(i)

THE CATALYTIC DECOMPOSITION OF

HYDROGEN PEROXIDE VAPOUR BY OXIDES

AND MIXED OXIDES

BY

ROBERT A. ROSS

A Thesis submitted in accordance w ith the regulations governing the

award o f the degree of Doctor of Philosophy in the Paculty of Science of

the U niversity o f Glasgow.

QOoOooo-
 ProQuest Number: 13850335

All rights reserved

INFORMATION TO ALL USERS

The quality of this reproduction is d e p e n d e n t upon the quality of the copy subm itted.

In the unlikely e v e n t that the a u thor did not send a c o m p le te m anuscript

and there are missing pages, these will be noted. Also, if m aterial had to be rem oved,
a n o te will ind ica te the deletion.

uest
ProQuest 13850335

Published by ProQuest LLC(2019). C opyright of the Dissertation is held by the Author.

All rights reserved.

This work is protected against unauthorized copying under Title 17, United States C o d e
M icroform Edition © ProQuest LLC.

ProQuest LLC.
789 East Eisenhower Parkway
P.O. Box 1346
Ann Arbor, Ml 4 8 1 0 6 - 1346
 ACKNOWLESDMENTS

The author d esires to express appreciation to Dr. A.B. Hart fo r h is

supervision and many stim u latin g discussions during the course of th is

work. Thanks are a lso due to Professor E.S. Spring, P .R .S ., and Dr. E.N.

Dodd, Royal Naval S c ie n tif ic S ervice, fo r th e ir in te r e s t and encouragement.

The receip t of an Admiralty grant i s g r a te fu lly acknowledged.

000G000-
 (Hi)

Iff SEX

PAGE
SMEARY 1

GENERAL UmOHICTIGET 4

(1) C atalytic decomposition o f hydrogen peroxide 5

(a) Liquid phase 5

(b) Vapour phase 11

(2) C atalysis and surface composition 15

(a) Defect structure o f oxides 17

(b) Oxide film growth on m etals 20

(c) C atalytic reaction s on s o lid oxides 21

(d) Physical properties o f oxide c a ta ly sts 25

( i) Magnetic s u s c e p tib ility andc a ta ly s is 25

( ii) Surface area and c a ta ly s is 27

(3) The present approach 28

E X PE R im m 32

(l) Apparatus and Method 32

(a) Plow system 32

(b) Metal evaporation u n it 49

(c) Magnetic s u s c e p tib ility balance 51

(d) Surface area apparatus 53

(e) S ta tic system 56

(f) P e lle t press 57

(g) Agate mortar 58

 (±T)

I 1 D E X (Contd.)

EXPERIMENTAL (Contd.) PAGE

(2 ) Materials 59

(a) Hydrogen peroxide 59

(b) C atalysts 60

( i) Metal s lip s 60

( i i ) Oxides 62

(iii)M e ta l film s 69

(iv ) Eqnimolar mixed oxides 65

(v) Trace im p arities 65

RESULTS

(1) General 66

(a) Determination o f reaction order 66

(b) Determination o f temperature c o e ffic ie n t 69

(c) R eproducibility -with, flow system 71

(2) C atalysis on oxide film s o f pure m etals 73

(a) Oxidised fla sh ed metal film s (Nickel and Copper) 73

to Oxidised metal s lip s 78

( i) N ick el/n ick el oxide 78a

(ii) Cu/Cu^O; Ag/Ag 0; Mn/knO; C0/C 0 O. 81

( i i i ) Zn/znO; Pe/PeO/Ee^O^ 83

(iv ) A l / A l ^ j Sn/Sn02 87

(c) Influence o f oxygen and water vapour on

metal s li p c a ta ly s is 93
 (v)

RESULTS (Contd.) PAGE

(3) C atalysis on pure oxide s lip s 98

(a) Structural examination o f "green* and""black" MO 99

(b) C atalysis on n ick el oxide s lip s 102

(c) PbO; m 20 y , CoO/Co^ 103

(d) ZnO; CdO; CuO; 107

(e) IgO; o^-AlgOy 108 '

(f) Influence o f oxygen and -water vapour on

bulk oxide c a ta ly s is 110

( 4 ) Equimolar oxide mixtures 114

(a) C atalytic a c t iv it y - Mn^O^ s e r ie s 114

(b) C atalytic a c t iv it y - CuO s e r ie s 119

(c) Magnetic s u s c e p tib ility r e s u lts 120

(d) Structural examination 121

(e) R elationship between liq u id and gas phase

C atalysis 123

(5) Influence o f foreign cations in the c a ta ly sis by MO 126

(6) E ffect o f h igh ly paramagneticcations onc a ta ly s is 130

( 7 ) Surface area measurements 131

(8) S ta tic system r e s u lts 134

 (v i)

I H D E X (Contd.)

PASS
d iscu ssio n 137

(1) The method o f in v e stig a tio n 137

(2) A c tiv ity of s in g le oxides in the c a ta ly sis of 139

( 3) The nature o f the c a ta ly st surface 147

(4 ) The decomposition mechanism and o v e r -a ll rate

co n tro llin g fa c to r s 152

(a) Expansion o f k in e tic s 170

(t>) I n i t i a l and c y c lic e ffic ie n c y changes 176

( 5) Arrhenius parameters 178

( 6) The Compensation E ffec t. 180

( 7) C atalysis on NiO and promoted NiO I 83

( 8) C atalysis on equimolar mixed oxides 187

( 9) Surface area and magnetic measurements 191

(10) Conclusion 193

REFERENCES 195

oooOooo-
 - 1 -

SUMMARY

The vapour phase decomposition o f hydrogen peroxide has been studied

on a wide range of metal oxide, mixed oxide and promoted metal oxide surfaces*

The in v e stig a tio n was conducted mainly in a flow system, in the temperature

range J8 - 184°C using high-purity nitrogen as the carrier gas, and HgOg p a r tia l

pressures not greater than 1 mm of Hg. Oxide surfaces were held on vacuum

fla sh ed metal film s , or "specpure” metal s lip s or e lse were presented to the

gas stream as a compressed s lip o f the bulk oxide*

There have been many stu d ies of H 0o decomposition in so lu tio n but few
2 d
in the vapour phase and these have been confined almost ex clu siv ely to work

w ith in a c tiv e c a ta ly sts such as g la s s , SiOg, SnOg and AlgO^. No d if f ic u lt y

was found in obtaining reproducible r e su lts with a ctiv e oxides for the

decomposition e ffic ie n c y ( d ) and a lso for the temperature c o e ffic ie n t (B^).

On the b asis of these two values the c a ta ly tic e f fic ie n c ie s of the oxides have

been compared. In order o f decreasing e ffic ie n c y we fin d for sin g le oxides

MagO^ t PbO AggO > CoO/COgO^ CUgO > NiO "green" ^ CuO v-

FegO^ CdO >■ ZnO = MgO > Sn Og > x-AlgO^ >•>* g la ss (pyrex)

This order taken in conjunction with the work of Garner, Stone, Wagner,

Schwab and others on the decomposition of NgO and the combination of CO and

Og lead s us to c la s s if y the oxides on the tendency they exh ib it to absorb or

r e je c t oxygen which i s in turn rela ted to the defect-type of the oxide.

We s ta te then in the terminology o f the semi-conductor f ie ld 8-

p type (PbO, CUgO e t c .) ^ n - t y p e (CuO, 2n0 e t c .) ^ -in su la to rs

Mg 0, c*C~ Alo0.,) as c a ta ly sts.

* 3
 - 2 -

The c la s s if ic a t io n i s examined, with reference to the treatment of

stoich iom etric NiO with Li^O and Ga^O^, which produce strongly p-type and

n-type oxide mixtures respectively* Two s e r ie s of oxides mixed in stoich im etric

proportion are a lso examined. These are the p re-sp in el mixtures MngQ^/XO,

where X«Cd, Zh, Pb, N i, and CuO/YgO^ where Y*A1, Fe, Co. The e f fe c t of

annealing temperature i s ca r efu lly examined in the former case and i t i s

shown th at the best c a ta ly sts are those in which in te r d iffu sio n o f ions has

begun between oxide c r y s t a llit e s , but no stable sp in el lias yet formed. The

most str ik in g e f fe c ts observed were in the synergesis obtained with the

ii> ii
Fe and Cu oxide mixture and the fa c t that in ert ZhO seemed to promote

the already extremely a ctiv e

On a l l c a ta ly s ts , temperature c o e ffic ie n ts were measured by making fa st

temperature changes. These values were, in most cases, dependant to some extent

on the temperature from which the change was made - the higher temperature

surfaee givin g the highest temperature c o e ffic ie n t. Energies of a ctiv a tio n

calcu lated from temperature c o e ffic ie n ts range from 4 - 34 K cals/m ol.

K in etic stu d ies a t constant temperatures cm both p and n-type oxides

v iz :- FegO^, CoO, NiO y ie ld :-

d ( PuiOj) u h .o * •f r j *

d t .

A mechanism is put forward to explain t h is . I t i s suggested that

c a ta ly sis depends on c y c lic tran sfer of electrons between the surface and

HgOg or i t s ra d ica ls and io n s. Og- or 0 adsorbed a t p-type oxides (witn

attendant p o sitiv e h o le s), or la t t ic e oxygen vacancies (with attendant electro n s)

 - 5 -

in the surface of n-type oxides, are suggested as su ita b le s it e s fo r tne

c a ta ly tic cy c le . Increase of oxygen pressure accelerated the c a ta ly sis both

on n-type and p-type oxides - probably actin g by ra isin g the surface

concentration of such sp ecies as Og •

The c a ta ly st surface i s in slow ly attain ed thermal equilibrium and also

i s subject to change w ith varying HgOg pressure. These e ffe c ts are observed

and discussed.

To a sse ss the p o s s ib ilit y of lin k in g the flow r e s u lts with those of

previous workers with s t a t ic systems, a b r ie f in v estig a tio n has been conducted

in a s t a t i c apparatus. The conclusion i s that comparison between s t a t ic and

dynamic systems i s not easy to make.

Information on the physical properties of the oxide systems has been

obtained from magnetic s u s c e p tib ility , surface area and X-ray measurements.

The sig n ifica n c e of which i s f u ll y discussed.

GEBJEEMi IM?EODU'CTI ON

There have “been few in v estig a tio n s of the heterogeneous c a ta ly s is of

hydrogen peroxide in the vapour phase. Those few have been conducted with

r e la t iv e ly in e rt surfaces such as g la ss and aluminium. Yet there seems to

be no inherent obstacle to the study by a flow method of the c a ta ly sis by a

wide range of surfaces, including very a c tiv e ones. Such a study would y ie ld

r e s u lts of in te r e st in the sp e cia l f ie ld of c a ta ly sis (So far studied in

d e ta il only in the aqueous p h a se.) and in the wider context of the general

mechanism of c a ta ly s is .

The present work describes a study along these lin e s . I t extends some

e a r lie r d e ta ile d work in th is laboratory on one oxide, and considers variation s

in c a ta ly tic e ffic ie n c y which are to be associated with chemical structure

im p u rities, the defect structure of the s o lid , surface changes occuring during

c a ta ly sis i t s e l f , and the e f fe c t of thermal pretreatment of the s o lid .

To begin w ith, a b r ie f o u tlin e of past work on the decomposition of

hydrogen peroxide in the liq u id phase w ill be given. Since in that f ie ld

much progress has been made towards a sa tisfa c to r y understanding of the

c a ta ly tic process. This w ill be followed by a section on previous vapour

phase work with hydrogen peroxide; and a review of the d efect structural

p roperties of s o lid surfaces and bulk oxides, which have recently come to be

recognised as being important in chemisorption and c a ta ly sis on oxides. In

t h is connection an ou tlin e w ill be given of some fundamental stu d ies on the

s o lid oxide catalysed decomposition of KgO "khe CO/O2 oxidation reaction

which are c le a r ly rela ted to the present problem.

 - 5 -

(1) C ata ly tic Decomposition of Hydrogen Peroxide

(a) Liquid Phase

Hydrogen Peroxide was discovered in 1818 by Louis - Jacques

Thenard, who observed that i t was rather unstable, presented a novel

form of chemical combination and that i t could be decomposed vigorously

by certain substances without d iscern ib le e f fe c t on the decomposing

agent (1 ,2 ) Between tne date of h is discovery and 1856, Thenard

published upwards of th ir ty papers describing the properties and

preparation of the compound* He also made comprehensive observations

on the r e a c tiv ity of towards many metals and m eta llic oxides,

including Ag^O, MqO^ and PbO, and noted that "in these decompositions,

chemical a ctio n i s evid en tly m issing; i t i s necessary then to attrib u te

these actions to a physical cause which might probably be e le c t r ic ity " .

This work of Thenards (5 ,4 ) co n stitu tes one of the e a r lie s t recognitions

of c a ta ly s is .

Since Thenard's early observations the advance of knowledge about

hydrogen peroxide decomposition has been hindered by experimental

d iffic u ltie s . Much work was carried out with hydrogen peroxide of

questionable p u rity. In many cases observations were not made in regard

to the e ffe c ts of concentration, temperature and the sta te of the c a ta ly st.

The e a r lie s t system atic work on d ilu te aqueous solu tion s was rep o rted

in a long c la ss ic a l se r ie s of papers by Bredig ( 5 ), onwards from I 8 9 7 .

He used metal so ls as ca ta ly sts which he recognised as being analogous

in action to the b io lo g ic a l enzyme, catalase, Bredig found that the

 - 6 -

decomposition was f i r s t order in HgOg concentration on most no"ble

metal c a ta ly s ts , and that there was an optimum rate at about pH 12*

These general f i ndings have been confirmed recen tly in d eta iled work

by T eleto f (6 ),

The b asis of the ea rly explanations of c a ta ly sis was an a ltern a tin g

oxidation - reduction cycle e .g . as proposed fo r c a ta ly sis by io d in e :-

Oxidations 2 HI + HgOg » Ig + 2 HgO.

Reductions 1g + HgOg * 2 HI + 0g.

Such explanations were only moderately su ccessfu l in explaining

the experimental resu lts* The f i r s t major advance in describing a

mechanism of hydrogen peroxide c a ta ly s is can be found in the c la s s ic a l

work o f Haber and W illsta ter (8) and Haber and Weiss ( 9 )* dealing with

the homogeneous c a ta ly sis by iron ions in d ilu te aqueous solutions*

To describe the decomposition by ferrous s a lt s in acid and neutral

so lu tio n the follow in g cycle was proposed:-

I’e++ « 2o2 - Fe5+ +0H~ +0H (1)

OH +h2°2 * HgO -tflOg (2)

ho2 = Og + HgO + OH (3)

■«202
OH - Pe5+ + QH“ (4)
The novel feature o f th is cycle was the use o f the rad icals HO

and HOg in a reaction scheme which permitted bimolecular step s more

p la u sib le than the simple oxidation processes represented above by

stoicheom etric equations. Another origin al suggestion incorporated

in the scheme was the importance of electron tran sfer steps as

represented by equations ( l ) and (4 ). Regarding the scheme as a chain

 - 7 -

process ( l ) and (4) are in itia t io n and termination reactions resp ectively

w h ilst (2) and ( 3 ) are propagation step s.

An outstanding development was found in the confirmation of the

production of OH rad icals in tne work of Baxendale, Evans and Park (10),

who showed that in the course of the above process vin yl polymerisation

could be in itia te d and OH detected in the polymer. Similar demonstrations

were made by Dainton ( l l ) fo r the decomposition in itia t e d by lig h t and

io n isin g radiation.

Later, Barb, Baxendale, George and Hargrave (12) modified the

o r ig in a l theory follow ing further work on the system iron 4- HgOg in

so lu tio n . They concluded that the reaction producing oxygen was not

th at of HOg with but HOg or Og with Fe^+. In ad dition, a fte r

further and extended experiments, Weiss (1 3 ) Las revised the origin al

I-j
mechanism to include a reaction between Fe and HOg as a chain-ending

step .

Evans, Hush and Uri ( 1 4 ) increased the understanding of the

postulated reaction steps by providing ca refu lly considered thermodynamic

data, based c h ie fly on experimental work of Baxendale, Evans and uri ( 1 5 )

fo r reactions between metal ions and the ra d ica ls.

The resu lts of these workers using k in e tic and the thermodynamic

arguments can be summarised for one ty p ica l c a ta ly tic process v iz . the

decomposition of d ilu te HgOg in acid solu tion by f e r r i c ions in the

follow in g scheme. The sequence of steps is due to the k in e tic argument,

the free energy changes are in s e rte d to in d icate the p la u s ib ility ci nio

s te p .
 - 8 -

JTe3* -m202 = HOg +Fe+ f +H+ (5)

&Gr a= 3IK. c a ls/m o l.

Fe44 +E2°2 » HO 4- OH” 4- Fe^4 (l)

&G * 20K. c a ls/m o l.

OH 4-H2 02 * H024- H2 0 (2 )

&G ss 19K. cals/m ol*

H02 =H + + 0g ' (6 )

&G = 3^. cals/m ol*

Fe54 4-02” = Fe44 4-02 (7)

&G = “34-K. ca ls/m o l.

Fe4 4 4-H0g - Fe4 4-HOg” (8)

&G - 7K« cals/m ol.

Fe4 4 4-HO = Fe4 4-HO” (4 )

&G = - 19 K cals/m ol.

The workzuig o f *klixs ciiaxn ms olioctixsiB. cLgjpghcLs on tfa.G fo3?iBQ#i#ion oi 9,

minute amount o f Fe44 in an electron tran sfer reaction (5)* This then

reacts in ( l ) forming hydroxyl radical which then produces perhydroxyl

in (2 ). I t may then replenish the l^e44 concentration by reducing Fe"4 ,

(7 ), v ia the anion, Og , of HOg which i s fa ir ly strong acid. The main

featu res of the c a ta ly tic process are represented by the sequence of

reaction s ( 7 ) , ( l ) and ( 2 ), the equilibrium ( 6 ) being assumed.

These p rinciple featu res, derived to explain homogeneous c a ta ly s is ,

have been used by Weiss (l6 ) to form the 'basis of a discussion of the

heterogeneous decomposition of HgOg solu tion on metal surfaces. The rate

co n tro llin g and chain in it ia t in g step was again suggested as an electron

 - 9 -

tr a n sfer to the ^ 0 ^ - the electron coming from the metal - with the

formation of OH and CH. The theoiy was te ste d by varying the e le c t r ic a l

p o ten tia l applied to a metal electrode immersed in so lu tio n and

observing the e ffe c t on the evolution of oxygen. I t was found that the

c a ta ly tic reaction was favoured by adjusting the p o ten tia l so as to

increase the a v a ila b ility of electrons at the metal surface. This

electro n tran sfer mechanism received added experimental support by the

work of Dowden and Reynolds (l7)> who observed that with a range of

Cu/Hi a llo y s , the copper rich mixtures were the best ca ta ly sts in

neutral solu tion between 20 and 80°C. The rate of ^ © 2 decomposition

was shown to increase with increasing copper content in accordance w ith

a process controlled by the rate of transfer of an electron from the

metal to the substrate.

Weiss (l6 ) suggested a lso that the formation of a dissolved mctw.1

ion might r e su lt from the i n i t i a l electron exchange between the m e ta l

and This complication is always present in solu tion work and had

previously been referred to by Wiegel (18) who showed that c a ta ly sis by

f in e ly divided s ilv e r was accompanied by solu tion of the metal u n til tne

product Ct + CTT~ had reached a steady value. J eiss interpreted Wiegel',

* Ag ‘ hSJ^-
r e s u lts as fo llo w si-

Ag (so lid ) HipOg = AS’*" Cal * ) +0H + OH (9)

Ag (so lid ) HrOH = Ag^ (aq.. ) +QH (lO)

H202 = H+ +H02~ (11)

aT" (ag..) +H02 = a£ +H02 (12)

By applying the steady sta te p rin cip le, i t may he shown t h a t t h i s cy cle
le a d s to the conclusion th a t the product C, - must assume a
Ag n02
steady value, Reactions (.11) and (l2 ) illu s t r a t e the importance of pH.

High pH’s favour c a ta ly tic decomposition of HgOg, and i t i s seen here

th at a high concentration of HOg, which would occur at high pH values,

should have th is e f f e c t ,

Broughton and Wentworth ( 1 9 ) studied the c a ta ly sis hy MhOg and

I|
showed that Mn accumulates in solu tion u n til the s o lu b ilit y product
2
^ reac^ie^* They further asserted , on the basis of ra d io -
1j
a c tiv e tracer experiments, that Mi in solu tion i s in a constant sta te

o f exchange with the manganese ions in the oxide surface, Broughton et a l.

( 2 0 ) were not able to find evidence for such an e f fe c t with cobalt oxides;

so in the absence of other support i t is not possible to conclude that th is

exchange e ffe c t i s general.

Recent work by Voltz and Weller (21) has been directed towards

obtaining a correlation , for aqueous HgOg decomposition, between c a ta ly tic

a c t iv it y and surface oxidation s ta te . The ca ta ly sts used were chromia

and chromia/alumina which had been p re tre a te d a t 5 0 C°C in an atmosphere

o f hydrogen or oxygen, and i t was found that the oxidised ca ta ly sts were

6-8 times more a ctiv e than the reduced ones. On the same theme, Liooi and

Selwood (22) showed th a t manganese oxide ca ta ly sts supported on HL.rO,,

*- j
gave a peak a c t iv it y in the decomposition reaction , which was shown by

magnetic s u s c e p tib ility measurements to occur a t 3,6^ In . They in te rp re te d

th is peak as being due to a maximum induced oxidation s ta te of the surface

h eld manganese.
 - 11 -

Schwab ( 2 3 ) examined the liq u id phase decomposition of by

the sp in els MgO. Feo0-, and ZnO. Eeo0 , , in order to assess the e ffe c t
25 2 y
I1 I-I-
o f replacement in th is structure of Mg b y Zn . The pure magnesium

f e r r it e was the "better c a ta ly st hut a lOfo replacement by Zn improved

i t s e ffic ie n c y to a peak JQf/o higher. In these in v estig a tio n s the authors

(21, 22 f 2 3 ) assumed that hydration and d isso lu tio n e ffe c ts were

unimportant. However, working in the liq u id phase always e n ta ils a

great deal o f uncertainty in th is aspect.

(b) Vapour Phase.

Hydrogen peroxide vapour, i f not d ilu ted too much by water vapour*

or other foreign gases, decomposes to water and oxygen with the

ch a r a cte ristic s of a flame as f i r s t demonstrated in the laboratory by

Hart (2 4 ), and la te r studied in d e ta il by S a tte r fie ld et a l. ( 2 5 ).

The k in e tic s of th is process have been d if f ic u lt to study because

of the in terven tion of the heterogeneously in itia te d decomposition.

Recent work, however, in flow systems has been able to sep arate the tuo.

S a tte r fie ld and S tein ( 2 6 ) showed th a t in glass v e sse ls a heterogeneous

process with energy of a c tiv a tio n approximately 10K. cals/m ol.

(calcu la ted assuming f i r s t order k in e tic s ), was superseded above 500 °C

by a surface independant reaction with E = 5Q'~6QK cals/m ol. S te in ’s

work o ffers support for the theory that decomposition is in itia te d ,

here thermally, by s p l i t t i n g the oxygen - oxygen bond to give two (hi

ra d ica ls which would require cals/m ol (14).

hs s t a t e d above, t h e r e nave be on com paratively few p r e v io u s

in v e stig a tio n s on the decompos i t ic-n 01 hydrogen peroxide by s o lid

 ~ 12 ~ '

C a ta ly sts in the vapour phase. In 1923> Hauser ( 2 7 ) p a s s e d t h e v a p o u r m ’

3 > ^2^2 0V<3r ^ p a c a i c a ta ly sts in the form of gauzes and found that

copper and iron destroyed l i t t l e while asbestos destroyed most of the

Such an observation could, of course, have l i t t l e relevance to

the chemical nature of the ca ta ly st since the surface areas were very

d ifferen t and not measured. In the same year, Hinshelwood and Pritchard

(28) made a b r ie f k in e tic study of the decomposition in g la ss bulbs at

76°C and found the reaction to be heterogeneous and of the f ir s t order

in concentration.

Later, Elder and Rideal (2 9 ) made a more d eta iled study of the

decomposition on quartz, g la ss and platinum wire in the temperature

o
range 80-100 C, using 6Qp vaPour* ®le k in e tic s were found to be

f i r s t order on the platimum and g lass surfaces, but on quartz zero order

was in d icated and the decomposition came to an end a t 15 y com pletion.

The quartz r e su lts were c r i t i c i s e d by Rosenberg and Kistiakowsky ( 3 0 )

who found poor r e p ro d u c ib ility in quartz v e sse ls at 85 “ 90 °C, Hot t i e

re a c tio n always went to com pletion.

Reuter and Gaukham (3 l) se t out to compare the decomposition cn

the same c a ta ly s t in the liq u id and vapour phases. Por the vapour phase

work they d i s t i l l e d 3 0 y over a c a ta ly s t c o n sist Rig of smooth or

p la tin iz e d platinum, or manganese or lead dioxide. They concluded that

the mechanism of the vapour phase decomposition was o s s e n tra ily s im ila r

to th a t in s o lu tio n . The re a c tio n was fcunu to be f i r s t order in bom.

ca ses, but the r e su lts published do not exclude the p o s s ib ilit y of

d iffu sio n 01 so tno oatuj.ys 0 sum ace as m y tne ec; ....on r a iio

c o n tro llin g f a c to r .
 - 13 -

iuackenzie and R itch ie (32) made a much more accu rate and d e ta ile d

study of th e system w ith q u artz a t p ressu res v ary in g from 0 .6 - 1 .2 mm jig.

The procedure involved in tro d u c tio n of hydrogen peroxide by v a p o risa tio n

from a quantity of h igh ly concentrated liq u id and the r a te of re a c tio n

was follow ed by th e p re ssu re change. They found the re a c tio n to be second

order with re sp e c t to HgOg in ’^eniPera'fcure range 15-150°C. (This is

the only case where second o rd er dependance was observed, as r e p o r t e d

l a t e r , s im ila r r e s u lts were found in a s t a t i c system during the p r e s .r e t

i n v e s t i g a t i o n , which are d if f e r e n tly i n t e r p r e t e d ) .

Again in s t a t ic systems Baker and Ouellet (33) &nd Giguere (34-),

carried out a wide se r ie s of in v estig a tio n s using d ifferen t forms of

g la ss and quartz, and with v e s se ls coated with t in and aluminium. The

reaction was generally f i r s t order in the range of conditions used,

(80-200°C and 50-100 mm. Hg pressure); and for the g la ss surfaces

apparent energies of a ctiv a tio n of 1 3 *4-~l9K« cals/m ol. were calculated

from the v e lo c ity constants at d iffe r e n t temperatures. Soft glass was

the most a c tiv e ca ta ly st with pyrex, tin and aluminium next in descending

order of a c t iv it y . Quarts, which at low temperatures gave a marked

period o f induction before any decomposition was observed, "was the le a s t

c a ta ly tic surface. However, Giguere*s work lends i t s e l f to the c r itisc ism

th a t hot chromic acid was used as the cleaning agent which v/ould probably

leave chromium oxides on the g lass s u rfa c e . This otherwise very accurate

s e r ie s of experim ents must be considered to be f a l l i b l e on t h i s p o i n t .

TicLane (35) has studaec. tne decomposition on pyrex graco ./a u.n_

w ith o u t a co a tin g of b o ric a c ia , in the temperabure r e g i o n -iT a-yia a an

a flo w sy stem w ith p a r t i a l p r e s s u r e s 1 -2 mm. Hg. The r e a c t i o n r a t e

was found to obey a f i r s t order r e la tio n , and was somewhat slower in the

coated v e s s e ls . The in h ib ito r actio n may be due to the conversion of the

b asic oxides in the g la ss surface to in active borates. H arris (3b) male

a b r ie f study of the decomposition on a quartz re a c tio n tube by

bubbling nitrogen a t one atmosphere through 65/* aqueous held at

52 °G. He found th a t the decomposition r a te was in c re a se d by coating

the tube with potassium chloride, and reported a h a lf l i f e of 0*5 se c s,

at 520°C in a clean v e s s e l.

Ysry re c e n tly , Giguere and Liu (37) have measured the ra te of

decomposition of HgOg vapour by a s t a t ic method, at pressures between 0.2

and 20 mm Hg in the temperature range 300-600°C, using g la ss v e s s e ls .

The reaction was found to be f i r s t order, with an a ctiv a tio n energy of

10-12K cals/m ol below 400°C and 48K cals/m ol above th is temperature,

which these authors take as the tra n sitio n point from the heterogeneous

to the homogeneous decomposition process. In th is instance the cleaning

procedure adopted was to tre a t the v e s se ls with hot fuming sulphuric acid

follow ed by hot concentrated hydrogen peroxide, which would seem to be a

more sa tis fa c to r y method than that employed in the e a r lie r work (34)*

I t w ill be evident from the above account that the vapour phase

work has not progressed far towards y ield in g a clear picture of the

mechanism o f the c a ta ly tic reaction. No d eta iled study of the surfaces

normally considered to be most a c tiv e , such as s ilv e r or manganese dioxide,

has been made. In addition, no conclusions have been drawn as to the

connection between the chemical nature of a ca ta ly st and i t s a c t iv it y .

 - 15 -

However, in the case of the work a t low pressures, f i r s t order

dependance on hydrogen peroxide concentration i s u su ally found which

suggests that the rate co n tro llin g step could he eith e r (a) gas phase

d iffu sio n of HgOg? (b) d iffu sio n through a layer of water vapour held

a t the surface or (c) adsorption of

In a l l these processes the rate of decomposition would depend on

the f i r s t power of the hydrogen peroxide p a r tia l pressure. D iffusion

seems to be ruled out by the high and variable energies of a ctiv a tio n
found by Giguere (34) and a lso by the s e n s it iv it y of the c a ta ly tic

a c t iv it y to minor changes in the treatment of surfaces. I t may then

be concluded that in the majority of the experiments reported, adsorption

i s probably the ra te-co n tro llin g process.

(2) C a ta ly sis and Surface Composition

Whether or not adsorption i s the k in e tic rate con trollin g step in a

p a rticu la r c a ta ly tic process, i t i s the unique step without which the reaction

i s not p o ssib le . Furthermore i t i s equally evident that i t i s chemisorption

which i s relevant since by d e fin itio n th is i s the type of adsorption which

in volves the chemical bonding o f the adsorbed sp ecies. (This is not to say

th at in a given process van der Waals adsorption may not be important as for

example by maintaining a loose surface layer of reactan ts}products or

interm ediates) •

The question which a r ise s i s what kind of surface feature w ill be

relevant to the chemisorption of an HgOg m olecule. We may approach th is by

considering what type of chemical bond weakening would lead to the

decomposition of Hp0o.
 - 16 -

As we have mentioned, thermal decomposition in the vapour phase

(McLane (35) and. S a tte r fie ld and S tein (2 6 ) ) seemsto he associated with

rupturing of the 0-0 bond, w hile lig h t is held to act sim ila rly (see Dainton

(ll) ) and so lu tio n phase homogeneously catalysed decomposition mechanisms

involve breaking of the 0-0 bond in an molecule a fte r acceptance o f an

electro n . Thus i t i s natural to consider as a f i r s t p o ssib le surface process.

H202 + 2S * 2S0H

Where S represents a surface s it e and SOH an OH radical adsorbed on such a

s ite . Since an molecule which has accepted an extra electron i s in an

unstable s ta te w ith regard to the 0-0 bond, (the extra electron goes in to an

antibonding o r b ita l and reduces the net bonding between the oxygen to zero ),

i t i s a ttr a c tiv e to consider that the most lik e ly precursive step to c a ta ly sis

is

B20 2 + 2S = 2S+ - H202'

. 1 1
= 2S*r...OH 2

in which the f in a l product can be thought of as a "loose" hydroxide - which

may in su ita b le conditions become normal hydroxide or react further to give

water and oxygen. Consideration o f th is mechanism may be l e f t u n til la te r

but i t would not be in a p p ro p ria te at th is stage to consider the properties of

oxides which may help to an understanding of the p ossib le nature of S.

The oxide surface i s , of course, a complex system. In addition to

_L 1 t _
surface metal io n s, M , M e t c . and oxide ions 0 , there w ill be surface

io n ic d efects of eith er kind (anionic or ca tio n ic ) with associated electron ic

ir r e g u la r it ie s . There w ill a lso be, on an oxide prepared normally, adsorbed

im p u rities such as excess oxygen or water. We sh a ll a lso have to consider the

 - 17 -

p o s s ib ilit y o f trace impurities#

(a ) D efect Structure o f Oxides#

In the past f if t e e n to twenty years a new conception o f s o lid s ta te

physics has arisen through the postulation of defect structures to explain

many chemical and physical properties of s o lid s , The theory of these

phenomena has been described in the c la s s ic a l work of Mott and Gurney (38 ),

and extensive experimental and th eo retica l in v estig a tio n s have been

completed by many workers which has been in tegrated in to a tr e a tis e on

the e x istin g conception of so lid s by Shockley (39)*

In a stoich im etric c r y sta llin e compound displacement of atoms from

th e ir regular l a t t i c e s it e s may occur. The resu ltin g in t e r s t i t ia l atoms

and vacant l a t t i c e s it e s are d istrib u ted at random through the c r y sta l.

Two simple lim itin g cases of defect formation are p ossib le

(1) Equal concentrations of vacant cation and vacant anion s it e s

(Schottky d efects)#

(2) I n t e r s t i t i a l cations or anions with a corresponding number of

vacant la t t ic e s it e s (Frenkel defects)#

Cases ( l ) and (2) may be considered to be inherent thermal d efects,

which a t normal temperatures are few in comparison to many other

deviations from stoichiom etry which have come to be recognised. For

example, in the case of metal oxides, excess metal may be incorporated

by converting some cations e f fe c t iv e ly into neutral atoms or in to cations

of lower valency, w h ilst the accomodation of excess oxygen involves

the presence of cations of higher valency. Each atom of excess metal

represents an extra cation plus a trapped electron in the la t t ic e and

 - 18

c o n stitu te s a f i l l e d impurity le v e l from which, by flu ctu ation s of

thermal energy, the electro n may he excited to the conduction hand of

the oxide. A cation of higher valency, in an oxide with excess oxygen,

i s a s i t e o f electron d e f ic ie n c y ; an empty impurity le v e l to which an

electron may he ex cited from the o rig in a lly f i l l e d valency hand of the

c r y s ta l. Electrons in the f i r s t case, and p o sitiv e holes (valency

electro n vacancies) in the second case are thereby rendered mobile,

g iv in g r is e to e lec tro n ic conductivity. On th is basis m etallic oxides

may he c la s s if ie d according to th e ir electro n ic (semi-conducting)

c h a r a c te r istic s.

Cuprous oxide i s an example of a '’p-type” semi-conductor in which

the e lec tr o n ic conductivity i s due to p o sitiv e holes a risin g from

oxygen excess or metal d efic ie n cy . The e le c tr ic a l n eu tra lity of the

c r y s ta l i s preserved by the existence of an equal number of cupric ions

and vacant cuprous cation s it e s in the la t t i c e , as fo llo w s:-

Cu+ ® Cu+ a Cu+

0= 0= 0“ 0=

Cu+ Cu+ ® Cu+ a

TShere ® = p o sitiv e hole

and O 35 vacant Cu* s i t e .

p-type oxides exh ib it a tendency for the metal to pass to a

higher valency sta te by adsorption of oxygen. The p o sitiv e hole

produced in the electron tran sfer d iffu ses in to the la t t ic e to some

ex te n t, w h ilst the adsorbed oxygen may be b u ilt in to the surface (41).

 - 19 -

This process has been studied very f u lly for Cu^O by Gamer and

h is coworkers (44, 46, 47) whose r e s u lts , among other th in gs, demonstrated

a marked r is e in conductivity when oxygen was adsorbed by Cu^O, This

could be explained by a creation of electron d efects thus

Cu+ + 02 * Cu+ ® -t 02~

where Cu*~ i s here a surface ion and the 0^ i s an unstable adsorbed

m olecule. The above workers preferred to consider that oxygen i s

rapidly d isso cia ted to form 0 or even 0* although the la t t e r ion i s

a s p e c ia l adsorbed sp ecies and not ordinary la t t i c e oxygen u n til metal

ion d iffu sio n has taken place to capture i t , Cu+ © i s a p-hole and i s

resp on sib le for e le ctro n ic conduction.

At low temperatures the l a t t i c e d efects w ill d iffu se n e g lig ib ly

slow ly but the electron s (or p -h oles) w ill remain mobile. The temperature

a t which io n ic d efects become mobile i s around 100°C for Cu^O su rfa ces.

I t may be judged from the temperature at which the oxide film on a metal

i s no longer f u lly p rotective and begins to grow slow ly.

In a normal, or n -ty p e, semi-conducting oxide the electro n ic

conductivity i s due to the movement w ithin the crystal of free electron s

or excess electrons which may a r is e from lo s s of oxygen or the in clu sio n

of s lig h t excess of metal atoms which then d is s o c ia te :-

M « M+ + e

Typical examples are ZnO or CdO* The excess metal atom i s probably to

be considered as in an i n t e r s t i t i a l l a t t ic e p o sitio n . The electron w ill

l- t -
normally be attached e ith e r to a normal M l a t t i c e ion from which i t can

escape f r e e ly w ith only thermal energy, an i n t e r s t i t i a l ion , or in the

 - 20 -

neighbourhood of m issing oxygen ions (vacant anion s it e s ) which w ill he

p a rtic u la r ly abundant a t the surface of the oxide c r y s ta l. For ZnO we

could have:-

an** 0* 0“ 2 d **

2n+ ✓

Zn4-1" 0= Zn*
\ '
Zn4" e se
✓

Where e * q u asi-free electron

Zn^ = i n t e r s t i t i a l zinc ion (There i s , at the moment, no

conclusive evidence to show whether th is zinc i s present as atoms or

io n s. ¥agner (42) suggests the ion Zd4"4* Gray (43)» Zn4*).

These n-type semi-conducting oxides have a tendency to lo s e oxygen

w ith the formation o f i n t e r s t i t i a l metal io n s, and q u asi-free electron s

which a c t as the current ca rriers.

A th ird type o f oxide comprises the h igh ly refractory ones lik e

A1203 and MgO. These oxides are in su la to rs, being quite non-conducting

a t normal temperatures but on heating in vacuo a t 1000°C they can lo se

oxygen and become sim ila r to the n-type semi-conductors.

For the purpose of t h is in v estig a tio n the tendency, inherent in

i t s properties as a semi-conductor, of the oxide to hold oxygen as

(>2 or 0 in an adsorbed condition w ill be considered to be an important

c r ite r io n in the c a ta ly sis of

(b) Oxide film growth on m etals.

As a source of evidence for the surface structure of cxides we

 - 21 -

must consider the work on the growth o f oxides on m etals. The p ra ctica l

work has been k in e tic and calorim etric and provides a complex f i e ld of

data. B a sic a lly , however, one general mechanism i s s u ffic ie n t to explain

the r e s u lts . Oxygen i s adsorbed at the surface of the oxide film and

being strongly electronegat iv e i t a ttr a c ts as electro n from the oxide.

An electron comes from the underlying metal to n eu tra lise the p o sitiv e

charge caused in the oxide surface and in so doing gives r is e to a cation

a t the m etal/oxide in te r fa c e . Electrons are transported eith er as

electron s or p -h o les. Metal cations must a lso be transported across the

film .

(c) C atalytic Reactions on S olid Oxides.

Through the ap p lication of these new ideas of the s o lid s ta t e ,

novel in terp reta tio n s have been advanced to explain the role of a ca ta ly st

in a heterogeneous reaction . Although a fundamental understanding has

not y et been completely attain ed ; there i s l i t t l e doubt that any in te r ­

p retation of the a c t iv it y of an oxide ca ta ly st in some reactions must

take clo se account of i t s properties as a d efect s o lid .

An extensive contribution in the f ie ld of g a s /s o lid reactions has

been made by Gamer, Stone, Gray and co-workers at B r is to l. They have

succeeded in e sta b lish in g a correlation between c a ta ly tic a c tiv ity and

e le c tr o n ic structure of certain metal oxides for the C0/02 reaction (44)

and the decomposition of 2^0 (43)* the CO/O^ reaction Gamer, Gray

and Stone (46, 47) proposed a reaction mechanism fo r the more reactive

c a ta ly s ts , e s p e c ia lly Cu20. They regarded the chemisorption of oxygen on

cuprous oxide as involving the process;-

02 + 2e = (f + <f (l)
 - 22 -

Wagner and Hauffe, (4S) had already suggested the i n i t i a l s te p i-

N20 + e = ( f (a d s.) + U2 (2)

to explain the c a ta ly sis of HgO by HiO a t 500 - 600°C. These workers

observed that the electro n ic conductivity o f the n ick el oxide was

enhanced during c a ta ly s is . Since i t was known that e le c tr ic a l conductivity

of these p-type oxides was due to an electron d eficien cy , i t was reasonable

to assume that i t would be increased during the occurrence of reaction ( l ) .

Gray (49) and Gray and Savage (50) used conductivity measurements to fo llo w

the adsorption and desorption of oxygen by the oxide which, on the

mechanism indicated by equation ( l ) , would a lso be accompanied by changes

in the number of electron defects in the oxide. The second stage in the

c a ta ly sis o f HgO decomposition begun as in equation ( 2 ), is

20 " (a d s.) » 02 + 2e (3 )

or 0“ (a d s.) + JffgO = 02 + U2 + e (4 )

and since Gray and Savage showed that the desorption of oxygen from

Cu20 was a very slow process i t may be inferred that reactions ( 3 ) and

(4 ) are probably rate co n tro llin g. This point has been substantiated by

Hauffe, Glang and Engell ( 5 1 ) who studied that decomposition of 1^0 on

a v a r ie ty of oxides including NiO. They found that the c a ta ly tic a c t iv it y

o f the oxide was improved by the addition of small traces of Li20, which

increased the p-type character, and thus would accelerate the desorption

rate through the creation of p o sitiv e holes able to consume electro n s.

Gamer, Stone and T iley (44) and D ell and Stone ( 5 2 ) examined the

C0/02 reaction on Cu20 and UiO film s present as a layer on the m etal.

They concluded that an important part o f the scheme was the reaction of

adsorbed oxygen and adsorbed carbon monoxide to give an adsorbed

 - 23 ~

-CO” complex.

CO+ 2 0 “ = CO* (5 )

This was rev ersib le below 100°C since the CO could be removed by

evacuation, but above 100°C i t reacted to C02* In the presence of

excess CO the complex r e a c tss-

C 0 ^ (a d s.) + CO - 2C02 (g) + 2e (6 )

The slow est step in the CO/Og c a ta ly sis according to these

in v e stig a to r s was the adsorption of oxygen. The v eiy slow desorption

o f oxygen was not involved. This would explain why C^O was a ctiv e at

lower temperatures (O-10Q°C) than for the N20 decomposition (200-250°C)

which can only proceed by the direct desorption of oxygen.

The c a ta ly tic oxidation of CO on KiO has a lso been studied by

Parravano (53) who suggested that at low temperatures (l00-180°C) the

rate co n tro llin g process was the adsorption of oxygen, w h ilst between

180-240°C the slow step was the removal of surface oxygen by carbon

monoxide to give C02. In addition, the system KiO + foreign ions

was in v estig a te d ( 5 4 ) in order to ascertain how the c a ta ly tic a c t iv it y

would be a ffe cted by modifying the oxide electro n ic type. In the low

temperature range a l l the ca ta ly sts gave the same value for the

a c tiv a tio n energy, but at the higher temperatures i t waa found that
monovalent cations increased w h ilst tr iv a le n t decreased the a ctiv a tio n

energy of the reaction. These findings are in apparent contradiction

to those of Schwab and Block (55) report that an increase in p-type

character lowers the a ctiv a tio n energy in the temperature range

250-400°C to 12K.cals/mol, w h ilst a decrease of p-type character by

 - 24 **

3-f
adding Cr , raised the a ctiv a tio n energy from 14 to 19 K. cals/m ol.

Stone (56) has tr ie d to recon cile these findings by suggesting

that the rate co n trollin g step i s , in each case, at a d ifferen t stage

in the reaction cy cle. However, i t may be possib le that the lack of

c a ta ly st s p e c if ic it y found by Parravano at low temperatures i s due to

the reaction being p artly subjected to transport con trol, which disappears

as the temperature i s raised. In addition, as has been pointed out by

Yerwey (57 )» when preparing an impurity type semi-conductor the foreign

ion should be about the same siz e as the ion fo r which i t i s su b stitu ted .

This w ill preserve the origin al la t t ic e without d isto rtio n or d islo c a tio n .

By adding Cr3+ and Li+ as trace im purities, Schwab and Block would not

appreciably a lte r the NiOl a t t i c e . On the other hand, Parravano used

_i_ .i- i -
Ig as an additive which has twice the ion ic radius of Hi , and th is

may have caused la t t ic e co lla p se, resu ltin g in a non-homogeneous structure

saturated with d efects. I f th is were so, then the ca ta ly sts used in the

two se r ie s of experiments are not s t r ic t l y comparable.

This impurity addition method of preparing ca ta ly sts of varying

defect properties has a lso been applied to the decomposition of H^O by

Wagner ( 5 8 ) and to the Hydrogen/Deuterium exchange reaction by Molinari

and Parravano (59) and Heckelsberg et a l . (60). In these cases

v a ria tio n s in quasi-free electron concentration in ZnO were accomplished

1 7 1. X _L

by su ita b le additions o f Li , A r or Ga • Measurements made by

Phrravano ( 6 l) on the reduction of n ick el oxide by hydrogen, have shown a

greater a ctiv a tio n energy with increasing p-type character of the oxide

by the addition of small amounts of Li20, Ag^O e tc . By the in clu sion of

th ese im purities i t i s possible to obtain a semi-conductor in which some

 - 25 -

o f the metal ions have been brought in to a higher valency s ta t e , p-type,

or one in which they have a tendency to pass in to a lower valency s ta te ,

n -ty p e.

F in a lly , the recent work of Gray and Darby (62) on the reaction

should be mentioned, as i t i s of in ter est from the mechanistic point of

view. The reaction was studied in the temperature range -21°C to 137°C

on an evaporated nick el film , surface oxidised. An. a ctiv a tio n energy

of 3*4 K .cals/m ol was determined and the rate was dependant on the square
11
root of the oxygen concentration. The a v a ila b ility of Hi ions was

suggested as being the most important rate determining factor; the

adsorption mechanism being suggested as :-

02 + 2 Hi44"^ 20“ + 2 Ni2 + ® (a)

20“ + 2 Hi44"^ 20= 4- 2 Hi24 ® (b)

and fo r the desorption the reverse process was held to occur.

(d) Physical properties of oxide c a ta ly sts .

I t i s w ell known that oxide c a ta ly sis i s very susceptible to the

conditions of pretreatment of the oxide. Among the physical properties

which have been used to characterise a given preparation are s p e c ific

surface area, e le c tr ic a l conductivity and Hall e f f e c t , crystal structure

and magnetic s u s c e p tib ility .

The la tt e r property seemed p articu larly relevant to many o f our

oxides which were based on tra n sition elements with several p ossib le

v a le n c ie s.
( i) [Magnetic s u s c e p tib ility and c a ta ly s is .

H uttig ( 6 5 , 66) studied the decomposition of methyl alcohol on the

 - 26 -

ZnO/Cr^O^ ca ta ly st system* He found that the c a ta ly tic a c t iv it y of the

mixture varied with the temperature of heat treatment, and these variation s

were c lo se ly accompanied by changes in the magnetic s u s c e p tib ility of the

mixtures* For the equi-molar mixture the a c t iv it y reached a sharp

maximum a t 400°C, a steep r is e in s u s c e p tib ility with in cip ien t

ferromognetism being observed a lso at th is temperature. These r e su lts

were critij&cised by Selwood ( 6 7 ) who suggested that paramagnetic im purities

may have been present in the mixtures and, to make an adequate d is tin c tio n ,

the magnetic s u sc e p tib ility measurements should have been determined at

d iffe r e n t temperatures.

Selwood (6 7 , 68) has reviewed the various techniques availab le for

making s u s c e p tib ility measurements, and has stressed the importance of

making determinations a t two or more f ie ld strengths, since although

paramagnetism i s independant of f ie ld , minute traces of ferromagnetic

im purities can be detected in th is manner. A graph of s u s c e p tib ility

against' reciprocal f ie ld strength w ill y ie ld the true s u s c e p tib ility of

the sample by extrapolating to in f in it e f ie ld . In addition, i t has been

pointed out that paramagnetism frequently varies in versely with the

absolute temperature, and by carrying out determinations at a se r ie s of

temperatures the presence of paramagnetic im purities in a given sample

can often be e a s ily detected.

This s u s c e p tib ility isotherm method has been applied by Selwood and

co-workers to study the magnetic properties of supported oxides of

manganese ( 6 9 ), and to the copper/chromium oxide spinel system ( 7 0 ).

The onset of spin el formation found at 1000°C for CuO. C r ^ has been
 ~ 27 -

disputed by Stroupe (7 l) who reports X-ray measurements in favour of

h is contention. It i s possible that Selwood!s a ctiv e ca ta ly st was a

s o lid solu tion of the two oxides, or merely C r ^ supported on CuO, but

from the point of view of ca ta lyst a c tiv ity i t should be stressed that

the type of structure i s unimportant, rather that stru ctu ral changes can

be related to c a ta ly tic a c tiv ity ,

11
Huttig and his collaborators in vestigated the c a ta ly tic a c t iv it ie s

o f mixed oxides of the composition XO/Y^O^, and related th is to various

physical properties, in addition to s u s c e p tib ility , such as crystal

structure and powder density e tc , ( 7 2 , 73)* Spinel formation was found

to occur around 800°C fo r most of the systems in vestigated and coincided

with lo s s of ca ta ly tic a c t iv it y . I t was concluded that although the

fin a l sp in el sta te was not so e ffe c tiv e in c a ta ly s is , the various stages

in the tra n sitio n from a mechanical mixture to th is structure were of

primary importance.

( i i ) (b) Surface area and c a ta ly s is .

The major contribution in th is f ie ld came from the c la s s ic a l work

of Brunauer, Bnmett and T eller (74-), and Emmett and Brunauer ( 7 5 ) on

m ultilayer physical adsorption. The method used by these authors was

su ccessfu l in measuring areas of not le s s than one square metre, gen erally

by nitrogen adsorption at liq u id a ir temperatures. Later, by using

ethylene a t - 195°C as the adsorbate, Duncan ( 7 6 ) and Wooten and Callaway -

Brown (77) were able to extend the method to measure areas in the region

o f 100 sq. cms. This treatment, althou^i not y ield in g absolute values

o f surface area, i s the best at present a v a ila b le. I t su ffers from the

d efect that Then the pores in a so lid are s u f f ic ie n t ly close to molecular

 28 -

dim ensions, the smaller c a p illa r ie s cannot he reached by the adsorbate

m olecules, and i s thus not applicable to surfaces having a roughness fa c to r

c lo se to unity*

Since these methods are based on physical adsorption they o ffe r no

means of determining the number of “a ctiv e s i t e s ” on a c a ta ly st surface.

Nor indeed are they any in d ication of the v a riety of adsorption s it e s which

may be availab le on a given catalyst*

Early conceptions, which regarded a c a ta ly st as having a fix e d surface

w ith a v a r iety of sp e c ia lly activ e "free valencies" (Taylor, JQ) due to

the p o sitio n of the ions being on edges or peaks of surface c r y s t a llit e s ,

have to be modified in the lig h t of the work o f Yolkenshtein (79) > 'who

emphasised the defect nature of a surface and drew particular a tten tio n to

the equilibrium which could be assumed to e x is t between various types o f

io n ic and electro n ic defects* In such cases a chemisorbed molecule i s an

impurity in the surface and gives r is e to s h ifts in the surface equilibrium .

Thus a surface with chemisorbed molecules attached has d ifferen t properties

from the same s o lid before the attachment*

Boudart (80) has extended th is view by concluding that a ctiv e centres

in a surface w ill be mobile and w ill be in a process of continual creation

and destruction. Naturally enough io n ic d efects and eq u ilib ria in v o lv in g

them w ill not be mobile a t the lower tenperatures and w ill o n l y be s lig h t ly

so fo r most oxides at the higher temperatures which we used in our

in v estig a tio n s (180°C)•

( 3) The Present Approach

Hart and McEadyen (81) in vestigated , in considerable d e ta il, the vapour phase

decomposition of hydrogen peroxide on cuprous oxide film s supported on copper

 - 29 -

m etal. This was used in the form of a gauze held normal to a flow ing stream

of gas containing HgOg vapour. They a lso made a few measurements on other

ox id ised metals av a ila b le as gauzes e .g . Ee, Ag and P t. They worked with

j a r t ia l pressures of in the region of 1 m . and in a temperature range

of 4G - 100°C. The a c tiv a tio n energy of the reaction on the CUgO covered

gauzes was determined as being about 15K. cals/m ol and the k in e tic s were f i r s t

order with respect to concentration.

The temperature c o e ffic ie n t of the reaction rate was found to vary with

the "previous history" of the c a ta ly st. This e f fe c t has been discovered by

many in v e stig a to r s in the f ie ld of heterogeneous c a ta ly s is , including Constable

who observed v a ria tio n s in a ctiv a tio n energy between 18-24K cals/m ol. for the

dehydrogenation o f eth yl alcohol a lso on cuprous oxide (82).

The exten sive r e su lts of Hart and McEadyen on the H^O^/Cu^Ocatalysed

rea ctio n , le d them to conclude that the property of the oxide which determined

i t s c a ta ly tic e ffic ie n c y was rela ted to i t s a b ilit y to adsorb oxygen, somewhat

analogous to the findings of D e ll, Stone and T iley (45) on the l^O decomposition

and the c a ta ly sis of the CO/O^ reaction on cuprous oxide.

The i n i t i a l aim of the present work was to extend the work o f Hart and

McFadyen to a wide range of metal oxides of d ifferen t types, so as to enable

a co rrela tio n to be made between the properties of an oxide and i t s c a ta ly tic

e ffic ie n c y . I t was r e a lise d that th is would involve the development of a

method su ita b le fo r the use of bulk oxides rather than oxidised metal gauzes,

so that ca r efu lly prepared pure oxides of known bulk properties could be used.

However, an important object was to maintain a lin k with the e a r lie r work,

and therefore in some cases a carefu l study was made of the c a ta ly s is

achieved by d iffer en t forms of a given oxide - oxidised metal s li p , f u lly

 - 30 -

o x id ised flash ed metal film on g la ss , and f in a lly , compressed bulk oxide

s lip .

A rising out of th is i t was e sse n tia l to give a tten tio n to the methods

of preparation, p a rticu la rly the thermal pre-treatm ent, of oxides and i t s

relevance to c a ta ly tic e ffic ie n c y .

The second aim was to study how to modify the c a ta ly s is by a given oxide,

through the addition of foreign oxides in eith e r trace or equi-molar amounts.

The object of th is was to lin k c a ta ly sis with s o lid oxide properties which

are known to be a ffec ted by such in clu sio n s. For example, the in ten tio n was

to study NiO with additions o f Irf^O f ° riner i s hnown to in crease,

and the la t t e r to decrease, the tendency of the oxide to hold adsorbed ox

a t i t s surface.

For the equimolar mixture stu d ies, CuOand Mn^O^ were chosen with equi-

molar amounts of a range of other oxides, e .g . ZnO, NiO, ^ 2 ^3 *

The f u l l id e n tific a tio n of what has been referred to above as "efficiency"

of a ca ta ly st i s a d if f ic u lt matter. The in ten tion in th is work was to

measure the fra ctio n of H^O^. decomposed under constant flow conditions and

to support th is with measurements of temperature c o e ffic ie n ts . In addition,

as a p o ssib le source of comparative data on d ifferen t oxide c a ta ly sts, the

surface areas of some of the s lip s wouldbe measured by a B.E.T. gas

adsorption method.
Paramagnetic measurements were also planned as a possible source of

information about the nature of the metal ions in mixed oxides, and X-ray

stru ctu ra l an alysis to give an in dication of compound formation in such

mixtures during heat treatment.

F in a lly , as an in d ication of the rela tio n between the flow stu d ies and
 - 31 -

the previous work using s t a t ic methods, i t was intended to carry out a b r ie f

study in a s t a t ic apparatus o f one or more su itab le c a ta ly s ts .

EXPHjRIMEIflEAIi

(l) Apparatus and Methods

(a ) Flow system# General d escrip tion .

A flow method fo r examination o f the ca ta ly sts was chosen in order

to d isc lo se changes of c a ta ly tic e ffic ie n c y with tim e, and to be able

to trea t these changes separately from the normal k in e tic s which assume

a constant c a ta ly st. In addition, p ractical advantage of such a system

was that i t would be p o ssib le, by su itab le choice of flow arrangements,

to elim inate gas phase d iffu sio n of the hydrogen peroxide to the ca ta ly st

surface as a p o ssib le rate co n trollin g fa cto r.

A method in which the vapour of pure or very concentrated hydrogen

peroxide was pumped over the ca ta ly st at reduced pressure presented the

main d if f ic u lt y that the ca ta ly st temperature during the decomposition

process would r is e w ell above that of the vapour stream. This could be

avoided by using vapour from solu tion s of low concentration hydrogen

peroxide, which would however, involve having always a large amount o f

water vapour present as a diluent in the gas stream. I t seemed preferable

to work with an in ert carrier gas such as nitrogen, at atmospheric

pressure. The p a rtia l pressure of the could be kept low enough

to lim it the temperature r ise on the ca ta ly st to 1 or 2 degrees.

Working at atmospheric pressure would permit easy p recision regulation

of the flow rate before the gas was passed through a saturator containing

hydrogen peroxide solu tion at a controlled temperature.

The apparatus eventually used i s shown in Fig. 1 ., up to and

including the pre-heater, and was a modified version of that used by

 - 33 -

Hart and McFadyen (81 )*

"White-Spot" nitrogen ( ’>99*99^ nitrogen, B ritish Oxygen Company),

was drawn from a cylinder equipped with pressure gauzes and needle

valves to adjust the rate of flow* This passed to a ten l i t r e stock

b o ttle ( r ) which acted as a storage v e sse l fo r the nitrogen, and a lso

served to reduce to a minimum slig h t pressure flu ctu ation s caused by the

manostat ( m) and the gas absorbers, not shown in diagram. Fine control

o f flow was obtained by ca refu lly manipulating the cylinder needle valves

to give a pressure of 125 ± 0 * 5 mms. Hg above atmospheric in the system,

which was kept at th is constant value by the manostat.

The manostat i s shown in Fig. 2. I t operated from a simple Hg

f i l l e d TJ - tube manometer with a cap illary in serted to prevent the

mercury see-sawing. With r isin g pressure in the storage b o ttle

e le c t r ic a l contact was ira.de in one arm, as in a thermostat bath regulator.

Through a rela y , th is caused a lea k -o ff to be opened. The le a k -o ff

con sisted of a ca p illa ry closed by a 100 gm. s te e l weight with s o ft rubber

attached to the underside, which pressed against the ground end of the

ca p illa ry . A solenoid, fix e d above th is weight, was able to raise i t

when energised. The relay operated on a difference of pressure of le s s

than 0.1 mm Hg and kept the pressure in the nitrogen reservoir to within

0 .2 mm o f water.
From the reservoir ( r ) the nitrogen could be drawn o ff in to any one

o f three streams as indicated in the diagram, v iz j -

(l) D iluent nitrogen to adjust the concentration of vapour entering the

reaction tube in determining the reaction order, and to "dry out” the

apparatus, e sp ec ia lly the reaction-tube, a fte r cleaning.

 MANOSTAT ASSEMBLY
z
u
Jn*jT

2.
FIG
nrnr

o
o O
CL Qf
hu
O'
 - 34 -

(2 ) The saturator stream in which nitrogen was bubbled through liq u id

hydrogen peroxide held at a constant temperature,

(3) The water vapour stream in which the nitrogen was passed through a

saturated solu tion of magnesium chloride held a t a constant temperature

which enabled a known amount of water vapour to be added to the to ta l

gas flow .

Each stream could be operated independently, or in conjunction

with one or both of the others. This was accomplished by su itab le

manipulation of stopcocks and screw c lip s con trollin g the gas flow from

the nitrogen reservoir.

The individual components in the flow system are d eta iled as

follow s:®

(iL^), (Ag), (A^). One l i t r e B34 round-bottom fla sk s used as secondary

nitrogen storage v e sse ls to reduce pressure flu ctu ation s to an absolute

minimum. They were connected to the reservoir via ca p illa ry tubing

which helped to in su la te the flow systems from the large storage v e s s e l.

(Y ,), (V2 ) , (Y ) . Eine adjustment n eed le-valves.

These gave fin e control of the flow rate having a s e n s it iv it y of 2 mls/min.

per degree of d ia l revolution at normal temperature and pressure.

(Y^) was la te r replaced by an Edwards needle-valve to give an even fin e r

degree of control, namely 0 .5 mls/min. per degree of d ia l revolu tion ,

which was required in adjusting the flow from the water saturator in

some of the la te r experiments.

(3^ ) , (f2 ) , (E^). D ifferen tia l pressure type flow-meters as described by

Farkas and M elville ( 8 5 ). The meters were constructed in pyrex g la ss

w ith limbs 30 cm. long, internal diameter = 0.5 cm* The flow ca p illa ry
was connected to the e x it limb of the meter by an A7 j o in t. Each flow ­

meter was securely bolted to a baseboard and scale from which the pressure

d iffe r e n tia l was e a s ily determined. The c a p illa r ie s used were 0.6 mm.

in diameter and o f various lengths from 5~15 cm* depending on the range

of flow -rates required, and could be e a s ily interchanged by removing the

q u ick fit spring and breaking the B19 join t which was lubricated by a

vacuum grease.

The c a p illa r ie s were calibrated d ire c tly by a moving soap film

technique. This was a simple and accuiate method in which the e x it

n itrogen from the flow-meter was passed just under the surface of a

"soap” so lu tio n , con sistin g of 40$ "Teepol”, 30$ glycerol and 30$ water.

This resu lted in the formation of a thin film which moved along a g la ss

tube, one metre long by 3 cm* internal diameter, with the gas flowr. In

order to prevent the film from bursting prematurely w h ilst moving along

the ca lib ra tio n tube, considerable atten tio n was paid to the c lea n lin ess

and freedom from grease of the tube surface, which was washed with

d i s t i l l e d water and dried with acetone. The time of travel between

sele cted fix ed points was noted. By choosing suitable nitrogen pressures

a graph was drawn o f flow rate against pressure d iffe r e n tia l on the flow ­

meter water manometer. The volume of the tube between the points was

determined by f i l l i n g i t with mercury and weighing.

A ltern a tiv ely , ca lib ration was accomplished by connecting a flow­

meter in ser ie s with another of pre-determined c h a ra cteristics. By these

methods i t was p ossib le to gauge accurately to within 0.5$, rates of flow

from 25-5 0 0 mls/minute which were required in the experiments.

The gas from the flow^meters passed through a further c a p illa iy to

 - 56 -

prevent the flow-meter reg isterin g minute pressure flu ctu ation s from

the absorber/bubblers*

(^2.)* (®2 ^9)• These were id en tica l splash traps which prevented water

from discharging into the apparatus in the event of a flow-meter being

a ccid en tly overloaded.

(^i)> C^)* Id en tical cleaning towers, 60 cms. long by 4 cms diameter,

to remove a l l dust p a rtic le s or water vapour carried over in the nitrogen

stream. The packing consisted of equal layers of f il t e r in g and drying

m aterials arranged as fo llo w ss-

(a) g la ss wool, (b) calcium chloride, (c) g la ss wool,

(d) calcium chloride, (e) g la ss wool, ( f ) magnesium perchlorate and

(g) g la ss wool.

(3)^) and (D^) were connected in se r ie s to shorter towers, 30 cm. x 4 cm.

i.d o , containing cotton wool and a 04 sin tered glass d isc which was used

to prevent any cotton wool fragments from being blown over in the in ert

gas stream.

(D^) A cleaning tower f i l l e d as for (D^) and (l^ ) but of smaller dimensions

being 50 cm. long and 2 cm. in ternal diameter.

(S1 ) The hydrogen peroxide saturator, con sistin g of a pyrex tube 15 cm.

long x 4 cm. i d . , the open end terminating in a B34 cone. The nitrogen

supply entered through a 5 mm. bore tube which passed through a B34

socket forming the head of the saturator; and was discharged in to pure

or very concentrated hydrogen peroxide solu tion through a G1 s in te r ,

which e f fe c t iv e ly distributed the in ert gas as tin y bubbles through the

body of the peroxide. The sin ter was placed about 2 cm. from the bottom

of the saturator, which contained 20 mis. HgOg*

 - 37 -

The saturator was immersed in a 3 l i t r e , s tir r e d , thermostat hath.

The temperature control was operated by an electro n ic relay which enabled

the temperature to be held to + 0.01°C. The bath was maintained a t a

temperature below that of the room (l2-18°C, as chosen) to prevent

condensation of the saturated vapour elsewhere in the system. The bath

was continuously fed from the cold mains water supply and the volume kept

steady by a constant le v e l device. I f the mains water temperature was

too high, iced water from a gravity feed was added a t rated o f 50 - H 0

mls/minute depending on the temperature required.

(S^) The water saturator, a pyrex tube 20 cm. long by 24 mm. i . d . , the

open end being a B24 cone; which contained a saturated so lu tio n of

magnesium chloride. The nitrogen in stream ( 3 ) entered the B24 sock et,

which formed the head of the saturator, through a 5 mm. bore tube which

tapered to a 2 mm. o r ific e at a distance of 2 cm. from the magnesium

chloride c r y s ta ls .

The saturator was immersed in a water bath, sim ilar to that used

fo r the H2 O2 saturator described under ( S ^ . Care was taken not to

exceed the saturated water vapour pressure under the p revailing atmospheric

conditions when adjusting th is temperature, - which was n ot, however,

n e c e ssa r ily below room.

A ll ground jo in ts in the apparatus up to th is point, with the

exception of the B34 jo in t in the peroxide saturator, were lu b ricated

w ith eith e r "Silicone” or "Apiezon M" grease. The jo in ts follow in g,

i . e . those which came in to contact with the hydrogen peroxide vapour,

were not greased; in order to prevent contamination and p ossib le

 TO
AbSOR&ER

FIG. 3-
FLOW REACTOR
(METAL FILMS)
 - 38 -

decomposition of the peroxide*

Streams ( l ) , ( 2 ) and ( 3 ) were now passed in to the common manifold

as shown in Pig* 1* Stopcocks were provided in streams ( l ) and ( 3 )*

Prom th is manifo ld the mixed gases passed in a common stream to the

preheater v ia a BIO jo in t.

(p) The sp ira l preheater which was a modified C3/12 q u ick fit condenser w ith

a c o il length of two feet* I t was used to Bring the temperature o f the

gases up to that ofthe c a ta ly st, w h ilst serving a lso to mix the gases

Before entering the reaction tuBe*

Two d ifferen t types of reaction tuBes were used in the experiments*

One was u t ilis e d in the examination of evaporated metal film s and the other

fo r metal oxide, and mixed oxide compressed s lip s .

The flo w reactor fo r metal film s.

This consisted of an 18 mm. in ternal diameter pyrex tuBe, Bent to

a righ t angle at the vapour entrance, which was coupled to the preheater

v ia a B14 socket (Pig*3*) The tuBe, 30 cm long, was in ter n a lly sealed

in to a B24 socket 4 cm Below the Bend. This socket formed the head of

the reactor and connected to the ca ta ly st chamber By the B24 jo in t.

(The metal oxide film to Be examined having previously Been deposited on

the w all of the chamber as a circu lar Band 5 cm. Broad. The Bottom edge

of which was a distance o f 10 cm. from the Base of the ca ta ly st chamber.)

The gas stream from the preheater passed in to the ca ta ly st chamber

through the pyrex tube and came in to close contact with the c a ta ly st while

flow ing through the 2 mm. annular gap. The efflu e n t gas from the reaction

was then Bubbled in to the gas absorbers through the two e x it tubes sealed

in to the B24 head.

 u
ui

01
a%
U.

REACTION TUBE
(com pressed o x id e s )
 - 39 -

The unit was immersed in a stir r e d o i l bath to ju st below the 3324

joint* The bath was heated by a 250 watt immersion heater and the

temperature controlled by a mercury or toluene regulator connected to the

mains through a "Sunvic” relay.

Reaction tube fo r oxides s l i p s .

I t was o r ig in a lly intended to examine the oxide ca ta ly sts impregnated

on a s i l i c a mesh support. Experiments were made in th is d irection by

b o ilin g d iscs o f the support with solu tion s of certain metal n itr a te s

follow ed by heating to the oxide. Cbi examination, i t was found that th is

method resu lted in irregular coverage o f the support by the oxide and i t

was v ir tu a lly impossible to prepare reproducible surfaces.

To surmount th is d iff ic u lty , and to obtain ca ta ly sts of sim ilar

surface areas, i t was decided to use the oxides in some form o f compressed

p la te or s li p . The reaction tube was designed accordingly, and the fin a l

u n it i s shown in Pig. 4.

I t consisted of a 2 cm. bore pyrex tube 17 cm. long, tapering over

5 cm. to a rectangular cro ss-sectio n , 1 cm x 0.25 cm x 10 cm. long. This

was blown out at the e x it to foim a 3324 socket. The ca ta ly st of dimensions

2 .5 cm. x 1 cm. x 0.1 cm. being situ ated 1-2 cm along the rectangular tube.

A vapour jacket enclosed the tube and constant temperatures were obtained

by reflu x in g the vapour of a r e d is t ille d liq u id , b o ilin g at atmospheric

pressure, in to the jacket.

The absorber.
The gas from the flow reactor was then passed in to a "bubbler" which

acted as an absorption unit for the peroxide vapour. These were constructed

in pyrex g la ss, 15 cm. long by 2.5 cm. diameter. The o r ific e of the
 - 40 -

bubbler being 0*5 cm. bore, and i t s feed-end connected w ith an A10 socket

to the reaction tube via an A10/B34 coupling* The e x it tube of the

bubblers was open to the atmosphere, thus allow ing the nitrogen ca rrier

gas to escape. I t was found that only one bubbler, containing 20 mis.

d i s t i l l e d water, was required to absorb a l l the e fflu e n t peroxide. When

two or more were added in series and the contents analysed by t it r a t io n

w ith 0.01N KMn.0^, no peroxide was detected in any but the f i r s t bubbler.

Temperature con trol.

In experiments with the oxidised metal film c a ta ly s ts , the reactor

tube was fix e d in p o sition in a therm ostatically con trolled o il-b a th ,

preheated to the temperature required. Dry nitrogen was then passed

through the system which was allowed to stand for approximately 5 minutes

to a tta in temperature equilibrium before commencing a run. The temperature

o f the o il-b a th was controlled to ± 0.01°C a t temperatures up to 100°C,

and over th is control was to + 0.02°C. Unfortunately, th is system did

not allow temperature changes to be made as quickly as desired in

determining the temperature c o e ffic ie n t. There being a time la g of 5

minutes required to give a steady temperature 20C° hi^aer when making

a change.
The temperature control with the s lip - reactor was to + 0.01°C in

the range 38 - 183°C. This was attained by reflu x in g the middle fr a c tio n

o f the follow ing r e d is t ille d e b u llio s ta tic liq u id s from a 500 ml B24

round bottom fla sk attached to the vapour - jacket; the b o ilin g p o in ts,

to the nearest 0*5°0; &re given in brackets

( l) Ethyl Bromide (38°C) (2) Chloroform (60°C)

(3) Carbon Tetrachloride ( l 6 . 5 ° C ) (4) Benzene (80°C)

 - 41 -

(5) Water (lOO°C) (6) Xylene (136°C).

(7) A niline (l83°C ).

In th is system, temperature changes were accomplished in le s s

than 60 seconds by quickly changing from a fla sk of one liq u id to

another preheated to near the temperature required. The temperature

of the vapour was ascertained from a B14 q u ick fit thermometer enclosed

in the vapour jacket. This was recorded as the temperature of the run

which could d iffe r s lig h tly from the above b o ilin g p oin ts, due to

contamination o f the liq u id s from the drainage of the vapour jack et.

I t was necessary frequently to r e d i s t i l l the liq u id s to r id them of

th is contamination.

With the pre-heater containing methyl alch ol b o ilin g at 64°C,

supplying a background heat to the peroxide vapour, i t was shown by the

use of therm istors, that the peroxide stream attained the temperature

o f the vapour jacket at a distance of 4 diameters along the reaction

tube, i . e . 5-6 cm. from the entrance with temperatures up to 100°C and

s lig h t ly more for temperatures of the higher order. Even a t 183°C the

peroxide vapour was w ithin 1C° of the ca ta ly st temperature by the time

contact was made.

Cleaning glassware.
In a l l work concerning the flow system extreme care with cleaning-

had to be taken in a l l experiments to obtain rep rod u cib ility of blanks.

Conventional methods of cleaning proved to be u n satisfactory e .g . chromic

a cid could not be used, as i t was shown that g la ss so treated decomposed

H2 °2 vapour, possibly by the formation of an in v is ib le chromic oxide skin

 - 42 -

on the w alls of the reaction v e s s e l, giving decompositions o f up to 5%

at temperaturesc.l80°C. Sim ilarly sulphuric acid was not s u ita b le ,

presumably because of metal ion im purities in th is acid which again cause

tube contamination and decomposition of the vapour. Cleaning with cau stic

a lk a lis was a lso attempted, without success, p ossib ly due to an etch ing

a tta ch on the pyrex g la ss surface exposing impurity occlusions and thus

making i t le s s stab le to hydrogen peroxide.

An e ffo r t to s t a b ilis e the reaction tube surface, a fte r cleaning

w ith the above mixtures, was made by trea tin g i t with 25 % boric acid

so lu tio n . This gave i n i t i a l l y a f a ir ly in a ctiv e coating but in about one

hour a t temperatures > 100°C. decompositions of 2-4% o f hydrogen peroxide

vapour were recorded. Treatment with sulphuric acid and ammonium per­

sulphate, which proceeds through the production of was f a ir ly u sefu l

but had to be discarded as sulphuric acid of s u f f ic ie n t ly high purity

could not be obtained.

The cleaning treatment eventually adopted con sisted of an ethanol

rin se follow ed by the addition of concentrated n it r ic acid , a n a ly tica l

grade. The nitrous fumes produced were allowed to permeate the system,

which was la te r f i l l e d with d ilu te HNO^ and allowed to stand for several

hours. This was followed by immersion and rin sin g w ith d is t il le d water,

and the g la ss was f in a lly dried at 100 - 150°C. I f necessary the cycle

was repeated several tim es.

A fter th is treatment no decomposition was observed on the g lass at

temperatures below 136°C and any which occurred between th is temperature

and 183°C was le s s than 3.5% involved a very small correction.

Thermal decomposition of H^O,, vapour was not an important factor sin ce

 - 43 -

t h is does not take place u n til temperatures of around 600°C are a tta in ed

(McLane, 29).

Blank Determinations.

The system was f i r s t purged with nitrogen w ith the reactor a t the

required temperature, in order to remove traces of moisture s t i l l p o ssib ly

occluded in the reaction-tube. This was continued for 30 minutes. The

dry nitrogen stream was then stopped and replaced by that from the peroxide

satu rator.

^he flowmeter was quickly adjusted to give the required flow -rate and

the apparatus allowed 15 minutes to s e t t le down. The gas absorbers were

changed every 5 minutes and the peroxide content of the gas stream

determined by a titr a tio n with 0.01IT KMnO^. Concordant r e s u lts were

u su a lly obtained with successive blanks, but i f th is was not the case,

the apparatus and e sp e c ia lly the reaction -tu b e, was recleaned. The flow -

rate was checked when necessary at points between the flow-meter and gas

absorber, by the use o f a s e n sitiv e rotameter which te ste d fo r leaks in

the system due to fa u lty jo in ts . Only very rarely was such a leak

detected.

A specimen se t of r e su lts and calcu lation s for blank runs was as

follow s

Temperature of Saturator

Concentration of ^ 0 2 in Saturator 99*90$

Flow Bate 300 mls/min.

Sample Time 5 mins.

Table 1 shows the experimental fig u res obtained with these

co n d itio n s:-
 - 44 -

TABLE 1

Time (mins) Reaction Tube T itre

Temperature ( C .). MLS.0 .OUT KMhO^

0 100 0

5 100 15*92

10 100 15*95

15 100 15*94

20 100 15.95

60 100 15.95

180 100 15*95

200 184 15.90

205 184 15.91

210 184 15.90

240 184 15.90

300 184 15.90

At the steady sta te at 10Q°C.

. _. . 15.95 x 17.01
Wt. o f H202 in vapourin 5 minutes = 100 x 1000
o -5
At the steady sta te a t 183C. = 2.72 x 10. ^gms.
. _. . 15o90 x 17.01
Wt. of E2 02 in vapour in 5 minutes - 100 x 1000

= 2 .7 1 x 10"^gm.

Comparison with th eo r etic a l.

P a r tia l Pressure o f H202 over 99*90$ so lu tio n .

» O.9 6 mm. (Kavanagh, 84)
0.96 x 1.5 x 54.02
Wt. of H202 in 1.5 litr e s - j 6q x #082 x 288
= 2.72 x IQ-?
 to
O
u-

o
o
CVj

ifl
z
o
£
£
H
Z UJ
u Id -3
O 0 u■ J
in ^ z
u 0 *
u 7
O' *
D

O'
U
a
£
u
■0“ ^
4K JJ>

t j O
U- Q ^ q_

o
o
O'
o
u
4
u
fcbSi
A L UJ c
ui if) I 6
9 >- »
a a
O' I o
0 U
Xu
ifl 0

O
i/)

o o o
rvj cu

( ^ 0 ^ WJ>1 N I O O ' <s n w ) 3 & i x i i M N V ^ 'g

 - 45 -

• • At the steady s ta t e , at 100°C, Blank. Decomposition = 0,

and j At the steady sta te , a t 184°C, Blank Decomposition

ii.g . 5 shows the deviation of the experimental blank from the

th e o r e tic a l value a t various concentrations of in the vapour phase,

and i t can be seen th at the blank decomposition was small in the temperature/

pressure region studied i . e . 38”183°C> aH.d 0.3 - 0.96 mm. Hg pressure.

The decomposition occurs only over 100°G and increases s lig h t ly with r is in g

peroxide concentration. I t was generally n e g lig ib le in comparison to the

c a ta ly tic decomposition, so that in very many cases i t could be n eglected

e n tir e ly . (These blank deviations r e fe r , o f course, to the decomposition

on pyrex g la s s ).

E ffic ie n c y of c a ta ly sts .

A fter obtaining a sa tisfa cto ry blank, the ca ta ly st was manoevred in to

the reaction tube with the aid of surgical tweezers, taking care not to

damage or break the leading edge of the s lip ; which was so placed in

p o sitio n as to give 1 mm. clearance between the ca ta ly st and the top and

bottom w alls of the reaction v e s s e l. Ihen th is had been accomplished, the

nitrogen flow was switched in to the hydrogen peroxide saturator, only a

few seconds being required to adjust the flow to the necessary value. "When

the temperature of the gas stream was te ste d during th is change no

flu ctu a tio n s were observed.

Ihen the HgOg vapour had passed over the c a ta ly st for 5 minutes, as

timed by a stop-watch, the efflu en t gas absorber was quickly changed over,

and the amount of P**®®®*^ estimated by a titr a tio n with 0.01K KMnO^.

Subtraction o f th is figure from the blank value gave the amount of vapour
decomposed* (These determinations were continued u n til the steady

decomposition value was reached* Generally, th is le v e l was attain ed

a f te r 20-25 minutes, and the measurements were continued for 60-70

minutes* I f the run was extended to 2 or 3 hours, steady changes in

a c t iv it y of about 1% were sometimes observed. However, slow changes of

t h is nature were not studied in these experiments*

I n i t i a l changes o f e ffic ie n c y *

During the f i r s t exposure of a ca ta ly st to vaPour> in itia l

f a s t changes of e ffic ie n c y were often determined by taking samples of

the efflu en t peroxide vapour every 60 seconds. Changes of absorber tube

could be made in about a second and there was n e g lig ib le disturbance of

the flow rate in the change, providing that the volume of water in the

absorbers was carefu lly pre-adjusted.

Determination of reaction order*

The order of the reaction was ascertained by a method involving

quick changes of concen^ra^^on» w h ilst keeping the flow rate constant.

Thus, w ith a flow rate o f 300 mls/min, and a hydrogen peroxide concentration
o f (h in the gas stream; several determinations were made in the manner

p reviously described. When a steady decomposition value had been reached,

the in ert gas flow through the saturator was cut to 200 mls/min.

i . e . to 2/3 Cj. and the to ta l flow brought back to 300 mls/mm. by the

a d d ition of dry nitrogen. Again, when a steady decomposition value was

reached 2/3 C.-» l / 2 Ch and sim ila rly l /2 C± ~ t l / 3 Ch. This was followed

by the reverse procedure, l/3 C^ "r l/2 Ch- ' 2/3 Ch ' Ch, by replacing the

dry nitrogen increments with peroxide saturated nitrogen from the H^O^

satu rator. Each o f these concentration changes could, with p r a c tic e , be

 - 47 -

completed in 15-20 s e c s.

Determination o f temperature c o e ffic ie n t.

The temperature c o e ffic ie n t of the reaction rate was found by making

quick temperature changes within the lim its of the decomposition. This

was accomplished by switching from one con stan t-boiling liq u id to another,

a f t e r the steady decomposition lim it bad been reached at a given

temperature. By pre-heating the change-over liq u id , the new temperature

of the c a ta ly st and reaction-tube was attained within 60 seconds, as

recorded on the top thermometer, and by a thermistor attached to the

c a ta ly st during t e s t runs at the various temperatures.

The decomposition was allowed to come to a steady value, taking

about 20-45 minutes according to the temperature, and the temperature

changed to a new value by again switching to a d ifferen t e b u llio s ta tic

liq u id . By th is method decomposition figu res were obtained, in the

temperature range 38 - 183°C, fo r each catalyst studied; thus perm itting

an evalu ation of the temperature c o e ffic ie n t of the reaction rate for

each stepw ise change.

In the case o f the metal film s i t was more d if f ic u lt to e ffe c t

immediate f a s t temperature changes. Even by employing two hot o il baths

i t took about 5 minutes to a tta in the necessary steady values. So in

t h is instance i t was not p ossible to a tta in the sudden e ffic ie n c y change

in a given temperature in ter v a l, as with the other reaction tube.

E ffe c t of oxygen and water vapour on the ra te.

In studying the e ffe c t of oxygen on the decomposition r a te, the

method employed was to change completely from nitrogen to oxygen as the

 - 48 -

ca rrier gas entering the storage v e sse l ( r ). The change from one gas

cy lin d er to the other ■was accomplished w ithin 5 seconds, and the time

required to give a steady pressure value in the flow system varied

s lig h t ly between 10-20 seconds. I t would be in correct to say that a fte r

a change from oxygen to nitrogen, the to ta l gas flow was immediately

composed of only one o f the gases; because o f the time required to

d isp la ce the e x is tin g gas in the storage v e s se l (r ).

In th is in stan ce, and w ith the water vapour ad d ition s, the runs

were a l l carried out with over 99% W w fresh ly d is t i lle d , u n sta b ilised

hydrogen peroxide*

Water vapour was added by passing dry nitrogen in to the saturated

so lu tio n o f Mg Cl^ a t a constant temperature, chosen so that the saturation

pressure at atmospheric conditions would not be exceeded. By adjusting

the flow o f the carrier gas passing in to the saturator, su itab le increments

of water vapour were added in the pressure region 1-16 mm, Hg. The t o ta l

flow was restored to 300 mls/min, by additions of a fix e d amount of

peroxide saturated gas and variable q u an tities of dry nitrogen from the

other streams.

Experiments to detect transport con trol.

The flow ra te, dimensions of the reactor, temperature range, and the

arrangement of the ca ta ly st were chosen so that surface processes and not

gas phase d iffu sio n of H202 to the surface con trolled the ra te.

I f gas phase d iffu sio n was ra te-co n tro llin g a f ir s t order process

would be found re la tin g rate of c a ta ly sis to gas phase concentration of

H202 a t a given flow ra te. A ll e ffe c tiv e ca ta ly sts of the same area would

be expected to give the same rate of decomposition. This was, of course,

 <i>
o u d
r~ cO o u
in

ul
5 a
i* o
d?
§ ul
-j N
a iP
Ul J
x
h- 2
d) £ 5
0
O
Ul O a
ul o
Cf ul
X u i—
r
3
a
ui 4
X
u ° 9 S
CjC
t
vl ul r 5
a
15 c* n r
o
3
eft if)
UI «
u
O ul a r >
CO
Ul OO*
ai a & r
(P
% fc ui
§ j M
O
w
a a o O

O O
<3
500

0, 01 001 ASOS-2
IK
 - 49 -

fa r from "being the case. At quite low temperatures the best c a ta ly sts

could achieve ■> 98$? decomposition while w ith the l e s s a ctiv e ones a much

higher temperature was necessary before decomposition could be observed

at a ll. This seemed to provide strong evidence that so long as

decompositions were not too close to 100% we could ignore gas phase

d iffu sio n as a con trollin g fa c to r.

Experiments to demonstrate the independence of k in e tic s on flow

v e lo c ity were made using gas flows in the region of 400-100 mls/min. at

various temperatures. P ig. 6 illu s t r a te s the r e su lts with c a ta ly sts of

widely d ifferen t type, a surface oxidised n ick el metal s lip and compressed

ZhO/Mt^O^. I t i s shown that fo r flow rates between 150 and 350 mls/min

the rate of decomposition, at a fix ed HgO^ concentration, i s independent

of flow r a te . As would be expected, above the higher flow lim it the

fr a ctio n of peroxide decomposed fo r a given time of stay a t the ca ta ly st

f a l l s s lig h t ly , presumably because average ^O^ molecules make fewer

c o llis io n s with the catalyst per second than they do at sm aller flow s, -

or put in another way molecules remote from the active surface when the

gas reaches the ca ta ly st zone are not able to d iffu se in to contact before

the bulk flow has passed i t . At flow s slower than the lim it a higher

decomposition i s observed. This suggests that a t these bulk flows the

time of contact with the ca ta ly st i s longer than i t appears to be -

perhaps due to back d iffu sio n .

o>) Metal evaporation u n it.

The technique adopted in the production of metal film s was sim ilar

to th at used by Pinch et a l. ( 85 ) and la te r by Trapnell (86).

 - 50 -

The apparatus shown in Fig© 7 consisted of a three-stage mercury

d iffu sio n pump hacked by a high e ffic ie n c y two-stage mechanical o i l pump

through a ^2^5 ^raP* T^e d iffu sion pump led through two vapour traps

immersed in Dewar fla sk s containing liq u id oxygen to the fla sh in g un it

attached by a B19 jo in t.

The fla sh in g unit head was a B24 socket in to which two tungsten

w ire s, 1 mm. diameter, were sealed. Both wires were encased in g la ss

c a p illa ry tubing. The tubing was held r ig id ly by g la ss rods, which

served to reta in the tungsten wires in p o sitio n . One wire was 5 era-

shorter in length than the other and was placed cen tra lly in the head.

The metal filam ent to be flashed was linked to the ends of these wires

so that deposition occurred ra d ia lly from the centre, with the c a ta ly st

tube surface interposed within the lim its of the mean free path of the

evaporating systems.

Mdway between the flash in g unit head and the B19 trap jo in t, a

B14 socket was sealed to take the Pirani gauge head which was connected

to an Edwards vacuum in d icator. The current used in the evaporation was

supplied by a 'Hank of accumulators which were connected to the tungsten

term inals v ia a rheostat and ammeter.

Evaporation technique.

The apparatus was evacuated for 16-20 hours accompanied by

in term itten t baking-out, (heating with a bunsen to the point of c o lla p se ).

-5
This enabled a vacuum b etter than 10 J mm. to be maintained when depositing

the film s. In a l l cases the deposition surface was cooled by liq u id

oxygen. This procedure was intended to avoid p r e feren tia lly oriented

film s (Beeck, Smith and Wheeler ( 8 9 ) ).

TLit\iGiST~,N
F I'-A M E N T S

„ Gl a s s
CAPILLARY
S upports

YSTAl f l a s h in g
POSITION

FIG. 7.
METAL FLASHINC UNIT
 - 51 -

The film thickness of the metals deposited were in the region 5”10yu

as calcu lated by d issolvin g the film in acid and analysing fo r ITi by

conventional methods. Deposition was completed in 15-20 s e c s, w ith a D.C.

current o f 5—7 amps. Films o f n ick el deposited in th is fashion were

transparent but on admission of a ir became black due to the formation of

an oxide a t the surface. Air was admitted to the system at - 183°C, which

was then quickly brought to room temperature.

By th is method film s of many metals were sp lu ttered , e ith e r as

"specpure" wires or as e le c tr o ly tic deposits on a clean tungsten filam en t,

0.01 mm. diameter.

(c ) Magnetic S u sc e p tib ility Balance.

The determinations were made with a Sucksmith type balance ( 8 7 ) as

used a lso by D ilke, Eley and Maxted (88).

The apparatus i s shown in F ig. 8, and was constructed from simple

J..

m aterials. I t consisted of a low extension micro-balance made from a

circu lar s tr ip of phosphor-brcnze which was 38, 42 or 46 s .w .g ., according

to the s e n s it iv it y required. The width of each str ip being 2 mm., bent

in to a c ir c le of 5 cm. diameter. The ring was supported at the top by a

s l i t in a brass rod and was enclosed in a brass box, 10 cm. x 10 cm. x

13 cm., with g la ss windows in three sid e s . Light from a galvanometer

lamp was r e fle c te d from mirror B to mirror A and then on to a v e r tic a l

sca le a t the focussing distance of 1 metre, by a system of plane mirrors

outside the box. A lig h t mica scale pan was attached to the lower

extrem ity o f t ie ring from which the specimen was suspended by means of a

fin e gLass or nylon thread.

The specimen was packed into a thin walled g la ss p h ial, 7 cm. long
o f
Li.
00

SUCKSMITH BALANCE
 - 52 -

x 1 mm* diameter of uniform cro ss-sectio n , The m aterials were thoroughly

ground in an agate mortar to pass JOQ B.S, siev e and ca refu lly packed in to

the p h ial hy tapping on the bench several hundred times in order to

minimise discrepancies in packing den sity, which i s the p rin cip le source

o f error in th is method.

Since an electro-magnet was not readily availab le the measurements

were made with permanent magnets of f ie ld strengths 10,600 dersteds and

1,500 oersteds resp ectiv ely . These were obtained from W01, Jessop & Sons

L td ,, S h e ffie ld , and were manufactured in the sin tered "Alcomax I I I 1’ allo y *

By means of a hydraulic l i f t (a modified lorry ja ck ), the magnet was

smoothly manoeuvred into p o sitio n , and the maximum p u ll, (when the tip of

the specimen was at the middle of the pole gap), was determined from the

galvanometer sca le d eflectio n . This was repeated u n til a constant

d e fle c tio n was obtained. The specimen was prevented from attaching to

e ith e r pole by means of two fin e nylon retain in g threads attached to the

top of the magnet by a mica support.

The apparatus was calibrated with a saturated solu tion of ferrous

amnioninm sulphate, fo r which X (magnetic su scep tib ility /g m )

where T * °K. In addition, the s u s c e p tib ility of various pure oxides

o f known values was determined at the two f ie ld strengths and agreed to

w ith in Vfo with I,C ,T, data.

The measurements were a l l made at room temperature and the f in a l

value fo r each system was taken as being the average given by three

separately packed specimens. The X values may be c r it ic is e d , on the

 - 53 -

grounds that determinations should have "been carried out a t d ifferen t

temperatures to obtain s u sc e p tib ility isotherms (Selwood, 6 j ) , This i s

a u se fu l method of detecting the presence of paramagnetic im purities but

was rather out^mth the scope of th is work* I t was f e l t that careful

measurements at two f ie ld strengths would be s u ffic ie n t to characterise

each substance, and a graph of X against l/H would in dicate whether any

ferromagnetic im purities were present*

The ca lcu la tio n of X was carried out in the follow in g manners-

Magnet moment induced/cc * kH.

Ihere k * volume su sc e p tib ility and H « f i e l d strength*

Magnetic moment induced in volume d v » kHov*

Now, force a ctin g o n o v , d f = moment .^=y

(Since T v = A J l)

In tegratin g over the whole length ( l ) of the specimen, we haves

i*e* F « -§• kAH^.

But F m $ w g , where Tw * p ull in gm. Wei^it and X (gm. s u s c e p tib ility )

specimen )

• • (W * Weight of specimen)

(d) Surface area apparatus and measurement.

The apparatus used i s bhown in Fig* 9 and was a modified version o f

th at proposed by Duncan ( 7 6 ) and Wooten and Callaway-Brown (7 7 ). Mercury

 O
u.
<r>

SURFACE AREA APPAR-TUS

 - 54 -

s e a ls toeing replaced toy vacuum stopcocks, and a se r ie s of three expansion

volumes introduced.

The measurements were made with ethylene adsorption at - 185°C, and

p r e limi nary t r ia ls were carried out to fin d a su itab le stopcock grease

which would not absorb th is gas. Various commercial high vacuum lu brican ts

were used, as were mixtures composed of fused d extrose/glycerol and mannitol

as recommended toy Tyrer (90). These were applied without su ccess, w h ilst

although very l i t t l e ethylene absorption was detected, the greases were

lia b le to retain moisture and a lso to carbonise w h ilst flaming out the

apparatus.

This d iff ic u lty was f in a lly overcome by using a fluorocarbon grease,

k in d ly presented by I .C .I ., Widnes. For example, on allow ing vapour/

grease contact for 12 hours, the pressure in the system remained constant

a t 5*5 x 10-4 mm. Hg, as measured on a McLeod gauge.

The apparatus was evacuated by a high e ffic ie n c y mercury d iffu sio n

pump backed by a mechanical o i l pump, attached to the vacuum manifold

through two liq u id a ir traps. The to ta l volume of the apparatus was

-6
2,100 c .c . and an evacuation to 8.2 x 10~ mm. Hg pressure when l e f t for
-5
60 hours changed to 7*9 x 10 mm. Hg, a reasonable leakage for a vacuum

apparatus of th is s iz e and one which would not in ter fere with the surface
area measurements.
The volumes of the various parts of the apparatus were determined

from that of the McLeod gauge - using the bulb volume o f the McLeod as

a gas p ip ette. The McLeod bulb volume i/ra,s previously calibrated a g a in st

a fla s k attached to the manifold which, in turn, had been calib rated by

weighing the mercuiy required to f i l l i t .

 - 55 ~

A 500 ml. fla sk containing ethylene, ( 99% p u rity, I .C .I .) kept

over dry sodium or was connected to the manifold v ia the relea se

v a lv e, the expansion bulbs and manifold having previously been subjected

to a hard vacuum accompanied by flaming for 45 minutes. When th is was

complete the bulbs were closed to the manifold and a liq u id oxygen trap

placed around U-tube. The ^2^4 ¥ias ^^Ien allow ed to issu e in to the system

and was su ccessiv ely evaporated, condensed and pumped o ff to remove

entrapped a ir . f in a lly , the ethylene was admitted to the expansion

bulbs which were then closed to the manifold.

The measurements were made by having a liq u id oxygen trap at a

constant le v e l around the specimen tube. The whole system was evacuated

and a very small amount of allowed in to the manifold (not s u ffic ie n t

to reach the condensation pressure) and the pressure measured on the McLeod

gauge by a cathetometer reading to .001 mm. Tap A was then opened and the

allowed in to the specimen side of the system. A fter about 15 minutes,

tap A was closed and the McLeod gauge reading taken. The adsorption was

then found by the usual d ifference method calculated from the to ta l gas

charge and the quantity of gas in the gas phase.

The gas charge was checked at the end of each run by measuring the

pressure with the specimen a t room temperature. According to Duncan ( 7 6 ),

with th is system errors due to thermal transpiration are generally

n e g lig ib le as i s the Knudsen pressure drop through liq u id oxygen co o lin g .

Ko minor gas law or glass adsorption corrections were applied; the

absolute accuracy of the surface area figu res being probably of the order

o f + 15$ .
 Ld
h -!
O 01
>“
g CO
U

&
II.

(/)i

OJ

ai5
 - 56 -

(e ) S ta tic system.

The equipment i s shown in Fig. 10 and was o f simple design, con­

stru cted in pyrex g la ss . I t consisted of a ca ta ly st chamber, A, 20 cm.

long x 2 .0 can. in tern al diameter, jacketed in an aluminium tubular furnace,

F, which was supplied from the mains via a "Sunvic*' input con trol, the

temperature being reg istered by a standard chromel/alumel thermocouple

tig h t ly jammed between the catalyst chamber and the c lo s e - f it t in g furnace.

The chamber was sealed to the vacuum lin e , in order to dispense with

greased jo in ts .

The oil-manometer, M, contained fr e sh ly d is t ille d dibutyl phthalate

and the peroxide reservoir, C, 12 mis. H^O^ (>99/^ w/w). The vacuum was

supplied by a sim ilar unit to that described under metal film s and was

reg istered on a Pirani gauge.

Method of measurement.

The HgOg in C was frozen by liq u id oxygen, melted and then pumped

o f f to remove entrapped oxygen. This cycle was repeated several tim es.

(The system was previously baked out and evacuated to 5*0 z 10 ^ mms Hg).

The peroxide reservoir was then brought to 20°C, tap T^ opened and

the vapour allowed to escape into the volume between T^ and Tg. Three-

way tap Tg was then opened to allow the gas in to the reaction chamber,

and the decomposition followed by the pressure change reg istered on

manone te r , M. Readings being taken every 20 seconds.

I t was o r ig in a lly intended to deposit the catalyst on the w alls of

the reaction chamber in the form of an oxide film , to obtain a reproducible

surface. However, in the experiments reported the ca ta ly st was 0.5 gm.

p-yjrip. ground to pass ^00 B.S. s ie v e , dusted over the in sid e of the tube
v / / v ~z
\
z

u
h
0
z
*

ul
o
X

m
o
Ll_

PELLET PRESS, ANVIL & PLUNCER

 - 57 -

with, s lig h t excess at the bottom of the v e s s e l. The in vestigations, were

carried out in the range 30 - 120°C and 0 . 5 - 2 mm. Hg pressure of

vapour.

The unit was constructed towards the end of th is research period, in

order to compare the e ffic ie n c ie s of high a c t iv it y c a ta ly sts in a s t a t ic

system w ith that found by the flow method, and so to examine the p o s s ib ilit ie s
o f obtaining a correlation with other workers (34, 3 5 ) on sim ilar s t a t ic

systems with in a ctiv e ca ta ly sts.

(f) P e lle t p ress*

Since the sin g le oxides and mixtures were to be examined as compact

s li p s o f m aterial, a suitab le p e lle t press had to be constructed fo r th is

purpose. Considerable d iff ic u lty was encountered in making the d ie, and

many designs were experimented with, in th is department and a lso by the

Admiralty, T .E.E., Greenock, whose assistan ce was most valuable in arrivin g

a t the f in a l form of press.

I t was found that a pressure of 17*5 to n s/sq . in s . was required to

give a compact s lip of material without using an adhesive. At th is pressure

d i f f i c u l t i e s arose through the buckling of the plunger and wear of the

in sid e of the d ie. The press eventually used i s shown in P ig. 11 and was

constructed in th is C ollege. The main body being of mild s t e e l , the

plunger, a n vil and four in serts being made of a high carbon, chrome s t e e l.

This was used to press the dry m aterial in to s lip s , 2 .5 cm x 1 cm x 0.1 cm.,

which were used in the flow apparatus.

The press was thoroughly cleaned a fte r each use with an abrasive,

follow ed by a thorough water rinse and drying with acetone. The p e lle t

surface was a t f i r s t gently rubbed with f i l t e r paper in order to remove

 - 58 -

any p o ssib le contamination due to contact with the s t e e l plunger* I t was

a ls o found that when a s lip was pressed between sheets of "iTuon” no

d ifferen ce was observed in the c a ta ly tic a c tiv ity * Thus by adopting

str in g en t cleaning procedures, p e lle t contamination was so n e g lig ib le as

to have no e ffe c t on the ca ta ly tic performance*

(g) Agate mortar.

This was a mechanical mortar driven by a 0*25 H*P. motor, manufactured

by Thos* Dryden and Sons, Preston, and kindly loaned for the purpose of

t h is research by Admiralty Materials Laboratory, Holton Heath, Dorset. It

was used in grinding oxides to a uniform p a rtic le s iz e and a lso served to

provide an intim ate mixture of two oxides by grinding together.

The mortar capacity was 5 gm. and in 12 hours grinding,gave a range

o f p a r tic le s iz e from l-5yu , startin g w ith A l ^ of 34- B.S* s ie v e . To

give the same reduction ra tio with PbO, e t c . a time of 7“8 hours

was required. The mortar was thoroughly cleaned a fte r each run by
scrubbing w ith water and detergent, then f in a lly washed and dried.
 - 59 -

(2) MATERIALS

(a) Hydrogen -peroxide.

This was Laporte's s ta b ilis e r -fr e e H.T.P. (High Test Peroxide)

supplied at 86 - 9Qfi w/w. In experiments involving additions o f water

vapour, and where r e la tiv e ly high, concentrations of peroxide vapour, c . l

m.m., were required in the gas phase; the H^Og was concentrated by

d i s t i l l a t i o n in the apparatus of Fig. 12.

800 mis. of u n sta b ilised H.T.P. were d is t ille d using a straigh t

tube r eflu x column in su lated with quartz wool, at 55 “'65 °C and 72 mm.

pressure. The peroxide d i s t i l l a t e was condensed in a fla sk cooled by an

ice/w a ter mixture. The o i l pump was protected by a drying tower containing

calcium chloride and gla ss wool, and by another column of quartz wool

impregnated with Ag^O and K MnO^ to decompose any peroxide vapour lia b le

to be drawn in to the pump. The apparatus was constructed in pyrex g la ss

and the evacuation tubing was e n tir e ly in thick-w alled p .v .c .

Bach d is t illa t io n la ste d s ix hours and resu lted in a concentrate o f

97 - 98% peroxide as determined by refra ctiv e index and t it r a t io n with

standard fflhQ,. About 60 mis. of concentrate was obtained from each run.

When 300 - 350 mis o f th is concentrate was c o lle c te d i t was concentrated

in the same apparatus u n til an Eg02 fra ctio n was obtained for which n^ =

1.4066 a t 25°C. This corresponded to 99*9^ ^2 °2 ( Gi^u®re and Geoffrion

(9 1 ), and Wynne - Jones et a l ( 9 2 ). The strength of th is peroxide was

checked from time to time by permanganate t it r a t io n , and i t was found

that the concentration did not a lte r by more than 0.5^ standing for

4-5 months. I t was found a lso that during use in the saturator the
QUARTZ I N S LI t- ATT 10 Ni
2
J

u»
Sa
&£
ah

C\J

FIC IZ.
HYDROGEN PEROXIDE CONCENTRATOR
 - 60 -

concentration of the E^Og S0^u^ 011 kept w ell up# This was because

any tendency to decompose was balanced by the s lig h t fraction ation

which occurs during passing of the nitrogen#

(b) C a ta ly sts,

These were o f three types v iz : - Metal s lip s , bulk oxides and

evaporated film s#

(i) Metal s l i p s .

”Specpurew metals were purchased from Johnson Matthey & Co., in a

form cut to f i t the reaction tube. The method of surface cleaning and

preparation d iffered fo r each material examined and was as follow s.

(a) Fe. The "specpure" iron p lates were etched with concentrated HC1,

and chem ically polished with Marshall’s so lu tio n , i . e . a warm solu tion
containing 1*5 o l s . H.T.P#, 2.5 gms. oxalic acid and 1 drop conc. E^SO^-

in 100 m is. water, then washed ca refu lly in d is t ille d water, dried, and

heated in an oven fo r one hour at 120°C. This would r esu lt in the formation

o f a film o f le Q / l e ^ Q y c.200A° th ick , as the surface oxide. Further

oeid ation to takes place at temperatures above 200°C as has been shown

by Nelson (93)#
(b) M. This was employed as a sheet of '’specpure" m etal, i . e . purity

greater than 99 . 99a/0, and was degreased in benzene, washed in d i s t ille d

water and heated in an oven a t 120 C for one hour. The oxide film was

considered as NiO approximately 50A° th ick , sim ilar to that found by

Rhodin ( 9 4 ) on copper.
(e ) £L. This was cut from a sheet of e le c tr o ly tic aluminium of 99

puxi-by, Xt was rubbed down with fin e sand—paper and then etched with

d ilu te n it r ic a cid , washed, dried with f i l t e r paper and heated in an oven

 - 6l -

at 120 C fo r one hour. This resu lts in an oxide film of optimum thickness

8 QA , co n sistin g of A^O^ or Part i a l l y hydrated Al^O^.

Alumi nium metal was a lso treated in hydrogen at 600°C in a tubular

furnace fo r 6 hours and quickly quenched before using in the flow system.

(d) Co. The surface o f the "specpure" cobalt plate was prepared in an

analogous manner to the Ni above and was considered to have a sim ilar oxide

surface film .

(e ) Sn. The t in p lates were made by melting AfR. t in granules, moulding

and machining to s lip dimensions. They were then rubbed down with fin e

sand paper, washed, degreased and dried at 120°C. The surface was then

considered to be covered with a film of SnOg.

(f) Ag* This specimen was 99*9^ s ilv e r and was treated with d ilu te HNO^,

washed and dried at 120°C for one hour. This would resu lt in a surface

covered w ith Ag^O, about 100-200A° thick (Cabrera and Mott, 4 1 ).

(g ) Cu» Tk® surface of the "specpure" s lip was prepared as in (b) and

(d ), avoiding the use of which oxidises the surface to CuO. The oxide

film would be CUgO and approximately 60A° thick as shown by Rhodin (94),

and in previous woifc in the laboratory by N ettleton (9^)»

(h) M l. Owing to the d iffic u lty of obtaining th is metal of high purity

in a form suitab le fo r examination in the flow system, i t was used as a

s l i p w ith a copper content of 8^?, which i s le s s b r itt le than the pure metal.

The surface was smoothed with fin e sand paper, treated with d ilu te HNO^,

washed and dried at 120°C.

I t w ill be clear from the above that the metals were a l l covered w ith

th a t oxygen layer which would be formed by standing or heating in a ir for a

 — 62 ^

short tim e. The film s would normally he protective i . e . they would not

tend to increase much under normal conditions.

( i i ) Oxides.

The method of preparation of an oxide ca ta ly st i s very important.

A tten tion must he paid to the temperature at which the ca ta ly st i s prepared,

the period o f time that the catalyst i s held at th is temperature, and the

gas atmosphere in the furnace (Huttig and Heinz, 97). A ca ta ly st is often

most a c tiv e in a reaction i f i t is prepared in the gaseous atmosphere in

which i t w ill he used c a ta ly tic a lly as has been shown by several

in v e stig a to r s ( 2 1 , 7 3 )*

Time and method of cooling a fter heat treatment w ill a lso be important

fa c to r s . A d ifferen t active sta te w ill resu lt i f a ca ta ly st i s allowed to

cool slo w ly , than i f the defects are "frozen" into the la t t ic e by quenching

(Nizboer, 98). In addition to heat treatment, the mode of i n i t i a l preparat­

ion and admixture of the two oxides is important. Such features as p a rtic le

s iz e and method of mixing (as powders or by co-p recip itation ) can a lso be

v aried .
In the experiments reported a mechanical mixing procedure was adopted.

Samples of the divalent oxides were either prepared by decomposition of

the A.H. n itr a te s e t c ., or by p recip itation of the hydroxides, and ig n itio n

to the oxide. The oxides thus prepared were ground in an agate mortar to

pass 300 B.S. sie v e .

( l) MtnO-, was prepared by ammonia p recip itation of the hydroxide from a 50 /k
— O .
so lu tio n of A.H. manganous n itr a te . The p recip itate was washed with

d i s t i l l e d water, b oiled and filt e r e d several tim es, then dried at 150 C.
 - 63 -

This product was then ign ited in a furnace in a flowing oxygen atmosphere

at 600 C for 40 hours. Subsequent analysis (by the bismuthate method)

showed of the triv a len t oxide. The change Mn70„ only takes

place a t 1000 C, and below th is temperature Mho 0 _ w ill resu lt from

ig n itio n o f any manganese oxide in a ir or oxygen.

(2 ) N ickel oxide presented sp e c ific d if f ic u lt ie s in preparation sin ce i t

rea d ily occludes oxygen into the la t t ic e and has been reported to e x is t in

many m odifications ( 9 9 )» I t i s now accepted that the structures Nio0 ..,

and Ei^O^ do not e x is t and that anomalies in the lite r a tu r e may be

explained by assuming a rhombohedxal unit c e ll in certain high temperature

preparations (lOO). Previous investigators (lO l) made no allowance for the

presence of chemically bound water in low temperature preparations of the

oxide from N i^ H ^ .

The oxide was prepared by ig n itio n of the A.R. n itra te in a ir , follow ed

by con trolled heating for 6 hours at 400°C. This removed a l l traces of

chem ically bound BgO, and the oxide was then ground to pass 300 B.S. s ie v e .

The r e s u ltin g "black" NiO was analysed by the dimethylglyoxime method and

corresponded to the formula EiO-j^-^, approximately 12$ excess oxygen.

When th is substance was fired in a ir at 800-900°C for 10-12 hours,

"green" NiO resu lted which on analysis showed a stoichiom etric ra tio of

n ick el to oxygen* Magnetic su sc e p tib ility and X-ray determinations were

made on these oxides and the resu lts are discussed la te r .

(3) By 0 This oxide was obtained from Johnson Matthey & C o., and was of 99•9}'^
—2 3
p u rity .
 ** 64- ^

(4) n 2e?. This was prepared by p recip itation of AJ (GH)^ by ammonium

hydroxide from a solution of the A.R. n itr a te . The p recip ita te was b oiled

and washed several times and calcined at 900°C to give Alo0 , in the
2 3
manner prescribed by Day and H ill (102).

(5) L a ^ . and (6) Ga^Q^. These were prepared by high temperature ig n itio n o f

the "specpure" n itra tes* The resulting white, homogeneous powders were

then stored in an evacuated desiccator, over $2®$* uzrtil required.

(7) O o Q /O o fly Cobalt oxide was prepared by "ignition of A.R. Co^G^)^*

6 HgO a t 1000°C in a ir . This resu lts in the formation of a black powder,

which i s considered to be ^0 ^ 4. with the spinel structure ( 1 0 3 ).

(8 ) ffegO^* The oxide was prepared by adding ammonia and ammonium chloride to

a so lu tio n of A.R. FeCl^. The precipitated hydroxide was f ilt e r e d , b oiled

and washed several times, dried in an oven at 130°C, and f in a lly ig n ite d

to the reddish-brown oxide a t 7C0«800OC in a tubular furnace*

(9) CuO. This was prepared by low temperature (c.450°C) ig n itio n of b asic

copper carbonate. The resu ltin g black oxide being stored in a d esiccator

u n til required.

(10) PbO and ( l l ) ZnO. Yellow lead oxide and white zinc oxide were prepared by

ammonium p recip ita tio n of the hydroxides from solutions o f the A.R. n itr a t e s .

The p re c ip ita te s were treated as in ( 8 ) above, followed by heating in a ir

in a s i l i c a crucible to give the oxides.

(l2 ) MgO. This was used as the A.R. oxide (Hopkins and Williams, L td.) and was

simply heated in a furnace at 500 ^ exPe^ EgO orCOg p ossib ly

present and then quickly cooled, and preserved in a desiccator.

 - 65 -

(l^ ) & Cinnamon —brown cadmium oxide was formed by ig n itio n of* A.R.

CdCRO^J^.-l-HgO in a ir at 50C“400°C. I t was also prepared by ig n itio n of

the A.R. carbonate at the same temperature, and as A.R. CdO as supplied

by Hopkins and ¥ illia m s , Ltd.

(1 4 ) ^igO. white powder was prepared by decomposition of the A.R. carbonate

a t 800°C in a stream of hydrogen, and stored in a desiccator over P0G.-.

( i i i ) Metal film s .

The metal film s were prepared by flash in g the metal from a “specpure”

w ire, or from an e le c tr o ly tic deposit on a clean tungsten filam ent. The

technique used was described under apparatus. The film s flash ed were of

Cu and H i, thicknesses 5“12yu, and were a l l deposited onto a surface

cooled in liq u id oxygen to avoid p referential orientation of s p e c ific

c r y s ta l p lan es.

(iv ) Equimolar mixed oxides.

The mixed oxides based on Mno0, were heat treated, a fte r mixing and

grinding fo r 5-6 hours, in a m ullite tubular furnace at a controlled

temperature in a p re-fired m ullite boat. The heat treatment la ste d for

6 hours in an in ert atmosphere.

The spin el-typ e systems based on CuO were prepared in a sim ilar

fash ion only d iffe rin g in that the sin terin g was carried out in a ir .

(v) Trace im p u rities.

The se r ie s of mixtures containing small percentages o f foreign cations

were prepared by intim ately mixing the required amounts in an agate mortar

fo r 5 - 4 hours and fir in g in a ir for s ix hours at 600°C. By th is method

the impurity i s dispersed in to the ”mother” oxide, although the dispersion

may not n ecessa rily be evenly distributed throughout the l a t t i c e .

 - 66 -

RESULTS.

(l) General#

In the experimental section , resu lts were reported in support of the

claim that with a l l the ca talysts (even the most a ctiv e ones), the decompositions

are co n tro lled by a surface process and not by gas phase d iffu sio n . Accordingly,

the fo llo w in g r e s u lts are displayed in terms of a heterogeneous chemical

reaction#

The two types of measurement, based on flashed film s on the one hand, and

metal and compressed oxide s lip s on the other, represent two d istin c t flow

patterns and the resu lts w ill not be compared q u an titatively. However, within

each s e r ie s the flow ch aracteristics of the system (contact time, type of flow )

are constant between one catalyst and another, and i t w ill therefore be con­

sidered v a lid to compare the ca talysts on the basis of the decompositions at

given temperatures and the temperature c o e ffic ie n t of the decomposition.

The ca lcu la tio n of the v elo c ity constant, k, and the derived value E^,

the energy of a c tiv a tio n , depend on knowledge of the order o f the reaction and

on other fa c to r s d ealt with more fu lly in the discussion. Here, however, the

method o f determining the order and E^ w ill be given in d e ta il.

(a) Determination o f Reaction Order.

D e fin itio n of “fra ctio n decomposed”, D»

Let the concentration of vaPour &as s^168,111

before reaching the catalyst * m o ls /litr e .

Concentration of vapour leaving the catalyst

= = f(h m o ls /litr e .
 ~ 67 ~

where f ( O ^ f <Cl) i s the fraction of vapour undecomposed.

We define D *= ( l - f ) as a measure of catalyst a ctiv ity *

I f the t it r e in O.OUf KMnO^ observed for a 5 minute absorption t e s t i s

Xm ls without the ca ta ly st. Thens-

Ci * ft x
.»where represents
the a n a ly tic a l fa cto rs.

I f y i s the t it r e a t a given time during the c a ta ly sis run, then:*

fC. - jfo

and thus f = y/x

and D = x -y
X

K in etic treatm ent.

IK
I<01

K low ----- > Ci fC. -----

2 _ l 1

I*------ *2“ — ^

Consider the gas flow past the ca ta ly st in the form of a rectangular

area as in the sketch. This w ill approximate quite c lo se ly to the actual

conditions w ith s lip s . I t is assumed that g a s/so lid equilibrium at any

zone such as BB1 i s completely attained.

The time taken by the flow front to pass from A to B = t^

t! H It M m » »» 11 " ” B to B' = A t

Xf c i s the average concentration in the increment of volume represented

by the p o sitio n of the flow front between B & B*. Then for f i r s t order

k in e tic s i f A c represents the concentration change in the time A t , we

state:*-
l\ c dc
= k]_c
at
In teg ra tin g and assuming that i t is valid to include in k^ any fa cto rs

a sso cia ted -with changes in surface i t s e l f along the s l i p , we have

" In C ■ k^t + constant.

vdiere t i s the time of flow across the catalyst to the point at which

the concentration i s C. The constant may be evaluated, since when

t = ©, C = C. and constant = - In C..

i i
A lso , when t - t 0 , C * fC ..
27 i
.** ©ver the whole length of the s lip from t = © to t= t2 , we have:-

, Ci . .
fC. ~ \ 2
l

i# e * \ = log l^D t1 )

Since tg i s a constant for a given flow ra te, i f f i r s t order k in e tic s

obtain then D, fra ctio n of H2@2 vapour decomposed, remains constant fo r

any value of Ch •

This may be represented graphically by p lo ttin g Ch against f Ch

(or x again st y ) . For f i r s t order k in etics the points w ill a l l l i e on

a str a ig h t lin e passing through the origin as illu s tr a te d in Fig. A.

For zero order k in e tic s :-

- #dt - ko

i . e . the decomposition rate i s independent of the peroxide concentration.

I t fo llo w s th at: Ci - C = kQt 2 « constant.

and DC^ = constant,

showing that B varies in versely with Ci. Zero order k in etics would appear

on the graph, F ig. A, as a series of points on a lin e p a ra llel to that fo r

micf f c f r ^ )
\j
( OR
ST
L_C AVI N G* CATAUV
VAPOUR
H 2 02
OP
CON CEN TRA TION

0 5 10

C O N C E N T R A T IO N OP H jO g V A P O U R P E A C H 'N S C V T A L V S T (o ^ X H lX l tit+ O

TH E O R E T IC A L . C U pV E fe EjHOWMNG RELATIVE P O S IT IO N S OF EaECONO,

F IR S T AND Z E R O OROER P R O C E S S E S .

F l G. A
 - 69 -

the ‘blank (not passing through the o r ig in ).

For 2nd. order k in e tic s the fraction of the decomposed w ill

in crease with in a rather complex fashion so that in changing from a

low Ch to a higher one the new 3) w ill he higher than i f 1 st order k in e tic s
p rev a iled .

We have not attempted to interpret the resu lt in terms of p recise

fr a c tio n a l orders since no steady resu lts leading to such an in terp reta tio n
ever a ro se.

(b) Determination o f the temperature c o e ffic ie n t.

The usual method of evaluating temperature c o e ffic ie n ts i s to obtain

the steady reaction rate a t various temperatures and then to p lo t the

logarithm o f the rate constant against the reciprocal of the absolute

temperature. The slope o f th is lin e being put equal to - E^/2.303R,

y ie ld s the value E^ which i s quoted as the "experimental a ctiv a tio n energy".

This value may often be very d ifferen t from the true surface process

a c tiv a tio n energy, because of variable heats of adsorption at the d iffe r e n t

temperatures, and in the method described above and E^ cannot be

adequately separated. In the sp ecial case of a zero order process E^ =

E . , since the overall rate i s not a ffected by the adsorption which i s

true*
regarded as complete, and consequently there i s no variable A H fa c to r .

The method used to find temperature c o e ffic ie n ts or a ctiv a tio n

en ergies in th is in v estig a tio n , was to follow the change of i n i t i a l decom­

p o sitio n a t a new temperature a fte r a fa st temperature change from a steady

decomposition value. By a minor extrapolation to zero tim e, the decom­

p o sitio n a t the two temperatures could be compared, i t was assumed, at a

constant ■surface condition and constant &H. Since the disturbance of

 - 7 0 -

surface equilibrium i s slow, by th is method the adsorption on the surface

could be considered to be momentarily "frozen", and the temperature co­

e f f i c i e n t would be equivalent to the activ a tio n energy for the surface from

which the change was made. A more d etailed treatment of th is topic w ill be
given in the d iscussion sectio n .

The temperature c o e ffic ie n t may be calculated in the follow ing ways-

Experiment shows the reaction to be f i r s t orderwith respect to B^O

concentration. Hence from ( l ) above, we haves—

\ - (2 .3 0 3 /t) lo g ( l/l- D ) . (1 )

How, sin ce = A exp. (-E^/RT)

or In « In A - E^/RT

Then, In kj_ - In ^ ^ (2 )

i f we may assume A to remain constant during thetemperature change.

(This w ill be discussed la te r but fo r the moment the assumption may be

accpeted fo r a fa s t temperature change). Hence by su b stitu tio n of ( l )

in (2) we haves-

EA - 4 .577. ^ ) lo g . log - lo g . log

which y ield sth e experimental a ctivation e n e r g y in terms of decompositions

and Dg temperatures T^ and T^ °K resp e c tiv e ly . F ig. 13 illu s t r a t e s

the general pattern of rate changes found, and the small extrapolation

used to fin d Dg imme d i a t e l y a fte r a temperature change.

The values of a ctiv a tio n energy obtained in th is fashion were

independant of the surface condition of the c a ta ly st. For example, an

Ag/AggO Pla-te when prepared as described e a r lie r had an E^ value for

 o
o
o
o

o
OO
o

N 0 lX lS 0 d iM 0 3 3 d
rO

o
to

—O

o
CV1

3 S V -L N 3 D 2 i3 d
o
R ecomposition p a t t e r n , fast temperature changes.
Ox i d i s e d C o p p e r S lip C a t a l y s t
o
rO
 - 71 -

4.7K .cals/m ol between 58-100°C. The same value was found a fte r the plate

had been h igh ly etched with H.T.P. solu tion which gave the ca ta ly st a

w h ite, pearly appearance - although the percentage decomposition a t the

same temperatures was very much d ifferen t.

(<0 R eproducibility with the flow system.

The steady value of the e fficien cy of a ca ta ly st could u sually be

duplicated with a sim ila rly but separately prepared sample to w ithin 5%

fo r m etals or metal film s, and with the bulk oxides to w ithin 1%. The

fo llo w in g r e s u lts demonstrate th is . Eor the metals and metal film s the

i n i t i a l behaviour was not q u antitatively reproducible but the same trend

was always exhibited. On the other hand, bulk oxides gave precise

rep ro d u cib ility in th is phase a ls o . The greater spread w ith oxidised

m etals may be held to in d icate that the metal i t s e l f had not been brought

in to a uniform sta te before use.

TABLE 2 . I llu s tr a tin g reproducibility of flow measurements with

separately prepared samples of the same specimen.

Oxide Temperature of Time (m ins.)

Sample Run (°C .) from i n i t i a l Percentage Decomposition
exposure of
catalyst to Sample 1 Sample 2
H2°2*

Zn0 /Mn2 05 60 5 6 4 .0 1 6 3 .5 0

N ti 10 5 6 .6 0 5 6 .1 0

M it 15 55*50 55.20
ft n 20 55.25 55.24
II H 25 55.25 55.24
 - 72 -
TABLE 2 (Continued.)

Oxide Temperature of Time (mins.)

Sample Run (°C .) from in it ia l Percentage Decomposition
exposure of
catalyst to Sample 1 Sample 2
H2°2-

O xidised 100 5 24.41 25.00

"specpure"
Pe s l i p n 10 20.04 36.50
it tt 15 35.52 37.25
« tt 20 35.05 3 6 .7 0

n tt 25 35.05 3 6 .7 0

MO ("green" ) 100 5 65.50 5 6 .5 5

it n 10 54.00 5 4 .5 0

it n 15 54.92 5 5 .6 3

n tt 20 54.01 5 4 .5 0

tt n 25 54.01 5 4 .5 0
 - 73 -

(2) C atalysis on Oxide Films of Pure Metals.

(a) Oxidised fla sh ed metal film s. (Nickel and Copper).

I n i t i a l changes of e ffic ie n c y .

The o rig in a l object in studying the c a ta ly sis on evaporated metal

film s was tw ofold. I t was intended to compare the c a ta ly tic a c t iv it y of

the bulk oxides with that of the same oxide on a pure, clean metal support •

Secondly, i t was hoped to measure the change of conductivity of the oxide

film during c a ta ly s is , in the manner used by Gray and Savage ( 5 0 ) and

Suhrmann and Schulz (104) to study the adsorption of various gases on metal

film s . A u n it was constructed for th is purpose but unfortunately could not

be used owing to the pressure of other experimental work.

The k in e tic s of the decomposition process were examined on copper and

n ick el film s and a d etailed in vestigation carried out on n ic k e l. The film s

were deposited onto a surface cooled in liq u id oxygen. During th is process

-5 -6
the pressure was kept in the region 10 - 10 mm. Hg., so that there

would be n e g lig ib le contamination of the film surface. Air was then admitted

and the film s brought quickly to room temperature. A fter th is process the

transparent n ic k e l and copper film s became black and bronze resp ectiv ely ,

due to the uptake of oxygen from the atmosphere. According to Beeck et a l.

( 8 9 ) film s deposited in th is fashion have a roughness factor of approximately

100 .
The general pattern of the decomposition changes on the copper film s

was sim ilar to that found by Hart & McPadyen (81) using Cu20 covered gauzes

(J ig s . 14 and 1 5 ). The e ffic ie n c y a fte r a quick drop recovered to a higher

steady v a lu e, dependant on the temperature of the run. On the other hand,

 E ffe c t o f T em p eratu re, o in a c t iv it y o f cop p er ($l)

100
D e c o m p o e i t i o i s i

4Q
QE
P e r c e n t a

ao 50 GO 70

E.FFECT T E V ' P E . i^ . A T U R . E ON A C T I V I T Y O F
o f OXlO'G ED
COPPER FI L.KA ( P R E S E N T FIC.
— &— ' .. ® — i
100 8 ------1— ...........
8 — 1
\
0

!C0‘VD D. GS’ c
\

SO
>) — ® — 40°c
g.— <x>— ^5 — 0 — c>c— ® — $
e c o m p o s it i o n

60 s
GO

./!
D

40

XI #T.^ --------- >L C

->o
*
e i? c e n t a q e

-
Y 1 *
0 — *
\ i «Jn -----------—
/ -----
80
f
P

________
rx /I A c,0 (bO 70

TIME
 - 74 -

the behaviour of tne oxidised nickel film s was more complex* Below 14-O^C ,

the c a ta ly tic e ffic ie n c y f e l l exponentially to a steady value in about JO

minutes* At 160 C a fa st increase was apparent which steadied to c.lOO^b

decomposition a fte r 4-5 minutes* At intermediate temperatures, c y c lic

v a ria tio n s occurred which eventually le v e lle d out* This i s illu s tr a te d

in F ig . 16 fo r a M oxidised film 5*7yu th ick , and geometric surface area

* 31 sq. cm*

Determination o f reaction order with film s,

■When the steady decomposition value had been reached at a given

temperature w ith a peroxide concentration of 0*88 mm Hg. in the vapour

phase; quick changes in concentration were made by reducing the peroxide

flo w and making up to 300 m is/ minute with dry nitrogen from the a lte rn a tiv e

stream* Fig. 17 illu s t r a te s the behaviour of an MO film at 100°C in

determining the k in e tic s .

Following a sharp change of HgOg concentration to h a lf the o rig in a l

va lu e, ( i . e . from Ch to C j/2), the fraction of H2 O2 decomposed remained

unchanged fo r some time. This suggests f ir s t order k in e tic s - the fra ctio n

decomposed was, in a given time of contact with the c a ta ly st, independant

o f the o rig in a l concentration. Soon, however the amount of c a ta ly sis rose

rap id ly and steadied out b r ie fly to a new higher point before sinking

slow ly to somewhere near the original amount. This cycle was repeated

once or twice before the rate s e ttle d at the fin a l value. The upper lim it

reached on the f i r s t cycle when compared with the ca ta ly sis before the

change to h a lf concentration indicated apparent zero order k in e tic s , i* e .

the t o ta l amount o f destroyed was independant of the concentration

 O
r

o
rO
(VI oO

o
v5

o
tO

(P
z

iti
*0
h

o o

o
0 -0 (VJ

o
o o
° *2 CVJ
oO
0 3^0^1*0330 H 3'DVi.N33S3d
 F IG . 17
o

o
o

£
J
40 L.
J
id
X &o
u I r ='
z
o 0 5 - . 5
u f
5? sa^ *
D Oo o
X hZ
o U
a u
o z
o IIU
>u
b
:r
>
I-
u
o < (0 **- C
^O^
^
u. 00
o c» <3 o O
z A A A J.
0
—t
vJrfi -*t
o
<VJ
£ b
O' Z
5 = * '
a

o o o
<VJ

NOtXl'SOdWODBQ 31DVXN3D»3d
 - 75 ~

reaching the c a ta ly st.

A change hack from Ci/2 to Ci gave the reverse of the above. F ir s t

a rapid movement to a lower decomposition percentage, suggesting again a

temporary change which could ea sily he taken as zero order. 'This was not

allow ed to revert to the f ir s t order value in the case illu s tr a te d .

However, on making the change Ci->Ci/2 i t remained steady at the origin al

" fir s t order" decomposition, a fte r a minor flu ctu a tio n . m en the con­

cen tration was lowered to C i/3, a sim ilar though le s s marked r esu lt was

recorded.

With the Cu^O film studied at 38°C, the in it ia l changes were the

same as for n ick el except that no rep etition of the f ir s t c y clic variation

was observed. A fter an i n i t i a l f a l l in a c t iv it y at Ci (p artial pressure

= 0.88 mm Hg.), the catalyst recovered in 15 minutes to a steady

decomposition value. Various changes of ^ ^ 2 concen^ra^^on gave few

d ev ia tio n s from a f i r s t order process. However, the changes Ci—^ Ci/2

and Ci/2 -~*C i/3 gave a decomposition r is e ind icatin g a pseudo zero order
process when compared with the steady value at Ci in the f i r s t case, and

Ci/2 in the second. In 10-20 minutes, these values deteriorated smoothly

to the o rig in a l " fir s t order" decomposition and then remained constant

a t th is value. The fin a l change, C i/3~ *C i, showed a s lig h t e ffic ie n c y

drop which in 15 minutes recovered to a steady le v e l ju st below the

o r ig in a l " fir s t order" decomposition (Fig. 18).

The order of the reaction on nickel film s was also examined at 80°C

and 140°C and the same variation in k in etic behaviour was observed, i . e .

there was always a tendency for the decomposition to r is e higher than the

f i r s t order value in lowering the concentration and to f a l l lower on r a isin g

 5

d
IL

ou oJ
(0 0r
*0 OJ
X . „
XL ~ '
JX.
% &v
J X
U 5 _ .
O if) tt 5
ftU lu ' '

op a o- <o to
a 4 CJ cO
0 o o o
u

o s
1J CT3
u. X1
rO a i .
u
y?
5 '
o a c£> ^ O
CP
J
dO r
U)
2 o OO
z
o
Or — oj rn
uf U
Q b-
5 ty Z
o o -
h
z CL
9 I
h
u l0
4* (iJ
ul or
tt Z

o 5
z
o
o
OJ 6
?
5
tt
y
h
bJ
a o U
o o a o
i/ j rvi

NCai^OdWOOaa aSV-LNBB&Bd
 - 76 ~

the concentration. Generally, however, these i n i t i a l flu ctu a tio n s

stead ied in time to the o rig in a l decomposition reg istered a t Ci.

In a l l the experiments on the metal film s the “f a s t change1* k in e tic s

were of the f i r s t order. Only temporary flu ctu ation s whicl}. were more

pronounced with IfiO than C^O, could lead to the suggestion of a zero

order process.

Temperature e f f e c t with o xid ised flash ed film s .

The temperature dependence of the reaction rate w ith MO was examined

in the region 60-l60°C, w ith a p a r tia l pressure of 0.88 mm Hg. o f HgOg.

The k in e tic s were taken as f i r s t order righ t through the temperature

range. Table 3 below gives the r e su lts obtained for the temperature co­

e f f ic ie n t in a d eta iled examination of a f u lly oxidised n ick el film 5*7yu

th ick .

TABLE 5# Experimental a c tiv a tio n en ergies, oxid ised nick el film

5.7^u th ick , (p .p . B-2®2 ~ ^*88 111111

Temperature Change (°K) Decomposition Change ($) A ctivation

Energy
T T (K .ca ls/m o l.)
1 2 D1 *2

433 423 99 92 22.0

423 416 92 82 19.3

416 403 82 59 17.4

333 333 7 17 11.3

353 373 17 28 10.9

Above 120°C the stepw ise temperature changes y ie ld a ctiv a tio n energies

in the region 17~22K,cals/mol. In the range 60-100°C the temperature

 - 77 ~

c o e ffic ie n t in d ica tes an apparent a ctiv a tio n energy of 11 K .cals/m ol.

Sim ilar r e su lts were obtained with the other oxidised n ick el film s

as shown in Table 4 fo r the temperature changes, 60-100°C and 120-l60°C.

TABLE 4 * Experimental a c tiv a tio n en ergies, oxid ised n ick el film s.

(p .p . H202 = 0.88 mm. Hg).

Oxide Thickness A ctivation Energy. K .ca ls/m o l.)

Film ( /a ) 60 - 100°C 120 - 160°C.

MO 7.22 11.2 20.3

MO 9.25 1 1 .4 22.9

MO 11.24 11.0 20.2

Cu20 7.63 12-14 -

The value obtained with the film , which was of the same order

as that found by Hart and McFadyen (81) with C^O covered metal gauzes,

was determined between 38-60°C, since the high a c t iv it y prohibited deter­

minations above 65 °C where lQO^b decomposition was recorded.

The values fo r the a c tiv a tio n energy of the various MO film s can be

seen from Table 4 to be r e la tiv e ly unaffected by small d ifferen ces in

film th ick n ess.

Surface areas of film s .

Attempts to measure the surface areas of the film s by the B.E.T.

method, using as adsorbent at 183°C> were u nsuccessful. A b etter

method has been demonstrated by Hideal and Trapnell (105) • I t involves

oxygen chemisorption, and the authors found that the surface areas of th e ir

film s were d ir e c tly proportional to the w eight. However as no comparative

 o o o
GO

N 0 I A ' lS 0 d W 0 3 3 a
o o
<\J

3'3M X N J3D a3cj

o
O

o
a

o
o
E ffect of temperature on activity of " S p e c p u r e 1'
N ickel sup showing i ni t i al c h a n g e s .
o
Ll.
 - 78 -

stu d ies were made w ith the flash ed film s, the surface area determinations

were not pursued#

(b) Oxidised metal s l i p s .

The metals examined were cut to f i t the flow reactor in Pig# 4#

The methods o f preparation and cleaning have been d eta iled under the

sectio n on m aterials.

I t i s proposed here to deal w ith MO/lfi f i r s t , as n ick el oxide was

taken as the model in a l l the experiments, including the bulk oxide and

impurity addition stu d ie s.

(i) M c k e l/n ic k el oxide.

I n i t i a l a c t iv it y changes and reaction order.

The course o f the changes o f the c a ta ly tic e ffic ie n c y on f i r s t

exposure to of the oxidised “specpure" n ick el s lip s was sim ilar to

that o f the n ick el film s. The same c y c lic changes of e ffic ie n c y were

observed at c,100°C and a lso sim ila r flu ctu a tio n s on changing the

concentration.

H g . 19 illu s t r a t e s the changes of e ffic ie n c y of the ca ta ly st before

reaching the steady decomposition le v e l a t various temperatures. F ig s.

20(a) and 20(b) show the v a ria tio n s found with the reaction order at 1J>6°C

and 183°C r e sp e c tiv e ly , with varying concen'tra^ ion8*

F ig. 20(a) shows the temporary increase in e ffic ie n c y of an Ni/UiO

"Specpure" s lip on making fa s t concentration changes at 138°C. The f i r s t

change a t point ( l ) gave a s lig h t a c t iv it y drop on doubling the o rig in a l

H2 O2 concentration. Following t h is , when the steady value had been

a tta in ed , a change back to Ci gave an increase of a c t iv it y which in 10

minutes f e l l to the o rig in a l " fir s t order" decomposition; and remained at

 SHOWING TEMPORARY INCREASE OF EFFICIENCY
ON DECREASING CONCENTRATION "SPECPURE*
N i AT ISSDC. POINT i 0-4-4. mm. 0 & 5 w w H^(o(o Hzo2
Z' ~ 0' 65MW. 0 - 4 - 4 - ..................
3> ‘ - 0 4 4 I H N . 0 2S ”
v ------------------------------
d e c o m p o sit io n

_ ..........

/
ra
>. n
Il

j
PERCENTAGE

) — O '— - (3 O Or pi R S T O R D E R ^ ^
K n c 7 O N E T IC S
z I

2Q-----------------------------------
-I--------------------------------- ;-------------------------------------------------------------------- ----------------------------------
0 20 4-0 <bQ 80
T ime (.Mi n s )
FIG. 20.*.
S>HOVNlNG, A P P A R E N T DEACTIVATION ON
inc rea sin g H2°2 concentration "s p e t c p u r e "m i at ISS°C .
P o i n t i o -b b o - 4 - 4 - i»m iv \ . h c i ^ l o b H p o ?
2 0-4-4- “ -H- 0- 23 ■< -P *
5 o-2^ " — 0 - 4 4 ................
D E C O M PO SIT IO N

— o
PE R C E N T A G E

o—

so

F IG .2 0 B .
 - 79 -

th is v a lu e, a fte r a temporary increase in e ffic ie n c y la s tin g 25 minutes,

on dropping the concentration to 2 C i/ 3 .

The apparent d eactivation of an M /liO ca ta ly st in runs o f increasing

concentration at 183°C i s illu s t r a te d in 51g.20(b ). The changes

C i—^ Ci/ 2 -....^ Ci/3 were accompanied by an i n i t i a l e ffic ie n c y in crea se,

e x a ctly analogous to those recorded a t 138°C. A fter the ca ta ly st had

reached a steady value a t C i/3, the reverse change to Ci/2 was accompanied

by a f a s t e ffic ie n c y drop which a fte r 25 minutes began to le v e l out and

approach the o r ig in a l " fir s t order" decomposition from below.

In a l l the experiments the departure from id ea l f i r s t order k in e tic s

was s lig h t over the whole process. Again, however, lik e the ITi/NiO film s

there was an i n i t i a l tendency fo r the ca ta ly st to "activate" on lowering

the concentration of peroxide vapour, and to "poison" on r a isin g the con­

cen tra tio n . In th is instance a ls o , unlike the oxid ised flash ed n ick el

film s , i n i t i a l c y c lic v a ria tio n s of e ffic ie n c y on making the concentration

changes were not apparent.

Temperature c o e f f ic ie n t .

The temperature c o e ffic ie n t of the decomposition on the oxidised

metal s lip s was found by making fa s t temperature changes in the sequences-

100°C —-^138°C - ~V 183°C — 100°C. In a l l cases the i n i t i a l changes o f

the decomposition at the new temperature were c a refu lly measured to enable

the evaluation of the rate change by extrapolation to zero tim e.

Table 5 below g ives the values found w ith an p a rtia l pressure of

0*96 mm Hg. at a constant flow rate of 308 rals/min.

 - 80 -

TABLE 5. Temperature c o e ffic ie n t n ic k e l/n ic k e l oxide; fa s t

temperature changes.

Temperature Decomposition a t T Decomposition a t Tgfck) A ctivation

Change (°K). Energy
I n itia l Steady I n it i a l Extra­ Steady (K. cals/m ol. )
T T Value Yalue Value polated Yalue
1 2
Yalue

373 411 11.2 8 .0 52.1 54.0 46 .4 11.7

411 456 52.1 4 6 .4 93.6 95.3 90.1 13.3

456 373 93*6 9 0 .1 7.5 5.2 9.1 15.7

373 456 27.5 12.2 87.5 90.0 86.1 11.8

456 373 87*5 86.1 5.6 4 .1 6.9 16.1

The f i r s t three values in the tab le r e fe r to fig u res obtained in the

course of a sin g le run on one sample of c a ta ly st. The la s t two figu res

r e fe r to a fr esh s l i p , which took 40 minutes to s e t t l e a t 1 2 decom­

p o sitio n at 100°C. When th is had remained steady fo r a further 20 minutes,

a rapid temperature change to 183°C was e ffe c te d and the steady value

a tta in ed in 15 minutes. Sim ilarly a fte r the drop to 100°C, the decom­

p o sitio n value became constant in 15 minutes at 6.9>o

0
The tab le shows th at the surface prepared at 183 0 "was associated

w ith a higher energy of a ctiv a tio n than the surfaces prepared at the lower

temperatures, (l38°C, 100°C). This e ffe c t was sim ila r to that described

w ith the NiO film s and although the actu al fig u res are not s t r ic t ly com­

parable because of the d ifferen t methods of surface preparation; they are

in very good agreement e sp e c ia lly a t the lower temperatures. Temperature

changes were made le s s quickly in the flash ed film determ inations.

 Ul \Z
m
h
h
o*- FIRST ORDER
o- 6 KINETICS ■
2
Y
O-OIN

4
MLS

0
0 OS 0 (o 0 9

PARTIAL. P ( ? e = jS U R e ltn.ni.Hg)

D ecomposition kinetics. C o & avjt S u p at 7 fic.

PO IN TS NUMBERED AT 5 MINUTE. In t e r v a l s .
FIG. 21 (a )

\z
KM 7 O4. TlTRE.

F i r s t Or d e r
Kin etic s O
S
O OIK

4
MLS

o
0.
partial - Pr essu r e Hg.)
D eco m p o sitio n k in e tic s , c o p p e r S u ip a t 7 Q>°C
 ICo
KM r^O^TlTREL

FIRST ORDER.
&
k i n e t i c s
O ' OI N

o
0 O <b

Decoiv\posi-noi>4 K i i s c t i c s , S i l v e r S l i p /\t£> 0 °C. f i c . 21(0

POINT4 f t r S M J N U T € Iw-reRVfM-S.
K.M,70 4 T1TRE

Fir st P eeper
X tfS E T IC E s
OOIN
MLS.

0 1 4 15,16

0 06

De c o m p o s i t i o n K i n e t i c s ^
FIC. 2 1 (d )
MANG,ANE.SE S l i p /\t G O °c .
 Other m etal/m etal oxid.es.

The other m etals examined in s lip form f a l l in to three c la sses

(1) Cu, Ag, Co, Mb. - A ctive Catalysts*

(2) Fe, Zn - Intermediate in A c tiv ity .

(3) A l, Sn - In active Catalysts*

A ll the m etals used would hear a p ro tectiv e film of oxide as shown

by the work of Mott ( 3 8 ) and fihodin (94, 106), as already described in

the se c tio n on m aterials. I t i s proposed to deal with each group

sep a ra tely .

) Class ( l ) .

The metals in th is c la ss a l l form s ta b le , coherent oxide film s of

lower oxides with a s lig h t tendency to pass to a higher valency sta te

a t the surface.

K in etic order of decomposition.

The k in e tic s of the decomposition on these surfaces showed much le s s

departure from f i r s t order behaviour on making changes of concentration,

than Hi/NiO. There was, however, a s lig h t tendency, exhibited by a l l the

specimens, towards d eactivation in making a change Ci/ 3 —* 2Ci/ 3 — ^ Ci.

P ig s . 2 l( a ) , (b ), ( c ) , (d) show the determination of reaction order on the

four m aterials at 'JQO0 or 60°C. The points are numbered in experimental

sequence at 5 minute in te r v a ls , point ( l ) represents the steady decom­

p o sitio n value a t Ci ( 3OO mls/min. through H2Q2 saturator; p a r tia l

pressure = 0.88 mm Hg) a fte r the i n i t i a l flu ctu a tio n s had s e tt le d .

I n i t i a l a c t iv it y changes.

The i n i t i a l e ffic ie n c y changes on f i r s t exposure of these ca ta ly sts

to vapour are shown in Pig. 22. The general pattern of the decom-
D E C O M PO SIT IO N

36
PERCENTAGE

30 40

INITIAL. A C T W IT y VARIATIONS CLASS 0 )

META l / M E T / M . O X ID ES
O m */m *o; ® C o /C o O ; ® c m / c m 2o ,

F ig . 2 d
 - 82 -

p o sitio n -was sim ila r w ith a l l of the specimens on which a fa s t e ffic ie n c y

drop quickly recovered to a steady value in 2 0 -2 5 minutes depending on the

temperature of the run. The higher temperatures generally a ccelera tin g the

attainment o f the steady decomposition value.

Temperature c o e f f ic ie n t .

The a c tiv a tio n energies obtained from the temperature c o e ffic ie n ts

in the range 38 - 100°C are summarised in Table 6 below.

TABLE 6. Experimental a c tiv a tio n energies for c la ss ( l ) metal/metal

oxides. Ih st temperature changes 38~100°C. P a rtia l pressure

^2^2 ■ 0.88 mm Hg.

Metal Probable Temperature Decomposition A ctivation

Surface Oxide Change (°C ) Change ($) Energy
(K .cals/m ol.).

Co (1 ) CoO 77 60 78.0 40.0 14.5

60 100 5 0 .0 9 0 .0 12.3

Co (2) CoO 77 60 72.5 55*0 14.7

60 100 5 0 .0 9 0 .0 12.3

80 58 4 0 .0 10.0 13.2

Qu Cu20 58 100 24.0 87.5 11.8

100 58 84.0 5.0 13.4

*
Mn Ml 0 77 59 64.0 47.5 4 .0

59 100 60.0 84.0 4.1

100 58 85.0 42.5 4.7

Ag20 58 100 19.5 5 1 .0 4.7

100 58 49*5 1 7 .0 4 .9
**
58 100 49.0 8 6 .5 4*6
 - 83 -

■ft
Contained Qc/o copper.

Etched.

I t should he noted that the temperature change from 100~38°C, could not

be accomplished w ith cobalt because at 100GC the decomposition was 100^6

with t h is oxide.

Again, i t i s evident that the surface prepared at the higher

temperatures i s a sso cia ted w ith a higher a c tiv a tio n energy than that

prepared a t 38°C. As can be seen from the figu res with two separate

cobalt p la te s , the values are h igh ly reproducible.

I t i s a ls o in te r e stin g to note that w h ilst highly etching the surface

o f the Ag/AggO p late gave a decomposition o f peroxide vapour at 38°C which,

formerly was reg iste re d a t 100°C; no appreciable a lte r a tio n was apparent

in the temperature c o e f fic ie n t .

( i i i ) Class ( 2 ) .

The m etals examined in th is c la ss were Fe and Zh, g iv in g r is e to

surface oxides ZnO and in the case of iron FeO/Fe^O^ as has been shown by

Nelson (93) f ° r treatment i n a ir below 200°C.

C a ta ly tic e ffic ie n c y and k in e tic order of decomposition.

These m etal/m etal oxides were le s s a ctiv e than those of cla ss ( l ) .

No decomposition was observed below ’JQ°0 and the k in e tic s were therefore

examined in the temperature range 100-183°C. The i n i t i a l exponential

decay which was observed with the a ctiv e oxides was not apparent in these

c a se s. In f a c t , steady c a ta ly sis was reached almost immediately, or, at

the most, 5 minutes from the beginning of the reaction . However, with

Zn/ZnO, but not w ith Fe/FeO/FegO^, a slow poisoning of the surface was

observed so th a t, for example, at 138°C in 2 hours the decomposition f e l l

 ~ 84 »

from 55°h zero. This slow deactivation was much accelerated in the

process o f order determination on Zn/ZhO and, in f a c t , "became equally

marked under these conditions fo r the iron surface.

On reducing the i n i t i a l concentration, (h , a f a ir ly constant e ffic ie n c y

was obtained, but an in crea sin g value from Ci/3 to Ci/2 or C^ showed a

sudden complete poisoning of the surface w ith both oxides. Tables 7 and

8 illu s t r a t e t h is phenomena fo r iron and zinc resp ectiv ely .

TABLE 7 . Showing d eactivation of iro n /iro n oxide in concentration

changes (lOO°C). Samples taken every fiv e minutes.

HpOp concentration Percentage Time from s ta r t _

(mm. H g.) Decomposition of run (m ins.) emar

0.98 3 4 .4 5

0 .9 8 34.5 10
0 .9 8 34.5 65 )
) F ir st order
O.6 5 35*0 70 ) k in e tic s

0 .6 5 3 4 .7 75
0 .6 5 3 4 .6 no )
) F ir s t order
0.33 34.6 115 ) k in e tic s

0,33 34.7 120

0.33 34.7 135
0.65 14.9 140 D eactivation
O.6 5 < 2 .0 145 "
0.33 0 .0 150 Poisoned
0 .6 5 0 .0 155 "

0.98 0 .0 160 "

 - 85 -

TABLE 7 Continued, T = 138°C.

HpOp concentration Percentage Time from sta rt

Hemarks
(mm, Hg) Becomposition o f runs (min.)

0.98 77*5 5

0.98 78.0 30 )
) F ir st order
0 .6 5 78.6 35 ) k in e tic s

0.65 78.5 40

0.65 78.1 65 )
) F ir s t order
0*35 78.4 70 ) k in e tic s

0.35 78.2 95

O.6 5 5 1 .0 100 D eactivation.

O.6 5 27.4 105 it

O.6 5 12.9 110 tt

O.6 5 3*7 115 it

O.9 8 0 .0 120 Poisoned

0.98 0 .0 155 11

I t can Be seen from the above tab le that whenever the surface becomes

in a c tiv e a t a given concentration i t remains so at a change to a higher

or lower concentration. I t seems to be permanently poisoned. The iron

s lip s did not show the i n i t i a l a c t iv it y v a riation s on lowering the

concentration that was displayed by the c la ss ( l ) metal/metal oxides.

Indeed, in these experiments the steady decomposition was reached almost

immediately on making a change.

 86 -

TABUS 8. Showing d eactivation of Zn/ZnO s lip in concentration changes

(l00°C ). KLve minute samples.

HgOg concentration Percentage Time from s ta r t Bemarks

fmm, Hg) Decomposition o f ran (mins • )

0.98 27.3 5 )
)
0.98 2 7 .2 10 )
1
0.65 2 3 .0 15
)
O.6 5 20.0 20 )\
) Steady
0.33 15*9 25 )\
D eactivation.
0.33 9.7 35 )
)
O065 3.1 40 ]
O.6 5 0 .0 45 Poisoning

0*98 0 .0 50 tt

0.98 2 5 .9 Overnight )

0.98 2 5 .3 10 !
\
0*65 1 9 .7 15 i
)
0.65 1 7 .2 20 ) Steady
)
0.65 1 6 .9 25 D eactivation.

0.33 14.0 30 1
0.33 (<2.0) 35 Poisoning

O.6 5 0 .0 40 »»

0 .0 50 11
0.65
 23.
FIG.
o
Q!
3
au
<vi Ul
a
r o tfl ul
s %
P
Ul -h
in ~o
tf
ay ul
* N
* =o
2 h
o ° Z
00 J) o
CJ «0 ili
Z (D
Z j
2 4 o
I a
u
Ul Ul
2
O'
3 ul
p O'
0? o
! 22
&
G
g ?
f
a o
4
of 5
IL .
o a
J
6Ul (0
a z
a o
id H ft?

O o O o O O
oO vj> CJ

N0IXI‘30cJCV033a 3^)V-LN33^i3d
 ~ 87 -

Table 8 in d ica tes that z in c /z in c oxide deactivates on lowering the

concentration to 2C i/3, and the surface i s more rea d ily poisoned thati iron

(which a t a steady conceni::ra'k;*-on does not tend to lo se a c t iv it y ) . It

can be seen that the zinc c a ta ly st recovered to a sim ilar a c t iv it y -when

l e f t overnight in the reaction tube, but quickly "poisoned” again in a new

run when exposed to vapour.

The steady d eactivation was too fa s t to permit sa tisfa c to r y order

determinations w ith zinc but i f the rate d eterioration i s allowed fo r ,

approximately f i r s t order k in e tic s are found.

Temperature c o e f fic ie n t .

The temperature c o e ffic ie n t of the reaction rate on iron /iron oxide

was found to be in the region 16-20K.cals/m ol fo r the temperature changes

shown in F ig, 23* For z in c /z in c oxide an estim ate was made on the

assumption of the reaction being f i r s t order:-

TABLE 9. R esults of quick temperature changes on Zn/ZnO.

(p.p.HgO^ *= 0.88 mm. H g.) Freshly prepared sample.

Temperature Decomposition Change A ctivation Energy

Change (°C) <*) (K cals/m ol)

100 78 26 3 33*9
78 100 3 25 33.5

(iv ) Class ( 3 ) - C atalytic a c t iv it y and k in e tic order of decomposition.

The two m etal/m etal oxides, based on aluminium and t in , in th is c la ss

were found to be even l e s s a ctiv e than zin c /z in c oxide. D etectable a c t iv it y

towards vapour was only displayed at temperatures above 140°C.

Aluminium, which had been exposed to the a ir fo r several days, gave no

 - 88 -

decomposition a t 140°C and le s s than 2$ at 183°C. A fter tr ea tin g th is

surface in a flow ing atmosphere of hydrogen at 500°C fo r several hours,

(w ith the object of a c tiv a tin g the su rfa ce), quickly cooling and tran sferrin g

to the flow apparatus; steady decompositions were obtained, e .g . 19$ a t 100°C.

A fter sw itching o ff the peroxide stream and lea v in g the specimen overnight

in a ir , the decomposition on sta rtin g the follow in g morning had fa lle n again,

t o 8$.

During a run in vapour of 90 minutes duration, the a c t iv it y f e l l

ste a d ily from 8% to 6.5$• ^he s lip ^ s then removed and heated at 500°C in

a ir , fo r two hours. On te s tin g a t 100°C, a decomposition of about 2% was

i n i t i a l l y recorded, f a llin g to zero in ten minutes. The s lip was again

removed, treated in hydrogen as above, and quickly returned to the flow

apparatus. Immediate decomposition of 23$ was found at 100°C. This remained

p r a c tic a lly constant fo r two hours a t which time 21$ decomposition was

recorded.

Tin, unlike a ctiv a ted aluminium, became very rapidly poisoned during

the course of continuous exposure to vapour. For example, origin al

decompositions of 40-55 cl° reg istered at 140°C, f e l l to zero in the course of

80-85 minutes. With th is behaviour i t was d if f ic u lt to e sta b lish the re

order on t i n / t i n oxide, sin ce bn making a concentration change almost

immediate poisoning of the c a ta ly st occurred.

The order w ith aluminium/aluminium oxide was determined on fresh ly

a ctiv a ted surfaces and the r e s u lts are given in Table 10.
 - 89

TABLE 10. Determination of reaction order on fresh ly a ctiv a ted

Al/Al^O^ surfaces.

Temperature HgOg concentration Percentage Time from sta r t

(mm. H g.) Remarks
o f run (°C) Decomposition o f run (m ins.)

138 0.96 20.8 5

138 0.96 20.7 15 ) F ir st
tt
) order
0.32 20.7 20 ) k in e tic s
tt 0.32 21.7 25
» 0.32 20.9 30
(t 0.32 20.7 35
tt 0.32 20.7 40 Steady decom­
p o sitio n
tt 0*64 8.9 45 D eactivation
tt 0.64 8.7 55 Steady de­
composition
tt 0*32. 8.9 60
tt 0 .3 2 20.7 65 C atalyst
Recovery
tt 0.32 20.7 75
tt 0.64 7.9 80 D eactivation
tt 0.64 8.4 90 Steady de­
composition
M 0*96 6.8 95 D eactivation
N O.96 0.0 100 Poisoned
1
1 O.96 0.0 r
t
120
 - 90 ~
TABLE 10 (C ontinued)

Temperature concentration Percentage Time from sta r t

Remarks
of run (°C) (mm H g.) Decomposition of run (mins)

138 0.96 23.2 5

n 0.96 21.8 15 ) f ir s t
) order
« 0.64 22.7 20 ) K inetics
n 0.64 21.9 25
» 0.64 21.8 35 ) f ir s t
) order
M 0.32 22.1 40 ) k in e tic s
H 0.32 21.8 55
n 0.64 9.7 60 D eactivation
M 0.64 8.3 65 it

If
0.96 7.4 70 n

n 0.96 2.8 75 h

it 0.9 6 0.0 80 Poisoned

« 0.96 0.0 95 ft

181 0.88 89.6 5

it 0.88 89.1 10 ) f ir s t
) order
it 0.59 8 7 .6 15 ) k in e tic s
ft
0.59 90.1 20
ft 0.59 88.8 25
ff 0.59 87.9 30 Steady
decomposition
 ~ 91 ~

TABLE 10 (C ontinued)

Temperature H2^2 concen^ra'tion Percentage Time for sta r t o f

Bemarks
o f run (°C) (mm Hg. ) Decomposition run (mins♦)

181 0.59 87.9 40 )P irst

)order
n 0.29 88.6 45 )k in etics
tt 0.29 88.1 50
n 0.29 88.1 55 Steady
decomposition
tt 0.59 71.5 60 D eactivation
n 59.4 65 n
0.59
H tt
0.59 47.6 70
ft
0.59 47.5 75 Steady
decomposition
tt 0.88 42.3 80 D eactivation

** 0.88 59.5 85 tt

ft 0.88 59.5 95 Steady

decomposition

183 0.29 90.1 15

Past
It
0.59 66.6 20 d eactivation
tt 0*59 63.4 25 D eactivation
H
0.59 63.4 55 Steady
decomposition
tt 0.88 52.1 40 D eactivation
n 0.88 50.9 45 tt
it 0.88 50.9 55 )Steady
) de c ompos i tion
it 0.59 51.7 60 )P irst order
H
)k in e tic s
0.59 51.2 65
It
0.59 51.1 75 )Steady decom-
)p o sitio n
It 80
0.29 51.5 )P ir st order
k in e t ic s .
 ~ 92 -

TABLE 10 (C ontinued)

Temperature concentration Percentage Time fo r sta r t of

Bemarks
Decomposition run (mins. )

183 0.29 51*2 95 Steady

Decomposit­
ion.

The v ariation s in the k in e tic Behaviour o f the aluminium surfaces

i s somewhat analogous to that found with iro n /iro n oxide, in that in runs

of decreasing ^ 0 ^ concentration f i r s t order k in e tic s were observed with­

out the in terv en tio n o f i n i t i a l e ffic ie n c y increases in d ica tiv e of an

apparent zero order process* In add ition , Al/AlgO^ (activated ) showed a

marked d ea ctiv a tio n , and even poisoning at the lower temperatures, on

making f a s t in creases of concentration.

Temperature c o e f f ic ie n t .

The experimental values found fo r the a c tiv a tio n energy of the

reaction on Sn/SnQ^and Al/Al^O^ (activated ) was found By assuming f i r s t

order k in e tic s and making quick temperature changes as shown

TABLE 11. Experimental a c tiv a tio n energies - Class ( 3 ) metal/m etal oxides.

P a rtia l pressure HgOg = 0.88 mm. Hg.

Metal/Metal Oxide Temperature Decomposition A ctivation Energy

change (°C) Change {%) (K .ca ls/m o l.)

Al/AlgO^ (normai ) 183 100 < 2 .0 0 -

Al/AlgO^ (activated) 100 136 6 .0 43 18.4
136 I 83 57.0 98.5 1 8 .3
100 136 7 .0 4 2 .0 18.1
136 183 24.0 92.0 I 8 .7
183 136 9 0 .0 20.0 19.1
 - 93 -

Metal/Metal Oxide Temperature Decomposition A ctivation Energy

change ( C) change ($) (K. cals/m ol)

Sn/SnOg 138 100 55-0 3 .0 2 6 .8

tt 100 138 4 .5 56.0 26.3

The f i r s t three fig u res in Table 11 re fe r to the same specimen before

and a fte r hydrogen treatm ent. The next three were obtained w ith a

sep arately prepared, hydrogen treated s li p .

(c ) The in flu en ce o f oxygen and water vapour on metal s lip c a t a ly s is .

The e f fe c t of added water vapour and oxygen cm the reaction rate was

examined with Ee, Co, A1 (a ctiv ated ) and Hi s li p s . In a l l the experiments

the e f f e c t o f adding H^Q vapour was to decrease the decomposition. This

d eterio ra tin g a ctio n increased w ith in creasing p a r tia l pressure of water

in the gas phase.

T ig . 24(a) shows the general pattern of the rate changes induced by

adding water vapour w ith the various c a ta ly s ts , and Table 12 gives the

steady experimental decomposition values a t the various water vapour

p a r tia l p ressures. As T ig. 24(a) shows, short periods o f 10-20 minutes

were necessary fo r equilibrium to be a tta in ed . So permanent e ffe c t of

water addition was discovered.

 2
s
il
0

O
Ui
5
?
h
u

arflOJ
S
UJ <
$ 4!
a

tfl 1
? o o
CM

Ix
O '
<u
s *& Ix
0
hi (J
h
z 0 lL
0 0 CD
Oi—©J9 r- 0
O' vX
o cb**?'*
o X I cJ i r f
a iP
4 o
y i = iD >-z .

z
O' 1 o
tii 03 m | a z
«■ o o
O i n Q-
s 0
? 0
U
a „ „ ? cT
o q 0 0
w ^ % u
*- Ul
u °oj ° oj qn
O' 0
Id x x * u
u O
a .0- JX °<\j (M
fi
G Ld jcL xL. I X 0
O o a o © o

N O lX lS O d V N 0 0 3 a 3 'D V X N 3 3 ^ 3 d
 - 94 ~

TABLE 12. Influence of varying amounts o f water vapour on the reaction

r a te . P a rtia l pressure = 0.48 mm. Hg.

C atalyst Temperature of Steady Decompos­ P a rtia l pressure Steady Decom-

Bun (°C) it io n 0 0 . No H^O. HpO added (mm Hg.) p o sitio n (%)
w ith H2 0

Ni/NiO 183 3 8 .1 7.58 14.9

» ti
3 6 .8 5.05 20.9
it it
37.6 4.06 22.5
i» ii
36.9 2 .8 5 2 5 .2
it it
37.1 2 .7 2 2 7 .6

W H 37.6 1.35 34.0

M It
0.78 35.7
37.1

Co/CoO 100 54.2 7 .5 8 30.3

n K 5.05 35.4
54.5
M H 54.8 4.06 38.1
M It
54.5 2 .7 1 4 5 .4
It It
54.8 1.35 4 9 .4
M It
O.7 8 53.3
54.5
Pe/FeO/ 183 ' 31.7 7.58 10.7
tt 31.8 5.05 1 6 .0

w It 32.7 4.06 1 9 .0
II tl 32.9 2 .7 1 23.1
11 11 32.9 1.35 2 7 .4
It II
32.3 0.78 28.9
 0-8
( s e c s -1
L og , jok

01 0 OS 0-4 0-6 0-6

L oq .l0 P M z O .

V A R IA T IO N OF RATE. C O N S T A N T Ns/lTM WATER

VAPOUR PRESSURE.

0 OX 101 S C O CO&ALT SU PS (jo o * 0

O O X ID ISE S N IC K E L S L I P S (IS S O
0 OXIDISED IRON S L I P S
£) F R E S H LV A C T IV A T E D ALUM INIUM S LI P S (i32> °c)

FIG. 2 4 ( b )
 ~ 95 -
TABLE 12 (C ontinued)

C atalyst Temperature o f Steady Decompos- P a r tia l pressure Steady decom­

Run (°C) it i o n $)*Ro H^O. H^O added (mmHg) p o sitio n (%)
with H^O

A l/A l205 183 17.4 5.05 8.5

(a ctiv a ted )
It II
17.4 3.80 10.8
ft h
17.4 0.78 16.7

In P ig . 24(B) log^Q (ra te constant) of the steady decompositions in

Table 21 axe p lo tted again st log^Q (p.p.H^O), where the p a r tia l pressures

o f water vapour are those o f the to ta l pressure o f water vapour present in

the system; found By adding the p a r tia l pressure of the water vapour

a r isin g from the decomposition to the values in column 4 of the above

ta b le . A s e r ie s of straight* p a r a lle l lin e s could be drawn of gradient

-0*5 with some d ev ia tio n , e s p e c ia lly fo r the more a ctiv e surfaces at the

lower p.HgO. This suggests that the decomposition rate varies approx­

im ately in v erse ly with the square root of the water vapour pressure,

i . e . Bate / - '"at * ( pH2 °) ”°*5

The experimental set-up did not allow fo r the evaluation of the rate

constant with varying q u a n tities of oxygen in the gas stream. The nitrogen

ca rrier was replaced t o t a lly by pure oxygen. Then, by determining the

ra te w ith oxygen and th at with nitrogen (plus oxygen produced in the

decom position), an in d ica tio n of the ro le of oxygen was obtained. In a l l

cases the decomposition was accelerated by oxygen. This procedure was

carried out fo r each specimen at two d ifferen t temperatures. The pattern

o f the a ccelera tio n i s shown in Fig. 2 5 ( a )., w h ilst the steady decomposition
 in
OJ

o UJ
Ll

OJ

ft'
y
o/
& r
4
0
s~ \
3z
a MS id
u 0 v!J
s
£>
h X
Or a
j zo
z (ft
o H
I- ty
u °u

rO < ■" £ ? UJ
uf £ a _
O' O' '
z a j
4 4!
0 U
j
z o
ui z y2
u u mj
0 j (T o
/ y 5
y >
x£
X
o
u p
u O2
u 4r
o 0 v9
u u) 5 *<i
u tP 31 <0~Oj
—
Z o in
UJ in iP
a X a h
_J o Z
Li­
o 5
1 ® 0 o a
oo o

NiOiJ-V^SOciVMODaa 3 ^ V ^ N 3 ^ ij3 cl
 - 96 -

values are given in Table 13*

TABLES 15* Influence o f added oxygen on the reaction rate*

(p .p . * 0.96 mm. Hg).

Temperature of Steady $ decomposition Steady decomposit­

C atalyst
Run (°C) ( 02 ca rrier) ion (Ug ca rrier)

Hi/ffiO 183 8 3 .1 55*6

tt
139 4 8 .2 24.0

A l/A l.O , 183 5 8 .1 41.5

( a c t)2 3
»» 140 28.9 18.0

C0/C 0 O 78 88.9 6 3 .I

n 38 39*2 1 3 .2

F e /F e O /F e ^ 139 66*6 41.5

it 100 25*6 6 .9

F ig . 25(a) shows that replacing oxygen by nitrogen as the carrier

caused the c a ta ly st e ffic ie n c y to d eteriorate to i t s origin al le v e l.

(Removing the added water vapour allowed the ca ta ly st to recover - F ig.

24(a) ).

In F ig . 2 5 (b ), k i s p lo tted against lo g (p*P*02 ) . The slope

o f the lin e s fo r the a ctiv a ted Al/A^O^ c a ta ly sts i s 0 .0 8 . The reaction

rate on the other c a ta ly sts was accelerated by oxygen addition to a

s lig h t ly greater extent and the slopes of these l in e s , which are nearly

a l l p a r a lle l, l i e between 0.12 and 0 .1 3 and may be taken as l / 8 .

Hence, Rate ex. (pOg)^^

fo r the c la ss ( l ) and c la ss ( 2 ) m etal/m etal oxides.
;so3s)
>|

O
J
o
-J

D5t

Eff e c t on c k t ^ u v s t s o f r e p l a c e m e n t opn^ b y Oj
\E> THE CARRIER G A S
® C o/C oO ;
<S> Ml/NiO
® i Fe/FeO/Fe2oa
A ^CTIV^TED " A.I/A.UO.

P IG
 - 97 ~

For the m etal/metal oxides of cla sse s ( l ) and (2 ), the dependence

o f the c a ta ly s is rate on ^2 "then he represented in the

form o f the fo llow in g equation

V8
v CH2 ° 2 > 02
' « = ‘ p^O *

iV.v. v . . k : •.xx ;-.

H UX'^.-X .

- ; :'a i " ' : . X - .Y ‘ \ '

dxdd ct: ■ ; > = . * . . irx: ;:.'W t. .; -v..; , ■ .

iift.vX ~i *•■,>_*■<X . ■ ..-i -i.':*

.•.U
■ f--, .■
■......... -i ' — ■
■
■
■■
• •■
•■
■
■

?:K- as Xo Xrr,' n / XvX Xr v; ;.a v-~' ■' 1 ‘

'•:xy ':•!-I: X* ■ " -d’ “-'-a yK y:€!'X). ■ ds - - • -■ V-’.1 !--d - ■-

to :v';;K-'5. o :o o y o ro o.v‘ o y y l;o - „ • o o to oo?:o o o

 - 98 -

(3 ) C ata ly sis on pure oxide s l i p s .

The c a ta ly sis was studied on a range of hulk oxides prepared as already

described under m aterials.

These included a wide range of sin g le oxides, mixed oxides of the type

XO/YgO^ and oxides containing trace im p u rities. I t i s proposed to deal with

th ese sep a ra tely , commencing with pure sin g le oxides which, lik e the m etal/

metal oxide specimens, may he subdivided in to three a c t iv it y c la sse s; v iz s -

I. K1Q; FbO; “ ^erX a c tiv e c a ta ly sts.

2o 2n0$ CdO; CuO; ^ 2^3 *■* F a irly a ctiv e c a ta ly s ts .

3 . IflgO; * ^ 2^ 3 * In active c a ta ly s ts .

Class I .

This group comprises those oxides found to he most a c tiv e .

N ickel Oxide.

Special a tten tio n was given to th is oxide because of the in te r e stin g

d ifferen ces between the green and black preparations. I f the oxide is prepared

below 400°C i t has a black appearance and contains an excess of oxygen, which

may be as much as 16% by weight estim ated on the bulk. (This may, of course,

be d iff e r e n t ia lly d istrib u ted as between the bulk and the surface o f a given

p a r t ic le .) I f the preparation i s carried out above 600°C and quickly cooled

to room temperature, an apple-green oxide r e su lts which on analysis showed no

deviation from stoichiom etry. This i s illu s t r a te d in the follow in g r e su lts

obtained by a n a ly sis with dimethylglyoxime:-

 ~ 99 -

TABLE 14. A n alytical r e su lts for "green" and "Black" n ick el oxide.

W t.of M ckel Wt.of M ckel Wt.of oxygen Mds. M ckel Mols oxygen Ratio
Oxide (gm.) in complex in oxide in oxide in oxide M ckel to
(gm.) (gm.) oxygen

0.05940 0.04504 0.01436 77 x 10-5 89.9 x 10~5 1:1.165

0 .1 0 7 2 0 0.08126 0.02594 139 x 10~5 162 x 10“5 1 : 1 .1 6 8

0.50510 0.24000 O.O65 IO 4 0 .8x 10~4 40.8 x 10~4 1:1.000

The f i r s t ; two r e su lts refer to "Black" MO and the th ird to "green" MO.

This d e ta il was expanded further By magnetic s u s c e p tiB ility and X-ray measure-

ments and the r e su lts are given Below.

(a ) Structural in v estig a tio n of "green" and "Black" MO.

The magnetic s u s c e p tiB ility r e su lts are shown in TaBle 15*

TABLE 15 Magnetic s u s c e p t iB ilitie s of "Black" and "green" MO at 20°C.

Magnetic S u sc e p tiB ility ( c .g . s . u n its ) Extrapolated F ield

Qxide Value (in ­ Strength
10,600 oersteds 1,500 oersteds f in it e f ie ld ) Be pen dance

MO (green) 9 .6 3 x 10”6 1 0 .0 5 x 10"6 9 .6 3 x 10“6 Independant

cgs. u n its.

MO (green) 9.63 x 10~6 10.09 * 10~6 9.63 x 10~6

cgs. u n its . Independant

MO(black) 1 2 .2 5 x 10“6 18.78 x 10~6 11.00 x 10~6 Ferro­

cgs. u n its . magnetic.

MO (Black) 1 2 .7 0 x 10“6 18.01 x 10“6 11.20 x 10~6 Ferro­

cgs. u n its. magnetic.

The determinations were made on four d ifferen t samples and the

values given aBove are the average o f three determinations at each

 CO
KD

Co
z
*
UJ
h* Co
f- «u
A
<r
a J
x o
a -J
hi
<x
cz a
or
o
K
O
X o
a. (£
O *U1 J
a
S az ”

o
P- a:
£Q CJT 7“ - z
<r iMi
cc til
f a£
X CT
f i e l d strength fo r every sample. The value found at 1,500 Oersteds for

the "green” MO was so s lig h t ly d ifferen t from the value a t 10,600

B ersteds, that t h is la t t e r value has "been taken as correct. In any case

i f th is was a rea l d ifferen ce "between the two determinations then the

ferromagnetism was very s lig h t in comparison to that of "black" MO,

This ferromagnetic behaviour o f "black" MO seemed to in d icate the

presence of m eta llic n ick el in the oxide as has been postulated by Klemm

and Hass (107) and supported by Selwood (108); but on making three

separate X-ray exam inations,with CuK^ rad iation in a powder camera, of

sep arately packed "black" MO specimens no trace of m etallic n ick el was

d etected . The X-ray powder photograph patterns are shown in F ig. B.

op p osite.

In these experiments a 9 cm. camera was used w ith a 0.25 mm,

diameter specimen packed in a Lindemann g la ss p h ia l. The p o sitio n of

the various arcs on the photograph was measured by a tr a v e llin g microscope

to 0.001 mm. and from th is data the d values were found from the Bragg

equation:-

d/n = A /2 sin 0 .

By the usual method u sing a master graph o f a cubic c r y sta l, a=0

■* 8A°, the d values of the two oxides were found to correspond to cry sta ls

o f u n it c e l l s iz e , 4.15 “ 4#2A°.

At the widest angle o f r e fle c tio n reg istered on the powder photographs,

fo r "green” i . e . stoich iom etric M 0, cL^q = d^^ =

 - 101 -

JL
♦ Unit c e l l s iz e = 0.805 (h2+k2+ l2 )

“ 4.1854°

and fo r ""black11 NiO, d ^ = d^^ * 0.801.

Unit c e l l s iz e = 4*1624°.

The value fo r "green” NiO i s sim ila r to that found "by Pensham (ll9 )>

a = 4*181A°, and i s s lig h t ly larger than th at given in the A.S.T.M.

index fo r "nickel monoxide" i . e . 4*1724°.

The lite r a tu r e on the stru ctu ral and magnetic properties of n ick el

oxide contains many anomalies, a r isin g from the fa c t that the excess of

oxygen accommodated in the oxide i s a variable fa cto r dependant on the

method and temperature of preparation. In many cases (99) no allowance

has "been made fo r the p o s s ib ilit y of hydrate formation in the subsequent

a n a ly sis or in heat treatm ent, which has le d to the p ostu lation of

formulae such as Ui^O^ and NiOg (109)* The existen ce of such structures

however, remains doubtful (llO ). That UiO can be prepared in a sto ic h io ­

metric r a tio of n ick el to oxygen i s evident from th is work and that of

Bhatnagar and Bal ( i l l ) .

That "black" n ick el oxide i s ferromagnetic in character i s estab­

lish e d ; but on the b asis of magnetic evidence alone i t is hardly

j u s t if ia b le to assume that th is is due to the presence of m eta llic n ic k e l.

The ferromagnetism may rather be due to the high oxygen content, which

has recen tly a lso been postulated by Perakis et a l . (112), and the

s lig h t ly sm aller u n it c e ll s iz e , reported here for "black" NiO ( i . e .

H i 0 i . ^ ,_) can be accounted fo r on the basis of the oxygen excess being

 v
9
OJ

d
Ll
O
&

o
1^
ll)
0
X
o 0
v9 J
bJ
y
u
lO
o z 2
<0 £
2
UJ
w
u) o'
o I 0
A
i/> rJO
UJ I

o Z I£
4
I 00
U <p
z
0O
<© n
(V * .a
o'
hU
Zv7
hi
U «
Zu
0 00
OJ
u
0O
a
4: h
<

W OU^OdW ODBQ 3 tDVXN33^)ac)

 - 102 -

accommodated by a balanced proportion of vacant cation s it e s throughout

the l a t t i c e , afford in g the p o s s ib ilit y of la t t i c e d isten tio n ,

(b) C atalysis on n ick el oxide s l i p s .

Both the "black11 and "green" n ick el oxides were quite a ctiv e towards

HgOg vapour. The "black" oxide on f i r s t exposure to the substrate gave

a high decomposition rate but over the course of a run a t 100°C of 90

minutes duration the "black" MO changed colour to a murky green, and

tended towards the e ffic ie n c y le v e l of the stoich iom etric oxide. Because

of the ra p id ity o f th is continuous d eactivation , i t was not p ossib le to

determine the reaction order on "black" MO w ith any s a tis fa c tio n .

M g. 26 illu s t r a t e s the behaviour, sim ilar with both oxides at the steady

e ffic ie n c y valu e, on making f a s t concentration changes, to determine the

reaction order a t 100°G. The change from C i(0.88 mm Hg. p.p. H^O^) to

2/3 Ci was accompanied by an e ffic ie n c y cycle before the rate s e ttle d to

the sane steady value as i n i t i a l l y exh ibited a t Ci, but the e a r lie s t

in d ica tio n a fte r a change was of a f i r s t order process, and remained so

a fte r a sim ila r flu ctu a tio n in the change to C i/ 3 . The reverse change

to 2 Ci/ 3 gave the f i r s t order requirement, but on restorin g the

concentration to the o rig in a l Ci a smooth f a l l of a c t iv it y was recorded

which s e tt le d to a new, lower decomposition value. The e ffic ie n c y cycles

found with these s lip s had sm aller amplitudes than those of the oxidised

metal film s, and at no time did the peaks approach a decomposition value

which would have in d icated zero order behaviour. However, lik e the m etal/

metal oxides the ca ta ly st always appeared to deactivate on restorin g the

o rig in a l concentration.
(

r
<VJ

(J
u.

— (?) o
■•J
z
t
uJ
id
0 Of
0

if)
z y
o
z I- o
j i
Id (0 > z
I
h z t z *
o 4: OJ u
3

^
0
a
z o
<0
5 6*
* £
or cf)

8
I» 5i
?! u

;e
u u

K 0Z
y
o
o o o o o O
«D OJ

M 0 \ i . l co 0 d \ M 0 0 3 Q H ^ V -L M 3 D ^ i3 c i
 - 103 -

Temperature c o e ffic ie n t of decomposition on NiO*

No two specimens of the non-stoichiom etric NiO gave p r e c ise ly the

same temperature co effic ien t* This was p ossib ly due to the continuous

change on exposure to hydrogen peroxide vapour* The values found lay

in the range 4~8K.cal s/mol © The tendency was always to s e t t le a t the

value o f "green" NiO which was 7“8 K.cals/mol* The follow in g tab le

illu s t r a t e s th is fo r runs with a se r ie s of separately compacted specimens.

TABLE 16. Temperature c o e ffic ie n ts for n ick el oxide*

Temperature (°C) Decomposition (*) Temperature

C oefficien t
Oxide T T m
1 2 X1
(K .ca ls/m o l.)
T2

NiO (green) 100 38 48 7 8 .1

NiO (green) 100 38 50 9 7.5

NiO (black) 100 38 65 36 4 .5
NiO (black) 100 38 70 46 4 .1

NiO (b lack / 100 38 64 18 6*8

green)

The la s t fig u re refers to a specimen o f "black” NiO exposed to

vapour fo r 4 0 minutes before making the temperature change*

Fig* 27 shows the i n i t i a l c y c lic e ffic ie n c y changes given by "green”

NiO, and the decay in time of the black oxide at d ifferen t temperatures,

(c) PbO, Hfa2 05, Co0/Co2 05*

These oxides were found to be very a c tiv e c a ta ly sts in the decom­

p o sitio n . PbO and MngO^ being perhaps the most a ctiv e ca ta ly sts of a l l

the sin g le oxides examined in th is work.

o m p o e it io n

C — 0 - 0 — € )— € - -
Q £c

® —® — ® —®
ercentage
P

G — 0 —0 — G

o 10 50 GO
T IM E ^ M !N S.^

S h o w in g in itia l c y c lic c h a n g e s on b u li< o x i d e s a t

© M t ij O j O C o O / C o2 0 3 ; ® P b O .
FIG,. 2 8 .

100
d e c o m po sit io n

70
percentage

50

40 14-0
Tn;Nipe.R/vTuR.E. (J3c )
S h o w in g d e c o m p o sit io n s at v a r io u s t e im p e ^ a t u r e s .
OXOES A'S FKG.29.
FIG, 29
 - 104 -

A ll the oxides showed c y c lic a c t iv it y va ria tio n s for 25-35 minutes

a t a given temperature before the steady decomposition value was reached.

This i s illu s t r a te d in F ig. 28 fo r the three oxides in runs a t 38°G, w h ilst

in F ig . 29 the oxides are contrasted according to the steady decomposition

given by each one at various temperatures. I t can be seen from F ig. 29

th at even a t room temperature (20°C) the oxides were exceedingly a ctiv e

c a ta ly s ts .

Table 17 below summarises the figu res obtained for the experimental

a c tiv a tio n energies in a f a s t temperature change fo r these oxides heat

treated a t 600°C.

TABLE 17* Experimental a ctiv a tio n en ergies, cla ss ( l ) oxides. Ih st

temperature changes (p .p . ^ 0 ^ = 0.86 mm Hg.)

Oxide Temperature (°C) Decomposition (%) Experimental

m m m m A^ctivation
vM
1 2 1 2 Energy (K .cals/m ol)

Mn205 59 37 95.0 76.5 6 .9

PbO 58 38 84.0 60.0 6.5

Co0/Co205 100 38 9 0 .1 41.0 5.9

I t can be seen from the above ta b le that the a c tiv a tio n energies for

these very a c tiv e c a ta ly sts do not vary much in the temperature region

100 - 38°C.

Magnetic s u s c e p t ib ilit y .

Table 18 gives the magnetic s u s c e p tib ility values found for the

various oxid es.

 15

IZ

tit
e FIRST ORDER
F Kinetics
O
s. <o
2
y
z
o
o
3
3

0
o OS
p p h 2 o 2 C m wt. H^')
D E C O M PO SIT IO N K IN E T IC S . FlG».2>0 ( c O
- 3 7 * c . P O I N T S AT 5 M lN U T E O N T E R V K U b

15

Ul
a
H
h
t
o
r. FIRST OROER
S Kin e t ic s
¥ <o
t
a
o
3 4 5 , 10,1!
«p
J
2

3
3 3 *4- , ^
K /O, CM M
OECOMPOST'ON
B'oO - i°>°; KT ^ m:s J~rp ■ vrG
J>v a •_ c r> / s^>
 - 105 -

TABLE 18. Magnetic s u s c e p t ib ilit ie s at 20°C. Class ( l ) oxides.

Oxide Magnetic S u sc e p tib ility Extrapolated Value Type

( e .g . s . u n its ) at in f in it e f ie ld .
10,600 o e r st. 1 ,5 0 0 o e r st.

PbO Diamagnetic
—6 -6
IfagOj 9 2 .0 x 10 108.7 2i 10 91.0 x 10 Ferromagnetic

Co0/Co205 30.9 x 10“6 30.9 x 10~6 3 0 .9 x 10“6 Paramagnetic

Order o f rea ctio n . •

The reaction k in e tic s were examined on IfajgOj 37°C. The

fin d in gs are shown in F ig s. 50(a) and (h) where i t can he seen that the

process o f fa s t concentration changes from Ci — > 0 i/5 showed no

deviation from a f i r s t order r ela tio n sh ip . The reverse changes C i / 5 5*

Ci/2 — ^ Ci point to the same f i r s t order dependance on ^ 0 ^ concentration,

a s lig h t lo s s of a c t iv it y being ind icated at the restoration o f Ci. No

c y c lic changes of e ffic ie n c y were observed.

Influence of preparation temperature.

Manganese sesquioxide was a lso studied with a view to obtaining a

correlation between c a ta ly tic a c t iv it y and the temperature o f pre-treatment

in the region 500-760°C. Each sample was fir e d fo r s ix hours at the

required temperature in an atmosphere of nitrogen. The follow ing tab le

summarisesthe r e s u lts , w ith reference to the steady decomposition

value found at d iffe r e n t temperatures fo r each specimen.

 ^ 00°c O ®

Q-

A
O <2>0— ® — <2* 0 9 -f cj t Z

*2 Ut)
0 Q/

OJ 1 ■
n rv
U W
V ^
^ u ul
«4 o !a
0 -0 tP
> g a
^ u
2 o
£ *
u
a tp
uj s 3
O' UJ o
3 I-
Ip r 5
a0 hz <f
>
* u*
U O'
. Ill
sa £s i/)
r 0o ft
ft
o o
o O o G o
00
N oatSO dw O D lC l 3'SVXK33^13d
 - 106 -

TABLE 19. ^Lea^“’^:rea'^ed- d ifferen t temperatures. Steady

decompositions (p.p.BgOg * 0.86 mm. Hg).

Temperature of Steady Decomposition Yalue ($) a t T°C

Preparation (°c ) 38°C 58°C 77°C 100°C

300 65 70 80 95
400 63 70 79 94

600 85 - 100 100

760 69 75 82.5 97.5

H.B. The D values in the la s t 4 columns are not su itab le fo r EL

A
determinations fo r -which rapid temperature changes are necessary.

As the ta b le shows the optimum e ffic ie n c y was reached in the ca ta ly st

prepared at 600°C. There was l i t t l e d ifferen ce in a c t iv it y between the

300 and 400°C samples, w h ilst the sin tered J60°C sample, resembled the low

temperature preparations in a c t iv it y . F ig. 30 shows the i n i t i a l e ffic ie n c y

changes o f these oxides before a tta in in g the steady value.

Magnetic s u s c e p t ib ilit y .

Table 20 s e ts out the magnetic s u s c e p tib ility data found for the

d iffe re n t oxides

TABLE 20. Magnetic s u s c e p tib ility (20°C) o f samples.

Temperature of X (cgs u n its ) x 10 ^ Dependanee on

I n fin ite
Preparation (°c) 10,600 o e r st. 1 ,5 0 0 o erst. F ield F ield Strength

300 8 9 .0 8 8 .7 89.0 -
400 89.1 8 9 .1 8 9 .I -
92.0 IO8 .7 9 1 .0 S lig h tly
600
Ferromagnetic
760 99.0 >250 95.0 Ferromagnetic
 ifi
lii
Of
3

f
A
£
Of
lii
a
2
U1 -
1- o a
<8> ~6
10 o u
3
o

i
€> O
O'
<
>

LA
I K
4:
2 o o
0 <\J
Or £
u N Li.
h
i t i tc
0 o
a ifl
I V 2 0
UJ
At q o < O X
P a
'I 1 / r \
0-O -8H — 0 CJ
a
2 i/i
o iT>
Li <
ui J
O Li

N0Ul90dW 0D3Q 3'3Vi-N-30^iBd

 « 107

There would appear from the s u s c e p tib ility data to be an apparent

stru ctu ra l change in the 600°C and 760°C samples, which i s more pronounced

in the la t t e r .

(&) Class ( 2 ) ZnO; CdO; CuO and J e ^ Q y

These oxides although not as a ctiv e as the f i r s t c la s s , s t i l l

decomposed H^Og (vapour) f a ir ly rea d ily in the temperature range 80-183°C.

Table 21 summarises the experimental data obtained with th is s e r ie s .

TABLE 21. Steady le v e l decompositions at various temperatures* Class ( 2 )

o x id es, (p .p . = 0.86 mm. H g.)

(STEADY
Oxide Percentage Decomposition
LEVEL.

38 °c 77°c 100°C 138°G 183°C

CdO 0 2.0 3 2 .0 55*5 -

Pe205 0 0 37*5 80.1 100

CuO 0 6.0 50.1 98.0 -

ZnO 0 0 14.0 - -

As the tab le shows the oxides so not become appreciably a ctiv e u n til

100°C, lik e the metal supported oxide film s of Fe and Zn. Apart from ZnO,

the c a ta ly sts when fr e sh ly exposed to the peroxide vapour showed no c y c lic

or other v a ria tio n in a c t iv it y ; the steady value being reached almost

immediatelyo ZnO when i n i t i a l l y exposed to HgOg vapour gave decompositions

a t 38 and 77°0 which f e l l to zero in 10 minutes at the lower and 30 minutes

at the higher temperature; sim ila rly at 100°C the oxide slowly poisoned,

as shown in Pig. 31 where th is behaviour i s contrasted with that of the

other oxides in t h is group. The figu res fo r ZnO given in the above table
 108 -

were the decomposition values a fte r 30 minutes exposure to ^ 0^ vapour*

Temperature c o e f f ic ie n t .

By making f a s t temperature changes with these c a ta ly s ts , the follow in g

fig u res were obtained fo r the temperature c o e ffic ie n ts

TABLE 22* Experimental a c tiv a tio n en ergies. Class (2) Oxides.

(p .p . = 0.86 mm. H g.)

Temperature Change (°C) % Decomposition A ctivation

Energy
Oxide T
T T (K .cals/m ol)
A1 2 Tl 2

CdO 100 183 32 87.5 10.8

CdO 183 138 80*1 42.2 11 .4

138 100 80.0 25.7 22*0

ye2°3
Fe203 100 78 37.1 0 -

CuO 101 130 50.9 95.2 16.5

CuO 130 77 82*1 8 .0 18.4-

Reaction order*

The order of the decomposition was studied on a l l four oxides and

showed l i t t l e or no d eviation from f i r s t order behaviour on lowering the

concentration from Ci to Ci/2 e t c . However, on making a change C i/3 —^C±/2~~t>

Ci, a l l the c a ta ly sts deactivated markedly, a phenomena which was not

apparent with the c la ss ( l ) oxides. In the sp e cia l case of ZnO, the

c a ta ly st showed almost complete deactivation even on making the origin al

Ci to Ci/2 concentration change. These points are illu s tr a te d in F ig. 32.

( e ) Class (3)

This comprises the in a ctiv e oxides MgO and a^-Al^O^. Neither of

th ese were c a t a ly tic a lly a c tiv e at 183°C when used as the fresh ly prepared
 15

12
M l . S . T \ T R E . ^ K M mO 4 . C . Nj \O O

9
Fir st o r o e r
Kin e t ic s n

4 ,5 ,6

0
0
PART l/M- PRESSURE HgOaL^M.
D E T E R M IN A T IO N O F K I N £ T \C S O S F e 2 O g AT \0 0 ° C .
S h o w in g l o s s o f a c t iv it y o n restor\nGi c*..
P o in ts \ t 5 M isa r e in te r v a u S > F IG . ^ 2 (<x)
J n/iOO
(k M h O ^ C .

FIRST o r d e r
K i n e t i c s -j
M u s . T\TKE

o 4 -S
AS 3t(.cx >
Ph^o^
C u O .f lT IOO'
 15

c. m / i o o )

3
4

F IR S T ORDER
t i t r e ( k m k 0

K in e tic s

6
M l= > .

0
0
PAiRTl A.L- P R E ^ U R .e H ? D o ( M . NA. H p ) N
determination OF KINETICS ON CdO. AsX IS3°e . P IG 5 2 (C.)
S H O W l N G i A C T I V I T Y L jO S S O N R A I S I N G
CONCENTRATION

A <D
C. N j l o o )
04
( < m h

FIRST-
ORDER
th tre .

K IN E TIC S
6,1
M US.

0 3&

DETERIORATION OF Zn 0 IN CONCENTRATION CHANGES

A - 10O°C ; 0~tSS°C. PIG S^CpO
 - 109 -

s lip s . However, they did become s lig h t ly a c tiv e a fte r prolonged (10 hours)

heat treatment a t 900~1000°C in a mull i t e tubular furnace in a ir , ( i t is

necessary to emphasise that the oxides were fir e d sep arately in m u llite

b oats, in a furnace with a fresh lin in g in order to avoid contamination

through im purities from previous ca lcin ation s "creeping" over the o x id es).

Alumina, h ea t-trea ted in th is way, would form eC-Al^O^ (102). I t was

a c tiv e fo r a short time in decomposing a steady fra ctio n of peroxide vapour

but became in e ffe c tiv e when the concentration was raised to a higher valu e,

e ,g . 0 ,4 4 mm. to 0,88 mm. Hg. The k in e tic s could not be determined because

of th is behaviour.

The c a ta ly tic e ffic ie n c y o f r f - A l ^ , rn lik e A l / l l ^ (a ctiv a ted ),

showed an almost lin e a r d eclin e in tim e. For example, an i n i t i a l decom­

p o sitio n of 37.5^ a t 140°C (0.88 mm. Hg, p .p . E^Og) wou^^ f a l l to zero

w ithin 50 minutes. The fo llo w in g tab le illu s t r a t e s t h is .

TABLE 23. D eterioration of o£“ -^2^3’ varf ° us ^2^2 concen^ra^^ons •

Reaction Temperature p a r tia l I n itia l Time to zero

Decom­
(°c ) pressure (mm.Hg) p o sitio n decomposition $Sns)

140 0.96 33.5$ 45

183 0.25 48.9$ 30

140 1.01 35.0$ 40

100 0.96 10.0$ 35

R efractory magnesium oxide when subjected to the same heat treatm ent,

gave no d etectab le a c t iv it y u n til 183°C. At th is temperature decompositions

between 43 a^d 45% were recorded with three separate specimens, which in
 - 110 -

75 minutes had f a lle n to a value o f 38 ~ 3

The k in e tic s o f the reaction on MgO were determined at 183°C for

concentration changes from G^Cp.p.H^O^ ® 0.88 mm Hg) to Ch/2 to C^/3 ,

the reverse procedure g iv in g almost immediate in a ctiv a tio n of the c a ta ly st

as i s shown in Table 24.

TABLE 24. Determination of Reaction order on MgO, heat treated at

1000°C fo r 10 hours, at 183°C.

Eo 0o p a r tia l Percentage Time from i n i t i a l

exposure to Remarks
pressure (mm. Hg) Decomposition vapour (m ins.;

0.88 43.1 5
0.88 42.8 10 ) F irst order
) k in e tic s
0 .4 4 42.2 15 )

0.4 4 42.1 20 )
1) F ir s t order
k in e tic s
0,29 41.7 25 )

0*29 41.7 30

0.4 4 4 .3 35 D eactivation

0.4 4 0 .0 40 No decomposition

0.88 0 .0 45 n

As the tab le shows, the concentration changes Ch to Ch/2 and Ch/2

to Ch/ 3 make no a lte r a tio n to the fra ctio n o f vapour decomposed (apart

that i s , from a small e ffic ie n c y decrease, considered to be due to slow

c a ta ly st d e -a c tiv a tio n ), and the k in e tic s are f i r s t order with respect to

3^2^2 concentration in each change.

(f) Influence of oxygen and water vapour on bulk oxide c a t a ly s is .

 (£30'

50
D EC O M PO SIT IO N
PERCENTAGE

00 40

EFFECT ocr H 2 0 V A PO UR O N R E A C T I O N R A T E . 5W O W IN Q P A T T E R N
Op PATE CHANGES |3.£>. H g » O 4- S mm Mg
0 "GREEn' (JsjpH^O T 5 -3T iwnvH ^
0 (,\a!p H^O = 4- iI m.m . h j .) F iC .3 3 C ® 0
H sO A D D E D X T P O IM T C , 5 , REMOVED XT 2=v4-.

H - ■-------- --------------- -----” ]

09
0

l/l
U
Id
IP OT
'

O
0
J
0
J 0-5

09 - .............J
02 0 4. 0-6 OS
I—OGi(^ b M
. ^sOji (jvVM. H
VARIATION S p R A T E CO N STA M T N
NITHp H jO
0 'G,REEN" Nt0. (J00°c^ 0 F ^O ^ (jOO°c> f ig . 2 > a (b )
 - Ill -

The e ffe c t of added oxygen and •water vapour on the c a ta ly sis was

examined on ^e^O^, ’’green” MO and ©{-AlgO^o The influence o f water vapour

on the reaction proceeding on the f i r s t two oxides i s shown in Table 25

below.

TABLE 25. Influence o f water vapour on reaction ra te. ’’Green” MO and

c a ta ly s ts . P a rtia l pressure ^ 0 ^ « 0.48 mm Hg.

Temperature Steady Decom­ P a r tia l pressure Steady Decom­

C atalyst of Bun p o sitio n (cjo) HyO a fte r addit­ p o sitio n (fo)
C °o) Ho HgO added. ion (mm Hg). w ith added H^O

MO 100 45*7 8.13 24.9

H tf 45.2 31.4
5.37
M ft 45.2 4.37 31.5
tf ft 2.82
44.7 36.5
If ft 1.41. 42.0
45.1

100 31.8 7-94 10.8

Pe2°3
If ft 32.4 4.17 18.1
tf ft 3 2 .0 2.82 19.5
ft M 31.9 1.41 2 7 .2

The value o f pH^O given in column 4 above refers to the p a r tia l

pressure o f I^O added plus that produced in the decomposition.

On o^-AlgO^ the reaction was immediately poisoned on adding EgO vapour

o f p a r tia l pressures 5 “ 4 mii Hg, and no reaction was recorded at 183°C.

P ig . 33(a) shows the general pattern of rate changes found with water

vapour ad d itio n s, and P ig. 33(b) shows the v a ria tio n induced in the rate

constant from which i t can be seen th a t, as with the m etal/oxide c a t a ly s t s : -

 - 112 -

Hate o<. (pE^O)"^

The influ en ce of oxygen on the reaction v e lo c ity was again determined

by fin d in g the steady decomposition le v e l given by a c a ta ly st with only

the oxygen produced in the decomposition, and that with the in e r t ca rrier

gas replaced t o t a lly by oxygen. The r e s u lts are shown in Table 26.

TABLE 26. E ffe c t on steady decomposition of replacement of nitrogen

by oxygen as the carrier gas.

(p.p.H 2 02 * O.9 6 mm. Hg).

Temperature of Steady Decomposition Steady Becomposition

C atalyst
Hun (°C) (fo) N2 ca rrier. (fc) ca rrier.

ITiO 60 26.4 57.0

HiO 100 5 0 .1 80. 9

*e2°3 100 58.0 64.1

On o( - Al2 0^, the in flu en ce of oxygen was only ju st apparent, in

th a t, the decomposition a t 183°C was raised from 5 to 52-53$ on f i r s t

a d d itio n , but the d eactivation continued on the same course as that shown

in Table 2 3 . With "black*1 HiO, w h ilst ad dition of oxygen accelerated the

decomposition, i t did not e f f e c t iv e ly a lte r the speed o f the colour change

black to green, by v isu a l observation. This s t i l l took about 100-120 mins.

at 100°C to become a murky green.

E ig. 34(a) shows the in flu en ce of oxygen on the decomposition pattern,

and 34(b) the v a ria tio n o f the rate constant, in which the lin e s are drawn

to form a p a r a lle l se r ie s showing th a t:-

Hate o< (pOg) 1/ 8

 decomposition

y Gy
Pe PC ent/vs e:

40 ~ ® /
0 —- 0 — (7)VC ,

— <g>— <s>
20

40 50

“T IM E . ( M l N S . ^
I N PLU EJN CE. OP OXV Q ELN OtN D E C O M P O S IT IO N PKTTEW N.
B "G ReEN " Ni'tO ( a t \ o o “c ^ <2 ) ' g p s ^ n ” n ^ o (,g o ° c ) &) F e .2 o 3 ( j o o 0c )
P O IN T S 1 ,3 ,6 O X Y G EN C ^ R B lE R Ga/VS N V T ^O G tE M C A R R I E R G.A.S .
" 2 , 4 -5 N IT R O G E N " “ -**' O X Y G E N
F I G . 5 4 (o -')

✓'N
7
ifl
u
hi
if)

04
I ■5
l o g io p o z (m fM m g .}

EFFECT OK C A TA LY 51E3 PA TE ^O P REPLACEM ENT

OP toy A S TMEL CASRPIEP. G A S

® ‘g r e e n " KtO 0 P e 2Oj FI G. 3 4 (js>.)

 - 113 -

I t i s p ossib le that further data would y ie ld b etter values fo r t h is

gradient, but i t appears f a ir ly certain to l i e in the region l / 6 - l/lO .

I t may be concluded that the o v er-a ll reaction rate on NiO and

may be represented by a sim ila r expression as found for certain m etal/

oxid es, v i z : -
1/8
(p IL O ) (p O „ )
Bate « K. — — ----- j —
(P^O) *
 rO «rt
111
O'
a
If)
in
0
o rTi £L
x
k 0 (8 )®
in

ifl
Ui
£
0 —0 o — z
i
V_>

u)
J
GO O ■CDl cvi p

O GO­

o q o'
fi d J3
N O CL

•© o
OD

Ul If)

3SVJ-M 3D^3d
 - 114 -

( 4 ) Equimolar oxide mixtures*

In th is sectio n the study o f the vapour phase decomposition of ^

h igh ly c a ta ly tic m aterials is extended to include a d eta iled survey of certain

mixed oxides of the type XO.Y^O^. In one s e r ie s the tr iv a le n t oxide used was

MagO^ to which equimolar additions of FbO, 2n0 and MO were separately added hy

a mechanical mixing and grinding procedure* The r e su ltin g mixtures were fir e d

at con trolled temperatures in an in e r t atmosphere fo r s ix hours*

In another s e r ie s the divalent oxide was fix e d as CuO and equimolar

ad dition s of Fe^O^, <*C and were made hy the same procedure as

ou tlin ed above* The heat treatment in th is case was carried out in a ir for

s ix hours*

(a) C atalytic a c t iv it y - MbgO, s e r ie s .

The s e r ie s based on M p r o v e d to be highly a c tiv e . The temperature

fo r 50$> decomposition was as low as 30“4-0°C fo r a l l the oxides examined.

The c a ta ly tic a c t iv it y was studied in the range 38 - 100°C and did not show

a high temperature dependence. F ig. 35 illu s t r a t e s the attainment of steady

decomposition le v e ls at d iffer en t temperatures. I t can be seen that the

various ca ta ly st surfaces ex h ib it an ir r e g u la r ity in behaviour before

le v e lin g out. This behaviour on sta r tin g up was e a s ily reproducible for

each c a ta ly st surface in fre sh runs at the same concentration as the

fig u re i l l u s t r a t e s .

The influence of temperature of preparation was a lso studied fo r the

mixtures and the r e s u lts are shown in F ig s. 38(a), (b ), (c ), in which

the steady le v e l of c a ta ly sis i s p lo tted again st the temperature of the run.

This is given fo r each mixed oxide p re-fired at the temperatures shown in

 100

_L__

(% )
d e c o m p o s it io n
steady

SO 73
REACTION TEMPERATURE^/
90 > IIQ

OF A C T IV IT Y WITH T E M P E R A T U R E OP P R E T R E A T M E N T
^

P b O 'M m . O o '. O - G O O °c PREPN

0 - 4-00°C » F \ G . 3< p ( c O
tf) - 7£>0°C •!.

100
0£CO M PO <bm O M (% >
STEADY

f 70 30
T
U E & E N D AE> 2><o(oC) ; N i Q/M vi.,0- CATA uYST
O 6oo°c. (g)4 0 o ° c u C> 7 2 0 '
F IG . 3G (b )
(°'d)
d e c o m p o s it io n
stea d y

53 7Q . 30 110
R EA C T tors T E M P ( 9C )

VARIATION OF A C T I V I T Y WITH T E M P E R A T U R £
OP PR E T R E A T M E N T Z H O /M n e Oz 0 < b00°C P R E P E R X T IO IN .
® 4 0 0 °£
• 50 0°c
O 7fc>0°C _
F i g . 3c»(c)
(00
d e c o m p o s it io in
S teady

400
C a t a c v s t PREPARATION TEM PER A TU R E
PEA K ACTIVITY CU RV ES
O pb0/
£3 Z.n0/(KttgO^
MiO/ltlnjOg

FIC .37
 - 115 -

the graph legend. The fig u res show that the mixtures were a l l very

a c tiv e . The h est ca ta ly sts were the PbO/lvlr^O^ s e r ie s in which the sin g le

oxides alone are of a high e ffic ie n c y , as has been shown e a r lie r in th is

se c tio n . The mixtures with ZnO and MO appeared to be equally a ctiv e

a t the same temperature. The graphs show that the samples prepared at

600°C were the most a c tiv e c a ta ly sts in a l l three in sta n ces. The preparat­

ion s carried out at 300° C showed l i t t l e or no d ifferen ce in a c t iv it y from

that of the 400°C preparations, w h ilst the mixtures p re-fired at 720 or

760°C were, in a l l three ca ses, the le a s t e f f i c ie n t . This i s illu s tr a te d

in F ig . 37 where the steady decomposition at 100°C i s p lo tted against the

preparation temperature of the ca ta ly st for the three s e r ie s , ( i t should

be noted here that the mixtures were a l l ground to pass 300 B.S. siev e

a fte r f ir in g in m u llite boats and before compressing in to s li p s ) .

K in etic order.

The k in e tic s of the decomposition reaction were examined on the

fo llo w in g three se le c te d specimenss- MO/Mn^O^ (prepared at 400°C);

ZnO/lfcigOj (300°C) and PbO/M^O^ (600°C). The r e s u lts are illu s tr a te d

in F ig s. 3 8 (a ), (b ), ( c ) , (d). The numbered points refer to consecutive

f iv e minute runs at three d iffe r e n t (0*80, 0.40, 0 .2 7 nun. Hg) peroxide

p a r tia l pressures a t a t o ta l constant flow of 300 mis of vapour per

minute. In the course of each experiment la s tin g about two hours there

was a very s lig h t a c t iv it y lo s s w ith a l l the c a ta ly sts shown when the

in itia l value was restored at the end of each run. ALlowing for th is

f i r s t order k in e tic s were observed in a l l cases.

Temperature c o e ffi c ie n t,

The determination of the temperature c o e ffic ie n t of the e ffic ie n c y

 10

T IT ^ E . s

6
r
\<W\u Oa

PlPST ORDER
4 K I N E T I C S *v
O- OlN

3 14-

04.
M LS.

0
o 03
PART1/VU PR ES^U TC WZOz
D E - T £ R r v \I N A c r t o r s o f B E ^ C T iO r 4 O R D E L R , Z vt O /M u j Oj
C.300°C P R E P A R A T IO N ^ AT <SQ°C
Points nunvsered iN'Expe.RirvNENTAu sequence F I G . S S ( cl^
AT 3M IM U TE INiTERVAL.S
10

&
T lT R E .

h
6
KfW-O.

«
!WLJ=>. O- OI M

4 F ir s t O r d e r
K in e tic s .

o 0 4 oa
Wh ^D^ ClV\ nv h ^ >
3»^,{'cx) TEMPeRNTURe -rr 7 T 5 ° c . FIG
 10

0 - OlN . VCMvtO^. Tt TREL

<3

FiR'srr oeoee.
4 , K lN E.T]C S> .
MLS

15,19

O
0 -4 OS
P ^ R T I K L P R E S S U R E H2 0 2 IN V tv\. H o }
D E T E R M IN A T IO N O P R E A C T I O N O R O e .« . O N N ^ . / M - H p O a
( 4 0 c f PR E P A R E a t i o n } AT i O o ® c * 5 FIGr
P o in t s a t S m in u t e in t e r y a u &
fo
M L S . 0 - 0 1 N KM^Q>4_ T I T r e ,

(o

pl RST OROER
K IN E T IC S , _
4

o
os
De t e r m in a t io n o p r e a c t io n o e d e r . o n
P b O / M v ^ O ^ ( _ G )0 0 0C 3 R .E P A R A T I Q N ^ A T 3>2J~C
cy. ■v ;,-*• C, T y_v, M \, \T»Sr Irvj-rc- P IG . SSC cO
 - 116 -

was made "by making rapid temperature changes in the region 38 - 100°C.

The r e s u lts are tabulated below fo r runs a t a constant p a r tia l

pressure (0*88 mm. Hg). The stepwise changes being in the sequence 77°C

— ?> 60°C — 100°C — ^ 39°C.

TABLE 27• Temperature c o e ffic ie n ts ZQ/Wol^O^ s e r ie s .

Preparation Temperature A c tiv ity A ctivation

Oxide Energy
Temperature (°C) Change (°C) Change (c/o) (K. c a ls/ mol)

77 60 78 64 5 .4

PbO/lShgO^ 300 60 100 68 90 4.8

100 39 87 60 3.1

78 59.5 80 64 5.7
PbO/^fiigO^ 400 59.5 100.5 68 95 5.9

100.5 38 90 55 4.1

77 59.5 86 70 5 .4

PbO/MrigO^ 600 59.5 100 78 97 5.3

100 39 93 55 4.7

77 60 72 64 3.2

300 60 100 70 90 4.3

100 38 84 58 3 .0

76 60 74 61 4.1

HiO/Mn^ 400 60 100 67 93 5.1

100 39 89 58 3.8
 - 117 -

TABLE 27. (Continued).

Preparation Temperature A c tiv ity A ctivation

Oxide Energy
Temperature (°C) Change (°G) Change ($) (K. ca ls/m o l)

76 60 80 69 5 .0

M0/lfci205 600 60 100 ‘ 72 95 4 .5

100 38 90 57 4 .0

78 60 69 56 5.1
Ui0/Mn205 720 60 100 61 85 4 .4

100 38 80 42 4 .5

78 58 83 71 4 .2

ZnO/kr^O^ 600 58 100 76 98 5.9

100 39 94 63 # 4 .0

78 59 69 51 5.9
Za0/W203 760 59 100 55 78 3.7

100 38 74 43 3 .4

77 60 .83 69 4 .4

CdO/l&igO^ 600 60 100 73 97 6.2

100 38 94 62 4 .0

The d ifferen ces in the values are often not s ig n ific a n t since

they l i e -within the range + 1 K .eal -which is a l l that can he claimed for

the rep ro d u cib ility o f th is value* There i s , however, a system atic

tendency fo r the value obtained from 100°C to be lower than the oth ers.
 - 118 -

This i s in opposition to what was normally found with pure oxides.

In view of the closeness of these three valu es, averages have been

taken and are recorded in Table 2 7 (a ).

TABLE 2 7 (a ). Average E^ values XO/Mn^O^ s e r ie s .

Preparation Temperature (°C) Mean A ctivation Energy

Oxide
(K.cal s/m ol).

PbO/Mn^ 300 4 .4

400 5.2
tt 600 5.1
HiO/ifagOj 300 3.5
tt 400 4 .3
n 600 4 .5
tt 720 4 .7

ZnO/lflhgO^ 300 4 .0
t» 600 4 .7
tt 760 4 .3

Cd0/Ma205 600 4 .8

I t can be seen from the above data th at the a c tiv a tio n energy seems

to increase as the c a ta ly st a c t iv it y decreases. The temperature

c o e ffic ie n t increases s lig h t ly with increasing sin terin g temperature,

analogous to the behaviour of IngO^ i t s e l f , which i s of such a high

c a ta ly tic power that an equimolar addition of an in a ctiv e oxide (ZnO) did

not cause any marked d eterioration in a c t iv it y . S im ilarly, the addition

o f a c tiv e oxides (NiO, PbO) improved the a c t iv it y by only a few percent.

 - 119 -

(l>) C atalytic a c t iv it y : - CuQ s e r ie s .

This se r ie s co n sisted of the mixtures CuO/Fe^O^, CuO/AlgO^ and

CuO/Co^^. Apart from the la t t e r they were le s s a ctiv e than the

s e r ie s .

F ig s. 39 (a ), (h) and (c ) show the coarse o f the c a ta ly tic e ffic ie n c y

in making f a s t temperature changes on a l l three m aterials to obtain the

experimental a c tiv a tio n en ergies, which are given in Table 28 below*

TABLE 28. Experimental a c tiv a tio n energies OuO/l^O^ s e r ie s ,

prepared a t 600°C in a ir .

Temperature A ctiv ity A ctivation

Oxide
Change (°c) Change (%) Energy (K .cals/m ol)

Cu0,Fe205 77 40 65 20 9.1 .
CuO.Fe^O^ 40 100 25 87 7.7

CuO.AlgO^ 14-0 179 50 95 14.3

Cu0.A120^ 179 140 93 48 15.1

Cu0.Co20^ 100 39 80 45 4<>2

»t 39 77 48 75 3.7

As Table 25 shows an equimolar mixture of CuO and Feo0, was much

23
more a ctiv e than these as sin g le oxides, which had to be studied in the

100-185°C temperature range. The e ffe c t was a lso to decrease markedly the

a c tiv a tio n energy required by CuO (l5 K .cals/m ol) and Fe^O^ (20K .cals/m ol).

The most in a c tiv e mixture was CuO. Al^O^ which only showed appreciable

a c t iv it y a t temperatures > 100°C and gave a sim ilar a ctiv a tio n energy to

th at o f CuO.
 IOO 100
Jp‘o«°*o-o-o-oo I19°c

so
n *c 14-SKcausIkaoi.
NOIX V<iOd WCD 3 a

' 0"0 "O*© €*O r

<bo T"TKc^u«b/wou
I .0-0-0-0-0 -
3. Kcau^ mow 40
40

t rrO O -O -O * '1
to ° 40*c
',0

X
O 40 •bo »£o
TIME (^iN<b)
FAST "TEMPERATUJ^E. CHANGES OH CuO|pCgO^ FA ST T E M P E R A s T a R E C H M G E S OH C u O jA IjO ^

FIG 3 9 Co.) F » G .3 9 i)

100
IOO°C
7 SO -*0^>-000-0-0 ITc *
0 ^'O-O-O-O-
t
(ft 60 4 2 4 ca^ / md, - V7kcau*/m°u.
O
a 'jyO-O--0-0-0-CK7
0 48 53°c •
U
hJ
a
vO iO
<K

40 <So 120
T IM E ( mi N s )
Fa s t t e m p e r a t u r e c h a n g e s on C u O /C o ^ a ^

F IG ,. B > ^ C O
 - 120 -

(c ) Magnetic s u s c e p tib ility r e s u lt s .

The magnetic s u s c e p tib ility measurements showed a str ik in g

correspondence between X and the temperature of preparation of the se r ie s

AO/M^O^. Thus, fo r the system PbO/Mi^O^, the s u s c e p tib ility tends to

decrease with in creasin g temperature but f in a lly reaches a steady valu e,

Zd.O/Md.^0^ f a l l s to approximately the same value for the specimen sin tered

at 760°C but tends to th is le v e l suddenly a fte r a progressive increase of

X, On the other hand, the system ITiO/Mn^®^ shows preliminary changes

before 600°C i s reached and then shoots in to the zone of ferromagnetism.

The r e s u lts are given in Table 29 and are in d ica tiv e of stru ctu ral changes

during progressive sintering,,

TABLE 29* Magnetic s u s c e p t ib ilit ie s AO/Mn^O^ s e r ie s at 20°C.

(in f in it e f i e l d ) .

Mixed X x 106 ( c .g . s . u n its ).

Oxide 300°C 400° C 500°0 6oo°c 760°C

PbO/MhgO^ 73*9 7 2 .2 19.5 19.6 -

0 „ Ferro- Ferro­
NiO/MnZQ 89.6 95*2 magnetic magnetic
ZmO/Mn^Qj 79.3 80o2 83.0 104.0 35.6
*
ZnO/MhgO^ 75.1 74.5 79.2 97.4 -

* These mixtures were examined a fte r exposure to vapour for two

hours a t 150°C, (flow rate = 300 mls/min; ~ ^.88 12111 ^ g). There was

a marked tendency fo r the s u s c e p tib ility of the oxide mixture to decrease

in each case a fte r being used in the c a ta ly s is . The e ffe c t may be r ela ted

to a change in oxidation s ta te of the ca ta ly st surface through a p a r tia l

 - 121 -

reduction by the a ctio n of the hydrogen peroxide.

The s u s c e p tib ility r e su lts obtained fo r the CuO/Fe^O^ mixtu re s are

given in Table 30 found from extrapolations to l/H « 0*

TABLE 50. Ifegnetic s u s c e p t ib ilit ie s CuO/YgO^ s e r ie s at 20°C.

( in f in it e f ie ld strength)

Magnetic S u sc e p tib ility (cg s. u n its)

Oxide
Room Temperature Sintered (600°C),

CuO 2,88 11 10“6 2.85 * 10“6

22.00 x 10“6 21.64 *10~6

-6 *
CuO.Fe20^ 6 .9 3 * io ~ 6 1 7 1 .8 * 1 0

* Ferromagnetic.

(d) Structural examination of equimolar oxide m ixtures.

An in v e stig a tio n of c a ta ly sis on mixed oxides would be a s s is te d

by information on stru ctu ral p roperties of these oxid es. Accordingly,

an X-ray in v e stig a tio n was carried out on the NiO/Mn^O^ s e r ie s and these

fin d in g s are discussed below.

The tr iv a le n t oxides Alo0, s Feo0, and Co00_, read ily combine with
23 23 23
CuO; 2*nO and CdO to form oxides w ith a d e fin ite sp in el la t t ic e (113)*

This l a t t i c e has been described by Greenwald, Pickart and Grannis (114-)*

I t co n sists of a cubic close-packed system of oxygen atoms with divalent

metal atoms situ a ted in tetrahedral p o sitio n s between the (llO ) planes,

and tr iv a le n t atone co-ordinated octahedrally to oxygen atoms, w ithin the

(llO ) plane between altern ate rows of oxygen atoms. The term "inverted

spinel" i s used to d istin g u ish a structure in which, in a unit c e l l , eigh t

 - 122 -

o f the divalen t ions are switched to octahedral p osition s with eight

tr iv a le n t ions in the tetrahedral in t e r s t ic e s . In addition to these

primary stru ctu res, p a r tia l inversions may occur givin g r is e to many

p o s s ib ilit ie s fo r l a t t i c e d isto r tio n .

The only reference in the lite r a tu r e to a sp in el structure of the

type A, MUgO^ was made over 50 years ago by Gorgeu (115). This author

prepared a s e r ie s of s p in e ls , w ith A = H i, Cu or 2n, by fu sing the mixed

sulphates and leaching out with a c e tic a cid . The structure when examined

by a microscope appeared to be of the cubic spinel type.

Since the spin el structure i s generally a sso cia ted with ferromagnetism

(73)> i t was f e l t that the MO/Mn^O^ sin tered mixtures would be su itab le

fo r an X-ray in v e stig a tio n , as these mixtures alone of the Mn^O^ s e r ie s

ex h ib ited ferromagnetism.

The X-ray data showed that w ith the preparations carried out below

600°C, the mixed oxide structure was inhomogeneous and con sisted of three

phases comprising the two sin g le oxides and a s o lid solu tion of NiO in

Mnn0 , . At rj6 0 °0 the structure had r a d ic a lly a lte red from that of the low

temperature preparations as can be seen from F ig. C, where the se r ie s of

powder photograph patterns are shown. I t was f e l t that the high temper­

ature preparation con sisted of a s o lid so lu tio n of NiO in Mn^O^ or Mn^O^.

No ch a ra c te r istic sp in el lin e s could be detected. However the formation

of the structure a fte r ca lcin a tio n a t 1000°C fo r some time has been

demonstrated fo r other systems by C a ste ll, Hilnot and Warrington ( ll 6 )

but was not pursued in th is work as the interm ediate mixtures are of more

s ig n ific a n t c a ta ly tic power as has been shown in the researches of

H uttig (73)* (The author i s indebted to Hr, P.J. Carter, Senior Lecturer,
 u
o
Li.
ILT

H
d
Os
UJ
a
2
ID
\~
er
Z
to 5
I ID
A. h
d z
A V)
Of- X *Q
o b o
J <*
A- i09 £
Id J'
z Id .
(A z o
>r cr
a oI
P- J
z
> XL
d O D
/ oH
X c*
Cl H
oz o?
d
uul z
>
h >
h 0
<L •2
A- 60

4
 “ 123 ~

Metallurgy Department, Royal College of Science and Technology, Glasgow,

fo r d iscu ssion on th is p oin t, and for advice on the in terp retation of the

X-ray photo gra phs)•

I t can he seen from Table 28 that a fte r sin terin g at 600°C the

CuO/fPegO^ mixture becomes ferrom agnetic. According to F o restier (117)

th is i s a true f e r r it e shown by X-ray examination to have the cubic spin el

p attern . In th is f e r r i t e , Cu2+ has approximately the same radius as Fe 24-

and there i s a p o s s ib ilit y of inversion of the p o sitio n o f the tr iv a le n t

and d ivalent io n s, so lea v in g a number of free v alen cies which w ill be

e f fe c t iv e in the c a ta ly s is . Unfortunately, time did not permit a more

exten sive study o f th is s e r ie s , and considering the in te r e stin g trend of

r e s u lts a more d eta iled examination might produce some u sefu l data.

(e) R elationship between liq u id and gas phase c a t a ly s is .

As a comparison between the gas phase decomposition and that in the

liq u id phase, q u a lita tiv e t e s t s were made with the mixed oxide se r ie s

A0/Mn2 0^. These were carried out by adding 0.05 gm. of oxide (300 B.S.

s ie v e ) to 3 m is. u n sta b ilised H.T.P. in a t e s t tubec The mode of reaction

was c a re fu lly observed, and the time of peroxide "disappearance" noted.

The r e s u lts are given in Table 31 below:-

TABLE 51. Liquid phase decomposition of 8670 w/w by AO/Ma^O^

__________ mixed oxides.___________________________________________________

Catalyst Temperature of Time of ^2^2 Remarks on mode of
preparation ( C) "disappearance" reaction .

ZnO/MttgO^ 300 38 • $ s e c s. V iolent decomposition,

11 400 35o8 n Slow for 8 s e c s ., then
v io le n t.
11 500 3 mins I n it i a l in h ib itio n ( 2 -J-mins)
then very f a s t .
^00 34.6 se c s. Immediate fa s t reaction .
M 760 ^48 hours Very slow reaction .
 - 124 -

TABLE 51. Continued

C atalyst Temperature of Time of Remarks on mode o f

preparation ( c) "disappearance” reaction

300 47*0 s e c s. Vigorous decomposition

tt 400 29.8 » Extremely vigorous.
H
600 3 0 .0 " V iolent reaction .
It
720 >48 hours I n it i a l surge, stead­
ing to slow r a te .

PbO/MhgO^ 300 21*7 s e c s . Exceedingly v io len t

it 400 2 3 .0 " tt tt

tt 500 2 5 .0 " I n it ia lly slow, ( l l s e c s )

then extremely f a s t .
n 600 c .2 0 .0 " E xplosive.

The r e s u lts show th at the 300~600°C preparations are the "best

c a ta ly sts as in the vapour phase. Generally, the mixtures pre-treated

a t 600°C were the most e f f e c t iv e , the PhO/Mo^O^ samples prepared at th is

temperature were v io le n t in the c a ta ly s is . I t may a lso he noted that the

720°C and 760°C sin te r s were r e la t iv e ly in a ctiv e as had previously been

recognised in the gas phase work, hut not to the same extent#

In general there seemed to he with most of the c a ta ly sts considered,

an i n i t i a l high e ffic ie n c y , due perhaps to surface reaction . I f th is

were s o , then the surface heat produced would he very large and the

c a ta ly s is would receive an i n i t i a l impulse to carry i t in to an ahiahatic

zone. This would not n e cessa rily he repeated at the next s ta r t , u n less

the sp e cia l rea ctiv e property of the surface was restored on standing.
 - 125 -

No example of such behaviour in the liq u id phase work was found, although

the pattern of behaviour has been e a s ily demonstrated in the case of the

vapour phase c a ta ly sis (M g. 35)*

 - 126

(5) The in flu en ce of foreign ions in the c a ta ly sis by MO.

The c a ta ly tic properties of "green” and "black" n ick el oxide s lip s have

been reported in a previous paragraph. The work described in th is section

concerns an attempt to modify the c a ta ly tic properties o f MO by su ita b le

ad d ition s of foreign ions which are well-known to modify the d efect structure

of the oxide.

Choice of foreign io n .

I t is important in choosing the foreign ion that i t should have a more

or le s s fix e d valency so that i t shows no tendency to consume or part with

electro n s (118). I t should a lso be about the same s iz e as the ion in whose

l a t t i c e i t is to be su b stitu ted . Accordingly, Li"*", with an io n ic radius of

0*60A°, and Ga^"*~, io n ic radius 0,62A°, were chosen as the ions most appropriat-
2+
in g to these conditions fo r in clu sio n in the MO l a t t i c e - i n which M has an

io n ic radius = 0,69A°. The method of preparation of these oxides has been

described under m aterials but i t may be noted here that a fte r f ir in g , the
MO/GagO^ mixtures remained green in colour, while Li^O addition resu lted in

the formation of a black oxide.

C ata ly tic a c t iv it y of "promoted" MO.

The MO + x % Li^O se r ie s proved to be very e ffe c tiv e in the c a ta ly s is .

Unlike "black" n ick el oxide, these oxides gave steady decompositions at a

fix e d temperature. There was no tendency, even on long (2-3 hours) exposure

to HgOg vapour, fo r the colour to change from black to green. The se r ie s

containing GagO^ as the trace impurity were le s s a c tiv e ca ta ly sts than those

w ith Li^O in c lu sio n s, but at a given temperature gave more decomposition than

stoich iom etric MOj pre-treated in the same way.

 o

a a
*
U
W
1
2
© EZ) t
v9
CP
o
II
o SL
m XL
pj
0
rJ
1
P
IxJ
J) O'
t
a in
d)
I 0 &
© —0 o a o
©{ OJ W / a
u j
j
F i- 4"
ifl r f)
ft O
a OJ
<
z U
o o 0 o
cu (\J OJ
*? . ^ .1 J
u
o d J S» '.0
2 OX CO
4 J
Z O f ? t
u 2
U? «O cd iO
>" 0 - 0
u + +• +
z o o o
y .X •,»
u 2 Z Z
£ ® 6 ® <=>
o ui
O
OJ
SiOi-Li^OciVNO^BQ BDVXNBO^a
 100

&o
(V0)

<b0
decomposition

40
steady

40 100 120 14-0

t e m p e r a t u r e (°c )

VARIATION IN ACTIVITY OP NICKEuOXIDEWITHDIFFERENT

amounts of Lt^O

O Stoichiometric NtO □ N j.O > \ -2 mol % Lj o

A Ato + 0 024 mol % L^O 0 •• -+ 2 4 " " ”
C)N{0 + 0- 24 mol % Ul2o 0 „ +II.5 " " "

F IG , 41
 - 127 -

The i n i t i a l e ffic ie n c y v a ria tion s on exposing the c a ta ly sts to

vapour follow ed a sim ilar course on hoth s e r ie s . Unlike the a ctiv e sin g le

oxid es, (PbO, MO e t c .) r e p e titiv e c y c lic e ffic ie n c y variation s were not

recorded, and in almost a l l in stan ces the a c t iv it y recovered to a steady value

a fte r an i n i t i a l steady d eclin e. This i s illu s t r a te d fo r various runs in

P ig . 4-0* which shows that in 20-25 minutes the rate had s e ttle d to a steady

valu e.

Decomposition determinations were nade a t various temperatures in the

region 38 - 158°C fo r a l l the ca ta ly sts in these s e r ie s , and in P igs. 41 and

42 the steady decomposition values are p lo tted against the temperature at which

they occuy, fo r the Li^O and Ga^O^ s e r ie s resp e c tiv e ly . As the figu res show,

the n ick el oxides containing 0.01^ ^a^O^ 0.01/t> hy weight gave the same
4-
a c t iv it y curve as that of the “green" MO. The e f fe c t of adding Li ions to

the stoich iom etric oxide was such as to markedly increase the c a ta ly tic a c t iv it y

as the Li"*" content increased. The e ffe c t of adding more than 0.004 molfo

Ga^O^ was to increase the a c t iv it y of MO s lig h t ly at each temperature, and to

remain a t th is value no matter how much Gao0T was added to the oxide. So that
* 2
the curve jo in in g up the a c t i v it i e s of the impurity s e r ie s , has been

taken a t the mean p o sitio n of a l l four p o in ts. I t can be seen from P ig. 42,

th at deviations from th is mean curve are very s lig h t .

Temperature c o e f fic ie n t .

The temperature c o e ffic ie n t of the Li"*" impurity s e r ie s was determined

by making temperature changes from 100°C to 38°8 to 60°C, and fo r the Ga^"*"

s e r ie s the changes were 100°C — ^ 60°C ----136°C. Since the decomposition

reg iste re d a t 3Q°Q with the la t t e r se r ie s was u sually n e g lig ib le . The r e su lts

o f th is in v e stig a tio n are given in tabular form below:-

 100

50
(% )
decomposition
STEA.DY

SD 10 0 120 140
TEMPERATURE (° c )

v a r i a t i o n in a c t i v i t y o r n i c k e l o x i d e w i t h d i f f e r e n t a m o u n t s
OF G a^^Oj , o - STOICHIOMETRIC M i.0 CHI — IHjO + 0 ’2 m o l %
A -N iO T -o o 4 .M o e.% G ^ o 3 0 _ m ^ 0 4 _ „

(£> - « + *04 ................ ® - •’ + 2-1 • •

PIG. 4 - a
 - 128 -

JL X-L.
TABLE 52. Influence of Li and Ga on the experimental a ctiv a tio n

energy of the reaction on MO.

Kapid temperature changes.

A ctivation Energy
Oxide Percentage Decomposition (K .cals/m ol.)
Composition Steady I n itia l Steady I n i t ia l
100°C 38°C 38 °C 60°C 100°C-38°C 38°C-60°C

MO "green" 5 0 .0 9 .0 - - 7.5 ± 0.5 -

M 0K 024 mol $ Li 20 42.5 1 2 .5 3.0 1 4 .0 7.5 ± 1 15 ± 3

" + .2 4 " $ L^O 80.0 28.3 28.5 49.2 6 .4 + 1 8 .0 + 1

« +1 .2 " " 7 0 .0 2 6 .1 2 5 .2 45 .0 6 .1 + 1 7.3 ± 1

h 4.2 . 4 « »t 57.5 43.5(60°C) - 5.5 ± 1

" + 1 1 .5 11 " 82.5 45.1 41.1 53.4 4 .8 + 0.5 6 .0 + 1

Oxide Steady I n itia l Steady I n i t ia l A ctivation Energy

Composition (K .ca ls/m o l.)
100°C 58°C 58°C 136°C 100°C-58°C 58°0-136ft

M Of.004 mol % Gao0, 6 0 .5 35.0 41.1 81.0 7 .0 + 0.5 5.7 ± 0.5

2 5
" + .0 4 mol % Ga203 75.5 45*1 (60°c) 45*5 89.9 6 .8 + 0 . 5 6 .0 + 0.5

it 4.0 .2 " " 61.5 25.1 36.7 78.5 8.5 ± 0.5 4 .6 + 1 .0

» +0.41 " " 6 2 .5 27.5 4 2 .6 88.2 9 .0 + 0.5 5.6 + 0.5

N4-2.1 •» « 6 0 .5 3 0 .0 82.0 1 2 .0 + 2 .0 8 .0 + 2 .0
17.5

The tab le shows that the e f fe c t of adding Li,?0 to MO i s to decrease

4-
s lig h t ly the a c tiv a tio n energy of the reaction as the amount of Li in crea ses.

This holds fo r an in creasin g and decreasing temperature charge. On the other

X-L
hand, Ga addition increases the a ctiv a tio n energy in a decreasing temperature

change, 100 - 58°C, hut for an increasing change the value remains r e la t iv e ly

con stan t. In comparing the two impurity e ffe c ts the decreasing temperature
 i - 10

^ --------- '1

2*4 i*G 0-2) 05 l(b Z4-

Mou^> Moi_% L t2o

EFFECT OP |MPL1C5>TV A D D IT IO N o n E .X PE.PlN \H .T sm vu.

ACTIVATION E N E R G Y OP " S R E E N '' NICKE.U 0 * 4 D E .
FIG 4 S

2-0 2-5 4 0
L O G ,10iC o M C . I W P liR .\T v ) + 3

EPPECT OF \N A P U P \T V A D D IT IO N O N E ^ F O P "(qHEEM*

L O Q K R ITH M IC P L .O T
O ’g ^ e e n ' n ^o
0 \O O IT IO N = >

(* Lt?0 KOQI“T\ONS
F IG .4 4
 - 129 “

changes are used since they refer in hoth cases to surfaces brought to a

steady s ta te at 100°C.

The e ffe c t of the Li^O and GagO^ trace additions on the experimental

a ctiv a tio n energy i s shown in F ig. 43 in which the v ariation of the la t t e r i s

p lo tte d again st the percentage impurity ad d ition , and in F ig. 44 where BA i s

p lo tted again st the logarithm o f the impurity concentration, ( i t may he noted

that the fig u res obtained fo r the decomposition in making the fa s t temperature

changes cannot he s t r i c t l y compared w ith the figu res obtained in the e ffic ie n c y

determinations which apply to a fresh c a ta ly st f i r s t exposed to vaPour

a t a constant temperature*)

Beaotion order*

The reaction order was studied on the NiO + 2.1 mol aL Ga^O, and 2 .4
2 3
mol ajo Li^O mixtures w ith an i n i t i a l p a r tia l pressure of 0.96 mm Hg.

The two ca ta ly sts behaved sim ila r ly on making f a s t concentration decreases,

in th at the same fr a ctio n of H^O vapour was decomposed at CL, 2CL/3 and C_/39

apart from a quickly damped c y c lic v a ria tio n on making the i n i t i a l concentration

changes. The rev ersa l from CL/3 —^ 2 CL/3 —^ CL was accompanied by an

34-
e ffic ie n c y drop which was s lig h t ly more pronounced with Ga as the impurity

and which was rem iniscent of 2n0 as a c a ta ly st. This i s illu s tr a te d in F ig s.

45 aB-cL 46 fo r runs a t d iffe r e n t temperatures.

 LO

o
U.

ioO
aAj
•«-»
j
■>0
(N
J
0
£
n in
uJ ■v>
OJ y
£z ■f vo
o I
2 £
ul 2
5 JI v9
til
h CJ (V
r o
OJ 4: tf
i
u OCJ fO
w fO s° -r»
i I .j
u
0 u u OJ u
£ U OJ 1 t
k ft i t * T
OJ a J t -£P rf) rO^
y (P . _>
z iP u O u ■j (J
Li ui J U OJ
u O f in r <0
2
0 a o j" fO'
u i
i iP
0 y
y z
a a
o
o
tt a fl
o a
co OJ

N 0 lll'5 O d \N O 3 3a 3"DV,LN3t>di3d
 o
4~1 V9

O
O il
cO
ifl u
tt z
ll
0
h

0 to
£ 0(VI
<
o
-o
ox
9 i/) J
^ uj o
»-
a U.
z OJ
+
o ^ o I ^Lrfl
* Id
t u a
J) * u u u
id $ oJ
OJ
1 I i i
o z a^ t t M
rO
(H ^
Q ■/ U ->i U
^ U ai U OJ

O . in is r
OJ
w _'cu m ^
1 *
d) *^-
id U t
t) tt o
O a a
J
i
a o:
/ <
o & a
001

OJ

N O lX lS O ciW JO D B a 3*D \ ^ S D & S cS

 - 130 -

(6) The e f fe c t o f highly -paramagnetic ions on c a t a ly s is .

This was undertaken to see i f the c a ta ly tic properties of an oxide could

he influenced by the trace addition of a highly paramagnetic cation . For th is

purpose 5y2°3 was ck °sen> as i t i s one °f the most paramagnetic oxides known;
Dy3+ having an e ffe c tiv e magnetic moment of 10*5 Bohr magnetons,

A 5 mol#5^ ad dition of was added to "green11 HiO and the mixture

heat treated a t 600°C, This was contrasted with a 5 mol % addition of Lao0,

to the same oxide; La3+ , e ffe c tiv e moment zero. Uo differen ce in the c a ta ly tic

behaviour of the two mixtures was d etected , both were very a ctiv e and gave

steady decompositions of 40/> 38°C and nearly JCffa at 7^°C, The temperature

c o e ffic ie n ts were 4~5 K .cals/m ol, However, these choices were perhaps

unfortunate in that the tr iv a le n t lanthanon ions are larger in diameter

2+
than Hi and would probably r e su lt in deformation, d islo ca tio n and ultim ate

co lla p se o f the HiO l a t t i c e .

Accordingly, i t m s decided to extend th is in v estig a tio n to CdO, which

O I Q
was chosen because of i t s valency s t a b ilit y . The io n ic ra d ii (Cd = 0.974 ;

0.98A0 ) are almost id e n tic a l.

The c a ta ly tic properties of CdO have already been described and i t was
be
shown to^a f a ir ly poor c a ta ly st. Ho improvement of a c t iv it y o f the c a ta ly st

was detected in adding 5 mol fo By^O^ as a trace impurity as can be seen from

-Fig* 47 > where the decomposition patterns of the two oxides are compared at

138°C, 100°C and 183°C.

 $
g
h u
z
u
z

E
\3
z
I
x
lA
<P
tP

5
o U rO
O
z £*
0 D
IP
z ovp
x
Q J
0
£
U (/■)
4-
1z o
U
1? J u
4 u
2 >
ui
O' J
£ >
u
> 7 o
J uJ
* if
a u u
a
a
U O
u y
o
h <
U w
u
Li «^P
U- U.
Ilf o
o a

M o \iiso c J V N 0 0 3 a
 - 131 -

(7) The surface area measurements.

I t may again he sta te d that the oxide s lip s -were a l l compressed under the

same pressure of 17*5 to n s/sq . in s . and i t would he expected that the surfaces

would not vary much apart from micro pores which would contribute l i t t l e to the

c a t a ly s is .

When examined m icroscopically (xlOO), the s lip s presented smooth surfaces

which did not vary from specimen to specimen. The same r e su lt was found when

the surface areas were measured hy a B.E.T. gas adsorption method -

183°C). Eleven c a ta ly sts were measured and the roughness factors found to

agree w ithin a maximum range of 2 tim es. A l l of th is roughness would n ot,

however, he a v a ila b le fo r c a ta ly s is . I t was concluded that the e a r lie r assumpt-

ion of the s im ila r ity of surface area was reasonable, and i t was not considered

necessary to extend the measurements to cover a l l the ca ta ly st specimens

examined.

I’rom the Brunauer, Emmett and T eller (74) treatment of m ultilayer

adsorption, we haves-

•p 1 + C - 1 .
v(po-p) ~ vmc ymG po

where v = volume a t N.T.P. adsorbed at pressure p.

v = n 11 n it when en tire surface i s covered by a

m °
unimolecular la y er.
m7 O\
po = Saturated vapour pressure (30.5 x 10 mm. Hg. for at -183 C).

C = exp. (E^ - E^)ET.

= Heat of adsorption of 1st la y er.

E = Latent heat of vapourisation.

JJ
 - 132 -

I f the adsorption follow s the B.E.T. treatment, then, a p lot of p /v .

(po-p) against p/po should give a straigh t lin e whose in tercep t i s l / v C and
' m
whose slope i s (C -l) Thus from the slope and in tercep t the two constants

vm and C can be evaluated! the former being the volume of the monolayer.

By p lo ttin g v again st p/po, vm can a lso be determined from the point of in fle x io n

on t h is curve.

Knowing the value of vm and using the conversion fa cto r of 57*2 sq.cm/cc.mm.

of adsorbed eth ylene, as calcu lated by Wooten and Callaway-Brown (77) from X-ray

data on the s iz e of the m olecule, the surface area of the adsorbent i s

e a s ily determined. The isotherms obtained are shown in F ig. 48 and were

g en erally of Type 2 although one example of a Type 1 or Langmuirian isotherm

was found w ith PbO, which r e s u lt i s probably not v a lid . The surface areas were

a l l found by the "point B" method, since nB,f was w ell defined, and the r e s u lts

are shown in Table 35 below.

TABLE 55* Surface areas of se le c te d ca ta ly sts as determined by the

"point B" method fo r C2H4/-183°C .

C atalyst S in terin g Volume of Mono- Surface 2 Roughness

Composition Temperature ( C) la y er (cc.mm) area (cm .) Factor

PbO 600 2o42 138.1 23

NiO/liBgOj 400 3.00 171.6 29
MgO 1000 1.98 1 1 3 .2 19
M 0P2.1 mol^b Ga20^ 600 2.77 156.6 26
HiO "green" 600 5.23 184.3 31
MhgOj 600 2.10 120.0 20

Ni04-2.4 mol % Ll^O 600 2.79 165.8 28

 (0

g
Ll

C“ ®
rf)
J
T+
C
J
O 0 0
H-
(P O
f\s
2
ft?
bi
I
h
0
5 fi
4 rO
mt OO
tf J
4 a0 *
til
u iy 0
4 z -
z
a ai ®
in if)
oO v9 CJ o
' uiim' 30 ) C s ^ a o s a v 3WrnoA
 - 133 -

TABLE 55 (Continued)

C atalyst S in terin g Volume on Mono­ Surface 2 Roughness

Composition Temperature ( C) layer (cc.mm) Area (cm. ) Factor *

Pb0/Mn205 400 2*32 132.2 22

&a0 600 2.11 120.5 20

CuO.FegO^ 600 2.01 110.9 18

1000 2.89 1 6 5 .0 27
A12°3

* Roughness fa cto r to nearest whole number*

I t i s evident from the above tab le that the roughness fa cto rs are in the

region 18-31> while most l i e between 20 and 30 i . e . a factor of l / 2 in the

surface area* No attempt was made to follow change of surface area with

sin te r in g because of the procedure adopted in preparing the ca ta ly st s l i p s .

 - 134- -

(8) S ta tic system r e s u lt s .

A few blank experiments were carried out with the empty v e sse l and no

decomposition was observed at temperatures up to 180°C.

This in a c tiv ity of the g la ss surface was in marked contrast to the r e s u lts

of Giguere (3 4 ), and was confirmed ca refu lly with a l l the pyrex reaction

chambers used.

A b r ie f in v e stig a tio n was carried out with ’'black11 and "green" NiO in

the temperature range 21 - 211°C. The r e su lts w ith the stoich iom etric oxide

were e a s ily reproducible but "black" NiO showed appreciable deactivation a fte r

being in contact w ith the peroxide vapour. With "green" NiO no decompositions

were observed in th is system -until temperatures greater than 100°C w h ilst

"black" NiO was a c tiv e a t temperatures c.30°C. The follow ing table shows the

r e s u lts obtained when a fresh ca ta ly st was exposed to HgOg vapour at 43°C*

TABLE 54. Decomposition of vapour on "black" n ick el oxide at 43°G.

S ta tic system.

Time (minutes) Total Pressure P artial pressure

(mm. d i-b u tylp h th alate) (mmt di^butylphthalate• )

0 16.75 16.5

1 17.75 14.5
2 18.50 13*0

3 19.00 12.0

4 19.50 10.5

5 20.00 9.5

6 2 0 .5 0 8 .5

7 2 0 .5 0 8 .5
 - 135 -

TABLE 54. (Continued)

Time (minuted) Total pressure P a rtia l pressure

(mm* d i-b u ty lp h th alate) (mm. d i-b u tylp h th alate. )

8 21*00 7.5
1° 21*50 6.5
12 2 2 .0 0 5.5

15 22*50 4 .5

20 23.50 2.5

25 24.00 1.5
30 24.50 0.5

35 2 5 .0 0 0 .0

Determination o f reaction order.

These experiments were made with an '’equilibrium oxide”, which was reached

a fte r several hours exposure of "black” MO to peroxide vapour. A d if f ic u lt y

here i s that the e ffic ie n c y of the so ca lled "equilibrium oxide" i s dependant

on the concentration of 2^02 vapour and could vary in the course of a run.

This uncertainty cannot be avoided in a s t a t ic system.

The determination o f the order is shown in P ig s. 49 & 50 where f i r s t and

second order p io ts are made. At f i r s t sigh t there would appear to be complete

support for second order. This was sim ilar to that found by Mackenzie and

B itch ie ( 3 2 ) fo r the decomposition on quartz with sim ilar p a rtia l pressures of

Ho0o . However, fo r a f a l l in C_ n to 50rfo of the i n i t i a l value a very s a t i s -

2 2
factory agreement w ith the f i r s t order law was obtained, which accords with the

flow r e s u lts .
 &
<*
a
LL

o
OJ

In
2

g
U {a
a
o
2
J aJ
n cy
o u
JUi o
0?
ifl
iJ
o
Q
L Z
z a
u
uI
i/)
z0
t s
8 8
1 j
S S
u y
o z
iO o
o 6 o
^ 3J.V-iVH.lHd -l/jLn9 - IQ ‘WVM^ ^n2 Hd
 O
in
13
I a

u}
ui• b
<
S 01
X 01
c ui
o- d 3
gd
/T
*P 2 P
z Z Z
1 a 0
(V F
c0 in
id J z
2
F o- a
Ui
If)
(J
b
Z
* s
Z J
o a
t 01
J) ui
a0 S
u
t 0
0
u
K
u or
O IL
OvJ 1.
Ci
_ o o
( 3JLxfnvvii^Hci nxi-n«a - i a 'w'w') 2o 2 hci -^jon
 - 136 -

I t i s f e l t that these r e su lts must be explained by assuming that slowness

of mixing caused the f a l l o ff in the rate of decomposition which appeals to

suggest second order, which may a lso explain the r e su lts of Mackenzie and

R itch ie ( 3 2 ); of course, gen tle f a l l o ff in the ca ta ly st e ffic ie n c y may a lso

be a fa c to r in th is apparent tr a n sitio n .

Temperature c o e f fic ie n t .

The temperature c o e ffic ie n t of the f i r s t order rate constant fo r the react

ions a t 176°C and 108°C leads to an value of 4.6K. eals/m ol. This i s a

s lig h t ly lower value than that found in the flow system (6*5 K .cals/m ol) fo r a

sim ila r oxide and probably r e s u lts from sim ilar reasons as those outlined above
 - 137 -

DISCUSSION

In th is sectio n the experimental r e s u lts w ill he examined, and discussed

in d e t a il.

The f i r s t se ctio n i s concerned w ith the method employed in th is in v e stig a t­

io n , fo llo w in g which the various fa cto rs which may he connected to form a b asis

fo r comparison o f c a ta ly tic a c t iv it y w ill be d iscu ssed. Other general section s

deal with the reaction mechanism, the Arrhenius parameters - lead in g to the

d iscu ssion of the Compensation E ffect; and the in terp retation of the other

physical data e .g . magnetic s u s c e p tib ility and surface areas. D etailed

d iscu ssio n i s given sep arately of n ick el oxide and sp in el-typ e systems as

c a ta ly s ts ,

a) The method of in v e stig a tio n .

Flow systems as methods of in v e stig a tin g heterogeneous c a ta ly tic reactions

in the gas phase have been widely used. P recise stu d ies such as those of

Schwab and P ietsch (l2 0 ) and Palmer and Constable ( l 2 l ) who both used a c a ta ly st

concentrated a t one point in a flow ing gas estab lish ed the method. A more

elaborate example o f i t s use was Twigg's (122) very d eta iled study of the

oxidation o f ethylene on Ag by using the c a ta ly st in the form of short lengths

o f w ire. Recent examples of fundamental stu d ies based on flow systems include

Bauffe, Glang and Engell ( 5 1 ) in an in v e stig a tio n of the decomposition on

various ca ta ly sts based on n ick el oxide. Parravano ( 5 3 ) an<l Kul'kova and

Temkin (123) on the CO/C^ oxidation on MO and the water gas reaction on

FegO^, and, very recen tly Darby and Kemball (124) on the methanol decomposition

on a F isch er- Tropsch c a ta ly st.

The advantages of the flow over the s t a t ic method are numerous. By

working w ith one flow rate throughout, a l l the c a ta ly sts were examined under
 - 138 -

the same flow conditions a t a given temperature. C atalysts could a ls o he

studied w ith steady HgOg concentration fo r any length of time n ecessary, so

that slow changes in the e ffic ie n c y of the c a ta ly st, i t s e l f , could he follow ed.

This i s , of course, im possible in a s t a t ic study in which the concentration,

and hence the working rate of the c a ta ly s t, v a ries during a run. As stated

previou sly the v a lid ity of s t a t ic stu d ies depends e n tir e ly on the assumption

that any changes in the ca ta ly st are e ith e r so f a s t as to fo llo w , without

appreciable la g , the steady f a l l in substrate concentration, o r/ that they are

so slow as to leave the ca ta ly st unchanged in the course of a convenient

experiment.

As shown in the r e su lts and experimental s e c tio n s , the flow method gave

e x c e lle n t rep ro d u cib ility of r e su lts with oxide and m etal/m etal oxide c a ta ly sts

which had been c lo s e ly con trolled in th e ir preparation. This, la t t e r point

must be emphasised sin ce a c a ta ly s t's a c t iv it y i s dependant on i t s "previous

h isto r y ” and rep ro d u cib ility depends on the care taken to match the pre­

treatment o f each sample. R eproducibility obtained in the b r ie f s t a t ic in v e st­

ig a tio n was not o f a high order even a fte r allow ing a ca ta ly st an origin al

s e t t lin g down period to "condition” the surface to

P ra ctica l advantages for these experiments have been referred to already.

The need to lim it temperature r is e at a ca ta ly st on which there i s a measureable

amount o f decomposition requires that a dilu ent gas be used. This could

introduce d i f f i c u l t i e s with gas phase d iffu sio n , which are read ily avoided in

a flow system.
 - 159 -

(2) The a c t iv it y o f sin g le oxides in the c a ta ly sis of

The r e su lts have shown that the e f fic ie n c ie s of the ca ta ly sts examined

vary over a wide range, and i t i s hoped now to draw a tten tio n to rela tio n sh ip s

between known p ro p erties, chemical and p h y sica l, and c a ta ly tic e ffic ie n c y . In

simple words we seek to answer the question - "What makes a substance a good

c a ta ly st ?»

But f i r s t i t must be decided what quantitative criterio n is to be used to

designate the e ffic ie n c y o f a given c a ta ly st. Consider the comparison between

MgO - no decomposition in the present apparatus a t 100°C, and PbO which gave

100 fo decomposition a t th is temperature, and therefore n eith er can be over­

lapping. The f i r s t need here then i s c le a r ly for a sa tisfa c to r y relation sh ip

between temperature and e ffic ie n c y fo r each c a ta ly st.

In the p a st, many methods have been used for d istin gu ish in g d ifferen t

c a ta ly sts in the same reaction in order to provide a b asis of comparison. The

choice of th is c r ite r io n i s n ot, by any means, a simple matter to decide, sin ce,

i t i s doubtful i f any sin g le measurement w ill provide a l l the data required to

formulate an adequate a c t iv it y s e r ie s . Hauffe, Glang and Engell (5 l) in a

study o f the HgO decomposition on metal oxides compared the c a ta ly tic a c t i v it ie s

by the amount of decomposition each specimen gave at various temperatures. In

a somewhat analogous manner Beeck ( 1 2 5 ) used the rate constant, which in the

present work may be su b stitu ted for °jo decomposition, as a criterio n of c a ta ly st

a c t iv it y fo r the hydrogenation on Cu, Au and Ag c a ta ly sts. D e ll, Stone

and T iley (45) have a lso formulated a se r ie s for the 1^0 decomposition in which

the e ffic ie n c y of the ca ta ly st i s considered to be in dicated by "the temperature

re q u ire d to produce measurable decom position", which is a more u se fu l method in

th e KgO ^ an -^2^2 ^ecomPosi ^ i Qn because of the wider tem perature range

 - 14-0 -

(250-700°C) found in the former fo r the various surfaces.

In addition to the above, very many in v estig a to rs have used the values

o f the a c tiv a tio n energies in a heterogeneous surface reaction to in d icate

c a ta ly st a c t iv it y . For example, Kemball (126) in in v e stig a tin g the ammonia/

deuterium exchange on P t, Eh, Pd, e t c ., and Couper and Eley (127) have used

the energy of a ctiv a tio n fo r the ortho-para hydrogen conversion to compare the

e f f ic ie n c ie s of Pd/Au a llo y c a ta ly s ts , which they la te r contrasted with the

magnetic s u s c e p t ib ilit y o f these samples.

In the present experiments the c a ta ly sis was studied on a l l the oxides at

d iffe r e n t temperatures and the steady decomposition found, a fte r the i n i t i a l

flu ctu a tio n s had s e tt le d . In ad d ition , by making rapid temperature changes,

experimental a ctiv a tio n energies were obtained which were considered to be

rela ted to the E^ appropriate to a p articular configuration of the surface from

which the temperature change was made. Thus, i f temperature changes were made

e ith e r up or down a t temperatures T^, T^ and °K then three s lig h t ly

d iffe r e n t E values were found fo r each temperature which was held to be

A
in d ic a tiv e o f the changing nature of the c a ta ly st surface a t T^, and T^ K

r e sp e c tiv e ly . This i s discussed more f u lly la t e r .

A comparison of D values im plies id e n tic a l t o ta l surface areas, and sp e c ia l

heat or other pre-treatment might, and probably w i l l , a ffe c t d ifferen t oxides

quite v a rio u sly . This has been shown in the case of Ag/Ag^O where the etched

c a ta ly sts gave a higher decomposition a t the same temperature than the "normal”

c a ta ly s ts . A lso, a ctiv a tio n in hydrogen of Al/Al^O^ resu lted in a ca ta ly st

which a t 183°C gave 85% decompositioniftg whereas the normal specimen gave

le s s than 2 $ .

However, the E fa c to r should be independant of the t o ta l surface area

 - 141 -

p resen t, although i t too i s dependant on pre-treatment hut le s s so than D.

Here, i t must he sta ted that the v a ria tio n s in D with temperature of pre­

treatment in the Mel^O^/AO s e r ie s cannot he rela ted to a d ifferen ce of area since

these c a ta ly sts were a l l sin tered before pressing in to s lip s .

I t i s assumed, as w ill he discussed la t e r , that at a l l the c a ta ly sts in

th is range there are a number o f p o ssib le reaction paths common to them a l l .

Each path i s a sso cia ted with an E^ whose value i s adjustable between certain

lim it s , decided by the type and proportion of the reaction paths and i s , in f a c t ,

fix e d by the s o lid c a ta ly st - by i t s chemistry and by i t s method of preparation

and heat treatm ent. The o v era ll process at one ca ta ly st w ill exh ib it an E^

which i s ch a r a c te r istic o f the surface, but when comparing two surfaces i t i s

necessary to bear in mind that we are not n ecessa rily dealin g w ith id e n tic a l

surface reaction paths. This w ill apply even to c a ta ly sts of the same chemical

co n stitu tio n although with much greater force to d ifferen t types of c a ta ly sts

such as Alo0 , and PbO.

£ 0
Thus, relevan t to each temperature, for each c a ta ly st we have two measured

values D and E^, which refer to a given sta te of each c a ta ly s t. How, within

each o f the three oxide groups stu d ied , i t i s f a ir ly simple to compare the

e ffic ie n c y of each member by the amount of decomposition each gave a t a certain

temperature. For in sta n ce, we may compare PbO, Ag^Q e tc . through the

decomposition each gave a t , say 38°C. Sim ilarly Fe20^, CuO and CdO may be

compared by th e ir e f f ic ie n c ie s at 140°C.

A ltern a tiv ely , a comparison of a c t iv it y could be made by considering the

temperature a t which each oxide gave a d e fin ite , say 5Q?o, decomposition.

This value i f not obtained experim entally may be derived by Interpolation or

 - 142 -

extrap olation from the a c tiv a tio n energy data which, however, im plies that the

nature of the c a ta ly st surface i s unchanged at the derived temperature, and

would therefore give a d iffe r e n t T value than that y ie ld in g a steady decomposit­

ion o f 50%

A s t r i c t comparison of experimental a c tiv a tio n energies is complicated

through the in terven tion o f variable heats of adsorption factors which w ill a lte r

with the oxide surface and the temperature of th is surface. In any case, i f

d iffe r e n t reaction mechanisms were involved for the various oxides then the

experimental E^'s could not j u s t ifia b ly be compared. This would be a major

o b stacle in the a n a ly sis of the rate data and would not s t r ic t l y allow the oxides

to be compared beyond th e ir separate groups, always assuming that in each group

the c a ta ly s is mechanism vrould be the same for every member.

Bearing the above provisos in mind, i t was decided that fo r the present

reaction the b est method of comparing the ca ta ly st a c t iv it i e s would be to use

the steady decompositions found at a certain temperature and, but with le s s

emphasis on, the apparent energy of a ctiv a tio n a t th is temperature.

How, considering the c a ta ly sis r e s u lts obtained w ith the metal/meta^

oxides of c la ss ( l ) i . e . the c a ta ly sts a ctiv e at room temperature to 100°C.

At 100°C, Co/CoO gave 10C$> decomposition. The oxides of Cu, Mi and Ag gave

decompositions from 80-9$>, w h ilst Hi/HiO (including i t in th is group) gave

11.5%. The apparent energies of a c tiv a tio n in th is region are 4-5 K .cals/m ol

fo r Mn/lfa.0 and Ag/Ag^O, and 12-16 K .cals/m ol for Co/CoO, Cu/Cu^O and Hi/HiO,

w ith the E^ values of the specimens increasing in the order w ritten . In th is

in sta n ces, the E^ fig u res provide the main b asis for the follow in g se r ie s as

the D values at 100°C were in very close proxim ity:-

 - 145 -

Mn/Mn.0 Ag/lg^O C°/C°0 > Cu/Cu^O ^>

Hi/HiO.

S im ilarly fo r c la ss (2 ), 2fa/4n0 at 138° C gave a percentage decomposition

after, 40 minutes of 28$ ( f a llin g ) , and an = 35“54 K .cals/m ol w h ilst for

Fe/Fe0/Fe205 these values are 78$ and 20 K.cals/m ol r e sp e c tiv e ly . Then, in

c la ss ( 3 ) Al/AlgO^ at 183°C gave ^ 2jo decomposition, and Sn/SnO^ 65-70$ a fte r

90 minutes, a f a llin g value, and an E^ = 26 - 27 K .cals/m ol. The E^ for

Al/Al^Oj was not determined because of lack of a c t iv it y in the temperature

range stu d ied , but in the sp e c ia l case of the a ctiv a ted Al/Al^O^ s lip at 183°C

decompositions o f > 90$ were found and an apparent energy of a ctiv a tio n of

18-19 K .cals/m ol.

So th a t, combining the data found w ith a l l the m etal/m etal oxides to give

an a c t iv it y s e r ie s , we have:-

Mn/MnO > Ag/Ag^O > Co/CoO > Cu/CUgO >

Hi/HiO ^ Fe/Fe0/Fe203 > A l / A l ^ (a ctiv a ted ) >

Zn/ZhO = Sn/Sn02 > Al/AlgO^ > g la ss

On the same b a s is , fo r the c a ta ly s is occurring on the pure bulk oxides

the fo llo w in g a c t iv it y s e r ie s may be formulated:-

M n ^ > PbO > Co0/Co205 > HiO "green" >

CuO > Fe205 > CdO > ZnO = MgO >

oC -A120^ >• g la ss (pyrex).

In the above s e r ie s the f i r s t four oxides gave decompositions of 85$, 55$

 - 14-4- —

and 1C$ a t 38°C r e sp e c tiv e ly , and had a ctiv a tio n energies at a temperature drop

to th is temperature of 5*9 - 8 K. cals/m ol. The next three oxides did not give

4C$ decomposition u n til above 100°G, apart from CuO, but at 100°C the decom­

p o sitio n s were CdO (32$), (38$) a^-d CuO (50$). The a ctiv a tio n energies

fo r r is in g and f a llin g temperatures were 11, 22 and 15 K .cals/m ol for CdO,

FegO^ and CuO r e sp e c tiv e ly .

2n0, I%0 and oC - A^O^ were in e ffe c tiv e c a ta ly s ts , the la t t e r two being-

subjected to heat treatment at 1000°C before they reg istered any appreciable

decomposition, which at 183°C was between 35 and 40$. The experimental EA's

were greater than 25 K .cals/m ol and could not be determined accurately because

of the in a c tiv a tio n incurred during contact with %2®2 vaPour*

Having formulated the ca ta ly sts in to these se r ie s of descending a c t iv it y ,

we must now consider i f there e x is ts a physical or chemical b asis by which we

may endeavour to explain th is d ifferen ce in p rop erties. That i s , we must now

consider answering the question posed in the f i r s t paragraph of th is sectio n .

Considering f i r s t the b a sic ity of the oxides, i t can be seen that they are

a l l b a sic, but i f th is alone was an accurate c r ite r io n then i t would be expected

that the tr iv a le n t amphoteric oxides Al^O^ and Co20 ^ e tc . would occupy sim ilar
p o sitio n s. This i s not the case, since there is a very wide d ifferen ce in

a c t iv it y between Mn^O^, CoC^/C^O^ ando£ -A^O^. So that the proton a ttra ctin g

p roperties of these surfaces are apparently not of much sig n ifica n ce in

determining the order of t h is s e r ie s .

As to the s t a b ilit y of the oxides, the ease of reduction by hydrogen of

the oxides of Fe, Co, Hi and Cu decreases in that order, which, apart from the

p o sitio n o f iron i s somewhat analogous to the decreasing e ffic ie n c y of the

c a ta ly sts given above. At the high e ffic ie n c y end of the l i s t is placed

 - 145 -

Ag/Ag^O, which begins to decompose at 240°C, and a t 300°C i s completely stripped

o f oxygen. On the other hand, alumina, an in a ctiv e c a ta ly st, is an extremely

stab le oxide w ith a very high heat of oxidation sim ilar to SnC^. However, PbO

i s ju st as sta b le as CdO or CuO and i s 3-4 times more a c tiv e in the c a ta ly s is .

So th a t, although there e x is t s these rela tio n sh ip s between a c t iv it y and oxide

s t a b i l i t y , certain anomalies are obvious which cannot be explained on th is

b asis alon e.

Be-examining the two s e r ie s , i t can be seen that from PbO to HiO there

e x is t s a common fa cto r in the electro n ic stru ctu re, in that they are normally

a l l p-type semi-conductors in which the stoich iom etic oxide tends to adsorb

oxygen. In the re st o f these s e r ie s from CuO to SnC^ we have only n-type and

in su la to r oxid es. Apart from ZnO, the n-type oxides were b etter ca ta ly sts

than the in su la to r s. ZnO and Zn/ZnO were in a ctiv e c a ta ly sts with an apparent

E^ o f 33<>8 K .cals/m ol only becoming appreciably a ctiv e a t 138°C, and in essence

there was l i t t l e differen ce in a c t iv it y between them and EgO. Otherwise, the

s e r ie s assume regular patterns in th a t, follow in g the p-type oxides we have a

group o f n -typ e, or oxygen lo sin g semi-conductors in which the metal is

u su a lly present in excess o f stoichiom etry; and then the in su lators MgO, Alo0,
^ 5
and Sn02 > which are normally non-conductive at ordinary temperatures.

The fa c t that oxide a c t iv it y in the c a ta ly tic decomposition of

vapour may be su c c essfu lly rela ted to the semi-conducting properties of the

oxide leads to a comparison o f the above data w ith that given by D ell, Stone

and T iley (45) fo r the NgO decomposition and by Stone ( 5 6 ) for the CO/O^

rea ctio n . For the former reaction the authors show that the oxide a c t iv it y

f a l l s in the sequence p-type > in su lators > n -type, w h ilst for the C0/ 02

oxid ation , Stone has shown the se r ie s to be p-type n -typ e>in su lators.
 - 146 -

So th a t, without d iscu ssin g th is comparison further in the meantime, i t would

appear from these a c t iv it y s e r ie s that the present reaction i s rela ted more

c lo s e ly to the CO/O^ oxidation than to the 1^0 decomposition, as far as the oxide

a c t iv it i e s are concerned.

Further to these p o in ters, i t has been shown that Al/Al^O^, when used as

the normal oxide covered m etal, i s in a ctiv e in the reaction giving about 2%

decomposition a t 183°C, but when th is same oxide was subjected to treatment with

hydrogen a t 600°C fo r several hours, i t became a ctiv e givin g 40 times the decom­

p o sitio n a t the same p a r tia l pressure of The Al/Al^O^ was thus activ a ted

by th is procedure, which owing to the s t a b ilit y of Al^O^ would certa in ly not

s tr ip the oxide from the surface. I t seems probable that some oxygen may be

removed, leavin g a surface with a metal excess or n-type character. The m etal/

metal oxide was then found to be sim ilar in a c t iv it y to Fe/FeO/Fe^Oy So that

th is treatment may be considered to have increased the electro n ic conductivity

o f the surface by producing aluminium atoms which d isso cia te into aluminium ions

and an equal number of q u asi-free electrons which may then in it ia t e the reaction .

S im ila rly , h eatin g MgO and oC -Al^O^ to 1000°C fo r some tim e, may give r is e

to anionic vacancies (.116), and when the oxides are quickly cooled to room

temperature the d efects may be frozen in to the la t t ic e as has been postulated by

Nizboer (9 8 ). These oxides may then a lso be compared to the n-type conductors,

but in any case they proved to be p r a c tic a lly in e ffe c tiv e ca ta ly sts in the

temperature range studied.

I t may then be concluded that fo r the sin g le oxides studied, eith e r as the

bulk m aterial, or as a skin on a pure metal or flash ed metal support, that the

most e ffe c tiv e c a ta ly sts are those in which the metal has a tendency to pass
 - 147 -

in to a higher valency s ta t e , p -type. (in additional support of th is i t may he

noted that non-stoichiom etric MO, i . e . NiQ with a large metal d efic ie n c y , was

2-3 times more a c tiv e on f i r s t exposure to than stoichiom etric MO, which

would have a tendency for le s s metal ions to pass to the higher valency, i . e .

le s s p-type character. However, in tim e, and coincident with the strip p in g of

oxygen from the ca ta ly st surface, the ’’black’' oxide became le s s active, and tended

to reach the e ffic ie n c y le v e l of the "green" m aterial. This may interpreted as

being due to the system becoming le s s p-type with the lo s s of oxygen from the

c a ta ly st su rface, there then being few er metal ions in the M 34- sta te w ith, in

consequence, a sm aller concentration of q u asi-free p o sitiv e h o le s .)

Following the p-type oxides we have those containing a metal excess

accommodated in i n t e r s t i t i a l la t t ic e p o sitio n s, CdO, 4n0 e t c ., and la s t ly those

oxides that are commonly termed in su la to r s, IflgO, o* -Al^O^ e t c ., which only become

measurably a ctiv e at low temperatures when treated by some means which might be

assumed to make them at le a s t s u p e r fic ia lly n-type semi-conducting.

(3) The nature of the ca ta ly st surface.

The a c t iv it y s e r ie s d iscu ssion of sectio n ( 2 ) would not be complete without

referrin g to the question of the nature of the actu al s it e s on the ca ta ly st

surface which may a ct as a trig g er in in it ia t in g the decomposition.

We must of course assume that the H^Og molecule i s adsorbed in such a way

as to involve the bonding of the molecule, or that i t reacts with a reactive

sp ecies already adsorbed in such a way as to i n it ia t e a chain of reaction steps

leading to decomposition and regeneration o f the i n i t i a l reactive sp e cie s. The

nature of the a ctiv e oxides suggests that electron in teraction between molecule

and surface, or s p e c ific surface s i t e , i s the important c r ite r io n .

 - 148 -

In the more straightforward case of c a ta ly sis by metals an analogous

s itu a tio n e x is t s , and much success has attended the theoiy o f Dowden (128) who

suggested that in reaction s in volvin g chemisorption on m etals, the degree o f

completion o f the metal d band was an important fa cto r in deciding the a c t iv it y .

This th eoiy received experimental support mainly in work by Dowden and Reynolds

(17), and Couper and Eley { 1 2 7 ). Later, Trapnell (129) concluded that chemi­

sorption employed the metal d band with a l l gases (CgH^, CO e t c . ) except

oxygen; and that d-character was therefore u n lik ely to be important in c a ta ly tic

oxidation rea ctio n s, where the metal work function or electron donating property

would control the ra te.

The p o sitio n regarding chemisorption on metal oxides is more complex and does

not permit any quan titative rela tio n sh ip to be formulated without reference to a

p a rtic u la r system. N evertheless, the in tera ctio n of the adsorbed molecule with

electron s or electron d efects a sso ciated w ith the oxide l a t t ic e has come to be

reco g iise d as an important fa cto r here. Thus, Gray and Darby ( 1 3 0 ) consider the

adsorption o f oxygen on NiO to occur p r e fe r e n tia lly on metal ions in volvin g

electro n tran sfer from the ion to 0^ giving 0 or with the s e ttin g up of a

p o sitiv e h o le. In th is instance a more complete tran sfer of the electron to the

adsorbed sp ecies i s p ostu lated. In the language of semi-conductor theory the

electro n i s trapped by the surface d efect.

Y olkenshtein (79) recognised the importance of su rface d e fe c ts in determ ining

the a c t i v i t y of chem isorption s it e s and th i s theory was l a t e r extended by Boudart

( l 3 l ) and Weisz ( 1 3 2 ) in to a more general p ic tu re of the e le c tro n ic n a tu re of th e

whole su rfa c e . Taylor and Thon (133) p o s tu la te d th a t only one type of a c tiv e

cen tre was re sp o n sib le f o r a given c a ta ly tic r e a c ti.c n , and th a t th is c o n siste d of

 - 149 -

a d is s o c ia tiv e ly chemisorbed sp ecies of only one reactant of the system. The

reaction then took place through the in teraction of one adsorbed reactant and an

unadsorbed sp e c ie s; sim ilar to the mechanism outlined by Twigg (134) fo r the

c a ta ly tic oxidation of eth ylene, in which he considered chemisorption of oxygen

alone was necessary; the reaction proceeding between a chemisorbed oxygen atom

and a van der Waals adsorbed molecule of the substance being oxidised .

"Whilst the Taylor/Thon theory presents a novel approach, i t would appear to

be unwise to suggest only one type of "active centre". U ntil further evidence

i s forthcoming i t w ill be assumed that adsorption may occur at various types of

lo ca tio n s on the surface, where ra d ica ls or ra d ica l ions such as 0 or 0^ (^n

the present work) may be a sso cia ted with d ifferen t chemical forces according to

t h e ir environment. Consequently, the "active s ite s " w ill not have a uniform

a c t iv it y , even although only one sp ecies i s p r e fe r e n tia lly chemisorbed. In

ad d ition , as has been pointed out by Laidler (135) "the Taylor/Thon theory does not

take in to account the v a ria tio n o f a c tiv a tio n energy with coverage of adsorbate

which i s w ell known experim entally from the work of Eoberts ( 1 3 6 ) and of Trapnell

( 137 ).

As can be seen from the above d iscu ssion no sin g le picture of the nature of

a ca ta ly st surface, and in p articu lar of "active centres", has yet been adequately

formulated. Each theory has i t s m erits, but needs to be ca refu lly applied in the

lig h t of the reaction being studied.

The present r e su lts have shown a varia tio n of a ctiv a tio n energy with

temperature. In the case of the m etal/metal oxide c a ta ly s ts , e s p e c ia lly , the

a c tiv a tio n energy appeared to increase as the temperature o f the surface, from

which the T change was made, increased. Thus, fo r example, with Cu/Cu^O the
 - 150 ~

follow in g r e s u lts were obtained (Fig. 13)s -

T. Change \ oals/* °1')

76 - 58°0 13.2 ± 0.5

58 - 100°0 11.8 + 0.5

100- 38°c 1 3 .4 ± 0.5

and w ith Fe/Fe0/Fe20^ (F ig. 2 3 ).

T . Change \ ^ * * •/" * ■ )■

101 - 130°C 1 6 .5 ± 0 .5

130 - ioo°c 2 1 .8 ± 0.5

These r e s u lts cannot be e n tir e ly due to the interven tion of variable heats

of adsorption fa cto rs a t the d iffere n t temperatures because of the fa s t temper­

ature changes, which would have the e f f e c t of "freezing" the adsorbate coverage

at any one T. In any case sin ce the reaction i s 1st order the coverage would be

small and would probably give l i t t l e more v a ria tio n in experimental E^ than +

0*5 K .cals/m ol in the range stu d ied . So that the phenomena cannot be explained

on th is b a sis.

Since i t i s not ju st a temperature range that gives a variation in E^, as

found fo r example by Smith and Taylor (138) fo r the Hg/Dg exchange on hnO, but

a change which i s connected with the temperature at which the surface is prepared;

then i t must be that a surface property is a ffected by a conditioning process at

the various temperatures. In other words, the ca ta ly st may be v isu a lise d as

having an irregu lar d istrib u tio n of adsorption s it e s on i t s surface which w ill

vary in a c t iv it y and importance 'with the surface tem perature. We may a ls o con­

sid er the s it e s to be of varying E as has a lso been p o s tu la te d by Koginskiiand

 - 151-

Tselinskaya (139) "to explain r e s u lts for the oxidation of CO on MnO^.

This i s not to say that d ra stic a lte ra tio n s of the surface are envisaged,

since at the temperatures under consideration (38 - 183°C) io n ic d efects are not

very m obile. Father, i t i s considered that the adsorbed species may a lte r in

i t s environment a t the various temperatures, and so we may envisage an adsorption

d istrib u tio n over the surface which in i t s e n tir e ty may be located at some time

or other a t p o sitio n s approximating to the whole surface, but which at a d e fin ite

fix e d temperature w ill aggregate at certain lo c a tio n s, or a t certain "active

s it e s " , and w ill be a sso cia ted with the variable chemical forces exerted at these

s ite s .

I t would be d if f ic u lt to id e n tify p recisely the exact nature of these

"active site s" but fo r the moment we w ill consider them to be p osition s where

electro n exchange i s favourable, such as one of the numerous types of electro n ic

d efects e x istin g in an oxide l a t t i c e . Thus we may haves-

(1 ) In excess electron trapped a t the surface in various ways.

e .g . a t a vacant anion s it e - co n stitu tin g an F - cen tre, or on excess

oxygen, or impurity of higher valency metal or lower valency non-metal.

(2) An electron d eficien cy or p-hole held on a surface cation vacancy, or

a t an impurity, or excess anion at the surface, or impurity of lower

valency metal e tc .

At the f i r s t type of defect we may have an adsorbed sp ecies or a located

reactio n cycle which can use the electron . The electron i s presumed to be

"borrowed" by the adsorbing system. In th is picture the energy of a c tiv a tio n

of the process may be lim ited by the a v a ila b ilit y of the electron - by the energy

required to r a ise the electron from i t s electron trap. Generally speaking oxides

which are.n -typ e conductors, and therefore in which trapped surface electron s
 - 152 -

may predominate as d e fe cts, do not free these w ith en ergies le s s than 20-40 K.

c a ls . Thus ZnO surfaces contain F-centres from which electron s w ill only be

r e le a s e d by about 30—40 K. c a ls o f energy (7)* At th e same time the vacant anion

s i t e s which form the electron traps are ju st the kind of surface d efect which

would be rap id ly poisoned by an OH or 0 atom "lowering" i t s e l f in to the vacancy

and more firm ly or permanently trapping the electron . This poisoning i s a very

ty p ic a l featu re o f n-type c a ta ly sis e s p e c ia lly onZnO which may bea "purer"

example than Feo0 , or CuO.

2 5
At the second type of d efect we may more u su ally have a p-hole held in the

v ic in it y o f an electro n a ttra c tin g adsorption sp ecies (e .g . 0 or 0^) 9 and i t

becomes the adsorbed sp ecies which controls the reaction . During c a ta ly s is t h is

sp e cie s may be replaced by another, and the properties o f the defect may as a

r e s u lt change, but poisoning should not be so complete as with the n-type oxid es.

(4 )The decomposition mechanism and o v e r -a ll rate co n tro llin g fa c to r s.

For the c a ta ly tic process i t s e l f , i . e . assuming a constant surface, the

follow in g general rea ctio n steps can be postulated one of which may be rate

c o n tro llin g .

(1) D iffu sio n o f to the ca ta ly st surface, in the gaseous phase or through

a stagnant film at the surface.

(2) Adsorption o f on the surface.

(3) Reaction, which may be a complex process in volvin g many steps any one o f

which may be ra te determining.

(4) Desorption o f products from the ca ta ly st surface.

( 5) D iffu sion o f products in to the gas phase.

P rocesses ( l) and ( 5 ) .

D iffu sio n i s often the rate co n tro llin g fa cto r in s o lid /f lu id in terfa ce
 - 153 -

reaction s when the v e lo c ity of f lu id flow i s the fa cto r determining the o v er-a ll

rate a t which the product i s obtained. For example, the c la s s ic a l work of rJ?u,

Davis and H ottel (l4 0 ) showed that the rate of combustion of carbon spheres in

a ir was transport con trolled for certain gas flow s. Later Katz, Biberdy and

Grant ( l4 l ) found evidence o f d iffu sio n control in the c a ta ly tic oxidation of

carbon monoxide by s ilv e r permanganate. Pure Ag MnO^ in the dry s ta te was only

s lig h t ly a ctiv e towards CO but on adding water vapour the reaction was accelerated ,

and the rate constant was found to vary with temperature in a manner ty p ic a l of

a d iffu sio n controlled reaction . However, due to the complexity of th is system,

i t would seem equally p la u sib le to explain the change o f with temperature

eith e r as being due to d iffu sio n through a water vapour barrier at the surface,

the r e la tiv e humidity was 80^; or to the blocking of some a ctiv e s it e s by the

water vapour which may condense in the low temperature ( - 3 •“> 11°C) region where

the small E^’s are found. R ecently, S a tte r fie ld , Resnick and Wentworth (142)

have found evidence fo r a transport con trolled process by follow ing the v a ria tio n

in Reynolds number of a mixed E^O/E^O^ vapour system flow ing through an un­

s p e c ifie d c a ta ly st cylinder in the temperature range 230 - 500°C. The experiments

were conducted on a p ilo t plant s c a le .

D iffu sion control im plies that the reaction at the phase boundary occurs

extremely rapidly so that i t s rate does not a ffe c t the observed v e lo c ity o f the
I
rea ctio n . This would mean that the free energy of a c tiv a tio n i s low, or that the

reactin g molecular sp ecies are already in an a ctiv a ted s ta te .

In ad d ition , i f d iffu sio n is the most important rate determining fa c to r , i t

i s to be expected that the observed v e lo c ity obeys the equation for f i r s t order

k in e tic s , d iffu sio n being a f i r s t order process. F in a lly , i f transport control

 - 154 -

i s important the temperature c o e ffic ie n t should be approximately equal to the

temperature c o e ffic ie n t of d iffu sio n .

The present r esu lts show that with the chosen experimental con d ition s, up

to fo u r-fo ld v a ria tio n s in the flow rate caused no change in the rate of d estru ct­

ion of #2^2* ^ow» reac-fcion i s f i r s t order with respect to concentration,

and follow in g the treatments of Zimmerman (143) and M ddiford (144), for the

chemical reaction a t the in te r fa c e ;-

a fe o j
^ ' ~aF^- - W

where * Concentration of a t the in terface

k « Chemical rate constant*

c

For the transport process,

a (H ^ )
dt \ (C - Cx ) (ii)

C b Concentration of fa r removed from surface,

kj. = Transport process rate constant.

From Uernst theory,

D iffu sion c o e ffic ie n t of

kt Thickness of d iffu sio n la y er.

Elim inating C^ from ( i ) and ( i i ) we have ;

_ a (a ,ft,) (kt kc )
C
dt = (k^+kc)

which i s a f i r s t order equation giving the reaction rate in terms of the combined

chemical and transport rate constants. I f the rate of reaction , - d ^ O ^ / d t ,

is independant of flow rate with C constant (as has been shown in experimental
 - 155

s e c tio n ) , th en :-

kt .k
-— -—
k.+k = Constant,
t c

fo r a given temperature which means that k^. kQ and the observed rate i s

determined by the rate of the chemical process.

In ad d itio n , the temperature c o e ffic ie n ts of the reaction rate show wide

d ifferen ces in value from 4 to 54 K .cals/m ol, depending on the oxide su rface.

Thus gas-phase d iffu sio n cannot be considered to be a major factor in th is

c a ta ly tic process involving a fa s t passage of gas molecules over a ca ta ly st

su rfa ce. N everth eless, the p o s s ib ilit y cannot be e n tir e ly n eglected in the

case of the more a c tiv e oxides e s p e c ia lly the sp in el s e r ie s , with S^ fs of

4-6 K .cals/m ol, where, although the reaction v e lo c ity was again independant of

flow r a te , k may not have been so very much le s s than k , .

c t
Processes (2 ). (5) and (4 ).

The reaction involves the transference of a gas-phase molecule to a

c a ta ly st surface from whence i t i s evolved only as products. We may represent

the o v era ll process as

H2 0 2 ( g ) + M - E2 02 ( a d s . ) (1 )

pl^Og m m1

where the relevant a c t iv it ie s are given below each s p e c ie s, and

HgOg (a d s .) — products ~i~ M (2)

This i s for n o n -d isso cia tiv e adsorption. The more ration al layout would be:-

H2 °2 ^ + 2M = 2 H0 (a d s.) (5 )
m
 - 156 -

and 2H0 (a d s.) = products + 2M (4 )

In support of th is we may mention the fa c t that Og i s b elieved to d isso c ia te on

adsorption on oxides.

Between ( 5 ) and (4 ) we may interpose the step s:

HO (a d s.) + E2 02 (g) * H02 (a d s.) + HgO (5 )

mM1

HOg(ads) + E2 02 (g) - HO (a d s.) + HgO + Og (6)

which would have the e f fe c t of in creasing the decomposition. I f in a l l these

cases we can assume the t o ta l a c tiv e s it e s to be only s lig h t ly covered,

i.e . m'st, 1 weobtain, fo r ( l ) and ( 2 )

'Bate = k2m* = kjpHgOg (7 )

or ( 3 ) + (4) Bate =Jsyp B ^ (8)

but i f we allow ( 5 ) and (6) to intervene

Bate « + k ^ ” pHgOg + k ^ ’pHgOg (9)

We may a lso assume, reasonable enough, that m1' = m,,f . Then ( 3 ) ajid ( 4 ) give,

a steady s ta te , no matter how much ( 5 ) and (6) contribute to the decomposition,

i i
m »» *s mw* = k ^ 2 .p H 2 0 2 .m (1 0 )

I f the decomposition is almost wholly due to ( 3 ) and ( 4 ), (9) reduces to (8 ),

but i f the steps ( 5 ) and (6) are mainly resp on sib le, ( 9 ) i s modified by (10)

and the assumptions with regard to mn and m,M lead to

Rate = 6 k^2 pHgO^2 ( ll)

I f reaction between gaseous and adsorbed ^ O g i s a v a lid p rocess, i* e . a

combination of steps ( l ) , (2 ), ( 5 ) and (6 ), a second order dependance on pHgOg

would be expected. Since n eith er of these la t t e r situ a tio n s has ever been met

with in th is work we may rule out the processes which lead to them*

I f now we wish to allow fo r a situ a tio n in which m is not equal to 1, i*e*

coverage o f the a ctiv e surface by adsorbed species i s appreciable, i t is necessary

to apply the Hinshelwood - Langmuir treatment assuming only that in the steady

s ta te m* (and m” , mm e tc * ) are constant.

The sim plest case ( l ) and ( 2 ) i s then modified by considering ( l ) to be at

equilibrium (constant = K),and m and mf to be rela ted by m + mf = 1 * We then

haves-

KpH202
m' - l+KpHgOg (12)

k2KpH202
and, Bate (1 3 )
MpEgOg

Quite c la s s ic a lly th is reduces to ( 7 ) with the added information k^ * k^E,

a t low coverage and gives a zero order laws-

Bate = kg (14)

a t high coverage. The case o f d iss o c ia tiv e adsorption follow ed by reaction (4 )

reduces to id e n tic a l laws (with appropriately changed constants) in the lim itin g

cases* The general case i s : -

(1 5 )
(l+ jKpH^O^)
 - 158 -

whence k^ i s seen to be equal to k^K.

Intervention o f ( 5 ) and. (4 ) g ives a sim ilar expression in the general case

with the r e s tr ic tio n mu = m,M, that i s : -

-r>TT A 5 /2
Rate - 5 , 6 . * . & z Oz V ‘ (l£)
1---------1 ------ :— '

Since the lim itin g conditions here g iv e , as b efore, orders eith e r u n ity

or greater than u n ity , i t i s s t i l l necessary to s e t/r e j e c t the schemesupon which

they rest*

The general conclusions from these elementary th e o r e tic a l considerations

a r e :-

(1) A f i r s t order r e la tio n between and surface s it e s seems most lik e ly .

That i s against d isso c ia tiv e adsorption,

(2) I f d iss o c ia tiv e adsorption occurs there must be a second order reaction

between adsorbed species to control the decomposition. This seems u n lik ely

in view o f the very rapid decomposition,

( 3 ) Gas/surface reactions lik e ( 5 ) and (6) seem disfavoured.

So fa r we have not considered how the roles of 0^ in a c c e le r a t in g , and

HgO in depressing, the rates of c a ta ly sis are to be explained.

We may now consider in some greater d e ta il the nature of the stage ( 2 ) in

the rate process in terms of a given oxide - a p-type oxide such as NiO.
2+ 2-
The surface w ill co n sist of Ni and 0 ions with a two to one r a tio o f

Ni '54- ions and vacant cation s i t e s , Now, the EgQg be adsorbed by electron

tra n sfer a t a metal ion in which case the decomposition may be in it ia t e d th u s:-

ITi1-1-------------------- H202 = Hi'H^ - - H 202" (1 7 )

Hi*4® ---------------- H202“ = Ni - - - OH- + OH (18)

 - 159 -

where © rep resen ts a newly created p o sitiv e h o le. This i s sim ila r to what

would occur with adsorption of oxygen. We could a lso w rite th u s:-

Ki++--------------------= Ni'H^-OH -f OH* (1 9 )

and may perhaps consider the p o s s ib ilit y o f : -

+ H202 - Ui4+ + H2 0 2+ (2 0 )

follow ed by:-

H2°2+ + ° 2” “ ^ + H02 ^21 ^

In a study of the decomposition of aqueous BgOg ( 0 ^ la b e lle d ) by MnOg

and ^*e 2 ^ e t c . as c a ta ly s ts , Dole et a l . (145) showed that the r e s u lts obtained

were con sisten t with an 0-0 bond fracture only. There i s no reason to suspect

that th is would be d iffer en t in the gas phase, and reaction s (20) and (21) would

then appear to be u n lik e ly . Reactions ( 1 7 ) and (18) are sim ilar to ( 1 9 ) only

d iffe r in g in that electron tran sfer takes place before d isso cia tio n in (l7)>

and in ( 1 9 ) coincides with d isso c ia tio n . I t i s doubtful i f these two processes

can be adequately separated. They are both types of surface oxidation, and, as

we show below, i t i s doubtful i f th is occurs to an appreciable exten t.

In a d iffe r e n t scheme, the may adsorb on an electron egative species

2-
on the surface such as a vacant cation p o sitio n an 0 ion or pre-adsorbed

sp ecies such as 0^ > 0 or OH held in the neighbourhood of q u asi-free p o sitiv e

2-
h o le s. Decomposition at an 0 l a t t i c e s it e can be considered to be u n lik ely

e n e r g e tic a lly , and i t would appear that the in tera ctio n between a vacant cation

s i t e and a peroxide molecule would be in s u f f ic ie n t to cause bond f is s io n . Thus,

the reaction may be considered to i n i t i a t e a t an 0^ or 0 pre-adsorbed sp e c ie s;

 - 160 -

an adsorbed OH rad ical w ill have a tran sien t existen ce but may be of importance

in a chain propagation mechanism. Thus the in it ia t io n step may be represented

as:-.

02“ ( a d s . ) ------------H202 = 02 (g) + 0H~ (a d s.) + OH(a d s .) (22)

or O" ( a d s . ) ------------H202 « -§02 (g) + OH*(a d s.) + 0H(ads.) (23)

These reaction s are considered to be e n tir e ly surface ones and would

contribute l i t t l e to the c ry sta l l a t t i c e d isorder, sin ce at low temperatures

d iffu sio n in to or from the bulk w ill be s lig h t . They are of course, surface

reduction rea ctio n s.

Experiment has shown that with "black” NiQ, oxygen i s stripped from the

surface so that the ca ta ly st becomes a murky green. In step with t h is , the

ca ta ly st slow ly lo se s e ffic ie n c y . Some of the surface oxygen on the other

p-type c a ta ly sts may a lso be stripped in th is same manner during the course of

the rea ctio n . I t may perhaps interchange with peroxide oxygen, furthermore,

i t has been shown that the presence of oxygen in the gas stream increases the

reaction r a te , perhaps by adsorbing on the ca ta ly st surface, as 02~ or 20*" and

creatin g p o sitiv e holes in the neighbourhood, thus making more s it e s availab le

at which reaction s (22) and (23) may then occur. These step s, (22) and (2 3 ),

must be considered to be as important as ( 1 7 ), (18) and ( 1 9 ) sin ce in creasin g

2+
the p a r tia l pressure of oxygen over a surface containing Ni ions should be

such as to y ie ld Ni* ^ ions as w ell as n eg a tiv ely charged oxygen s i t e s .

Once the decomposition has been in itia t e d givin g OH and OH , th e ir fa te has

to be decided, and doing so formulates the propagation step s. These sp ecies are

not considered to be chem ically bound at the surface for th is would give r is e to

slow build up of hydroxides. However, th is is not to say that the in tera ctio n
 - 161 -

of OH and OH with the surface can he n eglected , rather we may regard OH-” and

OH as being in a very rea ctiv e sta te held to the surface by weak e le c tr o s ta tic

and van der Waals fo r c e s, resp e ctiv ely ; e sp e c ia lly in the case of the high

a c t iv it y p-type oxides, some of which do not form stable hydroxides, (e .g .

Cu, Ag).

-Considering f i r s t the reactions of OH we have the p o s s ib il it i e s s -

OH (a d s.) + Hi44* ^ OH-(ads.) + Hi44® (24)

for d issolved and solvated sp ecies rather than adsorbed, th is would have

A& = -lO K .cals and &S » -60 c a ls . (Evans, Hush and Uri, 14), and we may

consider the error involved in u sin g these values for the adsorption process

as being small i f i t i s considered that the so lv a tio n influ en ces are replaced

by the surface e ffe c t of the adsorption fo r c e s. This i s therefore a highly

fe a s ib le process but not such that the surface a c t iv it y of OH would be e n tir e ly

n e g lig ib le . Thus, i f the fra ctio n of the surfaces covered are used fo r a c t i v it i e s

and i f ten per cent of the surface Hi ions are present as Hi 3+ , th is value of

AG gives the r a tio of OH to OH as approximately 10**^at 25°C, but rather more

3 o
than 10 at 100 C owing to the large entropy change.

Such amounts of OHwould be quite enough to permit chain reaction step s to

be based on th is sp e c ie s.

Thus the OH may r ea c ts-

OH (a d s.) + H202 (a d s.) * fH02 (a d s.) ++ H20 ((a

(ads•)
d s.) (a)
1- 4 -
to02 (a d s.) + 0 (a d s.) (b) (2 5 )

OH (a d s.) + 02 (a d s.) = OH . (a d s.) + 02 (g) (26)

AG = -61 K. c a ls .
 - 162 -

Reaction (26) would be sim ilar to one in which electron tra n sfer takes

place from 0 (a d s.) on the c a ta ly st surface. The HO^ produced in ( 25 a) may

then r e a c tj-

H02 (a d s.) + OH" (a d s.) « 1^0 (a d s.) + 0 2” (a d s.) (2 7 )

or H02 (a d s.) + 1^02 (a d s.) = (a d s.) + HO (a d s.) +02 (g) (28)

( 25 t ) o r( 27) in volves the regeneration of 02 , and could thus explain a c y c lic

c a ta ly tic process v ia ( 2 2 ), ( 25 a) and ( 2 7 ); (28) provides the p o s s ib ilit y of a

short chain rea ctio n . Desorption of 02 i s accomplished in the reactions (22)

and ( 2 3 ), in which 02 or 0 i s replaced by OH w ith OH adsorbing on an . adjacent

s ite . In th is scheme so fa r, we depend on reaction ( 2 7 ) or on migration of H^O4"

from 25(b) fo r the removal of OH (a d s .). I t must be noted that OH* (a d s.) may

be l e s s la b ile than 02 and 0 , and the steady replacement of 02~ or 0~ by 0H~

may r e su lt in a slowing down of c a t a ly s is . Thus, th is may correspond w ith the

f a l l in the c a ta ly s is rate on f i r s t exposure, or whenever the ^ 0 ^ concentration

i s ra ise d . We may, however, consider the p o s s ib ilit y of the follow ing reaction s j

CH" (a d s.) + E2 02 = H02" (a d s.) + H20 (2 9 )

follow ed b y :-

H02“ ( a d s .) + E2 0 2 = H2OK)2 (g) + 0H~ (a d s.) (30)

Such reactions may be sa id to occur in so lu tio n a s, for example, in the

slow decomposition of concentrated EpOg by a lk a li. They may, however, require

that OH (a d s.) should be in a s p e c ia lly reactive s ta t e .

F in a lly we have to consider desorption of E^O and other p ossib le reactions

of HO and H02 • Desorption of E^Q is probably very rapid and the rea ctio n :-
 - 163 -

2 OH* = 02" + 1^0 (31)

which w ill follow the formation o f any hydroxide l a t t i c e i s r e a lly a matter

of oxide growth. The same may he sa id of the p o ssib le ste p s-

2H02" - 02" + 02 + H20 (32)

which has been shown by McFadyen (81) and N ettleton ( 9 6 ) to occur a t a rate

n e g lig ib le compared with that o f c a ta ly sis ( l to 1000 on Cu^O a t 100°C).

This i s true also f o r j -

EQ~ + HO- - <?“ + E202 (33)

l l s o , as we have seen fo r NiO (b lack ), tends to reduce the normal p-type

surface, a well-known phenomena which already had been recognised in I 863 by

Brodie (146), so that these reactions are u n lik ely to be important compared

w ith j-

ho2“ + OH” - 1^0 + 20” (34)

follow ed bys-

20" + H202 = 02 (g ) + 20H* (3 5 )

as a propagation ste p , or a t high temperatures, we could consider the oxygen

desorptions-

20 ” +2 N i^ - 02 (g) + 2 (36)

as a term inating ste p .

The key to the mechanism i s probably to be found in the disturbance of

the surface equilibrium which any sudden change in ^©^concentration must bring
 - 164 -

about. Thus, as a r e su lt o f ( 2 3 ) or ( 2 4 ) and ( 2 5 ) 02 or 0 i s ste a d ily

replaced by OH • This need not n ecessa rily a ffe c t a l l the adsorbed oxygen

but only the most la b ile part. "When the change i s complete the electron

balance in the surface layers of the oxide w ill be disturbed r e su ltin g in a

s lig h t rearrangement of e le ctr o n ic d efects which w ill tend to reach an e q u il­

ibrium p o sitio n . This process w ill depend on d iffu sio n in surface layers and

would be very slow, although much fa ste r than bulk d iffu sio n .

We have, of course, shown experim entally that the reaction i s f i r s t order

w .r .t , HgOg and how oxygen and water vapour a ffe c te d the rate of the reaction*

R aising the oxygen pressure increased the r a te. This could be explained by the

surface concentration of oxygen being increased to come to equilibrium with

that of the gas phase. Hew s it e s of 0^ and, or 0 would be created which

would influence the o rig in a l electron tra n sfer step . On the other hand,

in creasin g the p a rtia l pressure of H20(v) decreased the r a te , perhaps by retard­

ing reaction s ( 2 7 ) and (28) which w ill reduce the speed with which s it e s w ill

become a v a ila b le by prolonging th e ir occupancy with 0H~.

On th is b asis i t i s h igh ly probable that the rate co n tro llin g step on the

p-type ca ta ly sts i s the o r ig in a l electron tran sfer to the substrate, and the

o v e r -a ll v e lo c ity of the reaction w ill be d ictated by the a v a ila b ilit y of

these electron sources or "active s i t e s ”.

In the above no committment was made as to the exact source of the

— — 4 -4 -v
electro n (0 , 0^ , M ). However, th is question has been ex ten siv ely studied

by Hart and McFadyen (81) with film s of Cu^O on copper. They concluded that

the o rig in a l electron tra n sfer takes place at low temperatures from a pre-
— _ ||
adsorbed 02 ion rather than 0 or M • They based th e ir argument on th er-
 - 165 -

modynamic considerations using data from Evans, Hush and Uri (14-), "but a lso

on the r e s u lts of Gamer and coworkers (44, 45? 46 , 47) which suggest that

CugO oxide surfaces w ill he covered a t temperatures of the order of 50 °C with

oxygen in the 0^ form. Farther, th is work suggests that p lain desorption

of oxygen i s always a slow process, which, as we mention elsewhere, may explain

why HgO c a t a ly s is , requiring as i t does the stra ig h t desorption ste p , i s so

much slower than CO/O^ c a ta ly sis or decomposition. CO/O^ (see Section I

equation (6) ) requires no desorption a t a l l , -while in the present scheme 0^

i s lib e ra te d in a replacement. I t would be l e s s easy to postulate th is with

0 as the relevant sp e c ie s.

Of course, the p o sitio n and r e a c tiv ity o f the 0^ w ill a ffe c t the energy

of a c tiv a tio n . The surface condition with regard to adsorbed oxygen m i l

change markedly in the region 58 - 183°C in vhich the experiments were conducted.

At 100°C and above oxygen i s rea d ily desorbed; the Og” sp ecies has therefore

a short l i f e . At the higher temperatures i t must be expected th at only the

most firm ly held oxygen w ill be e f f e c t iv e , and these may invoke a higher energy
I|
of a c tiv a tio n . At the higher temperatures s t i l l , the M ions may act as

electron donors.

The disappearance a t temperatures above 100°C of an a c t iv it y f a l l cannot

be rea d ily ascribed to steps lik e ( 3 2 ) because of the very high decomposition.

Reactions ( 2 9 ) and ( 3 0 ) may again be invoked. This suggests that whereas the

only reaction by which OH (a d s.) may further the c a ta ly tic process at low

temperatures i s ( 2 7 ) and (2 8 ), i t may be attacked by H^Og d ir e c tly a t higher

temperatures and i t therefore lo se s i t s r e la tiv e p a ss iv ity .

In reactions ( 2 5 ) and (28) we have the p o s s ib ilit y of chain decomposition

 - 166 -

in it ia t e d by an a c tiv e OH radical produced at the surface in (22). The

surface reaction i s , of course, a chain reaction in which the a c tiv e surface

centres are the ca rriers. Thus the production of a ctiv e r a d ic a ls, themselves

capable of continuing a chain, i s a branching of the surface process.

Reactions (2 2 ), (26) and ( 2 7 ) are in it ia t io n and propagation steps in the

normal surface - centred chain. There i s no termination stage in the surface -

centred chain but, in the radical branch, mutual in tera ctio n s between rad icals

may be regarded as term inating. The branch i s i t s e l f a stra ig h t chain, and no

fu rth er branching can be postulated. I t may be considered as proceeding w ithin

the orbit of the surface chemical fo r c e s, i . e . w ithin the chemisorbed layer or

w ithin the van der Waals la y er. The la t t e r i s preferred.

Whether or not the bulk of the decomposition proceeds via the chemisorbed

la y er ch ain s, i t must be clear that the c a ta ly st owes i t s e ffe c t to a b i lit y to

donate and accept an electron rea d ily . Thus we would appear to have a reason

why the n-type and in su la to r oxides are much poorer c a ta ly sts than C^O, ig^O,

etc.

I t should be noted that the value of 7«0 + 0.5 K.cals/m ol found as the

E^ fo r "green" HiO i s sim ilar to the value of 7 K .cals/m ol found by Wright

and Andrews (147) for the surface energy le v e l of electron s in HiO. Whilst

the value of 12 + 1 K .cals/m ol found fo r a surface i s sim ilar to the value

found by Rideal and Wilkins (148) fo r the E^ of surface conductivity in th is

oxide (9*5 K .cals/m ol). I t may be considered that surface conduction w ill

in volve sim ila r elec tr o n ic forces as those involved in adsorption.

For an n-type semi-conductor such as ZnO, the surface in equilibrium with

2+ 2-
oxygen w ill con sist of nn ,0 ions and vacant anion s it e s balanced by quasi -

free electron s; excess metal being accommodated in i n t e r s t i t i a l p o sitio n s.

 - 16? -

In certa in instances the anion vacancy contains a trapped electron forming an

F — cen tre. In th is case i t is u n lik ely that reaction w ill proceed by electron
2+ 2— —
tra n sfer from a Zn or 0 io n , or even from 0 (a d s.) — since the tendency

i s to lo se not adsorb oxygen, but i t may be that an molecule adsorbed in

the v ic in it y of an F-centre could i n it ia t e decomposition by accepting an electron ,

thus j -

^0^ - - - - - ( v .a . s . ) = HO (a d s.) + OH (a d s.) + ( v .a . s . ) (37)

where ( v .a . s . ) ** F -cen tre.

and ( v .a . s . ) = vacant anion s i t e .

However, th is electron w ill be firm ly held as known from the o p tica l

properties of ZnO which, according to S eitz ( 7 ), give a value of about 40 K.

e a ls fo r the energy o f io n isa tio n from an F-centre in the bulk of the oxide.

The value fo r a surface s it e would be lower, and perhaps not fa r from the 34

K .ca ls which i s the of ^ ^ 2 ^ecomPo s i^^on on ZnO.

Furthermore, th is picture permits an explanation for the ready decay of

e ffic ie n c y on a heat treated sample of ZnO during a run. Thus, we may consider

that an electro n removed from the F-centre to a surface sp ecies would be more

mobile than before and could eventually be captured by a p o sitiv e defect (e.g*

i n t e r s t i t i a l Zn**). So that the surface would gradually become stripped of

a ctiv e s it e s and would have no tendency to rep lenish these by d irect

a ctio n on the ZnO normal l a t t i c e .

The step s follow in g (37) would be sim ilar to those outlined above fo r the

p-type c a ta ly sts, namely ( 2 9 ), ( 3 0 )? (26), ( 25 a) and (28).

C learly, the above argument could not a lso be applied to the c a ta ly sis
 - 168 -

on CcLO or CuO, sin ce these are "better ca ta ly sts than ZnO and invoke energies

of a ctiv a tio n o f 12 and 15 K .cals/m ol, resp e c tiv e ly . At f i r s t sig h t th is

discrepancy in a c t iv it y between oxides which are a l l n-type may be d if f ic u lt

to understand, but in the lig h t of the surface reducing action of the

p o sitio n becomes clea rer. With CuO i t may be expressed simply th u s:-

2CuO + E2 0 2 = H20 + 02 + Cu20 (38)

The surface may not be e n tir e ly converted to cuprous oxide, but an

interm ediate Cu20/Cu0 s ta te may be envisaged which may then propagate the

reaction in the same manner as the p-type oxid es. There was no tendency for

t h is oxide to lo s e a c t iv it y during exposure to

The other n-type oxides examined, CdO and FegQ^, are sim ila r to CuO, and

■unlike 2n0, in that the oxides of lower valency are sta b le . So that wemay

again p ostu late surfaces containing Fe+ and Fe-H- metal io n s, sim ila rly with

CdO. I t may be, however, w ith the la t t e r that the oxygen i s stripped from the

surface and the excess cadium i s accommodated in i n t e r s t i t i a l p osition s as

atoms or io n s, which in i t s e l f im plies that more electron sources would be

availa b le to i n it ia t e c a t a ly s is .

This connection of c a ta ly tic a c t iv it y of the n-type oxides with the

a b ilit y of ^Og to reduce the surface givin g, as i t were, a p-n equilibrium

can a lso be used to explain the remarkable f a l l of c a ta ly s is on r a isin g the

concentration. For i f we reduce the EgOg concentration from Ch to Cj/3

we ease the surface reduction; the surface then at C^ /3 becomes s ta b ilis e d .

Then, the change back to (h i s accompanied by a big drop in c a ta ly sis from the

orig in a l Ch value which may be due to the lack of su itab le surface s it e s on

the c a ta ly st conditioned, as i t were, at C ./3 # This, of course, would not

 - 169 -

apply to the p-type oxides since they have an inherent tendency to pass from

M to M w ith the adsorption of oxygen, and r a isin g the p a rtia l pressure of

oxygen in contact with such a surface, i . e . in a change from C6/3 to CL , w ill

only serve to increase the concentration of q u asi-free p o sitiv e h o les, and

02~ or 0 io n s, to the o rig in a l CL valu e.

So far we have not touched on the probable mechanism of in itia t io n on the

in su la to r oxid es. Experiment showed that in the temperature range studied these

were in a ctiv e ca ta ly sts and only became a ctiv e when treated in some fashion as

may reasonably be held to make them n-type (see sectio n ( 3 ) )• The sources o f

electron s may then be E -cen tres, or reduced metal io n s, or in the case of

Al/AlgO^ as has been postulated by Mott (149) electrons may pass to the surface

by a quantum mechanical tunnel e f f e c t , while the oxide film i s s t i l l growing.

This being so , i t could explain the fin d in g that Al/Al^O^ when heated in a ir was

non-active (oxide film 80-100A° th ic k ), but when subjected to a fie r c e heat-

treatment in hydrogen and quickly cooled; a f a ir ly high c a ta ly s is rate was

found (oxide film = 40A°, say)# So long as the Al/Al^O^ was in contact with

HgOg* the oxide film would remain a t an intermediate thickness and c a ta ly sis

would be recorded. However, stopping the H^O^ flow and leavin g the ca ta ly st

in contact with a ir overnight would e f fe c t iv e ly replace a reducing atmosphere

by an o x id isin g one, which, in tim e, would allow the ca ta ly st to recover to i t s

o rig in a l oxide th ick n ess, and no c a ta ly s is would be observed on sta rtin g up at

180°C, because no electron sources would be fr e e ly a v a ila b le to in it ia t e the

decomposition. This i s not to say that a surface of uniform thickness is

v isu a lise d but that there w ill e x is t a ctiv e surface d isco n tin u ities and cracks

which in 24 hours w ill become, as i t were, f i l l e d up with an oxide film . This

irregu lar growth ox oxide film s on aluminium on exposure to a i r has been

 ~ 170 -

observed by Yernon ( 1 5 0 ), and i t may a lso be noted here that recent work by

Hart ( l5 l ) Has shown that hydrated oxide film s of aluminium growing in water

are accompanied by intergranular attack.

The r e su lts obtained with the in su lator oxides were in f a ir harmony with

those of Giguere (34) fo r Sn and AL surfaces in a s t a t ic system using H^C^

vapour concentrations and temperatures sim ila r to those here. The picture i s

markedly d iffe re n t to the work of S a tte r fie ld and Stein (26 ) who use high

p a r tia l pressures o f and H^O and fin d second order k in e tic s in

and complex in h ib itio n by B^O vapour. The in flu en ce of increasing p a rtia l

pressure of oxygen and water vapour in the present experiments was, as described,

only Just detectab le w ith the "normal” in su la to rs, due to lack of c a ta ly tic

a c tiv ity *

(a) Expansion of k in e tic s .

We may now attempt to complete the general k in e tic mechanism approach

begun above. The features to be accommodated are;~

(1) the f i r s t order dependence on j ^ a Q f o r rapid changes o f [s^O ,]

(2 ) the slow response to oxygen pressure changes which follow the approximate
1 /g
r e la t io n ! - rate of catalysisocpOg •

(3 ) the rather more rapid response to pHgO which poisons according to the

r e la tio n sh ip :- rate o f c a ta ly sis oc pH^O 2

(4 ) the complex changes which fo llow an instantaneous change o f

In the f i r s t place we may regard the peroxide decomposition as

occurring on a surface fix ed in composition. This surface i s subject to

slow change when I^C^, ^2 ^’ or ^2 Par"^^a-*- Pressures are changed separately

or together, but we may consider these changes from a k in e tic point of

view as separate p rocesses.

 The f a s t decomposition process may then he formulated as con trolled

hys-

+ s * = products + (39)

where S* i s an a ctiv e surface sp e c ie s, and i s able to return to S*

in a step which i s not rate co n tro llin g . Wehave discussed the nature

of S* already. I t may he 0 , 0^ , HO ametal ion , HgOg or an ^ 0^

dependent sp ecies lik e EO^ •

We then have from (39)

- a [ HgOg"] /a t = JqiHgOg mg* (4 0 )

I f Big#* the surface a c t iv it y of th is a ctiv e sp e c ie s, remains constant

during a very rapid change in pH^O^ we get the condition for a f i r s t order

rea ctio n .

On the other hand we have to observe that a r e la tiv e ly fa s t change

in the c a ta ly sis soon s e ts in a fte r the establishm ent o f a f i r s t order

re la tio n sh ip . This can only be due to a change in I t niay, however,

be induced in various wayss** v ia ^2 ^ 9 0^ or H^O. I t i s too fa s t to

be due to 0^ and too large to be due to B^O. Thus i t must be due to E^C^.

A simple p o s s ib ilit y i s that the surface is at f i r s t , a t low

temperatures anyway, capable of conducting a simple cycle based on 0

and Og •

0" + H2 0 2 (g) =0 ~ + H2 0(g) (41)

02"+ H ^ g ) = 0“ + HgOfe) + 02 (g) (42)

w ith eith e r of the reactions rate determining or with a balanced ra te.

Then poisoning could occur th u s:-

20" + B ^C g) = 2 OH" + 02 (g) (43)

in which OH i s a much sta b ler sp ecies than eith e r CT or 0^ *.

At high we may consider a regenerating s te p ;-

CH" + 'H202 C«) = 02H" + HgOCg) (29)

which may indeed be balanced so that the poisoning e f fe c t of B^O i s

excercised here. (44) could be succeeded by the slow s te p ;-

02H~ + OH” . 20" + HgOfe)

(34)
or 202H *s 02 + 0 + H20(g)

Steps lnJce (43 )> though quite rapid, are slow compared with c a ta ly sis

i t s e l f , and so for rapid rate changes we could represent the whole process

by ( 4 l) and (42). These lead d ir e c tly to a r e a lisa tio n of the importance

of 02 which would have an e f fe c t on the surface a c t iv it ie s of 0 and Q^",

and a ls o on other surface e le c tr o n ic e f f e c t s . Thus in the oxidation of

p-type m etals, Wagner and others have found that the rate of growth is

proportional to pO,,1/ 6 - pO ,1/ 8 . To explain th is fin d in g the rate i s

held to be proportional to the concentration of vacant cation s i t e s , and

p-holes - two of which are induced per electron by the oxygen* In our

case we may consider that the two sin g ly charged ions 0 , the two vacant

s it e s and the two p-holes a r isin g on adsorption o f our oxygen molecule -

s ix species in a l l , can a l l a ct as an anchor fo r a c a ta ly tic cycle

in volvin g HOg •
 Thus we would expect
l/6
BateoC p02 ' , which is s u f f ic ie n t ly clo se to the experimental

fin d in g . Equation (41) gives (39) where mg*. is now seen to he m^~

(assuming mQ- * mfi - ) . The poisoning scheme above may be analysed.

2
At balance, ( 4 3 ) and (34) g iv e :-

k43‘ mV ,pH2°2 = k34* m02H~- “OH" ^44)

while the balanced reaction ( 2 9 ) gives s-

mOpH~ = g2 9 ,pB2 ° 2 * mQH~ (45)

pH20

which g ives when su b stitu ted in (4 4 ):-

k43* “o"2* pH2° = kJ4 • K29m2°H

" (46)

su b stitu tin g fo r mg* (now m^-) in ( 3 9 ) we obtain, when is

approximately constant (e .g . when’= . l )

4
// lr TT
90
\
\
Bate = k,
■ f a y .

= k> pHo 0„ / 1 \ ® (47)

l2°2 ' 6v~)
Combining th is with the oxygen dependence we fin d

i / x1/6
Rate
- • ( p° 2 ) (4 8 )

Such an expression i s that ap plicab le to changes which are slow

enough to allow a l l slow surface eq u ilib ria to s e t t l e .

Equation (48) ap p lies only i f mQg~ i s constant, and th is i s surely

 - 174 -

only lik e ly to be a v a lid assumption when i t i s approximately equal to

un ity i<>e. a t a h igh ly poisoned surface such as w ill obtain at a low

temperature.

m0H~ from (45) g iv e s :-

k» pILO? Kp- % <)

Bate - —— (48a)
P a PHo0o .* PELO
HgO 2 2 ^29 ^2

(sin ce m
0
H~ -+m
o.H
* =^

This suggests that a zero order rate should come in when pg ^

^29 P^G i'*121’*' rader these circumstances

1
V
Rate CK Pg 0 (48b)

i . e . an accelera tio n not diminution in r a te , which may come in at much

higher p a rtia l pressures

I t should be mentioned that sim ilar rate equations to these may be

explained by assuming that the adsorption of the relevant species is

governed by e ith e r a logarithm ic (Temkin and Pyzhev, 160) or a power rate

law (Kwan, l 6 l ) . Ewan ( 1 6 2 ), fo r example, applied the power rate law to

the 2C0 + 0^ = 2 CO2 reaction on RiO.

The power rate law sta te s that the rate of adsorption of a substance

is given b y:-
-P
Rate * k^p 0 - k^ 0 (49)

where p i s the pressure o f adsorbate and 0 i s the fra ctio n of surface

covered. Other symbols are constants. At equilibrium th is law reduces

 - 175 -

to the Ereundlich isotherm. Eor the CO reaction i t i s necessary to

assvime a mechanism which Kwan took a s :-

= 20 ( 50 ^

0 + CO « C02 (51)

( 5 l) being slow . Then fo r the r e a c tio n :-

Rate = k pCO *= kpCO . p02 ^ 2)

which i s in agreement with high temperature r e su lts provided:-

“ ° ’5

an assumption shown to be reasonable by many examples of adsorption

r a te .

This method o f tre a tin g surface reactions becomes very complicated

when i t is necessary to consider several surface sp e cie s, and consequently

several p o ssib le interlocked reactions - as in fa c t we fin d i t n e c e s s a r y

to do. I t amounts to replacing the surface coverage, m in our Scheme,

P
by m^ fo r each surface rea ction . When th is complexity has added to i t

the uncertainty relevant to the numerical values of ^ and ft for each sp e c ie s,

i t seems evident that the treatment i s in ap p licab le. We might however

take fo r illu s t r a tio n our elementary scheme, namely that the rate of

c a ta ly s is i s con trolled only by the adsorption s te p :-

H202 + 2S* = 2 SOH

and t h a t s u b se q u e n t r e a c t i o n s a r e so r a p i d a s to be a t e q u ilib r iu m .
— o£

Then R a te - k pH 0 . 9 (53)
 - 176 -

and i f pg i s the p a rtia l pressure of which would be in equilibrium

w ith the chemisorbed sp ecies S.OH*

pe = KopH^O . p022 (54)

from the overa ll reaction equilibrium ,

-V13)
Also p m con st, 0 (55)
e

sin ce the freu n d lich type of equation r e la te s 0 and pg. Thus

k P ^ °2
Rate (56)
pH2C^. pOg*

i f ctj(« fp}= 0.5

C learly, although th is gives the correct dependence on both ^ 0^ and

H^O i t f a i l s completely fo r 02 *

To sum up, we may conclude that the prime consideration as to what

makes an oxide a good c a ta ly st in th is reaction , i s i t s a b ilit y to donate

electron s to the substrate from la b ile surface s i t e s ,

(b) I n i t i a l and c y c lic e ffic ie n c y changes.

I t would appear w ith the metal film s that two com petitive surface

processes are occuring, one dominant a t high and the other at low temper­

atu res. This f a l l in e ffic ie n c y cannot be due to the s e ttin g up o f an

equilibrium d istrib u tio n of d efects on the surface through migration from

the bulk, sin ce a t these temperatures (60- 180°C) io n ic type d efects are

h igh ly immobile, and the time la g in a tta in in g the steady value would not

be accounted fo r .

During the course of the reaction parts of the surface may be coveredty

interm ediate reaction sp ecies such as HOg or, and more e s p e c ia lly , HO- .
 - 177 -

I t i s n o t, however, considered that these sp ecies are chem ically bonded

a t the surface (fo r i f th is were so then the surface would become poisoned)

rather that they w ill propagate the reaction from the van der Waals layer.

The i n i t i a l decay may be due to in a ctiv a tio n proceeding through the

strip p in g of excess a c tiv e oxygen from the surface, and in support i t has

been shown that wblack,f MO w ill become a d irty green in about 50 minutes

w h ilst catalysin g vapour. This change must involve the slow removal

o f la b ile surface oxygen. The excess oxygen w ill read ily oxid ise 0^

and when the supply i s exhausted or has reached an equilibrium , the,

c a ta ly sis w ill assume a steady r a te .

A nalysis of the decomposition figu res on the MO and Ct^O flash ed

film s during the i n i t i a l f a l l in c a ta ly sis shows that they conform to a

law o f the ty p e:-

= K, exp (-AD) (a)

where D i s the fra ctio n decomposed a t time t ; A and K are constants.

On in teg ra tio n th is g iv e s:-
D « l/A . [ l n ( t + t o ) - 1 * 0 . 0>)

where t * l/AK.
o

This i s a form of the Elovich ( 1 6 5 ) equation subsequently shown by Taylor

and Thon (133) to apply to many c a ta ly tic rea ctio n s. The value of t is

gen erally found by t r ia l and error p lo ts u n til a straigh t lin e i s obtained

when graphing D against In (t+ tQ). In the present case when t Q = 0

equation (a) may be w ritten :-

lo g t = (A /2.3) D (c)
.
m in s

O
□
3

05

o 02 04 0 2. 10
f r a c t i o n H 2 o2 d e c o m p o s e d

E U O V lT CH TYPE P L O T S O F INITIAL. E F F I C I E N C Y
CHANGES ■ 0 OXIDISED, FLASHED NICKEL FILM AT I6 0 °c
(jN CfcCA ^E IN E F F I C I E N C Y )
f N i F ilm - io o ° c ^ e ffic ie n c y d e c r e a s e )
ft ” - (3 0 ° C (^ » )

FIG. 51
 - 178 -

and the graph of lo g t again st D should give a stra ig h t lin e . This is

shown in F ig. 51 for the slow f a l l in decomposition of a n ic k e l/n ic k e l

oxide fla sh ed film at 60°C, and fo r the r is e and f a l l of e ffic ie n c y of a

Cu/C^O film at 32°C, before the onset of steady ca ta ly sis*

(5) The Arrhenius parameters*

A common c r ite r io n in c la s s ify in g c a ta ly tic reaction s i s the temperature

c o e ffic ie n t or a c tiv a tio n energy o f the reaction . This i s calcu lated from the

rate o f reaction a t d ifferen t temperatures.

The in terp reta tio n of the meaning of the temperature c o e ffic ie n t of a s o lid

catalysed reaction i s well-known to be a d if f ic u lt matter. Recently, Trapnell

(152) and R ideal and Trapnell (153) have c r it ic iz e d the sig n ifica n ce given to

as found from the equations-

They suggest that a b e tte r in d ica tio n of c a ta ly tic power would be to

measure the temperature c o e ffic ie n t a t constant surface coverage of rea cta n ts.

This c r itic ism i s a real and v a lid one, since in very many cases the

fig u re s quoted for experimental a c tiv a tio n energies have been those obtained

by c a lc u la tin g E from the above equation when the reaction had reached a
•A.

steady value a fte r a temperature change. This complicates the system by the

in terven tion of the heat of adsorption of the reactan t. Thus, fo r a unimole-

cular surface reaction retarded by one of i t s products, i t may be shown

(Hinshelwood, 1 6 5 ) on a simple Langmuirian b a sis th a ts-

a In k
obs
(2 )
 - 179 -

where, E = True a c tiv a tio n energy o f the reaction#

» Heat o f adsorption o f the reactant.

AH2 «= Heat of adsorption of product which poisons,

and by equating (2) and ( 3 ), we see th a t;-

Ea = E 4 -A ^ - A Hg.

This expression may, as a f i r s t approximation be taken to hold fo r the

present reaction - which i s retarded by water vapour. (We are here ignoring

fo r the sake of s im p lic ity , the s lig h t a ccelera tin g role of oxygen.) The heats

of adsorptionAH^ and Aft? w ill themselves be dependant on the degree of cover­

age o f the surface w ith adsorbed m olecules, even although in a f i r s t order

process the adsorption i s sm all.

However, as i s w ell known, the degree of coverage of adsorbed molecules

changes slow ly in tim e, as has been shown, for example, in the work of Taylor

and Liang (154) where the volume of hydrogen adsorbed on ZnO took 16-17 hours

to move from ^ t o J ccs a t 184°C. So that the d if f ic u lt y of th is variation

inAH may be somewhat overcome by making an instantaneous adiabatic temperature

change in which the fr a ctio n of surface covered can be held constant. I t may

quite reasonably be expected to take 20 - 25 minutes before the coverage

adjusts i t s e l f to the new temperature, which i s the average time taken to reach

a steady decomposition value in these experiments a fte r a f a s t temperature

change.

In other words, in the present experimental set-u p , i t is possib le by a

minor extrapolation to “zero” time to obtain a temperature c o e ffic ie n t at

constant adsorbed volume fo r any given temperature i . e . at a constant A

and AHg Thus, i f th is condition i s achieved the E values found with

 - 180 -

d iffe r e n t c a ta ly sts are f a ir ly certain to be sig n ific a n t i f they are compared

by the values obtained at the same temperature. Since there i s no reason to

b eliev e that A a n d AHg a t 100°C, say, w ill vary much from one ca ta ly st to

another of the same a c t iv it y c la s s . We may then in terp ret our values on

the b asis given in sectio n ( 3 ).

(6) The compensation e f f e c t .

The c a ta ly sts examined gave a range of from 4K. cals/m ol with Mn^O^

to 34 K .cals/m ol for ZnO. For a sin g le ca ta ly st we fin d a variation in E^

according to the temperature of determination ( e .g . Ee^O^ gave from 16 to 22

K .cals/m ol. between 100 - 130°C) and for HiO c a ta ly sts containing trace

im purities a v a ria tio n in E^ was found in determinations at the sametemperature.

How, we write the Arrhenius equation:-

Rate « C n x A x exp. (~ E ./r t) (x)

2 2

where i s the energy o f a c tiv a tio n and A i s the frequency or entropy fa c to r .

In the case o f HiO + im purities we fin d , as shown in F ig. 52, that A changes

w ith so th at the lin ea r r e la tio n sh ip :-

E^ = constant. In A + constant

or In A = constant. E^ + constant

i s adhered to .F ig . 52 demonstrates th is for the reaction on HiO + x mol %

Ga^O^ and HiO + x mol% Li^O. For these s e r ie s a change in E^ fo r the temper­

ature drop from 100°C i s always accompanied by a sim ilar change in A, and the

points l i e on a str a ig h t lin e . This figu re a lso shows a sim ila r r e la tio n s h ip

derived from d ata given by Parravano (6 l) f o r the re d u c tio n o f ITiO (co n ta in in g

L io 0, Ag _0, A1_0 e tc . as im p u ritie s ) by hydrogen in the tem perature range

2 2 2 3
 r*> /f oOJ nCvl o o o rtO J oOJ
y * <* *. * i i 1
o W (7 |J (J ^
y„ J
U_ ^ ^ o'* ,N5 v*
2 4 ^ oP I I ^ J J J
o
r\ * o °o> aJ oJ ? o 2 5 ?
.^ £ £ E _ 5 '4-Z 2 2 £«
2 O O ^ C ^ - w OJ
wlI oJ
V)

t %
tU 2
o o 0 O <\j <3 J -
+ + -#- +
bJ h + + 4 <n
te a j -J =
<02 o
(3 \3
© O B t > O i 0 9 • 19 CM
s g
o
° 2
v> 4
0 w
2 tL
?0
*a 0£
*6
IL I
0

Ky
1 z
0 ui
rr tfl
o %
2 0
tft a
ai J IL
g a li
0

I t- _

<r
ill u S IX
0. ifl
S' 2
Z o

yu lO
U v I
u cL
u. O S
u
|_ IX
h u ft
X au
ill O I
z uy
Li O' O
a
5 x*
©
oO m u
• V 01'&©“!
 - 181 -

150 - 350°C and p a r t i a l p re ssu re s of 200 - 500 mm Hg.

This type of r e la tio n has long "been observed f o r heterogeneous systems

of th is kind. I t i s c a lle d the “Compensation E ffe c t” s in c e , as w ill read ily

be seen, a high E^, which would reduce the amount of reaction a t a given

temperature, i s accompanied by a high A which p artly compensates.

Barrer ( 1 5 5 ) has sta ted that th is e f fe c t has no deep sig n ifica n ce but

a r ise s from the fa c t that p ra ctica l measurements of rate l i e w ithin a narrow

range. Thus i f in (x) we p u t:-

Rate = k. n

so that k = A. exp. (~E^/ST).

and regard k as su b sta n tia lly constant while T varies over a wide range th en:-

ea
lcc 2c — lo g A ™ 2.303T

®A
i.e . lo g A - 4 .5 7 7 + c o n sta n t.

Such a rela tio n sh ip does f i t many s e r ie s of r e su lts e .g . decomposition

of NgO on (Cremer and la r s c h a ll, 1 5 6 ) and decomposition of on Pt/Al^O^

(Mikovsky and Waters, 157 )> hut i t f a i l s when a sin g le promoted c a ta ly st is

being considered (Fig. 5 2 ) , because here k i s by no means constant in comparison

with and A. Barrerfs c r itic ism was intended a p p a re n tly to apply to ranges

of c a ta ly sts a c tiv e over s e v e ra l hundred degrees.

Cremer (158), on the o th er hand, has argued t h a t the e f fe c t has a fund­

am ental s ig n ific a n c e and has discu ssed the r e l a t io n fo r many system s, (e .g .

decom position and c h lo rin a tio n c a ta ly s is ) .

 - 182 -

We must s t i l l , however, explain the change o f E. w ith temperature and

Jx
composition o f a given c a ta ly st and why A always changes in the d irectio n of

compensation. Much of Cremerrs d iscu ssion i s relevant to th is question. Thus,

although some of the d iscu ssion has been about f a ls e ca ses, there i s an e ffe c t

here which i s re a l and worth examining.

There are two sources of v a ria tio n in the A fa c to r , the concentration of

a c tiv e s i t e s and the entropy term. Thus i f we w rite the e ffe c tiv e process a s s-

H202 + A ^ *Products + A ., ( l )

where (HgOg - - - - - a )* represents an interm ediate complex which i s certa in ly

an adsorbed sp e c ie s, the absolute rate theory would gives-*

Rate - k x C (H2 02..................... A)* (2)

Now, K * C|
( h2 o2 ------------ a ) * / V ch202

and K* * exp. (~A^/RT)

SO Rate = k>0H202. C^. exp. (-&G/RT).

and i f a lso Bate ■ q

= kCA. exp. ( A s / r ) . exp. (~A2/RT)*

A lso, i f we as usual w rite the v e lo c ity constant in the Arrhenius styles*

d In k
■ * i
A /y \
dt “ ^
 - 185 -

we e a s ily fin d s-

« ke exp. ( A s/&) exp. (-E^/ rt ).

(y) in the form k = A exp* (-E ^ /r t).

shows that the A fa cto r has two variab le parts and A S.

Now, fo r an adsorption process we would not expect AS to vary much

sin ce i t in v o lv es, no matter what the s o lid , la rg ely the same type of process

as the condensation o f a gas m olecule. I t w i l l , however, vary s lig h t ly

according to -whether the i s held firm ly or in a mobile surface s ta t e .

We conclude that i s the main cause of A v a ria tio n .

A further point i s that reaction ( l ) and hence equation (2) refer to a

sin g le process on a sin g le type of a ctiv e s i t e , but we may be dealin g with a

range of processes each w ith a d ifferen t and C^. (Thus the e ffe c t of a

temperature change would be sim ila r to the e f f e c t o f temperature on conductivity

in a semi-conductor with a number o f d ifferen t kinds of impurity. The o v e r -a ll

rate would be in the form o f a s e r ie s s -

k = k*. C'A. exp. (~E»a/RT) + k " CA" exp. (~E"a/ rt ) + e tc .

(7) C atalysis on NiO and -promoted NiO*

In previous sectio n s reference has been made to some of the c a ta ly tic

r e s u lts found with the n ick el oxide system. In ad d ition , the resu lts sectio n

described certain stru ctu ral properties which have been in v estig a ted . In view

of the sp e c ia l a tten tio n given to NiO in th is work th is sectio n w ill deal w ith

the r e su lts obtained in greater d e ta il.

The c a ta ly sis was examined on flash ed Ni film s, specpure Ni metal and

 - 184 -

’’■black” and “green” NiO. At 100°C, “’black" NiO gave 87/b decomposition f a llin g

to a steady value of 63^ a fte r 55 minutes, “green” NiO gave 5Cffo steady decom­

p o sitio n , specpure metal 12$ and flash ed film 20-22$., The flash ed film cannot

he s t r i c t l y compared with the others because the resu lt refers to a d ifferen t

flow system. The a c t i v it i e s may then be placed:-

"black" NiO Z> "green" NiO Ni/NiO,

Here, as in most cases where metal/metal oxides and bulk oxides of the

same element were examined, the bulk m aterial was more a c tiv e .

The higher a c t iv it y of "black" over "green" n ick el oxide, and the lo s s

of e ffic ie n c y in time of the former, was probably due to the excess oxygen

which th is oxide contains. I t may be considered that the lo s s in a c t iv it y

which occurs in step with the colour change i s due to the stripp in g of the

la b ile oxygen from the ca ta ly st su rface. This oxygen i s chemisorbed, but

lo o se ly bound (Hauffe and B n gell, 159) sucl the onset of the steady decom­

p o sitio n le v e l may be coincident with the s e ttin g up of a le s s a ctiv e surface

where the oxygen content i s at equilibrium with that in the gas phase. In

which case, a l l the la b ile oxygen i s not removed and there s t i l l e x is ts

appreciable d eviation from stoichiom etry - the colour did not f u lly become the

apple green o f stoich iom etric n ick el oxide. The colour of the oxides containing

more than 0.5 mol % Li^0 in “green" NiO was unmistakably black but no lo s s of

c a ta ly tic e ffic ie n c y nor change of colour was observed, w ith passage of E^O^.

In ad d ition , the temperature c o e ffic ie n ts were e a s ily reproducible.

When Li20 i s added to NiO and fir e d in a ir the f in a l product has adsorbed

oxygen to such an extent that the N i/54- content is equivalent to the amount of

lith iu m added to the system (De Boer and Verwey, 118). Thus:-
 x = fr a ctio n MO replaced.

The concentration of p o sitiv e holes m i l increase with in creasin g lithium

ad dition .I t can he seen from J ig . 41 that the c a ta ly st e ffic ie n c y in th is

4-
reaction increased as the Li content increased, i . e . with increasing p-type

character o f the oxide, and consequently as electron in tera ctio n s it e s he come

more rea d ily a v a ila b le to i n i t i a t e rea ctio n . This explanation would appear to

f i t the experimental data, adds weight to the early a c t iv it y se r ie s (sectio n

(2) ) , and a ls o to the p ostu late that the rate co n tro llin g step i s electron

in te r a tio n between ca ta ly st and su b strate. The MO + Li^O ca ta ly sts may then

be b etter than "green" MO through the greater a v a ila b ilit y and r e a c tiv ity of

electron sources such as 0 or Og with which th is surface w ill be permanently

saturated.

As to the in s t a b ilit y of the "black" MO, containing no Li"*", i t may be

sta ted that the la t t e r cannot be considered to be a stab le or s p e c ific stru ctu re.

The oxygen content varies with the manner and temperature of i t s preparation,

and although the present ca ta ly st was s p e c ifie d by X-ray, magnetic and chemical

a n a ly sis , i t s exact composition could only be duplicated by rigorous r e p e titio n

of the experimental procedure. Thus though the "black" ca ta ly st i s i n i t i a l l y

more a c tiv e than any containing Li2 0 , i t soon worsens on exposure to ^ ^ 2 as

0 ,0 2 and p h oles are reduced s u p e r fic ia lly .

We may accept that as a r esu lt of the mechanical mixing, grinding and

thermal treatment accorded to the impurity s e r ie s , a uniform d istrib u tio n i s

achieved in the l a t t i c e and the induced valency changes would be d istrib u ted

throughout the bulk. So that i t may be considered that w h ilst "black" MO lo ses
 - 186 -

e ffic ie n c y through the lo ss o f excess surface oxygen, MO + LigO is stab le

sin ce the excess oxygen i s balanced throughout the la t t i c e by an equivalent

number of vacant cation s i t e s and q u asi-free p o sitiv e holes giving a stab le

stru ctu re.

The decrease of the experimental on MO + Li^O as the Li4- content i s

increased may be explained on the same b a sis as that used above to describe

the increase of e ffic ie n c y . On the other hand, the e f fe c t on the c a ta ly s is of

34-
adding Ga i s d if f ic u lt to in te r p r e t. The f i r s t e f fe c t i s to mop up already

e x is tin g p o sitiv e holes in MO, th u s:-

2© + Ga2 0j - 2 Ga+ (Hi) + 2 HiOt-JO^g)

where © » p o sitiv e h o le.

A ll the mixtures containing Ga20 ^ were pale green in colour and did not

a lt e r during contact with HgOg# The samples containing 0.004- mol %

showed ex a ctly the same a c t iv it y a t 60-138°C as the "green" n ick el oxide, which

leads to the conclusion that the impurity concentration was not great enough to

appreciably modify the Pehlordnung of MO, at any rate as far as the c a ta ly s is

was concerned. Further addition o f Ga3+ did n o t, as may be expected, produce

the opposite e f fe c t to that of Li*", in that the ca ta ly st e f fic ie n c ie s did not

34-
decrease ste a d ily with increasing Ga , but increased to a f a ir ly constant

le v e l independant of the amount of GagO^ added. This le v e l was about 1 .5 times

as high as pure MO and the same as fo r MO containing 0.24 mol % Li^O.

On the other hand, the experimental a c tiv a tio n energies for the fa s t
34-
temperature changes fo r the GagO^ se r ie s were found to increase with Ga

content from 7 to 12 K .cals/m ol fo r a T drop from 100 to 58°C, and to remain

f a ir ly steady at 6 + 1 K .cals/m ol for a T r is e from 58 to 136°C. I t may be

 - 187 -

th a t as w ith CuO, there i s a p o s s ib ilit y of a surface reduction by

-i_
u p settin g the d istrib u tio n o f elec tro n ic surface d efects giving Hi which, may

then rea d ily i n i t i a t e the reaction by lo sin g an electron to an adsorbed

m olecule. This cannot be considered very fe a s ib le , however, as i t would not

account fo r the in fe r io r it y of HiO as a c a ta ly st; nor can the existen ce of

4-
Ui be considered probable. However, i t may be supposed that the increase in

q u asi-free electro n concentration by adding Ga^O^ w ill not only an n ih ilate

certain p o sitiv e holes but, in many ca ses, w ill lead to the formation of

exciton s (in exciton is produced -when an electron is trapped in the e le c tr o ­

s t a t ic f i e l d o f a p o sitiv e h o le, Frenkel, 166), These would be p o te n tia lly

low energy source of electron s and could in it i a t e decomposition by electron

tra n sfer to an adsorbed molecule g ivin g a high e ffic ie n c y at 100°C (sa y ).

Then when the temperature i s dropped to 58°C, the surface i s momentarily

poisoned by an excess of OH or 0^ (a d s,) givin g a low rate and a high E^.

A fter about 25 minutes, the equilibrium i s reesta b lish ed and the rate recovers

to a higher e ffic ie n c y . Then, on r a isin g the temperature to 156°C the decom­

p o sitio n rate gives a steady high value almost instantaneously, due to the

absence of OH poisoning or fa ste r 0^ desorption. The ca ta ly st is then sim ilar

in a c t iv it y to "green" HiO and fo r the f a s t T change gives an id e n tic a l E^.

(8) C atalysis on equimolar mixed oxides.

Early p ra ctica l ap p lica tio n s of ca ta ly sts for decomposing concentrated

liq u id ^2^2 ma^e 1186 an<l Cu oxides in a matrix of cement

made highly a lk alin e with KOH, but the influence of composition and treatment

was only vaguely understood, and does not appear to have been in v estig a te d .

This choice of the above oxides could hardly have been improved, fo r ,

as we have seen, these are the most e f fic ie n t c a ta ly sts for gas phase H^O^
 - 188 -

d estru ction , What was not clea r was what advantage, i f any, was obtained

from the simultaneous use of these a c tiv e oxides; whether any synergesis was

involved i . e . whether the c a ta ly tic action of a mixture of oxides was greater

than the sin g le oxides used sep arately. Such e f fe c ts are, of course, common

in heterogeneous c a ta ly sis e .g . in Fischer-Tropsch c a ta ly s ts , in which mixtures

of oxides very sim ila r to these mentioned above have long been used. The

d ecisio n to study the mixed oxides by the use of s p in e l-lik e mixtures was

la r g e ly based on the a v a ila b ilit y o f the stu d ies H uttig (73) had made of

s p in e ls . We chose the two s e r ie s , one based on MngO^ and one on CuO, so as to

focus a tte n tio n separately on the divalent and tr iv a le n t io n s.

In the s e r ie s we studied the e f fe c t of annealing temperature on

the mixed oxid es, and demonstrated that i t is not the f in a l sp in el - lik e

structure which i s most a c tiv e but the interm ediate s ta te s which give the best

c a ta ly s ts .
Thus, as would be expected, i t i s the most highly d istorted
ti
c ry sta ls which are most a c tiv e . Following H u ttig's d iscu ssio n , we conclude ■

that the most a c tiv e mixture i s that which i s heated to the point where migrat­

ion of foreign ions into neighbouring c r y s t a llit ie s in a mixture has begun.

At th is p o in t, which seems to be a t 600°C fo r a l l the 0^ m ixtures, the

number and v a r iety of l a t t i c e im perfections w ill probably be g r ea test, though

in the absence of a very f u l l study of sem i-conductivity i t i s not p o ssib le

to be p recise about t h is . As was shown in the X-ray and paramagnetic work,

considerable stru ctu ral a lte ra tio n s were induced by sin te r in g , and these appear

to be rela ted to c a ta ly tic e ffic ie n c y .

The q u a lita tiv e r e su lts obtained when these c a ta ly sts were used to

decompose liq u id H.T.P. showed that the best gas phase ca ta ly sts were also

superior in the aqueous phase.

 ~ 189 -

The onset of ferromagnetism was not observed, w ith a l l the equimolar

mixtures based on but s u s c e p tib ility changes were, and may be a ttrib u ted

to the changes in the magnetic environment of d iffu sin g paramagnetic ions over

the surface, and perhaps a lso to valence induction (Anderson, 1 6 7 ). The lo s s

of c a ta ly tic a c t iv it y coincides w ith the destruction of the la b ile interm ediate

sta te s of the mixed oxides and the formation of a more stab le structure at high

(700-800°C) temperatures of sin te r in g . The a ctiv e centres w ill then have a

lower a c t iv it y anld th is may explain why the a c t iv it ie s of the mixed Mqo0_,

c a ta ly sts ste a d ily increased in the 3G0°C, 400°C and 600°C s e r ie s , and then

showed a marked f a l l with the 720 or r[60oC preparations.

With the se r ie s no very str ik in g e ffe c t of adding a second oxide

was observed. This was probably due to the very high a c t iv it y of Mn^O^ i t s e l f

(E^ * 6 .9 K .cals/m ol and IQOfo decomposition at 80°C fo r the 600°C s in t e r ) .

I t was at f i r s t sig h t surprising that the 2n0 mixtures did not show a bigger

f a l l in a c t iv it y compared with Mn^O^ i t s e l f . We must in fe r that even at 400°C,

there i s some surface migration so that the sample ex h ib its almost as many

a ctiv e s it e s as when i t co n sists e n tir e ly of Mh^O^.

A str ik in g r e s u lt w ith the Mn0Ch mixtures is the marked enhancement

of c a ta ly s is brought about in the ZnO/Mn^O^ s e r ie s when more f u lly annealed.

The in a ctiv e n-type oxide d e fin ite ly induces an outcrop of new and a c tiv e s it e s .

These new s it e s could a r ise eith e r in the Mn^O^ or ZnO c r y sta ls by introduction

o f im p u rities, or a lte r n a tiv e ly at in terfa ces between the p and n-type c r y s ta ls ,

which would appear to provide ex c elle n t s it e s fo r the type of dual s it e c y c lic

c a ta ly s is which seems most lik e ly to be involved, and which we have already

p ostu lated .

Apart from the case of the ZnO/DhigO^ m ixtures, th is Mn^O^ group o ffers
 - 190 -

l i t t l e o f str ik in g in te r e st from the c a ta ly tic view point.

The mixtures based, on CuO, an n-type oxide of interm ediate a c t iv it y , were

more in te r e stin g , e s p e c ia lly when the second oxide is Feo07 or Alo0^, though
^ k 5
here again when the a ctiv e Co^O^ was used i t s high a c t iv it y seemed to obscure

any mixture e f f e c t . These mixtures a l l gave str ik in g ly lower E^ values than

the separate oxides and were much more c a ta ly tic . Like the sin g le oxides and

m etal/m etal oxides examined, they a l l conformed to the general pattern, already

discu ssed , of s lig h t ly higher E^ value with in creasin g temperature of the

c a ta ly st (The Mr^O^ mixtures in some cases did not conform to th is behaviour).

CuO , had an average E^ = 4 .0 K .cals/m ol in the region 39-100°C; CuO.

Fe^O^ had E^ (a v g .) = 8 . 5 K .cals/m ol in the same temperature range; CuO. Al^O^,

a ctiv e in the 140-i79°C range, gave E^ = 15 K .cals/m ol, and unlike ^ -Al^O^

showed no tendency to lo se a c t iv it y on exposure to

We may consider the Co^O^/CuO and AlgG^/CuO mixtures to be sim ilar to

the MngO^/lfiO specimens in that s o lid solu tion s of the two oxides would be

formed giving an a c tiv e surface with many p ossib le electron sources to i n i t i a t e

decompositions (see sectio n 4(d) of r e s u lt s ) . On the other hand, CuO read ily

forms a sp in el with Fe20^ (F o restier, 117) and at the temperature of preparat­

ion ( 600 °C) may form a complex structure with some inversion of the p o sitio n

of di and tr iv a le n t ions y ield in g a highly d efect structure which is so

e f fe c tiv e in c a ta ly sis when compared with the individu al oxides. This provides

a p la in case of synergism and the e f fe c t may be to give r is e to s it e s requiring

a lower E^ than present in e ith e r oxide. Again, as in the case of 2h0/Mn,-,0^,

i t may be the junction between the cry sta ls which produces the most a ctiv e

c a ta ly s is . The alumina m ixtures provide the so le example of a d ilu e n t e f fe c t

and can be explained i f the alumina i s able to poison th e most a c tiv e s it e s

 - 191 -

on the CuO.

As sta ted e a r lie r in th is se c tio n , u n til more is known of the precise

e lectro n ic nature of these compounds, i t is not p ossib le to invoke a precise

d escrip tion o f the surface process without using tenucjs arguments, lh a t does

emerge from th is equimolar work i s that a combination of two p-type oxides or

two n-type oxides gives a b etter c a ta ly tic performance than the ind ividual

oxid es, w h ilst a combination of n + p or n + in su lator oxides is generally as

good, but not much b e tte r , a ca ta ly st than the b etter of the two individual

components.

(9) Surface area and magnetic measurements.

Regarding the surface areas i t may ju s t ifia b ly be concluded that as long

as the roughness fa cto rs of a given se r ie s do not vaiy much, i . e . by a fa cto r

of 0.5**2, surface area measurements w ill not be of much sig n ifica n ce unless

some correla tio n can be made between th is and the "active" area; which is

d if f ic u lt to achieve experim entally.

As regards the actu al method used to fin d the surface areas, i t was f e l t

that the point "B" method would s u ffic e to give the required data and t h is ,

of course, involves an error of s e le c tio n of "B". However, as has been pointed

out by Halsey ( 1 6 9 ) and Jacobs and Tompkins (l6 8 ), the B.E.T. plot of p/v

(pQ-p ) against p/pq i s now regarded as simply a convenient method for lo ca tin g

"B" and, in f a c t, Tompkins considers i t to be le s s v a lid than the method used

in th is work. The actual surface area r e su lts obtained here are regarded

merely as providing evidence fo r the sim ila r ity in roughness of the ca ta ly sts

examined.

The o rig in a l object in studying the magnetic s u s c e p tib ility was that i t
 - 192 -

was hoped to follow s u s c e p tib ility change, i f any, during c a t a ly s is , and the

apparatus was constructed w ith th is in mind. However, i t was soon apparent

th at such an in v e stig a tio n would i t s e l f be a long-term p roject, and i t would

not be p o ssib le to accomplish th is w h ilst making a k in e tic in v estig a tio n of a

number o f c a ta ly sts.

The magnetic s u s c e p t ib ilit ie s were then used to determine sample purity

and to give an in d ica tio n o f stru ctu ral changes induced by progressive sin te r in g .

The r e s u lts proved to be very u sefu l in th is resp ect.

-6
The magnetic s u s c e p tib ility o f ^ 2^3 was ^oun(i ‘to ^>e 89.0 x 10~ cgs.

u n its per gm., or 5«0 x 10■-3 cgs. u n its per gm. ion of manganese. With addition

o f equimolar r a tio s o f a d ivalent diamagnetic ion the s u s c e p tib ility for man­

ganese ion was increased, e .g . with lead oxide (300°C) the value i s 8 .0 6 x 10 ^

and fo r ZnO addition the peak value reached at 600 C was 11*74 x 10 cgs.

u n its . These high values were sharply decreased when the temperature of tr e a t­

ment was high enough fo r sin te r in g . Thus at 7^0°C the ZfoO/Mr^O^ s u s c e p tib ility

had f a lle n to 4 .2 6 x 10*"3 cg s. u n its per Mn ion .

Paramagnetic s u s c e p tib ility is a ssociated w ith electrons of unpaired sp in .

Mn3+ has 4 (3&/

/ \
electron s which would be unpaired, Mn-H- has 5> aud Mi4+ has 3 .

In increase of s u s c e p tib ility per Mn ion must be considered to a r ise from

reaction s in which the diamagnetic oxide gains or lo se s an electron perm itting

a contribution to the to ta l s u s c e p tib ility . These unpaired electron s may not

n e c e ssa r ily be associated with the metal ions but attached to io n ic vacancies,

e .g . as F - centres held in the diamagnetic oxide. On the other hand, simple

34-
donation o f an electron from an i n t e r s t i t i a l p o sitio n to Ivh could occur,

converting i t to the more su scep tib le Mn-H-•

 - 193 -

Again, in the absence of a complete a n a ly sis of an oxide system as a

semi-conductor in order to id e n tify l a t t ic e and electron im perfections and the

energy le v e ls o f trapped electron s or p o sitiv e h oles; the relation sh ip between

magnetic s u s c e p tib ility and c a ta ly s is can only be considered in a q u a lita tiv e

manner. Among the b est c a ta ly sts of the AO/Mo^O^ s e r ie s were those with the

highest paramagnetic s u s c e p t ib ilit y , i . e . the highest concentration of unpaired

electron s per Ma io n , e .g . PbO/MngO^ at 300°C and ZnO/ySo.^0^ at 600°C. These

unpaired electron s are c le a r ly a ssociated with c a t a ly tic a lly a ctiv e s it e s but

s u s c e p tib ility measurements t e l l nothing of the s t a b ilit y or r e a c tiv ity of these

s it e s . Some highly su scep tib le oxides may have a high proportion of the un­

paired electron s in t ig h t ly held and non-reactive p o sitio n s, and therefore be

poor c a ta ly s ts . However, i t i s quite c le a r ly esta b lish ed that s u s c e p tib ility

measurements are a u sefu l quide to the e f fe c t of temperature and other pre­

treatments in a ffe c tin g c a ta ly tic p roperties; e s p e c ia lly when dealing with

oxides of c lo s e ly rela ted typ es.

(10) Conclusion o

This th e s is describes work which has, fo r the f i r s t tim e, been directed

towards elu cid a tin g the mode of decomposition of hydrogen peroxide vapour on

a wide range of m e ta llic oxide surfaces in a flow system. The r e su lts show

that there appears to be a lin k between ca ta ly st e ffic ie n c y and electro n ic

structure of the oxides examined. To th is end, m odifications of the Fehlordnung

of n ick el oxide by the ad dition o f a lte r -v a le n t ions has yield ed substantive

data.

In ad d ition , the k in e tic s of the decomposition have been evaluated showing

the dependance of the reaction rate on the decomposition products. A p o ssib le

reaction mechanism and rate-determ ining step have been proposed.

 - 194 -

Certain physical and stru ctu ral properties have been in v estig a ted and the

r e s u lts c r i t i c a l l y applied to appraise the nature of certain c a ta ly st systems

whose preparations were c lo s e ly con trolled . Emphasis has been la id on the

n e c e s sity of having adequate knowledge o f the previous h isto ry of a ca ta ly st

in order to obtain reproducible r e s u lts and on the e s se n tia l need -to excercise

care in s e le c tin g cleaning f lu id s .

I t i s f e l t that the information emerging from th is work w ill contribute

to our expanding knowledge of c a ta ly st surface processes, which have been rather

n eglected in the p a st, when most a tten tio n has been directed at evaluating the

gas phase k in e tic s . Certain a sp ects, lik e sem i-conductivity and magnetic

s u s c e p t ib ilit y changes of the c a ta ly st during reaction , w ill bear in v e stig a tio n

and i t i s hoped that these avenues when f u lly explored w ill add further to our

knowledge o f heterogeneous c a t a ly s is .
REFEHEICES
/
(1) Thenard, Ann. Chim et p h y s., 1818, J3, 306.

(2) Thenard, Mem. Acad. S c i ., P aris, 1818, £> 345*

(3) The'nard & Thenard, Compt. rend., 1855> 341.

(4 ) Thenard & Thenard, Ann. Chim et p h ys., I 8 5 6 , 173*

(5 ) Bredig, HAnorganische Fermente", Engelmann, L eip zig, 1901.

(6) T e le to f, I .S . & G ritsan, B u ll. Khim. Obshchestva im Mendeleeva

1940, JJ, 22 (Chemical Abstracts 3^59)•

(7 ) S e itz , Revs. Mod. Phys. 1946, 18. 384*

(8) Haber & W illsta te r , B e r ., 1931, 6 4 , 2844.

(9 ) Haber & Weiss, Proc. Roy. Soc. A ., 1932, l^Z*332.

(10) Baxendale, Evans & Park, Trans. Thr. S o c., 1946, 4 2 , 155.

(11) Dainton, J. Phys. & C olloid . Chem., 1948, ^2, 490*

(12) Barb, Baxendale, George & Hargrave, Trans. Far. S o c., 1951, £ l

(13) Weiss, Ear. Soc. D is c ., 1947, 2., 212.

(14) Evans, Hush and U ri, Quart. Revs. Chem. S o c., 1952, jJ, 186.

(1 5 ) Baxendale, Evans & Uri. Trans. Far S o c., 1949, 4£, 236.

(16) Weiss, ib id , 1935, JJ1, 1547.

(17) Bowden & Reynolds, Far. Soc. B is c ., 1950,, ,8,184 .

(18) W iegel, 4. Phys. Chem., 1929, 143A. 81.

(19) Broughton & Wentworth, J. Amer. Chem. S o c., 1947 , 69 . 741.

(20) Broughton, Wentworth & Farnsworth, i b i d ., 1949, 71. 2346.

(21) Voltz & W eller, i b i d ., 1954, J 6 , 1586.

(22) Mooi & Selwood, ib id , 1952, J 4 , 1750.

(2 3 ) Schwab, La Chimica e Industria, 1953, 810.

 - 196 -

BEEEBEHCES (C ontinued)

(24 Hart, Nature, 1949, l 6 £» 876.

(25 S a tte r fie ld , C eccotti & Feldbrugge. Ind. Eng. Chem., 1950, 4 £ , 1040.

(26 S a tte r fie ld , and S tein , J. Phys. Chem., 1957, 61. 537*

(27 Hauser, B er., 1923, ,£6 , 888 .

(28 Hinshelwood & Pritchard, J. Chem. S o c., 1923, 123, 2725*

(29 Elder & B ideal, Trans. Far. S o c., 1927, 25. 545*

(50 Rosenberg & Eistiakawsky, J. Amer. Chem. S o c., 1937, 59. 422.

(31 Reuter & Gaukhman, J. Phys. Chem. (T3.S.S.R.) , 1933, 465*

(32 Mackenzie & R itch ie , Proo. Roy. Soc. A ., 1946, 18£, 207*

(33 Baker & O^iellet, Can. J. R es., 1945, 2^j>B, 167.

(34 Giguere, ib id , 1947, ^ B , 135

(35 McLane, J. Chem. Phys., 1949, ±1, 379.

(36 H arris, Trans. Far. S o c., 1948, 44, 764.

(37 Giguere & Liu, Can. J. Chem., 1957, 35. 283*

(38 Mott & Gurney, "Electronic Processes in Ionic C rystals”, Oxford, 1950.

(39 Schockley, "Electrons & Holes in Semi-Conductors", van Nostrand,

New York, 1951.

(40 Anderson, Ann. Reps. Chem. S o c., 1946, 104.

(41 Cabrera & Mott, Rep. Prog. Phys., 1949, 12, 163 .

(42 Wagner, J. Chem. Phys., 1950, ^ 8 , 62.

(43 Gray, "Chemistry o f the S olid State", ed. Garner, Butterworths,

London, 1955, PP.133*

(44 Garner, Stone & T iley , Proc. Roy Soc. A ., 1952, 211. 472.

(45 D e ll, Stone & T ile y , Trans. Far. S oc., 1953, 42, 201.
 - 197 -

REEEEENCES (Continued)

(46 Gamer, Gray & Stone, Far. Soc. D is c ., 1950, J3, 246.

(47 Gamer, Gray & Stone, Proc. Roy. Soc. A ., 1949, 197 . 294.

(48 Wagner & Hauffe, Z. Electrochem ., 1938, 44, 172.

(49 Gray, Proc. Roy. Soc. A ., 1949, 197 . 314.

(50 Gray & Savage, Far Soc. D is c ., 1950, £ , 250.

(51 Hauffe, Glang & E n g ell, Z. Phys. Chem., 1952, 201. 223*

(52 D ell & Stone, Trans. Far. S oc., 1954, ,£0, 501.

(53 Parravano, J. Amer. Chem. S o c ., 1953, 1443.

(54 Parravano, ib id , 1953, 75. 1452.

(55 Schwab & Block, Z. Phys. Chem., (N .F .)., 1954, i , 42.

(56 Stone, r e f . (43) PP.399*

(57 Verwey, "Semi-Conductors", Butterworths, London, 1951, PP-151.

(58 Wagner, J. Chem. Phys., 1950, 69.

(59 Molinari & Parravano, J. Amer. Chem. Soc, 1953, i a , 5233 .

(60 Heckelsberg, Clark & B a iley , J. Phys. Chem., 1956, 60, 559*

(61 Parravano, J. Amer. Chem. S o c., 1952, 1194-

(62 Gray & Darby, J. Phys. Chem., 1956, 60, 209.

(63 Pink, Ann. Reps. Chem. S oc., 1949, 46 , 34.

(64 Emmett, "Catalysis 1", Reinhold, New York, 1954, PP*31«

(65 H uttig, K olloid - Z ., 1 9 4 1 , £ 4 , 258.

(66 H uttig, ib id , 1942, 95, 262.

(67 Selwood, Chem. R evs., 1946, ^8, 41.

(68 Selwood, r e f . ( 6 4 ) , pp.353*

(69 Selwood, Moore, E llis & Wethington, J. Amer. Chem. S oc., 1949, X , 693 .
 - 198 -

REFERENCES (Continued)

(70) Selwood, H ill & Boardmann, ib id , 1946, 68, 2058.

(71) Stroupe, ib id , 1949, 5^9*

(7 2 ) H uttig, Eunice & K itt e l, ib id , 1935, 5X» 2470*

(7 3 ) H uttig, Ear. Soc. D is c ., 1950, 8,, 215.

(7 4 ) Brunauer, Emmett & T e lle r , J. Amer. Chem. S o c ., 1938, 60, 309.

(75) Emmett & Brunauer, ib id , 1937, 52> 1553*

(76) Duncan, Trans. Ear. S o c., 1949, 45. 879*

(77) Wooten & Callaway - Brown, J. Amer. Chem. S o c., 1943, 6 5 , 113.

(78) Taylor, H .S ., ib id , 1931, j g , 578.

(79) Volkenshtein, 2hur. P iz . Khim, 1949, 917*

(80) Boudart, J. Chem. Phys., 1950, 18, 571.

(81) Hart & Mc.Padyen, In course of p u b lication , Glasgow.

(82) Constable, Proc. Roy. Soc. A ., 1925, ^ 8 , 355*

(83) Parkas & M e lv ille , "Experimental Methods in Gas Reactions”,

McMillan, London, 1939, PP. 102.

(84) Kavanagh, Scatchard, and Ticknor, J. Amer. Chem. Soc. 1 9 5 2 , 2 4 , 3715.

(85) Pinch, Quarrell & Wilman, Trans. Par. S o c ., 1935, 1051

(86) Trapnell, ib id , 1955, 3^8.

(87) Sucksmith, P h il. Mag., 1929, .8, 158.

(88) D ilk e, Eley & Maxted, Nature, 1949, 161. 804.

(89) Beeck, Smith & Wheeler, Proc. Roy. Soc. A ., 1941, ,!][£» 67 .

(90) Tyner, J. Chem. S o c., 1 9 H , 22,9 -1-^53.

(91) Giguere & G eoffrion, Can. J. R es., 1949, 2J8, 168.

(92 > Easton, M itchell & Wynne - Jones, Trans. Par. Soc., 1952, 48, 79&
 - 199 -

REFERENCES (Continued)

(93) Kelson, J. Chem, Phys., 1937, £ , 252.

(94) Rhodin, J. Amer. Chem. S o c., 1950, 5102.

(95) R ef. 58, pp.262.

(96) N e ttle to n , T hesis, Royal College of Science & Technology, Glasgow, 1953.

(97) H uttig & Heinz, Z. Anorg. Chem., 1948 , 2££, 223*

(98) Nizboer, Proc. Phys. S o c., 1939, ^1, 575*

(99) Leblanc & Sachse, Z. Eleetrochem ., 1926, J>2, 59 & 204#

(100) Rooks by, Nature, 1943, 152, 304.

(lG l) Bennett, Cairns & O tt, J. Amer. Chem. S o c., 1931, 53,, H79»

(102) Day & H ill, J. Phys. Chem., 1953, 2L> 946.

(103) Howell, Proc. Roy. S o c., 1923, 104. 134.

(104) Suhrmann & Schulz, Z. Phys. Chem., (N .F .), 1954, 69.

(105) Rideal & Trapnell, Proc. Roy. Soc. A ., 1951, 205. 409.

(106) Ehodin, J. Amer. Chem. S o c., 1951, j^» 3143*

(107) KLemm & Hass, Z. Anorg. Chem., 1934, 219. 82.

(108) Selwood, "Magneto chemistry", In terscien ce, Hew York,1956,PP-313*

(109) Howell, J. Chem. S o c ., 1923, 125. 669 .

(110) Pinch & Sinha, Trans. Par. S oc., 1957, 823*

(111) Bhatnagar & Bal, J. Ind. Chem. S oc., 1934, 1A, 603.

(112) Perakis, Serres, Parravano & Wucher, Compt. Rend., 1956,242, 1275*

(113) W ells, ‘'Structural Inorganic Chemistry", Oxford, 1950, pp.381.

(114) Greenwald, Pickart & Grannis, J. Chem. Phys., 1954, 22, 1597*

(115) Gorgeu, B u ll. Soc. Chim., 1903, 29_, 1111.

(11 6 ) C a ste ll, Dilnot & Warrington, Nature, 1944, 155. 6 53*
 - 200 -

B M M C B S (Continued)

(117 F o r estier, Compt. Rend,, 1931, 192, 842.

(118 De Boer & Yerwey, Proc. Phys. S oc., 1937, 42> 59.

(119 Fensham, J. Amer. Chem., 1954, 1 6, 9 69 .

(120 Schwab & P ietsc h , Z. Phys. Chem., 1926, P21, 189*

(121 Palmer & Constable, Proc. Boy. Soc. A ., 1924, 106 . 250.

(122 Twigg, Trans. Far. S o c., 1946, 4 2 , 284.

(123 Kulkova & Temkin, Zhur F iz. Khim., 1949, 2£, 695 .

(124 Darby & Kemball, Trans. Far. S o c., 1957, 22.* 832#

(125 Beeck, Far. Soc. D is c ., 1950, JU 118.

(126 Kemball, Proc. Roy. Soc. A ., 1952, 214. 413«

(127 Couper & Eley, Far. Soc. D is c ., 1950, J3, 118

(128 Dowden, J. Chem. S o c., 1950, 242.

(129 Trapnell, Proc. Roy. Soc. A ., 1953, 218, 566 .

(130 Gray and Darby, J. Phys. Chem., 1956, 60, 201.

(131 Boudart, J. Amer. Chem. S oc., 1952, 74. 1531*

(132 Weisz, J. Chem. Phys., 1953, 21, 1531.

(133 Taylor and Thon, J. Amer. Chem. S o c., 1953, 75. 2747*

(134 Twigg, Proc. Roy. Soc. A ., 1948, 188, 92.

(135 L aidler, r e f . ( 6 4 ) pp. 19 0 .

(136 Roberts, Proc. Roy. Soc. A ., 1935, 152. 445*

(137 Trapnell, ib id , 1951, 206 , 39*

(138 Smith and Taylor, J. Amer. Chem. S o c., 1938, 60, 362.

(139 R oginskii & Tselinskaya, Zhur. F iz. Khim., 1948, 22, 1350.
 - 201 -

REEEKBNCES (Continued)

(140 Tu, Da-vis and H o ttel, Ind. Eng. Chem., 1934, 26, 7 4 9 .

(141 Katz, Riberdy and Grant, Can J. Chem., 1956 , 1719 .

(142 S a tte r fie ld , Resnick and ?/entworth, Chem. Eng. Prog., 1954, 50. 4 6 O.

(143 2immerman, J. Phys. Chem., 1949> 562.

(144 R iddiford, ih id , 1952, J 6 , 745.

(145 P ole, Rudd, Michow and Comte, J. Chem. P hys., 1952, 20* 961.

(146 Brodie, P h il. Trans., I 8 6 5 , 1SE> ®37*

(147 Wright and Andrews, Proc. Phys. S o c., 1949> 446.

(148 Rideal and W ilkins, Proc. Roy. Soc. A ., 1930, 394*

(149 Mott, Trans. Ear. S o c., 1940, j 6., 472.

(150 Vernon, ib id , 1927> 25. 152.

(151 Hart, R.K., ib id , 1957, 1020.

(152 Trapnell, Quart. Revs. Chem. S oc., 1954, ,8 , 404.

(153 Rideal and Trapnell, Par. Soc. D is c ., 1950, J3, 114*

(154 Taylor and Liang, J. Amer. Chem. S oc., 1947» 6 9 . 1306.

(155 Barrer, J. Chim. Phys., 1950 , 47 > 444.

(156 Cremer & M arschall, Monatsch, 1951» ,82, 840.

(157 Mikovsky and Waters, quoted by Cremer, Advances in C a ta ly sis, VII,

Academic P ress, Hew York, 1955 > pp.81.

(158 Cremer, Advances in C atalysis VII, Academic Press; Hew York, 1955> PP«75<

(159 Engell and Hauffe, 2 . Electrochem ., 1953> 57 > 773*

(160 Temkin & Pyzhev, Acta Physicochim, 1940, 12, 327*

(l6 l JKtaui), J. Res. I n s t. C a ta ly sis, 1953 > J.? 16.

(162 Ewan, J. Amer. Chem. S oc., 1956, 60, 1033*

 - 202 -

REFERENCES (Continued)

(163) Elovich & Zhabrova, Zhur. F iz . Khim, 1939, 1^*1716, 1775*

(164) Stone, r e f. (43)> PP*385»

(1 6 5 ) Hinshelwood, "Kinetics o f Chemical Change”, Oxford, 1949* pp#2l6.

(166) Frenkel, Phys. R evs., 1931* 1276.

(1 6 7) Anderson, Far. Soc. P i s e ., 1950, £, 305*

(1 6 8 ) Jacobs & Tompkins, "Chemistry of S o lid S ta te”, Ed. Garner,Butterworths,

London, 1955, PP*95*

(169) Halsey, Far. Soc. P i s e ., 1950, j8, 54*