195 A Spectro-photometricComparison of the Emissivity of Solid and Liquid Copper and of Liquid Silver at High Temperatures with that of a Full Radiator .
By C. M. Stubs , M.A. , M.Sc .
( Communicated by Prof. F. G. Donnan , F.R.S. Received January 9 , \#151 ; Read January 30 , 1913 .
) ( From the Muspratt Laboratory , University of Liverpool .
) Introductory .
In a paper which appeared recently in these c Proceedings'* Dr. Prideaux and the author communicated the results of a spectro-photometric comparison of the emissivity of solid and liquid gold at high temperatures with that of a full radiator , or " black body .
" Por solid gold , the curve for relative emissivity against wave-length was shown to be similar to , but not identical with , that for absorptivity at low temperatures determined by other workers ; whether the difference was due to a difference in the structure of the surfaces examined , or to a real temperature coefficient of the absorptivity , remained an open question .
At the high temperatures employed no appreciable temperature coefficient of the relative emissivity could be discovered , for either the solid or liquid metal .
At the melting-point , however , there is an abrupt change in omissivity , the curve for the liquid differing considerably from that for the solid .
In the present paper is described a similar investigation of the emissivity of copper and silver .
Apparatus and Method .
The general method pursued and the apparatus used were the same as described in the former paper , with the exception of such changes and modifications as will now be referred to .
The " black body " used in the present research was of the Lutnmer-Kurlbaum type.f It had certain peculiarities which perhaps are worthy of description .
Instead of the usual porcelain tube , an iron tube 58 cm .
long , 5 cm .
in internal diameter , and with walls 3 mm. thick was used ; the radiating surface consisted of a well-fitting iron block , 2 cm .
in thickness ; the radiating chamber was 4*7 cm .
in length , the nearest diaphragm to the radiating surface being of iron , 4 mm. thick , and with an aperture 1*6 cm .
in diameter .
* ' Roy .
Soc. Proc. , ' 1912 , A , vol. 87 , p. 451 .
t ' Ann. d. Phys. , ' 1901 , vol. 5 , p. 829 .
VOL. LXXXVIII.\#151 ; A. P 196 Mr. 0 .
M. Stubs .
Emissivity of Solid and Liquid [ Jan. 9 , In front of the radiating chamber were four other diaphragms , made of sheet nickel , which withstands oxidation well .
The temperature was measured by a thermocouple , enclosed in a thin-walled tube , which passed through the diaphragms immediately below the apertures , and fitted into a hole bored in the radiating block ; on the reverse side , a second checking thermocouple passed through suitable diaphragms and fitted into a hole bored in the centre of the radiating block .
The various diaphragms and radiating block were kept in their relative positions by stout iron rods , but could be moved collectively along the iron tube ; the radiating chamber could thus by actual trial be made to coincide with the region of uniform temperature .
The heating was by means of two windings of " nichrome " No. 22 resistance wire , on independent circuits , insulated from each other and from the iron tube by several layers of asbestos paper ; the outer one was , as suggested by Waidner and Burgess , * wound more closely at the ends than in the middle .
To sum up the advantages obtained , these were ( 1 ) evenness of temperature distribution through the nature of the winding , the conductivity of the walls of the radiating chamber , and the freedom of movement of the latter relative to the tube , and ( 2 ) automatic lining of the walls of the chamber , on heating , with iron oxide , which possesses a good degree of " blackness .
" The fact that at high temperatures a narrow hole bored in the centre of the radiating surface was invisible showed that very approximately full radiation was obtained .
The whole " black body " might probably with advantage have been made on a somewhat smaller scale .
Both copper and silver had to be kept in a reducing atmosphere , in order to avoid surface oxidation of the former , and the dissolution of oxygen in the latter , resulting in a considerable lowering of freezing-point and consequent uncertainty in temperature-measurement , and possibly also in a change in emissivity .
In the case of copper , the metal was contained in a silica pot inside a vertical iron tube 40 cm .
long and 5 cm .
in internal diameter , and closed at the bottom .
This tube possessed a hollow water-cooled cap which could be screwed tightly on .
Through the centre of the water-chamber was soldered a tube sufficiently wide to allow spectro-photometric observation of the copper ; other smaller tubes passing through the water-chamber were for the thermocouple , and for ingress and egress of hydrogen .
Over the central aperture was laid a thin microscope cover-glass , vaselined down at the edges , through which observations were taken .
A suitable tap allowed the observation-tube to be closed from the air when the cover-glass was removed for cleaning .
This arrangement proved quite satisfactory , and a slow stream of hydrogen passing * 4 Bull .
Bur .
Standards/ 1907 , vol. 3 , p. 165 .
1913 .
] Copper and of Liquid Silver at High Temperatures .
197 through the tube and out without burning kept the copper from oxidation , -The width of the iron tube was such that the reflection of its walls was not visible in the copper mirror .
The image of the under surface of the water-cooled cap , illuminated by the light of the furnace , would , however , be visible .
This surface was , therefore , lampblacked ; though as a matter of fact a simple calculation showed that the glowing parts of the heated tube were so far from the top that no appreciable influence would be exercised on the apparent emissivity of the copper , even if the bottom of the water-cooler were far removed from blackness .
In order to balance the loss of heat by conduction to the water-cooled end , the wall of the upper half of the iron tube was turned thin , and the whole tube then wound with " nichrome " resistance wire over asbestos paper , and placed within a large vertical platinum-wound resistance furnace whose heating-tube was 22 cm .
long and 12 cm .
in diameter .
With the aid of the heat generated by the passage of an auxiliary current through the winding on the tube , the desired temperatures were easily obtained .
In the case of silver , where access of oxygen below a red heat has not to be avoided , the metal was heated in a furnace whose heating-tube was a silica one wound with " nichrome " wire .
On to this tube fitted closely the water-cooled cap described above .
Any appreciable access of air was thus avoided^ and a reducing atmosphere was kept by placing a quantity of powdered graphite beneath the crucible containing the silver .
The metal was observed , as in the case of copper , through a thin cover-glass .
As was necessary for comparative purposes , the " black body " also was viewed through this glass .
The latter was in all cases cleaned from time to time .
The same care as described in the former paper was taken to calibrate and check the thermocouples , and the speetro-photometer with its comparison lamp .
Any change in the latter wTas checked and allowed for by making observations of " black body " radiation several times during the course of the experiments .
In measuring the temperature of the metals only one thermocouple was used , which was inserted about 1*5 cm .
into the metal .
Owing to the sharp temperature-gradient above the surface the temperatures indicated were a little too low , but were corrected , as described in the former paper on gold ( p. 456 ) , by observing the indication at the melting-point.* The corrections necessary were small , usually about 3 ' .
* In the case of oopper , allowance was made for the depression of freezing-point caused by dissolved hydrogen .
The latter amounts to 0*54 mgrm .
per 100 grm. of metal , according to Sieverts and Krumbhaar ( ' Zeits .
phys .
Ohem./ 1910 , vol. 74 , p. 292 ) .
From the depression constant given for copper by Heyn ( ' Zeits .
anorg .
Chem. , ' 1904 , vol. 39 , p. 20 ) , this would involve a depression of freezing point of less than 0*5 ' .
P 2 198 Mr. C. M. Stubs .
Emissivity of Solid and Liquid [ Jan. 9 , A method was devised of compensating automatically for the troublesome variable thermal E.M.F. of the potentiometer , mentioned in the former paper ( p. 457 ) , which , as such thermal E.M.F. 's occur frequently in these instruments , may be worth a brief description .
AiA2 and BiB2 ( fig. 1 ) are two of the pairs of circuit terminals on the potentiometer .
The thermal E.M.F. of the instrument tends to create a small difference of potential between Ai and A2 , or between Bi and B2 , whichever pair may be .
put into the potentiometer circuit by the circuit switch .
Ei is a battery which runs down through the high resistance R and the low slide-resistance XY .
With the circuit switch on AiA2 , and the main potentiometer current temporarily disconnected , the slider Z is adjusted so that the fall in potential along ZY is equal to that between Ai and A2 , due to the thermal E.M.F. ; the galvanometer will accordingly give no deflection .
If now E2 be the source of the E.M.F. to be measured , and connections be made as in the diagram Fig. 1.\#151 ; Diagram of method of compensating for thermal E.M.F. it is clear that between Bi and B2 there will be a fall in potential equal to E2 , together with the fall along ZY ; this latter will exactly balance the thermal E.M.F. , and consequently the reading of the instrument will correspond with the true E M.F. E2 .
Results and Conclusions .
( 1 ) Copper.\#151 ; About 400 grm. of pure electrolytic copper were used , the internal diameter of the containing pot being 4*2 cm .
and the depth of metal about 3*5 cm .
The solid copper surfaces whose emissivity was measured were prepared as follows :\#151 ; The block of previously fused electrolytic copper was turned flat in the lathe , then treated successively with four grades of emery paper , and finally polished on chamois leather with a little " Globe " metal polish , the surface being washed with benzene after polishing .
The use of rouge was avoided , as it is liable to tarnish soft metals , and actually did so in the previously investigated case of gold .
Brilliant mirrors were obtained which , on heating , showed practically no trace of filming due to impurity .
1913 .
] Copper and of Liquid* Silver at High Temperatures .
199 Even the best mirror was , however , marked with a few fine scratches .
Tate , also , * has recently found it impossible to get a perfect copper mirror by polishing .
The area occupied by these scratches would , however , be so small as scarcely to affect the emissivity ; and , in fact , they were invisible when the metal was heated to glowing .
The surface showed no apparent deterioration until within about 10 ' of the melting-point , when re-crystallisation rapidly set in , involving reflection of light from the furnace walls by the roughened surface .
A perfect mirror surface of liquid copper , free from film , was obtained without difficulty .
In Table I are given the values obtained for the relative emissivity of solid and liquid copper , calculated as shown in the former paper ( p. 458 ) .
The capitals at the head of the columns indicate , in alphabetical order , the order of the experiments .
The approximate temperatures are also given .
In this and subsequent tables emissivities and absorptivities are stated in percentage numbers .
Table I.\#151 ; Relative Emissivity of Copper .
Wave-lengtk " \#171 ; * .
Solid .
Liquid .
D. 989 ' .
c. 991 ' .
p. 1053 ' .
Mean .
B. 1090 ' .
A. 1127 ' .
c. 1X74 ' .
Mean .
700 10 -i 7*7 9*4 8*7 10-6 13*0 12 *4 11 *8 675 10 -7 95 11 *1 10 -2 12 -3 13 9 13 -0 12-9 650 11 *6 10 *4 12 *4 11 *2 14 *8 15 2 14 *6 14 *8 625 14 *7 13 *0 15 '0 13*9 17 *1 17 *8 16 7 17 -1 600 19 *3 17 -3 20 *1 18*5 21 *0 21 -0 19 *7 20 *5 575 24 *6 23 *7 25 *7 24 *4 25 *1 25 *0 24 *4 24 *8 550 .
31*7 30 *8 31 *9 31 *3 29 *8 29 *5 28 -6 29 *3 525 35 *2 35 *5 36 *7 35 -7 33 0 34 -0 34 *9 34 *0 500 41 *5 38'9 42*1 40*4 37 *4 38*1 40 *2 38 *7 475 \#151 ; \#151 ; \#151 ; \#151 ; 37 *4 47 *3 \#151 ; 42 *4 The solid copper mirror used for experiment D was not so perfect as that used for C , and one or two small red spots due to markings on the surface were visible amid the otherwise greenish light from the copper .
In calculating the mean , therefore , only half weight is given to the numbers in the first and third columns .
It is probable that even so the " mean " values err on the high side .
In calculating the mean values for the liquid only half weight is assigned to experiment A , there being reason to suspect the corresponding " black body " readings of less accuracy than usual .
The * 'Phys .
Rev. , ' 1912 , vol. 34 , p. 321 .
200 Mr. C. M. Stubs .
Emissivity of Solid and Liquid , [ Jan. 9 , mean values of the relative emissivity at the various wave-lengths are shown graphically by the Curves II and III in the diagram ( fig. 2 ) .
It is evident that , as in the case of gold , the emissivity of copper is discontinuous at the melting-point , the difference , however , between that of the solid and of the liquid being less marked .
No previous quantitative I. Absorptivity , solid copper ; II .
Relative emissivity , solid copper ; III .
Relative emissivity , liquid copper ; IY .
Relative emissivity , liquid silver .
measurements have been made of the radiation from solid copper ; but some on liquid copper have been made by Burgess.* The following table embodies his results :\#151 ; * Wave-length Relative emissivity of liquid copper at in ma .
1075 ' .
1125 ' .
1175 ' .
.1225 ' .
650 17 15 14 13 550 47 38 32 28 * 'Bull .
Bur .
Standards , ' 1909 , vol. 6 , p. 111 .
1913 .
] Copper and of Liquid Silver at High Temperatures .
201 The wide disparity between his values and the author 's is apparent ; nor in the author 's results is there any evidence of such a decrease of relative emissivity with rise in temperature as Burgess shows .
His results are undoubtedly in error , due largely to lack of monochromatism of the coloured glasses used in his pyrometer ( which in a case such as the present , where emissivity varies rapidly with wave-length , might introduce considerable errors ) , and probably also to the far less favourable experimental conditions of his work .
The reflectivity of copper at ordinary temperatures has been measured by several , the surface being prepared by polishing in the work of Hagen and Rubens , * Minor , f and Tool .
J The first named determined the reflectivity directly , the other two by calculation from other optical constants .
The following values for the absorptivity ( 1 \#151 ; R , where R = reflectivity ) are given by their results :\#151 ; Wave-length in / i/ x. Hagen and Rubens .
Minor .
Tool .
Wave-length in Hagen and Rubens .
Minor .
Tool .
700 9*3 575 29*8 i 660 \#151 ; \#151 ; 15 *4 560 \#151 ; \#151 ; 39 *9 650 11 *0 \#151 ; \#151 ; 550 40 *5 41 *6 \#151 ; 640 \#151 ; \#151 ; 15*9 540 \#151 ; \#151 ; 42 *4 630 \#151 ; 19 *5 \#151 ; 535 \#151 ; 43 *8 \#151 ; 620 \#151 ; \#151 ; .
17 *2 520 \#151 ; \#151 ; 44-0 600 16-5 \#151 ; 18 -6 500 46 -7 44-5 44 *9 589 *3 \#151 ; 25 *9 \#151 ; 480 \#151 ; \#151 ; 46 *1 580 \#151 ; \#151 ; 23 *3 450 51*2 49 *5 \#151 ; \#166 ; The values of the different workers do not agree very well , the differences being no doubt largely due to variation in the conditions of preparation of the surface .
All agree , however , in deviating strikingly from the values for the relative emissivity of the solid metal given in Table I. Hagen and Rubens ' values are shown graphically by the curve I in the diagram .
The deviation is in all cases greatest for the shorter wave-lengths .
Thus for wave-length 550 fifi ( green ) all three workers agree in placing the absorptivity at about 41 per cent. ; the author 's value for relative emissivity is 31 per cent. Again , in the region 560 to 590 the author 's curve does not show the remarkable flexure which is characteristic of the others ' , though there are signs of a slight bend .
As may be shown by a simple calculation , the discrepancy cannot be due to the temperature of the surface being lower than indicated by the thermo* 4 Ann. d. Phys. , ' 1902 , vol. 8 , p. 17 .
t 4 Ann. d. Phys. , ' 1903 , vol. 10 , p. 609 .
+ 4 Phys. Rev. , ' 1910 , vol. 31 , p. 14 .
202 Mr. C. M. Stubs .
Emissivity of Solid and Liquid [ Jan. 9 , couple .
For at ( say ) 1000 ' C. the total energy radiated by a " black body " is , by Stefan 's law , 5*3 x 10"5 x 12734 ergs per square centimetre per second .
The thermal conductivity of copper at ordinary temperatures is such that , with a temperature-gradient of 1 ' per centimetre , 0'91 grm. calories , or 4*2 xl07x 0*91 ergs , pass per square centimetre per second .
Thus , if X be the temperature-gradient in degrees per centimetre which a " black body " at 1000 ' and with the conductivity of copper would need for radiated energy to be continually replaced , 5*3 x 10"5 x 12734 = X x 4-2 x 107 x 0*91 , whence X = 3*7 .
Since now most of the energy radiated by a " black body " at 1000 ' C. is in the infra-red , where the reflectivity of copper is very high , and emissivity therefore very low , it is safe to say that copper would radiate less than 1/ 10 as much as a " black body " ; and the temperature-gradient necessary is reduced to less than 0*37 ' per centimetre .
Or , making every allowance for change in conductivity with temperature , etc. , a temperature-gradient of less than 1 ' per centimetre would be required , whereas an error of 17 ' would be needed to explain the difference given above between the values at wave-length 550 yaya .
Again , loss of polish on heating would result in reflection from the furnace walls , and therefore a greater instead of a less value for the relative emissivity .
The conclusion is unavoidable that the absorptivity in the visible spectrum of a polished copper surface changes on heating .
Such a change in the case of gold was suggested , though left open to question , by the results given in the former paper ( p. 461 ) .
It is significant that the relative emissivity curve for solid copper tends to approach the absorptivity curve at long and short wavelengths , but to make the intermediate flexure which is characteristic of all copper surfaces , even unpolished ones , * much less marked .
This tendency is carried out to a much greater degree in the curve for liquid copper , which is without flexure , but still approaches the absorptivity curve at extreme wavelengths .
Similar remarks apply in the case of gold .
It seems at least plausible that , however invariable optical constants may be with temperature in the case of such metals as platinum , *)* the atomic or molecular peculiarities which cause the rapid variation , in the visible spectrum , of the optical constants of copper and gold are diminished as the temperature rises and the kinetic energy of the atoms increases , a further diminution appearing suddenly when inter-molecular constraint is further relieved at the melting-point .
An investigation which the author has not the facilities for carrying out , * For an example of these , the recent work of Tate ( 4 Phys. Rev. , ' 1912 , vol. 34 , p. 321 ) may be referred to .
t Rubens , 4 Phys. Zeits .
, J 1910 , vol. 11 , p. 139 .
1913 .
] Copper and of Liquid Silver at High Temperatures .
203 but which may be suggested as a way of definitely settling the behaviour of copper , would be , after determining the emissivity of a heated solid copper surface , to let it cool down , and directly determine its reflectivity , or better still , to determine the reflectivity at various temperatures up to the melting-point .
The experimental difficulties , though great , should not be insuperable .
II .
Silver.\#151 ; About 400 grm. of silver were used .
A good mirror was easily prepared , but lost its polish on heating to redness , reflecting light from the furnace walls , and preventing measurement of the true emissivity .
This instability of the surface was noted by Drude , even when the metal was heated in hydrogen .
It is perhaps connected with the noticeable volatility of the metal .
The emissivity of a clear surface of liquid silver was , however , measured .
The results are tabulated below , along with Hagen and Rubens ' ( loc. cit. } p. 16 ) directly and Minor 's ( loc. cit. , p. 614 ) indirectly obtained values for the absorptivity ( 1\#151 ; R ) of solid silver mirrors .
Table II .
Wave-length in fx/ ji .
Relative emissivity of liquid silver .
Absorptivity of solid silver .
1004 ' .
1060 ' .
1117 ' .
Mean .
Hagen and Rubens .
Minor .
700 6*66 7 -22 6 99 5 -4 675 \#151 ; 6-90 7-41 7*15 \#151 ; - .
650 6-87 7 22 7 '30 7*18 6*5 \#151 ; .
' 625 6-77 6*97 7 '37 7*09 \#151 ; ' \#151 ; 600 7 -03 7*17 7'58 731 7*4 5*0 ( for 589 '3 ju/ i ) 575 7-39 7*75 7'74 7*67 \#151 ; \#151 ; 550 7-77 8-06 8-27 8-09 7 3 5*8 525 7'63 8*49 9 '03 8*53 \#151 ; \#151 ; 500 \#151 ; 8*17 9'48 8*82 8*7 6*8 450 \#151 ; \#151 ; \#151 ; ' \#151 ; 9*5 8*3 The relative emissivity of liquid silver is shown graphically in the diagram .
Since the reflectivity of silver is very high , a small error in its determination would involve a large error in the absorptivity , as given above ; because of this uncertainty the values for the absorptivity are not shown graphically .
It will be seen that the relative emissivity of the liquid is throughout somewhat greater than the absorptivity of the solid .
There is no other remarkable feature about the emissivity of liquid silver , except its extreme smallness throughout the visible spectrum , as might be anticipated from the high reflectivity of the metal .
This lowness of emissivity made measurements , especially near the melting-point , very difficult ; it is therefore doubtful Emissivity at High Temperatures .
whether the apparent slight increase in relative emissivity with rise in temperature , shown in the table , is real .
On this account also only half weight is given to the measurements at 1004 ' , in calculating the mean .
As in the former paper for gold , it is thought of interest to calculate the " black body " temperature , S , of the metals at their respective melting-points .
Table III.\#151 ; " Black Body " Temperature of Copper and Silver at their Melting-Points .
Wave-length in fx/ x. Copper .
Melting-point 1083 *4 ' C. Silver .
Melting-point 960 *7 ' C. S ( solid ) in 'C .
S ( liquid ) in 'C .
S ( liquid ) in 'C .
700 896 917 792 675 912 928 799 650 924 942 804 625 943 956 808 600 966 973 816 575 988 989 823 550 1007 1003 831 525 1018 1015 838 500 1028 1026 845 475 \#151 ; 1033 \#151 ; * Summary .
1 .
The emissivity of solid and liquid copper and of liquid silver at high temperatures , relative to that of a full radiator at the same temperatures , has been measured throughout the visible spectrum .
2 .
As in the case of gold , the emissivity of copper is discontinuous at the melting-point , the " relative emissivity " curve of the liquid showing no flexure .
3 .
The curve of " relative emissivity " of solid copper at high temperatures differs considerably from that of absorptivity at low temperatures ; it possesses a much less marked flexure in the green , and it is suggested that this is due to the same causes which ultimately bring about the total absence of a marked bend in the curve for the liquid .
4 .
Contrary to Burgess 's results , no appreciable temperature coefficient of " relative emissivity " was found for liquid copper over a range of 100 ' .
5 .
The " relative emissivity " of liquid silver is throughout remarkably low , but seems to be somewhat greater than the corresponding values of the absorptivity of solid silver at ordinary temperatures .
Crystallisation of Substances which form Mixed Crystals .
205 6 .
" Black body " temperatures of solid and liquid copper and of liquid silver at the respective melting-points are calculated .
The author desires to express his thanks to Prof. F. G. Donnan for his advice and help , and to Mr. L. Spencer for his invaluable assistance in making some of the measurements .
On the Spontaneous Crystallisation and the Melting- and Freezing-point Curves of two Substances which form Mixed Crystals and whose Freezing-point Curve exhibits a Transition Point.\#151 ; Mixtures of p-Bromnitrobenzene and p-Chlornitro-benzene .
By Miss Florence Isaac .
( Communicated by Sir Henry A. Miers , F.R.S. Received January 18 , \#151 ; Read January 30 , 1913 .
) The following paper is a continuation of two previous papers in which the melting- and freezing-point curves for two pairs of substances , each of which forms mixed crystals , have already been determined .
In the first of these papers* mixtures of naphthalene and / 3-naphthol were examined and found to form a continuous series of mixed crystals and to give curves of Roozeboom 's Type I , in which the melting and freezing points of all mixtures lie between the melting points of the pure substances .
In the second paperf mixtures of azobenzene and benzylaniline were examined and found to form mixed crystals of two kinds , whose melting-and freezing-point curves exhibit a minimum or eutectic point ( Roozeboom 's Type Y ) .
The present paper deals with mixtures of a pair of substances\#151 ; p-brom-nitrobenzene and jp-chlornitrobenzene\#151 ; which form mixed crystals and give freezing-point and melting-point curves belonging to Type IV of Roozeboom , in which , though the melting and freezing points of all the mixtures lie between those of the pure substances , the curves show a break or discontinuity corresponding to the existence of two kinds of mixed crystals , as in Type V. * ' Journ. Chem. Soc. , ' 1908 , vol. 93 , I , p. 927 .
+ 'Roy .
Soc. Proc. , ' 1910 , A , vol. 84 , p. 344 .

