Chemisorption of nitrogen on iron catalysts in connection with ammonia synthesis

CHEMISORPTION OF NITROGEN ON

IRON CATALYSTS IN CONNECTION

WITH AMl\IONIA SYNTHESIS

PROEFSCHRIFT

TER VERKRIJGING VAN DE GRAAD VAN DOCTOR IN DE TECHNISCHE WETENSCHAP

AAN DE TECHNISCHE HOGESCHOOL TE DELFT, OP GEZAG VAN DE RECTOR MAGNIFICUS DR. O. BOTTEMA,

HOOGLERAAR IN DE AFDELING DER ALGEMENE WETENSCHAPPEN, VOOR EEN COMMISSIE UIT DE SENAAT TE VERDEDIGEN

OP WOENSDAG 1 JULI 1959 DES NAMIDDAGS TE 4 UUR

DOOR

JOSEPH JOHANNES FRANCISCUS SCHOLTEN

GEBOREN TE AMSTERDAM.

11)/2

DRUK-, V.B.B., nEINB DER A' • GRONINGEN

-DIT PROEFSCHRIFT IS GOEDGEKEURD DOOR DE PROMOTOR PROF. DR. J. H. DE BOER

CONTENTS

INTRODUCTION

1. Objective

2. Chemisorption and catalysis 3. Division of the thesis

CHAPTER 1. CRITICAL SURVEY OF THE WORK OF PREVIOUS INVESTIGATORS

A . . Preparation and reduction of the catalyst B. Surface structure of the reduced catalyst C. Kinetics of nitrogen chemisorption on iron

catalysts and films

D. Nitrogen chemisorption isotherms and heat chemisorption

E. Hydrogen chemisorption and hydrogen-deuterium exchange on iron catalysts F. The state of the chemisorbed nitrogen G. Kinetics of ammonia synthesis

CHAPTER 1I. APPARATUS AND METHOD OF EXPERIMENT ATION

A. Introduction

B. The high vacuum balance C. Buoyancy corrections D. Flow corrections

E. Method of experimentation

F. :purification of gases and preparation of the catalyst 1 1 2 6 9 9 10 22 of 29 34 37 38 42 42 42 44 47

52

56 CHAPTER lIl. REDUCTION OF THE CATÀLYST AND DETERMINATION OF THE SURFACE STRUCTURE 559

A. Reduction 59

CHAPTER IV~ THE KINETICS OF THE ADSORPTION

AND DESORPTION OF NITROGEN 70

A. Introduction 70

B. Adsorption kinetic s 70

C. Desorption kinetics 77

D. Discussion of the results 81

CHAPTER V. THE RATE-DETERMINING STEP IN

AMMONIA SYNTHESIS 107

A. Introduction

B. Method of experimentation C. Analysis of the experiments,

results D. Discussion 107 108 and experimental 109 118 APPENDIX - Examination into a possible retar

-dation of the chemisorption and reaction rates

owing to diffusion 122

SUMMARY 130

SAMENVATTING 135

LITERATURE 141

CHEMISORFTION OF NITROGEN ON IRON CATALYST IN CONNECTION WITH AMMONIA SYNTHESIS

J. J. F. Scholten

ERRATA

p. 4 _{Line 8 from bottom; for 16}

_g-2

_{read: 16.2}

_g-2.

p. 9 _{First sentence section A; read: The starting material. ...} p. 18 _{At the bottom of this page part of a sentence has been}

p. 20 p. 39 p. 41 p. 50 p. 58 p. 76 p. 82 p. 85 p. 87 p. 88 p. 93 p. 96 p. 99 omitted; read: ₀

, could be desorbed completely at 500 C (see Ch. IV). The amount of H_{2 dissolved in Fe after evacuation at} 5000_{C is likewise extremely low.}

Table IE, column 4; read: m 2/ g. Line 2 from bottom; read: Ct ::: g!h

The equation following (1-24) should be numbered(I-25). Line 4 from top; read:

Frankenburg remarked th at Ct can also be calcu-lated from ... .

The equation should be numbered (11-8). Line 2 from bottom; read: traces: Cr and Co Equation (1-5) should read:

~f:::

22.3 expo (107.20/ R) expo (-9200-727000)/ RT.

I Equation (IV-15); for fj read: fj.

Equations (IV-26) and (IV-28)a should be interchanged. Equation (IV - 31); for

r;

read :

iJ.

Table IV Ct , last column; read:-f1 a

Line 2 from top; Boltmann constant, read: Boltzmann constant.

Line 13 from top; for "change" read: "charge".

Line 12 from top; read: were 0 ::: _{0.1 at room}

temper-ature and (J ::: 0.2 at 100o C. p. 102 _{Under equation (IV-55), read:}

cr, the symmetry number, ::: 2. p. 105 _{Equation (1-23) should read: g}

Ct : : : -Tem. g+h Equation (1-12) should read:

(dO.

p. 105 Equation (1-13) should read: dO

(dt) des. = kd' expo (h (J).

p. 111 The first equation from the bottom should read:

p. 118 p. 120 p. 124 p. 129 p. 129 (gN)(J -gH 2 = 1 2 7

The equation should be numbered (V-7). Below equation (V-l3), read:

By filling in P N ~~ . . . .

2 .

Equation (A4); for Db read: Dk' Equation (A 7) should re ad :

~to=

21.9 P_N expo (132.4 OjR) expo (-132.90jR) -1

2 expo (-5250j RT)min

Line 18 from top; read: Equation (IV - 8), describing the

1

IN'T'RODUCTION

1. Objective

In his book "Geschichte der Ammoniaksynthese "1 , A1win Mittasch accentuates the fact that an industria1 ammonia synthesis was rendered possible on the one hand by the fundamental research of Ostwald, Nernst and Haber in the field of gas equilibria, and on the other by the laborious. empirical work carried out in order to find. a suitab1e c ata'fyst.

As early as 1911 the synthesis of ammonia on a techni-cal stechni-cale was started in Ludwigsjlafen (Germany), but at present~ after almost fifty years, we have to acknow-1edge the fac t that our understanding!,of the mechanism

and the kinetics of the reaction is stillfar from complete. Hence, a fundamental study of this synthesis rema:ins important, not only as a foundation for further perfec-tion of the process, but also as a means of obtaining a better insight into heterogeneous catalytic processes in general.

In volume 111 of the well-known handbook "Catalysis",

~dited by P. H. Emmett, W. G. Frankenburg 2 gives a detailed summary of the work performed in the field of ammonia synthesis from 1820 onwards. As appears fr om a recapitu1ation of the research since 1940, published in the same book 3, much interest in this subject has existed a1so oflate years. This recapitu1ation, compiled by re-searchworkers of the cata1ysis department of the"Central Laboratory" of "Staatsmijnen" (the chemical research centre of the Nether1ands State Mines), gives also a resumé of the work carried' out in the Netherlands. The investigation described in the present thesis is a continuation of the work just meIltioned, and relates in particu1ar to kinetic problems.

It is commonly accepted that the chemisorption rate of nitrogen on the ironcatalyst is the rate-determining step of the ove r- all reac tion. About this chemisorption proces itself we have only incomplete information, and it has been one of our aims togive a further extension of our know1edge of this reac tion- step. In addition, also the kinetics of the reverse process, viz. the de-sorption of nitrogen, was studied, a subject which had previous1y attracted only little attention. The last part of this thesis deals with measurements providing an

...

2

insight into the connec tion between ni trogen chemisorption on iron and the synthesis of ammonia ..

2. Chemisorption and c atalysis

It is impossible to give a brief outline of the complicated field of heterogeneous catalysis and chemisorption on metals. Therefore we will restrict ourselves to giving a short treatise on some important concepts and methods of experimentation,by;which the non- specialist reader should be enable to follow our argumentation; the various con-cepts will be illustrated with examples from ammonia synthesis. In the discussions at the end of each chapter we will enter more deeply into the problems.

In studying heterogeneous catalysis 'we have to deal with processes which take place at the boundary of a (gene rally solid) catalyst and a gaseous or liquid phase coniaining the reaction components. In most cases it will be necessary for one or two of the components to be adsorbed on the catalyst surface by chemical forces, the same forces that play a part in normal compounds. As a result of c •• emi~orption the molecules become more disposed to react, or, in other words, the activation energy of the over- all reaction is lowered.

This effect is well illustrated by the catalytic action of iron in ammonia synthesis. From the equilibrium

N2 + 3 H2 ~2 NH3 + 24 Kcal

it is clear that the reaction producing ammonia is pro-moted by the· application of a high pressure and a not-too:"high reaction temperature.

For example, at 300 Atm. and 2000 _{C the equilibrium}

mixture contains about 90% NH3.However, because of the high energy of activation needed for making the reaction proceed to the right (at least 100 Kcal/mol N 2), such an equilibrium is never reached. The only way to effect this is drastically to lower the ~ctivation energy. If iron is used as the catalyst this energy amounts te> only 25-30 Kcal/mól.

The actionof the c~talyst is to be understood as follows: The extreme slowness of the reaction between nitrogen

3

and hydrogen if no c atalyst is used is due to the chemie al inertia of the nitrogenmolecule .. However, in the pres en-ceof iron the nitrogenis chemisorbed on the iron surface~

This proce~s requires only a'r~ther low activation energy, viz. about 30 Kcal/mol. The nitrogen bound on the iron surface has weakened orperhaps completely broken N-N bonds. However, the iron-nitrogen bond is not very strong, as an activation lower than 20 Kcal/ mol suffices forre-ducing the resulting surface complex to ammonia.

Diagrams of the energy levels are given in Fig. a., both for iron and vanadium.

'RON "itralton N2·3H, ( s) ~ yanodl\lmnJtrld. YANADltJtroII

Fig. a. (on the analogy of' the schemes given by Frankenburg2 . Possible diagrams of the energy levels (activation ~ner­ gies and heats of reaction) in ammonia synthesis, with iron and vanadiUm. as catalysts.

The numbers denote Kcal. per mole of N~.

As will beclear from this picture, it is essential for an efficient ammoniacatalyst that the bond it produces be-tween nitrogen and metal be not too strong; in the c,ase ofvanadium the nitrogen-metal bond is so strong,that the hydrogenation requires an activation energy of 106 Kcal/mol N2. Moreover. vanadium forms a so-called "bulk-nitride", whereas, under the samecircumstances, iron produces a "surface-nitride". Somewhat paradoxial-ly it may be said that iron isa good cataparadoxial-lyst for the synthesis of ammonIa owing to the fact that it shows .'little tendency to nitride formation.

If theactivityofacatalystisto be used to full adv~ntage

the .surface-to-weight ratio ' of this catalyst must,be

prepa-4

ration techniques. e. g. decarboxilation or dehydration of compounds by which pores and/ or cracks are formed. In the case ofironcatalysts for the synthesis of ammonia

. a large surfac e. of sufficient thermostability is obtained by ~tarting from magnetite to which aluminium oxide is added as a "promoter". As there are many types of promotion in catalysis the term "promoter" is difficult to define. The promoting action of the aluminium oxide . will be dealt with in chapter lIl.

A method -now in common use- of measuring the surface areaofa catalyst by means of low-temperature gas ad-sorption was originated by the Amerîcan investiga.tors Brunauef' Emmett and Teller (the socalled B. E. T. -method) . An analysis of the mechanism of physical multilayer adsorption enabled them to draw up an equa-ti on by' means of which the volume adsorbed in a mono-. layer. Vm • can be computed from the adsorption iso-therm. This equation has the following shape:

= (1 ).

where V is the total volume adsorbed at pressure p; Vm is the volume adsorbed in a monolayer;

Po is the saturation pressure of the adsorbed gas;

c is a constant. related exponentially to the heat of adsorption and the heat of liquefaction of the ,adsorbate.

From Vm wecan calculate the surface area of the cata-lyst if the cross section of the molecules in the first layer iskriown; this cross sec tion c an be c alculated. e. g .• from the density of the liquefied adsorbate.

For the molecularcross sectionofnitrogenO'N' whichis often used for the determinati.on of surface areas. Emmett· and Brunauer5)j' found . 16.zR-2 from the density of liquefied nitrogEm at - 1960 _C.

However. H. K. Livingston6 prefers the valtie of 15. 4,R -2 for the cross- sectional area of nitrogen. which is i;n accordance both with the numbe_rfound from a calori-m~tric. non-adsorption determination by Harkins.. and Jura7• andwithacomputationofO':N'from the two-dimen-sional van der Waals constant b:8

Another important factor in determining the suitability

of acatalystis its'pore size distrilJution.1t isdesirable

that the greatest part of the catalyst surface be formed by pores of such dimensions that the transport of the reaction components is not retarded by a slow diffusion through the pores. An elegant method of determining pore size distributions was elaborated by Barrett c. s. 9. By a careful analysis of thedesorption of a gas physically ad-sorbed in multilayers and condensed by capillary conden-sation, these investigators succeeded in finding a method for computing the pore size distribution from the measur-ed desorption isotherm.

,

Another procedure, specially suited for the determination

of the distribution of pores with radii above c. 25

)?,

was

first described by Ritter and Drake 10 . Here, the pore size distribution is found from the volume of mercury that penetrates into the pores at pressures varying from

one to . three thousand atmospheres. By way of example

Fig. b.shows the pore size distribution of a doubly

pro-moted ironcatalyst measured by Z'wietering and KOksl l ,

who found a non- symmetrical distribution with a peak at

r

=

1-20 .R~ Fig. b. I I , , I I 100 200 300

Pore size 'distribution of an iron eatalyst eontaining:

2.90;0 Al203 and 1.1% K20.

V = mereury-filled volume in em3 r = pore radius inR.

Besides determining the surface area and the pore struc-ture it is of ten necessary to be able to distinguish diffe-rent groups on the surface, as it is possible that only a

6

specified part of the surface is catalytically active. Emmett and Brunauer5 showed that small quantities of aluminium oxide pr~sentin anironcatalyst as promoter, accumulate on the surface in unexpectedly high concen'-tration. They succeeded in determining the extent' of the catalytically active part of the surface area (free" iron surface) from the amount of carbon monoxide chemical-ly bound by it. A; detailed description of this method will be given in chapter 1.

In mentioning the above methods we wanted only to refer to some technique s which are avaible to the inve stigator in lhe development of technically useful catalysts, and also in fundamental catalysis- research.

Moreover, to obtain adeeper insight into catalytic reac-tions, it is of ten necessary to have a more exact know-ledge of the kinetics of the chemisorption of the reactants, and also of the heat of chemisorption, Q. In.both cases the "degree of occupation"O(also: coverage) plays a im-portant role. 0 is defined as the ratio of the number of atoms or molecules adsorbed on the surface, to the maximum numbèr that can be adsorbed.

Inmanycases aroughly linear decrease of Q with in-creasingcoverage Oisfound. De Boerl~ and Boudart13 have given a theoretical basis to this phenómenon. How-ever, the relation between Q and 0 isoften more I

compli-cated. and linearity is c ertainly -not a general rule; if diff~rentcrystallografic planes are exposed and there is the possibility th at the adsorbed atoms or molecules are bound in different ways, a simple relation is not. to be éxpected.

-Anothercomplicationis that the behaviour ofevaporated me tal films, of ten used in fundament al reaearch, differs considerably from that of catalysts of the same metal. One of the reasons may be that different crystallographic planes are exposed or that a promoter influences the properties of the metal; ho wever this may be, .it is cer-tain that the fact that a higher degree of purity can be reached in the preparation of films is the most important cause.

In chapter I we wil! illustrate the difference between

. films and catalysts as regards the chemisorption ofni-trogen and hydrogen on iron. .

3. Division of the the sis

re-7

searches on ammonia synthesis which are important in connection with our own·work. The limited value of Em-met and Brunauer' s Em-methpd for the determination of the free iron surface of iron catalysts is :demonstrated. It is shown that Kummer ,and Emmett' s determination of the amount of hydrogen left on th~. surface of an iron catalyst af ter reduction and degassing at c. 5000 C, was probably a determination of the hydroxyl-groups on eh,~

promoter surface. Finally it is demonstrated that the Elovich equation gives a better description of the adsorp-tion rate of nitrogen on iron catalysts than does the "power rate law"favoured byKwan, and that the chemi-sorptionisotherms, also determined by this author, are better described by an equation of the Frumkin- Slyghin

type then by a Freundlich- equation.

Chapter II gives ~description of the special gravimetrie

aRpar~üus wl'!,ich was used in our investigations. It is ex-p'erime.ntally proved, that the flow correc tions whic hare necessary when working with streaming gases, are described by Poisseuille's ·,law for a tube 01 annular section. A description is given of the preparation of the catalyst and the purification of the gases used.

Chapter lil deals with the reduction of the catalyst and thedeterminationofits surface structure. By the use of an extremely long reduction time, acatalyst was prepared with amuchhigher activity towards nitrogen , chemisQrp-tion than reported hitherto in literature. A new method isproposed for the determination of the Al203 part of the catalyst surface.

Chapter IV deals with the adsorption and desorption kine-tics ofnitrogen both onhighly active catalyst samples and samples of lower ac tivity. It is shown ~that by the use of an intensive reduction method a catalyst was ob~ained

with a chemisorptive behaviour closely approaching that known to be exhibited by iron films. !Up t~ a nitrogen coverage of aboutO. 30 the : adsorption and desorption rates are rather weU described by Eyring' s "'absolute reaction rate theory". With the help ot' the :

chemisorp-,8

tion isotherms. found by equalizing the adsorption and desorption rates, a computation is made of the entro-py of the adsorbed nitrogen particle_~. From the change of the entropy with coverage it follows, ithat it is, likely that the first portion of êhemisorhed nitr~ogen is ' ato-mically bound on the surface. and that the binding state of nitrogen at higher coverage is a mobile predissocia-ted one.

Chapter V deals with synthesis experiments duringwhich

,the nitrogencoverage on the catalyst was gravimetrically determined. It is experimentally proved that nitrogen chemisorption is indeed the rate- determining step 'in ammonia synthesis. as was already supposed by Bru-nauer and Emmett. about twenty five years ago.

In the Appendix it is finally shown that the chemisorption and synthesis rates we measured we re not reta'rded by a to slow diffusion of the gases.

9

Chapter I

CruTICAL SURVEY OF THE WORK OF. PREVIOUS INVESTIGATORS

In the introduc ti on we have already pointed out th at quite recentlyextensive summaries ··of the research work car- . ried out in the field of ammonia synthesis were published. Hence we wil! here give a condensed survey· only, and accentuate those researches which have a bearing on our own work. In conjunc tion with this survey we will make some critical remarks which will be further elabor.ated in the following chapters.

A. Preparation and reduction of' the catalyst The starlng material is magnetite, prepared by burning pure Fe which is elec tric aily fused with small amounts of promoter components, the mQst important among which is A1203. If Al203 is used only we obtain what is mostly called a "singly promoted catalyst". In the fusing process the Al203 goes into solid solution with the magnetite; it is taken up in the form of 'Y _{Al 203 in the magnetite laUice,}

which has a ~pinel struc ture. 14 ~roz:n the research carried out by Hall, Tarn and Anderson we know that· during the reduc tion the acce ssible sur-face area increases linearly with the degree of reduc-tion, but decreases slightly during the removal of the very last traces of 02. The free iron surface area also shows a linear increase; however, when a high degree of reduc tion has been reached it extends more rapidly. Obviously, the last traces of 02 are mainlybound at the surface. Promo-ters have a retarding effect on the reduction rate, in particular those dissolving in the lattice; this effect in-creases with increasing promoter content. Hence it is possible that catalysts differing in promoter content, when reduced under exactly the same conditions, ultimately show different degrees of reduction. In chapter 111 the degree to which the last traces of 02 are removed from the surface wil! be shown to have a great influence on the activity of the catalyst. Perhaps this explaines Zwietering' s15 findings ·with a series' of singlypromoted catalysts, that the catalitic activity per square meter of free iron surface area was not constant, but decreased with

10

increasing amount of Al203. A second possible explanation is, however. that anincrease in the amouilt of Al203 has a negative influence on the intrinsic catalytic ac ti vity of iron. Latchinov and Vvedenskii16 record that the use of a high space veloci-ty during the reduction of the catalyst has a favourable effect on the catalytic activity. In this case the partial water pressure over the catalyst surface is low. and the retrogressive reaction of the equilibrium

will be slower'. Here also. the greater activity of the ca-talyst may be accounted for by the fact that a higher de:..

gree of reduction is obtaine.d.

A higher reduction temperature win also acceler<ate the

reduction rate, as it has a favourable effect on the

acti-vity. On the other hand, a rise in temperature results in a higher sintering rate of the catalyst so that a smaller surface area is obtained. Consequently, there win in general exist an optimum reduction temperature, varying

between 450 and 5500 _{C, dependirig on the catalyst type.}

Another important fac tor is the purity of the hydrogen used

in the reduction; if this hydrogen contains contaminants

that can be chemisorbed by the iron surface, a catalyst of

lower activity win be formed. If the H2 is contaminated

with ca O. \

10

~2, we get a catalyst surface occupied by

25-30% N2 1 . This will, of course, not influence the

catalitic activity, but N2-chemisorption rate measure-ments may be obscured by it.

In conclusion we may say the that method of reduc tion (the space velocity used, purity of the H2 and the reduction time and temperature)will influence the chemiqorption and

the catalytic properties of the catalyst. It may therefore

not be expected that different investigators will always arrive at quantitatively the same results in studying the chemisorption ofN2 and the synthesis ofNH3, even if they use catalysts of exactly the same composition.

B. Surfac est r u c tu r e 0 f th e r e d u c e d c a tal y st

In the reduction process the catalyst obtains aporous (1-Fe

structure with a relatively large accessible surface area of aboutlO-20m 2 /g. The external volume of the catalyst

11

is not altered by the reduc tion; shape and dimensions of the original spinel lattice is preserved19 .

The func tion of the Al 20 3 promoter is to be understood as the preservation of the-thermostability ofthe lattice, i. e. the growth and the sintering of the Fe crystallites is pre-vented by a thin but strongframework of-y-AI203, so that also at higher temperatures a large surface area is main-tained.

The surface of the reduced catalysthas rather acomplica-ted structure. It is formed partly by Fe atoms which may be 'present in different crystaUographic planes. As has been explainedinsectionA, afraction of the iron surface .may still be covered with 02' to a degree dependent on the reduction method used. In addition, in spite of the fact that only little Al203 is contained in the catalyst, this Al203 proves to occupy alarge part of the surface 20. In a catalyst containing about 5% of Al203 promoter, for instance, practically all the Al 203 accumulates at the surface, covering 50% of it. Finally, the promoter sur-face is probably covered partly with hydroxyl groups, as is the case also with the surface of pure 'Y-AI 20 3 21 22; this coverage wiU be a func tion of the temperature and the time of heating of the catalyst in H2 and in vacuo. From what has been said above, it is clear th at on the surface of a singly promoted catalyst we can alreadyexpect four different atoms or groups of atoms!

We shall now examine in what manner we c'an get quanti-tative and qualiquanti-tative information about these groups.

1. Determination of the tree Iron surface areA

The free iron surface area can be determined by a method devised by Brunauer and Emmett5 20 (see Fig. Ia): First the adsorptionisotherm of CO at -1830 _Cis

measu-red. At this temperature the gas is bound both physically and chemic ally; this is se en from the fac t that by pumping at -780 _{C only part ofit can beremoved from the surface.}

If, afterpumping at -780 _{C, asecondisothermis deter-.}

mined at -1830 _{C, this isotherm proves to run at a}

con-stant distance under the first. Af ter repeated pumping at 780 _{C this second isotherm is reproducible; ·-it obviously}

40 30 10 V (cm3 ) Volume of CO adsorbed ~ 10 12

0---

o~ _ P(cmHg) 20 30

Fig. Ia. AdsoI'ption of CO on an Fe catalyst.

I. Total adsorption at -1830 _C.

II. Physical adsorption af ter evacuation at -78.0 _C.

represents the physically bound CO. From the distance

between the two isotherms we can derive the volume of

chemically bound carbon monoxide, VCO'

This volume, V CO, and the volume of N;::! physically

ad-sorbed in a monolayer at-1830 _{C, Vm (N2). measuredon}

an unpromoted cat-alyst, enabled Brunauer and Emmett to

derive the relation

v

CO (pure iron)

Vm (N2) (pure iron)

= 1 25 _{. . . . (I -1).}

It follows from this relation that the free iron surface,

0Fe, for a promoted catalyst is given by

where

VCO

V

_co

o

tot. . (1-2),

= the volume of CO (N. T. P. ) chemie ally bound on

the promoted catalyst;

V_{m (N 2)} = _{the volume ofN2 physically}

absorbedinamono-layer at -1830 _C;

'1:3

Themethod described above is the only.one known to us tor determining me free iron surface «rea; however we have som-e praètiGal ana theoretical objections to it.

a.

, A practical difficultyis that'after chemisorption of CO it isimpossible to restore the original form of the catalyst· if one tries to remove CO by desorption. part of it is re-leased as iron carbonyl and the remaining catalyst has a smaller surface area. Therefore. the method can be used only once. at the end of a series of measurements.

b.

A second problem is presented by the length of time that is to be used in removing the physically adsorbed CO by pumping at-780 _{C. Emmett and Brunauer23 recommend} one to two hours; in this case isotherm Il (see Fig. Ia) would practically coincide with the physical adsorption isotherm of Na the N2 molecule having about the same . cross':'sectionaL area as the CO molecule.

The amount of CO ac tually removed by pumping for one to two hours wil!. however. always be influenced by the total amount of' catalyst to be degassed and by the vacuuIn reached in pumping. In chapter lil it will be shown from experiments that the amount of CO removed by pumping depends rather strongly on the pumping time (and the tem-perature).

Therefore we considerit a better procedure to subtract ~e physic al N

a

-

ad.sorption isother~ from the total. CO lsOtherm (No. I). lnstead of subt2îctmg the poorly defmed physical CO adsorptionisotherm . In this case we have to multiply the adsorbed quantities of N2 by a factor of about 1. 05. as appears from the almost constant ratio of the physical adsorption of N2 and CO on different inert

adsorbents (table Ia).

Table I cr

.

Reference adsorbent 'V_m(CO)/V_{m (N2)}

P. Zwietering Al203 1.056 ,( unpublishedl . Emmett.P.~ Bruna~er.S _8i02 1.043 11 charcoal. 1. 038

•

...

Vror./V_(N2 ) d

aeference

Brunauer, S;Emmett, P. E.

J. Am. Chem. Soc. 59, 310, ~937

Brunauer, S;Emmett, P. H. J. Am. Chem. Soc. 62, 1732, 1940

Kummer, J. T. ; Emmett, P. H.

J. Phys. and Colloitl Chemistry 55,342,1951

Podgurski, H. H.;Emmett, P. H.

J. Am. Chem. Soc. 57,159,195.3 Podgurski, H. H. ; Emme'tt, P. H. J.Am. Chem. Soc. 57,159,1953

Podgurski, H. H. ;Emmett, P. H.

J. Am. Chem. Soc. 57,159,1953

al. Menf '. Catalyst No 973 No 973 No 910 No 927 No 97'7 Pure Fe-wire roughness factor 2.5 Pure Fe-wire roughness factor 6 Table I

fJ

hes bv E Impurities "unpromoted" 0.150/0 _{Al 20 3} "unpromoted" 0.15,,/0 _{Al 203} "unpromoted" "unpromoted" "unpromoted" 0.15-0.20"/0 Al 203 O. Ol % Ni 0.091 % Cr 0.0030/0 Al 0.0050/0 Si 0.0030/0 Sn 0.01 % Ni 0.001 % Cr 0.0030/0 Al 0.005% Si 0.003% Sn

mmett and coworkers

Total _VCO Surface Reduction _at'ea(m 2 _{/ g)} V m (N2) 24 hrs;300-350 o C 54 hrs;375-500o C 0,80 1. 25 92 hrs"up to 4000 Cl! 0.97 1. 25 0.82 1.30 0.66 1. 18 4600 C 0.52 0.64 5200 _{C 0.74}

o

. :) .

4500 C 0.88 (.78 4250 C 1. 07 0.44 6200 _{C 0.025 0.13} 4600

_C

_{0.06 0.63}

In this table Vm (N2 )

=

the volume ofthe mQnolayer of

phy-sically adsorbed N 2 (at ~1830 C). in cm3• Vm(CO} = the

volume of the monolayer of physically adsorbed CO (at

-183 0 C). in cm3 .

c.

The unpromoted c.atalyst used by Brunauer and Emmet in

the determination of equation (I-i) contained 0.15% 'A120 3'

Sec ti on B2 ofthis chapter gives information which indicates

that such a catalyst has only a small quantity of A1203 on

its surface.

Itis strange. however. that thiscatalyst does not show a

constant V co/Vm(Nz) - value. This appears from tabie

If3. where we calcu,l ated this value from measurements

which Brunauer and Emmett made in a latter stage of their

rese~rch. Apparently the value of the quotient V

col

V m (N 2)

strongly depends on the reductioncondi~ions. in particular

on the temperature. and 1. 25 is only about the hip;hest value

found. r;ierefure we . .Jelieve t/jat eqLwtion (I -1) canilut

ûe used for detcrmi .. illé t/je free sur face AreA uf jJrO/110 ted

CA t ,'dy st s.

It will now be clear that the same behaviour ,of the

V CO!Vm (N2),-vaIuemaybeexpected for catalyst samples

olsomewhathigher A1203 cQ.ntent. e. g. O.;W-l %.It follows

that the determi.lation of the free iron surface is espe-cially doubtful if applied to catalysts of low A1203 con-tent.

The nesults listed in table 113 might be due to a slow diffusion,

at higher reduction temperature, to the catalyst surface of the'

remaining smal! quantity of Al203 and possibly other

conta-minants. the amount being a function of terr [Jeratnre and t;~ -.

A second e.xplanation of the decréase c~ VCO/V_{m (N2) wil! be}

given in chapter 111. ,

The pure iron wires (see the two bottom rows of table

ll3l

had

very sm all surface areas. Possibly, the degree of CO

chemi-sorption is low in this case. a~ the smaH iron surface area

can hardly compete with he surface area of the contaminants.

If the roughness' Îac tor of the wire is raised by etching (see

'16,

d.

Different a-Fe planes contain different numbers of Fe atoms per cm 2 . For instance:

[111] face: nFe = 0.71 x 101'5 [110] face: nFe = 1. 7 . x 10 15 [100] fac-e: nFe = 1.2 x 1015

Even if the question as to number of CO molecules that can be chemisorbed per Fe atom on these planes is left out of cOI;lsideration, it wiU be clear that different distributions of exposed planes result in different numbers of chemis-orbed CO molecules per cm 2 . Yet, in passing from for-mula1-1 to 1-2 it was tacitly assumed that an unpromoted catalyst has always the same distribution of exposed planes as a promoted one. However, this is unlikely, as we know from Bravais' law that the more dense planes will grow at the cost of more open planes, which effec t is a function of temperature, and probably of promoter content.

It maybeconcluded fr om b, c and d that the only informa-tion that can be obtained by Brunauer and Emmett's method is the maximum number of CO molecules which can be chemisorbed by an iron catalyst. This number gives us only a rough approximation,of the extent of the iron surface. Neither the free iron surface area in m 2

f

g, nor the percenta-geoffree ironcan be calculated from it with reasonable ac-curacy.

Also Westrik. and Zwietering19 studied the problem of CO chemisorption on an uripromoted Fe catalyst. They star-ted from pure magnetite*) which was verycarefully reduced at 2250 _{C for 2000 hrs. The renuction was completed by}

short-time reductions at higher temperatures. From their experiments we calculated table 11'; it is seen that Westrik and Zwietering arrived at practically the same results as Emmett and coworkei's: the highest V COfVm(N?) - value found is 1.20 (Emrrett 1.25), and this number decreases again wh en higher reduction temperatures are used. ,

. ..

:I' Analysis of the sample used by these authors proved it to be very pure indeed: it contained only 0.04% Ali03'

17

Table I l'

Table I l'

VcojV_m(N2), calculated from Westrik and Zwietering's experiments

Catalyst Reduction Total surface

...

_YCO

(After pre-reduction area

at 2250 _{C for 2000 hrs)} Vm (N2) m2 jg Pure magnetite 40 hrs at 3000 _{C 2.96 1.18} 0.04% _{A1 20 3} 88 hrs at 3000 _{C 2.66 1. 20} 128 hrs at 3000 _{C 2.48 1.18} 40 hrs at 3500 _{C 2.10 0.98} 84 hrs at 3500 _{C 1. 70 0.99} 40 hrs at 4000 _{C 0.85 0.79} 105 hrs at 4000 _{C 0.80 0.78}

*

The cross-sectional area of a phys~c~lly adsorbed N 2 molecule was taken to be 16. 2

J(-. 0-2

Talnng 16.2 1\ to

7be the surface area of a physically

adsorbed N2 molecule, . Westrik and Zwietering found for the number of CO molecules per cm2, ncO, (ifV CO/V (N2)=

1.20): m

nco = O. 74 x 10 15 .

As this experimental value was found to be in good agree-ment with the value c alculated for the [111] -plane(nCO[111]=

O. 71 x 1(15 ), it was concluded that the Fe particles obta:in-ed by careful robta:in-eduction of magnetite expo se almost exclu-sively [111] -planes. This opinion was fortified by the crys-tallographic observation that during and after reduc ti on of a single crystal of magnetite the shape and size of the origirial crystal had been preserved (pseudomorphosis),

and that the [111] -planes of the Fe crystallites proved to run parallel to the surface .

However, at present, we, and also the authors themselves,

*

.• Personal communication.

la

hold a different view. First of all. objec.:tions band c made in connection with Brunauer and Emmett's work afPly also in this case. Secondly. from the fact that the[111 - faces of the iron crystallites run parallel to the macroscopic boundary planes of the erystal. it does not follow that the surface of the crystallites is composed of [111] -planes. Finally. also a combination of different planes at the sur-face can lead to a numberofO. 74x'1015 COatomsperCîn l .

2. Determination of tileamount of residual H2 retained oy

Fe catalysts alter reducti~n

KUx:q.p1er and Emmett25 ) determined the amount of H2 left on,the surface o(a~umber ofpromoted and unpromoted Fe catalysts r{ter 79'hrs. of reduction at 425-50(10 C

(

,

S.V'.

1000 hrs- ). and evacuation for a few hours at the final reduction temperature. The experiments were performed bycirculating.D2 over the catalyst surface at 300-5000 C, and next de,termining the H2 enrichment of the D2 mass-spec trom~tic'ally or katharometric ally. The c atalysts were. ind'eed, found to give H2 enrichment. to a degree strongly dependent on their type. From these authors experimelltalresults we calculated table ló

About the origin of the H 2 Kummer and Emmett donot make a definite statement. but lhey. sum up the following

possi-bilities: '

(i) H2 chemisorbed on the iron surface.

(ii) H2 chemisorbed on the promoter surface. f (iii) H2 enrichment of the D₂ by exchange with H20 bound on the prcmoter surface.

It will be demonstrated that possibility (iii) is the most plausible.

ltis unlikely that an Fe catalyst which has been heated at 5000 C for several hours still contains H2 on its surface or in the bulk phase: the heat of chemisorpti~n _{of H2 on} iron and Fe catalysts is Bnly 20 Kcal/mol 2 (8H2,~ 0), on Al203 27.5 Kcal/mol. 2 . These heats of chemisorp-tion are so low that H2 is certainly desorbed at 5000

ei

from our own experiments it followed, for instance, that N2' which is bound with a heat of chemisorption of c. 44 Kcal/mol at 8N2 = O. could be desorbed completely at

, 0 0

oe.

(~

CL-

~J"

Table I ti

Residual H2 retained by iron catalysts af ter reduction at c. 5000 _{C for 70 hrs, and evacuation'at 400-500}0 _C

Catalyst No 910 927 238 422 423 954 a. b -. c -.

Promoters Evacuation Total free Total Exchange H2 rapidly

af ter iron promoter temperature exchanging

reduction surface surface

area area hrs (oe' m 2 m 2 °C cm3 "unpromoted" 1.~ 469 5.30 5.05 450 0.02 "unpromoted" 3 450 18.22 10.76 306 0.01 0.92"/0 Si02 16 500 33.12 93.56 306 0.23 1. 55"/0 A1203 2 500 112.6 239.4 500 1. 80 0.580/0 Zr02 2:_{26"/0 A1 203 24} 502 6~:i. 2 380.9, 504 3.20 0.620/0 Si02 10"/0 A1 203 16 500 54.7 156.5 306 3.10

a. _{An "unpromoted" catalyst cim still contain 0.15"/0 A1 203'}

b. Thefreeiron surface area was calculated'by Brunauer and Emmett's

method. See, however, the objec~ions made in section BI of this chapter.

c. The promoter surface area was found by subtracting the free iron surface area from the tot al surface area.

H2 rapidly exc~anging per m of promo! surface cm3 4 x 10- 3 1 x 10-;j 2.4 x 10- 3 7.5x10-3 8.4x10-" 20 x 10-;J er

...

<&

20

5000 _{C is lik~wise extremely low.}

Armbruster 28 studied the solubility of H 2 in very pure Fe made from iron carbonyl. The solubility s, in rrticro-moles of H2 per 100 g of metal, was given by:

Log

(+)

where 1454

+

1.946 -T p

=

H2 pressure in mmi T 1:: absolute temperature. (I - 3),

From(I-3)wecancalculate that at 5000 _{C and aH2 pressure} óf 1 atmosphere only 7 xl 0- 3 cm 3 of H2 is dissolved in 1

2

g

of Fe catalyst. Taking the mean surface area to be 10 m / g, the H_{2 enrichment of D?...t caused by dissolved H2, can} be onIy 7 x 10- 4 cm3 per

nï:&,

while by degassing the sam-ple, this quantity is still further reduced. (see formula 1- 3).

On the other liand, the presence of H20 on the surface of -Y-AI?-fii and its exchange witli D2 is very well known. Mills c. s. give the following table for the amount of H20 left on the surface of -Y-AI203 af ter evacuation at different times and temperatures:

Table IE

Evacuation _{%H2O surfaCv t rea} H2 per ~2 (cm3 )

Temp. Time (m 2 g) 4500 _{C 16' hrs 2.29 304 95} x 10- 3 450 64 1. 80 294 75 x 10- 3 550 16 .25 305 52 x 10- 3 550 64 0.98 290 42 x 10- 3 650

16

0.61 293 25 x 10- 3 650 93 0.37 284 15 x 10- 3

The degree of H~ enrichment, expressed in cm3 H2 (N.T.P.) per m of 'Y-AI203, and determined by D2 exchange at 4500 _{C - 600}0 _{C, is givenin the last column.}

Startingfrom table IE we can calculate the H2 enrichment which may be expected from the AI2.0 3 - promoted Fe catalyst No 954 (see table Ió), assummg that the H2 en-richment is caused onlyby H 20 still bound on the promoter

21

surface . Catalyst 954 was heated at 5000 _{C in H2 and in}

vacuum for c. 86 hrs; in this c ase we may expec t,

ac-cordi~g to table IE, an exchange oCc. 50 x 10- 3 of H₂

per m of promoter surface area. The figure found by

Kummer and Emmett (see last column, bottom line tabfe

16) is of the same order of ma~nitude, but somewhatlower

viz. 20 x 10-::i cm 3 ofH2.per m~ of promoter surface area . •

In table I X we list some of our own observations, which

were made incidentally. As is seen from the last column,

Tabl.1 ~

H2.

exchanctnc per

,:

:

c:..

~loc:,3h::!ir~~~:u:~z;.0!t c;~ê~~rr;:~:: at 5000

C

for 170 bra a. b.

HZ etthanp per m2 ol

-Sample Promoter. Free·iron Promoter EXchana.

aurface area . aurface ~. temperature promoter aurlace area

mal. m 2/. oe cm3 7-Al203 211.0 400 42.8 x 10-3 Fe-catalyat 3.1'110 Al203. 8.10 12.0 400· 28.5 x 10- 3 Fe-catalyllt 14.4" Al203 12_,.50' 11.1 ·400 31. 3 x 10-' Fe-catalyllt ''unpromoted'' (0.24'110 Al2O,) 1. 83 2.4 400 0.38 x 10-' •• Ca1culated b,. Bnanauer and EmmeW8 methocl.

( ••• tb. obJecUon. in llectlon Bl ot thi. eh.pter).

b. Found by 8Ubtractin, tb. free-iron aurface ~. tl;om tbe total aurface area. .

Dur results are in quantitative agreement with the yiew

that the H2 enrichment i._{s caused by H20 bound on the}

promoter surface . In chapter 111 we shall return to the

observations listed in table IX.

In accordance with Kummer and Emmett's results for "unpro-moted" catalysts (see table 16, catalysts Nos. 910 and 927),

our "unpromoted" catalyst with an Al 203 content of 0.240/0 gave

a very low H2 exchange per m 2 of promoted surface area (see

table IX, bottom row). This may be explained as follows:

In section B we showed that the determination of the free iron

surface area, and consequently of the ·promoter surface area,

is a doubtful affair, particular1y BO in the case of catalysts of

22

of V co/V m (N 2) as demonstrated in table I{3for acatalyst having anAI203 content of 0.150/0 is observed in this case. This may strongly influence the ratio: H2 (cm 3)/promoter surface area

(m2 ).

From the fact thatcatalysts with 0.150/0 and O. 240/0 A1203 cause

• only a very low H2 enrichment. of D2' we are inclined to

believe that these catalysts have, in fact, a very low percentage

of A1a03 on their surface.

c.

Kinetics of N2 chemisorption on iron

ca-talysts and films

The first few measurements of the rate of chemisorption of N2 on technical Fe catalysts were made by Emmett and Brunauer29 ; they date back to 1934. The reduction

pro-gramme they adopted for the doubly promoted Fe

cata-lyst was as follows: reduction in H2 between 350 and 4000

C, until the water produc tion was only 0.1 mg per hour;

evacuationat 4500 _{C for half an hour. The N2}

chemisorp-tion rate was measured volumetrically at 350, 400 and

4500 _{C and 1 atm. N2 pressure. The activation energy}

proved to be c. 16KcaljmolN2. Immediatelyafter these

adsorption measurements the N₂ was removed by r-

e-duction and the rate ofthe NH₃ synthesis was determined

by passing a N 2 - H₂ mixture (1: 3) over the catalyst and

measuring the rate fitrimetrically.

The rate of NH3 synthesis was found to be practically

equal to the rate of adsorption of the first portion of N2, measured at the same temperature and pressure. On

account of this it seemed reasonable to assume the N2

chemisorption to be the rate-determining step in the NH3 synthesis.

At a later stage Brunauer, Love and Keenan 30 analyzed

these N2 adsorption measurements. It proved possible

to describe the adsorption rate by the so-called Elovich-equation:

where

dv/dt = c exp(-gv)

v·= adsorbed volume of N2i

t

=

time;

c and gare contants.

23

The activation energy of adsorption was found to change lipearly fr om 10 Kcal/mol N2 at low values of v to 21.5

Kcal/mol N2 at higher values. of v.

As is clear from equatior(I-4Jhe adsorption rate depends

strongly on the amount of N2 adsorbed. This amount being unknown under reaction conditions, Emmett and Brunauer could not actually prove the N2 chemisorption to be rate-determining. However, this pioneer work was of the highest importance for a deeper understanding of NH3 synthesis; in 1940 the supposition of a rate-deter-mining N2 chemisorption was used by Temkin and Pyshev31 as the principal postulate in their derivation of a successful kinetic formula for NH-3 synthesis.

Afterwards the matter was taken up again by Zwietering and Roukens 32, as the experiments of Emmett and

Brunauer gave a far from complete picture citre adsorption

process, and moreover an important part of their me a-surements was disturbed by what Brunauer and Emmett themselves called an "unknown irregularity or poisoning factor". Zwietering and Roukens started from an iron catalyst containing 0.850/0 A1203' The catalyst was re-duced. at 4000 _{C for 24 hours. and at 450}0 _{C for 24 hours}

with H2 ofnormal purity. finally at 5500 _{C with extremely}

pure H2' prepared by diffusion through Ni at 10000

C:

The space velocity* during reduc tion was 1200 hrs-1 •

The N2 chemisorption rate was measured at a constant

pressure of 20 cm Hg, at 200, 225 and 2500 _C.

All rneasurements could be described by the following equation:

'

de

~

_-, 22.3 _{e x p . . -}(1072

e)

_expo (-9200'-72700

'e

_{_}

J

_mln.

.

-1' • (1-5)

d t ' R RT / '

where . \

e =

N2 coverage ori the Fe surface;

R

=

gas constant in calories;

t

=

time in minutes;

T

=

temperature in OK.

The equation is valid for 1 cm Hg nitrogen pressure in the

occupation range

e

=

0.08- 0.25, and between 200 and 2500

C. A limited extrapolation to lower or higher

temperatu-*

The space velocity is defined as the ratio of the gas ;rate (in cm3 (N. T. P. )/hour) to the catalyst volume in ~m3.

24

res is probably admissible. Equation (1-5) points to a linear change of the activation energy with a:

Ee

=

9200 + 72700

e

(1- 6).

A full discusion of equation (1-5) will be given with refe-rence to our own adsorption experiments, in chapter IV. Here, sQme preliminary remarks will be made only.

If the reac tion temperature is constant, equation (1- 5) points to a linear change of the logarithm of the adsorp-tianratewithe. Inthis respect it is comparable with the Elovich equation, which describes the chemisorption rate of gases on metals for many cases 33 34. However, equation (1-5) contains also a temperature-indepêndent term, exp(107.2ejR), whichcompensates for the strong increase of the activation energy with e. Such a "compen-sation effect" is a widely observed phenomenon in chemi-calkinetics. As appears from a study by Cremer 35 ,the interpretation may be given along diffe rent line s. In chap-ter IV we shall return to this phenomenon.

The work of Zwietering and Roukens differs from the study of Brunauer and Emmett in some other respects besidesits more complete analysis of the kinetics. For both reduction and chemisorption highly purified gases were used, resulting in reproducible adsorption runs. Zwietering and Roukens measured also the free iron surface area of their sample; this enabled them to express the adsorption rate as afunction ofcoverage (see equation (1-5.). Finally, like Brunauer and Emmett, they determin-ed the NH3 synthesis rate, using the same sample as in the adsorption experiments.

Incomparing the NH3 synthesis rate with the N2 chemi-sorptionrate, thedifficult point was, again, that the co-verage during synthesis,

es, was unknown. Therefore,

the authors had to be content with ascertaining th at ad-sorption and syn~hesis rates were equal for an experi-ment at 325 0 C i fes was O. 14, and for an experiment at 2750 C i fes was 0.11. These figures for as we re not unlikely and lent new support to the belief that the N2 chemisorption is rate determining.

Also in Japan an interest has been taken in this chemisorp-

25

rate of N 2...manometric ally on an Fe ca talyst promoted wi th

1. 06% K2U, 1. 82% Al203 and 0.14% Si02, 'at pressures below 2 mm Hg. The catalyst was reduced at 5500 _{C for one} month, in a circulating stream of H2 purified by' passing , the gas through a heated Pd-thimble. ~ _{The H 20 formed} was frozen out by liquid air traps, attached at both sides of the sample bulb. The adsorptlon Jrate was measured at 400, 425, 450 and 5000 _{C. with N2 prepared by}

decom-position of NaN3. The rates measured at 450 and 5000 _C

were corrected for simultaneO.us desorption of N2; this was done by means of the chemisorption isotherms mea-,

sured between 300 and 5000 _{C. T_he occupation range}

studied lay between 0 = 0.03 and 6 =

O.li.

.

The author described his measurements by an equation of the general form:

d P , 0(

_ _ N ... 2_

=

ka P_N (Po - P_N ) (1-7),

d t 2 2

. , ~ , .

which he called the "power rate law". Po is the initial pressure;

P N2 is the N2 pressure ~t time t; ka and ex are constants.

In 'the very sm all occupation range studied the constant

ex changed three times39 , (see, Fig. Ib)

0< 8<0.025 ex = 0 o -0 ,1.5 0.025 <; 0<0. 08 ex= 1.4 log(, ~*')

I

-tO , -05 \ _,

_,

\

o 0.08<6<0.11 ex

=

3 ' 05

Fig. Ib. Chemisorption rate of nitrogen as a funetion of eoverage 6 on a logarithmie seale, as given by T. Kw~, J. Res. Inst. for Catalysis, Vol. lIl, 1953, p. 18.

2.6:

However. as alreadypointed outby Zwietering3• Kwans's·

measurementsmay be described in a much better way by the Elovich equation40 :

dP . . . .

-_--.:.N-.!2=-

=

ka P

_N

exp_

[-P-'(

Po - P

N

>]

(1-8).

d t 2 ' - 2 .

(see Fig. Ic)

where

f3

is constant throughout the occupation ranp.e

stu-died by Kwan. and(P 0-PN ) is proportional to

e

*

- - 2

-0.0

-0.5

-0.

Fig. Ic. The same points as give.n in Fig. Ib, on a·semilog-scale. Also other measurements reported by Kwan can be

accu-rately described by the Elovich equation. as shown"in-Fig.

ld. where againbrokenlines re sult when the "power rate

law" is applied.

As appears froIll Fig. Ie. the activation energy values calculateJ from Fig. Id. rather well join the activation

'"

.. _{log (Po- PN2)} = -0.18 corresponds to 8 = 0.074.

27

Chemleorptlon rate of n1trosen as a function of coverage 9 «

constructed trom mea8urernents bx T. XI',sn. J. Hes. Inst.

tor CatalY81s. Vol. lIl. 1955. p. 114 - 115. table 2 an1 fig. :;

( 1 d P) , 1 log,-

p

'

dt mln. -OA

t

,

+", -0, ++

~ ~

\\0 ·

_,

_,:.~

•

.

~

-1.2

..

~'

• • 500 oe + x x + -1.6

\

~~

,

0

~~

450°C x'xx -2,0 _~x

\

~400°C

-2.4 \ 300°C 0 - 9 0,02 0,06 0,10 0.14

Fig. Id. Chemisorption rate of nitrogen as a function of ,coverage

9, constructed from measurements by T.Kwan, J. Res.

lnst. for Catalysis, Vol. lIl, 1955, p. 114 - 115, table 2 and fig. 3.

energies reported by Zwietering.

The kine tic s of N 2 chemisorption on evaporated lie film s

were studied by Greenhalgh, Slack and Trapnell . These

authors found three different types of chemisorption on films: a rapid, weak, reversible adsorption, neglihle at room temperature and above; a rapid strong adsorption,

irreversible below 1000 _{C, and a slow sorption of the}

ac-tivated type.

The last-mentioned slow sorption, which was identified with the ac tivated N 2 adsorption on Fe catalysts, aUained

28 30 E (Kcal/mol N2)

t

x 20 x x 10 _ 9 0.1 0.2 0.3

Fig. Ie. Activation energy of N2 adsorption as a function of cove-rage

X } P. Zwietering and J. J. Roukens.

o

Trans. Farad. Soc. 50,

+

178 -187, 1954

• t

Calculated fr om Kwan's ~

J

measurements, Fig. Id . equation where dvJdt

=

K

~

exp (-l' v) v

=

adsorbed volume of N 2; t = time; p = pressure;

K and l' are constants.

)

29

This equation again, is of the Elovich type. Howeve r, the fac t that the adsorption rate was proportional to the square root of the N 2 pressure is quite remarkable. In chapter IV it will be shown that all ov.r adsorption me asurements on Fe catalysts point to a proportionality with the N2 pres-sure. Thecause ofthe discrepancybetween the results on films and those on càtalysts is not clear.

D. N2 chemisorption isotherms and the heat of chemisorption

lsotherms of N 2 on Fe catalysts were determined by Emmett and Brunauer29, who measured the maximum aIIlount of gas adsorbed at constant temperature as a functionofpressure. In consequence of the activated na-ture of the adsorption process the isotherms could only bemeasuredathighertemperature; 330, 400and 4500 C. The heat of chemisorption was calculated f~om the iso-therms by the Clausius- Clapeyron equation

where (. d

In

p \

=

\ d T

Je

p

=

N2 pressure; T = absolute temperature;

e

.

::;

N 2 coverage;

~())= heat of chemisorption at coverage

e ;

R

=

gas constant. .

(1-10),

Emmett and Brunauer' sisotherms gave only limited infor-mation, as poisoning and other irregularities were not fully ruled out in their investigation. A detailed evalu~­

tionoftheseisothermsby Brunauer, Love and Keenan3U shows the heat ofchemisorption to decrease linearly from an initial value of about 44 Kcal/ mol N 2 for the very first stage of adsorption, to a value of about 32 Kcal/mol N 2 at "higher coverage".

An original method for the determination of N2 chemi-sorption isotherms was devised by the Russian investi-gator.s Romanushkina, Kiperman and Temkin42 .

(a).

In this formula Fe [N2] represents" N₂, chemisorbed on

the surface of the catalyst. This . equilibrium, which

the authors called an "adsorption-chemical equilibrium",

can be represented by the sum of two other equilibria, viz. :

and

(c ). Hence, if we know the position of the equilibria (a) and (c), we can calculate equilibrium (b), which represents

th~ N2-chemisorption isotherm. Unfortunately, also

these investigators report poisoning of their catalyst

samples. However, their most reliable measurements

lead to the following results:

No. of experiment Q (9N2 "Iow") Q (ON2 "high").

Kcal/mol Kcal/mol

I 50.9 32.9

VI 52.7 25.0

VII 59.8 17.-4

A typical isotherm taken from thei,r work is given in

Fig. If; it is seen that in the greater part of th~

occu-pation range (0 = 0.20-0.80), () eq. is a linear function of

log PN2' '00 <

"' "3 -2

I

Fig. If. N2 chemisorption isotherm on an Fe catalyst. according to Temkin c. s. 42. T = 3500 _C.

o

:

experiment No. IV.

·31

This re sult agrees with the chemisorpt-ion isotherm pro-posed by Slyghin and Frumkin43 :

(1-11 ) (I -11) is found by equalizing the Elovich equations for adsorption and desorption of N_{2 :}

( de

- -

J

- k P exp'{-g9)' (I-12) ( ) d t ads. - ;p. Hs.

de

- - - k d e xp. ( he) d t des. (1-13)

substituÜng f for (g+h) and a o for ka/kd.

The bends in the isotherm at 0 = 0.20 and (J = 0.80 ( see Fig. I-f) can be made understandable by intro-ducing into e~ations ~I-12) and (1-13) the "Langmuir-terms" (1-0) and 0 , which represent the change with growing coverage in thé adsorption and desorp-tion chances, respectively. In this case equation (I -11) bec omes

e -

2 log

f'ë

9)

+

T

loga.

~2

(1-14) Tne term 2log (.J.u~causes a deviation of linearity at

. very low and very high coverage. .

Also Kwan37 measured N2 chemisorption isotherms on an Fe catalyst in connee tion with his kinetic experiments already referred to in section C. Using the "power rate law '·' for both adsorption and desorption kinetics, viz.:

(~~

_{tJ .}

_.

-0<.. (1-15), ka. P N _a

e

and \

(~l

Ket

e

~

(1-16), d t des. and putting

(dO/ dt)ads. = (dO/ dt)des.

he arrived at a chemisorpiion isotherm of the "Freundlich type "39:

P

=

log

(k

de,:'

+

(a(

+~)

log

e _

(1-17), log N₂ kaJ:..)

32

Kwanmeasured chemisorption isotherms at 400, 450 and· 5000 _{C, and plotted his experimental points on a}

logarith-mic scale (equation 1-17). In Fig. Ig we 'demonstrate, however, thatupwards of 9~0.15 Kwanls points are per-fectly described also by the Slyghin-Frumkin equation; moreover, the overall shape of the isotherms is now in

0.5 0.5 SN, 0.4 0.3 0.2

~

~o. • 0.1

~5Y

••• ..Ai -3 •

"

"

•• _ log P N, (mmHg) o +1 o •

Fig. Ig. Chemisorption isotherms of N2 on a promoted Fe catalyst, constructed on .a semilog-scale, from Kwan's

experi-mental points. 0 : 6730K

• : 7230K

6. : 7730_K

accordance with those of Temkin c. s. (see Fig. If).

By means of the Claushis- Clapeyron equation the heat of chemisorption c an be c alculated from the isotherms in Fig. Ig, although the points are rather widely scattered:*;

() Q (Kcal/mol) 0.1 40 0.2 0.3 0.4 24 18 15

In chapter IV we wiU show that Kwanl.s Q-values are'in rather good accordance with our own observations.

".. The temperature constancy in Kwan' s experimen~s was only

From the survey given above it may be cpncluded that the information available in literature about the

chemisorp-,tion on Fe catalysts is only limUed. A second difficulty is that the pict1,lre found from measurements 'on evapo-rated Fe films essentially deviates ,from that obtained with c atalyst,s:

Workingcalo;rimetrically on Fe films. Bagg and Tomp-kins44 found heats of <.;nemisorption decreasing r6ughly exponentially from 70 Kcal/mol at (J=o to 16 Kcal/mol at (J

=

0.15. which is in accordance with Beecks results. 45

,(see Fig. Ih). '

80 C(Kcol/mol) 40 a 20 o b 0,' 0.2 03 04

Fig. Th. Heat of chemisorption of N2 on evaporated Fe films a : .J. Bagg and F. C. T.ompkins. 44

o : O. Beeck. 45

b : A second type of N2 chemisorption at - 1960 _C found by Beeck, in accordance with Porter and Tompkins' findi~gs.

N2 begins to be chemisorbed at a measurable rate at oo,C already. whereas on Fe catalyts the start is obse'rved on-lyat about '2000 _{C29. Moreover Beeck4E • and also Porter'}

and Tompkins46 • struckupon a second type of N2 chemi-sorptiononfilms at -1960 _{C. (see Fig.lh) which has never}

been detected on Fe catalysts.

However. inchapterI,II we will show that the chemisorp-tion behaviour of an Fe catalyst is modified be the use of a high space velocity 'and: ,a long reduction tim,e. and that the difference betweencatalysts and films is made

con-'.

.34

siderably smaller by it.

E. H2 chemisorption and H 2--D ₂ exchange on

iron catalysts

1. It _{is not certain that H 2 chemisorption on the Fe} c

ata-Iyst is an essential reaction step in NH3 synthesis; it is possible that H2 from the gas phase reacts directly with the chemisorbedN2 (Eley-Ridealmechanism); anotherpo s-sibility is the appearanc e of a so-c alled Langmuir- Hinshel-woodmechanism, in which onlychemisorbed H2 reactswith

chemisorbed N 2 . Perhaps both mechanisms together play apartintheformationof NH3. However, whether it is e s-sentialor not, there wiU always be H2 chemisorption on the

catalyst during synthesis, and this is important in conne

c-tion with our own synthesis experiments (chapter IV), where we calculated the quantity ofchemisorbed N2 by

sub-trac ting the weight of chemisorbed H2 from the total adsor-bed quantity.

Therefore, we will here give a short summary of what is knownaboutH2chemisorption. LikeforN 2, the picture for H2 is complicated; again, astrong discrepancy exists be-tween the results on films and on catalysts.

Extensive studies of the chemisorption of H2 on differe~

Fe catalysts were carried outby Emmett and coworkers 48. On singly promoted, doubly promoted and unpromGted catalysts, two types of activated adsorption have been found, type A taking place at a measurable rate at - 900 _C

. and above, type Bat 1000 _{C and above. The activation ene}

r-gy for type A was about 10 Kcal/mol; the adsorption of type B "was rapid ", with a heat of adsorption of about 8.5 Kcal/mol. It is remarkable that only on singly pro-moted èatalysts (propro-moted with A1203) a third type of H2 chemisorption was found (type C), at -1960 _{C and above.}

Also the mutual influence of chemisorbed gases was studied indetail 20 49. Here we will deal only with the results for N 2 and H2.

N2' preadsorbed on unpromoted and doubly promoted

ca-talysts gives rise to a lowering of both type A and B H2 chemisorption, in such a way that the total number of adsorbed atoms remains about constant. With singly

pro-moted catalysts again an exception is observed! Here, previous chemisorption of N2 increases type B

chemi-35

sorption to suchan extent that the total number of adsorbed H2 atoms exceeds the n~mber _{of atoms adsorbed on} N2-free catalysts by 30-500/0.

Kwan50 . stuctied the chemisorption óf type B hydrogen on a catalystpromoted with 1. 060/0 K 20, 1. 820/0 Al203 and 0.4% Si02. As was the case for N2' he found chemisorp-tionisothe·rms of the Freundlich type, the heat of chemi-sorption being given by

Q

=

~ 16; 7. log

e

(I-18)(see Fig. li, curve d) In a calorimetrie study of the H2 chemisorption <?n Fe. films, Beeck51 came to different results. At room

tem-. perature he found an approximately linear decrease of Q with

e

(see Fig. li, curve a), which is in

40

Q(Kcal/moi}

I

_ 8

0.2 0.<5 0.8 1.0

Fig. Ii. Heats of chemisorption of H2 on Fe films a: Beeck, 0., "Advances in Catalysis"

11, pag. 177.

(Calorimetrically at roomtemp. ) b: Bagg, J., Tompkins, F. C.,

. Trans. Farad. Soc. 51, 1071, '55.

(Calorimetrically)

-c: Porter, A. S., Tompkins, F. C., Trans. Farad. Soc. 217A, 529, '53. (m~ome tric ally)

d: Kwan, T., J. Res. Inst. for C~talysis,

IV, 207, '57. . .