ASSOCIATION is a term used in a specialized sense in chemistry to denote the union of like atoms of an element or molecules of a compound to give a more complex molecule hav ing the same chemical properties. The term polymerization is not quite synonymous, for it has a wider application (see later) . Thus the molecule of white phosphorus is produced by the asso ciation of four atoms of the element to give and the molecules of water are said to be associated because they are complex mole cules, each consisting of two or three simple 11,0 molecules. The reverse process of complex molecules breaking down to simple ones is known as dissociation, but this term is not necessarily the converse of the term association, since it includes also the break ing down of a molecule into simpler unlike species (ions, atoms, or molecules).
The phenomenon of association is found in solids, liquids, and gases, but its extent can only be determined with certainty in the case of gases; its existence can be detected in the case of liquids, but its extent cannot be determined definitely; whilst in solids, its existence is inferred both because a solid is probably at least as complex as the corresponding liquid or vapour, and because, in many cases, X-ray examination shows that several molecules unite to form a crystal unit.
The existence of association in gases was inferred from vapour point density determinations by E. Mitscherlich and J. B. A. Du mas, but its extent was only clearly demonstrated when S. Canniz zaro showed how to apply Avogadro's law to such cases. Chiefly owing to the work of Victor Meyer, a mass of data was then accumulated on the molecular complexity of vapours at tem peratures up to 2,000°C. Thus, the vapour densities showed that the molecules of most metals were monatomic at tempera tures slightly above their boiling points, that phosphorus was in the vapour state, arsenic As, at 1,700°C., and sulphur under reduced pressure at 200°C., decreasing by stages to at about I,700°C. This decreasing complexity with increasing temperature is quite general, and even stable molecules like and begin to break up to Cl and Br above r,000°C. ; so also at very high temperatures and show a very slight dissociation to simple atoms. Among compounds, cuprous chloride was shown to be Cu, in a state of vapour, and ferric chloride to be Fe, Cl, at about 500°C., whereas many metallic chlorides and bromides were unassociated; acetic acid was (C2H402)2 just above its boil ing point, but at about 200°C.; phosphorous, arsenious, and antimonious oxides were and Sb40e at low temper atures, the second being at i,800°C. and the last remaining complex even at this high temperature. Phosphoric oxide seems to persist as P401, at very high temperatures. In many cases the equi librium between the higher and lower type of molecule has been shown to conform to the law of mass action (see CHEMICAL ACTION) at each of a number of temperatures. Moreover, accord ing to this law, increase of pressure should decrease dissociation, i.e., favour association, and this also has been verified. The asso ciation of nitrogen peroxide can be followed visually by the loss of colour as the deep brown is cooled and assumes the form of the colourless N20, in increasing proportions.
In a few cases, vapour-density results have led to erroneous conclusions; thus, mercurous chloride was thought to be HgC1 until it was shown that the molecule was really but that it dissociated into unlike molecules (Hg2C12---> and not into like molecules (2HgC1). According to H. B. Baker, how ever, perfectly dry mercurous chloride does not dissociate but remains as in the vapour state (see DRYNESS, CHEMICAL).
The association of liquids can be deduced from a variety of phenomena. J. D. van der Waals showed that a certain relation should hold between the critical temperature and pressure and the vapour pressure at any given temperature, and the fact that marked deviations were found by S. Young in the case of certain liquids containing hydroxyl groups (e.g., water, alcohols, and acetic acid) was ascribed to association in the case of these liquids. Similarly, the abnormally high boiling point of water (I oo ° C.) , when compared with that of hydrogen sulphide (— 6o° C. ), for instance, leads to the same conclusion. Further, F. Trouton showed that for "normal" liquids the equation ML/T = 20.7 held with fair accuracy at the boiling point (M=molecular weight; L=latent heat of vaporization; T=ab solute temperature) ; marked deviation from this relation is shown by the above hydroxylic liquids and is regarded as evi dence of association.
The precise extent of association in the liquid state was first studied by Sir W. Ramsay and J. Shields, using a modification of a formula proposed by R. Eotvds. If y is the surface tension, M the molecular weight, and v the specific volume at any given temperature T, then they found that the "molecular surface energy" y(Mv)3 changes at a definite rate with change of tem perature : dy (Mv) l/dT = k, the constant k being approximately 2.121 for all non-associated liquids. In many cases, k was less than this value, and a factor x was introduced, such that the use of Mx instead of M gave the correct value for k. The necessary value of x decreased with rise of temperature and was held to denote the degree of association, and the results were in agree ment with the foregoing qualitative methods. Quantitatively, however, the method is open to many objections; P. Walden showed that k was not a constant, even for normal liquids, and was dependent upon the molecular weight, varying from 2 up to about 6, this high value obtaining for tristearin. G. M. Bennett and A. D. Mitchell showed that the "total molecular surface energy," K= (y--T.dy/dT) (Mv) was constant over a wide range of temperature for any one unassociated liquid, and that K was an additive function of certain atomic and structural con stants even for substances which gave high values for Ramsay and Shields's k ; K was not constant for associated liquids, :;ut although the method could be applied to the evaluation of the degree of association of some liquids, it failed in the case of hydroxylic liquids. Walden used the specific cohesion yv and found that = I • i 6 2 for normal liquids, and this con stant did not suffer from the same disadvantages as that of Ramsay and Shields. By a modification of this he deduced de grees of association for many substances, but some (e.g., benzene, 1.85) seemed improbable as judged by other methods.' Longuinescu found that for normal liquids T/Acn= ioo, where d is density at boiling point, T (in absolute degrees), and it is the number of atoms in the molecule. His results for it led to values for the degree of association which were similar to those deduced from other methods. E. C. Bingham, from viscosity data, and J. Traube, from considerations of atomic and molecular volumes, deduced similar results. The general trend of the f ore going methods seems to indicate that water is chiefly and but this is doubtless a statistical average for all sorts of molecules from to, possibly, (see WATER).
Still more modern views, based on dielectric constants, internal pressure, and other physical properties; tend to supplant all the foregoing results in attributing a considerable but constant de gree of association to many of the liquids hitherto regarded as "normal," and a similar but variable degree to other liquids.
Although it is not possible to give any very definite information as to solids, yet many cases are on record in which they give com plex molecules in solution. Thus, benzoic acid exists as double molecules in benzene solution; trimethylammonium chloride and bromide undergo four- or five-fold association in fairly concen trated bromobenzene solution; phosphorus and sulphur are re spectively P4 and S8 in carbon disulphide solution ; but most metals are monatomic when dissolved in mercury or molten tin. 'In dealing with surface-tension measurements, it must always be remembered that the molecules in the surface layer may differ from those in the bulk of the liquid.
In dealing with solutions, however, so many new factors are in volved that the results should not be accepted as conclusive without careful consideration—the solvent plays an important and imperfectly understood part in such cases. For this reason, it is doubtful how to interpret G. Oddo's results that water has an association factor of 1•2---2.0 when dissolved in several organic solvents.
It is important to note that association in the solid state is a fundamental concept of A. Smits's theory of allotropy (q.v.), and that the results of both this author and H. B. Baker seem to show that association in the solid and liquid states depends on the degree of dryness (q.v.), just as it does in the gaseous state.
As implied at the outset, the terms association and polymeri zation are not always interchangeable, for the latter was applied by J. J. Berzelius to cases where the percentage compo sition remained the same, but the properties (and molecular weight) were different, whereas in association there is no well marked change in the chemical properties, as far as we are able to tell. In the article POLYMERIZATION, instances are given of organic substances which differ greatly from their "polymerides"; on the other hand, in the cases dealt with here, such differences are not sufficiently pronounced to be detectable as a rule, although the different forms of phosphorus and sulphur may possibly have to be classed as polymerides on account of their altered solubility properties (see ALLOTROPY) . It may be that, in the future, im proved experimental technique will be able to assign different properties to each of the different stages of complexity in the molecules of all the substances discussed here.
(A. D. M.)