T H E STRUCTURE OF WATER BY S. W. PENNYCCICK
I n recent years attention has been recalled to the problem of valence and co-ordination numbers, and it is generally accepted that (with very few exceptions) each element can exercise, and does exercise, valence forces in excess of that fixed number which had for so many years been assigned to it. The oxygen atom possesses (and exercises) a very active auxiliary valence and forms numerous molecular compounds, sometimes referred to as oxonium compounds, in which the oxygen‘ atom can show a maximum co-ordination number of 4. The evidence in favour of a strong residual oxygen field is now quite extensive.’ Further, hydrogen has a co-ordination number z , ~and therefore in many of its compounds i t can still exert an attraction on the lone electron pairs of other elements. The evidence for the existence of these molecular compounds is also very exten~ive.~The simplest compound of these two elements is H-0-H, and this paper concerns itself largely with the simple explanation of various phenomena that follow from the assumption that the auxiliary valence fields of both hydrogen and oxygen in the water molecule are not only very active, but also have a strong attraction for one another.
The Structure of the Water Molecule The numerous unique properties of water,-the high dielectric constant, the solvent activity, the remarkable and general catalytic activity, and even the physiological importance-must finally be traced to the structure of the individual molecule itself, and an examination shows that this structure is in most respects unique. I n the first place the molecule is polar and is therefore unsymmetrical, and as there now seems to be little doubt that the saturated oxygen atom possesses tetrahedral structure4, the water molecule is most satisfactorily represented by the tetrahedron in Fig. I , where the electrons are in pairs a t tetrahedron corner^.^ (The oxygen nucleus with its two helium 1 See chiefly Sagnet: Compt. rend., 58, 381, 675 (1860); Friedel: Bull., ( 2 ) 24, 166, 241 (1875); Hantzsch:Z. physik. Chem., 61, I 7 (1907); 65,41 (1908); Homfray: J. Chem. SOC. 87, 1443 (1905); Irvine and Moodie: J. &hem. Soc., 89, I578 (1906); Falk and Selson; J. Am. Chem. SOC.,37, 1732 (1915); Kendall and Booge: 38, 1712 (1916); Tschelinzeff: Bull., (4) 35, 741 (1924);37, 181 (1925); Morgan and Smith: J. Chem. Soc., 125, 1997 (1924); Picard: Helv. Chim. Acta, 7, 800 (1924); Sidga-ick and Callow: J. Chem. SOC., 125, 527 (1924); 127,908 (1925). Sidgwick: J. Chem. SOC.,123, 726 (1922);Trans. Faraday Soc., 19, 1 1 , 469 (1923). 3 See chiefly Lowry: J. SOC.Chem. Ind., 42, 43 (1923); Lowry and Burgess: J. Chern. SOC., 123, 2111 (1923); Lowry a n d c u t t e r : 125, 1466 (1924); Morgan and Smith: J. Chem. Soc., 125, 1997; Sidgwick and co-workers: 527 (1924); G. 1\J. Lewis: “Valence”; Werner: “Sew Ideas in Inorganic Chemistry”. Picard: Helv. Chim. Acta, 7, 800 (1924); Main Smith: “Chemistry and Atomic Structure”. Huggins: Phys. Rev., ( 2 ) 27, 281 (1926).
1682
S . W. PENNYCUICK
electrons is omitted). The simplicity of the total electronic structure in this molecule must be emphasised, for like the CH, molecule it possesses only the two closed shells of neon with a total of I O electrons. This simplicity of structure, in conjunction with the foregoing, allows us to assume that all the auxiliary valence effects are due directly to the two lone oxygen pairs or to the bivalence of the hydrogen nuclei, the two helium electrons in the inner shell having a very minimum of disturbing action. The polarity of the molecule is obvious from the structure above, and hence water can attach itself to other molecules, as experiment indicates, either nthrough its own negative pairs, or through its own positive hydrogen nuclei, which accounts for the “positive and negative” affinity of the water molecule as outlined by Briggs.’ Further, the molecular of the water molecule is relatively small .-L_-.-.-._ volume 0” (according to FraenkeP, the diameter is 1.7 X I O - ~ * cm.), and as there are four active points of auxilFIQ.I iary attack, it is not surprising that the molecule shows such great activity. Numerous simple indications, such as the great difficulty experienced in removing the last traces of water from asubstance, and the remarkable catalytic properties of traces of water vapour, point to the fact that the atoms in the H-0-H molecule are in especially favourable positions to use their auxiliary fields. The characteristic peculiarities of the structure outlined above will be shown to be quite sufficient to account for various of the unique properties of this compound.
A
.-
.
The Structure of Ice The X-ray analysis of ice supports Bragg’s view3 that each oxygen atom is surrounded by four atoms of hydrogen, and that each hydrogen atom lies symmetrically between two atoms of oxygen. Thus in the ice crystal each oxygen atom is exerting its maximum co-valence of four, and each hydrogen atom its maximum of two, or, what is the same thing, each water molecule is completely utilising its four auxiliary bonds. It is significant (from the point of view of the tetrahedral water molecule) that the crystal structure of ice is directly comparable with that of the diamond. I n fact the former can be obtained from the latter by replacing each carbon by an oxygen atom, inserting a hydrogen nucleus between each pair of oxygen atoms, and making a very simple parallel planar shift (Figs. 2A and 2B). Just as the carbon atoms in diamond may be said to form hexagonal, six carbon rings, so the oxygen atoms in ice form hexagonal six oxygen rings, but with an alternating hydrogen nucleus. Ice thus forms a network in three dimensions of H-0-H rings, each containing six H-0-H molecules. It is quite probable (Bragg) that in such a structure the primary bonds are of the same strength as the secondary and that therefore the whole is symmetrical. It is important to stress the simJ. Chem. SOC.,93, 1564 (1908);115, 278 (1919). (1918). BProc. Phys. SOC.London, 34, 98 (1922).
* J. Rum. Phys. Chem. Soc., 49, 87 50, 5,
THE STRUCTURE OF WATER
1683
Fro. z u l l g . zl3 is B photograph of a model of the ice crystal (after I-dragenatoms (or nuclei) with a saturated co-vsleiice of z. The 8tnmtUre is even more regular than that of the diamond, and thebmt-skpcd (hcnae"e) rinse, here containing 6 oxygen stom8, stand out. Tho top layer of S t o m LY identysl mth, and mperposable on, tho bottom Isyer, and the stmetwe repents itself every third layer.
1684
S. W. PENNYCUICK
plicity of this structure, and to emphasise again that the only bonds between the molecules appear to be the active auxiliary bonds with which H-0-H forms its numerous co-valent compounds. If these be the only bonds between the molecules in the solid state, then we can have no reason for assuming that the liquid molecules are held by other than the same co-valent bonds. (For, speaking generally, we should expect that known auxiliary bonds play an important part in liquid structure, but that in the case of solids owing to the decreased molecular energy, quite new bonds due to outer electronic rearrange-
FIG.3
ments might come into action. I n the structure of ice there is no evidence of such new bonds, and hence there is no need to presuppose such in water structure). It has been suggested1 that the association of molecules in liquid H-0-H is due to the particular secondary bonds under discussion. But in view of the foregoing is there any necessity to imagine the existence of any other form of bond between the water molecules in general? It is attempted in this paper to show that the various properties of water can be explained on the assumption that the molecules are held together solely through the attractions between the hydrogen nuclei and the oxygen lone electron pairs.
The Condensation of H-0-HMolecules When a gaseous H-0-H molecule a t condensation temperature enters the auxiliary field of a second, then assuming the tetrahedral structure, combination can take place a t the apices of the tetrahedra concerned. The (H-O-H)z molecule so formed could exist as one of the structures shown in Fig. 3. The structures B and C require an approach of the oxygen nuclei, and general chemical evidence shows such to be rare even in the case of strong primary fields. It is therefore probable that structure A more accurately represents the facts (Picard). Owing to the simple and symmetrical structure of the molecule (two hydrogen nuclei and two lone pairs), the types of double mole-
* Latimer and Rodebush:
J. Am. Chem. Soc., 42, 1431 (1920).
+
.. -
-.
H
:0:
+
H
H:O
..
,
:-
H
The HCl molecule is now more Dolar and sufficiently active to attach itself to the NHs lone pairs, as in H .. (2).
NHI
+ H C l . Hz0 e
..
..
H : N : H:C1 : H : O :
..
H
.. H
and finally, H
..
..
Or again, when a gaseous H-0-H molecule is captured through its auxiliary field by some foreign surface, its exposed field is now more active than the normal, and further as i t has surrendered some of its kinetic energy, it now
1686
S. W. PENNYCUICK
acts as an efficient condensing agent towards other water molecules; hence the condensation of water vapour on nuclei. It may also be concluded that the strong attraction which exists between water and almost all substances, causing water to be classed’ as Ira special contamination” which is completely removable only with very great difficulty, is due primarily to the water itself, for in addition to the fact that the latter has such strong auxiliary fields, these active fields are both “positive” and “negative” and can thus neutralise practically any foreign field.
The Structure of Water When a third molecule of water approaches an H-0-H pair it adds on through a hydrogen nucleus or a lone electron pair in the foregoing way. Hence we get (H-O-H)3 and then IH-O-H)a and so on. These tend to form polar chains as below - .. + *. I .. :O:H :O:Hi:O:H :O:H :O:H :O:H H ’. I H I H .’ I H
1, 1,
”
Although the end molecules of the chain are the most active, the interlying molecules are not wholly inactive, and the lesser tendency towards the formation of side chains plays an important part in determining the mobility of the liquid. (In the solid state the growth of the side chains is of fundamental importance and gives the system a definite structure). So far as we know, the individual molecules in the liquid state possesses translational, vibrational, and a certain amount of rotational energy. The total molecular energy, limits the length of the polar chains and continually breaks the longer ones at the weaker links. But there is another factor influencing the size of these chains. The two dimensional regularity in the above figure is diagrammatic only, the actual chains will be irregular and diqtorted because of tetrahedral structure, and if the ends of a chain combine or neutralise one another the resulting closed chain or ring will be relatively inactive. Various considerations lead us to believe that such closed chains play a very important part in the structure of water. The hexagonal structure of the ice crystal together with the tetrahedral structure of the water molecule, point definitely to the conclusion that if such stable rings,are formed in water, they must contain six oxygen nuclei and be represented by Fig. SA. Such an (H-O-H)6 ring bears a striking resemblance to the inactive benzene ring, particularly when it is kept in mind that the alternating hydrogen in the former ring is quite inactive (co-valence z ) , and probably occupies a small volume. Now the general evidence points undoubtedly to the association of the water molecule, and, in Trans. Faraday SOC.,6 , 71 (I~IO),it is concluded after some exhaustive discussion, that water consists of some (H-O-H)3 (“ice”) molecules, a large percentage of (H-O-H)* molecules, and some single molecules; and some such idea is generally accepted a t the present day. The 1
Smite: “Theory of Allotropy”
(1922).
1687
THE STRUCTURE O F WATER
structure of ice shows however, that the term ice molecule has no meaning, but that if there any unit of compound molecular structure in the ice crystal that unit is not the (H-O-H)a molecule but the (H-O-H)6. It must be here stressed that while various methods depending on variations in the surface tension, in the boiling-point, in the latent heat, in the viscosity, in the cohesion, and in other physical properties, all agree that water (among other liquids) is abnormal and that the abnormality can always be explained by some form of association; the value of the degree of association is quite another matter. In fact all of the known methods give association values which are open to serious question. Their merits and demerits have n
n
Y
Y
B
A FIGB.SA and sB
often been discussed, and are reviewed by Tyrer’. Tyrer concludes that none of the usual methods is sufficiently theoretically sound to give a reliable value for the association factor. He also shows that a new boiling-point equation which he developes gives the association factor of water throughout its bulk (not at the surface only) as having a minimum value of six. It would be well a t this stage to ask the question, ‘(What do we really understand by liquid association?” A little consideration shows that the idea is far from self evident. For instance all liquids must be associated in a general sense by virtue of their being liquids, for in all liquids there must be attractions between the molecules and hence a statistical association. Smits submits very definite evidence that complexity in the liquid state is a “perfectly general’’ phenomenon, and the evidence indicates that even in the so-called normal liquids, such as benzene, there is an inner equilibrium between different aggregates of molecules. Further, Duclaux2 maintains that the equation of state can be deduced equally well by substituting the idea of molecular association, for van der Waals’ conception of internal pressure. Now in the crystal the atom is the only real unit of structure, and each atom under the attractions of its neighbours possesses zero resultant field. In the liquid state however, the molecule preserves its individuality, although the intermolecular forces may, and do, act through particular atoms. @.ping primarily to the greater molecular energy and the consequent greater mobility, liquid systems lack definite structure and the resultant field on any particular Mag., (6)20, (1910);2. physik. Chem., 50, 80 (1912);J. Phys. Chem., 19, 1‘‘Phil. 3.Ishy*. Radium, (6) 6, 199 (1925). 522
81
)*
1688
S. W. PENNYCUICK
molecule is not zero. We must conclude then that in any liquid, each molecule is at any instant more strongly attached to some one particular neighbour and less strongly to other neighbouring molecules; and that the whole system forms a three-dimensional continually-changing and branching network in which the linkings are of varying strength, and in which it is difficult to imagine the existence of a single or completely uncombined molecule even as a rare occurrence. Every normal liquid must possess this kind of association, but it is questionable whether the term “degree of msociation” can have any real quantitative meaning in these cases. Van der Waals’ considers that the molecules in normal liquids tend to form loosely combined groups which for an instant behave as single molecules. He calls this quasi-association, and further expresses the belief2 that no equation of state is possible unless one postulates an association of molecules to larger complexes. Further, Garver3 finds that the normal liquids are all associated; whilst Tyrer suggests that the quasi-associated molecules in any liquid may have a mean kinetic energy less than that of the external (gas) molecules, and suggests this as a real distinction between quasi-and true association. Accepting as we must, that normal liquids show a loose structure and through this a quasi-association, due inherently to intermolecular attractions, where then lies the difference in structure of the abnormal or so-called associated liquids? The abnormalities displayed by the latter as revealed by the various methods already mentioned, and the fact that these abnormalities always indicate a larger molecular weight than normal (although perhaps of uncertain or variable value), point undoubtedly to the conclusion that there is a real difference between their structure and that of normal liquids. Yet we have no reason for believing that such real association, in the case of water at all events, would be due to other than the usual auxiliary fields. The author suggests that real association does take place through auxiliary fields, but has a quantitative meaning, only when two or more united polar molecules present an abnormally weakened field to the surroundings and thus have independent existence within relatively wide time limits. The difference between associated and normal (quasi-associated) liquids then lies in this,though both show molecular combinations through auxiliary fields, the compound molecule in an associated liquid presents an abnormally weakened field to the surroundings, while the compound molecule (if the term may still be used) in a normal liquid possesses a strong field by means of which it tends to grow and to become immediately incorporated into the whole liquid system. As it is quite impossible to obtain any idea of the degree of this quasi-association, the latter liquids are referred to as normal. I n the case of associated liquids we might well imagine several molecules combining so as to neutralise their active fields by forming a ring, as in Fig. 6. Such a compound would behave as a solute molecule, moving through the solution as an independent unit, until such time as it was opened up and thus Proc. K. Akad. Wet. Amsterdam, 13, Pt. I, 107 (1910).
* “Die Zustandsgleichung” (1911). a
J. Phys. Chcm., 16, 454,669 (1912).
THE STRUCTURE OF WATER
benzene ring structure in a normal (quasiassociated) solvent. We might even regard water as an initial attempt of the H-0-H system to build up its hexagonal crystal structure; and whilst the
’
1689
1690
S. W. PENNY CUI CK
to be less active and thus exhibit smaller tendencies to satisfy their maximum coordination 2 . As the hydroxyl ends of the molecules are the more active, these liquids distinctly resemble water and form very similar coordination compounds. We should expect association to take place through these the strongest of the auxiliary fields, and hence by analogy with water we should expect, first, polar compounds as in Fig. 7A, and finally inactive ring compound molecules after the manner of Fig. 7B. 1
R
FIQ.7-4
FIQ.7R
As the general investigations on crystal structure have not as yet given any insight into the bonds between the molecules nor the relative atomic positions in solid alcohol, the above structure can only be regarded as a suggestive type of alcoholic association. Some of the organic acids however are especially interesting, for although the various quantitative methods for association show serious disagreement for most liquids, they agree remarkably well in pointing to the bimolecular state of various organic acids, particularly acetic (Tyrer). This agreement is probably due to the fact that whilst the other associated liquids are equilibrium mixtures of associated and non-associated molecules, i.e. are really solutions, acetic acid in particular exists almost wholly as (RCOOH),. These acids also show very marked association in the gaseous state, acetic acid again showing practically twice its normal molecular weight. When we add to the above, the evidence put forward by Innesl and by Brown and Bury2 that organic acids in hydrocarbon and other normal solvents, show, a t higher concentrations, abnormally large molecular weights which reach a maximum of twice the normal and never exceed this, then we must conclude that the bimolecular condition is an exceedingly favourable state for the organic acid molecule. Now, Brag@ shows that in the solid state organic acids are always united with their COOH ends in pairs; and there is in fact every evidence that the active fields of organic acid molecules are associated with these polar COOH groups. So that if we assign the strong auxiliary effects to the oxygen lone electron pairs and hydrogen nuclei of these groups, and imagine two J. Chem. SOC., 81, 682 (1902).
* J. Phys. Chem.,30, 694 (1926). * Chemistry and Industry, 1926, 245.
1691
THE STRUCTURE OF WATER
molecules to combine as in Fig. 8 (Lewis) we obtain a compound molecule which satisfies the necessary criteria for true association. It is of interest that we have the six kernel ring again, this time with two hydrogen nuclei. Owing to the organic acids exhibiting association even in the gaseous state, we must conclude that their auxiliary fields are exceedingly active, and the above rings very stable, and hence it is not surprising that in the liquid state acetic acid should show evidence of total polymerisation to form the molecule (RC0OH)z. R
R
4
Some of the Properties of Water The various properties of aqueous systems may now be considered in terms of the above form of association, and as is seen below the evidence is strikingly in favour both of ring association and of the activity of the oxygen lone electron pairs and the hydrogen nuclei of the water molecule. The Effect of Pressure on the System An increase in pressure decreases the abnormality, i.e. the association, and a t very high pressure, 2500 atmospheres, the liquid behaves as normal'. If association were due to the formation of more or less stable chains in the system, it would be reasonable to expect that pressure would increase and not decrease the association. If however, as is here suggested, molecular association is due to the relatively stable rings outlined above, and only to these rings, then, as the latter cause the structure to be more open, the effect of a pressure increase ie to compress and break these rings and thus cause the association to disappear. In the same way an increase of pressure on ice partially breaks the crystal ring structure and the ice melts.
The Maximum Density of Water The existence of a maximum density a t 4' may be connected with the above form of association in the following way. As Bragg has pointed out, the crystal structure of ice is extremely empty and it is very easy to imagine a less rigid arrangement of water molecules occupying less space. When ice melts a t oo, some of the associated rings still persist; this causes the structure to be more open and therefore the density to be less than if there were na such association. This is in agreement with the work of McLeod2 who shows that Bridgman: Proc. Am.Acad., 47, 538 (1912). Tram. Faraday SOC.,21, 145 (1925).
1692
S. W. PENNYCUICK
association (in the case of water) results in a gain in volume “owing to the bulky nature of the associated molecule.” As the temperature is raised the stability of these rings, and hence the association, decreases, and thus the system tends to contract or to become more dense. On the other hand there is the opposing effect, namely the increase of molecular energy with temperature and the consequent wider spacing of the attracting centres. It is very probable that the minimum shown a t 4’ on the temperature-volume curve is due to the alternate predominance of these opposing effects, and that from oo to 4’ the contraction due to the breaking of the associated rings is greater than the expansion due to the usual kinetic separation. Above 4’ the latter change predominates. The Latent Heat Although the crystal bonds are necessarily of great strength the energy required to melt one gram of ice is not abnormally great. This is evidently due to the fact that the process of fusion is not a breaking of all the intermolecular bonds; in fact few bonds may be actually broken, many may be loosened and some quite unaffected. On the other hand the process of vapourisation necessitates the breaking of all the intermolecular bonds, and as some of these have persisted from the solid state the very great latent heat of steam is not surprising. The value of the fraction of the latent heat that can be ascribed to the dissociation of the complex molecules a t the boiling point is invariably very high, and is believed to be about 3 1 5 of the total latent heat (Tyrer). The Effect of a Solute Speaking generally it may be said that a solute is held in solution because i t possesses auxiliary fields which the solvent is able to neutralise; and the extreme solvent power of water is thus due to the exceptional activity of the spare fields of the water molecule. Hydration and solvation not only accompany but are actually the fundamental cause of solution. Now when a substance is introduced into water, it will be attacked by the more active, Le. the unassociated molecules, and thus there will be a change in the water equilibrium and a decrease in the number of associated molecules. The decrease in the degree of association on the addition of a solute, can account satisfactorily for the corresponding lowering of the temperature of the maximum density of water, and the decrease in compressibi1ity.l Water of Crystallization The fact that the most numerous of the salts that contain water of crystal~ the total), are those that contain 6, 7 or 1 2 molecules lisation, (about 4 0 7 of of water has often been commented upon. Werner has incorporated these hydrates into his general co-ordination theory, and writes them
Rhodes2, however, points out the inadequacy of this theory as applied to these 1
See Qrucker: 2. physik. Chem., 52, 641 (1905). 122, 85 (1921).
* Chem. News,
I693
THE STRUCTURE O F WATER
compounds, and suggests formulae of the following type, where the six water molecules exist as a unit, Fig. 9. Such six H-0-H rings with all the auxiliary linkings through the oxygen atoms were first suggested by Kohlrausch. In view of the foregoing it is very probable that six H-0-H rings (with the auxiliary linkings as in Fig. SA) exist as units in the crystal. Crystals with twelve molecules of water contain two such rings, while crystals with seven molecules of water contain one such ring and have one odd water molecule, the latter assumption being based on the fact that hepta-hydrated salts have one molecule of water differentiated from the other six, e.g. ZnSOl. 7Hz0 on heating loses /OH\ 6 H 2 0 a t IOO', and the seventh water molecule at red heat. ---CI
11/
M
4
"i"
=-.E-----; - __ c1
The Self-ionisation of Water HoH I n the ice crystal it appears that all the H O H/HoH inter-atomic bonds are of equal strength (Bragg), and that therefore each hydrogen FIG.9 nucleus is equidistant from its two adjacent oxygens. Accordingly when the ice system is melted, the hexagonal associated (H-O-H)6 molecules, which are carried unbroken into the liquid system, are still held together as units by equivalent forces, or in other words the primary and the auxiliary bonds in the ring compound molecule (Fig. I O ) are indistinguishable. Such a compound molecule could then
\
\
H
FIG.I O
break (under equal chance) either as in A or as in B, Fig. IO, to give in each case exactly the same products, namely six molecules of water. There is however this interesting difference, six water molecules could unite as shown by the dotted lines in h above and then would stand an equal chance of breaking as in B, in which case each water molecule will have interchanged an hydrogen nucleus with a neighbouring molecule, or in other words, each hydrogen nucleus in the ring will have changed its auxiliary bond into a normal primary, and its primary into an auxiliary. Owing to the symmetrical structure of the above compounds, such an interchange could not of itself give rise to any abnormal or unique effects in the system. The greater part of water, however, does not consist of these associated and relatively stable rings, but of irregular polar chains (rings in the making), represented approximately as in Fig. I I .
'694
--
S. W. PENNYCUICK
A
: .. 0 : H
B w -
: 0 : 1 j L : ..O : H
---H
H
"
..
H
H
:..0 : H
FIG 1 1
The relative strengths of the primary and auxiliary bonds in these chains is of course quite unknown, but i t may be assumed (keeping the associated molecules in mind) that as the chain grows, the primary bonds grow weaker and the auxiliary stronger. In such a continually changing system, it may so happen, when conditions are especially favourable, that the polar chain instead of breaking a t its normal breaking point A, breaks a t B, the hydrogen nucleus concerned becoming attached to quite a different water molecule. The conditions would then be represented, momentarily and very incompletely bv
Such a rupture would account for the self-ionisation of water. It is evidently of very rare occurrence, for the ionisation constant of water indicates that a t ordinary temperatures only one hydrogen nucleus out of every I ,ooo,ooo,ooo is so transferred at any instant. The Di-electric Constant Although the rupture of polar chains to produce the (hydrated) hydrogen and hydroxyl ions is very rare, the normal rupture of such chains must be of remarkably frequent occurrence. When such a normal rupture occurs, e.g. a t the arrow head in the simple chain
..
1I
.. H : O : H : O :
H
..
H
the binding hydrogen must be displaced for a very short time from its normal position in its own particular water molecule. Such a continually occurring displacement accounts for the high d.e.k. of water. It is of interest to notice that of the other common liquids which show high di-electric constants, the chief are the simpler alcohols and the simpler fatty acids. These all possess lone oxygen electron pairs and ionisable hydrogen nuclei, and as has been shown probably combine through such in the liquid state. Their high di-electric constants are then probably due to a similar cause to that outlined above, namely to a short-time displacement of a hydrogen nucleus under the influence of the field due to an oxygen lone electron pair.
THE STRUCTURE OF WATER
1695
The Abnormal Velocities of Hydrogen and Hydroxyl Ions In virtue of the foregoing it is evident that the ions of water, cannot be dissociated from the contiguous water molecules which brought them into existence, or in other words these ions are always hydrated. In the same way the hydrogen ions of acid solutions are always directly associated with the surrounding water molecules. A hydrogen ion would then be more correctly written in some such form as
L
H
H
and an hydroxyl ion as
J
.. : .. O:H:O .. : H : O : H H H
where the ionic groups are continually increasing or decreasing the amount of water with which they are associated. Now, in these hydrated ions i t is impossible to say which of the (removable) hydrogen nuclei (four above), may be taken as representing the hydrogen ion, or which of the removable hydroxyl radicals (two above) may be the hydroxyl ion, rather must we assume the whole hydrate to carry the positive or negative charge. Under the stress of a difference of potential these ionic groups migrate towards the two poles with probably normal ionic velocities; but the difference of potential may have (in the case of these two particular ions) an added abnormal effect. For if A represent
1! +
A
B
the positive pole of an applied E.M.F. and B a hydrated hydrogen ion, then of the four removable hydrogen nuclei, those furthest from A will be the more
readily removed or the more active; and during the normal continual hydration and dehydration, water will be added on a t this active end and given up at the end nearest to A. This is equivalent to an abnormal increase in the velocity of the hydrogenion. A similar explanation holds with the hydroxyl ion. It will be noticed that whilst other ions, such as Ka+ and Ca++ (undoubtedly hydrated), have an individual existence in aqueous systems, the hydrogen and hydroxyl ions in such systems do not exist as unchanging individuals, but, owing to the nature of the solvent, the particular nucleus which may be arbitrarily taken as representing a hydrogen ion a t any one
1696
S. W. PENNYCUICK
instant, may exist a t another as part of a water molecule, the liberated hydrogen nucleus becoming the hydrogen ion. Perhaps we have here the most common example of tautomeric hydrogen. The catalytic effect of the hydrogen ion on hydrolyses in general, is probably intimately connected with this tautomeric change; for a water molecule undergoing such a change would be in a peculiarly unstable (loosened) condition, and as such would be particularly open to internal rupture'.
Summary2 The wide, general evidence of the activity of the auxiliary fields of the hydrogen and oxygen atoms in the water molecule, and its bearing on the idea of the association of such molecules is discussed. The difference between quasi- and true association is pointed out, and it is concluded, in conformity with the X-ray structure of ice, that the associated molecule in liquid water is the (HZ0)O molecule, and that water is a mixture of relatively stable compound molecules of benzene ring structure in a normal (quasi-associated) solvent. The ring structure of associated acetic acid molecules is shown to be in conformity with this view. The evidence derived from the effect of pressure on aqueous systems, the addition of a solute, the maximum density of water, the latent heat, the dielectric constant, the self-ionization, and the abnormal velocities of the hydrogen and hydroxyl ions, is shown t o be strikingly in favour of ring association and of the abnormal activity of the auxiliary fields of the oxygen and hydrogen atoms. Physical Chemistry Laboratories, University of Adelaide, S o u t h Australia. See Pennycuick: J. Am. Chem. SOC.,48, 6 (1926). * M y attention has recently been called to a paper by Simons and Hildebrand: J. Am.
Chem. Sot., 2183, (1924) where it has been shown from vapour density measurements, that hydrogen fluoride vapour consists of an equilibrium mixture of H F and (HF),. I n conjunction with the foregoing, this is very significant, particularly when one considers the close similarity in electronic structure between the hydrogen fluoride molecule and the water molecule.
