solutions. VI. Experiments at interfaces of another kind.

I. Introduction

1. In the first five-year periods of the present century, the conviction took shape that gases and dissolved substances at the phase boundary have a concentration different from that within the phase. Thus it followed that adsorption phenomena had to be taken into account, and at the same time the question remained unresolved whether these phenomena should be explained by the Gibbs law, expressing the dependence of concentration on surface tension, or regarded as the consequence of a chemical combination due to supplementary valences. Nothing was said about the arrangement of the molecules; for the most part it was assumed that they were arranged at random, just as within the phase. It was a great advance when attempts began to be made to examine more precisely the question of the orientation of molecules and, in many cases, when it proved possible to establish it. For this we are indebted above all to Langmuir¹. Even earlier Hardy² had pointed to such a possibility, and, at the same time as Langmuir, Harkins³ drew important conclusions from this idea.

The study of the properties of thin oil films on the surface of water led to the question of oriented molecular layers. These films were investigated with particular success by Rayleigh⁴, Pockels⁵, and Devaux⁶. Pockels developed a method by means of which it was possible easily to obtain these films and to vary their thickness. For this purpose a rather narrow flat rectangular trough is taken, which, as far as possible, is filled completely with water. On the surface of the water there is placed a very small drop of oil or of a solution of oil in a volatile solvent, insoluble in water, such as, for example, ligroin or benzene; after its evaporation, there remains on the surface of the water

* Article in volume 12 of Ergebnisse der exakten Naturwissenschaften. Translation and notes by A. S. Akhmatov.

extremely thin layer of oil. The magnitude of the oil-covered surface depends on how far the drop has spread over the surface of the water. It can be varied at will by means of narrow, well-fitted, and, if possible, non-wettable strips of paraffin-coated glass, which are moved normal to the long side of the cuvette and thus make it possible to decrease or increase the oil-contaminated portion of the surface. For oils insoluble in water, the thickness of the film is readily calculated from the weight of the drop placed on the surface and the size of the area.

Rayleigh and Pockels investigated surfaces that were covered with layers of castor oil of various thicknesses. The magnitude of the surface tension could be measured without disturbing the arrangement by determining the weight of the weights required to detach a circular plate from the surface.

Fig. 1. Change in the surface tension of water by thin layers of oil.

to detach a circular plate from the surface. The dependence of the surface tension \(\sigma\) (in dyn/cm) on the thickness of the oil layer \(\tau\) (in Å) is shown in Fig. 1.

So long as the amount of substance introduced onto the surface is small—and the layer is therefore very thin—the surface tension changes imperceptibly (region \(AB\) in Fig. 1); beginning with a certain definite thickness \(\tau_1\), corresponding to point \(B\) of the curve, the surface tension decreases sharply, approaching, at the larger value \(\tau_2\), a certain limiting value; this limiting value of the surface tension corresponds to the surface tension of pure oil covering the surface of the water in a thick layer. The thickness \(\tau_1\), from these measurements, proved to fluctuate from \(10\ \text{Å}\) to \(40\ \text{Å}\) and more. For high-molecular substances, such as oils, a thickness of \(10\ \text{Å}\) corresponds approximately to the transverse dimension of a molecule, whereas a layer of thickness \(\tau_2\) corresponds to many layers of molecules. A very important consequence of these experiments\(^7\) was the demonstra-

evidence for the existence of monomolecular layers; these latter possess properties that can readily be studied experimentally.

II. Results of capillary-chemical investigations of layers of insoluble substances

2. Two measures contributed to the success of Langmuir’s work. First, instead of oils he subjected to investigation organic substances well studied with respect to their chemical structure (higher alcohols, fatty acids, etc.); in doing so he

Fig. 2. Langmuir’s balance.

studied a very large number of similar compounds; moreover, Langmuir did not confine himself merely to measurements of the surface tension of a surface contaminated by the given substance, but constructed an apparatus that made it possible to measure directly the difference of surface tensions between a clean surface and a surface bearing the substance under investigation, the said measurements being made while the thickness of the surface layer of the organic substance was varied arbitrarily.

These Langmuir balances, as such a two-dimensional differential manometer is often called (Fig. 2), consist of a rectangular trough filled to the brim with water or with some aqueous solution; over the surface of the water two partitions in the form of paraffined glass strips can be moved. One of them, \(B\), is movable and is rigidly connected with the arm of the balance beam; the second, \(A\), rests on the edges of the trough. On the surface of the liquid between these partitions a layer of the organic substance is obtained. The pulling force with which the surface tension of the clean surface \(H\) behind

the partition \(B\) acting on the surface bearing a thin layer of the substance under investigation can be compensated by placing a load of the proper magnitude on the pan of the balance.

It proved very useful to regard the difference in surface tensions not as a tensile force, but as a pressure which the molecules of the organic substance exert on the partition \(B\). This pressure may be regarded as osmotic in two-dimensional space. The partition \(B\) is, so to speak, semipermeable: permeable to water molecules, which can penetrate under the partition, and impermeable to the organic substance, which is insoluble in water.

In order that the difference in surface tensions should remain constant, i.e. in order that the organic substance should not seep through the slit \(D\) between the partition \(B\) and the edges of the trough, a sufficiently strong current of air was driven through the tubes \(F\) and \(F_1\). By moving the partition \(A\) one could vary the thickness of the layer on the surface, and thereby also the force acting on the partition \(B\). The scale \(M\) made it possible to measure the magnitude of the surface.

Knowing the magnitude of the surface between the partitions and the edges of the trough, as well as the weight of the organic substance placed on the surface of the liquid, one can calculate the area attributable to each molecule. In addition, from the weight of the load placed on the pan of the balance, one knows the “expansion pressure,” which the surface layer under the given conditions exerts on the partition \(B\). In the figure below (Fig. 3) is shown the dependence between the pressure (Ausdehnungsdruck) \(s = \sigma_M - \sigma_L\), in dyn/cm, and the area occupied by each molecule \(\omega\) (in \(\text{Å}^2\)); \(\sigma_M\) denotes the surface tension of the clean surface, \(\sigma_L\) the surface tension of the surface contaminated by the substance under investigation.

For many substances comparatively simple and characteristic curves were obtained. So long as a relatively large area falls to each molecule of the organic substance, \(s\) is small. In this region \(AB\) of the curve, \(\sigma_M\) and \(\sigma_L\) differ little; this region corresponds to the region \(AB\) of the curve in Fig. 1. The layer can be strongly compressed without increasing the load. This state of the layer changes abruptly when, upon further displacement of the partition, point \(B\) is reached. Now the layer already offers appreciable resistance to compression. \(S\) can be increased from \(B\) to \(C\) with a very small change in \(\omega\). Only under the action of a more considerable pressure is the layer compressed and finally occupies a smaller area—\(CD\). From the foregoing it follows that at point \(B\) the thickness of the layer corresponds to the thickness of a monomolecular layer. Moreover, from the character of the curve shown in Fig. 3 one may conclude that the molecules in the monomolecular layer are so closely pressed against one another that they no longer have the possibility of moving in the plane. Hence the considerable resistance to compression and the invariability of \(\omega_1\). This quantity can easily be obtained by extrapolating

of the rectilinear portion of the \(CB\)-curve. In the region \(CD\) the layer is disrupted, i.e., is no longer strictly monomolecular—numerous overlayers have formed. This latter circumstance was confirmed with the aid of the ultramicroscope on films of palmitic acid.⁸

A sharply expressed state, characterized by the value \(\omega_1\), called condensed, was described still more fully by comparing various organic substances with one another. It turned out that substances with very different molecular weights and different chain lengths gave identical curves and practically identical \(\omega_1\). Table 1, compiled from Langmuir’s first measurements, illustrates this.

The substances listed possess one common property: they are all polar; their molecules have one pronounced hydrophilic end (the \(OH\) of alcohols, the \(COOH\) of acids) and the other hydrophobic, in the form of the group \(CH_3\). This led Langmuir to the idea that the molecules compressed into a monomolecular layer are oriented; the hydrophilic groups are turned toward the water, and the hydrophobic groups outward—toward the gas phase. Since these groups are the same in all the substances, then, consequently, despite the difference in the structure of the molecules and in the length of their chains, they occupy identical areas on the surface, equal to \(\omega_1\).

Fig. 3. Simple form of the \(S-\omega\) curve.

1. On the basis of these experiments two main conclusions could be drawn: first, the necessity of always taking into account the possibility of orientation of molecules at the phase boundary, and second, thanks to Langmuir’s work, an apparatus was constructed with the aid of which a quantitative study of the state of a substance in two-dimensional space became possible. For the study of the condensed state, Langmuir balances, operating with accu—

TABLE 1

Name of substance	Formula	\(\omega_1\)
Myricyl alcohol	\(C_{30}H_{61}OH\)	27
Palmitic acid	\(C_{15}H_{31}COOH\)	21
Stearic acid	\(C_{17}H_{35}COOH\)	22
Cerotic acid	\(C_{25}H_{51}COOH\)	25
Cetyl palmitate	\(C_{15}H_{31}COO — C_{16}H_{33}\)	23

...with an accuracy of up to \(0.5\ \mathrm{dyn/cm}\), are sufficient. For studying the dependence cited above over its entire extent (Fig. 3), and especially the segment of the curve \(AB\), a more sensitive apparatus with an accuracy of up to \(0.01\ \mathrm{dyn/cm}\) is required. Mainly Adam\(^{9}\), and also Rideal\(^{10}\), Marcelin\(^{11}\), and Langmuir\(^{12}\), were able to accomplish this by means of a number of measures. Blowing air into the slit \(D\) proved too crude a method; it is much better to place in the slit between the partition and the walls of the cuvette two platinum strips, completely free, as thin as possible—for example, \(2\)–\(3\ \mathrm{mm}\) wide and \(0.003\ \mathrm{mm}\) thick. In addition, as a compensating force pressure, the above-mentioned authors used the elastic resistance of a twisted thread*.

As often happens in the detailed study of a new phenomenon, it turned out in further investigations that the phenomena described above are more varied than had previously been supposed. In what follows we shall consider some characteristic cases, and first of all we shall have in mind only substances that are high-molecular and insoluble in water and in aqueous solutions. We shall first deal with certain features of the condensed state.

Adam’s more precise measurements gave, for a large number of substances in the condensed state, \(\omega_1 = 21\ \mathrm{\AA}^2\) \((21\cdot 10^{-16}\ \mathrm{cm}^2)\). These were saturated fatty acids; among them the acid with the shortest chain which still gave, at room temperature, the same value \(\omega_1\), was tridecanoic acid (\(\mathrm{C}_{12}\mathrm{H}_{25}\mathrm{COOH}\)); as an example of an acid with a very long chain which, for the simple course of the curve similar to that shown in Fig. 3, gave the same \(\omega_1\), we may name lignoceric acid (\(\mathrm{C}_{23}\mathrm{H}_{47}\mathrm{COOH}\)). Further, high-molecular amides with the terminal group—the “head”—in the form of the \(\mathrm{CONH}_2\) group; methyl ketones with the “head” \(\mathrm{COH}_3\), and dibasic esters with the group \(\mathrm{COOC}_2\mathrm{H}_5\) at both ends. Here also

* Platinum may be replaced by tinfoil or, for example, aluminum. The method of blocking the surface film in the slits between the walls of the cuvette and the floating partition has an essential significance for the sensitivity of the apparatus. It has recently been proposed to use, for this purpose, silk threads floating on the surface of the water and lightly greased with vaseline (I. Guastalla, Comptes r. d. l’Acad. 189, 241, 1929).

* The apparatus of Adam and of Marcelin, which are torsion microbalances, are very close to one another. Adam’s apparatus, however, is considerably more complex in construction: the twisting of a horizontally arranged thread and the motion of the float are connected by means of a special, finely constructed armature capable of rotating a mirror used for readings by the zero method. In Marcelin’s apparatus, a light transverse bar is attached to a thread stretched vertically, and this transmits the elastic twisting force to the mica float. Both of the instruments mentioned (for the cited literature, see A. Marcelin, “Oberflächenlösungen,” pp. 205–220, Dresden 1933) are unsuitable for simultaneous measurements of condensation pressure and of the gas film, and require a considerable amount of time for this purpose. Recently, a new method for studying thin layers has been proposed, based on the study of the damping of capillary waves on the surface of a liquid (Gorter and Seeger, Koll. Z. 58*, 257, 1932).

also include triglyceride chains, for which in the maximally condensed state each molecule likewise accounts for \(21\ \text{Å}^2\). A substance characterized by four parallel chains is pentaerythritol tetrapalmitate:

\[ \begin{array}{c} \mathrm{CH_2OCO\cdot C_{15}H_{31}}\\ \mathrm{CH_2OCOC_{15}H_{31}}\\ \mathrm{C\!-\!CH_2OCOC_{15}H_{31}}\\ \mathrm{CH_2OCOC_{15}H_{31}} \end{array} \]

For it the \(S-\omega\) curve shows a steep rise already at \(100\ \text{Å}^2\); at a pressure of approximately \(20\ \text{dyn}/\text{cm}\) the curve becomes still steeper and, by extrapolation, makes it possible to find \(\omega_1 = 80\ \text{Å}^2\), which corresponds to the quadrupled value of \(\omega_1\) given above.

For many other substances the course of the curves is obtained in the same way; only the value of \(\omega_1\) differs. Sometimes, however, the curves run more gently. As examples we shall cite: nitriles—the “head” is \(\mathrm{C \equiv N}\), \(\omega_1 = 28\ \text{Å}^2\); phenols and other para-substituted benzene derivatives with heads:

\[ \begin{array}{ccc} \begin{array}{c} |\\[-2pt] \hexagon\\[-2pt] \mathrm{OH} \end{array} & \begin{array}{c} |\\[-2pt] \hexagon\\[-2pt] \mathrm{OCH_3} \end{array} & \begin{array}{c} |\\[-2pt] \hexagon\\[-2pt] \mathrm{NH_3} \end{array} \end{array} \]

For them \(\omega_1 = 24\ \text{Å}^2\). Cholesterol and some of its simpler derivatives also belong to this group (Adam and Rosenheim); for them \(\omega_1\) is equal to \(41\ \text{Å}^2\).

According to Adam\(^{13}\), it is very probable that in the condensed state corresponding to \(21\ \text{Å}^2\), it is the chains themselves that are displaced to the maximum, independently of the size of the head; whereas at larger values of \(\omega_1\) \((>21\ \text{Å}^2)\), the size of the heads does not permit closer approach; in this latter case we therefore have the maximum possible approach of the heads of the molecules. In Rideal’s opinion\(^{14}\), one should take into account the degree of inclination of the chains and the possibility of their mutual overlap.

In many cases, when a layer is compressed, the condensed state is not reached at once, and it is possible to distinguish many states of different degrees of condensation. As an example let us take iso-oleic acid, the film of which was obtained on the surface of a dilute hydrochloric acid solution (Fig. 4).

The region of greatest condensation \(BC\) again corresponds to \(\omega_1 = 21\ \text{Å}^2\). In addition, there is a rectilinear segment \(BB_1\), which gives an area per molecule equal to \(29\ \text{Å}^2\). It is considered that the latter corresponds to close contact of the heads,

whereas at higher pressure the heads of the molecules are arranged in such a way that the chains come into closest contact. Ethers and alcohols behave similarly to what has been described. Ethers, except for \(\omega_1 = 21\ \text{Å}^2\), also give a value \(\omega_2\) somewhat greater than \(22\ \text{Å}^2\); cetyl alcohol likewise has \(\omega_2\) slightly less than \(22\ \text{Å}^2\).

The laws of compression of thin layers depend very much on a number of influences. Thus, for fatty acids on the surface of distilled water, if it has only just been poured into the cuvette (on the surface of freshly poured water as yet unexplained hysteresis phenomena have been observed[^15]), the curve of Fig. 3 is obtained well, whereas on the surface of a dilute HCl solution the curve of Fig. 4 is obtained. These phenomena[^16] are still difficult to explain.* Lyons and Rideal[^17] consider that on the surface of pure water fatty acids, being partly dissociated, are more deeply immersed by their heads in the water, and therefore the maximum packing is observed at once; on the surface of an acid solution, owing to the decrease in dissociation, the molecules are less deeply immersed and therefore, upon compression, packing of only the heads is observed.

In order to decide whether the film is a liquid or a solid, one observes whether small particles of talc and similar powders scattered over its surface are displaced by a slight blowing. One may also, following Fölmer and Kessler, observe how readily small iron particles placed on the surface of the film move in a magnetic field. Whereas some investigators consider the solution of such a question easy and assume that the melting point of a film is practically equal to the melting point of the same substance in bulk, other authors believe that in many cases this question is not easily decided and that it is difficult to conclude whether the film is really solid or whether it is similar to a soft, plastic mass. It is possible that here the circumstance is manifested that these layers are similar to layers of a mesomorphic phase, i.e. the molecules in solid films are not oriented as strictly as is the case, for example, in a thin crystalline platelet, but are arranged next to one another more disorderly.

Allotropic transformations are often connected with characteristic changes in the magnitude of \(\omega_1\). Thus, octadecylacetamide \((\mathrm{C_8H_{17}NHCOCH_3})\) below \(17^\circ\text{C}\) gives the normal value \(\omega_1 = 21\ \text{Å}^2\)—the layer is only slightly compressible and, according to the talc test, must be recognized as solid. Above \(17^\circ\), \(\omega_1 = 24\ \text{Å}^2\)—the layer is strongly compressible and behaves

* Bresler and Talmud, on the basis of a number of experiments they carried out with various electrolytes, as well as with glucose, attribute the influence of the “substrate” to dehydration of the polar groups of the molecules of the surface layer (S. Bresler u. D. Talmud, Koll. Z. 63, 323, 1933).

as a liquid. Hexodecylacetamide behaves in the same way, with the only difference that the characteristic temperature point lies at \(9^\circ\). High-molecular derivatives of urea, for example \(C_{20}H_{41}NHCONH_2\)—eicosylurea, show the opposite relations: at low temperatures, about \(25^\circ\), \(\omega_1 = 26\ \text{Å}^2\), at \(35^\circ\), \(\omega_1 = 21\ \text{Å}^2\). The interval \(30\)—\(35^\circ\) is the transition region, accompanied by sharp hysteresis phenomena. For such substances the temperature point of transition falls with the length of the chain. Films at both higher and lower temperatures are only slightly compressible. Adam explains these phenomena by the fact that, on heating, the heads of the molecules are capable of becoming deformed and thus allowing maximum packing of the molecules.

Cases of hysteresis in these phenomena are in general rare; the curves, both in stretching and in compression, pass well along one and the same path. An exceptional case of hysteresis has been described for dodecylphenol \((C_{12}H_{25}C_6H_4OH)\): on compression \(S\) rapidly increases and only after several minutes reaches a lower value; when the surface \(S\) is increased, it drops sharply and then rises again to a certain limiting value.

Fig. 4. Complex form of the \(S\)—\(\omega\) curve.

1. Having considered the features of the condensed state, we shall now pass to the second type of states of thin layers, which is observed when it is possible for the substance under investigation to occupy the maximum possible area on the surface of a liquid. This state on the curve of Fig. 3 is expressed by its right-hand part, beginning at point \(A\) and beyond. Measurements in this region are possible, of course, only with an especially sensitive method, for example Adam’s. It turned out that the behavior of the molecules under these conditions obeys surprisingly simple laws. If the surface is sufficiently large, it may be assumed that the organic molecules can move independently of one another. As has already been noted, the expansion pressure of the layer, which is experienced by the movable frame, may be regarded as osmotic pressure on a semipermeable wall in two-dimensional space. For the relation between the surface concentration of the molecules \(\Omega\) (per 1 mole) and the pressure \(S\), an equation was obtained which fully corresponds to the laws of ideal gases, as well as to van ’t Hoff’s law for dilute solutions. This is:

\[ S\Omega = RT. \tag{1} \]

Here \(\Omega\) is the area occupied by one mole of the substance; \(R\) and \(T\) have their usual meanings. It also follows from this equation that the lowering of the surface tension by the organic substance

proportional to its concentration. Traube^18 had already earlier found this dependence for truly dissolved, capillary-active organic substances, and pointed to features of similarity with Boyle’s law, while Devaux^19 emphasized the possibility of studying force interactions between molecules by means of investigating thin layers of insoluble substances. Langmuir,^1 however, was the first to derive equation (1) and to investigate it on adsorption layers of truly dissolved substances. If \(S\) is calculated in dyn/cm, and \(\Omega\) not per mole, but, as we have done up to now, per one molecule in \(\text{\AA}^2\), then \(S \cdot \omega\) for room temperature should have the value \(400\) ergs.*

Adam^20 found cases for which this gas law is fulfilled almost exactly. Measurements carried out by Marcelin and Delaplace^21 created the impression that, although equation (1) is applicable, \(R\) is much smaller than the gas constant. According to Adam and Jessop^22 this is explained by the fact that the apparatus used was not fully adequate for such measurements. New experiments by Guastalla^23 also gave the correct value of \(R\); Delaplace, however, confirms, though for a comparatively small temperature range, that \(S\omega\) is proportional to the absolute temperature.

The good fulfillment of equation (1), obtained by Adam and Jessop, was observed for certain esters of dibasic acids, such as

\[ \mathrm{C_2H_5OOC(CH_2)_{10}COOC_2H_5,} \]

\[ \mathrm{C_2H_5OOC(CH_2)_{11}COOC_2H_5}^{**}. \]

At low pressures, below \(0.3\) dyn/cm, the product \(S \cdot \omega\) is \(400\) with an accuracy up to \(10\%\). The fact that precisely these substances obey the gas law so well has the following basis: both ends of the molecule are hydrophilic and therefore the molecules lie flat on the surface, remaining separated. It is all the more surprising, of course, that in the condensed state these substances have a value \(\omega_1\), corresponding to the greatest compaction, equal to \(21\ \text{\AA}^2\). The attraction between the chains must be very great, in any case great enough to completely overcome the affinity of the other hydrophilic end for water. In addition, the molecules of these substances are relatively small and therefore do not occupy a large area. If, consequently, equation (1) is fulfilled, then the \(S\)—\(\omega\) curve, in accordance with the \(p\)—\(v\) curve for an ideal gas, must have the form of a hyperbola.

* According to the kinetic theory, the energy falling to each degree of freedom of an ideal gas, calculated per molecule, is

\[ e=\frac{R}{2N}=0.66\cdot 10^{-16}\ \text{erg}, \]

where \(N\) is Avogadro’s number; for a two-dimensional space we have \(e=1.32\cdot 10^{-16}\); taking \(1\ \text{\AA}\) as the unit of length, we obtain \(S\omega=1.32T\), which, for example, at \(18^\circ\mathrm{C}\) gives \(S\omega=384.1\).

** Diethyl esters of decanedicarboxylic and brassilic acids.

The similarity to gas laws becomes still clearer if one follows the dependence of \(S\omega\) on \(S\). It is extremely similar to the dependence of \(pv\) on \(v\) for gases, i.e., according to the van der Waals equation, at first, as compression increases, the attraction between molecules makes itself felt, and then the area occupied by the molecules themselves makes itself felt still more strongly; thus at first, with increasing \(S\), there is a decrease of \(S\omega\), i.e., the action of the forces of attraction, and then a strong increase of \(S\omega\), since upon compression of the molecules the correction for the area \(\beta\) actually occupied by them becomes more and more noticeable. For intermediate values of the pressure \(S\), the equation

\[ S(\Omega-\beta)=RT. \tag{2} \]

is well applicable.

Among other substances, fatty acids on the surface of dilute HCl and their ethyl esters on the surface of water approach the fulfillment of these ideal laws at small values of the pressure \(S\)^{24}. Equation (2) is further fulfilled for benzophenone on the surface of mercury^{25}; in these experiments the pressure was not measured directly, but determinations were made of the surface tension of the pure surface and of the surface contaminated with benzophenone, i.e., \(\sigma_M\) and \(\sigma_L\). The area occupied by the organic molecules was calculated on the basis of the amount of adsorbed substance. This latter was measured directly with the aid of microbalances from the loss in weight of a benzophenone crystal that was in contact with the surface of mercury*.

The assumption following from the facts set out above—that the adsorbed molecules in these cases move freely over the surface—finds confirmation in the following experiments. If one places a crystal, or better a layer of fine crystalline dust, of some high-molecular capillary-active substance on the surface of water covered with talc particles, then the spreading of this substance over the surface can be detected by the displacement of the talc particles that it produces^{25}. The rate of spreading can be expressed by the equation:

\[ \frac{dS_t}{dt}=k(S_\infty-S_t), \tag{3} \]

where \(S_\infty\) denotes the pressure ultimately attained, and \(S_t\) the pressure present at the given moment. It should be noted that in the spreading of liquid substances over the surface of water we are dealing with a hydrodynamic process; the spreading liquid carries along the boundary^{27}. Only the observations of Volmer just mentioned, concerning the loss in weight of a benzophenone crystal when it is in contact with the surface of mercury, cannot be interpreted otherwise than as the spreading of an organic substance

* Losses in weight by evaporation into the gaseous medium, because of their smallness, may be disregarded (boiling point of benzophenone—305° C).

along the surface, accompanied by the formation of an adsorption layer. The fact that benzophenone molecules can move also along the surface of a solid body, such as, for example, a crystal of benzophenone itself or, for example, glass, has been proved by direct observations^28.

1. The properties of thin films in states intermediate between the two limiting ones—the ideal gaseous state and the condensed state—are very diverse and in part still difficult to explain. The properties of a number of substances at low temperatures are well understood. If the condensed state of the layer is disturbed by increasing the surface, then usually, over a considerable range of values of \(\omega\), constancy of \(S\) can be observed. Only at very large values of \(\omega\) does \(S\) again decrease, and in just such a way as if this part of the curve represented the end of the hyperbolic curve \(S\omega\). Such relations can be well explained: as soon as the condensed layer is given an area larger than that required for the molecules to be placed closely next to one another, evaporation into the two-dimensional space occurs, characterized at constant temperature by a constant pressure^29. This two-dimensional vapor pressure remains constant as long as there are at least some remnants—islands—of the condensed layer*. Only when these latter have completely evaporated can the pressure \(S\) again decrease. In excellent agreement with the representation just mentioned is the finding that the two-dimensional vapor pressure decreases with increasing chain length, while the region of \(\omega\)-values for which the vapor pressure remains constant increases.

Both facts are clearly seen from Fig. 5, which shows the results of measurements of a series of organic acids from tridecanoic
\(\mathrm{C}_{12}\mathrm{H}_{25}\mathrm{COOH}\) (in the figure denoted—\(\mathrm{C}_{13}\)) to palmitic
\(\mathrm{C}_{15}\mathrm{H}_{31}\mathrm{COOH}\) (in the figure—\(\mathrm{C}_{16}\)); the measurements were made at very close temperatures: at \(14.5^\circ\mathrm{C}\) for the first three acids and at \(12^\circ\)—for palmitic acid. The two-dimensional vapor pressure falls from \(0.3\ \mathrm{dyn}/\mathrm{cm}\) for tridecanoic acid to \(0.04\ \mathrm{dyn}/\mathrm{cm}\) for palmitic acid.

However, for many substances already at room temperature, and for the majority at higher temperatures, other, difficult-to-understand relations occur^30. On stretching a thin layer, the condensed state does not pass directly into the state of two-dimensional evaporation; rather, the following course is observed: after a steep fall of the pressure \(S\) in the condensed region there follows, over a small interval of \(\omega\)-values, a more gradual fall of \(S\); then a steeper fall of the curve is resumed—

* According to Devaux’s observations, evaporation of this kind into two-dimensional space (of a drop of liquid situated on the surface of another liquid) is accompanied by phenomena quite analogous to boiling.

He believes, however, that in two-dimensional space the Thomson dependence between the curvature of the liquid surface and the elasticity of the vapor is observed.

of the curve over a comparatively small segment of the abscissa axis, and only after this does the curve enter the region of constant two-dimensional vapor pressure.

Fig. 5. Dependence of the two-dimensional vapor pressure on the chain length.

Fig. 6 shows that similar relations also hold if the temperature of a film of myristic acid \((\mathrm{C}_{13}\mathrm{H}_{27}\mathrm{COOH})\), placed on the surface of an acidic solution, is raised.

Fig. 6. \(S\)–\(\omega\) curves of a substance with a liquid-expanded layer.

Thus there is also a second region of low compressibility of the layer, which on extrapolation gives the value \(\omega_2 = 48\ \text{Å}^2\). Many other substances, such as alcohols, nitriles, amides, etc., also have \(\omega_2 = 48\ \text{Å}^2\); phenol shows \(\omega_2 = 39\ \text{Å}^2\). This region is often called the region of the liquid-expanded layer (flüssig

ausgedehnte Schicht; liquid expanded film). The anomalies discovered by Langmuir¹ in oleic acid are explained by the fact that this acid, already at room temperature, gives such liquid-expanded layers. Langmuir established that the value of $\omega_1$ for this acid is much greater than $\omega_1$ in Table 1, namely equal to $46\ \text{Å}^2$; he explained this by saying that the double bond, playing the role of a hydrophilic group, is also turned toward the water, owing to which the need of the molecules for the area they occupy increases.

There is no unanimity in the explanation of the phenomenon described³¹. Whereas condensed films may be either solid or liquid, liquid-expanded films apparently always remain liquid. The surface they occupy is smaller than that which would be required if the molecules lay flat near one another; but it is, of course, considerably larger than the surface occupied by closely packed, vertically standing molecules, as occurs in the condensed state. It thus seemed that the molecules are arranged in some way obliquely with respect to one another. Adam would like to assume that, as the temperature is raised, the condensed layer gradually loosens, and since in this process the heads of the molecules interact more strongly than the chains, the latter, being in rapid thermal motion, push the molecules apart from one another*. This interpretation would require the assumption that liquid-expanded layers are built of molecular groups that occupy a larger surface than the molecules in the condensed state. It remains surprising that the magnitude of the area occupied by one such group changes so little and that, in this way, the value of $\omega_2$ is constant to such a high degree.

From these liquid-expanded layers Adam³² distinguishes “vapor-expanded layers” (dampfförmig gedehnte Schichten; vapour expanded films). For these, when the area is increased, the usual gaseous state immediately appears, without an intermediate state of constant two-dimensional pressure. Such relations are found in many esters, as, for example, in ethyl palmitate.

All these phenomena are highly sensitive to external influences. Thus, liquid-expanded films of oleic acid on the surface of a dilute acid solution turn into nearly normal gaseous films if permanganate is dissolved in the water³³. This is readily explained as follows: the chemical interaction of permanganate with the double bonds of oleic acid causes the oleic acid molecules to spread over the surface not in a semi-straightened position—

* Supporters of this hypothesis of rotating chains are Müller (Nature 19. II. 1932) and Bernal (Bernal, Nature 11. VI. 1932), who believe that in hydrocarbon crystals, near the melting point, the chains begin to undergo rotational motion, which should lead to an increase in the area they occupy by approximately $19.6\ \text{Å}^2$.

tion and melting, which is very favorable for the appearance of gaseous properties in the film. Conversely, it is observed that if cholesterol is added in sufficient quantity to fatty acids under conditions in which they usually form liquid-expanded films, their layers remain condensed.^34 Apparently, the solid, vertically standing molecules of cholesterol hinder the incipient loosening of the condensed film of the fatty acid.

III. Results of capillary-chemical investigations of layers of soluble substances

1. Up to now we have spoken of organic substances that are practically insoluble in water, or of substances that are extremely difficult to dissolve. It is possible, however, to observe many of the phenomena described above also for organic substances of low molecular weight, truly soluble, which give capillary-active solutions with water. This was done already by Langmuir in his classic work. Here we are concerned with the properties of those layers which are formed by adsorption of organic molecules at the boundary with the gaseous phase. In this case the quantities that usually characterize the phenomenon are not measured directly. They are judged from the surface tension of the capillary-active solution. This can be done on the basis of the well-known Gibbs equation, which, owing to the applicability of the van ’t Hoff law to dilute solutions, has the form:

\[ a=-\frac{c}{RT}\cdot\frac{d\sigma}{dc}. \tag{4} \]

Here \(a\) is the amount of substance adsorbed per \(1\ \mathrm{cm}^2\) of surface, and \(c\) is the equilibrium concentration in the solution. This equation is derived on the basis of the general principles of thermodynamics. It turned out to be difficult to confirm it experimentally, since it is difficult to measure the quantity \(a\) at the liquid/gas phase boundary. The amount of substance adsorbed on \(1\ \mathrm{cm}^2\) of surface is very small; moreover, the exact separation of the intended and sufficiently large surface from the rest of the liquid is by no means a simple task. McBain^35 tried to do this by cutting off, as with a microtome, the upper layer of the liquid and subjecting it to analysis. In this method, a knife placed exactly under the very surface of a capillary-active solution of various substances (caproic acid, phenol, \(p\)-toluidine, etc.), contained in a long rectangular cuvette, passed under the entire surface, as if cutting off the surface layer. The values of \(a\) obtained in this way agreed well with those calculated on the basis of equation (4).

For the most part, it is necessary only to calculate the value of \(a\) on the basis of equation (4). For this purpose one uses yet another relation, which expresses the connection of \(\sigma\) with the concentration \(c\)

capillary-active substance. Very often the formula given by Shishkovsky[^36] is used:

\[ \Delta=\frac{\sigma_M-\sigma_L}{\sigma_M}=b\cdot \ln \left(\frac{c}{c_1}+1\right), \tag{5} \]

where \(b\) and \(c_1\) are constants, with \(b\) varying little for different substances, whereas \(c_1\) is a quantity characterizing the capillary activity of the organic substance. If equation (5) is differentiated and \(\frac{d\sigma_L}{dc}\) is compared with the derivative \(\frac{dc}{dc}\) from equation (4), we obtain:

\[ a=\frac{b\cdot\sigma_M}{RT}\cdot\frac{c}{c+c_1}. \tag{6} \]

For large concentrations one may neglect \(c_1\) and obtain:

\[ a_\infty=\frac{b\,\sigma_M}{RT} \tag{7} \]

—an expression containing only constant quantities. From equation (7) it follows that at high concentrations the amount of adsorbed capillary-active substance is constant. This conclusion may be regarded as being in agreement with the fact that the \(\sigma\)—\(c\) curve of a capillary-active substance in aqueous solution (Fig. 7), in the region of large concentrations, has a flat segment \(BC\), which is explained by adsorption saturation.

Fig. 7. \(\sigma\)—\(c\) curve.

From the considerations that lead to equations (6) and (7), and from certain others that we do not set forth here, it follows that Shishkovsky’s formula is based on the following assumptions[^37]: the adsorption layer must be monomolecular; therefore, in the limiting case, it can contain only a definite number of adsorbed molecules, and in the region of small adsorption the matter is rather one of the interaction of adsorbed molecules with solvent molecules than of the action of organic molecules in the adsorption layer upon one another. Thus, according to Frumkin[^38], the deviations from Shishkovsky’s formula that are observed for substances of higher molecular weight—with 8 carbon atoms and more—are explained by the fact that attraction between the molecules of the organic substance must be taken into account.

There are, therefore, grounds for assuming that adsorption saturation corresponds to the condensed state.

The area occupied under these conditions by each molecule—\(\omega_1'\)—is easily calculated from the value \(a_{\infty}\); it is equal to:

\[ \omega_1'=\frac{1}{N\cdot a_{\infty}}=\frac{RT}{Nb\sigma_M}, \tag{8} \]

where \(N\) denotes Loschmidt’s number. For fatty acids, which practically all have identical \(b\), the value \(\omega_1'=31\ \text{Å}^2\) was obtained, which, consequently, is considerably larger than \(\omega_1=21\ \text{Å}^2\) for maximally compressed insoluble films.

There are justified doubts as to the applicability of Shishkovskii’s formula, which extends Gibbs’s equation, in the form given above and applicable to dilute solutions, excessively far. Schofield and Rideal therefore proceeded differently\({}^{40}\): they calculated the adsorbed amount of, for example, alcohol in an aqueous solution on the basis of Gibbs’s equation in its general form, in which the activity of the solution is taken into account through its corresponding vapor pressure. In this case, as the concentration increases, a constant adsorption maximum is not obtained, as formula (5) requires; although a maximum is indeed observed, with a further increase in concentration the curve descends and shows an unchanged course at ordinate values considerably smaller than the maximal ones.

How this complicated course of the curve should be explained remains unclear. It is possible that, as the concentration of alcohol increases, an ever greater number of water molecules are attracted to the surface, so that a layer of water may form lying directly beneath the upper (alcohol) layer.*

The area occupied at the adsorption maximum by each molecule proved to be \(24\ \text{Å}^2\). This already agreed better with the state of maximum packing under vertical orientation of the molecules than the figure cited above for fatty acids. For phenol, the area occupied by each molecule at the adsorption maximum was found\({}^{42}\) by the same method, and also proved equal to \(24\ \text{Å}^2\); this value agrees well with the value \(\omega_1\) given above for benzene derivatives with an OH group and a hydrocarbon chain in the para-position.

1. It is easier to subject to investigation another type of state of the adsorption layer, by attempting to establish how far equation (1) is fulfilled for dilute solutions. The pressure \(S\) is calculated as the difference between the surface tensions of pure water \(\sigma_M\) and of the solution \(\sigma_L\); \(\Omega\), and also \(\omega\), from the value \(a\) on the basis of Gibbs’s equation, which here may safely be applied in the form of equation (4). The theory here has been well confirmed by experiment\({}^{43}\) for an entire series of substances, for example, for dilute (for the most

* This explanation, put forward by Adam\({}^{41}\), is based on the assumption that the premises made in the derivation of Gibbs’s equation are violated—namely, the assumption of the absence of adsorption of water.

parts in weak solutions of HCl) of fatty acids from butyric to capric. If the curve \((S\omega; S)\) is extrapolated to small values of \(S\), then the value of \(S\omega\) for room temperature approaches the theoretically required one, i.e. 400. How great the similarity with the gaseous state is here is shown by Fig. 8, which, in a somewhat modified form, is taken from one of the works of Schofield and Rideal. On the axis of ordinates is plotted not \(S\omega\), but the ratio

\[ \frac{S\omega}{RT}, \]

which, according to equation (2), at ideal dilution should assume a value equal to unity.

Curves 1 and 2 refer to caprylic and capric acids at \(25^\circ\); curve 3 is the curve \(\left(\frac{pv}{RT};\ p\right)\) for gaseous \(\mathrm{CO_2}\) at \(100^\circ \mathrm{C}\).

Fig. 8: Comparison of the \(S\Omega - S\) curve with the \(pv - p\) curve.

At large values of \(S\), which in most cases are directly obtained from measurements\({}^{44}\), equation (2) is fulfilled. In it the quantity \(\beta\) represents a correction for the area occupied by the molecules themselves; with its aid the area occupied by each molecule is calculated—it is equal to

\[ \frac{2\beta_{45}}{N}. \]

Folmer, using this method of calculation, obtained for the lower alcohols, fatty acids, amines, etc. a value of \(\omega_1'\) averaging about \(30\ \text{\AA}^2\); Schofield and Rideal, for the lower members of the fatty-acid series, on the average approximately \(24\ \text{\AA}^2\). These figures would seem to indicate that molecules in such dilute adsorption layers still retain a sufficiently pronounced vertical position, which is very surprising in connection with the questions we have just discussed.

1. In comparing the capillary-active properties of different substances belonging to the same homologous series, a new phenomenon is encountered, for the explanation of which the assumption of a known orientation of molecules in the adsorption layer is very convenient. Fig. 9 shows the characteristic course of the curves for the lower fatty acids.

It is clearly seen how the lowering of the surface tension, i.e. the capillary activity, regularly increases along the direction from the beginning of the series. The course of the curves is entirely analogous for many other organic substances, such as, for example, alcohols, al-

hydes, amides, esters, etc. This regularity is called Traube’s rule. If one compares the lowering of the surface tension for a given concentration, it is easy to find that it increases each time by approximately a factor of three when we pass from one substance to another possessing one extra group \(\mathrm{CH}_2\). It has already been mentioned that the constant \(C\) in Shishkovsky’s equation may be regarded as a measure of capillary activity; \(C\) is the concentration of the capillary-active substance which produces a lowering of the surface tension \(\left(\dfrac{\sigma_M-\sigma_L}{\sigma_M}\right)\) of \(14\%\). The use of Shishkovsky’s formula here is justified, since Traube’s rule is satisfied for small and medium concentrations. Thus the reciprocal of \(C_1\),

\[ v=\frac{1}{C_1}, \]

with capillary activity increases in a geometric ratio, while the molecular weight increases in an arithmetic one. Consequently, the equation is satisfied:

\[ v=e^{kM}, \]

where \(M\) is the molecular weight, and \(k\) is a constant. For two substances following one another in a homologous series, whose positions are therefore characterized by the indices \(n\) and \((n+1)\), we have:

Fig. 9. \(\sigma\)—\(c\) curves of the homologous series of fatty acids.

\[ \frac{v_{n+1}}{v_n}=e^{k(M_{n+1}-M_n)}. \tag{9} \]

On the other hand, Langmuir showed that the capillary activity stands in a similar relation to the magnitude of the work \(\lambda\), which corresponds to the transfer of 1 mole of an organic substance of concentration \(C\) from the interior of the solution to its surface. Consequently, the following relation holds:

\[ \frac{v_{n+1}}{v_n}=e^{\frac{1}{RT}(\lambda_{n+1}-\lambda_n)}. \tag{10} \]

From equations (9) and (10) we obtain:

\[ \Delta=\lambda_{n+1}-\lambda_n=RT\cdot k(M_{n+1}-M_n), \tag{11} \]

i.e., in a homologous series, for each adjacent pair of substances there is a constant difference in the work required to transfer 1 mole of a capillary-active substance from the interior of the solution to the surface; in a homologous series the magnitude of the work \(\lambda\), consequently, increases linearly with molecular weight.

Langmuir drew from this the natural conclusion that the chains of these organic molecules in dilute surface layers lie flat on the surface. In such a case all \(\mathrm{CH}_2\) groups are equivalent in their action, and one may expect that each new \(\mathrm{CH}_2\) group will require the same increase in work. Thus, in dilute adsorption layers the molecules lie flat, just as is assumed for insoluble substances—for example, esters of dibasic acids; upon compression the molecules gradually straighten, and finally, at saturation of the surface, i.e., in the condensed layer, they are practically oriented vertically.

Whereas the vertical position of molecules in compact adsorption layers is doubted by no one, the position of molecules lying flat on the surface, considered as a general rule for dilute adsorption layers, is by no means a commonly held opinion. As indicated above, a number of investigators (Volmer, Schofield, Rideal) arrived at the conclusion that the area per molecule for gaseous films of truly soluble organic substances of low molecular weight, calculated from the value \(\beta\), for which equation (2) is applicable, indicates an almost correct vertical position of the molecules.

The following facts led Cassel\({}^{46}\) to the same conclusion: the capillary-active action of paraffin vapors (from normal butane to normal heptane) obeys Traube’s rule and follows a two-dimensional equation of state, which argues for a lying position of the molecules and for a strong attraction between them; thus one cannot simply use equation (2), but must introduce into it some term expressing this kind of intermolecular interaction. For water-soluble capillary-active substances, such as, for example, the lower alcohols, fatty acids, etc., on the contrary, equation (2) is applicable without corrections. Since attraction is ascribed to the chains, it remains unclear why, for soluble substances of low molecular weight, if their molecules really occupy a lying position, intermolecular attraction is not observed, as is the case for paraffins. Cassel explains this by saying that the molecules of these soluble substances do not lie horizontally, but rather stand vertically. The fact that Traube’s rule is nevertheless fulfilled must then have another explanation. The quantity \(\lambda\) increases in proportion to the number of \(\mathrm{CH}_2\) groups because, in proportion to the surface (or volume) of the \(\mathrm{CH}_2\) group, the work required for the passage of the organic molecule to the surface through the system of van der Waals forces acting between water molecules increases. Apparently, me-

...the mechanism of the process mentioned is very different from that by which insoluble paraffin molecules are bound to the surface of water.

On the basis of a comparison of the amount of work required to transfer organic molecules from the solution to its surface (or into the gas phase), Frumkin^47 concludes that there is attraction between the surface and individual \(\mathrm{CH}_2\) groups, and this speaks in favor of the conception of the molecules lying flat. This point of view, however, is based on an extrapolation which, in our opinion, has been extended too far.

As will be clear below, much, in addition to what has just been mentioned, indicates that high-molecular water-insoluble molecules in adsorbed layers, even at large dilutions, lie almost horizontally, whereas low-molecular, truly soluble substances have molecules which, even in dilute layers, are in practice also standing vertically.

Fig. 10. Adsorption isotherms of fatty acids on \(\mathrm{Si}_6\mathrm{O}_3\mathrm{H}_6\) (Siloxen).
Curve labels: valeric acid; butyric acid; propionic acid; acetic acid; formic acid. Axes: \(a\) and \(c\).

Regardless of whether Traube’s rule has the same explanation in all cases or not, we must emphasize that it is observed for an astonishingly large number of processes at phase boundaries, and not only for the case of lowering the surface tension of liquids (such as benzene^48 and mercury^49) with respect to aqueous solutions of capillary-active substances. The connection between adsorption and the lowering of surface tension, expressed by the Gibbs equation, also accounts for the fact that Traube’s rule is fulfilled in many cases of adsorption on a solid surface^49. In Fig. 10, as an example, adsorption curves are given for fatty acids on \(\mathrm{Si}_6\mathrm{O}_3\mathrm{H}_6\) (Siloxen), which is a hydrophobic, water-insoluble, and strong adsorbent. On the ordinate axis is plotted the adsorbed amount of substance \(a_1\) (which, in contrast to \(a\), is calculated per 1 g of adsorbent); on the abscissa axis—the con-

centration of the substance in the solution after equilibrium is reached1. Similar relations have been observed for the adsorption of many organic substances on charcoal. These facts make it possible, on the basis of the Gibbs equation, to relate adsorption on a solid surface also to surface tension.

Traube’s rule is applicable only to nonelectrolytes and weak electrolytes. Evidently, special conditions are required for it to hold also for ions. Thus, surprisingly, it holds for the anions of benzenetoluene- and ethylbenzenesulfonic acids:

\[ \begin{array}{ccc} \begin{array}{c} \text{benzene ring}\\ \mathrm{SO_3'} \end{array} & \begin{array}{c} \mathrm{CH_3}\\ \text{benzene ring}\\ \mathrm{SO_3'} \end{array} & \begin{array}{c} \mathrm{CH_2{-}CH_3}\\ \text{benzene ring}\\ \mathrm{SO_3'} \end{array} \end{array} \]

This is observed in the lowering of the surface tension of water by the sodium salts of these acids, in their influence on the cataphoresis of particles of ferric hydroxide sol \((\mathrm{FeO.OH})\), and also in their influence on its coagulation2. Both of the latter phenomena are closely connected with the adsorption of these anions on the surface of the particles. These ions are characterized by the following peculiarity: the sharply hydrophilic \(\mathrm{SO_3}\)-group compels the ion to orient itself strictly with one end toward the water; the rigid benzene nucleus, as follows from the properties of the phenols mentioned repeatedly above, is very much inclined thereby to assume a vertical position relative to the surface; thus the other end of the ion, with the hydrophobic chain, is necessarily turned toward the gaseous or solid phase.

IV. Comparison of the results of investigations obtained by methods of capillary chemistry with other investigations

9. Up to now we have spoken of investigations that concerned only measurements of surface tension or of differences of surface tensions; moreover, phenomena at the interface between liquid and gaseous phases were considered almost exclusively. Below it will be shown to what extent other methods are capable of supplementing and extending our information in the field under consideration, and what the phenomena are that occur at various phase boundaries.

A very valuable contribution is provided by the results of electrocapillary investigations. Frumkin3, in connection with the earlier work of Kendrick4, showed how the change in the potential difference at the boundary between liquid and gaseous phases can be measured. We shall not enter into consideration of his method. The matter here concerns so

called the $\varepsilon$-potential, a certain force $\Delta \varepsilon$ at the phase boundary (Phasengrenzkraft), which is directed vertically to the surface of their separation. Its absolute value cannot be measured; one can only determine its changes relative to the initial magnitude of the potential of the water/air interface. This boundary-phase force depends in a definite way on the composition of the very outermost surface layer, i.e. the adsorption layer. Consequently, on the basis of the corresponding electrical measurements it is possible to confirm much of what had been established for the properties of adsorption layers from measurements of surface tension. As an example we shall cite only the following: organic, capillary-active and, therefore, well-adsorbed substances affect the magnitude of $\Delta \Sigma$ much more strongly than weakly adsorbed inorganic electrolytes; since $\Delta \Sigma$ increases with an increase in the amount of adsorbed substance, in homologous series we again encounter Traube’s rule when we compare the concentrations of different substances of such a series at equal $\Delta \varepsilon$. In some cases, for adsorption layers of truly dissolved substances, it is possible to draw a conclusion about the orientation of their molecules; in doing so it should be borne in mind that these molecules are polar. Polarity, insofar as it is characterized by dipole moments obtained on the basis of dielectric constants, should not be confused with the opposition between the hydrocarbon chain and the hydrophilic end of the molecule, which in many cases is decisive for characterizing the capillary-active properties of many organic substances. Thus, the dipole moments of alcohols, beginning with methyl alcohol and ending with normal hexyl alcohol, lie within the limits $1.62—1.66 \cdot 10^{-18}$, without showing any regular progression; at the same time their capillary activity, in accordance with Traube’s rule, increases by a factor of $3^5$. Perhaps the polarity of molecules, when one has in mind the opposition of their hydrophobic chains to hydrophilic “heads,” should be characterized by a special term; such substances could be called hydropolar. Frumkin explains the appearance of $\Delta \varepsilon$ for these truly dissolved organic substances, which do not belong to the electrolytes, in the following way: the molecules of these substances, being polar, consequently have a considerable dipole moment; when oriented at the surface, they are turned with their negative hydrophilic groups (the OH groups of alcohols, the COOH groups of fatty acids, etc.) toward the water, and with the positive hydrophobic CH$_3$ groups outward. And indeed, from measurements of $\Delta \varepsilon$ it follows that the positive charges are turned toward the gas phase, and the negative ones toward the water. In this way, however, one cannot draw a final conclusion as to whether the molecules stand strictly vertically or lie almost flat on the surface, since we know only that the negative groups are turned somewhat more toward the water. More

data is provided by the study of cresols^54. \(p\)-cresol

\[ \begin{array}{c} \mathrm{CH_3}\\ \text{(benzene ring)}\\ \mathrm{OH} \end{array} \]

with the largest dipole moment \(m = 1.81 \cdot 10^{-18}\), gives a large value of \(\Delta \Sigma\), equal, for example, to \(224\ \mathrm{mV}\) at concentration

\[ C = 37.5\ \text{millimoles per } L, \]

whereas \(o\)-cresol

\[ \begin{array}{c} \text{(benzene ring with } \mathrm{CH_3}\text{ and } \mathrm{OH}\text{)} \end{array} \]

which has the minimum \(m = 1.54 \cdot 10^{-18}\), gives \(\Delta \Sigma = 19\ \mathrm{mV}\) at \(C = 50\) millimoles per \(L\); \(m\)-cresol

\[ \begin{array}{c} \text{(benzene ring with } \mathrm{CH_3}\text{ and } \mathrm{OH}\text{)} \end{array} \]

occupies an intermediate position—\(m = 1.76 \cdot 10^{-18}\), \(\Delta \Sigma = 148\ \mathrm{mV}\) at \(C = 40\) millimoles per \(L\). In this case all three cresols lower the surface tension practically equally. Apparently, \(p\)-cresol is well oriented, and its molecules stand almost vertically, whereas the molecules of \(o\)-cresol lie almost flat on the surface. If this were not so, then \(o\)-cresol, in accordance with its relatively high moment, should have caused a considerably larger value of the potential jump \((\Delta \varepsilon)\). In agreement with the above is also the fact that the area per molecule in the state of maximum packing, calculated from the value \(a\infty\), is, for \(p\)-cresol, considerably smaller than for \(o\)-cresol.

In measurements on thin layers of insoluble substances, results similar to those given above were obtained^55. Thus, upon compression of the layer, \(\Delta \Sigma\) increases up to a certain maximum value, which does not change upon further compression. This maximum value should therefore be attributed to the condensed state. In Frumkin’s measurements, for example, for myristic acid, saturation was reached at the value

\[ a = 7.6 \cdot 10^{-10}\ \text{mole per } 1\ \mathrm{cm^2}; \]

from Adam’s data, according to which there are \(21\ \text{\AA}^2\) per molecule, the value obtained is

\[ 7.7 \cdot 10^{-10}\ \mathrm{mole/cm^2}. \]

Very characteristic is the rise of \(\Delta \varepsilon\) with an increase in the number of molecules per unit area, which was found by Ghu^56 in measurements on trilaurin (Fig. 11).

Between \(\Delta \varepsilon\) and the dipole moment \(m\) of the molecule one should expect the following dependence: if the axis of the molecule is inclined to the surfac-

ness at an angle \(\alpha\), then the moment normal to the surface is \(m_1=m\cdot\sin\alpha\); if there are \(n\) molecules on \(1\ \mathrm{cm}^2\) of surface, then the moment of the double layer is \(n\cdot m\cdot\sin\alpha\), and for \(\Delta \varepsilon\) we obtain*:

\[ \Delta \varepsilon = 4\pi\cdot nm\sin\alpha . \tag{12} \]

In Fig. 11, \(\Delta \varepsilon\) is plotted along the ordinate axis, and \(n\) along the abscissa axis. The curve from 0 to point \(A\) rises very slowly; correspondingly, the angle \(\alpha\) at small surface concentrations is very small; the molecules, consequently, lie flat. In this initial region the molecules almost do not interact, and therefore \(\Delta \varepsilon\) is proportional to \(n\). From \(A\) to \(B\) the rise is no longer linear; the curve rises considerably more steeply; not only \(n\) increases,

Fig. 11. Dependence of \(\Delta \varepsilon\) on the surface concentration of molecules.

but also \(\sin\alpha\). The molecules, compressed ever more densely, straighten more and more, striving to assume a vertical position. Between \(B\) and \(C\), \(\Delta \varepsilon\) again becomes proportional to \(n\). The molecules now stand vertically; \(\sin\alpha\) no longer changes, but the molecules are still capable of becoming denser. At \(C\) saturation sets in and at the same time constancy of the quantity \(\Delta \varepsilon\) is attained. According to Schulman and Rideal\(^{57}\), the dependence of the electric moment normal to the surface \((m\sin\alpha)\) on the number and arrangement of molecules on the surface is still more complicated; however, they too arrived at the conclusion,

* The double electric layer may be regarded as a plane (atomic) capacitor with capacitance \(c\) and charge \(e\); then \(\Delta \varepsilon=\dfrac{e}{c}=4\pi el\), where \(l\) is the distance between the plates (the capacitance is referred to unit area, and the dielectric constant is taken equal to unity). The product \(el\) is the dipole moment, equal to \(el_1\sin\alpha=m\sin\alpha\). Consequently, we finally obtain:

\[ \Delta \varepsilon=4\pi\cdot n\cdot m\cdot\sin\alpha . \]

that in the two-dimensional-gas state the molecules of capillary-active high-molecular substances lie flat on the surface.

All these experiments with great probability indicate that the molecules of insoluble high-molecular substances in dilute surface solutions lie horizontally, and upon compression up to the condensed state straighten out and orient themselves vertically; at the same time it is highly probable that the molecules of truly soluble, low-molecular capillary-active substances, even in dilute adsorption layers, stand almost vertically.

1. In investigating the phenomena observed for condensed layers, the area occupied by one molecule can be found, i.e., in other words, by means of these measurements we are able to judge some of the dimensions of molecules. The question arises to what extent the results obtained in this way agree with results obtained by other methods, for example by X-ray analysis. The answer is that, as regards the order of magnitude, the agreement is good; in details, however, there are discrepancies. These discrepancies consist in almost all cases in the fact that the quantities obtained by X-ray analysis in the study of crystals prove to be smaller than the corresponding quantities found by capillary-chemical methods. Thus, various experiments carried out by capillary-chemical methods on insoluble derivatives of phenol, and also on soluble phenol itself, give for the magnitude of the cross section of the benzene nucleus \(24 \ \text{Å}^2\); X-ray photographs of naphthalene and anthracene crystals give for this quantity \(21.5 \ \text{Å}^2\). The value found for the condensed state of very numerous substances, \(21 \ \text{Å}^2\), is attributed to the area occupied by the group \(\mathrm{CH}_3\). From X-ray measurements on hydrocarbons, fatty acids, and other substances, a cross section of the carbon chain equal to \(18.5 \ \text{Å}^2\) was obtained. At present it is not yet clear how this discrepancy should be explained. For the most part it is explained by the fact that the molecules in the surface layer do not stand strictly vertically, but are inclined to the surface at some angle, which causes the apparent exaggeration of the cross section. This explanation is, of course, hardly probable for derivatives of benzene. The aforementioned discrepancies, moreover, are often also explained by the fact that in monomolecular layers there are water molecules between the molecules of the organic substance, owing to which the same strict orientation as occurs in crystals cannot take place.

From the magnitude of the area occupied by one molecule in a densely compressed, well-oriented surface layer, the thickness of the layer, and at the same time the length of the molecule \((\delta)\), can, of course, be obtained:

\[ \delta = \frac{M}{N \rho \omega}. \tag{13} \]

Here \(N\) is Loschmidt’s number, \(\rho\) is the density of the organic substance.

\(\delta\)—for simple chains, such as the chains of alcohols, fatty acids, etc., is the greater the longer the hydrocarbon chain; if \(\delta\) is divided by the number of carbon atoms of the chain, we obtain the distance at which the \(\mathrm{CH_2}\) groups are situated from one another. This distance, calculated from Adam’s measurements\(^{58}\), is equal to \(1.4\,\text{Å}\); by X-ray analysis\(^{59}\) the value \(1.27\,\text{Å}\) was obtained for it. This distance is thus smaller than the distance between the carbon atoms of diamond, in which the atoms are very closely packed and for which the corresponding distance is \(1.54\,\text{Å}\). From this one may conclude that the carbon atoms in molecular chains are arranged in a zigzag fashion.

For the case of the condensed state of adsorbed layers, if equation (8) is taken into account, we obtain:

\[ \delta=\frac{M}{N\rho\omega_{1}}=\frac{b\cdot\sigma_{M}}{RT}\cdot\frac{M}{\rho}. \tag{13a} \]

For the homologous series of amines, mono-, di-, and triethylamine:

\[ \begin{array}{ccc} \begin{array}{c} \mathrm{CH_2CH_3}\\[-2pt] \diagup\\[-2pt] \mathrm{N{-}H}\\[-2pt] \diagdown\\[-2pt] \mathrm{H} \end{array} & \begin{array}{c} \mathrm{CH_2CH_3}\\[-2pt] \diagup\\[-2pt] \mathrm{N{-}CH_2CH_3}\\[-2pt] \diagdown\\[-2pt] \mathrm{H} \end{array} & \begin{array}{c} \mathrm{CH_2CH_3}\\[-2pt] \diagup\\[-2pt] \mathrm{N{-}CH_2CH_3}\\[-2pt] \diagdown\\[-2pt] \mathrm{CH_2CH_3} \end{array} \end{array} \]

there was obtained, with the aid of the value \(b\) found on the basis of Shishkovsky’s equation, a remarkable regularity\(^{60}\). The values of \(b\) for these three substances are related as:

\[ 0.286:0.179:0.132=2.17:1.36:1; \]

their molecular volumes \(\dfrac{M}{\rho}\) give the inverse ratio:

\[ 1:1.35:8.12. \]

It then follows from equation (13a) that the length of the molecules of these three amines in the surface layer is the same. The molecules lie with the nitrogen atoms turned toward the water, and with the \(\mathrm{CH_3}\) groups outward.

The packing of the layer is more considerable than for the molecules of the previously mentioned organic substances; the \(\omega_1\) values were obtained as \(19\)—\(31\)—\(42\,\text{Å}^2\). For the \(\mathrm{CH_3}\) group of monoethylamine the value of \(19\,\text{Å}^2\) is much closer to that found from X-ray analysis and equal to \(18.5\,\text{Å}^2\), than the value of \(21\,\text{Å}^2\) mentioned by us. For the two other amines an even smaller area is accounted for by each separate \(\mathrm{CH_3}\) group.

1. From the foregoing it follows that the connection between the structure of organic molecules and those relations which are observed

for adsorption layers, especially in the compacted, condensed state, is not always entirely definite. Thus, for example, one may recall that esters of dibasic acids show a normal value of \(\omega_1\) for maximum compression, so that one has to assume that they are turned toward the water by only one of their ends; whereas, as a consequence of the hydrophilicity of the opposite ends of the molecule, one might have expected another (larger) value for the area occupied by it. Nevertheless, in a number of cases capillary-chemical relations make it possible to draw very valuable conclusions about the structure of molecules of high-molecular substances.

As an example we shall point to the cyclic ketones found by Ruzicka. On the basis of their chemical properties he came to the conclusion that in compounds having the structure \(C_nH_{2n}CO\), all the carbon atoms are closed into a ring, and the attraction between the long segments of the chain determines their close position near one another. It was interesting to test the validity of such an assumption by the capillary-chemical method. The substances mentioned belong to clearly expressed hydropolar ones. The CO group must be turned toward the water, and both chains, joined in a ring and extending outward, must occupy twice the area relative to a single chain; there should occur:

\[ \omega_1 = 2 \cdot 21 \,\text{\AA}^2 = 42 \,\text{\AA}^2 . \]

For the high-molecular ketones with 29 and 30 carbon atoms this is indeed observed\(^{61}\); for thin layers of these substances on the surface of water, simple \(s-\omega\) curves are obtained—like those shown in Fig. 3, with a steep rise, and the value \(\omega_1\) for the ketone \(C_{29}H_{58}CO\), both at \(1^\circ\) and at \(11\text{—}15^\circ\), proved to be \(43\,\text{\AA}^2\); for the ketone \(C_{28}H_{56}CO\) at \(1^\circ\)—\(45\,\text{\AA}^2\), at \(11\text{—}15^\circ\)—\(47\,\text{\AA}^2\). The length of the molecule also proved to correspond to the expected value. It was equal, for the ketone with 30 atoms, to \(19\,\text{\AA}\); for each \(CH_2\)-group there is consequently \(1/15\) part, i.e. \(1.27\,\text{\AA}\), which is in good agreement with the above value obtained by X-ray analysis. Ketones with a smaller number of carbon atoms,—with 14—21 atoms,—have \(\omega_1 = 50\text{—}60\,\text{\AA}^2\); the rise of the curve is much steeper. It is possible that here we are dealing with “liquid-expanded” layers. In explaining the structure of some high-molecular alcohols, the capillary-chemical method was also applied\(^{62}\) with success.

V. Experiments with thin layers obtained from colloidal solutions

1. Up to now we have dealt only with such substances which, including salts and high-molecular ones, were nevertheless truly dissolved. It is possible successfully to perform capillary-chemical

methods described above to the investigation of high-molecular substances, which ordinarily yield only colloidal solutions. In this case one proceeds in the usual way, i.e., a drop of a solution of the substance under investigation in an organic solvent, in which the given substance gives a colloidal or semicolloidal solution, is allowed to spread over the surface of water. Thus, for example, Katz^63 worked in his investigations on cellulose derivatives. However, surprising as it may be, such a problem can be solved much more simply. Thus, for example, one can make a drop of an aqueous colloidal solution of a protein spread over the surface of water or of an aqueous solution, if measures are taken to prevent the protein from penetrating into the interior of the liquid. For this purpose the pipette containing the colloidal solution must be carefully brought into contact with the surface, inclining the pipette, and the drop placed on the surface must be made to spread only by means of gentle blowing. Here two phenomena may be utilized: first, that a liquid with a slightly lowered surface tension, such as an aqueous protein solution, spreads well over the surface of a liquid with a somewhat higher surface tension; and, second, that the diffusion rates of such a colloidally dissolved substance are very small, and owing to this the process of penetration of the protein from the surface into the bulk of the liquid takes a long time. In addition, in many cases the protein apparently changes chemically—denatures—and thereby becomes less soluble. Gorter and Grendel^64 were the first, using Langmuir’s balance, to carry out in this way an investigation of a whole series of proteins.

Let us now consider the very diverse phenomena observed for proteins. Their great sensitivity to changes in pH causes the properties of thin layers of protein also to depend to a high degree on the pH of the liquid in the trough. For egg albumin,^65 for example, under isoelectric conditions (pH = 4.7) a very thin film is obtained, for which each milligram of protein corresponds to 0.9 m²; whereas in the region of acid reactions a minimum of approximately 0.1 m² was found at pH = 2.8; in the alkaline region the surface area corresponding to 1 mg of protein decreases almost as sharply; at pH = 6 it is equal to 0.15 m². In many cases simple $S-\omega$ curves are obtained (similar to the curve in Fig. 3) with a pronounced value of $\omega_1$; this is observed, for example, for hemoglobin both on a neutral liquid and on a strong solution of hydrochloric acid. The value of $\omega_1$, calculated per molecule, is extraordinarily large, even much larger than might be expected, assuming a cubic form of the molecule. Thus, taking the molecular weight of hemoglobin to be 68,000, $\omega_1$ on a surface of 0.1-n* is—

* On the question of the molecular weight of hemoglobin and its derivatives, see The Svedberg, Koll. Z. 67: 1, 2, 1934.

of solid HCl is equal to 13,000 Ų at a layer thickness of 6–10 Å, whereas for a cubic molecule one might have expected values of $\omega_1$ equal to 2000 Ų and a thickness of 45 Å. It should be noted, however, that these films are, ultramicroscopically, completely homogeneous. They are therefore not built, as under other conditions one often has occasion to observe, from individual colloidal particles scattered over the free surface. Such relations are usually explained as follows: the protein particles at the surface are deformed, and precisely many of them, if not all, lie, being turned with hydrophilic groups (COOH and NH$_2$ groups), toward the water.

Cellulose derivatives behave in a similar way. With the aid of the ultramicroscope it can be found that there are very diverse kinds of protein films: quite heterogeneous ones, with randomly distributed islands, and also fairly homogeneous ones. To the latter belong, for example, those which Katz obtained on the surface of water from a solution in chloroform (triacetylcellulose, lichenin acetate, etc.). The thickness of the layers was surprisingly small, reaching only 5–9 Å; this corresponded to the thickness of a glucose molecule. On this basis, and, moreover, taking into account the structure of cellulose, one could conclude that in thin layers the individual macromolecules of cellulose—which is also true for its derivatives—are arranged flat on the surface. These circumstances may be interpreted in favor of Staudinger’s views, according to which colloidal solutions of cellulose derivatives chiefly contain macromolecules. It is not excluded, however, that the micelles present in solution, i.e. certain unit formations built from groups of macromolecules, separate into individual macromolecules upon spreading over the surface of the liquid phase.

VI. Experiments at Interfaces of the Second Kind

1. Up to this point we have spoken chiefly about interfaces between water (or an aqueous solution) and air; in conclusion we shall consider phenomena observed at interfaces of another kind. Some experiments carried out on the surface of mercury have already been mentioned above; there are also other experiments of the same character$^{66}$; in general, they have not yet led to successful results. Mercury as a liquid is poorly suited for the Langmuir balance, since solid surfaces are not wetted by it. The experiment is also greatly hampered by the extreme sensitivity of the surface tension of mercury to traces of impurities.

The state of a two-dimensional adsorption layer at the interface of two liquids has been studied in considerable detail. Thus, Schofield and Rideal$^{67}$ [continued] Harkins and King’s measurements$^{68}$, which concerned the influence of an oilary

acid at the water/benzene interfacial tension, recalculated by the method set forth on p. 758. It turned out that equation (2) is satisfied; moreover, if the lateral attraction of the molecules can be neglected, the same cannot be done with respect to the area occupied by the molecules themselves.

The characteristic difference between organic molecules which, not being hydropolar, give capillary-active solutions (paraffins), and molecules which, in addition, are also hydropolar (alcohols, acids, etc.), is also manifested in the difference between these liquids with respect to their cohesive properties, on the one hand, and their adhesive properties relative to water, on the other[^69]. The measure of cohesive attraction is twice the value of the surface tension. The measure of the adhesive cohesion \(f\) between two liquids \(A\) and \(B\) is:

\[ f=\sigma_A+\sigma_B-\sigma_{AB}, \]

where \(\sigma_A\) and \(\sigma_B\) are the surface tensions of the two liquids in the pure state, \(\sigma_{AB}\) is the surface tension at the \(A/B\) interface (for liquids miscible in all proportions, \(\sigma_{AB}=0\)). A number of investigators have established that cohesive cohesion changes little from liquid to liquid. The values for paraffins, alcohols, methyl ketones, acids, and nitriles lie between \(37\text{–}55\ \frac{\mathrm{erg}}{\mathrm{cm}^2}\). The values of adhesive cohesion relative to water vary more substantially (Table 2):

TABLE 2

Name of substance	\(f\left(\frac{\mathrm{erg}}{\mathrm{cm}^2}\right)\)
Paraffins	35–45
Alcohols	95
Methyl ketones	85–90
Acids	90–100
Nitriles	90

The large figures for hydropolar substances are explained by the strong attraction between their hydrophilic groups and water. One might, however, have expected that between the OH groups of alcohols and the COOH groups of acids there would be a noticeable attraction. Since this is not reflected in the phenomena of cohesive cohesion, one must suppose that on the surface of these liquids the mentioned groups are strongly overlain by \(\mathrm{CH_3}\) groups. The differences, however, in the magnitude of cohesive cohesion are nevertheless so large that a strict orientation of the molecules with the \(\mathrm{CH_3}\) groups outward cannot be admitted.

Such a supposition is also confirmed by the fact that members of a homologous series of hydropolar substances in the form of pure liquids have only slightly differing values of surface

tension, whereas in dilute aqueous solutions these substances, according to Traube’s rule, lower the surface tension of water to extremely different degrees. But here as well, the conclusion of Langmuir and Harkins, made on the basis of the facts set forth above, that in pure liquids the molecules are oriented with the CH₃ groups outward, gives rise to doubt. Thus Sedgewick\(^{70}\) showed that the conclusions which, on the basis of such an orientation of the molecules, could have been drawn regarding the total reserve of surface energy of an organic liquid are not confirmed for a large number of liquid benzene derivatives.

From the magnitudes of cohesive and adhesive attraction one can calculate the so-called coefficient of spreading, which indicates the degree of tendency of one liquid to spread over another; it is equal to the difference between the magnitude of the adhesive \((f)\) and cohesive \((f_1)\) attraction:

\[ p = f - f_1 = \sigma_B - \sigma_A - \sigma_{AB}. \]

Since hydropolar structure plays a prominent role in adhesive phenomena, it must also influence the coefficients of spreading precisely in the sense that sharply hydropolar substances should also have large coefficients. This is in fact observed\(^{71}\) for alcohols, fatty acids, and other substances. It is striking that non-hydropolar paraffins (hexane, octane) have a positive value of \(p\), whereas higher paraffins and other organic liquids (carbon disulfide, methylene iodide) have a negative coefficient \(p\), i.e., in other words, they do not have the ability to spread.

The ability of organic liquids to spread over the surface of water is closely connected with their ability to spread over the surface of solid bodies, in particular over the surface of metals\(^{72}\), and also with the question of the stability of the thin films formed in this case. These phenomena are also of interest from the technical point of view for elucidating lubricating action\(^{73}\). The hydropolar structure of molecules and their orientation in adsorption layers apparently play an important role here as well; numerous contradictions, however, still do not allow one to give a clear picture of lubricating action as the result of the presence of oriented layers*.

* The orientation of molecules of an organic substance on the surface of a solid body was established with complete obviousness by a number of authors and by very different methods; thus Trillat did this by X-ray methods (see also Metallwirtschaft 7, 101, 1928), Kalman and Kreidl by the method of measuring dielectric constants. Recently Rupp has arrived at the same results by using the method of electron interference; processing the data obtained with the aid of Bragg’s equation, he established changes in the lattice constants with time, which naturally should be connected with the latent period of Hardy, whose theory was recently developed by Deryagin. A completely clear connection between the magnitude of the frictional moment and the orientation of molecules (“orientation effect”) was discovered

Many investigators are inclined to accept that the orientation of molecules is also important for explaining the stability of emulsions[^74]. As is known, very stable emulsions are obtained by adding emulsifiers, i.e., substances that form adsorption layers at the boundary of the two liquid phases. These emulsifiers, such as soaps and saponin, are almost exclusively hydropolar substances, for which one may expect the hydrophilic groups to be turned toward water, and the hydrophobic ones toward the organic liquid (or mercury). The substances under consideration, however, are almost all colloids or semicolloids, and therefore it remains an open question whether the character and arrangement of micelles in the boundary layer is not more important than the assumed orientation of macromolecules.

In certain other cases, the orientation of hydropolar molecules perhaps has greater grounds for being recognized. Thus, the increase in the stability of the so-called Oden sulfur sol (the particles of this sol apparently consist of a solution of sulfur in polysulfide hydrogen[^75]) by adding pentathionic acid \((\mathrm{H}_2\mathrm{S}_5\mathrm{O}_6)\) can be explained by the fact that this acid is adsorbed in the form of oriented layers; in this case the sulfur atoms are turned toward the colloidal particle, and the hydrophilic oxygen-containing groups toward the water[^76]. The colloidal particle is thereby made, in general, hydrophilic, and the colloidal solution consequently more stable. In a similar manner one explains the very far-reaching action of tannin[^77] and other substances similar to it, which consists in the fact that tannin makes the particles of hydrophilic sols (agar and others) more hydrophobic; it thus sensitizes the sols with respect to the coagulating agent, making them less stable. Tannin is also constructed hydropolarly: on the one hand, it contains a hydrophilic residue of the grape-sugar molecule, and on the other, a hydrophobic galloyl residue*. It is highly probable that the hydrophilic ends of its molecules are turned toward the surface of the hydrophilic particle, while the hydrophobic ends are situated closer to the water; owing to this, the particles, including their adsorption layer, would be made more hydrophobic, and the colloidal solution as a whole less stable.

Finally, the remarkable phenomena observed in surface catalysis have been connected with the orientation of reacting molecules[^78]. Such are, for example, the retardation of the reaction of KI with the acid sodium salt of \(p\)-sulfodibromo-hydrocinchoninic acid on carbon, and also the acceleration on carbon of the reaction of KI with \(\alpha\)- and \(\beta\)-dib-

researches of Vieweg (Vieweg, Arch. Eisenhüttenwes 2, 805, 1929) when measuring the rectifying action of a lubricating layer which, under asymmetrical conditions, possesses one-sided conductivity. Woog, Karplus, Kost, and others proposed various schemes of the structure of the lubricating layer.

* According to E. Fischer, tannin is pentadigalloylglucose: \([(\mathrm{OH})_3\cdot \mathrm{C}_6\mathrm{H}_2\cdot \mathrm{CO}\cdot \mathrm{OC}_6\mathrm{H}_2(\mathrm{OH})_2\cdot \mathrm{CO}]_5\cdot \mathrm{C}_6\mathrm{H}_7\mathrm{O}_6\).

propionic acid. It is possible, however, that in these cases insufficient attention has been paid to the capillary activity of the initial substances and the final products of the reaction.

LITERATURE

1. Langmuir, Chem. metallurg. Engng. 15, 468, 1916; Journ. Am. Chem. Soc. 39, 1848, 1917.

2. Hardy, Proc. Roy. Soc. Lond. (A) 86, 634, 1912; 88, 330, 1913.

3. W. D. Harkins, F. E. Brown a. E. C. H. Davies, Journ. Am. Chem. Soc. 39, 354, 1917; W. D. Harkins, E. C. H. Davies a. G. Z. Clark, Journ. Am. Chem. Soc. 39, 541, 1917.

4. Rayleigh, Proc. Roy. Soc. Lond. 47, 364, 1890; 48, 127, 1890; Philos. Mag. (5) 30, 386, 1890; 48, 331, 1899.

5. Agnes Pockels, Nature Lond. 43, 437, 1890; 46, 418, 1892; 48, 152, 1893; 50, 223, 1894; Wiedemanns Ann. 67, 668, 1899; Naturwiss. 5, 137, 149, 1917.

6. H. Devaux, A. R. Smithson Inst. 261, 1913.

7. Rayleigh, Philos. Mag. (6) 48, 337, 1899.

8. Zocher u. Stiebel, Z. physik. Chem., Abt A 147, 401, 1930.

9. N. K. Adam, Proc. R. y. Soc. Lond. (A) 99, 336, 1921; 101, 425, 1922; 103, 676, 687, 1923; 119, 628, 1928; Adam a. Dyer, Proc. Roy. Soc. Lond. 106, 694, 1924; Adam a. Jessop, Proc. Roy. Soc. Lond. 110, 423, 1926; 112, 362, 376, 1926; 120, 473, 1928; Adam, Berry, Turner, Proc. Roy. Soc. Lond. 117, 532, 1928; Adam a. Rosenheim, Proc. Roy. Soc. Lond. 126, 25, 1929; Adam, Kolloid Z. 57, 125, 1911; 61, 168, 1932.

10. Schofield a. Rideal, Proc. Roy. Soc. Lond. (A) 109, 57, 1925; 110, 167, 1926; Cary a. Rideal, Proc. Roy. Soc. Lond. 109, 201, 318, 331, 1925; Mougin a. Rideal, Proc. Roy. Soc. Lond. 114, 690, 1927; Lyons a. Rideal, Proc. Roy. Soc. Lond. 124, 322, 333, 314, 1929; Goard a. Rideal, Journ. chem. Soc. Lond. 14, 780, 1668, 1925; Rideal, “An Introduction to surface Chemistry”, Cambridge 1926.

11. A. Marcelin, Ann. Physique (9) 1, 19, 1914; (10) 4, 459, 1925.

12. Ann. Physique (9) 14, 164, 1920.

13. Adam, Berry a. Turner, Proc. Roy. Soc. Lond. (A) 117, 536, 1928.

14. Lyons a. Rideal, Proc. Roy. Soc. Lond. (A) 124, 33, 1929.

15. Adam, Proc. Roy. Soc. Lond. (A) 101, 466, 1922.

16. Langmuir, loc. cit.; Adam, Proc. Roy. Soc. Lond. (A) 99, 336, 1921; 101, 452, 1922; Hartridge a. Peters, Proc. Roy. Soc. Lond. (A), 101, 348, 1922; Gorter u. Grendel, Proc. Acad. Amsterd. 29, 1262, 1926; Egue u. Hågg, Philos. Mag. (7) 4, 667, 1927.

17. Proc. Roy. Soc. Lond. (A) 124, 322, 333, 344, 1929.

18. I. Traube, Liebigs Ann. 265, 27, 1891.

19. H. Devaux, A. R. Smithson Inst. 273, 1913.

20. Adam a. Jessop, Proc. Roy. Soc. Lond. (A) 112, 376, 1926.

21. A. Marcelin, Ann. Physique (10) 4, 496, 1925.

22. Adam a. Jessop, Nature Lond. 117, 484, 1926.

23. E. R. Guastalla, Acad. Sci. Paris 189, 241, 1929.

24. Adam a. Jessop, Proc. Roy. Soc. Lond. (A) 110, 423, 1926.

25. Volmer u. Mahnert, Z. physik. Chem. 115, 239, 1925.

26. Cary a. Rideal, Proc. Roy. Soc. Lond. (A) 109, 301, 1925; Adam a. Jessop, Proc. Roy. Soc. Lond. (A) 110, 441, 1926.

27. Brinkmann u. v. Szent-Györgi, Bichem. Z. 139, 274, 1923; Cary a. Rideal, loc. cit.; Landt u. Volmer, Z. physik. Chem. 122, 398, 1926.

1. Volmer and Adhikart, Z. Physik. 35, 170, 722, 1925; Z. physik. Chem. 119, 46, 1926. Moll, Z. physik. Chem. 136, 183, 1928.

2. Adam and Jessop, Proc. Roy. Soc. Lond. (A) 110, 423, 1926.

3. Labrouste, loc. cit.; Adam, Proc. Roy. Soc. Lond. (A) 101, 516, 1922.

4. Langmuir, Colloid. Sym. Monogr. 3, 70, 1925; Cary and Rideal, Proc. Roy. Soc. Lond. (A), 109, 326, 1925.

5. Adam and Jessop, Proc. Roy. Soc. Lond. (A) 112, 362, 1926.

6. Adam and Jessop, Proc. Roy. Soc. Lond. (A) 112, 371, 1926.

7. Leathes, Journ. of Physiol, 58, No. 2/3 VI, 1923.

8. J. W. McBain, and C. W. Humphreys, Journ. physik. Chem. 36, 300, 1932.

9. Szyszkowski, Z. physik. Chem. 64, 385, 1908.

10. F. Haber, Journ. Franklin Inst. 437, 1925.

11. A. Frumkin, Z. physik. Chem. 116, 466, 1925.

12. Bury, Philos. Mag. (6) 4, 980, 1927.

13. Schofield and Rideal, Proc. Roy. Soc. Lond. (A) 109, 57, 1925.

14. Adam, “The Physics and Chemistry of Surfaces”, p. 131.

15. Goard and Rideal, Journ. chem. Soc. Lond. 127, 1668, 1925.

16. Schofield and Rideal, Proc. Roy. Soc. Lond. (A) 109, 57, 1925; 110, 167, 1926.

17. Volmer, Z. physik. Chem. 115, 253, 1925.

18. Adam, Proc. Roy. Soc. Lond. (A) 101, 652, 1922.

19. Cassel and Formstecher, Kolloid-Z. 61, 18, 1932.

20. Frumkin, Z. physik. Chem. 116, 501, 1925.

21. Harkins and Humphrey, Journ. Amer. chem. Soc. 38, 242, 1916.

22. Freundlich, “Kapillarchemie”, 4th ed. I, pp. 257, 408.

23. Kautsky and Blinoff, Z. physik. Chem. (A) 139, 497, 1928.

24. Freundlich and Slottman, Z. physik. Chem. 129, 305, 1927.

25. Frumkin, Z. physik. Chem. 109, 34, 1924; 112, 190, 1924; 116, 485, 1925; Kolloid-Z., 35, 340, 1924.

26. Kenrick, Ztschr. physik. Chem. 19, 625, 1896.

27. Frumkin, Doude and Kulvarakaja, Z. physik. Chem. 123, 321, 1926.

28. Frumkin, Z. physik. Chem. 116, 485, 1925.

29. Guyot, Ann. Physique (10) 2, 589, 1924.

30. Schulman and Rideal, Proc. Roy. Soc. Lond. (A) 130, 259, 270, 284, 1931.

31. Adam, Proc. Roy. Soc. Lond. (A) 101, 463, 1922.

32. Hengstenberg, Z. Krystallogr. 67, 583, 1928; the literature is also given there.

33. Schnell, Z. physik. Chem. 127, 121, 1927.

34. Bücher, Katz and Samwel, Z. physik. Chem. (B) 5, 327, 1929.

35. B. C. J. G. Knight, Nature, Lond. 125, 951, 1930; Adam, Journ. chem. Soc. Lond. 164, 1933; Adam, Kolloid-Z. 61, 175, 1932.

36. Katz and Samwel, Naturwiss. 16, 592, 1928; Liebigs Ann. 472, 241, 1929; 474, 296, 1929.

37. Garter and Grendel, Acad. Amsterd. Versl. 29, 1262, 1926; 32, 770, 1929; Trans. Faraday Soc. 22, 477, 1926; Biochem. Z. 201, 391, 1928; A. H. Hughes and E. K. Rideal, Proc. Roy. Soc. Lond. (A) 137, 62, 1932.

38. Gorter, van Ormondt and Dom, Acad. Amsterd. Versl. 35, 838, 1932.

39. Devaux, Proc. Soc. Sci. Bordeaux 1927; Journ. Physique 9, 37, 1928; Sheppard and Keenau, Nature, Lond. 121, 982, 1928; Keenau, Koolloid-Z. 47, 289, 1929.

40. Schofield and Rideal, Proc. Roy. Soc. Lond. (A) 109, 69, 1925.

41. Harkins and King, Journ. Amer. chem. Soc. 41, 983, 1919.

42. Hardy, Proc. Roy. Soc. Lond (A) 88, 303, 1913; W. D. Harkins; F. E. Brown and E. C. H. Davies, Journ. Amer. chem. Soc. 39, 354, 1917. W. D. Harkins, G. L. Clark and L. E. Roberts, Journ. Amer. chem.

Soc. 42, 700, 1920; W. D. Harkins, and Fwing, Journ. Amer. chem. Soc. 42, 2539, 1920; W. D. Harkins, Z. physik. Chem. 139, 647, 1928.

1. Sugden, Journ. chem. Soc. Lond. 125, 1167, 1924; Trans. Farad. Soc. 22, 486, 1926.

2. Harkins and A. Feldmann, Journ. Amer. chem. Soc. 44, 2665, 1922; W. D. Harkins, Colloid. Symp. Monogr. 6, 17, 1928.

3. Bulkley and Snyder, Journ. Amer. chem. Soc. 55, 194, 1933.

4. W. B. Hardy and I. K. Hardy, Philos. Mag. (6) 38, 32, 1919; W. B. Hardy, Philos. Mag. (6) 40, 201, 1920; W. B. Hardy and I. Doubleday, Proc. Roy. Soc. Lond. (A), 100, 550, 1922; 101, 487, 1922; 104, 25, 1923; W. B. Hardy and Bircumshaw, Proc. Roy. Soc. Lond. (A), 108, 1, 1925; W. B. Hardy, Journ. chem. Soc. Lond. 127, 1207, 1925; W. B. Hardy, Kolloid-Z. 46, 268, 1928; I. Doubleday, Kolloid-Z. 121, 2875, 1922; I. Doubleday, Proc. Roy. Soc. Lond. (A) 106, 341, 1924; Trillat, Ann. Physique (10) 6, 5, 1926; Kallmann and Kreidl, Z. physik. Chem. (A), 159, 322, 1932.

5. W. D. Harkins, E. C. H. Davies and G. L. Clark, Journ. Amer. chem. Soc. 39, 586, 1917; W. D. Harkins, Science 59, 463, 1924; Finkle, Draper and Hildebrand, Journ. Amer. chem. Soc. 45, 2780, 1923; W. D. Harkins and N. Beeman, Proc. Nat. Acad. Science 11, 631, 1925; Journ Amer. chem. Soc. 51, 1674, 1929.

6. Deines, Kolloid-Z. 62, 145, 1933.

7. Freundlich and Scholz, Kolloidchem. Beih. 16, 234, 1922.

8. Kruyt, Kolloid Z. 31, 338, 1922; Kruyt and Bungenberg de Jong, Z. physik. Chem. 100, 250, 1922.

9. Kruyt and van Duin, Rec. Trav. chim. Pays-Bas 40, 249, 1921; van Duin, Rec. Trav. chim. Pays-Bas 47, 715, 1928; Kruyt, Z. Elektrochem. 35, 539, 1929.