T H E PROBLEM OF T H E SURFACE TENSION OF MERCURY AND T H E ACTION OF AQUEOUS SOLUTIONS ON A MERCURY SURFACE.* BY R. S. BURDON AND M. L. OLIPHANT. (Receizied January I I th I 9 2 7 .) The following is a brief survey of the present position of the knowledge concerning the value of the surface tension of mercury together with an account of further experiments on the spreading of aqueous solutions of acid upon a mercury surface and observations of the effect of different gases on the spreading of water and of dilute acids. Apparent contradictions of the theory of Hardy and Harkins as to the spreading coefficient are also pointed out and it is shown how Antonow’s rule can be reconciled with the observed phenomena. The Surface Tension of Mercury. This has received more attention than that of any other substance except possibly water.The wealth of papers on the subject is equalled by the remarkable lack of uniformity in the values obtained. The subject is of interest at the present juncture because of the suggestion of certain workers,l* 2 that the behaviour of a mercury surface will eventually be explained in terms of orientation of the atoms. Many writers have remarked on the extreme difficulty of obtaining and keeping a “clean” mercury surface but it is probable that in the Fast too much of the diversity in the values obtained for the surface tension of mercury has been attributed to contamination and faulty manipulation. This is emphasised by the fact that more recent workers taking the most elaborate precautions do not obtain results in agreement with each other.In considering the great number of papers on this subject two de-terminations that must receive serious consideration are those of HarkinsY3 by the method of drop weight ; and of PopescoY1 by the method of the large drop. The latter method depends only on the difference in height of the top and the equator of a large circular drop of mercury resting on a flat plate or in a circular depression. In such a case the form of the upper part of the drop (which alone is used in measurement) must be independent of angle of contact with the lower plate. The drop-weight method is also claimed to be independent of contact angle and both are classed as ‘‘ static ” methods4 For mercury against its own vapour Harkins obtains a value * A contribution to the General Discussion on Phenomena at Interfaces.-See Vol.1 Popesco A m . de Phys. 1925 3 402. 2 Perucca Comptes Rendus 1922 175 519. 3 Harkins 3’. Am. Chem. SOC. 1920. 4 Rideal Intro. to Surface Chemistry 4. XXII. p. 433. 20 206 PROBLEM OF THE SURFACE TENSION OF MERCURY 474 dynes per cm. while Popesco’s value is 436.3 dynes. I n dry air, Harkins obtained values lying between 459 and 464 while Popesco’s value for a drop formed in dry air is 5 1 7 five seconds after formation the value falling rapidlyat first and then more slowly to approximately 440 after one hour. If the drop was formed in vacuo however the subsequent admission of air caused no rise in surface tension. For drops formed in air at less than atmospheric pressure the initial surface tension appears to vary approximately linearly with the pressure of the air.Popesco’s explanation is essentially that 436 dynes/cm. is the value for a mercury surface in which the atoms have become oriented and have reached a fairly stable state. The presence of gas however tends to prevent orientation by the impact of gas molecules on the surface and hence gives larger values for the surface tension. In a gas the orientation of the surface atoms goes on progressively and most rapidly in hydrogen The values obtained by Popesco 5 and 10 seconds after formation in Hq 02, and N are : Gas. After 5 secs. After 10 secs. H2 5 1 0 47 7 0 2 525 505 N2 540 5 24 Meyer in 1898 using the method of ripples obtained values that can be reconciled with Popesco’s measurements but not with his conclusions.Meyer’s values would be for a surface only a small fraction of a second after its formation and hence if the only effect of the gas is to impede orientation Meyer’s values should be higher for all gases. His values actually are higher for H2(554) but lower for 0,(505) and N2(504). Now though it is most im-probable that an adsorbed layer should have the effect of raising surface tension yet a comparison of Meyer’s and Popesco’s figures indicates that possibly a mercury surface on exposure to a gas grows rapidly in surface tension (reaching a maximum much sooner in H than in other gases) and then falls. In any case it becomes of great interest to know what happens to the surface tension of a mercury surface during the first five seconds after its formation.After checking by means of the apparatus described later ( p 210) that Popesco’s values could be reproduced to a close approxima-tion the authors endeavoured to follow the behaviour of mercury under various times of exposure to dry air (or other gases) by means of the drop-weight method. An apparatus was manufactured to enable mercury drops to be formed from the same tip either in gases or in vacuo at any desired rate. Independently of theory as to the connecticn between drop weight and absolute value of the surface tension it may be assumed that drops of the same liquid formed at equal rates from the same tip should have weights whose differences are very nearly proportional to the differences in surface tension.5 Hence by taking weights for drops formed at say one per second in dry air and then in vacuo it was expected to detect the effect of one second’s exposure to air.By varying the drop rate the effects of different times of exposure would be found. ResuZts and Discussion.-The weights for ten drops were found to be very consistent but for all rates of formation from three drops per second to one drop in 30 seconds the weight of a drop formed in air exceeded the weight of a drop formed in vacuo at the same rate by only approximately 5 Not exactly since the surface tension changes without change of density and hence drops formed at the same rate are not exactly similar in form. (Rideal p. 14. R. S. BURDON AND M. L. OLIPHANT 207 one per cent. I t was of course known that previous workers Harkins,l Hogness and Cenac had found very small differences between the drop weight of mercury in air and vacuo but it seemed possible that the long time taken in forming the drop might be a cause of this since the surface tension falls rapidly with exposure to air.The present work shows that the drop-weight method fails entirely to show up differences in the value for air and vacuum though these differences are shown up by the fact that a “large drop” of mercury is always deeper when formed in air than in vacuo. The negative result may mean that there is something wrong with the drop-weight method as applied to mercury or that the effect of air on the surface is really not present in the drop-weight method. The latter alternative is not impossible since the formation of a drop at the end of a tube involves the continual expansion of the surface and even where the drop is formed very slowly the actual motion of breaking away is quite rapid.A number of observers have commented on the way in which an expanding mercury surface differs in properties from a stationary surface,* A drop of water which fails to spread when placed on a mercury surface is often caused to spread by the action of pouring the mercury out of the dish, this involving the creation of new surface. Regarding the absolute value of the surface tension of mercury in vacuo it may be said that different observers using the large drop method agree very closely with 436 dynes/cm. Results from the method of maximum bubble pressure (in H2) agree generally with those of the drop-weight method while measurement of rippIes and waves on a jet tend to the value found by the big drop method.Concerning the drop-weight method as applied to mercury two points call for comment. Firstly the method depends on the assumption that the capillary rise method gives the correct value for the surface tension of water, but the fact that careful workers9 have obtained values which differ with the kind of glass used for the tube does not seem to have been disposed of by modern workers. Secondly the method has been based on measurements of liquid drops of approximately unit density which hang from the outer circumference of the tube and there does not seem to be much experimental evidence for the belief that drops will have the same form for liquids of high density that do not wet the tube.At present it may be said that the big drop method is the simplest for following changes of surface tension but it is not possible to take a reading for some seconds after the drop is formed. A method of getting over this difficulty is still being sought. As to absolute values there does not appear to be any objection to the mathematical basis of the method but it is unfortunate that it has been used very little except for molten metals (chiefly mercury). Measurements on less dense sub-stances by this method would be valuable. The drop-weight method for mercury must still be held suspect until an explanation is found for the approximate equality of drop weight in air and vacuo. Aqueous Solutions on the Surface of Mercury.I n a previous paper one of us 10 described some phenomena presented The following by solutions spreading over the surface of mercury in air. is a report of further work on this subject. J. Am. Chern. SOC. 43 (1921) 1621. 8 Schofield Phil. Mug. March 1926. Ann. Chim. et Phys. (rgq) 299. g Volkmann Ann. Physik 66 1898. lo Proc. physical SOC. 38 Feb. 1926 208 PROBLEM OF THE SURFACE TENSION OF MERCURY It was shown that pure water spreads very slowly solutions of neutral inorganic salts and of acids rapidly and solutions of inorganic bases not at all. Drops of very dilute solutions of inorganic acids spread rapidly to cover a definite area and then stopped the area covered during the rapid stage being approximately I sq.cm. for each lol* atoms of monobasic acid present the figures being for HNO 0.90 x 1 0 ~ ~ atoms per sq. cm. for HC1 1-0 x 1014 and for HBr 1.2 x 1014 each molecule of acid covering approximately ten atoms of mercury. A number of acids both organic and inorganic have now been tested using '0001 normal solutions and in general the differences have been much less than those indicated above. In fact for all the acids tested it may be said that a drop of -0001 normal solution placed on a clean mercury surface spreads rapidly to occupy an area very closely equal to I sq. cm for each 1014H ions given by the com-plete neutralisation of the acid. The area covered per molecule for H,S04 and other dibasic acids is twice as great as for HCl moreover the solutions of the organic acids oxalic succinic and phthalic spread as fast and cover as great an area as the sulphuric acid.In order to test whether any progressive change in behaviour occurred with progressive change in the acid molecule readings were taken with -0001 normal solutions of the five organic acids formic acetic propionic, butyric and valerianic using drops from the same pipette and observing the area covered during the rapid stage of spreading. The following results were obtained : Number of Molecules per sq. cm. covered x 1014. Acid. Formic H.COOH . . -96 Acetic H.CH,.COOH . . 1'02 Propionic H(CH,),COOH . . *98 Valeric H(CH,),COOH . . - '97 Butyric H(CH,),COOH . . . -92 The differences obtained are only of the same order as the possible experi-mental errors involved in the dilution and measurement of area.Even at the dilutions used these acids differ by over 50 per cent in their degrees of ionisation yet they appear identical in their behaviour on a mercury surface. A drop of any one of them within about one second spreads to a definite area and stops the action being completed approxi-mately as quickly for valeric acid as for hydrochloric. This may indicate either that the neutral molecule is concerned in the spreading or that the mercury absorbs the ions available with extreme speed and fresh molecules then dissociate. EdectricaZb Aided Spreading.-The very general appearance of the number 10 14 ions per sq. cm. directed attention to one of the eiectrical phenomena recorded in the early paper viz. that if the positive terminal of a battery be placed in the mercury and the other in the drop of solution, then spreading is caused if the mercury be positive but retarded if the mercury be negative.(In the latter case a forced spreading occurs if the voltage be high enough the drop immediately contracting when the circuit is broken.) I t was decided to try the effect of forcing 1014 electrons through a drop of distilled water on the surface of mercury. A platinum terminal was placed in the mercury and a liquid terminallo placed in a drop of water on the mercury surface. First of all the effect of discharging a condenser through the drop was tried but the results were not very consistent largely owing to the uncertainty as to when the condenser ha R. S. BURDON AND M.L. OLIPHANT 209 completely discharged. Next a battery safety resistance and microamnieter were placed in series in the circuit. A drop of distilled water was placed on the mercury the terminals inserted and current switched on for ten seconds and then the area of the drop was measured. The difference between this area and that of a similar drop left on open circuit for the same time gave the extra area covered owing to the current. A current of 16 microamps. caused an increase in area of one sq. cm. per second for pure water on a clean mercury surface. But one electron = I ‘5 7 x 10 - 19 coulombs and therefore a current of 16 microamps. in one second delivers : 16 x 10 - 6 -+ 1.57 x 10-l9 = 1.0 x ro14 electrons. I t may be then that absolutely pure water is perfectly neutral so far as its tendency to spread on mercury is concerned but that it will spread to cover one sq.cm. for each 10 l4 monovalent negative ions present whether these are provided by an acid dissolved in the water or whether they are sorted out from the water itself by means of an electric current. If any factors are present to prevent spreading e.g bases in solution or contamina-tion on the mercury surface then the areas covered are much less. The rate of spreading without current is not affected by earthing the mercury or the solution or both. Mechanism of Spreading. I t is difficult to picture any mechanism accounting for spreading since it is practically inconceivable that all the acid molecules in a drop of solution could reach the surface in the short time involved.I t was first suggested by Mr. A. V. SlaterlO that the definite area covered was due to adsorption of ions with attendant water molecules thus accounting for the large area covered. At first the fact that the drops spread to an appreciably larger area on warm mercury seemed to contradict this hypothesis but it is quite possible that at the higher temperature the water itself supplies the extra ions. The increased spreading due to this would not depend on the number of extra ions present in the water at any instant but on the rate at which these ions when absorbed at the surface are replaced by further dis-sociation of the water. It seems necessary to postulate an action analogous to the ‘( squeezing action ” suggested by Edser to account for the spreading.The adsorbed ionic groups with the mercury atoms to which they are attached are free to move on the surface with only fluid friction and hence if the attraction of the underlying mercury surface extends beyond the primary adsorbed layer this latter will be jostled out and spread will occur and continue so long as the supply of negative ions lasts. Two observations support this view : (I) So far as can be seen by eye the drop spreads more slowly at first, the rate of growth of diameter increasing with the area covered. Now if spreading is due to any action at the edge of the spreading drop then at most the diameter should increase uniformly whereas if spreading is due to forces between the comparatively more distant parts of the drop exerting a I ‘ squeezing-out ” action then the linear rate of spreading should increase with the area over which this squeezing action can occur.I t is proposed later to determine if possible the connection between instantaneous area and rate of spread by means of a cinema picture of the spreading drop 210 PROBLEM OF THE SURFACE TENSION OF MERCURY (2) If when one drop has spread to a disc and stopped a second drop of solution be placed on the middle of the disc then the solution spreads to cover double the area in a time shorter than that taken by the spreading of the first drop. Lycopodium sprinkled on the first disc indicates that the spreading takes place from the centre as would be expected on the squeez-ing theory. On this view adsorption takes place on a continually renewed surface, as the adsorbed ions move outwards carrying with them the mercury atoms to which they are attached.This seems in accord with the observation that water will spread on a mercury surface for which the spreading coefficient of Hardy and Harkins is negative. This point will be discussed later. If there already exists on the surface an adsorbed film of liquid or solid, due to contamination etc. spread will only occur so far as the spreading iu FIQ. I. substance is able to compress this film. fi the adsorbed film is gaseous the life of any gas molecule in the surface is exceedingly short at ordinary temperatures and it is easily replaced by a more stable ion or molecule so that even distilled water will spread slowly on a surface which has been in contact with a gas for 18 hours.The adsorbed layer of negative ions may be covered with a secondary layer of positive ions so that the whole phenomenon is equivalent to an adsorption of the neutral molecule as found by Schofield8 and others. Some light has been thrown on these prob-lems by experiments made with the apparatus described below by means of which it is possible to form surfaces of mercury under a variety of conditions and to observe the be-haviour of water or other liquids dropped upon them. Simultaneous measurements of the height of the big drop of mercury used are also possible so that the changes in surface tension produced during spreading can be fol-lowed. The vessel “ A ” is a large Pyrex beaker with the bottom removed and plate glass end-plates cemented on with red sealing wax.A large drop of mercury can be prepared on the slightly concave surface ‘‘ S ” by turning on the tap ‘(T” and allowing mercury to flow in from the reservoir “R”. The tap ‘‘ T ” is about 80 cms. below the level of the side tube “ L,” so that although it is impossible to grease the tap no air will leak in even when the apparatus is evacuated. The tube “ M ” is connected through a two-way tap and drying tubes to a vacuum pump and also to a reservoir of gas pro-vided with purifying and drying tubes. The ground glass join “ G ” is mercury sealed and can be closed by a ground cone of glass or provided with a pipette c c P ” as shown so that drops of liquid can be allowed to fall upon the mercury surface without admitting any trace of air.After any experiment the mercury is drained away from “ S ” through the side tube ‘‘ L ” and the glass plate repeatedly washed with clean mercury. The plate glass windows allow the height of the drop to be photographe R. S. BURDON AND M. L. OLIPHANT 2 1 1 in the manner described by P0pesco.l The whole apparatus is mounted on a levelling table so that the concave surface '( S " can be set truly horizontal. Although thick-walled rubber tubing and a clip are used to admit liquids through the pipette (as it is not possible to grease a tap) it was found that by tightly screwing up this clip the apparatus could be evacuated to a '' black vacuum " by means of a condensation pump and a high degree of evacuation could be maintained for hours without running the pump.The mercury surface found in a dried and purified gas is remarkably free from contamination and water spreads rapidly and uniformly right up to and even over the edge. The argon used in the experiments was purified by passage over heated metallic calcium. The oxygen was a carefully dried commercial product but otherwise was used as it left the cylinder. Results. The E&t of Gases OH Spreading.-Water spreads rapidly on a surface prepared in pure dry air or argon but spreads even more rapidly on a sur-face prepared in oxygen. Oxygen appears to play an important part in the spreading in many cases. For instance when a solution of HCl which is not too dilute spreads and evapprates in air or oxygen a white film of calomel is left behind on the surface.This film does not appear to con-taminate the surface to any large extent for a second drop of acid will spread pushing the solid layer to one side. On the other hand if the drop is formed in argon that is free from oxygen and the HC1 is placed on the surface it spreads as before and evaporates but no visible deposit is formed, though the behaviour of a second drop of acid shows that the surface is badly contaminated. The effects of oxygen are most marked when the mercury surface is prepared in vacuo and the gas then admitted. The Spreading Coeficient and Antonow's Rule.-It appears at first that neither of these rules is applicable in general to mercury-water or to mercury-acid solution interfaces. For instance water spreads well on a surface of mercury freshly formed in a gas such as air or argon and if we accept the values for the surface tension of mercury given by the large drop method viz.500 dyneslcm. for a fresh surface in air 427 for the interfacial tension water-mercury the spreading coefficient of Hardy and Harkins has a positive value : However for a drop prepared in vacuo gas being subsequently admitted (a = 436) the spreading coefficient must have a negative value of about 60. Spread always occurs though very much more sluggishly than in the first case. If the gas admitted be argon the surface tension possesses the same initial value of 436 dynes/cm. but spread does not now occur although the drops of water flatten slightly. I f oxygen be admitted spread occurs quite rapidly so that it appears that here again oxygen is a controlling factor.Water also shows a positive tendency to spread in vacuo on a drop formed in vacuo. When water spreads on a surface prepared in vacuo or on mercury which has been exposed to air for some time the spreading coefficient is negative and the water must be spreading against a pressure of some 60 dynes/cm. if the interfacial tension possesses its normal value and the ad-sorbed film of water molecules is hence subjected to a lateral compression of this amount. As has already been pointed out it is probable that sprea 2 1 2 PROBLEM OF THE SURFACE TENSION OF MERCURY occurs always on a freshly renewed surface so that it would seem permis-sible to assume the normal value of the interfacial tension.In several ex-periments on spreading in such cases a rise in the height of the large drop, indicating an increase of surface tension has been noted. This rise is very small of the order of 4 or 5 dynes only but it is consistently observed for older drops. It is difficult to form a satisfactory idea of the processes musing this increase in surface tension of the mercury. Mr. R. Mitton working in this laboratory has measured by means of a sensitive torsion balance the pressure exerted by spreading drops of acid on mercury and finds for the spreading coefficient a value of 30 or 40 dynes which is in general agreement with the above conclusions. In the various experiments in which water has been observed to spread on mercury the surface tension of the latter has varied between about 500 495 490 $ 485 ti al 480 * 475 470 x w c x 460 0 1 2 3 4 5 6 7 8 9 Time in minutes.(I) Graph of fall of S.T. of a large drop of Hg in dry air. (2) Graph of fall of S.T. of a large drop of Hg with a droplet of water spreading on its surface. FIG. 2. dyneslcm. for a freshly prepared surface in argon or air to about 400 for a surface which has stood in contact with a gas for some hours. Antonow’s rule states that for a liquid which spreads on mercury the interfacial tension between it and mercury is equal to the difference between their individual surface tensions. The interfacial tension would then have to vary between 427 and 327 dynes/cm. which is improbable but if we conclude that spread always occurs on a continually renewed surface the difficulty disappears to a large extent for the S.T.of a freshly formed surface is greater than 500 dyneslcm. and the interfacial tension possesses its normal value. Conclusion. The present paper is written as a contribution to a general discussion on surface phenomena. I t draws attention to outstanding difficulties with regard to the surface tension of mercury; it may be said to be impro K. S. BURDON AND M. L. OLIPHANT 213 bable that agreement as to the absolute value of this quantity will ever be reached by mere repetition of measurements. The outstanding need at present is for a theoretical basis to account at least qualitatively for the differences between accurate determinations made by different methods. The account given of further work on the spreading of aqueous solu-tions on mercury suggests on the one hand the existence of forces exerted over greater than molecular distances and on the other hand further sup-ports the view that the ion is the determining factor in the spreading of these liquids. In conclusion it is desired to express our thanks to Mr. H. Finlayson for advice and for assistance in obtaining pure samples of the acids used, and to Mr. F. J. Wauchope for much assistance in connection with both apparatus and cbservations. Ih Uziversity of AdeZaide