]\gt ; On the Determination of Viscosity at High Temperatures .
By E. FAWSITT , D.Sc .
, Ph. D. , Lecturer on Metallurgical Chemistry in the University of Glasgow .
( Communicated by Professor Andrew Gray , F.R.S. Received January 22 , \mdash ; Read February 13 , 1908 .
) A large number of viscosity determinations have been carried out with liquids at temperatures near the ordinary room temperature , and some observations at lower and higher temperatures have also been made by similar methods to those adopted for working at the usual temperatures .
easurements at temperatures higher than to C. present considerable difficulties , and until the present year this subject had not even been by experimenters .
As I was anxious to determine the viscosity of substances which melt at temperatures up to a white heat , I lately set myself to work out a satisfactory method of determination .
The present communication contains a description of the method used .
The method is suitable for the measurement of the viscosity of liquids which are not very viscous\mdash ; not more than say times as viscous as water , and is especially designed for the determination of the viscosity of molten metals and salts .
The determination of the viscosity of salts up to 1200o C. , or even higher , can be quite satisfactorily carried out by this method .
Determinations of the viscosity of metals are much more difficult , owing to the impossibility of preventing a certain amount of face oxidation .
The smallest trace of surface oxidation will completely spoil a series of observations , and the prevention of oxidation is really the chief difficulty in such determinations .
Previous Work in this Branch .
The usual method adopted for the determination of viscosity at ordinary temperatures depends on the rate of flow of the liquid through a capillary tube .
The determination is easy to carry out , the calculation of the viscosity from the time of flow is simple , and the experimental results are in close agreement with the laws of flow deduced from theoretical considerations .
The glass capillary has been used up to C. by Beck* for the determination of the of fused mercury salts .
With a Jena glass tube the method has been used as high as 60 C. for the determination of 'Zeit .
fiir Physikal .
' 1907 , vol. 58 , pp. 425\mdash ; 441 .
On the Determ , of Viscosity at High Temperatures .
291 the viscosity of lead salts .
* A platinum capillary has been used as high as C. by Goodwin and Mailey .
It is probable that this is about the limit of temperature to which the capillary method may be used , and in any case it did not appear that this method was likely to be employed with good results for metals of even comparatively low melting point .
A rough method of determining the viscosity of molten silicates , depending on the rate at which a platinum wire sinks in the liquid , has been used by Doelter .
An advance on this is an interesting apparatus used by Arndt , S who measured the rate at which a platinum cylinder fell in the liquid .
These methods of Doelter and Arndt can only be used for very viscous substances\mdash ; for liquids whose viscosity is more than 1000 times that of water .
They are of no use for measurements on ordinary salts and metals , which appear , in general , to have a viscosity not more than 100 times that of water .
Method .
The method used is based on the method originally given by the modiiications introduced being due to the special nature of the determinations .
In Coulomb 's method a horizontal disc is allowed to execute horizontal vibrations about a vertical suspeIjding wire attached to its centre .
The viscosity of the liquid can be calculated from the rate of decay of amplitude .
In recent years this method has not been used to reat extent for ordinary viscosity determinations , and this can probably be explained from the following facts:\mdash ; ( 1 ) The disc used has to be fairly large ( 15 to 20 cm .
diameter is usual ) , and this involves the use of rather large quantities of liquid ; ( 2 ) the calculation of viscosity from the observations of experiment is much more involved than in the case of the capillary method ; ( 3 ) an error made in the observations produces a much greater error in the ( calculated ) value of the viscosity .
While it is not possible to rid of these last two difficultieH , this method is nevertheless capable of good results , and the tinle necessary to carry out a determination is certainly less than that occupied by a determination according to the capillary tube method .
The apparatus as employed by Coulomb consisted of a vertical rigid metal rod several inches in length , which was fixed at its upper end into the centre * Lorenz and Kalmus , ' Zeit .
fur Physikal .
Chem 1907 , vol. ) , pp. ) 'Amer .
Chem. Soc. Trans 1907 , vol. 11 , pp. 211\mdash ; 223 .
Sitz .
der Kais .
Akad .
Wiss .
Wien , ' 1905 , vol. 114 , p. 629 .
S 'Zeit .
fur Elektrochem , ' 1907 , vol. 13 , pp. 578\mdash ; 582 .
'Mem .
de l'Institut Not .
des Sciences et Arts , ' 1800 , vol. 111 , p. 246 .
Dr. C. E. Fawsitt .
On the nination of [ Jan. 22 , of a graduated metal disc , and at its lower end into another disc which was ungraduated .
The centre of the upper disc was provided with a small screw arrangement for the reception of the suspension wire , the upper end of which was fixed to a rigid support .
The apparatus was first of all made to oscillate in air , and the logarithmic decrement of the amplitude obtained .
The lower disc was then submerged in a number of liquids , and the logarithmic decrements noted .
After deducting the value for air from that obtained for each liquid , it was possible to compare the viscosities of liquids whose densities were known .
The method has been considerably improved and made much more sensitive by Stokes , *Clerk Maxwell , and Meyer , and it has been shown that the viscosity of a liquid is given by the equation , ( 1 ) where and are the logalithmic decrements of the apparatus in the liquid and in air , , and are constants depending on the apparatus , and is the density of the liquid .
This method is found to give results which agree well with eory except in the case of very viscous liquids .
In adapting this method to high temperature work the following difficulties have to be encountered:\mdash ; ( 1 ) It is not practicable , when working on a laboratory scale , to keep large quantities of liquid at a high temperature ; it is desirable , therefore , to make the method work with an amount of liquid which will be taken by a crucible , say 3 inches deep and inches in diameter .
The dimensions of the immersed solid must therefore be small .
( 2 ) The solid whose oscillations are to be damped , and other ompanying parts , cannot be made too small , as the sensibility of the apparatus is thereby decreased , and aJso as the molten salts or metals corrode the parts immersed too rapidly .
( 3 ) The liquids , especially the metals , havs a comparatively high density , and the apparatus must be made to sink in these to the proper depth .
A consideration of the decay of amplitude in torsional oscillation , when applied to viscosity determinations , shows that it is made up of three parts : the first due to the defective elasticity of the wire , the second to the viscosity of the liquid , and the third due to the loss of energy consequent on a simple movement of translation being imparted to the liquid .
If 'Camb .
Phil. Soc. Trans 1850 , vol. 9 , II , p. 8 .
'Phil .
Trans 1866 , vol. 156 , pp. 249\mdash ; 268 .
'Pogg .
Ann 1866 , vol. 126 , p. 177 .
1908 .
] Viscosity High Temperatures .
represent the amplitude and the time , the decrease in amplitude from these causes is iven as follows:\mdash ; - where , and are constants .
Combining these we have . .
( 2 ) If A is small , and if is small , i.e. , if the translation movement communicated to the liquid is small , or ( ) where and are the values of A at the beginning and after time .
In this case the logarithmic decrement for successive oscillations is constant .
If the velocity of movement of the liquid is considerable , the last term cannot be neglected and the integral has the form .
( 4 ) In this case the arithmic decrement is not constant for any one series of observations , but varies with the amplitude .
As it is necessary that the oscillations should conform to equation ( 3 ) , the of motion communicated to the liquid has to be kept to a .
It follows that the body submerged must be of such a form that a horizontal section throu , point will be circular .
A circular disc , a sphere or a cylinder are all quite suitable , but after some trials with these I have returned to the disc , it best suited to the present purpose .
The diso used was 26 mm. in diameter and 1 to 3 mm. thick .
For temperatures up to 40 C. the diameter of the stem ( which forms the axis of rotation ) was 1 to 3 mm. ; for higher temperatures it was to mm. It is very important that the disc should be an exact circle , and should rotate truly about its axis .
It is not possible to obtain for the manufacture of the disc any material which is heavy enough to sink in metals , and at the same time resistant enough to stand the corrosion and heat of the liquid in which it is placed .
The only way out of this difficulty seemed to be to weight the apparatus externally .
The amount of this weight and its distance from the disc must be snch as to keep the centre of gravity of the rigid part of the apparatus low possible , otherwise the apparatus will not swing evenly of Dr. C. E. FawsitL On the of [ Jan. 22 , any particle of the disc becomes an ellipse instead of a circle , thus making the motion correspond to equation ( 4 ) instead of to equationl ( 3 ) .
The wei , ht used was a small iron cylinder clamped on to the stem about to 2 inches above the disc .
For temperatures up to C. iron is quite suitable as a material for the disc .
For temperatures up to a white heat one may use either fireclay , or a mixture of fireclay and plumbago .
In making a series of observations with this apparatus , the disc is allowed to sink about half an inch below the surface of the liquid .
The amplitude of the oscillations is indicated by a pointer ( wire ) at right to the top of the iron rod which carries the disc , and the pointer moves above a circular scale divided periments varied betwee , and 9 seconds .
The period of the oscillations in the liquids used hardly val.ied at all .
The period in the case of mercury was only per cent. greater than in the case of water .
1908 .
] Viscosity at High The Calculation of the from an } of the Logarithmic Decrement .
In obtaining the logarithmic decrement , Briggs ' logarithms have been used throughout .
The constants , and in equation ( 1 ) are best determined empirically from determinations on three liquids of known viscosity and density .
Knowing , and , the viscosity of any other liquid can be determined by solving the quadratic equation where The expansion of the disc at high temperatures is small , but a slight correction has to be made on account of the slightly larger surface in contact with the liquid .
If .
be the radius of the disc , the surface in contact with the liquid is approximately equal to The coefficient of expansion for the surface is to twice the coefficient of linear expansion for the material composing the disc .
The coefficient of linear expansion for iron is , and for fireclay it is probably about .
Each unit of face will therefore increase by per in the one case and in the other .
The logarithmic decrenlent is proportional to the surface in contact with the liquid , and a slight correction must therefore be made on this according to the temperature of experiment .
nentat .
There has occasionally existed in the minds of previous workers with the oscillation method some doubt as to whether this was a real test of the internal friction of the liquid .
If any slipping takes place at the surface of contact between the solid and the liquid , the method is really of very little use .
This point was therefore tested first of all .
The experiments show that slipping does not take place to any detectable extent ; this holds good even for liquids like mercury , which do not " " wet\ldquo ; the solid .
We can therefore assume that the layer of liquid next the solid moves with the same velocity as the solid itself .
If any slipping took place the amount of slip would be different for different liquids and for different solid surfaces .
A number of different discs gave the following results:\mdash ; VOL. LXXX.\mdash ; A. Dr. C. E. Fawsitt : On the of [ Jan. 22 , Table I. Material of disc .
rough , smooth , with smooth s Any differences in the results of Cases 1 , 2 , and 3 may be put down to experimental error .
The somewhat higher figures obtained in the fourth case are due to the disc being thicker ; it will be noticed that the proportionate increase is practically the same for all the liquids .
It is assumed from these results that there is no slipping at the surface of contact .
The method was next tested with liquids of known cosity .
Taking three of these as standards the viscosity of the others was calculated .
The results are given in the next table .
The values of density and viscosity in the table are taken from Landolt and Bornstein 's tables with the exception of some density values obtained from the researches of Patterson .
* Table II.\mdash ; Iron Disc 26 mm. diameter .
Period of oscillation , seconds .
Using water , chloroform , and mercury to calculate the constants from , we have We may now use these values to calculate the viscosity of benzene and alcohol from the observed logarithmic decrements .
The values obtained are 'Journal Chem. Soc for the last few years .
1908 .
] Viscosity at High and respectively , which compare well with the values given in Iable II .
The same apparatus gave the following results for mercury at temperatures:\mdash ; Table III.\mdash ; Mercury at High Temperatures .
ture .
Logarithmicdecrement .
Density .
according to * ' Ann. de ] ' Physik 1881 , vol. 14 , pp. 1\mdash ; 12 .
An example of a determination of viscosity at a somewhat higher temperature is given in Table .
The values of the density and viscosity for sodium nitrate have already been determined by the ] method.* These values are given in the table with the values of viscosity calculated from the logarithmic decrements using the same density values .
The disc used was one of fireclay , 27 mm. in diameter .
The constants , and , as determined by oscillations in water , mercury , and chloroform , were Table \mdash ; Sodium Nitrate ( Melting Point The numbers in the last two columns agree within the experimental error .
These results show the availability of this method for the determination * Goodwin and Mailey , .
cit. VOL. LXXX.\mdash ; A. Dr. T. E. Thorpe .
[ Mar. 5 , of viscosity up to the highest temperature at which a platinum capillary has been used .
As the fireclay or graphite disc remains uncorroded at temperatures of 1000o C. and over , the method is quite well suited for high temperature work .
I am now on viscosity measurements with a number of salts and metals , and hope before long to report on these .
Any expenses connected with this ation have been defrayed by grants from the Chemical Society and from the Carnegie Trust for the Universities of Scotland .
BAKERIAN LECTURE for 1907.\mdash ; On the Atomic Weight of By T. E. THORPE , C.B. , .D .
, F.B.S. ( Delivered June 20 , 1907 ; MS .
received in completed form , March 5 , 1908 .
Although there has been a considerable amount of discussion , based upon spectroscopic considerations and on its supposed mode of genesis , respecting the place of radium in the system of the elements , and inferentially , therefore , concerning its atomic weight , we are indebted for the only direct experimental determinations of this value hitherto made known to the discoverer of the element , Mme .
Curie .
Her first observations , published in 1902 , were made on about 90 milligrammes of the chloride , and furnished the value 225 as the mean of three fairly concordant experiments .
* In the autumn of last year Mme .
Curie communicated to the French Academy the results of a second series of estimations .
These were made upon much larger quantities of the careful]y purified chloride ( about 4 decigrammes ) and afforded the value as the mean of three closely concordant determinations In 1906 , at the instance of Sir William , then President of the Society , and by the aid of the kind interest shown by H.R.H. the Prince of Wales , the Austrian Government placed about 500 kilogrammes of pitchblende residues from the mine at Joachimsthal at the disposal of the Royal Society .
These residues were sent to be worked up by M. Armet de Lisle at his factory at Nogent-sur-Marne , by the method employed by him in the case of the ' Ann. de Chim .
et de Phys vol. 30 , 1903 .
'Comptes Rendus , ' 1907 , vol. 146 , p. 422 ; 'Le Radium , ' October , 1907 .

