A Scientific Approach To The Preparation Of Bulk Mixed Oxide Catalysts

485

G. Poncelet, P. Grange and P.A. Jacobs (Editors), Preparation of Catalysts III

© 1983 Elsevier Science Publishers B.V., Amsterdam - Printed in The Netherlands

A SCIENTIFIC APPROACH TO THE PREPARATION OF BULK MIXED OXIDE CATALYSTS Ph. COURTY and Ch. MARCILLY Institut Franqais du Petrole, B.P. 311 - 92506

RUEIL-MALMAISON (FRANCE)

Procatalyse, "Les Miroirs", 18 boucle d'Alsace

Paris La Defense zone 3

92400 - COURBEVOIE (FRANCE)

I. INTRODUCTION The purpose of this paper is to present an approach, as scientific as possible, to the selective synthesis of mixed oxide catalysts, i.e. those, which are uncommon ,

made of mixed oxides or true solid solutions and those, which

are more numerous, made of a complex association of mixed oxides (or solid solutions) and of single oxides.

We will restrict our discussion to bulk

mixed oxide catalysts. In our earlier publication of 1975 (1, 2), we had pointed out that homogeneity is a necessary condition for selectively obtaining mixed

oxide catalysts.

However, this condition is not always sufficient to characterize the microscopic composition of the catalysts. neity (on a 1 to 10 nm scale).

Some of them

~ndeed

show a controlled heteroge-

This is the case of the multicomponent

ammoxidation catalysts made of the association of Bi, Fe, Co, Mg and/or Zn molybdates in various compositions (3) and of some methanol synthesis catalysts made of the association of Zn aluminate (4) and of the CU solid solution 20-ZnO (5, 6). In a general manner, the superficial composition of the "ready for use" catalyst, which can be deduced from ESCA studies for instance, is different from the bulk composition.

This is often emphasized by the interaction bet-

ween the catalyst and the reaction medium. In all cases, the mixed oxide catalyst is the result of a complex synthesis method comprising

many elementary steps.

These steps must lead to the

desired state of the following characteristics : bulk structure, texture, superficial and bulk composition and metal oxidation state.

This implies the

fixing and the controlling of the main parameters of each elementary step and a systematic characterization of all the intermediary precursors.

In the

first part of this paper, some of the main steps will be studied. The mixed oxide coming from the last thermal activation step is usually different from the catalyst which has a steady-state interaction with the

486 reaction medium.

In the second part, transformations Qf the catalyst in contact

with the reaction medium

will be illustrated by a few examples.

II. THE VARIOUS METHODS OF SYNTHESIS OF BULK MIXED OXIDE CATALYSTS

Among the numerous methods reviewed in the literature (2), we will focus more particularly on those described in Fig. 1. only in specific cases.

Procedures b, c and d are used

Procedure a, which is by far the most common method,

will be illustrated by most of the examples given here. SOLUTION CONTAINING SOME PRECURSOR METALLIC IONS

~

__

Ib~_d

I

COPRECIPITATION

I

COMPLEXATION lacid_alcohols addn.)

GEL FORMATION

I

CAGINGJ

WA~~NG

OR~NG

I

CASlNGJ

DR~NG

I

MALAX;NG WITH OXIOES PRECURSORS

I

CAGING'

I

REMOVAL OF VOLATILE COMPOUNDS EXTRUSION OR I BALLS AGGLOMERATION Jl CAG'r

I

I

DRYING

I

THERMAL ACTIVATION

THERMAL ACTIVATION decomposilion ) ( of the complex.

\

THERMAL ACTIVATION

THERMAL ACTIVATION (solid.stale reaction)

~

ADDITION OF OTHER REMENTS lmpregnalioft

malaxrn; balls ogglomerotian

I

CAGING' I DRYING

I

THERMAL ACTIVATION '"

---(OPTIONNAll

FORMING PROCESS

, extrUSion,) (tabIet'i"' balls agglamerallon I . (AGING,

I

DRYING

I

THERMAL ACTIVATION

I CATALYST - READY FOR USEFig. 1

Different methods of synthesis of bulk mixed oxide catalysts.

487 The complex sequences shown in Fig. 1 for procedures a and, to a lesser extent, c and d, include the following main elementary steps : - separation of the hydrated precursor (Liquid to Solid reaction) - washing - hydrothermal transformations of the hydrated precursor (aging) - drying - thermal activation forming

- possible addition of other elements to the precalcined mixed oxide hydrothermal transformations.

During these elementary steps, important modifications of the solid occur.

In

the case of supported metallic catalysts, complex phenomena occurring during impregnation, drying and thermal activation are not perfectly known ; as well as the bulk and superficial composition of metallic association, the homogeneity and the dispersion may change, but the texture and the structure of the carrier are,in general,rarely modified.

In the case of bulk mixed oxide catalysts ho-

wever, the eVOlution is even more complex, as all these properties can change simultaneously.

This is one of the reasons why, in general, few publications

deal with the synthesis of mixed oxide catalysts and why the know-how is very rarely given. The fundamental study of the evolution of the catalyst precursor during each elementary step (the product obtained at each step becoming in turn the

precur~

sor for the next) has been,and still is,the subject of a great deal of research which should lead to a better understanding and, in many cases, to a better control of its transformation.

We will limit ourselves to the known general

principles of its production and illustrate them by a few examples. 11.1. Obtaining the hydrated precursor The aim of this step is to obtain a hydrated precursor which is as close

as

possible to the ideal precursor having the following qualities - relative proportions of active agents similar to those of the final catalyst - homogeneity - during decomposition, it should keep its homogeneity and permit the production of the active agents in the optimal states of oxydo-reduction and dispersion. The researcher has two main general ways to reach these targets : - separation of a solid phase within a liquid phase (discontinuous heterogeneous reaction)

: coprecipitation

- continuous transformation of a solution in a hydrated solid precursor (continuous homogeneous reaction) by gelification or complexation. Coprecipitation, the most common method, does not always allow the production

488 of a hydrated precursor with the desired composition.

In this case, one can

either make the coprecipitate change to this desired composition,by hydrothermal transformation for instance,(but such a transformation is not always easy to control) or use two other methods : gelification or complexation. 11.1.1. Coprecipitation 11.1.1.1. General principles Coprecipitation consists of precipitating simultaneously at least two metallic compounds within a solution, u5ually an aqueous one.

The supersaturation which

is necessary for the formation of the coprecipitate can be obtained : - physically:

by a change of the solution temperature

- chemically

by the addition of a compound which decreases the solubility

by solvent evaporation (common ions, pH modifiers, etc •.. ) by mixing solutions, the salts of which react to give an insoluble compound (double decomposition) • Precipitates can be very different from each other in terms of their morphological, textural and structural characteristics, but they can all be described in relation to two extreme cases - crystallized precipitates which are more or less hydrophobic - amorphous precipitates (gels, coagulates,flocculates) with an hydrophilic character. crystallized precipitates (chromates, molybdates, tungstates, phosphates, etc ..• ) have often a rigidly defined stoichiometry.

Examples of solid solutions

the composition of which can vary continuously and extensively, are not numerous. The

existence of such solid solutions which implies at least a partial misci-

bility

of the various compounds, is difficult to foresee.

MITSCHERLICH's

(1818) and GRIMM's (1924) rules, which express respectively the necessity for chemical similarity and for cases,

structural isomorphism, are obeyed in a few

but are far from being general.

Amorphous precipitates are not easily characterized by X-ray diffraction or by electron diffraction.

They usually have a flexible composition and are of-

ten unstable or metastable under their own conditions of preparation.

Many

precipitates are indeed intermediate between the two extreme categories defined above : they can move more or less rapidly from the amorphous state to a more organized one (7). Morphology, texture and structure depend on many parameters.

One of the most

important of which is supersaturation,which is crucial for the two main steps of the formation of precipitates : - nucleation (8) which is essentially homogeneous at high supersaturation and

489 heterogeneous at low or medium supersaturation - particle growth which occurs more or less with agglomeration and/or coalescence of particles. Among other factors which playa role in the quality of the precipitate, but to a lesser extent, we can mention the following : pH, temperature, nature of reagents, presence of impurities, method of precipitation, etc ... Supersaturation is a critical parameter for particle size, i.e. for texture. If it is high, the consumption rate of reagents by nucleation is much greater than that of particle growth : particles are numerous but small.

On the contra-

ry, if it is low and if solid impurities are absent (i.e. no heterogeneous a few but large particles are obtained.

nucleati~n),

Supersaturation is also very important for the structure of precipitates. During precipitation, many compounds can be obtained with several possible structures which are more or less metastable.

OSTWALD-LUSSAC's rule (9)

and

GOLDSMITH's principle of simplexity (10, 11) foresee that the higher the supersaturation the less stable and organized the structure is (higher simplexity).

As a matter of fact, when supersaturation is very high, dimensions of the

critical nucleus are very small and can be incompatible with the minimum level of information that must be included in this nucleus to form an organized structure (12).

In such conditions, only a metastable and poorly organized

phase can develop.

This phase can then change to another with less entropy

by hydrothermal transformation : this is the case of many metallic hydroxides and of the hydrothermal synthesis of zeolites which usually starts with a rapid gel formation.

Moreover, an intermediary metastable phase can favor

and orientate the heterogeneous nucleation of a more stable phase. In addition to the three characteristics mentionen above,

those of composi-

tion and, above all, of homogeneity must be added for coprecipitates. pitation rarely allows us to obtain a good macroscopic homogeneity.

CopreciIn a sys-

tem with two components A and B, the dAZ/dB z ratio of A and B within the elementary coprecipitate Z (which can be formed at any moment), and the A+S/B+S

ratio in the corresponding mother liquor S, are related to each other

by the derivative of the DOERNER-HOSKINS equation (13, 14) A+S B+S

, where

:

A is the heterogeneous distribution coefficient.

If the precipitation rate is low,

A~

1 : the coprecipitation is selective

and the coprecipitate is heterogeneous because the first crystals that appear do not have the same composition as the rest that follows.

If the precipita-

tion rate is high, the reaction· is limited by diffusion and A small

1 for ions with

differences in charges and sizes : the coprecipitation is unselective

490 and the proportions of components A and B in particles, as well as in the mother liquor, do not vary from the beginning to the end of the operation. Heterogeneities of the coprecipitate can also result from the method of coprecipitation : for instance, local heterogeneities of composition

may occur

in the system when reagents are mixed. 11.1.1.2. Pratical realization In a practical point of view, the main parameters are : nature and relative proportions of the metallic salts, concentrations, pH, temperature and the duration of the operation.

Other less obvious parameters may also be important

for instance, type and duration of stirring as mentioned mixed oxides precursors by JENSEN et al.

(15).

for iron molybdenum

This is consistent with our

own observations (not published). Fig. 2 presents three REACTION PROCESS FLOW _ SCHEME

FIXED PARAMETERS

VARIABLE PARAMETERS

Tempero/ure

pH

practical approaches to coprecipitation and indicates the

o

concen/rolions resid,nc, liru AlB rolio in s/urr,

fixed and variable parameters.

Devices

A and B are particularly well adapted

BATCH (,ariable pH) COPRECIPITATION

for producing crystallized hydrated pre-

pH

T,mperolure AlB rolio

concenlrolions resid,nce lime

cursors which are usually obtained at low supersaturation

G)

(high dilution, high

BATCH (fixed pHl COPRECIPITATION

temperature). Device A is used rather for

pH

Temperolure conCflnlrolion residence lime AlB rati«

CONTINUOUS (fixed pH) COPRECIPITATION

double decomposition at variable pH between soluble salts of metals (Fe, Co, Ni, Cu, etc .•. ) and heteropolyanionic complexes

Fig. 2 : Scheme and main parameters of the coprecipitation processes

and,more particularly, for the synthesis of multicomponent metal catalysts, during which the feed compo-

491 sition can be modified.

For both procedures, concentrations and residence time

are variable ; therefore, differences of crystalline organization between the start and the end of the precipita{ion are difficult to avoid as a result of precipitate aging in the presence of its mother liquor. Device C, which corresponds to a perfectly stirred chemical reactor, is the most satisfactory.

It operates under steady-state conditions (residence time,

concentration, pH, temperature) and allows continuous production and better control of precipitate aging in the mother liquor.

It is used rather for pro-

duction of amorphous or poorly organized precursors.

These precursors which

have a flexible composition are obtained at high supersaturation and low temperature. Sequencial precipitation, where a first precipitated precursor is used to initiate the nucleation of be

a

second precursor with a different composition, can

performed in devices B or C i.n controlled pH conditions.

This method is

described in several patents (4, 16). 11.1.1.3. Examples of coprecipitation a) introduction Catalysts for the synthesis of methanol or for the

low temperature

shift

conversion usually contain a large excess of copper oxide associated to zinc oxide and to trivalent metal oxides like those of aluminium and/or chromium. They can also contain various other additives. be written

Their bulk atomic formula can

as (5, 6)

(cuO) 1 (znO)0.05_0.7

~

A I 20 3

cr

20 3

j

0.05-0.4

They are usually obtained by precipitation at constant pH. Publications H4 (GHERARDI et al.), H5 (PETRINI et al.) and H7 (SHISHKOV et al.) of the present congress deal with the coprecipitation of various crystallized precursors, their characterization and the study of their decomposition products (17, 18, 19). Towards 1977, KLIER et al. 'mentioned.

(5, 6) characterized some of the precursors above

Table I, which is non-exhaustive, sums up their results and those,

more recently obtained by many other researchers (5, 6, 16 to 24).

Only one of

the seven products identified in Table 1 contains three metals : CU, Zn and Al or Cr.

It is probably difficult to obtain this particular one in a selectiveWly.

This precursor, partly described by KLIER et al.

(5) for the Cu-Zn-AI system,

has a lamellar structure, isomorphous to HYDROTALCITE Mg A1 C0 4 H 2(OH)16 6 20 3, which can be indexed in the hexagonal system (space group R3m). This phase, which is also described by TRIFIRO et al.

(17) in the present congress for the

CU-Zn-AI system, will be called (HC) in our text.

GERHARDITE and Na-Zn hydro-

492 xycarbonate (table I) are not suited as precursors because the NO; and Na+ ions favor the sintering of the active phase during the successive steps, especially during thermal activation.

(see Fig. 4).

TABLE I Crystallized phase formed by coprecipitation of Cu(Co) Cr AI(Zn) based hydroxycarbonates References

Formula

Name

3+ H M (OH) 16 C0 3-4 20 2 ++ ++ ++ ++ Cu ,Co ,Zn ,Mg A1 3+ cr 3+ Fe 3+

HYDROTALCITE TYPE PHASE (HC)

Cu 2+

MALACHITE TYPE PHASE (ROSASITE)

2-x

COPPER-ZINC HYDROXYCARBONATE (AURICHALCITE)

Zn 2+ (OH)2 C0

x

2+) (Cu 2+ zn 5-x x

(5, 16, 17, 20, 21)

(6, 22)

3

(OH) 6 (C0

3)

(18, 20')

2

BASIC COPPER NITRATE (GERHARDITE)

(6, 23)

, ZINC HYDROXYCARBONATE (HYDROZINCITE)

(6, 24)

i SODIUM-ZINC HYDROXYCAR-

(17)

ALUMINIUM HYDROXYCARBONATE (SCARBROITE)

(19)

NATE

b)

preparation of hydroxycarbonates from nitrates at variable pH

Precipitation curves of Cu, Zn or Al hydroxycarbonates are so different that it is very difficult to obtain a tion at constant pH.

homogeneous precursor other than by precipita-

Fig. 3 presents the neutralization curve of a solution

containing the three metal nitrates by disodic carbonate at 70°C : the three neutralization waves show that precipitation is heterogeneous and that each hydroxycarbonate precipitates essentially on its own.

However, it can be obser-

ved that the precipitation pH for Cu and Al (4.4 and 2.9) are much lower than those obtained when hydroxycarbonates are precipitated alone from a solution containing only one of these two metals (respectively 6.7 and 5.5). carbonate

precipitates at a normal pH (7 to 7.5).

Zn hydroxy-

A study by X-ray diffraction

of the hydrated precursor (Fig. 4.a) which looks homogeneous on a visual scale, shows that it contains essentially GERHARDITE and a small proportion of the (HC) phase described in Table I.

Analysis confirms the presence of large quantities

of nitrates. A mild calcination (350°C, 3 hrs) of this heterogeneous precursor leads to a well crystallized mixture of sintered CuO (TENORITE) and ZnO (ZINCITE)

(Fig. 4.b).

493

••

pH

T.70·C N03-/CO"z-

8

2 C03 2- II;

0.2

0.4

0.6

08

nh M O+

• CuO

• ZnO

Ij

(Mn+. AI"· ...Znt' ...Cu z·)

7

Calc.oxides mix!.

i \:~i

",--,---,-".:",

1.0

Fig. 3 : Neutralization with sodium carbonate (0.4 M) of Cu 2+, AI3+, Zn 2+ solution (0.35 M).

Cu AI In

Fig. 4 : X-ray diagrams of the hydrated precursor (a) and oxides mixture (b) from neutralization sample (Fig. 3).

f

Calc oxide

(31IJ

Cu ColnAI

Calc. spinel

:'

f)recursor .

•

,,-_',....- ;_.~.l ;\;.' ~il(;i.\t~; ,~;~--:,

~------:-:-:.~,:........,~.~.~r~_~----:-:

• {He] ternary phase malachite and lor rosasite phases

:\".~ ~\:;!. ~;:J~_ . _'.-~ __ y-;-_~.~- - :;:;

Fig. 5 : x-ray diagrams of Cu Zn Al precursor (a) and mixed oxide (b).

Fig. 6 : x-ray diagrams of Cu Co Zn Al precursor (a) and mixed oxide (b).

494 Alumina is not visible.

The relatively high value of the specific area of the 2.g- 1) calcinated product (50 m suggests that alumina has not sintered. Study of the precursor rich in GERHARDITE by electron microscopy shows

presence of large crystallites (up to 1 of crystals of TENORITE

the

Calcination leads to the formation

~m).

(CuO) of similar morphology.

KLIER et al.

(6) who pre-

pared his Cu-Zn mixed oxides according to the same method (slow addition in 90 mn of a 1 M carbonate solution in a 1 M nitrate solution at 80-85°C, with an increase of pH from 3 to 6.8-7), also obtained major proportions of GERHARDITE when the atomic ratio CU/Cu + Zn

~

0.5.

CAMPBELL had already pointed out, as

early as 1970 (25) that such a method leads to precursors in large crystallites and that it is preferable to operate in the opposite way (addition of a nitrate solution in a carbonate solution) in order to obtain finely divided precursors. c) batch precipitation of Cu-Zn-Al precursors Study of the batch precipitation of Cu-Zn-Al hydrated precursor shows that the crystallized ternary (HC) phase selectively appears onlyin a narrow range of compositions(l~.

If the atomic ratio Cu/Zn

<

1 and if the pH is

cons-

main~ed

tant (Fig. 5.a), a complex biphasic precursor is obtained, containing the (HC) phase and a Cu-Zn binary phase, ROSASITE as previously described by KLIER et al. (5, 6). (table I)

This phase is isomorphous and difficult to distinguish from malachite ; copper hydroxynitrate is

no longer formed.

After washing,

drying and mild thermal activation (300-350°C), a divided phase is obtained 2.g- 1 which has a specific area between 80 and 120 m and contains the three oxides in which Cu and Zn are partly combined (5, 6).

Only TENORITE (CuO) is

visible in the X-ray diagram (Fig. 5.b). d)

cobalt introduction in ternary precursors

Cobalt introduction in Cu-Zn-Al precursors results in a selective and highly homogeneous (HC) phase (16, 21). Fig. 6.a presents the X-ray diagram of this hexagonal structure (a = 0.305 nm, c = 2.24 nm, space group R3m) isomorphous with

the ternary structure (Cu-Zn-Al) mentioned above. The strong intensities

of (0, 0, 1) reflections and

scanning electron microscopy (STEM) (Fig. 7)

that this precursor has a lamellar structure. on STEM reveals it is highly homogeneous.

show

X-ray fluorescence microanalysis

After washing, drying and

moderate

thermal activation (350-450°C) a homogeneous and well divided spinel-type phase (a = 0.810 nm) is obtained (Fig. 6.b) its specific area is between 150 and 2.g- 1 200 m ; TENORITE is no longer visible. e) influence of operating conditions of coprecipitation on the crystallinity of the Cu-Zn-Co-Al hydrated precursor. The level of organization of coprecipitates depends on the nature and on the relative proportions of the various metallic ions.

For a fixed composition,

495

Fig. 7 : Electron micrograph (STEM) of (HC) phase and related microanalysis

':[003J

OPEIWING COMJITKlNS

[006J

1[012J i i

(015) I I

~R

Cu_Co_ALZn based TYPE OF precursor; [He] phose

:[018J

SODIUM

SATU- (%wtl

RATION

~Ii:'

T"C (Fig.21 RANGE

IN

~~D

OXIDE

[1019) [OTIJ] B 60-90

lOW

0.0050.015

CIS 50·80 MEDIUM 0.02' 0.05 10-40

HIGH

0.1' 0.3

Fig. 8 : Influence of the operating conditions of coprecipitation (supersaturation level, duration and temperature of reaction)on the crystallinity and the residual alkali content of Cu Co Ai Zn (CH) precursors.

496 the structural organization will depend, as indicated in the general principles

(§ II.l.l.l.), on the operating conditions and especially on supersaturation and on the length of the reaction.

Fig. 8 (26) which concerns hydrated precur-

sors of Cu-Zn-Co-Al mixed oxides (catalysts for the synthesis of alcohols described in ref. 16 and 21), shows that changes in operating conditions allow us to control the crystallinity of the precursor precisely without modifying its degree of homogeneity.

These precursors have the lamellar (HC) structure

previously described. 11.1.2. Complexation To avoid the usual imperfections of coprecipitation methods, a more general method was

developed a few years ago (27, 28, 29).

It permits the production

of an amorphous solid compound with a vitrous structure and of homogeneous composition, without physical discontinuity (phase separation) from the starting sOlution.

This is achieved by evaporating a solution containing various metal-

lic salts in any proportion and a complexing acid.oc

alcohol.

This method was

used for the preparation of many mixed oxides (29) and many bulk and supported catalysts (30, 31, 32). The metallic elements to be combined are added as soluble salts (or as reactive compounds) to an aqueous solution containing 0.5 to 2 gram equivalent of acid per gram equivalent of metal.

The following complexing acids are mainly

used: citric, maleic, tartaric, glycollic and lactic acids. complexing,

With citric acid

the evaporation under vacuum controlled conditions of a solution

results in a vitrous, transparent, amorphous solid precursor which is a mixed hydroxycitrate of the various metals. very homogeneous and isotropic.

Its composition and its structure are

Its thermal activation leads to the desired

mixed oxide. Table II presents the composition of such a citrate-type precursor containing Al and Y as a function of treatment conditions (29).

Such

decomposition is

progressive if metals active for oxidation (Cu, Ag, Fe, Co, Ni, etc ... ) are absent from the precursor.

In presence of such metals, thermograms of deeompo-

sition of the vitrous precursor usually show one or several plateaus; the semi-decomposed precursor does not contain any clearly illustrated by Fig. 9 (33, 34).

more

nitrate ions.

This is

In the presence of metals with oxydo-

reduction properties, it is possible to control thermal decomposition only by decreasing the concentration of nitrate ions or by dispersing the acid OC alcohol complex on a porous carrier (32).

497 TABLE 2 Composition of a citrate-type precursor of the Y AI0 of treatment conditions

Conditions of treatment

Product

In1 tial solution

3

perowskite as a function

Formula (calculated after analysis)

T '" 40°C

(COO'Iplete dissolution)

Al~~132

Y~~125

(00;)0.426

(C6H507)~~115(C6H807)O.126{H20)O.658

Semi -deccepcsec , amorphous

T = 135°C -

10 hrs (partial decomposition in vacuum)

Al~~132

Y~~125

(t«>~)O.OO3

(C6H507)~~154

Mixed oxide

T = 500°C, 4 hrs

rv

Amorphous ,"l transparent vi trous precursor

T :: 60°C - 20 brs (drying in vacuum)

opaque precursor

(OB-l O. 30 6

A10 - 0.0028 A1 3 20 3)O.125 (perowskite substoechiometric phase)

(air calcination)

11.1.3. Gelification by polymerization in solution In a few cases, it is possible to prepare an amorphous hydrated precipitate having a strong hydrophylic character that favors its interaction with the mother liquor and its transformation by tridimensional

reticulation in a

homogeneous hydrogel retaining the majority of the solution in its net. Such is the case for iron molybdate gels (pure or with additives) which were developped at I.F.P. methanol to formol

(35) for the production of catalysts for the

(36). Fig. 10

oxidation of

(36, 37) shows that gel formation only occurs

in a narrow range of composition and operating conditions.

Main parameters

are the Mo/Fe ratio, the nature of salts, the possible additives and their relative proportions, the temperature, the concentrations and the type of stirring.

Aging increases reticulation and homogeneity; the hydrogel becomes

hard and brittle

(conchoidal break characteristic of hydrogels).

Depending

on the concentrations and Mo/Fe ratios, the precipitate will partly or totally dissolve to give a gel. atomic ratios Mo/Fe

~

A true metastable solution is transiently formed for 1.5 (37).

The aged gel is then dried to give a bi own

transparent, homogeneous xerogel containing less than

wt % water and which

produces the activated catalyst after a succession of other elementary steps (Fig. 1). 11.2. Hydrothermal transformation of the hydrated precursor 11.2.1. General principles on hydrothermal transformations These transformations cover all spontaneous reactions between a solid and an aqueous solution (0-300°C) or water vapor (200-500 0C) at atmospheric pressure or under pressure. calcinated solid.

They concern the hydrated precursor as well as the

498

weigh!

loso% 0

M~

~'., .~

~<,

\

50

~O

~f~'

0

100

®

Fe (at)

Ik

0

COLLOiDAL PRECIPITATE .... solution .... elastic,transparent, homogeneous gel

120.

\ Ni

--

~

(T.IO_20·cI

'.

©

roc 100

200

300

40Q

500

COLLOiDAL PRECIPITATE .... opaque, hard,britfle gel

2

Fig. 9 : Effect of oxidation active metals (Fe, Ni) on the thermal decomposition of amorphous citric complexes (33, 34)

(T.IO _ 20'C)

.... transparent.homogeneous gel (T.2O _ 400CI

@

0

COllOiDAL PRECIPITATE .... heterogeneous gel+ precipitate

3

AGING

II

T.SO_BO·C

PHzOllfP sot.

4

II "

mechanical stirring

II

II

[M004]2[Fe]3+ion.g.L-'

15

0.5

2

25

Fig. 10 : Effect of the concentrations and Mo/Fe ratios on gel formation for iron molybdate precursors. Inthe

presence of aqueous solution or water vapor, a solid spontaneously

modifies with a decrease in free energy.

This results in changes of one or

more of the following characteristics : bulk or superficial composition, precursor homogeneity, texture, structure.

Hydrothermal transformations aim at

acting as selectively as possible on these characteristics. Any change in the free energy, sum of two terms ; tne first one,

6 6

of the solid can be considered as the

G~,

v'

G

is related to the number and to the

strength of bonds within the solid; the second one, 6G , is assumed to be s to the change of specific area 6 (6Gs = k , 6:f)

'I

proportional

0

When the hydrothermal transformation is essentially a textural one (the structure does not change : 6GVN 0) , 6 solid

increas$~and

f

is negative : the particle size

of the

the specific area decreases. This happens when autoclaving

a pseudoboehmite hydrogel in the presence of liquid water between 200 and 300·C, the specific area decreases from 310 to 72 mL.g- 1 ; if a mineralizing agent

499 (NaOH

0.01 M) is added, the specific area can decreases to 13 m2.g- 1 after

42 hrs (3B). In structural hydrothermal transformations, there is either a solvation

(Ii G :;II!: 0). If Ii G v v be positive: this results

change or a modification of the structural organization

IliGvl

is negative and if

is high enough,

.A~could

in an increase in the specific area as in the case of zeolite synthesis from gels. In general, hydrothermal transformations proceed in three steps ; -"breaking of the bonds at the solid surface,solvation - diffusion of

and/or dissolution

more or less solvated species

- integration of the species into a new particle or a new structure. In the liquid phase, dissolution or diffusion are the limiting steps. Adding mineralizing compounds (acid, base, salt) and/or increasing the concentration gradient between both solid phases are ways of increasing the rate of these steps. 11.2.2. Examples 11.2.2.1. Hydrothermal transformation in the presence of aqueous solutions As early as 1939, H. FORESTIER and J. LONGUET (39) found out that hydroxides coprecipitated or in physical mixture, kept in boiling water can react with each other and transform into crystallized mixed oxides.

Ferrites with spinel-

type structure are thus obtained. by aging a mixed hydroxyde of or Ni between 60 and 220°C. Ni cr

Fe and Zn, Co

can be obtained by the same method even

204 at room temperature in about 13-15 days (40).

The aging of mixed hydroxides

of Al and Ni at 100°C results in the formation of a crystallized hydrated precursor having a lamellar structure isomorphous to ANTIGORITE Mg Si 0S(OH)4 2 3 and close to that of the (HC) hexagonal phase. Calcining this precursor at 1000 0C leads to the Ni Al

spinel. Binary couples Zn-AI and Mg-AI 204 give a similar precursor (HYDROTALCITE for Mg-Al). Fig. 11 and 12

(41) present the X-ray diagrams of the products originating

from coprecipitating Zn-AI hydroxy carbonates (70°C, constant PH) and by aging them. If aging is short, a mixture of a spinel is obtained.

(HC) - like phase and of the Zn Al 204 A B houmaging at BOoC in pure water favors a selective

growth of this (HC)- like phase, without noticeably changing the proportion of spinel.

After washing, drying and calcining (450°C, 3 hrs) , the desired 2.g- 1). spinel) is obtained in a divided form (260 m 204 Nuclei of Zn Al spinel formed during coprecipitation certainly favor the 204 transformation of the precipitate into the spinel phase.

mixed oxide (Zn Al

Crystallized hydrated precursors containing copper are metastable.

Their

drying must be performed in particular conditions. Aging them at GO-BO°C in the presence of water, causes their hydrolysis (Fig. 13) and leads to a recrystal-

500

• [HC] precursor spinel

[311J

aging. B hrs.lBO"C

[2201

[400]

b

[4~21~

[511] [440] ,

b

• [HC] precursor

ZnAI204,[CHj ~rec.

spinel

aging, 0.2 hr.l20"C

aged precursor

Fig. 12 : X-ray diagrams of (HC) precursor (a) and activated Zn A1 20 4 mixed oxide (b)

Fig. 11 : Effect of aging on crystal(HC) type prelization of Zn A1 204 cursor

..

•

CuO

• [He] malachite and/or rasasite

a

Fig. 13 : X-ray diagrams of Cu Al Zn, (CH) type dried precursors : a properly dried b : dried under high steam partial pressure

501 lization of black cupric oxide which is easily identified by X-ray diffraction. The crystallization of the (HC) phase is improved by this treatment. Another type of undesirable aging can occur "if the desired mixed oxide has a high rate of sintering in the condition~" by TRIFIRO et al.

of its formation.

As shown

(42), such a reaction occurs with the hydrated precursor of

the Fe-Mo catalyst and leads to a stoichiometric Fe (M00 mixed oxide in a 2 4)3 2.g- 1) poorly divided form (3.5 m and with poor catalytic properties.

11.2.2.2.

Hydrothermal transfornlation in the presence of water vapor

Examples of hydrothermal transformations of solids in the presence of water vapor are numerous.

In a lot of cases, they result in

sintering characterized

by a decrease of the specific area at nearly constant pore volume. A specific hydrothermal effect has been observed by with a low sodium content (~O.7 water vapor (43).

calcining NH zeolite 4Y wt %) under a controlled partial pressure of

As shown in Fig. 14, calcination above 450°C using very dry

air leads to a tremendous decrease in the specific area due to the collapse of the structure.

A metastable and very disorganized HY form is obtained after

calcination at 500°C under a small partial pressure of" water vapor (about 30 Torr)

; the

porosity of such a metastable product is mostly intact

2.g- 1). sorption = 21.7 wt % at 25 Torr and 2SoC, S = 712 m The evolution 6H6 of this metastable form can be directed in two different ways depending on the (C

treatment : - complete collapse of the structure if calcination is performed above 500°C under very dry air - stabilization of the structure if calcination is continued at 500°C with a high partial pressure of water (about> 150 Torr).

The zeolite is thus cha-

racterized by a well organized structure, an important microporosity (C 6H6 2.g- 1) sorption = IS.4 wt %, S = 672 m and a high thermal stability (up to about SOO°C) . If NH is calcined from 150 to 500°C under high water partial pressure 4Y ( > 500 Torr) a limited collapse of the structure is observed (C sorption 6H6 2.g- 1) 12.2 wt %, S = 406 m (43, 44). All these results can be explained by the intervention of two competitive modification processes : one is aluminum extraction from the framework, the other is migration of silicon created by the removal of aluminum. on the pperating conditions. is much faster than silicon

toward holes

Relative rates of both processes depend

The structure collapses if aluminum extraction migration.

crystallinity is maintained if the

rates of both processes are of the same order of magnitude.

502

SAMPLE

nr

CALCINATION final

lTemp atmosphere °C

.

wt%

adsorbed C6H6

SAMPLE CHARACTERISTICS S m~g_l

1

450 very dry air

< 1 <10

2

500 slightly wet air

21.7 712

« 30torr H2O)

3

product 2

500 reea/eined 18.4 672 under wet air (>200torr H2O

4

X .ray Diffraction

J ~

1Jld II

air... water

500 vapor from 12.2 406 150°C

>500tarr H2O)

.1.

".~J.

I

"J

Fig. 14 : X-ray diagrams of NH zeolites after air calcination under various 4Y conditions 11.3. Washing of the hydrated precursor This elementary step aims at removing compounds (by-products of coprecipitation) retained in the hydrated precursor and which are undesirable because of the various following risks : - inhibition or modification of activity and/or selectiVity of the final catalyst - bad evolution of the precursor during the next elementary steps. Compounds to be removed can be divided into two groups : - those dissolved in the mother liquor still present within the porosity ions, mineral or organic molecules. - those fixed at the surface of the hydrated precursor

mainly ions.

503 The first ones can easily be removed by simple washing with distilled water. But the ease of washing will depend

·on the nature of the precipitate.

Amorphous precipitates like hydrogels or coagulates are in general hydrophilic, voluminous and their sedimentation is slow.

They are difficult to wash out

because of diffusional limitations in their particular porous texture.

On the

contrary, well crystallized precipitates decant more easily and can be washed out without difficulty.

This case is well illustrated by Fig. 8 which shows

that the sodium content of washed Cu-Zn-Co-Al

mi~ed

oxides decreases when the

crystallinity of the precursors increases. Removal of ions of the second group requires special washing conditions which will be briefly described.

An amphoteric oxide placed in water acts either as an

anionic or cationic exchanger depending on pH and ionic strength conditions. Its surface, which is indeed essentially

positively or negatively charged,

is neutralized by ions from the solution (45 to 48).

The isoelectric point

(which corresponds to an overall neutral surface containing an equal number of positive and negative charges) is reached when the pH value is equal to pHi. If pH

>

pHi' the solid surface is negative and the solid acts as a cationic

exchanger.

If pH

<

pHi' the solid surface is positive and the solid becomes

an anionic exchanger.

Two methods can

theoretically be used to remove unde-

sirable ions : - replacement of these ions by ions of the same kind, which are not inconvenient or which decompose easily during calcination : this is ionic exchange - washing under pH conditions where undesirable ions are not fixed. anion can be eliminated by washing with water at pH

>

Thus an

pH., because in this ~

condition, the solid is negatively charged and does not fix anions.

For ins-

tance, nitrate, sulfate and chloride anions can be eliminated by ammoniacal washing.

Sodium and calcium can be eliminated by acidic washing.

Suspensions of amphoteric oxides are washed in most cases by the second method which makes the removal of the strongly adsorbed ions easier.

Suspen-

sions of non-amphoteric ion exchangers like silica rich silica-alumina and zeolites, are washed according to the first method.

Their framework is indead

negatively charged whatever the pH may be, due to the valence 3 and the coordination 4 of aluminum.

In both kinds of washing, particular attention must be

paid to some secondary reactions that might occur during the contact between the hydrated precursor and the washing solution (hydrolysis, selective tions)

dissolu~

(49).

11.4. Drying Drying is aimed at removing water filling the pores as well as adsorbed water.

Obviously, it is rather difficult to dissociate this elementary step

and its secondary effects, especially textural and structural hydrothermal

504 effects. Classical drying is performed under atmospheric pressure with a high air flow rate and for long periods.

Particular techniques such as spray drying

or vacuum drying have also been developed. Drying of well crystallized compounds with low or medium. porosity is usually an easy operation ; texture and structure are not affected if drying is fast enough to limit hydrothermal effects.

On the other hand, drying of amorphous

or poorly crystallized precipitates and of the organic complexes previously mentioned can be responsible for considerable morphological, textural and structural evolutions.

In this case, this elementary step is important because of

its secondary effects. Specific phenomena which occur during drying of a hydrogel are due to interfacial tensions which develop within the porosity.

Below the critical

temperature of water, drying is characterized by a contraction of the solid and therefore usually leads to a xerogel of high density and low porous volume. Above the critical temperature, capillary tensions no longer exist and an aerogel of low density and high porous volume is obtained (50).

Effects

of interfacial tensions can be limited in several ways : - hydrothermal pretreatment resulting in an increase of pore diameter (at constant structure) - addition of a substance which lowers capillary forces (51). Secondary effects can be limited by drying at low temperature under vacuum or low pressure, or by fast drying like spray drying. 11.5. Thermal treatments

The objective of thermal treatments is to transform the previously dried precursor into a mixed oxide (or an association of single and mixed oxides) having the desired characteristics of texture and structure.

During these

treatments, volatile or decomposable compounds still present after the preceding step, are eliminated. It seems that there are no general rules concerning thermal treatments. It is only to be noted that some interferences often appear between the following reactions : - thermal decomposition of the precursor - formation of the mixed oxide - thermal sintering of the mixed oxide. Many parameters among the selected operating conditions (temperature, pressure, atmosphere. procedure of activation, solid-gas contact technology, etc .•. ) can playa decisive role on these three reactions. In other respects, in severe thermal conditions (T

~

400°C for instance), solid state reactions

505 or composition modifications by sublimation of volatile compounds (such as M00

v etc •.• ) can occur. Suitable programming of the operating conditions 3, 205' could be established with the aid of a study of the thermal decomposition of the precursor (TGA, DTA) and of correlations between temperature and sintering, like those presented in Fig. 15 for Fe-Cr-K mixed oxides which catalyse ethyl-

benzene to styrene dehydrogenation (52). As

in supported metallic catalysts, the kinetics of sintering depends

largely on the partial pressure of water vapor and also on the presence of various compounds (alkaline ions, nitrate ions, etc ... ) in the solid and/or gaseous phases.

This is the case in thermal decomposition of organic complexes

and of the Fe-Mo xerogels previously mentioned,

which may have retained in

their porosity some of the by-product ions of the precipitation.

These cases

require particular thermal and/or hydrothermal treatments (36). Another example presented in Fig. 16 (26) concerns the influence of thermal sintering on the degree of the crystalline organization of a divided AB

-£ AO spinel phase containing an excess of the bivalent ion A (A B = Al).

20 4,

= Co, Cu Zn

In spite of this excess above stoichiometry, these phases remain

remarkably homogeneous and highly dispersed when the temperature and the length of calcination increase.

If the same hydrated crystallized precursor

contains significant quantities of alkaline ions, even in the absence of nitrate ions, a similar preparation concluded with a mild calcination (350·C, 3 hrs) leads to two distinct phenomena : - demixing of cupric oxide having a TENORITE structure easily identified by X-ray diffraction - sensible sintering of the spinel phase 2.g- 1 from 190 to about 110 m A study

the specific area decreases

of this product in electron microscopy coupled with X-ray fluorescence

microanalysis reveals the CuO demixing areas and shows that the composition of the spinel phase remains homogeneous (Fig. 17)

(21, 53).

These phenomena

of demixing and sintering will be emphasized if the mixed oxide is calcined at higher temperature. To conclude on thermal treatments, let us remember that the solid state reactions mentioned

in Fig. 1 can be accelerated by artificial nucleation.

In the case of the Ni cr et al.

(54).

2

04 sYnthesis, this has been illustrated by CHARCOSSET

11.6. Possible addition of other elements to the precalcined mixed oxide. Forming. Hydrothermal transformations. Usually, the thermally activated mixed oxide still has to go through a few operations before obtaining the "ready for use" catalyst.

These operations

506

O.

average pore diameter l¢ pAl

cumulated(Vp cm 3g- l ) porosity

( 0.2000 AIm)

Vp

5000

025 4000

3000

0.20

specific area

4

(5 m2g " )

3

2000

s

1000

hrs. oir calcinotion at 900

.r-c

950

1000·C

Fig. 15 : Fe Cr K oxides based catalyst for dehydrogenation of ethylbenzene to styrene. Influence of the calcination temperature on the textural properties.

, ~[hkIJ spinel

"I

;]\1 phase

,j 1\\ ;1'\\\

.; !\\Vt~~~

"'\\~l ;; I :°11

i:j

I

:q ,

\

[311]

[2~0]

;

11\

::J

[440] [4001 I

i ~;~I

I

I

II

:I

'1\I \

II~'I

I I

I I

II

Iii

i II

I

I

L',.. . . ..... : :1~'~:

OF CALCINATION (M 2G-')

[51ll

I \

)1I :~\.

TEMPERATURE SPEC.

AND DURATION AREA

I

450OC/20hrs.

158

4500C/3hrs.

170

350OC/3hrs.

190

::

I

Fig. 16 : Cu Co Al Zn mixed oxide (spinel structure). mal sintering on the degree of crystallinity.

Effect of ther-

507 are mainly : - addition of other elements to the calcined

mixed oxide by means of

impregnation or malaxing with a solution - the forming of the catalyst after a controlled moistening of the mixed oxide (balls agglomeration, extrusion, etc .•. ). During both operations (the two can be combined in some cases (55»

or during

subsequent drying, hydrothermal transformations of the activated mixed oxide may occur.

11.6.1: Deposition of metals on the precalcined mixed oxide Rules which govern impregnation and malaxing have been extensively described elsewhere (48, 55).

A fundamental study of ion exchange properties of mixed

oxides must be undertaken for each particular case. 11.6.2. Forming by controlled moistening Many pUblications deal partly with this elementary step (55, 56).

The mixed

oxide must have a hydrophilic character and its moistening must lead to a phenomenon of setting or allow the formation of a paste.

homogeneous and thixotropic

Both properties are characteristics of an imperfect state of crystalli-

zation.

If the mixed oxide does not present at least one of these properties,

it must be formed by tabletting. 11.6.3. Hydrothermal transformation of the calcined oxide A genuine example of hydrothermal transformation of the calcined

mixed

oxide is that of an industrial catalyst for the synthesis of methanol (Fig .18)(41). In most cases, the crystalline structure of this oxide is deeply modified by a simple moistening followed by a mild drying.

A partial reversal is observed

as well as recrystallization of the (HC) phase isomorphous to HYDROTALCITE. This is the reason why it is advisable to stock this catalyst in a dry atmosphere. As early as 1953, TERTIAN, PAPEE et al.

P and ~

for alumina.

(38, 57) described a similar reaction

aluminas, obtained by dehydration

hydrated aluminas at 200-400°C

un~rhigh

of crystallized

vacuum, can be transformed by rehydra-

tion and suitable aging, either into BAYERITE (T = 25°C, 14 days, liquid water) or into BOEHMITE (T

=

200-300 oC, liquid water)

(38).

To our knowledge, such a

reaction, well known for alumina, does not seem to have been described for mixed oxides. The few following examples concern again the mixed oxides used as catalysts for the synthesis of methanol. Zn Al

204

spinel phase, the preparation of which has been described in the

508 second part (calcination of the crystallized hydrated precursor 3 hrs at 450°C) is a well divided 2.g- 1) (260 m and homogeneous solid.

Its X-ray diagram is presen-

ted in Fig. 19.a (41).

Simple

moistening of this oxide, followed by aging at 80°C sence of

in the pre-

liquid water, resul ts in a'

recrystallization of the (HC) phase,whereas (Fig. 19.b) the degree of crystalline organization of the spinel phase does not change ; calcination must be

per~

formed at temperatures higher than 450°C and for a long period so as to avoid recrystallization by rehydratation.

Fig. 17 : Electron micrograph (STEM) of Cu Co Al Zn alkalinized mixed oxides and related microanalysis •

• CuO [HC) phase

Fig. 18 : Effect of moistening, aging and drying on the crystalline structure of an industrial catalyst for methanol synthesis.

Fig. 19 : x-ray diagrams of air calcined Zn A1 before (a) and after 204, (b) moisten1ng, aging and drying.

509

Fig. 20 : X-ray diagrams of air-calcined Cu Al Zn mixed oxide (Fig. 5) before (a) and after (b) moistening, aging and drying .

• CuO

• [Hqternary phase

.

. .a

e .

.....

• •\:\.. . . . . __"'_./.-.<'.., ........>:

~

-'

CuZnAI

.

...

.......b

Calc.oxide

.'

With Cu-Zn-Al mixed oxide, the preparation of which has been previously described (Fig. 5), a similar phenomenon is observed (Fig. 20) case, i t might be due to

(41).

In this

selective recrystallization of hydrated zinc alumi-

nate, because copper aluminate neither recrystallizes in a hydrated state nor in an anhydrous state when calcination temperatures are lower than 450°C, and its rehydration causes a partial hydrolysis in CuO + Al(OH)3 (58, 41) without any formation of the (HC) phase. tallizations which

a~e

Water is essential for these recrys-

not observed in non aqueous medium.

Mixed oxide catalysts are usually thermally activated at temperatureS

which

are not too high in order to obtain a high state of division, that is to say a poor state of crystallization.

So it is not very surprising that such solids

are reactive enough to be hydrothermally transformed. III. TRANSFORMATIONS OF MIXED OXIDES IN THE REACTION MEDIUM 111.1. Introduction The 'ready for use'catalyst made of the thermally activated mixed oxide is, in most cases, very different from the steady-state catalyst which is in equilibrium with the reacting medium under catalytic operating conditions. In mild oxidation, reaching the steady-state does not cause an extensive modification of the bulk or superficial composition of the catalysts. But in many other catalytic operations, the composition, the texture and the structure undergo important changes owing to interactions between catalyst and reactants, products and by-products.

Among these modifications, the following can be

emphasized : - formation of divided alloys or metals (hydrogenation, CO + H catalysis) 2 - formation of single or mixed sulfides and/or sulfur-metal association (hydrotreating) - superficial coking and structural modifications (dehydrogenation)

510 - carburization (CO + H catalysis) 2 - nitridation (NH synthesis or decomposition) 3 - modification of bulk and/or superficial composition (thermal effects on volatile oxides, chemical or physical migration, attack by reactants or products). It is not the purpose of this communication to deal with catalyst deactivation in the reaction medium.

This has been treated in

in various specific congresses (59, 60).

many publications

We will restrict ourselves to modi-

ficat.ions which lE:,ad to the ini.tial steady-state.

Study of these modifications

implies a precise characterization of the steady state catalyst (bulk and superficial composition, texture, structure, oxidation degree of metallic elements, mechanical properties, etc ... ). This leads us to two remarks : - the steady-state depends, in most cases, on the operating conditions. It is sometimes reached after a long period (up to 1000 hrs) and is very difficult to characterize lIin vitro" - there are no general rules which permit foresight into and control of the catalyst evolution toward the steady-state.

This evolution is often even

much more complex than the preparation of the catalyst. The following

three examples will illustrate the main problems which can occur.

111.2. Molybdenum or vanadium based

catalys~for

mild oxidation processes

Under reaction conditions, many mixed oxide catalysts for mild oxidation undergo significant modifications of their superficial composition and very often oxide sublimation (M00

V which deteriorates the homogeneity and 3, 205) the textural properties of these catalysts.

For instance, migration of M00 has been observed in the Fe - Mo0 2(M004)3 3 3 oxide combination used as a catalyst for methanol to formol oxidation (61, 62, 63), in Mo0 - V supported mixed oxides used as catalyst for benzene to 20 S 3 maleic anhydrid oxidation (64) and also in the M.C.M. catalysts (Bi, Co, Fe, Ni, molybdates) for ammoxidation of propene (65).

In the latter case, an

enrichment of the surface in Mo, Bi and Fe is observed.

These composition

modifications are due to the thermal instabilit:y of molybdate phases in the operating conditions.

At least two parameters are decisive:

- the average temperature of operation and the temperature of the hottest point of the catalytic bed - the partial pressure of oxygen. The partial pressure of oxygen is often decreased when recycling a part of the used air after the water absorption of reaction products. raising

This allows the

of the lower limit of ignition of the reaction medium and the use

511 of higher concentrations of reactants, but has two consequences - an increase in the heat of the reaction per unit volume of catalytic bed - an increase in the rate of the reduction of the catalyst, in the MARS and VAN KREVELEN - type equilibrium (66) oxidized Cat. + reduced Cat. +

Cat. + products

reactants~reduced

oxidized Cat.

02------------~.~

(M00 - Mo0 oxide mixture, the reduction 2 4)3 3 species (62, 63) is accelerated and results in a

For instance, in the case of Fe toward the

IX - or

demixing of Mo0 The c~talyst

3

p-

Fe MoO4

which mOVes inside the solid. stabilization under process conditions can be improved by

formula modifications and by using milder operating conditions (lower concentration of organic reactant, dilution of the catalyst, etc •.. ).

In all cases, the

best economical compromise must be sought. 111.3. Catalyst for ethylbenzene to styrene dehydrogenation This bulk catalyst is composed of ferric oxide promoted with chromium oxide and stabilized by potassium.

Moreover, it

can contain various other textural

and/or structural promoters (67, 68, 69). The "ready for use" catalyst is in general obtained by malaxing iron hydroxide or oxide with a solution containing promoters. ded, aged and then dried.

A

The paste obtained is extru-

solid-state reaction (procedure d of Fig. 1) is

performed in severe thermal conditions (7OO-1200°C). Fig. 15 (paragraph 11.5) presents correlations between temperature and sintering for one of our catalysts. An X-ray diffraction study of the activated catalyst (Fig. 21) shows that it contains essentially the IX -Fe

HEMATITE phase (52). 20 3 Ethylbenzene to styrene dehydrogenation is performed with this catalyst

under the following operating conditions LHSV

=

0.3-0.7 h- 1, Steam/oil ratio (wt)

water vapor improves the performance is reached after about 1000 hrs. period, a slow modification of the

T = 580-590°C, P = 2 bar, 2.

Dilution of ethylbenzene by

and limits coking.

The true steady-state

During the first 300 hrs of this transitory crystalline structure occurs

catalys~s

whereas its texture is not significantly affected (Fig. 21). is reduced

to the ferrimagnetic Fe

phase.

The IX-Fe phase 20 3 Chromium and potassium oxides

30 4 Cr 04' Reduction of Fe does not go beyond 2 2 20 3 because the molar ratio PH2/PH~ is always less than 0.1 (52). Simulta-

interact and lead to

IX K

Fe 0 3 4 neously, secondary reactions of dealkylation under water, which lead to benzene and toluene

decreases,and the molar selectivity in styrene increases in a

significant way. more selective (Fig. 22)

(70).

After 1000 other hours of operation, the catalyst is much although the structure obtained after 300 hrs is not modified

512 x

X [104J ./00

[1I0J .50

CAT.AFTER

300 HRS. PILOT TEST

X (116]. 60

X mognelile(Fe,O.) .. promotors

[021].25

X [02.]40

X

(300]35 X X (214]35

o K2Cr040(

X[113] 30

60

40

50

20

e[311]100 30

NEW CAT. e [440]. 40 e(120J30 e(511] 30

hemotile( Fe20,c<) . promotors e

e (400].10

..

• (422]10

60

50

40

30

Fig. 21 : Fe Cr K oxides based catalyst for dehydrogenation of ethylbenzene to styrene. Effect of the stabilization (in reaction medium) on the crystalline structure.

Benz...

'!o(mol.)

90 80 70 60 50

0_._ 0_._

0_.-.-.-

0 - . - _ . _ 0 .......

I _'-0-'_0_.-01

i

i

T.59cJot p. 2.bar, TaSeo-c

___.. . . . _.

40

I

I

i r.s'O"C

I

I

j~

i

I

/i

I

°j I

,

30

i

20

•i I i

1

/

P.2Dor.

I

TEAMIOIL..2(wl)

1

200

..o_. __ .-oyield T.aeooc

LH$V. O.lth-1 STPIl/OILdtwt)

rr.6OO"C I

li~1O , i i Ethytbenzone 1::----', i'--'-'",,__, I

15

P.z bart

I

"'o__ o__ oj

I

0----

i I I i i i~.

0

__

conversiCJfl 0

__

Sw- yield

0

__

0

__•

i , l ' /

!

I I

I I

!..

··.411

1300

5

'~\,

I i i ' ,

~ __-------:

100

Toluene

•

I

f ~ LHSV.o..,tr-1 j STfAN I

I .".""",

:=.. .

j 0 .... 1.,//

I/ OOO hru

I-

STUNJOIL= 2(wI)

I I

I

I

LHS\I'=O..l4.-I

°'-_0 __ 0 __ 0 _ _ 0 __

I

I

S ...!!.~".!_~r!c!!!~!_._o_

i

T _ yiold "°"_0'"

0

.... 0

i~,--___o__--!~~~~

14

Fig. 22 : Dehydrogenation of ethylbenzene to styrene. and steam-treatment on performance .

1600

.~.o

__-r~,17

hours of operation

Effect of aging

513 At the end of this transitory period, injection of ethylbenzene is stopped and the catalyst is kept under water vapor at 600°C for

48 hrs.

After such

steaming, the catalyst presents again a relatively high dealkylating activity and a new 300 hrs period is necessary to reach the previous steady-state. Analysis of gas collected during steaming reveals the presence of a significant proportion of CO which probably results from superficial decoking of the 2 catalyst according to the water-gas and shift reactions; moreover, these reactions are catalyzed by alkaline metals. To a lower extent, one other reaction produces CO 2

but, under steaming and operating conditions, the amount of K transformed 2C0 3 is quite low (71). Therefore, the steady state of the catalyst probably results from two consecutive phenomena: (70). - a slow reduction of Fe into Fe 20 3 304 - a selective coking under water of sites active for steam-dealkylation. This steady-state is metastable.

A slight decrease of the steam/oil ratio

is sufficient to increase drastically the coking of the centers which are active for dehydrogenation and therefore to reduce heavily the production of styrene. The catalyst activity can be partly restored by steaming. The optimization of such a catalyst implies a setting up of correlations between the physicochemical properties and the stability of dehydrogenating activity under the most economical operating conditions.

The proportion, the

distribution and the degree of combination of the alkaline elements are some of the main parameters of this optimization (67). 111.4. Catalysts for the synthesis of C

to C alcohols 1 6 We will summarize here some recent results (21) obtained in the study of

the transitory phase of some catalysts in the synthesis of C to C alcohols. 1 6 These catalysts are composed of Cu-Co-Al and possibly Zn mixed oxides promoted by alkaline metals.

In paragraph 11.5., an

electron microscopy study of the

activated catalyst was presented: except for some copper oxide demixing, this catalyst is very homogeneous.

A T.P.R.

(Temperature Programmed Reduction)

study shows that the reduction of cupric oxide to copper is complete at 240°C ; the reduction of CoO to cobalt about 700°C.

= 6-10

So, i t appears that under the

MPa, H = 2, GHSV 2/CO gen is never complete. P

starts at 380°C and is complete at

=

process conditions (T = 260-300°C,

4000 hr- 1 NTP), the reduction of CoO by hydro-

When the catalyst, prereduced by hydrogen, comes into contact with

~he

514 synthesis gas under the process conditions, the following phenomena are observed :

2 - reduction of co + to metallic cobalt

- exothermic chemisorption of CO on metallic cobalt formation of cobalt carbonyl for certain compositions, which results in a cobalt migration - transitory reaction of the very exothermic methanation. Study of the catalyst stabilized in the reaction medium reveals a decrease of the cobalt content in the spinel phase, which is not easily detected by X-ray diffraction.

~Very

small crystals 1 to 3 run in size, containing Cu and

Co and possibly Zn, are formed (Fig. 23).

Under special operating conditions,

formation of large crystallites of cobalt carbide can be revealed by X-ray diffraction.

Fig. 23 : Electron micrograph (STEM) and related microanalysis of Cu Co Al Zn mixed oxides catalyst after H reduction and 2 aging in (CO + H medium. 2)

In this last example, transformation of the mixed oxide is extremely complex and

characterization of each intermediate of the solid transformation

is indispensable for the understanding and the control of its evolution.

515 IV. GENERAL CONCLUSION The preparation of mixed oxide catalysts, which is usually very complex, can be described as a succession of many elementary steps which are governed by more or less empirical rules and where the evolution of composition, structure and texture often occurs simultaneously.

A systematic and rigorous characteri-

zation of each intermediate is necessary to set up a continuous relationship between the first hydrated precursor and the final activated mixed oxide. No results have been presented on mixed oxides, the surfaces of which have properties (composition, degrees of oxidation) different from those of the bulk.

In general, such modifications occur during drying or thermal activa-

tion and sometimes during reduction of the oxide (72). The "ready for use" catalyst can undergo many transformations when put into contact with the reaction medium.

The examples given in the third part show

the diversity, the complexity and the specificity of these transformations. In some cases, the steady state catalyst is so different from the activated catalyst (Cu-Co crystals of example 111.4) that other ways of synthesis can be considered where the last precursor (the activated mixed oxide) is not always indispensable.

As a matter of fact, products thus obtained can give the opti-

mal steady-state catalyst more easily than

the activated mixed oxide.

Such

an approach, which is relatively new, could initiate in some specific cases the perfecting of new preparation processes. Mixed oxides occupy an important place in heterogeneous catalysis.

If alu-

minosilicates are excluded, mixed oxides catalyse the main following reactions synthesis of methanol and of C to C alcohols, FISCHER TROPSCH synthesis, the 2 6 hydrogen chain and some mild oxidations. They are also carriers and cocatalysts of many metals.

More recent works show that fields of application of mixed

oxides in catalysis are more and more diversified.

Important developments

in

research into the synthesis of mixed oxides are expected, more especially as many other fields (electronics ,microprocessors , magnetism, nuclear, etc ... ) initiate new research into this subject. Whatever the method of preparation, there must be a good compromise between scientific, ,technical and economical imperatives as shown in the following diagram :

process conditions

--

catalytic performances

tt

final catalyst optimization

1!

~

preparation procedure (costprice of catalyst)

process ec cnomy

516 REFERENCES Intern. Symp. on Scientific Bases for the Preparation of Heterogeneous catalysts (editors B. Delman, P.A. Jacobs and G. Poncelet) Elsevier, Amsterdam 1976. 2 Ph. Courty, Ch. Marcilly, General Synthesis method for mixed oxide catalysts, in ref. 1, 1st Symp., 1976, pp. 119-145. 3 M.W.J. Wolfs, J.B.C. Van Hooff, in ref. 1, 1st Symp., 1976, pp. 161-171. 4 I.C.I. Patent U.S. 3.850.850, 1972. 5 S. Melta, G.W. Simmons, K. Klier, Proceed. 7th Congr. on Catalysis, Tokio, 1980, pp. 475-489. 6 R.G. Herman, K. Klier, G.W. Simmons, B.P. Finn, J.B. Bulka, T.P. Kobylinski, J. Cat., 56, 19708, pp. 407-J2~. A.C.S. ANAHEIM Meeting, March 12-17, 1978, pp. 595-615. 7 G.K. Boreskov, Scientific Basis of Catalyst Preparation, in ref. 1, 1st Symp. 1976, pp. 223-250. 8 A.C. Zettlemoyer, "Nucleation", Edt. Marcel Dekker Inc., N.Y., 1969. 9 H. Furedi-Milhofer, 4th Intern. Conf. on .Surface and Colloid Science, Jerusalem, Israel 5-10 jUly 1981, in IUPAC - Pure and Applied Chemistry, Pergamon Press, 53, 1981, pp. 2041-2055. 10 J.R. Goldsmith, J. Geol., 61, 1953, 439. 11 D.W. Breck, "Zeolite Molecular Sieves, Structure Chemistry and Use", John Wiley and Sons, London, 1974. 12 Ph. Caullet, J.L. Guth, G. Hurtrez, R. Wey, Bull. Soc. Chim. Fr., 7-8, 1981, 1.253-257. 13 B.A. Doerner, W.M. Boskins, J. Am. Chern. Soc., 47, 1925, 662. 14 A.G. Walton, The Formation and Properties of Precipitates, Intersciences Publishers div. of.J. Wiley and Sons, 1967. 15 E.J. Jensen, K. Johansen, Man-Yu Topso~ J. Villadsen, This symposium (publ. F6 - Communication withdrawn) . 16 I.F.P. Patent NE 82.05368, 1982. 17 P. Gherardi, O. Ruggeri, F. Trifiro, A. Vaccari, G. Del Piero, B. Notari, G. Manara, This symposium (publ. H4). 18 G. Petrini, F. Montino, A. Bossi, F. Garbassi, This symposium (publ. H5). 19 D.S. Shishkov, N.A. Kassabova, K.N. Petkov, This symposium (publ. H7). 20 A.S.T.M. 14-191, 22-700, 25-521. 20'A.S.T.M.17.743. 21 Ph. courty, D. Durand, E. Freund, A. Sugier, Catalytic reactions on one carbon molecules, Symp. Bruges 06/82 (to be pUblished in J. of molecular Cat.' 22 S.D.H. Donnay, H.M. Ondik, Crystals datas, Determination tables, N.B.S. WASHINGTON D.C., 2, 1973. 23 I.M. Vasserman, N.I. Silant'Eva, Russ. J. Inorg. Chern. 13, 1968, 1041. 24 S. Ghose, Acta Crystall. 17, 1964, 1051. 25 J.S. Campbell, I.E.C. Process Res. Develop. 9(4), 1970, pp. 588-95. 26 Ph. Courty, D. Durand, E. Freund, B. Rebours, A. Sugier, Unpublished results, 1981. - C. Durand, B. Rebours, Ph, courty, Unpublished results 1982. 27 I.F.P.-C.E.A. Patents, F.l.604.707 (1968), F.2.045.612 (1969). 28 Ch. Marcilly, Ph.D. Thesis, Grenoble, 1968. 29 Ph. Courty, H. Ajot, Ch. Marcilly, B. Delmon, Powder Technology, 7, 1973, pp.21-38. 30 I.F.P. Patents, F.2.086.903 (1972), U.S.4.122.110 (19~8). 31 Procatalyse Patents, NE.71-13858, (1971), NE.72-12020, (1972). 32 Ph. Courty, B. Raynal, B. Rebours, M. Prigent, A. Sugier, I.E.C. Prod. Res. Dev. 19, 1980, pp.226-231. 33 J. Droguest, Memoire de licence en sciences chimiques, LOUVAIN (1972). 34 B. Delmon, J. Droguest, 2d Intern. Conf. on "Fine Particles". 35 Ph. Courty, H. Ajot, B. Delman, C.R. Acad. Sc., 276C, 1973, pp.1147-49. 36 I.F.P. Patents, F.1.600.128 (1968), F.2.060.171 (1969), F.2.082.444 (1970), U.S.3.716.497 (1973), u.S.3.846.341 (1974), U.S.3.975.302 (1976), U.S.4.000.085 (1976), u.S.4.141.861 (1979).

517 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52

53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72

Ph. Courty, H. Ajot, B. Delmon, Unpublished results (1967). R. Tertian, D. Papee, J. Chim. Phys., 1958, p.341. H. Forestier, J. Longuet, C.R. Acad. Sc., 208, 1939, p.1729. J. Longuet-Escart, Bull. Soc. Chim. Fr., 1949, p.153. Ph. Courty, Unpublished results, 1981. F. Trifiro, P. Forzatti, P.L. Villa, in ref. 1, 1st symp., 1976, p.148. F. Ribeiro, Ch. Marcilly, 5th Intern. Conf. on zeolites, Recent progress reports, Naples june 2-6 1980, pp.135-138. UNION OIL of Calif. Patent (HANSFORD), u.S.3.506.400, 1970. A.W. Adamson, Physical Chemistry of Surface, Intersciences Publishers Inc., N.Y., 1960. D.J. Shaw, Introduction to Colloid and Surface Chemistry, 2nd edition Butterworth and Co, LONDON 1970. S.J. Gregg, The Surface Chemistry of Solids, Chapman and Hall, LONDON 1961. J.P. Srunelle, Preparation of catalysts by adsorption of metal complexes on mineral oxides, in 2d Int. Symp. on Scientific Bases for the preparation of Heterogeneous Catalysts - Elsevier 1979, pp.211-232. H. Boerma, Preparation of Copper and Zinc Chromium oxide catalysts for the reduction of fatty acid esters to alcohols, in ref. 1, 1st symposium, 1976, p.105-118. D.R.M.E. (French State), F.2.050.725, 1969. E.J. Newson, J.V. Jensen, The effects of preparation parameters on the oxidation activity of catalysts made by coprecipitation, in ref. 1, 1st symposium, 1976, pp.91-103. Ph. Courty, J.F. Le Page, Relationship between average pore diameter and selectivity in iron - chromium - potassium dehydrogenation catalysts, in 2d Int. Symposium on Scientific bases for the preparation of heterogeneous Catalysts - Elsevier 1979, pp.293-305. M. Bisiaux, Ph. Courty, H~ Dexpert, E. Freund, Unpublished results, 1981. H. Charcosset, P. Turlier, Y. Trambouze, J. Chim. Phys., 107, 1964, pp.1249-1256, 108, 1964, pp.1258-1261. J.F. Le Page et al., "Catalyse de contact", Ed. Technip, 1978. Ph. Courty, P. Duhaut, Rev. I.F.P., 29(6), 1974, pp.861-877. D. Papee, J. charrier, R. Tertian, R. Houssemaine, Congres de l'aluminium, Paris, juin 1954. G. Barrera, Ph. Courty, B. Rebours, A. Sugier, Unpublished results, 1981. Conference on catalyst deactivation and poisoning - May 24-26 1978, Lawrence Berkeley Laboratory - BERKELEY, California 94720. International symposium on catalyst deactivation. Antwerp oct. 13-15 1980, (Studies in surface science and catalysis 6. B. Delmon and G.F. Froment editors) Elsevier 1980. S. Peirs, Ph.D. Thesis, Lille, 1970. Ph. Courty, J.F. Le Page, unpublished Results, 1976. N. Burriesci, F. Garbassi, M. Petrera, G. Petrini, N. Pernicone, in ref. 60, pp.115-126. A. Bielanski, M. Najbar, J. Chrzaszcz, W. Wal, in ref. 60, pp.127-140. T.S.R. Prasada Rao, P.G. Menon, J. Cat. 51, 1978, pp.64-71. P. Mars, D.W. Van Krevelen, Chern. Eng. Sci., 3, 1954, p.41. I.F.P. Patent, U.S.4.134.858, 1979. SHELL Patent, U.S. 4.052.338, 1977 - 4.098.723, 1978. GIRDLER Patent, D.T.2.406.280, 1977. Ph. Courty, Ph. Varin, J.F. Le Page, Unpublished results, 1980. GMELIN, 22(4), 1937, p.841 (potassium). E.M. Thornsteinson, T.P. Wilson, F.G. Young, P.H. Kasai, J. Cat., 52, 1978, pp.116-132.

ACKNOWLEDGEMENTS The authors wish to thank all these who contributed to the realization of this paper as well as Dr. E. Freund and Dr. G. Martino for helpful discussions.

518 DISCUSSION A. BOSSI: In your lecture you presented much structural characterization data (mainly by XRD) of Some mixed oxide systems. It seems that you neglected to take into account the surface characterization data (obtained by AES, XPS, etc .. ) which appear more suitable for correlations with the catalytic activity. Have you any comments on this? P. COURTY: The correlations we have presented deal mostly with either preparation or stabilization procedures and XRD or STEM and XRD microdiffraction characterization. For mixed oxides based catalysts, some doubts exist on the results obtained with surface characterization (AES, xPS, etc .• ) as high vacuum conditions are required to get these results; therefore, partial reduction or modifications of the surface composition cannot be neglected. On the other hand, XRD under controlled atmosphere gives an inclusive but accurate answer. L. RIEKERT: Can we differentiate between those variables of the preparation procedure which are important with respect to the steady state, in contact with reactants, and those which are not? P. COURTY: No general rules allow us to select between the numerous parameters of the preparation procedure as well as between the numerous ones of the procedure which permit to reach the steady-state of the catalyst in the reaction medium. The same catalyst, for example, can be used for very different reactions which can be "homogeneity demanding or homogeneity non-demanding" ones. In iron molybdate based catalysts, homogeneity is very important when used in methanol oxidation to formaldehyde. Homogeneity is much less important for uses in reducing conditions (NOx reduction to N2 or NH3 decomposition). ZHAO JIUSHENG pH of CU, Zn, cipitation pH control it in

: When you make the Cu-Zn-AI catalyst precursor, the precipitation Al with CO~are not the same. How do you determine the prein order to obtain the homogeneous composition and how do you the coprecipitation reactor ?

P. COURTY: Your question deals mostly with the know-how for making such catalysts. Therefore, one can say that coprecipitation pH, (pH)M is a "good compromise" between the extreme values of individual precipitation pH. (pH)M allOWS the best homogeneity and results in minimizing metal losses in the mother liquor. (pH)M can be controlled through a precise adjusting (or monitoring) of the flow rate ratios during the coprecipitation reaction. F. TRIFIRO: What do you think is the importance to start from the hydrotalcite structure? Does it stabilize CuO or metallic Cu? Does it determine the formation of CuO with small crystallite size ? P. COURTY According to your own results (1) binary copper-zinc based precursors and ternary copper-zinc aluminum ones produce CuO crystals of equivalent sizes (about 7.5 nm) after thermal treatment. The comparison between these methanol synthesis based catalyst precursors has to be done in terms of stability. The stability of the performances of Cu-Zn-AI based catalysts is much higher than that of Cu-Zn based catalysts. Alumina is supposed to reduce the rate of sintering of CuI-Zn I I based active species. Therefore, the use of the ternary precursor allows the best dispersion of the Cu-Zn active sites inside the alumina matrix and reduces the sintering rate during the operation. (1) O. Ruggeri, F. Trifiro and A. Vaccari, J. of Solid State Chern. 42, 120 (1982). J. SCHEVE: I cannot completely agree with your statement, that there is no general rule for controlling the sintering process. According to our experience, the old rule established by Huttig operates well i.e. using the ratio between heating temperature and the melting point of the substances in question. You can

519 gain a lot of insight in what would happen during firing your catalyst. you comment on this ?

Would

P. COURTY: I have only mentioned that "three are no general rules which allow foresight into and control on the effects of thermal treatments". Taking into account the ratio between heating temperature and melting point allows a partial foresight of the level of sintering; but, on the one hand, sintering level depends also on many other parameters (size and shape of elementary particles, crystallinity degree, preferential orientation, effect of impurities on crystal growth .•. ) and, on the other hand, at least two other reactions (thermal decomposition of the precursor and formation of the mixed oxide) occur simultaneously. The effect of these two latter ones seems more difficult to predict.

J. GEUS: Your pH curve for AI3+, cu 2+, Zn 2+ indicates three plateaus before the increase to the Zn 2+ precipitation level proceeds. We have evidence that copper hydroxy-nitrate is precipitating already at pH level below 4 and that the hydroxy-nitrate becomes unstable at pH level of about 8 when no carbonate is present. With carbonate ions present the hydroxy-nitrate becomes unstable already at a pH level of about 5. I therefore should like to suggest that the second plateau in your curve is due to precipitation of the hydroxy-nitrate and the third plateau is either due to a partial reaction of the hydro~-nitrate to the hydroxy-carbonate or to a combined precipitation of Zn 2+ and Cu + evolving from the hydroxy-nitrate. P. COURTY: We believe that the three plateaus correspond to the successive precipitation of alumina, copper and zinc based hydroxy-carbonates. Therefore, we cannot exclude that some copper is simultaneously precipitated with alumina during the first precipitation step. This could explain the very high stability of such "alumina doped" copper hydroxy-nitrates which resist either acidic or alkaline washing, without any decomposition to hydroxy-carbonates. These properties are very different from those you have mentioned in your comment for pure copper hydroxy-nitrate. B. GRIFFE DE MARTINEZ: We have been working on copper oxide-chromium oxide catalysts in the oxidation of CO to CO2, We have prepared copper oxide (CuO) , copper oxide-chromium oxide (CuO-Cr203) and chromium oxide alumina supported catalysts in different weight percents. We have found that the last ones are not active, whereas the former ones (CUO and Cuo-cr203/A1203 are very active at a reaction temperature of 200°C, especially those in the range of 10-20 wt %. Unfortunately, the CUO/A1 20 3 catalysts become almost deactivated after they are submitted to a water vapour treatment at 400"c. On the contrary, the CuO-cr203/ A1203 catalysts do not lose much activity. It must be said that the CUO/A1203 catalysts are reactivated after being submitted to the reaction conditions. We are at the moment carrying on research to try to give a scientific explanation for these differences. Would you please give a comment on this ? P. COURTY : The interactions between copper and chromium oxides (CuCr204 formation) are much stronger than those between CuO and A1203' Thus a stabilization of Cu-Cr species (bulk or alumina supported ones) can be invoked to explain these discrepancies.