]\gt ; 566 Dr. George Senter .
The Role of Diffusion in the [ Mar. 7 , the lowest parts were near the ends\mdash ; the two nodes .
When the stationary wave was of the same length as the trough two mounds arose , one near each end , and so on .
It is suggested that the tidal ridges in estuaries and the chains of sand banks under the sea are formed in this way , by stationary or nearly water waves , and that the sand dunes of the seashore and of the desert , and the clouds in a " " mackerel sky\ldquo ; may be similarly the products of stationary air waves .
" " The of Diffusion in the Catalysis of Hydrogen Peroxide by Colloidal Platinum By GEORGE SENTER , Ph. D. , B.Sc. ( Lond. ) .
Lecturer on Chemistry at St. Mary 's Hospital Medical School .
Comnunicated by Sir WILLIAM RAMSAY , K.C.B. , F.RS .
Received March 7 , \mdash ; Read March 30 , 1905 .
According to a theory of reaction-velocities in heterogsneous systems recently put forward by Nernst , the observed velocities are those with which diffusion and convection renew the reacting material at the boundary of the two phasss , the actual chemical change at the boundary being very rapid in comparison .
Nernst is of opinion that the same considerations apply to the catalytic decomposition of hydrogen peroxide by colloidal platinum and similar substances , but bas not gone fully into the discussion of this particular class of heterogeneous reactions .
In a recent paper on the effect of poisons on the catalysis of hydrogen peroxide by haemase , I have discussed the probable mechanism of catalysis by colloidal particles , and have arrived at the conclusion that the experimental results obtained by Bredig and his pupils and by myself are best accounted for on Nernst 's hypothesis .
In an important paper on this subject , Sand , starting from certain assumptions regarding the size and nature of the particles in a colloidal platinum solution , has calculated the minimum value of the velocity-constant ( which we may call ) to be expected on Nernst 's hypothesis , and finds that it is at least 16 times as great as the velocity-constant obtained experimentally by Bredig .
S Hence , he concludes that Nernst 's hypothesis does not apply to this particular reaction ; the observed velocity is really that of a chemical action .
* ' Zeitschrift fur physikal .
Chemie , ' vol. 47 , p. 52 , 1904 .
'Roy .
Soc. Proc vol. 74 , p. 201 , 1904 .
'Roy . .
Proc vol. 74 , p. 356 ( 1905 ) .
S Bredig and Muller von Berneck , ' Zeit .
fur physikal .
Chemie , ' vol. 31 , p. 258 , 1899 ; Bredig and K. Ikeda , ' Zeit .
fur physikal .
Chemie , ' vol. , p. 1 , 190 ] .
1905 .
] of drogen C'olloidal .
56 In another part of his paper Sand discusses the of Bredig that the reaction-velocity increases more rapidly than concentration of the catalysor , and concludes that this can only be accounted for by increased convection owing to the brisker tion of oxygen when large amounts of catalysor are present .
In the )resent paper it is that when , lvhich is a of the maximum rate at which hydrogen peroxide can be QuppIied to ' the particles by diffusion , is great in comparison with the observed velocity-constant , the hydrogen peroxide concentration at the surface of the particles will be maiutained by diffusion at which does not differ appreciably from the co1lcentratio in the solution , so that increased convection will have practically no effect on the observed -velocity .
From this it follows that if , Sand maintains , convection does modify the reaction- velocity considerably , cannot be large in comparison with , since convection can only modify the vaIue of the former constant .
The question here considered is somewhat analogous to that of a slow chemical action succeeded by a rapid one .
Under such circumstances it is the velocity of the slow ction which is measured , and it is .
clear that an increase in the rate of the rapid reaction will have no effect on the observed velocity : It is not a priori evident that the same considerations apply to a slow chemical action , accompanied by rapid diffusion , since the constants of the two changes are not directly comparable , though both have reference to the rate of tnsference of hydrogen peroxide .
If ( as dejined be times as as th of the chemical action at the snrf of the , ihe concentration at be maintained a less than of the average conc ( inh.ation in solution .
Let be the observed velocity-constant , as defined below , have a value which does not differ from the average con- centration of hydrogen peroxide in the solution , then we have , for the rate of fall of concentratiou in the main bulk of liquid , Further , let be the velocity-constant of the chemical action , anl the concentration of peroxide close to the of any particle , then , if we ttssume that the reaction velocity is to we find for the rate of of concentration the due to chemical action ( which is of coursc equal to the observed rate of fall in the main bulk of liquid , since hydrogen peroxide is only removed at the boundary ) , . .
If this loss is just compensated by diffusion inwards townrds the 568 Dr. George Senter .
The particles , we find , from Sand 's Equation ( 19 ) , p. 367 , that the amount removed by each particle in the time is where is the diffusion coefficient of hydrogen peroxide , is the radius of the particle , and , where is at such a distance from the particle that the concentration there is , which does not differ appreciably from the average concentration in the solution .
The amount removed from unit volume containing particles ( where is times as great , or ( 4 ) , where diameter of a particle .
Hence ( 5 ) .
Further , since KD ; we have ( 6 ) .
If is infinite ( Nernst 's hypothesis ) , and from Equation ( 5 ) we obtain for , which in this case is , the value ( 7 ) , which Sand shows is at least 16 times as large as the velocity-constant determined by Bredig .
Hehce , if Sanl 's value forbe accepted , we must conclude that is not zero , and that experiment therefore does not justify Nernst 's hypothesis .
From Equations ( 5 ) and ( 7 ) we can readily find the relation between and D. If we deal with the case where is 16 times as large as the observed -constant , it is obvious that , or In general , if is times as large as ( 8 ) .
The way in which the velocity of a heterogeneous chemical reaction depends upon the relative of the diffusion\ldquo ; constant and the " " chemical\ldquo ; velocity constant.\mdash ; From Equation ( s ) we find that if is great in comparison with , the concentration at the surface of the particles , with a minimum of stirring , is practically equal to the average concentration in the solution .
Since , it follows that in this case *Sand , , p. 362 .
1905 .
] Catalysis of Hydrogen Peroxide by Colloided .
569 the chemical velocity-constant is scarcely different from the observed velocity-constant K. Just as in the case of successive inter-dependent chemical actions , there are , therefore , three cases to distinguish:\mdash ; ( 1 ) If is great in comparison with ( Nernst 's hypothesis ) , the observed reaction-velocity depends only on , it is that with which diffusion and convection bring the reacting material to the boundar .
( 2 ) If and are comparable in value , the observed velocity will depend on both , and will , consequently , be modified by alterations in either .
( 3 ) If is small in comparison wibh , the observed reactionvelocity depends only on , and will , consequently , not be affected by such changes as stirring may produce in the value of So far as we gone , it would seem that the platintlm oatalysis of hydrogen peroxide is an example of Case ( 3 ) .
The reaction-velocity increases much more the .\mdash ; Bredig has found that , on doubling the amount of his catalysor , the reaction-velocity is trebled instead of merely bei1lg doubled , as one would expectif the catalysor particles act independently of one another , Sand , apparently with good reason , has suggested that this is due to convection currents set up by the rapid evolution of oxygen bubbles in solutions containing large quantities of catalysor .
This view as to the great effect of convection currents cannot , however , be easily reconciled with the conclusion arrived at above , that we dealing with a " " chemical\ldquo ; reaction-velocity .
We have seen that , in the case considered , , with a minimum of stirring , is of the average concentration in the solution .
Ths effect of stirring would be to shorten the diffusion-layer , and bring nearer to the average concentration ; but it is clear that the eatest possible iteration of the reaction-velocity from this will only amoumt to of its valuc , which is quite insufficient to account for Bredig 's resuIts .
In searching for an explanation of this difficulty , ws must first consider what evidence there is for the suggestion that the of proportionality belween and the catalysor concentration is due to the disturbing effect of co1lvection currents .
If tlns is not the case\mdash ; if an on OUIlds can be found for tho deviation in question\mdash ; then we must regard the observed velocity as in all probability a chemical velocity .
If , on the othe ] , the of proportionality is to the effect of convection in modifying the reaction-velocity , it is that canuot } ) in comparison with , and must search for a possible error in the assumptions which have led to this conclusion .
The Part Convcction \mdash ; As Sand has pointed out , ' ' the fact that Kincreases mBredig terneck , .
sor 570 Dr. George Senter .
The Role of Diffusion in the [ Mar. 7 , tion can only be accounted for by supposing that the process taking place on the surface of a partiole influences that of its neighbours by some non-chemical means and it is difficult to imagine any other way in which this can be eflected than by the production of convec- tion currents .
Sand suggests that these ourrents are set up in the liquid by the bubbles of oxygen evolved , but this cannot be regarded as a complete explanation of the observed facts .
On this view we would expect to increase with increasing peroxide concentration ( owing to the more rapid evolution of oxygen , but this is not the case .
It has been found that , though only to a small extent , with increased peroxide concentration .
Further consideration shows , however , that this apparent contradiction can be satisfactorily accounted for on the convection hypothesis .
We must , as a matter of fact , consider thres possible ways in which may be affected by the processes in question:\mdash ; ( 1 ) The particles themselves are in Brownian motion ; they are continually moving into fresh portions of the solution , and thus tend to keep a constant concentration outside the adherent layer through which is taking place .
Further , owing to their motion , they stir the liquid as a whole , and thus , with increasing concentration , the average thickness of the layers on all the particles would be lessened , so that would increase more rapidly than the catalysor concentration .
Owing to evolution of bubbles of oxygen in the stronger solutions , and consequent of the liquid , the diffusion path would be shortened , and , as we have already seen , ould increase more rapidly than the platinum concentration .
The formation of bubbles on the surfaces of the particles would interrupt the diffusion-layer and tend to retard the action .
This process on one particle would not , however , affect that on another , so that , as far as this effect alone , the reaction-velocity would be proportional to the platinum concentration .
In solutions containing the same amount of catalysor and different peroxide concentrations , we wouId expect this disturbing cause to increase with the peroxide concentration , so that the greater velocity to be expected from increased stirring might be more than neutralised by the cause iust mentioned .
We have thus a explanation of the observed fact that decreases slightly with increase of .
The fact that increases considerably during the action , which has up to the present not been satisfactorily accounted for , is probably due to a combination of the two convection effects just discussed .
* Bredig and Ikeda , , p. 4 .
Senter , , p. 213 .
Compare Spring , ' Bull .
Soc. Chim .
Belg vol 19 , 219 , 1900 .
Compare Bredig and Ikeda , loc. , p. 5 ; Engler and Wohler , ' Zeit .
Anorg .
Chemie , ' vol. 29 , p. 13 , 1901 .
1905 .
] Catalysis of Hydgn by Colloidal .
571 In solutions of hydrogen peroxide so dilute that the oxygen evolved would be insnfficisnt to the solution , we would expect to be proportional to the ysor concentration , and independent of the peroxide concentration .
This has not been directly determined for platinum , but Sand , by collating resuIts in different parts of Bredig 's paper , has shown that in solutions containing vety little platinum , is approximately proportional to the platinum concentration .
It is strictly true of the haemase catalysis ; only when the evolved oxygen is more than sufficient to saturate the solution do deviations from the simple logarithmic formula appear .
The fact that all the observations in the platinum and haemase catalysis of hydrogen peroxide are thus satisfactorily accounted for on the hypothesis of convection currents , seems a strong argument in favour of the view that the latter play a considerable part in the action .
Possible errors in the ass ?
bmptions used in ulating 1.\mdash ; If the above views as to the part played by convection currents are accepted , we must investigate the assumptions which have led to the conclusion that the value of is in comparison with that of K. Sand arrived at this result by substitution of probable minimum values for and , and a maximum value for , the diameter of a particle , in Equation ( 5 ) , There would seem to be at least three possible sources of error in the calculation of a numerical value for .
( 1 ) The value cm.2 per sec. ) assumed for , the coefficient of diffusion of hydrogen peroxide , may be too large .
( 2 ) The avcrage diameter of the particles may be greater than the nlaximum value assumed by Sand .
( 3 ) Only part of the surface may be active towards hydrogen peroxide ; the taken may thus be too large .
These three })ilities will now be shortly considered:\mdash ; ( 1 ) Although the diffusion coefficient of hydrogen peroxide has not been determined , there can be very little doul ) the values obtained with substances of similar molecular weight , that the value assumed for is not too large , so that an error from this cause is very unlikely .
( 2 ) The size of the partides.\mdash ; Since , according to quation ( varies inversely as the square of the diameter of the particles , the velocity calculated on Nernst 's hypothesis would correspond with the velocity actually observed if the average diametsr of the particles be four times the maximum value assumed by Bredig and Sand .
Bredig mentions that his solutions .
showed no snspended particles when examined under a microscope apable of detecting objects of the meter 0 it ought to be borne in mind particles Senter , ' Zeit .
physikal .
Chemis , ' vol. 44 , p. 257 , 1903 .
' Anorganiscbe ermente , ' Leipzig , 1901 , p. 21 .
572 Dr. George Senter .
The of in the [ Mar. 7 , in motion may not be so readily detected as stationary particles , yet the uncertainty from this cause would be least in the case of large particles , so that it seems very improbable that particles of the diameter could escape detection under the above conditions .
Zsigrllondy and Siedentopf have investigated the magnitude of the particles in colloidal solutions of gold prepared by reduction ; they find many particles of the diameter and others still smaller .
Stockl and have independently arrived at the same result .
So far as I am aware , colloidal solutions , prepared by Bredig 's method , have not been examined by Zsigmondy and Siedentopf 's method , but it does not seem probable that such solutions would differ markedly , as regards the magnitude of the particles , from those prepared by reduction .
It may , therefore , be concluded that the want of agreement between and cannot be accounted for by too low a value having been taken for the magnitude of the particles .
( 3 ) The suggesfed inactivity ofpart of the surface hydrogen peroxide.\mdash ; Equation ( 7 ) has been derived on the assumption that all the particles are active towards hydrogen peroxide .
If only some of the particles are thus active , the total reacting surface , which is proportional to for eqnal particles , would be diminished , and thus would attain a value more nearly coresponding to K. ( It should also be noted that and consequently would be diminished by the process discussed under the third heading in the section on convection , p. 570 , but this effect would , in any case , be small and would completelv disappear for solutions in which oxygen is not evolved in bubbles , so that we cannot account in this way for the difference between and K ) .
NernstS has already suggested the possibility that part of the platinum is inactive , and some positive evidence to this effect is to be found in papers by Bredig and Fortner and by Liebermann .
The first two investigators found that the palladium catalysis of hydrogen peroxide is greatly accelerated by previous treatment of the palIadium with hydrogen , and the same is true , though to a much smaller extent , of the platinum catalysis .
The latter result has been confirmed by Liebermann .
Bredighas also found that platinum which has been poisoned with CO is more active after recovery from the poison than platinum which has not been so treated .
* The limit of visibility under the microscope will probably depend upon the number of the particles present in a given volume , as well as on the nature and magnitude of their motion .
'Zeit .
fur ochemie , ' vol. 8 , p. 686 , 1902 .
'Zeit .
fur physikal .
Chemie , ' vol. 30 , ) , 1899 .
Compare also obry de Bruyn , ' Receuil Travaux Chim .
Pays .
' vol. 19 , p. 251 , 1900 .
S , p. 65 .
Berichte .
deut .
chem .
Ges vol. 37 , p. 805 , 1904 .
'Berichte .
deut .
chem .
Ges vol. 37 , p. 1519 , 1904 .
' Anorg .
Fermente , ' p. 79 .
1905 .
] Catalysis of Eydrogen Peroxide by Colloidal Platinum .
573 The cause of this increased activity must be , for the present , largely a matter of speculation .
It may possibly be due to the breaking up of aggregates of particles and consequent increase of the reacting surface , but a more probable explanation is that the snrfaces are partly freed from some impurity which prevents action on the peroxide .
The suggestion may be made that this surface impurity is platinous oxide .
It has been shown by Mond , Ramsay and Shields , and by Wohler , that this oxide results when platinum is heated in air , and , having regard to the method of preparation of Bredig 's solutions , it is to be expected that the platinum will be partly oxidised .
This is supported by thp observation of Bredig that a trace of his colloidal platinum is dissolved by sulphuric acid .
Engler and Wohler have shown that platinous oxide acts much less energetically on hydrogen peroxide than platinum itself .
It seems probable that , as many investigators have suggested , an intermediate oxide of platinum is formed in the platinum catalysis of hydrogen peroxide , but most likely a higher oxide , perhaps which can react rapidly with the peroxide .
This would correspond with Manchot'sS explanation of the great accelerating effect of ferrous salts on oxidations as compared with salts .
He has shown that the action proceeds by intermediate formation of a higher oxide of iron , and that the catalytic effect is much lessened when part of the iron has changed to the ferric condition .
Whether this suggestion applies or not , it is clear from what has been given above that there is reason for supposing that the entire platinum surface is not active under ordinary conditions , and this seems the most probable explanation of the difference between as calculated by Sand , and the observed velocity-constant .
Summary oend It has been shown that the deviations from the simple logarithmic formula in the catalytic decomposition of hydrogen peroxide by coIloidal platinum are due to disturl ) ances caused by convection currents .
It has further been shown that when the velocityconstant calculated on Nernst 's diffusion hypothesis is great compared with the chemical velocity-constant , increased convection can produce no eciable effect on the observed reaction-velocity .
In the case under consideration , therefore , since increased convectiou modifies the observed reaction-velocity , there must be some error in the assumptions which lead to the concIusion that the diffusion velocity- * ' Zeit . .
Chemie , ' vol. 2 p. 681 1898 .
'Zeit .
Anorg .
Chemie , ' vol. , p. ( 1903 ) .
'Zeit .
Anorg .
Chemie , ' vol. 29 , p. 13 , 1901 .
S 'Annalen der Chemie , ' vol. 325 , p. 93 , 574 The Role of Diffusion in C'atalysis of Hydrogen Peroxide .
constant is great in comparison with the chemical velocity-constant .
This error is probably to be found in the assumption that the whole surface of the platinum is , under ordinary conditions , active towards hydrogen peroxide .
We are not entitled to assume , from the above considerations , that Nernst 's hypothesis is true for the platinum catalysis , but only that the diffusion-velocity is not great in comparison with the chemical velocity Cases 1 and 2 , p. 569 ) .
Other considerations , however , such as the small value of the temperature coefficient , make it probable that the above hypothesis does apply to this particular action .
Further support for this view may , perhaps , be fonnd in the fact that the deviations from the simple logarithmic law in catalysis by platinum have their exact analogy in the haemase catalysis .
On the " " chemical\ldquo ; velocity hypothesis it would seem rather remarkable that two catalysors of so different origin should show exactly similar behaviour , but this becomes at once intelligible on Nernst 's hypothesis , according to which the chemical action plays quite a secondary part in the reaction-velocities in question .
In conclusion , I wish to acknowledge my obligations to Professor A. W. Porter and Dr. Sand , with whom I have had the advantage of discussing the subject-matter of the present paper .

