]\gt ; On the -chemistry of Flame High By W. N. HARTLEY , F.R.S. , D.Sc .
, College of Science , Dubhn .
( Received January January 31 , \mdash ; Received in revised form March 22 , 1907 .
) The majority of the metallic elements have two distinct spectra , one of lines , the other of bands .
In some spectra , bands are the most prominent feature in association with but few lines ; in others the lines predominate over the bands .
Of those substances with bands in their spectra there are many which may be reduced in quantity to almost a minimum without the bands disappearing from the flame or from the uncondensed spark ; from this it would appear that there is a.very intimate relationship between the bands and lines of the same element .
It has been supposed that such hands as cannot be attributed to definite salts are the spectra of oxides or even of hydrides , but many facts have been accumulated from which it is evident that this is not rect .
On the other hand , many of the elements which possess complex molecules may readily be understood to emit molecular or band spectra , but some further explanation is when the molecule of the element is monatomic .
The present communication deals only with one part of this subject , being merely a record of observations on temperature flame reactions , and principally those of the alkali and alkaline earth metals .
The spectra of the alkali metals at the highest temperature in the flame of the oxyhydrogen blow-pipe more nearly resemble those obtained from the arc than any other spectra of the same elements .
Thus , all the lines of lithium Taphed by Kayser and Rung in the arc have been obtained in the flame from lithium nitrate .
This is evidence of the high temperature obtainable in the flame , but if further evidence were desired , we have it in the great facility with which both gold and silver can be vapourised and condensed from the vapour , the large number of arc lines of iron , practically all those that are reversible , which appear in the flame , and the volatilisation of dium and iridium in wei.(yhable quantities .
Part I.\mdash ; BAND SPEcTltA OF THE ALKALI AND ALKALINE EARTH METALS .
Lenard* has recently described the coloured glow by globules of fused salts in the Bunsen flame , the particular parts of the flame from 'Annalen der Physik , ' 1905 ( IV ) , vol. 17 , pp. 197\mdash ; 247 .
-chelnistry of High .
243 which the glow emanates , with the spectrum of the from different metals .
The spectra of the glow from metallic vapours have been photographed repeatedly in the oxyhydrogen flame spectra of all the alkali metal salts , * and those of the alkaline earths , but Lenard states that he has been able to find the maximum of brightness of the continued rays of potassium , at or near , as it appears on the reproduction of my photographs ( Plates 6 and 7 ) .
On referring to the plate , I do not see that the maximum is at 5500 , and the strong continuous rays are so evenly distributed over so wide a range of spectruzn that it is better to refer to the letterpress of the publication .
The continuous spectrum of lithium extended from about to 3200 ; of sodium very strongly from 6020 to and continued to ; of potassium it was intense from about 4600 to 3400 , but continuous rays were observed from about .
As importance has been attached to the kind of photographic plate used , it is as well to state that isochromatic pJates of Edwards ' .
or of the Ilford make , and Cadett and Neall 's spectrum plates , were most generally in use , but when the plates were not isochromatic , the fact was recorded on the photograph .
On the photograph referred to by Lenard , the plate was less sensitive to the green than to the and red rays , so that a weak space is seen between and .
The continuous spectra have been carefully examined on more than 40 different photographs of the alkali-metal salts , including sulphates , nitrates , carbonates , and chlorides ; also potassium cyanide , ferrocyanide , ferricyanide , and fluo- silicate .
A hitherto undescribed feature in the spectra which is unquestionably due to each metal , is the occurrence of a multitude of bands in the rays of lithium , sodium , and potassium , which occupy spaces within the above-quoted wave-lengths .
The photographed spectra of potassium chloride , as emitted by different flames , were studied .
In no case did the nature of the salt , the support used , or the nature of the flame , make any difference , except in lengtl ) ening or shortening the exposure which , jriods , was proved to have same effect as more or less of the salt into the flame .
It is deemed of importance that this should be recorded , because different metals , and also different salts of the same metal , under the same conditions show different results , and this arises principally from the rence in Flame Spectra at High Temperature 'Phil .
Trans 1894 , vol. 185 , , pp. 161\mdash ; 244 Prof W. N. Hartley .
Thermo-chemistry of [ Mar. volatility of the salts , or the comparative ease with which they are thermo- chemically decomposed .
It is also important that the salts be pure , otherwise the spectrum of one may interfere with the other , the two spectra become confused , and the bands are not visible .
liithium Salts.\mdash ; Various specimens of the sulphate , carbonate , chloride , and nitrate , were used , but most of the spectra were raphed from the nitrate .
Latterly the salts obtained have been so pure that the sulphats showed hardly a trace of any other metal , even sodium , and no calcium .
* They show all the lines attributed to lithium by Kayser and Runge .
The structure of the background is that of fine lines and very narrow bands extremely close together , extending from to .
But it must be observed that a feeble group of the water-vapour lines may be seen on other spectra extending to 3466 , therefore to be sure that these may not be mistaken for lithium bands , it is better to place the margin of the latter below and to disregard the rays of greater refrangibility .
The lithium bands , therefore , without doubt , extend from 3610 to The lithium spectrum , exclusive of the lines , is a very weak one , so that an extended band of continuous rays is more often photographed than a discontinuous spectrum consisting of a multitude of fine lines and narrow bands .
Sodium Salts.\mdash ; The banded spectrum of sodium has been observed on three photographs , but it is only feebly visible .
A complete array of the bands was raphed from very pure rock salt , and from a specially prepared pure sodium chloride .
Like the lithium bands , they are very narrow and close together ; they extend from rays a little less refrangible than down to about .
A beautiful banded spectrum has recently been photographed from sodium carbonate in the oxyhydrogen blow-pipe flame .
Some feeble lines accompany the bands , and the principal lines and ] are enormously extended on each side so as to form a broad band .
O. H. Basquin observed a series of lines in the spectrum of the sodium arc when surrounded by an atmosphere of hydrogen ; they lie between and 3800 .
" " This is probably a complicated fluting of heads of , less clearly marked than usual , and running towards shorter wave-lengths.\ldquo ; It is more than probable that the bands which I have observed are of the These salts were kindly presented to me by Mr. W. E. B. Blenkinsop , the Managing Director of the firm of May and Baker , Ltd. , of Garden Wharf , Battersea , London .
Astrophys .
Journ 1901 , vol. 14 , p. 1 .
In lieu of any better hypothesis , Basquin supposed that the banded spectra of reral of the metals , including sodium , are due to hydrides being formed in the arc .
Sodium hydride cannot exist in a state of vapour ; it readily uudergoes dissociation , which commences at 20 and is complete at 422 , under ordinary barometric pressures ; the bands , therefore , are not due to sodium hydride .
1907 .
] Spectra High Temperatures .
245 same nature as those which he , by the aid of a large concave grating , resolved into lines .
There is an important erence between the spectrum of sodium taken from the metal when burnt in a Bunsen flame and that obtained in the oxyhydrogen blow-pipe flame , but it appears to be due solely to the lower temperature of the former .
With metallic sodium , no lines but the principal yellow ones are visible , these are very broad and dense .
It is extremely difficult to introduce sodium into the oxyhydrogen blow-pipe flame , because it melts so and escapes .
The sodium spectrum from salts gives not only the and lines more widely ext , ended on either side than is the oase with the metal , but also the subordinate series of lines , the band of continuous rays , and with the highest temperature of the flame , the series of hands in these rays in addition .
Potass Salts.\mdash ; The potassium bands are very easily from many different salts .
They are themselves wider , and also wider apart than those of lithium and sodium , and the structure of the banded spectrum varies reatly with the quantity of the substance used , and with the exposure given .
Many lines accompany the stronger banded spectrum .
The potassium bands , as obtained from weighed quantities of potassium chloride , frequently extend from to .
This effect is obtained from of the salt , but with 1 amme or an exposure of 5 minutes , the bands may stretch down into the red .
With an exposure of 1# minutes there is barely an indication of bands , and when the quantity of the chlolide is reduced to ramme , there is nothing but a continuous spectrum to be seen .
Potassium cyanide and ferrocyanide with the same exposure both yielded very dense spectra , but on that account the bands are not seen distinctly .
They are well seen in spectra photographed from potash alum , potassium carbonate , and potassium ferricyanide .
Between the wave-lengths of two cadmium lines , and , ten of these bands were counted .
They are not sharp at the , and the spaces between , successive bands decrease as they tend towards the violet .
They appear , in fact , like a continuation of the subordinate series of lines , 47968 , , and and Dewar ) , but they are diffuse .
There is an absence of any fine lines such as are seen in the lithium and sodium spectra .
Spectra from sodium and potassium carbonates have recently been photographed with Wratten and Wainwright 's panchromatic spectrum plates , which give the lines well defined with the continuous rays from wave-lengths and of potassium to a line which belongs to sodium .
There is a strong band in the red , extending from to , fading away rather abruptly on the less refrangible side , above are seen to be composed of two groups ; the one nearer the red appear* like the continuation of the subordinate series of lines with the spaces .
between them decreasing , and the second group more towards the violet , where the bands do not appreciably vary in width or in distance apart .
Rubidium Salts.\mdash ; Sulphate and chloride of great purity yielded merely a continuous spectrum in which there are no bands , but the lines are very broad and strong .
Coesium Salts.\mdash ; The same effect is seen with the corresponding salts of caesium .
Banded Spectra from the Alkaline Earth Salts .
The banded spectra observed with barium chloride and strontinm sulphate are reproduced , but with a dispersion of only one prism on Plate 6 , in\ldquo ; Flame Spectra at High Temperatures.\ldquo ; * In course of a study of the oxyhydrogen flame spectra of calcium , strontium , and barium compounds , made in 1902 , bands were constantly observed in minerals containing barium , just as the bands of lead are constantly seen in the flame spectra of minerals containing lead and , subsequently , the barium bands were from the flame spectra of very carefully prepared compounds which were spectroscopically pure .
The bands in the continuous background of rays are different from those of the alkali metal group .
In calcium and strontium the most conspicuous bands are broad and disposed somewhat irregularly , but in detail the spectra of calcium , strontium , and barium compounds show bands of two varieties , the one broad overlying narrow bands ; when the quantity of material is small , only the narrow bands are seen , and when the slit is not narrow enough they appear as diffused rays or a continuous spectrum .
As in the lithium , sodium , potassium group , the calcium , strontium , and barinm spectra differ much in intensity , and follow in the same order as the alkali metals in this respect , which is that of the numerical values of the atomic weights of the metals , the higher the value the greater the volatility of the compound .
The strontium and barium chlorides exhibit the banded spectra best , but they are well seen also in sulphates , carbonates , and oxides .
The best mineral in which to study the bands of calcium is the natural sulphate , selenite .
It should not be overlooked that compounds yield their own spectra , such as chlolides , under certain conditions , phosphates and fluorides , as , for instance , ferrous phosphate and calciulJl fluoride .
The oxides and also emit their own spectra , but 'Phil .
Trans. 1894 .
1907 .
] Flame Spectra at High Temperatures .
247 whereas the former shows bands , the latter shows lines only .
This has been made the basis of a method of separating alkali metal flame spectra from those of the alkaline earth metals .
* Bands characteristic probably of the sulphates may be foumd to overlie the metal bands .
Several different compounds of barium have been found to yield the same banded flame spectrum when a sufficient quantity of substance is introduced into the flame , and all strontimn salts also yield in like circumstances the same flame spectrum .
When the sulphates of the alkaline earths are submitted to the flame , they all yield the banded spectra , and in the case of calcium and strontinm sulphates , also certain lines in addition , but barium sulphate frequently shows only the bands without the barium line ; therefore , the lines appear more readily than the bands with calcium salts , and the bands more readily than the lines with bariuun compounds .
In fact , barium behaves in this respect very much as lead does .
There is an important difference in the chemical nature of the three sulphates after being heated in the flame : thus , the barium sulphate is entirely reduced to sulphide , so that we arrive at the fact that the same spectrum is common to chloride , oxide , and sulphide .
But the chlorides are converted by the water-vapoul ' in such a flame into oxides , so that we have simply the oxide and sulphide to consider .
The strontium sulphate is reduced partly to oxide and partly to sulphide under precisely the same conditions , and the calcium salt is simply converted into oxides .
As the xide produced from the chloride and sulphide produced from the sulphate of barium yield exactly the same banded spectrum , it is evident that we are dealing with the spectrum of the only constituent in common , which is the metal .
It is quite inconceivable that the oxide and sulphide should identical spectra of the respective compounds because the molecular weights of the compounds are different .
The same observation applies to the compounds of strontium , which yield spectra with lines and bands , but we know that strontium sulphate , after it has been removed from the flame , has by less than one-half been reduced to sulphide ; it is , therefore , probable that in this case as in that of barium , the spectrum is caused either by the dissociation of the sulphide in the flame or by a chemical reduction .
Calcium sulphate is not simply deoxidised in the flame , like the corresponding salts of strontium and barium , but is converted into lime , yet this compound exhibits both lines and bands .
' Chem. Soc. Trans 1893 , vol. 63 , p. 138 .
Prof W. N. Hartley .
of of Thermo-chemical Recluction of Oxides in the Flame .
It has already been shown that aluminium can be reduced from its oxide in the oxyhydrogen flame , that it colours the flame and emits the two lines and 3946 .
It was concluded that bands which have been attributed to the oxide are in reality metal bands .
* Bands of beryllium and lanthanum were also reduced from their oxides ; the greater volatility of metallic beryllium caused the band spectrum to be photographed without the lines .
I have now arrived at the conviction , from additional evidence preseJltly to be adduced , that calcium and strontium are also produced in the oxyhydrogen flame thermo-chemical reduction of their oxides and sulphides .
The question of the reduction of the oxides of the dyad and triad roups is quite different from that of the oxides of the alkali metals , because compounds of the latter are not only easily volatilised , but it has been proved by St. Claire Deville that in the manufacture of the alkali metals , the oxides are thermo-chemically dissociated , and then reduced by carbon entering into immediate combination with the dissociated oxygen .
Spectro-chemical Evidenoe from and Dehydrated Salts .
By an anhydrous flame is meant one in which water is not a product of combustion .
Any salt which yields the spectrum , or imparts the characteristic colour of the metal to such a flame , is excluded from the view of Arrhenius that vapourised salts are first hydrolysed by water-vapour in the flame , that the hydroxi thus formed undergo ionisation , and the spectra are caused by the metallic cations .
Smithells , Dawson , and Wilson first applied this test to alkali salts , by heating them in the flame of burning cyanogen .
Lenard also has observed that sodium salts colour the flame of carbon disulphide .
The authors aforesaid suggest that a chemical reduction takes place and the metal then colours the flame .
I have extended this enquiry by placing anhydrous salts in the flame of carbon monoxide .
Smithells has shown that , if this gas be carefully dried , it will not inflame when ; from a jet , but if the jet be heated the gas ignites .
It has also been shownS that carbon monoxide stored over water in a asholder w not inflame in air if the jet be small like that of a Bunsen or ordinary\ldquo ; bat 's wing\ldquo ; burner , but it will burn from a tube if the orifice be wide enough , 'Banded Flame Spectra of MetaIs , ' p. 346 , and Plate XXXL This was observed by Pollok in its reduction by the arc .
'Roy .
Soc. Proc 1899 , vol. 64 , p. 142 .
Hartley , .
" " On the Temperature of Certain Flames , \ldquo ; 'Chem .
Soc. Trans 1896 , 1907 .
] ab .
249 and even from a " " bat 's wing\ldquo ; burner when the pressure is small ; in these latter circumstances the flame is very small .
In view of these facts , pure carbon monoxide was most carefully dried , first with a column of pumice saturated with oil of vitriol , and then with two large tubes of phosphoric anhydride .
The gas was conducted into a platinum tube with an internal diameter of 6 mm. and a length of 30 cm .
It was found that under these conditions the gas was very difficult to inflame , but it did burn ; nevertheless , to increase the temperature and render the same steady by combustion , the tube was heated red-hot by a Mecke burner placed 3 inches from its orifice .
On introducing into the flame recently nited carbonates of lithiull , sodium , and potassium , even without heating the platinum tube , beautifn1 spectra were observed , such as are visible in the flame of a Bunsen It was very difficult to see the red line of potassium about , but the green glow and a group of green lines about 5S00 were observed .
Experiments were tried in both the inner and outer cones , but only when the substance was heated at the tip of the inner cone could the group of greenl lines be seen .
In this instance it is not reasonable to suppose that sufficient water could gain access to the flame to give rise to hydrolysis , and , in fact , hydrolysis could not take place .
The same experiments were repeated with the oxides , carbonates , nitrates of calcium , strontium , and barium ; but although the platinum tube was heated to bright redness , there was no trace of a spectrum of either bands or lines in any case .
Calcium , strontium , and barium chlorides , on the other hand , showed superb spectra of the chlorides , apparently identical with those produced by heating the same salts in a coal-gas flame saturated with hydrochlori acid .
The temperature of this carbon-monoxide flalne , which from 00 to 75 mm. long , is higher than that of a Bunsen burner of the same size , * and can raise a larger mass of material to a higher temperature .
Here we have evidence that the metals of the alkalies behave quite differently from those of the alkaline earths ; , strontium , and baJium chlorides are volatile without decomposition , for they fuse and remain clear liquids without showing turbidity for as long as is necessary to observe their spectra .
Inasmuch as they appear to yield their own spectra , they behave like calcium fluoride when placed in the oxyhydrogen blow-pipe flame , and doubtless other similar fluorides and phosphates .
The most important facts ascertained so far are , first , the alkaline earth oxides are not Platinum wire mm. in tbickness can be melted into a globule in any of the flame .
second , the flarne is a powerful reducing one to all oxides with a heat of : formation not gleater than that of zinc oxide , but the oxides in question are not reduced ; thirdly , the temperature of the flame is higher than that of a burner , and over a area it is nearly as high as that of the oxyhydrogen flame ; fourthly , the chlorides are volatile in an anhydrous flarne , but the stability of the chlorides in such a flame is somewhat doubtful , they may undergo dissociation .
The action of the cyanogen flame is very different from that of the flame of carbon monoxide , because is a very powerful reducing agent .
When calcium oxide was introduced on a platinum wire into the flame of cyanogen in air , there was no action low down in the flame , but when the substance was placed at the tip of the inner cone .
there was a magnificent display of the same spectru1n as that seen when calcium metal is burnt in air , and with no trace of the cyanogen bands .
The effect on baryta was tested in a similar manner , by making pure barium carbonate into a stiff paste with water , and moulding it round a loop of platinum wire .
On placing this in the flame at the same point , an equally fine spectrum of barium was obtained .
In both instances the spectra were identical with those photographed when the same compounds were placed in a Mecke tlrner , but the brilliancy and intensity of the rays were greatly increased .
Beyond all doubt these spectra are the spectra of the metals .
The deductions from these facts are , that the lines and bands seen in the flame are certainly not the spectra of the oxides , because if the difference between this and the carbon monoxide flame were simply one of temperature , the emission spectra of oxides and sulphides should be quite as easily produced in a carbon monoxide flame as in one of hydrogen .
Finally , carbon monoxide was burnt with oxygen in a Deville blow-pipe and still no spectrum of calcium or strontium was seen , except when traces of chloride were present .
The salts were suppol.ted on very thin slips of Donegal kyanite .
The presence of hydrogen in the flame is , therefore , a necessary element in some chemical reaction which leads to the reduction of the sulphides or oxides to the metallic state .
The temperature of these flames , which ean melt platinum , is sufficient to effect the thermo.chemical tion of water-vapour , and it is probable that in this lies the cause of the spectra appealing .
Spectra of the metals observed in a carbon monoxide flame on the grandest .
scale are those which were photographed from the Bessemer\ldquo ; blow but in this case the spectra are not due to the heating of a substance in a pure gasi ' 1907 .
] Flame at or even in a xture of gases , but are the result of the internal combustion of impurities in the lletal , which causes the vapours to be carried off in a stream of intensely heated carbon monoxide mixed with and a proportion of hydrogen , all of which gases , except the nitrogen , are burnt in the air .
The average vapour pressure of water in the blast to about .
1 per cent. of the air ; this is reduced by the metal bath to hydrogen and carbon monoxide and , as the temperature is an exceedingly one , it is not surprising that a line spectrum of calcium seen in the oxyhydrogen flame has been photographed in the Bessemer flame .
the flame being a mixed hydrogen and carbon monoxide one .
The lines are , owever , only Ilamely , and uments from Bessemer flame observations alher lead to usion , on account of the many other elements present in the flame , and they must therefore be rejected or eated with discrimination .
It will be observed that in the groups Li , , and Ca , , Ba , the facility with which the bands are produced increases with the atomic weights , likewise the volatility of metals increases in the same order , and as , in the first group , the elements are monatomic , the relative vapour densities are half their atomic ; accordingly we have 7/ 2 , 23/ 2 , 39/ 2 as the relative quantities of matter in the flame if they have all the same volatility and if their oxides undergo dissociation with equal facility ; but the least easily dissociated is the least volatile lithium oxide ; we may therefore say that the vapour pressure of potassium will be at least six times that of lithium in any flame in which the two oxides are placed under the same conditions .
As there is very little difference between the heats of formation of the groups of oxides , , and , the required for their reduction may be considered to be the same for each oxide .
The respective volatilities of the oxides or of the metals is not , neither are we acquainted with their vapour densities ; accordingly we may assume them to be either monatomic 01 diatomic molecules : , if the latter , then the relative quantities of vapour in ) flame will be in the proportion of the atomic , or 40 , , and ; if they are monatomic , half these quantities , which gives three and -half times as much vapour of barium in the flame as of calcium , under the same conditions in either case .
If , however , we attempt to compare the apour densities of the alkali metals with those of the dyad ooroup as judged by their spectra , we are at ollce met with a difiiculty by the reactions in the flame by which the metals are set free , being different for group .
Prof W. N. Hartley .
Thermo-chemistry of [ Mar. Part \mdash ; ON THE CHEMICAL EDUCTION 0F THE ALKALINE EARTH SULPHATES .
For the action of gases in the flame on the insoluble sulphates , the minerals heavy spar , celestine , and selenite were chosen .
Pieces of each measuring about 15 mm. were held by platinum forceps in the oxyhydrogen flame for a definite period , and theu dropped into a measured volume of water .
The unaltered nuneral was insoluble in each case , but that which had reduction , whether to oxide or sulphide , dissolved with a strongly alkaline reaction ; the metal in solution was therefore determined volumetrically with standard acid .
From the dissolved salts obtained from heavy spar , sulphuretted hydrogen was evolved in large quantity , somewhat less in the case of celestine , and hardly a trace with selenite .
The following tabulated statement shows the results obtained:\mdash ; Anhydrous salt .
Time in flame 15 mins .
15 mins .
10 mins .
15 mins .
Metal in solution grm. Metal as sulphide Practically 10 per cent. Practically all .
none If numbers be taken proportional to the above quantities , they are found to be approximately in the ratio of the atomic weights of the elements , thus : gramme : gramme : gramme Ca : Sr : Ba A sample of celestine was heated in the flame for minutes and dropped into water , acidified with acetic acid , and the sulphuretted hydrogen precipitated as lead sulphide which was filtered off , converted into lead sulphate and weighed .
The excess of lead was removed and the excess of strontium determined as carbonate .
Weight of lead sulphate gramme .
Strontium as Strontium carbonate Strontium as These figures show that per cent. of the strontium in solution is in the tstate of sulphide .
The quantity of strontium reduced from sulphate in 35 minutes was gramme , whereas that reduced in 15 minutes was gramme , 'or the longer period in the flame has yielded the smaller product .
1907 .
] Ftame Spectra at High Temperatures .
253 : Rednction of Sulphate to by Hydrogen .
A current of dry hydrogen was passed over a weighed quantity of 1precipitated dried barium sulphate , heated to redness in a platinum tube .
It was soon observed that sulphuretted hydrogen was evolved , probably according to the equations following :\mdash ; ( 1 ) ( 2 ) Sulphm.etted hydrogen was still being evolved after a period of eight hours , when the experiment was stopped .
, initial gramme } ramme .
, residue unchanged , , Ba Difference , substance reduced , , Ba ) from , found Ba in , found ) Ba in , found Total Ba in reduced The reduced is thus seen to have been composed of gramme , and gramme ; or about two-thirds sulphide and onethird oxide .
There can be no doubt that the initial action of the hydrogen is the reduction of the sulphate to sulphide , and that probably the whole the sulphate is reduced ; the secondary action , whereby the sulphide is converted into oxide with evolution of sulphuretted hydrogen , is sl.ow and gradual with the barium compound , but takes place with greater facility with .
strontium , and with completeness in the case of calcium .
When heated to bright redness in a platinum tube ttnough which a current of nitrogen and of carbon dioxide is passed , bnrium sulphate in neither instance undergoes any change .
When submitted to a white heat in a platinum tube at a reduced pressure for 15 minutes , barium sulphate evolved no and the substance showed no alkalinity and was quite unchanged .
These experiments show that the hydrogen of the flame is the reducing agent by which the spectra are produced .
Part III.\mdash ; THE 0F THE 1N FLAME REACTIONS .
In studying the thermo-chemistry of flames , it is necessary to consider three separate questions:\mdash ; ( 1 ) The effect on compounds of high temperatures solely .
254 Prof W. N. Hartley .
Thermo-chemistry of [ Mar. 22 , ( 2 ) The chemical action of other reacting substances in the flame gases on the compound in the flame .
( ) The combined of temperatures and reducing gases .
The thermo-chemical notation used is that of Dr. James H. Pollok .
* The solid , liquid , ) seous , and dissolved states are indicated where necessary by a dot , dash , , or circle , placed below the formula of the substance .
The heat evolved in the formation of a compound is indicated by placed before its formula , which signifies energy .
The heat of a reaction is indicated by enclosing the equation in brackets and placing before it .
signifies a ramme degree of heat , or large calorie , similarly is a gramme degree of heat , or small calorie The chemical formula for a substance means , as is usual in this connection , its gramme-molecular weight .
In discussing the process of hydrolysis of alkaline chlorides in the flame , we must take into account the following values and place the reaction at a temperature approaching 2000o , then and The heat of formation of solid potassium and sodium hydroxides is thus represented\mdash ; ; .
similarly , NaOH The hydrolysis is then The corresponding value for NaOH is These reactions are not very strongly endothermic and therefore may occur , and are reyersible .
Among the common salts of lithium the sulphate has the greatest stability , nevertheless in a gas it is reduced to oxide .
Fused lithium sulphate was found to be a perfectly neutral substance , but when a bead of the salt had been heated in the flame of a Mecke burner for an hour it yielded a strongly alkaline solution which did not evolve sulphuretted hydrogen when acidified , nor did it lfervesce .
It therefore containe , the oxide .
On taking a larger mass and heating it on kyanite in the oxy-coalgas blow-pipe flame , care being taken that there should be no free carbon , the 'Sci .
Proc. Roy .
Dublin Society , ' 1899 , and ' Chem. Soc. Trans 1904 , vol. 85 , p. 603 .
For the convenient symbol C. , or gramme degrees Celsius , instead of heats units , or calories , and kilogramme degrees of heat , C. , we are indebted to Mr. J. .
Buchanan , " " On Steam and Brines ' Trans. Roy .
Soc. Edin 1894 , vol. 39 , Part III , pp. 529\mdash ; 573 .
1907 .
] Flame High ) es .
fused salt became reduced to sulphide .
From this it ) ears that the normal process of reduction of the sulphates in the flame is iven i the two equations referring more particularly to barium sulphate .
The difference iu the heat of formation of litbium oxide and sulphide respectively , and ; this shows how readily the former may be converted into the latter by the action of water-vapour , , by reason of its an exothermic action .
The flames from coal-gas , hydrogen , and carbon luonoxide respectively , when burning in air , have a temperature as high as that of the point of platinum , ) C. This temperature exceeds of the furnaces , in which sodium , potassium , , zinc , and magnesium are reduced and distilled on an industrial scale .
In the manufacture of sodium and potassium from the respective carbonates by means of carbon , the oxide is vapourised and chemically dissociated , thus , but the action is not reversible , owing to the presence of carbon , which immediately forms carbon monoxide with the oxygen , thns\mdash ; and ; therefore , on the whole we have But and so that where the carbon is oxidised to carbon monoxide only , we have a considerable endothermic action , thus\mdash ; , endothermic .
Although this equation does not take into account the fact that the is already dissociated , this endothermic action may easily tend towards an exothermic character by the further combustion of the carbon monoxide to carbon dioxide , thus\mdash ; or fur one atom of If we suppose the operation to be a simple one of reduction , we have\mdash ; The reduction of in like manner would result in an endotherrnic value of .
The same reasoning does not apply to lithium oxide , for all attempts to obtain lithium by a similar process of reduction have entirely failed .
The reason is , that lithia more nearly resembles the alkaline earth oxides in its chemical properties ; it is also the least volatile and least fusible and An important fact may be observed here , namely , that whereas the banded spectrum of potassium is comparatively easy to obtain , that of lithium ia very difficult , more difficult than that of sodium , and even the continuous rays are very weak .
The same must be said of rubidium , with the additional remark that its band spectrum has never yet been seen .
* This , undoubtedly , in part arises from the high value for the heat of formation of the oxides of these metals , and it is obvious that these particular properties of lithium serve to explain why the continuous glow observed by Lenard in the case of lithium is weaker than that of either sodium or potassium , which led him to remark that it was probably a property peculiar to that element .
Even rubidium carbonate was reduced by mixing with carbon and heating a particularly high temperature , but the greater volatility of rubidium compounds may facilitate their dissociation .
It is quite evident , then , that with the fusible and volatile oxides of sodium , potassium , and rubidium , reduction can be effected at a tempel .
not exceeding a white heat and probably about 1400o C. , certainly below thetemperature of melting platinum , and below the highest temperatures of the coal-gas , the hydrogen , and the carbonic oxide flames .
Hence I conclude that the banded spectra of lithium , sodium , and potassium observed at the temperature of the oxyhydrogen flame are spectra of the metals .
The Alkaline Earths .
It has been shown that these substances are of a different nature and their salts are sharply divided into two classes , the haloid and the oxy-salts , , each with a different behaviour .
The Salts.\mdash ; These are volatile in flames supplied with the halogen hydrides without decomposition .
They are also volatile without decom- .
position in a dry carbonic oxide flame .
Here , then , we may probably have the spectra of salts and not of the metals .
In a hydrogen or coal-gas flama the salts are hydrolysed and converted into oxides by heat ; this is a fact well known .
If we examine the heat of formation of the ohlorides , we find it to be , for calcium and strontium respectively , and A nole appears on one of my photographs that there are just faint indications of bands in the strong continuous spectrum .
: 1907 .
] Flame Spectra at Hi. .
257 while the oxides are and The barium chloride has not been determined , but the oxide is and we may therefore safely say that the energy of formation of the chloride does not differ much from the number deduced from strontium chloride or about 187 The effect of water-vapour in the flame is an endothermic reaction when the chloride is a solidL ; and where towards 2000o C. , and The corresponding value for is .
When the chlorides are vapourised in the flame , these endothermic values must be much less and the reactions are easily reversed , as we know by experience wheu gaseous hydrochloric acid or much ammonium chloride is passed into the flame .
We can see clearly also why it is that the carbonic oxide flame does not decompose these salts , the heat of combination of the chlorides being much in excess of that of the oxides , and the reducing power of carbonic oxide , which is the converse of the energy of formation of oarbon dioxide , much less , thus : It is different with cupric chloride , as this in time is reduced to copper .
The Oxy-salts.\mdash ; These are all resolved into either oxides or sulphides , or mixtures of the two , when heated in the flame .
The energy of formation of the oxides is as follows:\mdash ; , and there being very little difference between them , but these values are all below those of lithium and rubidium , where and We have now to take into consideration the reducing power of hydrogen : at high temperatures in flame reactions .
There are two values for the production of gaseous water , the smaller of which is at a temperature 2000o , which we may consider to be the highest temperature all flames capable of melting platinum\mdash ; and * A. Guntz , ' Comptes Rendus , ' 1903 , vol. 136 , p. 1071 .
VOL. LXXIX.\mdash ; A. 258 Prof W. N. Hartley .
Thermo-chemistry of [ Mar. 22 The decreasing value at high temperatures is evidence of the dissociation of water-vapour , which we know does occur at the temperature of white hot platinum .
Now both the above values are below that obtained by the combustion of carbon monoxide .
; it is , therefore , u1mecessary to consider a coal-gas flame apart from one fed with hydrogen , for many of the reductions and dissociation phenomena that can take place in the latter may occur in the former .
The high temperature tends to the dissociation of other oxides than water .
and the alkaline earths are most certainly volatilised in Deville 's blow-pipe flame , whether they are first dissociated or not .
Furthermore , reduction processes are carried on in the solid material , which have not hitherto been suspected , such the reduction of strontium and barium sulphates to sulphides and of beryllia and alumina to the state of metal .
In the metallurgical process of zinc distillation , the presence of carbon mixed with the ore is necessary .
The energy of formation of zinc oxide is .
Supposing the reduction to take place through the medium of carbon monoxide , we have the following:\mdash ; .
II .
Endothermic .
In this case we are considering a solid oxide , not one in the state of vapour ; nevertheless we know that reduction does occur though the reaction is endothermic .
By recent observations it has been shown that zinc oxide is vapourised at 1400o and rapidly at 1700o and this must greatly facilitate the reductiion process .
The reduction of cadmium is more easily accomplished than that of zinc , and hence the cadmium distils over first .
The obvious reason for this is that the heat of formation of the oxide is about 21 less than that of zinc oxide and ; we have , then , the following equation:\mdash ; The reduction of cadmium is a feebly exothermic process .
Let us take another instance , that of the reduction of a solid oxide such as alumina .
The combination of this with oxygen gives an amount of energy which has not hitherto been measured directly , but the heat of formation of the coIloidal hydroxide is known , and it must be .
accepted as being somewhat * F. O. Doltz and C. A. Grammamn , ' ' 1906 , vol. 3 , , also pp. 372\mdash ; 375 1907 .
] Spectra High greater than that of the oxide , but only slightly greater .
It is expressed by the following equation:\mdash ; All attempts to reduce alumina by means of carbon alone , or carbon acting jointly with carbon monoxide on an industrial scale have failed , though many such processes have been devised .
The fact may be accounted for by the following equations:\mdash ; or endothennic ; or , for 1 atom of oxygen removed , Substituting for solid carbon , gaseous hydrogen , we have not a very great difference .
Substituting gaseous carbon monoxide for solid carbon , we obtain or , for 1 atom of oxygen removed , of this The actual energy supplied in these reactions before the aluminium can be reduced is for carbon , carbon monoxide , and for hydrogen respectively , .
The carbon monoxide value is thus shown to be less than that in either of the other two reactions .
It already been proved , first , that alumina is reduced to the metallic state in the oxyhydrogen flame : secondly , that the reduction takes place more easily when the alumina is mixed with some dense form of carbon .
In the first instance , the flame alone supplied hydrogen only as the reducing agent , consequently , the energy required for every atom of oxygen removed was in the form of extraneous heat which was necessary not only for initiating the process , but also it .
When solid carbon was used , this was a somewhat higher figure , being 8 , but this very reduction process with solid carbon would result in the formation of carbon monoxide which could then operate upon another molecule of alumina , which places the extreme limits of the energy required for reduction as between 63 and .
From this it would appear that any r.educing process ?
be carried out in the oxyhydrogen by action if extraneous energy required to initiate naintain the action does not per atom of to be remored .
* This is the figure for the zinc reduction , that calculated for aluluina is 82 .
the flame , namely , beryllia .
The energy of formation of the oxide is but its composition is Pollok has measured the heat of formation of the chloride and of the heat of dissolution of beryllium in aqueous hydrochloric acid .
Comparing these results with the heat of formation and solution of aluminium chloride , we obtain the following figures 1\mdash ; Heat of formation Heat of solution , Similar comparisons , made with magnesium and zinc chlorides , give Heat of formation Heat of solution These figures show that in each case a fall in the heat of formation and solution accompanies the rise in the atomic weights .
Now and the difference , , shows that the energy of mation of the oxide is by that amount greater than the , and as we may assume the value to be approximately correct .
But and also and the difference between magnesium chloride and oxide is and between zinc chloride and oxide , we may assume then that and this number is very probably more correct than that derived from the figures obtained from a comparison of the difference between the oxide and chloride of aluminium , because magnesium and zinc belong to the same group as beryllium , and aluminium does not .
The metal is certainly reduced from the oxide by the oxyhydrogen flame , and , as far as we know , beryllia is not a volatile oxide , nor is it fusibla The reduction must , therefore , be the result of a purely chemical action , and it must also be of a highly endothermic character , as may be seen by the equation following : Endothermic .
: 1907 .
] Flame Spectra at High It may be inferred from the evidence afforded by the tlame spectra , that at these high temperatures the infusible and non-volatile metallic oxides are in a peculiarly active condition , having absorbed a amount of energy from the flame , so that they are approaching the condition of dissociation and are , therefore , subject to chemical interactions with the reducing gases , which otherwise would have no power to reduce them .
But in the cases referred to , the reducing gas is hydrogen , and in all probability it is not the supplied to the flame in the molecular condition , but hydrogen in the atomic or nascent state , resulting perhaps from the dissociation of water , the product of combustion in another part of the flame , which carries with it the additional charge of energy which enables it to initiate and complete the reduction process .
( 1 ) The oxides of calcium , strontium , and barium , are not dissociated by heaG alone , because they show no spectrum in a carbon monoxide flame ; ( 2 ) They are reduced by the combined action of heat and hydrogen in the oxyhydrogen flame and by the action of cyanogen in the cyanogen flame ; ( 3 ) The flame coloration is due to the metal , because not only is the flame spectrum from lime essentially the same as that of the metal calcium , but also the heats of formation of , and have very nearly the same value , and that where calcium oxide can be reduced the other oxides could , on that account , undergo a similar reduction .
Whether the compound of strontium or barium in the flame be a sulphide or an oxide , the spectrum is emitted , but there is some uncertainty as to whether the barium sulphide is not converted into oxide by water-vapour in the flame .
The explanation given by Lenard , of the flame coloration by the alkali salts , appears to be inapplicable to the coloration of the carbon monoxide flame by the haloid salts of the alkaline earth metals .

