66 Sir J. Dewar and Dr. H. O. Jones .
[ Dec. 19 , In conclusion , I wish to record my indebtedness to Professor H. A. Wilson for much kind advice and encouragement given during the course of these experiments .
On a New Iron Carbonyl , and on the Action of Light and of Heat on the Iron Carbonyls .
By Sir James Dewar , M.A. , Sc. D. , LL. D. , F.R.S. , Jacksonian Professor in the University of Cambridge , and Humphrey Owen Jones , M.A. , D.Sc .
, Fellow of Clare College , and Jacksonian Demonstrator in the University of Cambridge .
( Received December 19 , 1906 , \#151 ; Read January 24 , 1907 .
) This paper contains an account of the results of the continuation of the experiments on the action of light on the liquid iron carbonyl ( iron penta-carbonyl ) and the action of heat on the resulting solid compound , differro-nonacarbonyl , Fe2(CO)9 , which were described in a paper on " The Physical and Chemical Properties of Iron Carbonyl , " communicated to the Society in 1905.* The experiments on the action of light on iron pentacarbonyl under varied conditions have resulted in new and interesting observations , and approximate measurements of the velocity of the reaction induced by light have been made and compared with that of other reactions induced by light .
The action of heat on solid iron carbonyl , which we described in our former communication as producing an intense green colour , which colour Dr. Mond stated that he had observed casually , has been further examined , and the experiments now described have led to the discovery of a new compound of iron and carbon monoxide , iron tetracarbonyl , which forms dark green crystals of remarkable stability , exhibiting some very striking properties .
Action of Light on Iron Pentacarbonyl .
It was shown in the former communication that the action of light on iron pentacarbonyl alone or in solution in solvents , with the exception of nickel carbonyl , is represented by the equation 2Fe(CO)5 = Fe2(CO)9 + CO. Carbon monoxide is evolved and differro-nonacarbonyl , an orange red crystalline compound , is produced and separates out .
When pyridine was * 'Roy .
Soc. Proc. , ' A , 1905 , vol 76 , p. 558 .
On a New Iron , etc. 1906 .
] used as solvent , the differro-nonacarbonyl remained in the deep red solution unless very much iron pentacarbonyl had been used , when some crystals were deposited .
The fact that solid is not deposited from a pyridine solution of iron pentacarbonyl under the action of light is attributed* to the solubility of the differro-nonacarbonyl in pyridine , since it was shown that solid was deposited when the solution contained more than 50 per cent , of the pentacarbonyl .
This has been confirmed by using a saturated solution of the differro-nonacarbonyl in pyridine as solvent for the pentacarbonyl .
A 20-per-cent , solution of iron pentacarbonyl in this mixture , on exposure to light for a long time , deposited crystals .
Owing to the deep colour of the solution , the action is naturally a very slow one .
A 20-per-cent , solution of iron pentacarbonyl in pyridine alone would not have deposited any of the solid .
When nickel carbonyl was used as solvent , no change occurred ; the solutions were of a much paler colour than solutions of the same concentration in other solvents , and it was suggested that the two carbonyls united to form an unstable compound { Fe(CO)5Ni(CO)4 } which was analogous to Fe2(CO)9 and was unaffected by light .
Also it was shown that above 60 ' C. light had no apparent action on solutions of iron pentacarbonyl , and that the reaction induced by light was reversed in the dark .
The later experiments were directed towards confirming and extending the above observations .
Action of Light on Iron Pentacarbonyl at Higher Temperatures .
1 n the former communication ( p. 574 ) it was stated that if iron pentacarbonyl or its solutions in ether or petroleum ether were heated to any temperature between 60 ' and 100 ' C. while exposed to light , then no separation of solid occurred , even after long exposure and subsequent cooling in the dark .
This remarkable behaviour has been further confirmed by exposing tubes to light while immersed in the vapours of various liquids and observing whether solid was deposited or not , while hot or after cooling in the dark , , and also by using tubes provided with a manometer of the type used for the velocity measurements described below .
Usually , petroleum ether solutions , containing about 10 per cent , by volume of iron pentacarbonyl , .
were used ; ether solutions were also used .
In ether vapour , 30 ' to 35 ' C. , solid is slowly deposited , and the manometer shows an increase of pressure .
On bright days , an appreciable amount of solid had formed after five minutes .
* Loc .
cit. , p. 575 .
F 2 Sir J. Dewar and Dr. H. O. Jones .
[ Dec. 19 , In the vapour of petroleum ether , 45 ' to 50 ' C. , the solid was more slowly formed , but in half an hour on the same day a distinct amount of crystalline deposit had formed , and the manometer again indicated an increase of pressure .
In acetone vapour , 56 ' C. , no solid was deposited after exposure for five hours to bright sunlight , no increase of pressure was noticed , and no solid was deposited when the tube was allowed to cool in the dark .
In chloroform vapour , 61 ' C. , no change was observed after six hours ' exposure to bright sunlight , no solid separated when the tube was cooled in the dark , and the manometer showed that no increase of pressure had occurred .
That the deposition of solid is a delicate and trustworthy indication of any change is proved by the fact that whenever any solid could be observed the manometer indicated an increase of pressure .
The solid is extremely sparingly soluble in the two solvents , petroleum ether and ether , and so is deposited as soon as formed .
Further , above 46 ' C. , the solid is gradually decomposed , as will be described later , producing a green solution ; no trace of this colour was ever observed in the tubes exposed to light at any temperature between 50 ' and 100 ' C. , provided they contained no solid before they were heated .
An experiment was made in which a solution of iron pentacarbonyl containing a little of the solid differro-nonacarbonyl was heated to 56 ' C. and then exposed to light .
A green colour was produced , but no solid was deposited after several hours .
It is noteworthy that neither on bright days in December nor in July could we observe any decomposition at temperatures above 56 ' C. ; that is , no definite shifting of the transition temperature due to the difference in the intensity of the light could be observed .
These experiments will be discussed later , when the experiments on the reverse reaction have been described .
Approximate measurements of the velocity of the reaction were made in solutions in petroleum ether and in pyridine at two concentrations , and the rate in the two solvents compared .
Since no artificial light had been found to effect the change at all rapidly , it was necessary to depend on sunlight , the intensity of which naturally varied considerably .
Tubes of the same size were made to hold the solutions , and these were filled withi solutions of arbitrary concentrations which were in the ratio of 2 :1 , so that information as to the " order " of the reaction might be obtained .
These tubes were sealed on to a capillary tube , closed at the other end and containing a small pellet of mercury ; measurements of the movement of this mercury index enabled us to determine the pressure inside the tubes .
The tubes were set On a New Iron , etc. 1906 .
] side by side in a glass tank of water and exposed to direct sunlight and examined at intervals until no further change took place .
Calculation of the fraction of the whole decomposed x/ a , where a represents the final increase of pressure and is a measure of the initial concentration of the iron carbonyl , and x is the increase of pressure at any time and is a measure of the amount of decomposition that has taken place , shows clearly that the reaction proceeds as one of the " first order , " since it will be seen that the fraction for each concentration is the same within as narrow limits as could be expected in the circumstances .
The agreement is naturally not good at the beginning , since the solution must become saturated with the gas .
It is probable that great supersaturation is prevented in the case of petroleum ether by the separation of solid ; but in the case of pyridine , where no solid separates , this disturbing effect , which tends to make too small at the beginning , is much greater , and good agreement is only obtained near the end of the reaction .
Another fact must also be taken into consideration as affecting the result : in the case of petroleum ether more solid separates from the more concentrated solution than from the other ; in the case of pyridine it acquires a deeper colour , so that in both cases less light penetrates the more concentrated solution ; in both cases , therefore , the amount of change is relatively less in the more concentrated solution towards the end .
This is shown by the greater values of xfa for the dilute solution near the end .
The following two sets of experiments are sufficient to indicate the kind of result obtained:\#151 ; I. Solvent\#151 ; Petroleum Ether .
Time .
A. Concentration 2 .
B. Concentration 1 .
Pressure .
xja .
Pressure .
x\a .
2 mins .
atmos .
1 01 o-ooi atmos .
1 -02 0-004 12 " 1 -30 0-033 1 -07 0-014 22 " 1 72 0-079 1 -30 0-060 32 " 2 -25 0-147 1 -47 0-094 42 " 2-46 0-16 1 -81 0-16 55 " 2-70 0-19 2 -05 0-21 112 " 3-86 0-32 2-69 0-34 142 " 4-26 0-36 2 -80 0-36 182 " 4-76 0-41 3-25 0-45 247 " 5-40 0-47 3 -55 0-51 28 hours 7 36 0-70 4-33 0-67 2 days 8 1 0-78 5 -2 0-84 00 10 -1 1 -o 6-0 1 -o Sir J. Dewar and Dr. H. O. Jones .
[ Dec. 19 , II .
Solvent\#151 ; Solvent\#151 ; Pyridine .
Petroleum Ether .
Time .
0 .
Concentration 2 .
D. Concentration 1 .
E. Concentration 2 .
Pressure .
xja .
Pressure .
x/ a. Pressure .
xfa .
atmos .
atmos .
atmos .
5 mins .
1 -oo \#151 ; 1 -oo \#151 ; 1 -03 0-014 7 33 1 -oo \#151 ; 1 -oo \#151 ; 1 -10 0-014 10 33 1*03 \#151 ; 1 -oi \#151 ; 117 0-023 20 33 1 T3 \#151 ; 1 -02 \#151 ; 1 -25 0-034 30 13 1 T6 \#151 ; 1 -03 \#151 ; 1 -47 0-064 50 33 1 -31 \#151 ; 1-06 \#151 ; 1 -92 0T26 70 33 1 -49 \#151 ; 1 *12 \#151 ; 2-80 0-24 85 33 1 -60 \#151 ; 1 '24 \#151 ; 3-02 0-27 105 33 1 67 0 13 1 -33 0-093 3T3 0-28 135 33 1-81 0-15 1 -46 0-13 3-57 0-35 175 33 2-47 0-28 2-21 0-34 4-41 0-47 255 33 3-18 0-43 2 -43 0-40 5-00 0-55 325 5 J 3*35 0 '46 2-70 0-48 5-30 0-58 26 hours 4-78 0-76 3 -47 0-70 5-78 0-65 52 33 5*12 0-81 4-07 0-86 6-83 0-80 00 6-09 1 -oo 4-56 1 -oo 8-35 1 -oo A tube set up in precisely the same way with nickel carbonyl as solvent showed during several months ' exposure only very slight changes of pressure , which were due to temperature changes .
In order to compare the decomposition of iron carbonyl with other reactions induced by light , we compared the rate of evolution of gas from a solution of iron pentacarbonyl in petroleum ether with that from a solution of ferric chloride and oxalic acid .
The latter action , which has been examined by Lemoine , * proceeds according to the equation 2FeCl3 + H2C204 = 2FeCl2 + 2HCl-f-2C02 , is exothermic , and at the ordinary temperature only takes place under the influence of light .
For the purpose of the approximate comparison , equimolecular solutions of iron pentacarbonyl in petroleum ether and of ferric chloride were used .
In order to make the latter reaction approximate more nearly to a reaction of the " first order , " two molecular proportions of oxalic acid were used ( more cannot be added , since the solution then changes its colour and ferrous oxalate is deposited during the later stages ) .
Both solutions are yellow , but the colour of ferric salt solution is much deeper than that of the iron carbonyl at the same molecular concentration , hence the comparison is not quite fair , since the amount of change is proportional to the light absorbed , and the latter solution absorbs more light .
* 'Ann .
de Chimie et de Phys. , ' 1895 , [ 7 ] , vol. 6 , p. 433 .
1906 .
] On a New Iron , etc. Two solutions were used , containing respectively 4 per cent , and 10 per cent , of iron carbonyl , and the rate of change compared with two ferric chloride and oxalic acid solutions of corresponding equimolecular concentrations .
The solutions were placed in tubes of the same size , which were placed side by side in a tank of water , and the gases evolved were collected over glycerine .
The volume of gas evolved from the ferric salt solutions is twice that evolved from the corresponding iron carbonyl solutions .
The ferric salt solutions were saturated with carbon dioxide before being introduced into the tubes , yet it is probable that the supersaturation of the solution which is set up causes a lag in the evolution of gas at the beginning .
Later , the deposition of solid from the iron carbonyl solutions cuts off some of the light and further diminishes the rate of this action as compared with the other .
The results of six sets of experiments show that there is no very pronounced difference in the sensibility to light of the two reactions in the more concentrated solutions , but that in the more dilute solutions the ferric chloride and oxalic acid mixture is the more sensitive .
The following table gives the fractions of the total amount of gas evolved from the solutions during the same time:\#151 ; Fe(CO)5 , 10 per cent. Fe , 2'86 per cent. 1IoC,0 " 4'6 per cent. Fe(CO)5 , 4 per cent. Fe , 1*12 per cent. H2C204 , 2*86 per cent. 0 24 0-19 0-12 * 0-28 0-28 0-25 0-26 0-40 0-31 0-37 0-40 0-67 0-40 0-55 0-65 0-85 0-66 0-75 0-75 0-95 0 71 0-79 0-82 0-94 The figures for the dilute solutions do not represent readings taken at the same time as those for the more concentrated solutions .
From the above numbers no very definite conclusion can be drawn , but it is probable that iron carbonyl is more sensitive to light than the ferric chloride oxalic mixture , seeing that the decomposition is nearly as rapid in the former case as in the latter , and the amount of light absorbed by the ferric chloride and oxalic acid mixture is greater than that absorbed by the iron carbonyl solutions .
A rough comparison of the velocity of the decomposition of iron carbonyl with that of the reaction between mercuric chloride and oxalic acid , which was measured by the development of pressure due to the carbon dioxide 72 Sir J. Dewar and Dr. H. O. Jones .
[ Dec. 19 , showed that the former reaction proceeded very much more rapidly than the latter under the same conditions .
Reversal of the Reaction .
The velocity of the reverse reaction , Fe2(CO)9 + CO =\#177 ; 2Fe(CO)5 , which proceeds in the dark , was studied by using the same tubes that were used for the examination of the direct action ( p. 68 ) .
The results show that this is an extremely slow reaction at the ordinary temperature .
Tube E , above , containing petroleum ether , had a pressure of 3 atmospheres after three months , so that only 73 per cent , of the gas had been reabsorbed by the solid ; after 11 months the pressure was still l-68 atmospheres , so that 92 per cent , of the products had recombined .
With pyridine as solvent , the reverse action was more rapid at the beginning , but did not proceed so far .
Thus , after 15 days in the dark , tube D , containing pyridine , had a pressure of 1-8 atmospheres , representing 51 per cent , recombination , while in the petroleum ether tube E only 28 per cent , recombination had occurred in the same time .
The two tubes , A and B , containing petroleum ether , were used for studying the rate of the reverse reaction at higher temperatures .
B was heated in ether vapour .
In two days the pressure fell from 6 to 4T atmospheres , in five days to 2'8 atmospheres , in 10 days to 1*6 atmospheres , and after 16 days the reversal was complete , the solid had disappeared and the pressure had fallen to 1 atmosphere .
Similar results were obtained with other tubes .
Attempts were made to study the reverse reaction at higher temperatures ; in carbon disulphide vapour ( 46 ' C. ) the reaction appeared to proceed more rapidly , but development of a green colour showed that already the decomposition of the differro-nonacarbonyl by heat was beginning , and this naturally vitiated the results .
Results of precisely the same kind were obtained with methylal as solvent .
The solid differro-nonacarbonyl is slightly soluble in the solvent , and on heating gives a red solution\#151 ; it is therefore intermediate in its behaviour between petroleum ether and pyridine ; as in the case of pyridine , the reverse action is also more rapid in the early stages than it is in ether or petroleum ether .
These observations on the greatly increased velocity of the reaction between differro-nonacarbonyl and carbon monoxide at temperatures little above the ordinary temperature are of great importance in their bearing on the absence of any action of light on iron pentacarbonyl above 56 ' C. Iron carbonyl shows no selective absorption , but completely absorbs the more refrangible rays .
On a New Iron , etc. 1906 .
] It was found by spectroscopic examination of iron carbonyl and its solutions that there was no observable difference in the amount of light absorbed at 15 ' C. , and at 60 ' or 100 ' C. The absence of action of light above 56 ' C. must be due , therefore , either to a greater stability of the iron pentacarbonyl molecule at higher temperatures or to the displacement of an equilibrium represented by the equation 2Fe(CO)5 ^ Pef + CO , in the direction of the liquid carbonyl at the higher temperatures .
The fact that the action of carbon monoxide on the differro-nonacarbonyl is so much accelerated by an increase of temperature makes it probable that this is the true explanation .
Lemoine* has shown that the reaction between ferric chloride and oxalic acid exhibits no appreciable acceleration due to increase of temperature ; the action of light on the silver halides is also little affected by temperature .
If the decomposition of iron pentacarbonyl also has no temperature coefficient , which is not improbable , since we were able to observe the action at \#151 ; 180 ' C. , and the reverse action has the normal temperature coefficient of chemical reactions , then the absence of any change at temperatures above 56 ' C. is easily explained .
This explanation is compatible with the assumption that the decomposition of the pentacarbonyl by light is an exothermic change , so that the reverse reaction must take place with absorption of heat , and an increase of temperature would favour the progress of the endothermic change , i.e. , the recombination of carbon monoxide and the differro-nonacarbonyl .
We are continuing the experiments on these and other chemical reactions induced by light and applying other methods in studying them , the results of which we hope to describe later .
Decomposition of Iron Pentacarbonyl Absorbed in Alumina .
During the course of some experiments on the absorbing power of various substances for gases and vapours , it was found that alumina absorbed 2'5 per cent , ot its weight of iron pentacarbonyl from gases saturated with the vapour at the ordinary temperature .
On exposing this alumina to light ( out of contact with air ) it became coloured , first pink , then red , reddish purple , and finally , a dark reddish purple .
Experiments showed that during this action carbon monoxide was evolved in approximately the theoretical quantity , it care were taken to shake the alumina frequently so as to expose it all to light .
* Loc .
cit. 74 Sir J. Dewar and Dr. II .
O. Jones .
[ Dec. 19 , Th solid differro-nonacarbonyl formed in the alumina possibly forms a kind of " lake " with the alumina .
The coloured alumina retains its colour when boiled with pyridine , although alumina will not remove the colour from the red pyridine solutions of differro-nonacarbonyl .
Warm concentrated hydrochloric acid removes the colour , gas is evolved , and ferrous chloride is formed ; nitric acid reacts with the coloured alumina in the cold , but sulphuric acid has very little action on it .
Action of Heat on Dif As stated by Mond and Langer , and confirmed in our previous communication , the solid , when heated alone , breaks up into iron pentacarbonyl , iron , and carbon monoxide .
It was observed that the iron pentacarbonyl so produced was coloured green , but that when distilled the green colour was removed from it .
This decomposition occurs at about 95 ' C. Quantitative experiments were made in which a weighed quantity of the solid was placed in a porcelain boat and heated in a stream of hydrogen to a temperature of about 100 ' to 120 ' C. by means of a toluene or an air bath ; the volatile iron pentacarbonyl was thus carried off as vapour and passed through a glass tube , bent twice on itself and heated in an air bath to a temperature of about 200 ' C. The boat and the coiled tube were weighed after the experiment , and so the amount of left iron and the amount of iron deposited from the volatile carbonyl were determined .
The boat was then heated in a current of oxygen until there was no further change of weight and so a second determination of the iron made as ferric oxide .
The following are typical experiments performed in this way:\#151 ; gramme .
gramme .
gramme .
0-2981 Fe2(CO)9 gave 0-0754 Fe in tube and 0-0223 Fe in boat 0-4505 " 0-1211 " 0-0378 0-6555 " 0-1215 " 0-0725 It is evident that this method does not give consistent results , and even if it did it would probably give values for the iron deposited in the bent tube from the iron pentacarbonyl that were too high , since it has been shown that the iron so deposited always contained carbon .
Other experiments were then made in which the iron pentacarbonyl was collected and weighed .
A weighed quantity of differro-nonacarbonyl was placed in one limb of a U-shaped tube , which was then exhausted and sealed .
This limb of the tube was heated and the iron pentacarbonyl which was formed was condensed in the other limb by cooling this in liquid air .
The iron pentacarbonyl and the On a New Iron , etc. 1906 .
] residue left in the other limb of the tube were weighed .
The residue was found to be iron in such a finely divided state that it was pyrophoric .
The following experiments are typical .
In the first experiment the solid was heated to 100 ' C. and in the second to 110 ' to 115 ' C. gramme .
gramme .
gramme .
0-685 Fe2(CO)9 gave 0-567 Fe(CO)5 and 0D55 Fe 0-587 " 0-453 " 0*039 " In the former communication it was stated that the decomposition might be represented by the equation 2Fe2(CO)9 = 3Fe(CO)g + Fe + 3CO .
The above results confirm this , since the amounts of Fe(CO)5 formed on this view would be 0'583 and 0-503 gramme respectively , and the amounts of iron would be 0-053 and 0'045 gramme respectively .
Since it is shown later that , by the action of heat on differro-nonacarbonyl in presence of solvents below 100 ' C. , a compound , Fe(CO)4 , is produced , it is quite possible that the decomposition was produced in two stages , represented by the equations 2Fe2(CO)9 = 2Fe(CO)5 + 2Fe(CO)4 and 2Fe(CO)4 = Fe(CO)5 + Fe + 3CO .
This view is supported by the fact that when differro-nonacarbonyl is heated to 100 ' C. in a tube filled with carbon dioxide and fitted with a small mercury index manometer , at first the solid is decomposed with formation of a green liquid without any appreciable development of pressure , but as the heating is continued , pressure is gradually developed , iron is formed , and the green colour disappears .
When heated to 100 ' C. with carbon monoxide under a pressure of 75 atmospheres , the solid is at once converted into liquid iron pentacarbonyl and no green colour is produced .
This shows that the recombination of carbon monoxide with the differro-nonacarbonyl is favoured by pressure and the formation of the green compound usually produced is prevented in this way .
These facts are here restated and supported by further evidence in order that they may be compared with the results of experiments on the action of heat on differro-nonacarbonyl in the presence of solvents .
When the cystals of differro-nonacarbonyl are heated with a solvent such as ether or toluene , change begins at a much lower temperature , below 60 ' C. , and the liquid acquires a dark green colour ( as stated above , the green colour is developed slowly , even at 46 ' C. ) .
When enough of the solvent is present Sir J. Dewar and Dr. H. O. Jones .
[ Dec. 19 , the whole of the solid dissolves and no iron or other solid substance is produced ; the solution has such an intense colour that it is opaque , except in very thin layers .
It is clear , therefore , that we have here a different reaction to deal with .
The green solutions on exposure to light lose their colour and deposit the orange crystals of differro-nonacarbonyl and a small quantity of a black solid .
By carrying out experiments in a closed tube filled with inert gas and fitted with a mercury manometer , it was shown that no change of pressure was caused by the action of heat or differro-nonacarbonyl ; hence no gas can have been evolved .
Long continued heating , however , causes a disappearance of the green colour , and slight pressure is developed .
Similar results were obtained when benzene , petroleum ether , brombenzene or iron pentacarbonyl was used as solvent .
When pyridine , alcohol , acetone or acetonitrile was used as solvent , the solution obtained was not green , but some shade of red or brownish red , but again no gas was evolved .
When nickel carbonyl was used as solvent , the solution obtained was yellow , with only the faintest trace of green colour , a black solid was left , probably iron , and gas was evolved .
With chloroform , ferrous chloride was produced , so that the solvent had entered into the reaction ; solid products were also produced when nitrobenzene was used as solvent , and with ethyl acetate or methyl aniline a greenish brown solution and a solid were produced .
The green solutions in toluene , ether or petroleum ether were found in some cases to deposit beautiful green crystals , which , when separated , were found to contain iron and carbon monoxide and to redissolve in the solvent to give a green solution .
After numerous trials , the best conditions for the separation of the maximum amount of these crystals were found , and the following summarises the process used for their preparation .
The solid differro-nonacarbonyl was sealed up in a tube , the air from which had been displaced by carbon dioxide , with toluene in the proportion of 1 gramme of the carbonyl to 20 c.c. of toluene .
The tube was then heated gradually , with occasional shaking , to about 95 ' C. and kept at that temperature until all the reddish yellow solid had disappeared , the tube being meanwhile shaken from time to time .
If shaking is omitted or if the tube be heated to too high a temperature or for too long a time , a quantity of dark coloured powder is formed , but with care this may be almost entirely avoided .
The tube is then allowed to cool slowly , when fairly large crystals of the On a New Iron , etc. 1900 .
] green compound separate from the solution and , after washing with the solvent and drying in a desiccator , were used for analysis .
The yield of the green crystals obtainable is only about 20 per cent , by weight of the differro-nonacarbonyl used .
The carbon monoxide was determined by combustion , care being taken to heat the boat long enough to oxidise any carbon that may have been retained in the iron .
In this case the iron almost entirely remains in the boat and , in some cases when there was no observable loss , good determinations of the percentage of iron were obtained by weighing the ferric oxide ; in other cases there was obviously a loss , indicated by deposition of iron on the combustion tube near the boat , and the percentage of iron was low .
Independent determinations of the iron were made by decomposing a weighed quantity of the crystals with nitric acid and igniting .
The following results were obtained by these methods:\#151 ; ( 1 ) 0T078 gramme gave 0T118 gramme C02 and 0-0513 gramme Fe ( 2 ) 01214 \gt ; \gt ; 0-1258 " 0-0442 ( 3 ) 01744 \#187 ; 0-0845 " Fe203 .
( 4 ) 0-1828 \gt ; ) 0-0868 Found .
CO. Fe .
' ( 1 ) 66-1 33-3 ( 2 ) 65-9 \#151 ; ( 3 ) \#151 ; 33-9 ( 4 ) 33-2 Fe(CO)4 requires ... 66*66 33-33 1 he compound is , therefore , iron tetracarbonyl or some polymer of this .
Ihe decomposition of differro-nonacarbonyl in presence of solvents such as toluene and ether is probably to be represented by the equation Fe2(CO)9 = Fe(CO)5+Fe(CO)4 .
lurther evidence in support of this was obtained by carrying out the decomposition in tubes filled with carbon dioxide and fitted with a mercury manometer .
No change of pressure was produced by the decomposition .
On exposing the tubes , after heating and now containing a green solution , to light , the orange crystals of differro-nonacarbonyl were deposited and pressure was produced , showing that during the action of heat iron pentacarbonyl had been produced .
Iron tetracarbonyl crystallises in short , dark green , lustrous prisms , the density of which was determined by the flotation method in a mixture of ethylene dibromide and alcohol , in both of which the compound is almost Sir J. Dewar and Dr. H. O. Jones .
[ Dec. 19 , insoluble and was found to be 1-996 at 18 ' C. ; the molecular volume is therefore 84 .
If we suppose that a molecule of Fe2(CO)9 ( molecule volume 174 ) decomposes into a molecule of solid iron pentacarbonyl ( molecule volume 128 ) and a molecule of the new iron tetracarbonyl , 174 volumes become 212 and there is consequently an expansion of 38 volumes , or about 22 per cent. Iron tetracarbonyl is a substance of remarkable stability .
On heating to about 140 ' to 150 ' C. it decomposes into carbon monoxide and iron .
When heated under a pressure of 150 atmospheres of carbon monoxide it underwent no visible change until a temperature of 129 ' C. was reached , and then only turned slightly brown ; it was not found possible to reconvert it into iron pentacarbonyl .
Reactions of Iron Tetracarbonyl .
The compound is not attacked by a concentrated solution of hydrochloric acid even on boiling ; it is attacked by cold concentrated nitric acid , but only by concentrated sulphuric acid when heated .
In the last case carbon monoxide and ferrous sulphate are produced .
Aqueous solution of alkalies , whether hot or cold , had only a very slight action on the compound .
Molecular Weight .
Iron tetracarbonyl is soluble in many organic solvents , to which it imparts an intense colour ; as it appeared to be decidedly soluble in benzene , it was thought that a molecular weight determination might be made by the cryoscopic method .
, On examining the solubility , however , it was found to be small , only about 4 per cent , in benzene and much less in acetic acid and alcohol .
Attempts were , however , made to determine the molecular weight , and proved definitely that the molecular weight must he very high .
For instance , a solution of 0-031 gramme of the compound in 12-88 grammes of benzene had a freezing point only 0"003 ' C. below that of the solvent .
Now if the molecule were Fe(CO)4 , ( 168 ) the depression of freezing point should be 0-076 ' C. , so that the molecule would appear to be of the order of { Fe(CO)4}2o , or , at any rate , the compound is a polymer consisting of many Fe(CO)4 units .
Behaviour of Iron Tetracarbonyl towards Solvents .
Iron tetracarbonyl dissolves in benzene , toluene , petroleum ether , ether , acetonitrile , ethyl acetate , methyl and ethyl aniline , acetone , nickel carbonyl , and iron pentacarbonyl to give solutions of a dark green colour .
1906 .
] On a New Iron , etc. It is much less soluble in ethyl acetate and ether than in the hydrocarbons and carbonyls .
The solutions in the above solvents are all stable to heat at 100 ' C. for short periods , but the colour gradually disappears and iron is deposited if the heating be long continued .
This decomposition is quickly completed at 140 ' C. The iron tetracarbonyl can be recrystallised easily from toluene , petroleum ether or nickel carbonyl , when due care is taken to prevent oxidation , to which the solutions are very sensitive .
Pyridine , on the other hand , dissolves iron tetracarbonyl quickly and gives a green solution , which on standing for a few minutes turns red , which change of colour is brought about rapidly by warming .
Alcohol behaves similarly ; only a faint green colour is obtained , as the substance is very sparingly soluble in alcohol ; this solution becomes red on boiling .
Solutions of the iron tetracarbonyl in the solvents , in which it forms green solutions when mixed with pyridine in sufficient quantity , turn red on standing .
It is probable , therefore , that the iron tetracarbonyl when dissolved in pyridine exists in a different molecular state from that in which it exists in the green solutions , that is , it is dissociated into simpler molecules or is combined with the solvent .
The same is true of solutions in alcohol .
Now it was found that acetonitrile and acetone also gave red solutions instead of green when heated with the differro-nonacarbonyl .
Presumably the iron tetracarbonyl is present in these solutions in the same molecular state as in the pyridine solutions , but the effect of these solvents is only strong enough to bring about the conversion of iron tetracarbonyl into this state at the moment of its formation and not at a later stage .
It is noteworthy that iron tetracarbonyl dissolved in either nickel carbonyl or iron pentacarbonyl gives a green solution ; hence , it follows that this solid iron tetracarbonyl is different from the hypothetical Fe(CO)4 , which it was suggested in our former communication* was the first product of the action of light on iron pentacarbonyl , and which then combined with this to form { Fe(CO)5Fe(CO)4 } or differro-nonacarbonyl .
The green solutions of iron tetracarbonyl exhibit a characteristic absorption band in the yellow , while the red solutions show no selective absorption .
Action of Light on Iron Tetracarbonyl .
Ihe green solutions gradually lose their colour on exposure to light , and a metallic looking solid , probably iron , is deposited .
The change is very * Loc .
cit. , p. 576 .
VOL. LXXIX.\#151 ; A. a Mr. F. W. Aston .
Experiments on the [ Nov. 29 , slow and the solutions were too dilute to allow of the detection of evolved gas with certainty : a small amount of a reddish solid , probably ferric oxide , was frequently observed , which was due to oxidation by traces of dissolved oxygen .
The red solution in pyridine appears to be quite stable to light ; no change can be observed after long exposure .
Experiments on the Length of the Cathode Dark Space with Varying Current Densities and Pressures in Different Gases .
By Francis William Aston , A.I.C. , Research Scholar at the University of Birmingham .
( Communicated by Professor J. H. Pointing , F.R.S. Received November 29 , \#151 ; Read December 13 , 1906 .
) During some experiments with various types of vacuum tubes the author was led by the behaviour of one in particular to believe that conditions were possible under which the length of the cathode dark space might be an accurately measurable quantity .
This tube is shown in fig. 1 .
The aluminium disc cathode K is movable and exactly fits the tube containing the anode A. If it is placed to the right in the bulb and a current from a coil passed , tire dark space assumes a highly indefinite and irregular form , such as is indicated .
If , Fig. 1 .
however , K is slid to the left , right into the tube containing A , the boundary of the negative glow becomes a very definite plane parallel to the cathode , its distance from the cathode being the same for all positions of the latter so long as it is inside the tube and the current from the coil kept constant .

