ヘキサアルミネート系高温熱焼触媒材料の開発

九州大学学術情報リポジトリ

Kyushu University Institutional Repository

ヘキサアルミネート系高温熱焼触媒材料の開発

町田, 正人

https://doi.org/10.11501/3088193

出版情報：Kyushu University, 1991, 博士（工学）, 論文博士 バージョン：

権利関係：

Chapter 3

References

1 D.C.Bradly, R.C.Mchrotra, and D.P.Gaur,"Metal Alkoxldes," Academic Press, New York (1978), p.199.

2 B.E.Yo1das, Am. Ceram. Soc. Bul l.,54, 289 (1975).

3 B.E.Y_oldas, J. Am. Ceram. Soc., 65, 387 (1982).

4 K.S.Mazdlyasni, Ceram. Int'l, �. 42 (1982).

5 D.W.Johnson, Jr., Am. Ceram. Soc. Bull. ,64, 1597 (1985).

6 Y.Ozaki, Kogyo-zairyo, 29, 85 (1981); 29, 101 (1981).

7 J.B.Blum and S.R.Gurkovich, J. Mater. Scl.,20, 4479 (1985).

8 S.Sakka, Science, 79, 188 (1985).

9 D.B.Williams, "Practical Analytical Electron Microscopy in Materials Science," Verlag Chemie Int'l, 1984.

10 Y.Bando, Seramikkusu, 21, 54 (1987).

11 K.S.Mazdiyasni, R.T.Dolloff, and J.S.Smi th, J. Am. Ceram. Soc., 52, 523 (1969).

12 M.Machida, K.Eguchi, and II.Arai, Chern. Lett., 1986, 1993.

13 M. Mach ida, K. Eguchi, and II. Arai, J. Catal. , 103, 385 ( 1987).

14 M.Machida, K.Eguchi, and JI.Arai, Bull. Chern. Soc. Jpn., 61, 3659 (1989).

15 E.M.Levin, C.R. Robbins, and ll.F. McMurdJc; Fig.206 in Phase Diagrams for Ceramists, Amer. Ceram. Soc., 1964.

16 M.Machida, K.Eguchl, and II.Arai, J. Am Ccram. Soc.,71, 1142 (1988).

17 R.C. Ropp and fl. CarroJl , J. Am. Ceram. Soc., 63, 416 (1980).

18 II.Yama_shi_{t a,} A. Kato, N.Watanabc, and S. Matsuda, Nippon Kagaku Kalshi, 1986, 1169 (1986).

19 J. Y. Chene-Ching and L. C. Klein, J. Am. Cermn. Soc. , 71, 83 ( 1988); 71, 86 (1988).

20 R.G.Gosslnk, Mater. Res. Bull., 10, 35 (1975).

CRYSTALLOGRAPHIC ANALYSIS ON THERMAL STABILIT Y OF IIEXMLUMINATE

Chapter 4

CRYSTALLOGRAPHIC ANALYSIS ON THERMAL STABILITY OF HEXAALUMINATE

4.1 Introduction

The relation between surface area and crystalline phase of aluminates was briefly discussed in Chapter 2. Since the large

surface area retention is the common feature of hexaaluminate compounds, their thermal stability appears to be closely related to the crystal structure. Owing to this nature, hexaaluminate is believed to be a suitable material for the study of the struc­

ture-powder property relations. In this Chapter, several attempts were directed towards the elucidation of crystallographic charac­

terization on the thermal stability of hexaaluminat compounds.

Electron microscopy was employed to study the morphology and the orientation of hexaaluminate fine particles. The grajn growth of hexaaluminate powders was followed as the change of surface area, crystal size distribution, and surface microstructure during an isothermal heat treatment at 1300 °C. This also gives an important practical information on the heat resistance of hexaaluminate powders in a long range. On the other hand, the

general f ormulation of the mechanism of sintering and grain growth requires the information on solid state diffusion. Sec­

ondary ion mass spectroscope (SIMS) analysis on oxygen self­

diffusion in the different crystal direction of the single crys­

tal was performed in order to know the relation between the crystal structure and the mass transfer route of hexaaluminate.

From these results, the thermal stability of hexaaluminate fine

Chapter 4

particles was discussed from the crystallographic aspect.

4.2 Experimental

4.2.1 Heat Treatment of Powder Samples

The powders of hexaaluminate and its manganese-substituted sample, sr0.8Lao.2MnA111o19_a, were prepared by hydrolysis of metal alkoxides as described in Chapters 3 and 5. The powder

sample pressed into a disk was submitted to isothermal heating at 1300 � for 9800 h in air. After heat treatment of certain peri­

od, following characterization was performed. Specific surface area was measured by the BET method using nitrogen adsorption.

For the observation of surface microstructure, a high resolution SEM (Hitachi S800) was used. The crystal orientation and crystal size distribution of primary particles were obtained by TEM (JEOL JEM4000EX and JEM2000FX, H V_EM laboratory, Kyushu Uni versity) observation wlth selected area electron diffraction.

4.2.2 Single Crystal Growth

The material used for 18o diffusion-annealing and subsequent SIMS analysis was single-crystal barium hexaaluminate. For the single crystal growth of barium hexaalumlnate, a sintered rod was prepared from cool isostatic pressed powder mixtures of Baco3;

r-Al203 by heating at 1₄5₀ �- The chemical composition of the starting material was adjusted to Ba0.75Al11o17.25, as reported by Iyi et al.14-16 Single crystal of barium hexaaluminate was grown by the floating zone appar atus (Nichiden Kika i C o .) equipped with a xenon arc lamp. Crystal growth was operated at a constant rate (3 mm/h) in air. The transparent and colorless

crystals as grown were a rod-like shape with ca. 5 mm in diame­

ter and 20-30 mm long. Although they contained imperfection due

CRYSTJ\LLOGI�APIIIC ANALYS IS ON THERMAL STABILITY OF IIEXAALUMINATE

to microbubbles and cracks, large (ca.3 X 3 X 2 mm3) clear parts with good crystalline quality were obtained. Crystal growth axis was the [hkO] direction which corresponds to the cleavage plane of the single crystal. Crystal structure was confirmed to be a single hexaaluminate phase with no impurities from powder X-ray diffraction of the crashed crystals. Their chemical composition

was same with that of starting material as revealed by energy­

dispersive X-ray analysis.

4.2. 3 Diffusion Annealing and SIMS Analysis

Clear parts of th_e single crystal were cut in parallel and normal to the (001) plane, which was identified by the back reflection Laue method. Crystals thus obt ained were carefully

polished with a diamond paste (0.5 �m) and were preheated at 1200 �for 5 h in air to reduce the surface stress. For diffu­

sion-annealing, the sample was ^p ace_{d inside a} Pt hold r in _a quartz tube. After introducing a 18o-cnriched _gas into the tube, the sample was heated at 1200, 1400, and 1500 � with a xenon arc lamp. An isotopic exchange wlth a quartz tube was negligible in this condition.

The 16o;18 o rat i o in the diffusion-anneal d sample was analyzed by means of SIMS (Cameca,IMS-4F, National Institute for Research of Inorganic Materials, Tsukuba). T_he sample was coated with 40 nm of Pt metal to keep the sample surface at constant potential during the measurement. Also, an electron beam was irradiated during measurement to suppress the charge build-up on the sample. The SIMS analysis was operated under the following

typical conditions : primary ions cs+, primary ion energy 10 keV, Pri

m

ary ion current 5-20 nA. Secondary ion detection was limited to a central quarter area of 100 �m square raster-scanning of

Chapter 4

the primary ion beam. Intensities of negative secondary ions of mass/charge ratio at 16 and 18 were recorded as a function of time. After each analysis, the depth of sputtered crater was measured with profilometer (Rank Precision Industry, Talystep) to obtain depth profiles of the concentration ratio (18o;16o).

4.3 Crystal Structure _of Ilexaaluminate Compounds

The purpose of this Chapter is to clarify a crystallo­

graphic origin of the large surface area retention of hexaalumi- nate powders. For the latter discussion, crystal structures of hexaaluminates, i.e., ,a-alumina and magnetoplumbi te, are de­

scribed in this section. The previous studies provide substan­

tial knowledge on the crystal structure of hexaaluminates.3-6 Basically, ,a-alumina and magnetoplumbite consist of s pinel blocks and mirror planes, which are stacked alternatively along c axis to form a pseudo-layer structure as shown in Fig. 4.1.

A spinel block is composed of Al3+ and o2- lons, having the same rigid close-packing of structure as spinel oxides. Four planes of oxygen ions in a cubic close-packed sequence comprise a slab within which aluminlum j ons occupy octahedral a_nd tetrahedral positions. Each spinel blocks are weakly held t6gether by a

rather open layer (mirror plane), in which various large cations such as alkali, alkaline earth or rare earth metals can be _accom- modated. The major difference between ,B-aJumtna and magneto- plumbite lies _in content and arrangement of the ions within the

mirror plane layer. Thus, cheml caJ formulae of /3 -alumina and magnetoplumbite are ideally expressed as MA111o17 and MAl12019 (M

large cations), respectively.

Because of insertion of the loosely bound monolayer, hexaa- luminate compounds will show quite different physicochemical

Mi rror plane

Spinel block

CRYSTALLOGRAPHIC ANALYSIS ON THERMAL STABILITY OF IIEXAALUMINATE

® M (large cation)

•

AI

0 0

Fig. 4.1. Crystal structure of hexaaluminate compounds (,B-alumina and magnetoplumbite (MP))

Chapter 4 CRYSTALLOGRAPHIC ANALYSIS ON THERMAL STABILITY OF HEXA/\LUMINATE

properties from those of spinels, which can be seen in the two­

dimensional cationic conduction4, thermal expansion, single crystal growth7 and so on. Jt also expe cted that the layer plane appears in a crystal shape because iLs loosely bound struc­

ture leads to a lowest surface energy. The following sect ions revealed that the structural anisotropy plays a key role in grain

growth and sintering behavior of hexaaluminate powders.

4.4 Morphology and Orientation of Hexaaluminate Particles

Crystal morphology of h exaaluminate fine p articles was investigated by TEM observation. Figure 4.2a shows the alkoxide- derived BaAl12019 crystallized as thin planar particles with thickness of about 20 nm which is ranging from one fifth to one tenth of the diameter. Selected area diffraction patterns were taken when an electron beam is perpendicular or parallel t o the facet (FJgure 4.2b). An ·ncidcnt beam normal to the facet showed th hexagonal diffraction patterns due to the [001] zone axis, whereas an incident beam parallel to the hexagonal facet showed the dlffraction pattern due to the [110] zone axis. The d-spac-

ings from electron diffraction in these directions agreed with theoretical value within an experimental error. Thus, the facet

is in parallel to the (001) plane of hexaaluminate structure.

Such a anisotropic crystal shape was also observed for other hexaalum1nates and their derivatives, such as Mn-substituted hexaalurninates.

Figure 4.3 shows Lhe structural images of a typical hexaalu­

minate sample, Sro.sLao.zMnAl11o19_a • in which the Al site was partially substituted by Mn as mentioned in Chapter 5. The inci­

dent beam normal to the (110) plane showed the evidence of an array of spinel layers wi Lh ca. l nm thick along the ^c axis. On

a

----------

� �

^{----------____}

^{2 200nm}

!001] 1

b

[001]

�L---20

[11 OJ

Fig. 4.2. TF.M micrographs (a) and selected area diffraction (b) of alkoxide-derived DaAl12019·

Chapter 4

the contrary, the image of flat basal plane symmetrically orient­

ed with [001] parallel _{to the} incjdent beam is characterized by ca. 0.6nm spaced white dots ir1 a hexagonal ar_ra_y. From these

images, it can be clearly observed that the basaJ surface of facets is just in _parallel to mJrror planes.

From this anisot ropic shape of crys t a l li tes, t he g rain growth along the c axis is _expected to _{be s}tr_on^gly s u ppress ed in hexaalumi nate structures. This is likely a reason for the large s u rface area of this material, because the p_la_na_r morpl1ology with a larger aspect ratio (thjckness/diameter ratio) res u l ts in a larger specific su rface area. The planar morphology is the common feature of the hexaalu minate famlly and other isomorphous com- pounds, i.e., hexagalete and hexaferrite.8-10 For instance, the

mixture of large planar particles and small particles so called

"duplex _stru^cture"l l is ^pro_duce_d in a sin tered polycrys talline Na·/3-aJumlna. In this rnaLerlal, sLackjng of layers along the c axls proceeds at a slower _rate _as compared to the spinel block format ion. For the elucjdatlon of the anisotropic crystal shape

of he_xaalumin_ates

,

the information on _their grain growth process was examined in the next section.

4.5 _{Crys tal} Growth of llexaalu rninate d_{u rin}g _{Hea t} Treatment

An isothermal heat treatment of hexaaluminate powders was carried out _at 1300 � for 9800 h. The surface area of Sro.sLao. z­

MnAlllol9-a is plotted as a functLon of he_{atin g pe}r_iod

(

^Fig.

4. 4). Although the _he_at treatment gradually reduc_ed the surface

area, 4200 h of heating was necessary to reduce to the half (ca.

10 m2/g) of the ini tial s u rface area. This is _enough to demon­

strate the substantial potential Lty of thermal stability in a long period of high-temperature opera tion. SEM observation showed

- 64 -

CRYSTALLOGRAPHIC ANALYSIS ON THERMAL STABILITY OF IIEXAALUMINATE

-

1nm

� 1ncident beam

, ,.. .., -

.• y

I ^,.

i J.

�

�

�

,.._ ^.....

:'II .. �.,�.,.

v .. � II ,,�

11

'I .. '.} ^... ,. -ti � ..

1

incident

t

^beam

;;.:

�

'-.;_

:- � :tl

�� ^£r

t'l: :!lotlt:Jf! .

.::""

� �·

� :i ^il �� ,;:ill

rJ ^P,. X:�

"' ,) lr"'. ^:t

Jl

.il

1 ]'').1

'L'

•

m:'

I!;_ �} ... o.iC. ...

.... :�:���1nm

Fig. 4.3. H igh_-resol_utio_n transmission electron mjcrographs of alkoxide-derived Sro. sLao. zMnAlllol9-a·

20

I 16

01 N

E

Cl) 12

<1J 0

L. Cl)

<1J 8

u nj '+-L.

__o

rJ> ::J 4

6000 8000 10000 Heating period I h

Fig. 4.4. ^Change in ^surface area of Sro.sLao. zMnAlllol9-a during heat treatment at 1300 �.

- 65 -

Chapter 4

that the particle size of the hexaaluminate increased with the heating period as was evident from the appearance of large hexag­

onal facets, whereas a local agglomeration was hardly observed even after heating for 9800 h. Since the sample showed no crystallographic change during the heating, the loss in surface

a rea apparentl y resulted from the grain growth of the p lanar particles.

Changes in crystal size distribution and crystal morphology of hexaalumlnate in the heat treatme nt were studied with TEM.

Crystal size was ^measured for one tmndrcd crystals when an elec-

tron beam is normal or parallel to the (001) plane. These values correspond to the crystal diameter and the crystal thickness of planar facets, respectively. figure 4.5 shows the dependence of

the average crystal size of Sro.sLao.2MnAl11019-a on a heating period. The crystal diameter increased more rapidly than the crystal thickness in the initial heating process. Ifowever, the grain growth along both directions was almost saturated above 4000 h of heating. The planar morpholog^y of hexaaluminate obvi-

ously results from thls anisoLropic grain growth ln the initial slntering p�ocess.

In order to evaluate the change in crystal morphology, an average aspect ratio (thickness/diameter ratio) of the facets was plotted in Fig. 4.6. A ^large number of cr^ystal^s in a fresh sample showed the smallest aspect ratio of ca. 0.1. However, the iso- thermal heat treatment sprea d the distribution of the aspect

ratio and shifted the peak of distrib ution to a larger side.

Finally, the average aspect ratio reached to ca.0.25 after 9800 h of heating, which is about twice larger than that of the initial value. This relative increase of the thickness leads to decrease in surface area because the specific surface area of facets

- 66 -

CRYSTi\LLOGRi\PI!IC ANALYSIS ON THERMAL STABILITY OF HEXMLUMJNATE

500 .---�

400

E

c

-

(lJ 300

N lf)

�

200

lf) >..

l-

u

Diameter( 1c)

Thickness (//c)

8000 10000

Heating period

I

h

Fig. 4.5. Change in average particle size of Sro.sLao.�Mn­

Al11019^-a during heat treatment at 1300 �-

0

-+--' u ClJ 0. V1

<I:

0.3

2000 4000 6000 8000 10000 Heating period I h

Fig. 4.6. Change in aspect ratio of Sro^.sL^ao^.2M^nAl1¹0¹⁹-a during heat treatment at 1300 �-

- 57 -

Chapter 4

depends on thickness rather than on diameter. In the case of hexaaluminate planar particles, however, an increase in the thickness is significantly suppressed by the anisotropic grain growth along the (001) plane. Thjs is why the large surface area can be retained for a long period.

The surface microstructure of the powder sample after

9800

_h

of heating is well reflected by an anisotropic grain growth as was observed with a high-resolution SEM (Fig. 4.7). The basal plane (parallel to

(001)

plane) of the facet is flat and smooth except for several parallel growth steps. The thickness of these unit steps is less than 10 nm which corresponds to the stacking of a few spinel blocks. On the contrary, a curved side plane of

the facet contains a lot of streaks on the surface. This surface roughness indubitably indicates that the grain growth proceeded significantly on t he side plane of facets rather than on the basal plan as speculated below. Hcxaaluminate structure con- sists of the stacking of oxygen close-packed spinel blocks as shown in Fig. 4.1. Continuous spinel formation along the (001) plane is expected to easily proceed because of the absence of possible barriers to the lattice construction. Stacking of each

blocks, however, will proceed more slowly due to discontinuous 'nsertion of mirror plane layers. Therefore, on the basal plane, the grain growth appears to proceed through the formation of single step and a subsequent two-dimensional step growth on a smooth surface. Whlle such a layer growth on the basal plane

brings about the increase ln thickness of a hexagonal facet, its growth rate will be lower than that parallel to the facet. Conse­

quently, the anisotropic grain growth appears to closely related to the difference of the growth mechanism between basal and side planes of the facet.

- 68 -

001

a

CRYSTALLOGRAPHIC ANALYSIS ON THERMAL STABILITY OF I!EXAALUMINATE

- . � ----....,

b

Fig. 4.7. _Surface microstructures of hexagonal facets of

sr0_8-

Lao.2MnA111019-a after heat treatment at 1300 � _for

9800

h. a) The basal plane with parallel steps, b) The side plane with streaks. High resolution SEM with a field emission gun was used.

4.6 Relation between Crystal Growth and Oxygen Diffus on 4.6.1 SIMS Analysis on Anisotropic Oxygen Diffusion

Since it becan1e apparent that the anisotropic crystal growth depends on the anj sotropj c crystal structure, whole c rystal growth problem is reduced to a mass transfer route to the growth surface (basal and side plane of the facet). Generally, the sin­

tering rate is usually controlled by the mass transfer of con-

- 69 -

Chapter 4

stitute ions with a lowest diffusion rate. Since a rate-deter- mining step in sintering of the related spinel compound (MgAlz04) is oxygen diffusion ln the lattice, the information on the oxygen diffusion in hexaalurnlnates needs to evaluate the mass transfer mechanism. To determine the oxygen self-diffusion coeffic i ent i n parallel (//c) _{and normal} ( J c) to the (001) plane of the single crystal BaAl12019· the depth profile of oxygen concentra­

tion was measured by SIMs12,13 after diffusion annealing in 18o­

enrlched gas. Detailed conditions of the diffus i on annealing and SIMS analysis are summarized in T^ab le 4.1.

Depth profiles of 18o concentration in the two crystallo- graph i c directions of the barium hcxaaluminate are shown in Fig.

4.8. Since the profiJometer measurement of sputtered craters showed a fairy flat bottom with a well-defined shape, these figures reflect ^{real c}oncentrati^on profiles as a funct i^on of the dlffuslon distance. From Fjg, 4.8, i t ^is apparent that 18o pene-

trat i on i n the �c _direction was more significant than the //c direction regardless of the diffusion annealing temperature ( 1 2 0 0 -1 5 0 0 OC ) • T h e ex t en t of p en e t r a t i ^on i n b o t h d i r e c t i on s

increased w i th an iⁿcrease i n the anneal i ng temperature. After every diffusion annealing, the surface 18o concentrat i on deter­

mined by linear extrapolation approximately corresponds to a gas­

phase concentration. Th1s means that J8o transfer from the gas phase to the solid phase was controlled only by solid state dlffus i on in the bulk.

When the surface concentration was in equilibrium as in such a case, the effect of surface exchange reaction can be neglected.

Thus, the diffusion coefficient can be calculated from the depth profiles of 18o;16o by equation of Fick's second low of i nfin i te solid as follows17,18.

- 70 -

CRYSTALLOGRAPHIC ANALYSIS ON THERMAL STAI3ILITY OF HEXAALUMINATE

Table ^4.1 Diffusion-annealing and SIMS analysis conditions

Analysis Diffusion anneal Oz pressure Sputtering 18o2;(18o2+16o2) Diffusivity direction Temp.

I

_oc time/s /1o3 Pa depth/ _tL m gas phase /crrf sec-1

1/c

�c //c

�c //c

�c

1200 7200 1200 7200 1400 7200 1400 7200 1500 7200 1500 7200

1.0

0

([) 0.6

r-+

a::P

r--- 0.4

-

coo r-

0.2

2.0 2.0 2.0

2.0 2.0

2.0

c

0.1

2.06 2.04 4.02

6.76 4.46

3.56

0.2 Depth I

1-Jm

0.97 0.97 0.96

0.96 0.95 0.95

I!C direction

0.3 0.4

1.0 r---,

([)0

+ 0.6-

�

0

-- 0.4

�0 0.2

0.1 0.2

Oepth/�m

..LC direction

0.3 0.4

9.09Xlo-15 5 .14XHf 14 2.59X1o-14 9. 35X1cr13 7.32Xlo-13 2. 92x1cr12

Fig. 4.8. Depth proflles of 18o concentration i n a BaAl1z019 single crystal. D i ffusion anneal i ng at a) 1200 °C, b ) 14 0 0 oc ^' c ) 15 0 0 oc .

- 71 -

Chapter 4

[C(x, t)-Co]/(Cg-Co)=1-erf(x/2(Dt)0.5) ( 1)

where x is the penetration depth, t the duration of the diffusion annealing, C(x,t) the concentration of 18o at x, Cg the gas-phase concentration of 18o, Co natu^{r a l} concentration of 18o, D the oxygen self-diffusion coeff icient and erf the Gaussian error

function. Diffusion coefficients thus obtained are plotted as a function of reciprocal diffusion temperature in Fig. 4.9. It should be noted that the observed diffusivity in the �c direc­

tion was ca.l0-40 times larger than in the //c direction. The diffusivity along the c axis is close to that in MgAlz04 with the spinel structure reported by Oishi and Andol9. Since the MgA12o4 spinel data cited here were determined by gas-phase analysis of 18o;l6o ratio during diffusion annealing, the diffusivity is considered as the average value in all the crystal axes.

4.6.2 Structural Aspect of Anisotropic Crystal Growth

From Fig. 4.9, oxygen diffusion coefficients in the � c direction of barium hexaalurninate obey the following equation.

D ( cm2 I _{sec)= 0. 00} J _lexp (-290 ( kJ /mol) /l\T) ( 2)

On the other hand, oxygen diffusion in the //c direction of barium hexa alurninate can be approximated to that in MgAlz04 reported by Olshj and Andol5.

D(cm2/sec)=0.89exp(-440(kJ/mol)/RT) ( 3)

The oxygen diffusion in the //c direction analogous to that in the spinel can be deduced from the crystal structure (Fig. 4.10),

- 72 -

CRYSTALLOGRAPHIC ANALYSIS ON Tl!ERMI\L STABILITY OF I!EXAI\LUMINATE

Tj"U1

('I

10

§

¹²

0 0 01

I 14 I-

1900

Temperature/ K

1700 1500

0

16---�1---�IL_ ______ _L ______ __J

5.0 5.5 6.0

1011/T /K-1

6.5 7.0

Fig. 4.9. Temperature dependence of oxygen diffusivity in BaAl12019 single crystal.

0 Bao.7sAlllo17.25 ( �c. this study) 0 Bao.7sAl11011.25 (�c. this study) --- MgAl204 (Ref. 15)

--- Mg diffusivity in MgAl204 (Ref.20)

in which cationic and anionic configurations of the spinel block projected on (140) are similar to those of MgAl204 projected on (321). In the diffusion in the //c direction of hexaaluminate, oxygen ions should run across the close-packing oxygen layers, as well as in _the spinel aluminate, and mirror plane layers

alternatively. However, the above result means that the diffusion rate along the //c direction was mainly determined by that in spinel blocks. In spinel blocks with the close-packing of o2- ions, oxygen diffusion likely requires higher activation energy than in the mirror plane.

- 73 -

Chapter 4

0000000000

0000000000

Qp�OC60C60�0

0 0 0 0 0

ODODODODOD

0000000000

1�0�0�0�0�0�

doobobobob

00000000000

oqoqoqoqoqo oqoqoqo�o�

0 0 0 0 0

DODODdOODO

rn rn rn rn oo

oc5oeooo&c5o

0000000000 _y

00C60C60C60C600

0 0 0 0 0

o0oCO

^o

COdodo

u 00000000000

:E opo�opopoo ��

Q} 0 0 0 0 0 °o ⁰ .�

eoeoooeoeoo

²

0000000000 � 0000000000

C60C60C60C60C6o opopopopopo

0 0 0 0 0

o0dodododo

@) ro ru @:) 00 oooooooooo

obobo�oboDo

0C6°C6°C6°C�PC6

0 0 0 0 0 o o o o o o o o o o

<ododododoo �oqoqoqoqoo

0 0 0 0 0 0 0 0 0 0

° �

^o0

d! �

^o0

0000000000 ouooououoo

o Mg o AI

0

⁰

Spinel projected on (321)

@Ba o AI

0

⁰

Beta alumina projected on (140)

fig. 4.10. Crystal structure of 8-alumina and spinel aluminate.

Comparison of the two equations clearly show� that some preferential diffusion routes with a lower activation energy lie in the _l_c direction of hexaalumlnate. This corresponds to the

mirror plane with an open structure, which lies on in _�c direc­

tion. The o2--o2- distance in the mirror plane (ca.0.56 nm) is quite larger than that in the spinel block (ca.0.26-0.28 nm). In addition, it is well-known that nonstoichiometry in this struc­

ture arises from the poss1bility accommodating variable quanti­

ties of oxygen i o n s and large cati ons in the m ir r o r plane.

Several researchers reported lhat there are some intermediate

- 74 -

CRYSTALLOGRAPHIC ANALYSIS ON TIIERMJ\L STM3ILI TY OF I!EXMLUMINATE

sites available for o2- ions to occupy.4 These structural con- siderations conclude that the oxygen ions preferentially diffuse

through the mirror planes between spinel blocks.

Anisotropic oxygen self-diffusion in hexaaJuminate compounds was not reported so far. The onJy study concerning this is on chemical coloration of singJe crystal Na-�-alumina reported by De Jonghe et al.16,17 They revealed that bJeaching (reoxidation) of the chemically colored (reduced) single crystal proceeded predominantly along (001) plane. This result implies anisotropic diffusion of oxygen or oxygen vacancy and supports our explana­

tion . It can be pointed out _that the anisotropic oxygen diffu­

sion is expected to be seen in other �-alumina as well as magne- toplumbite compounds.

It is tempting to speculate on the relation between the grain growth and the oxygen diffusion of hexaaluminate com­

pounds. Solid state oxygen diffuslon, which is rep orted to be the r a t e- de t ermining step durin g the initial sln tering of MgA12o4,18,19 is also believed to be a key role in the grain growth of hexaaluminate powders. If the oxygen ions as well as other ions would prefer to migrate through the mirror plane, it becomes the preferential transport route of these growth units from the bulk to the surface as shown ln the sch matic illustra- tion in Fig. 4.11. As discussed in the previous section, the grain growth mechanism on basal and side plane of hexaaluminate facets seems different, reflecting the pseudo-layer structure.

In such a situation, the anisotropic mass transfer wlll further

enhanced the preferential grain growth along (001) plane. These are the plausible explanation why the planar morphology with a large aspect ratio can be retained in the hexaaluminate com­

pounds.

- 75 -

Chapter 4

Spinel block

02-.

0: 10n

Fig. 4.11. Schematic illustration of crystal growth under oxygen diffusion limitation.

4.7 Conclusion

This Chapter revealed that anisotropic grain growth, which results in planar morphology with a large aspect ratio, is the reason for the large surface area retention of hexaaluminate powders. From electron microscopic observation, hexaaluminate crystallized as thin planar particles of which basal planes are exactly parallel to the

(001)

plane. Anisotropic grain growth was confirmed by a morphological change during heat treatment in a long period

(9800

h

)

and is also obviously seen in the differ-

- 76 -

CRYSTALLOGRAPHIC ANALYSIS ON THERMAL STABILITY OF IIEXAALUMINAT E

ence in surface microstructures. The anisotropic grain growth reflects the pseudo-layer structure, i.e., the rate of spinel block formation is higher than that of stacking of spinel blocks along ^c axis due to insertion of mirror plane layers.

SIMS analysis on oxygen self-diffusion in the single crystal hexaaluminate

higher in the

showed that the diffusion rate was obviously

�c

direction than in the

//c

direction. Moreover, the diffusivity in the

//c

direction is similar to that in MgAl204 spinel oxide. These results indicate that the loosely packed intermediate monolayer between closely-packed spinel blocks, which spreads in parallel to the

(001)

plane, is likely a preferential diffusion route of oxygen ions and thus other con­

stituents. The anisotropic grain growth of hexaaluminate appears to be enhanced by this preferentiaJ mass transfer process in a sintering process.

- 77 -

Chapter 4

References

1 N.Iyi, S.Takekawa, Y.Bando,and S.KJmura, J. Solid State Chern. ,47, 8(1983).

2 N. Iyi, Z. Inoue, S. Takekawa, and S. KiJhura, J. Solid State Chern. , 52, 66 ( 1984).

3 N.Iyi, S.Takekawa, and S.Kimura, J. Solid State Chern., 83, 8 (1989).

4 R.Collongues, J.Thery, and J.P.l3oilot, in "Solid Electrolytes," ed. by P.

Hagenmuller and W. Van Goal, Academic Press, Inc., New York, 1978. p.253.

5 J.M.P.J.Verstegen, J.L.Sommerdik, and J.G.Verriet, J.Lumin.,

�.

425(1975).

6 A.L.N.Stevels and A.D.M.Schrama-de Pauw, J. Electrochem. Soc., 123, 691 (1976).

7 F.Laville, D.Gourier, A.M.Lejus, and D. Vivien, J. Solid State Chern., 49, 180 ( 1983) .

8 Y.Sumlyoshi and M.Ushio, J. Am. Ceram. Soc.,73, 3015 (1990).

9 M.llervleu, D.Groult, B.Raveau, and G. Fuchs, J. Solid State Chern., 62, 261 ( 1986).

10 H.Ikawa, T.Ohashi, M.Ishimori, T.Tsuruml, K.Urabe, and S.Udagawa,in ''High Tech Ceramics," ed. by P.Vincenzini, p.2137, Elsevier, Amsterdam, 1987.

11 R.W.Powers and S.P.Mitoff in "Solid Electrolytes," ed. by P.Hagenmuller and W. Van Gool, Academic Press, Inc., New York, 1978. p.123.

12 M.Arita, M.Hosono, M.Kobayashi, and M.Someno, J. Am. Ceram. Soc., 62, 443 ( 1979} .

13 II.IIaneda and C.Monty, J. Am. Ceram. Soc., 72, 1153 (1989).

14 J.Crank, Mathematics of Diffusion. Clarendon Press, Oxford, England, (1956).

15 K.Ando and Y,Oishi, J. Chern. Phys. ,61, 625 (1974).

16 L.C.De Jonghe and A.Buechele, J. Mater Set., 17, 885 (1982).

17 L.C.De Jonghe, A.l3uechele, and M.Armand, Soljd State Ionics, 18 R.J.Bratton, J. Am. Ceram. Soc., 52, 417 (1969).

19 R.J.Bratton, J. Am. Ceram. Soc., 54, 141 (1971).

20 R.Linder and A.Akerstrom, Z. Physik. Chern., 18, 303 (1958}.

9&10, 165 (1983).

- 78 -

MATERIAL DESIGN OF COMBUSTION CATALYST BY STRUCTURAL MODIFICATION OF IIEXAALUMINATE

Chapter 5

MATERIAL DESIGN OF COMBUSTION CATALYST BY STRUCTURAL MODIFICATION OF HEXAALUMINATE

5 .1 Introduction

In the preceding Chapters, hexaaluminate compounds were revealed to possess an excellent thermal stabJlity so that they are useful for a high temperature catalyst support. The next

step of this study is to develop the combustion catalyst by using hexaaluminate. This subject Includes the introduction of cata­

lytically active components and the control of catalytic proper­

ties. How to introduce active components is a crucial problem as well as the selection of active components. Since nobJe metal catalysts should be excluded due to the high volatiJity above 1000 °c,1,2 some metal oxides with suffJcicnt th rmal stability ls preferred as an active component. Even in active oxide cata ysts like perovskite type oxides3-6, however, their supported cata­

lysts will suffer from thermal deactivation because of ease of sintering7 and/or solid state reactions with support materials as described in this Chapter. A new concept of cataJyst design is strongly required for overcoming the difficulties in conventional catalyst system. As revealed in Chapter 4, the large sur face area of hexaaluminates is originated from their crystal struc­

ture. This means that the design of thermally stable catalysts should be also based on the hcxaaluminate structure.

The material des ign proposed here is the structural modifi­

cation of hexaaluminate as schematJcally shown in Fig. 5.1. A wide variety of catalytically active cations, e.g., Cr, Mn, Fe,

- 79 -

Chapter 5

00

Mirror plane

Spinel block

M = Mg. Mn. Cr. Fe. Co. Ni. Cu.· Ga A. A" = Na. K. Ca. Sr. Ba. La. Y. Nd

Fig. 5.1. Structural modification of hexaaluminate by cation­

substitution.

- 80 -

MATERIAL DESIGN OF COMBUSTION CATALYST !3Y STRUCTURAL MODIFICATION OF IIEXAALUMINATE

co, Ni, and Cu, can situate the Al site of hexaaluminate.9-12 Moreover, various large cations, e.g., alkaline, alkaline earth, and/or rar e e a r t h e l e m e n t c a n o c c u p y t h e m i r r o r p l a n e site.8,13,14 These cation-substituted samples with a variety of constituents are quite suitable for controlling catalytic proper- ties, being one of interesting approach for the material design of combustion catalysts. Catalytic activity and related physico­

chemical properties of the cation-substituted sample were inves­

tigated from a view point of the reduction-oxidatjon behavior.

From these characterization, discussion is directed towards the elucidation of the relation between catalytic activity and crys­

tal chemistry of hexaaluminate.

5.2 Experimental

5.2.1 Preparation of Samples

A perovskite-type oxide, La0.6sr0.4Mll03, was support d on BaAl12 o l9• which was prepared from BaC03/7-Al203 calcined at 1450 �(Chapter 3), by conventional impregnation using the me al acetates. Series of cation-substituted as well as unsubstituted barium hexaaluminates, BaMAl11019-a (M=Al, Cr. Mn, Fe, Co, and Ni) were prepared by a modified alkoxide process as follows.

Barium isopropoxide (Ba(OC3H7)2) was obtained by reaction between Ba metal and 2-propanol in a nitrogen stream. Calculated amounts of Ba(OCsH7)2 and Al(OC3H7)3 were d1ssolvcd in 2-propanol and were kept in 80 � for 5 h. An aqueous solution of acetate or nitrate of transition metals (substituents) was added to the alcoholic solution of metal alkoxides. The precip itate thus formed with gelation was evaporated to dryness. Structural modification of Mn-substituted hexaaluminate was demonstrated on BaMnxA112_x0ls-a and A1_xA'xMnAl11o19_a (A = Ba, Sr, Ca, La, K:

- 81 -

Chapter 5

A'= Sr, La, K, Ca), which were prepared in a same manner. All catalyst was calcined at 1300 � in air prior to their use for the catalytic reaction.

The crystal structure of calcined samples was determined by X-ray diffraction (Rigaku Denki, 4011) with CuKa radiation. The specific surface areas of calcined samples were measured by the BET method using N2 adsorption.

5.2.2 Catalytic Reaction

Catalytic activities were measured in ^a conventional flow system as described in Chapter 2. The combustion activity is expressed as temperatures, T10% and Tgo%· at which the methane conversion to carbon dioxide is 10% and 90%, respectively.

5.2.3 Transmission Electron Microscopy (TEM)

A transmission electron microscope (JEOL, JEM-2000FX) was used for studying crystal morphology, crystal orientation, and crystal size distribution as mentioned in Chapter 4.

5.2.4 Temperature Programmed Desorption (TPD) of Oxygen

Temperature programmed desorption (TPD) of oxygen was meas­

ured in a flow system. Prior to the measurement, the sample (ca.

2 g) was treated in a oxygen stream (50 rnl/min)

at

800 OC for 1 h and subsequently cooled to room temperature. After evacuation, the sample was heated at a constant rnte of 10 deg/min in a

helium stream (50 ml/min). The desorbed oxygen in an effluent gas was detected with TCD cells as function of the desorption temperature.

- 82 -

MATERIAL DESIGN OF COMBUSTION CATALYST BY STRUCTURAL MODIFICATION OF HEXAALUMINATE

5.2.5 Thermogravimetry

(TG)

The oxidation states of Mn ions was determined by thermogra­

vimetry (Shimadzu DT-40) equipped to a flow system. Prior to measurement, the sample was heated at 1000 � in a stream of dry air to remove physisorbed water and then cooled to room tempera­

ture. After evacuation, the sample was heated again at a constant rate of 10 deg/min in a H2 stream (30 ml/min). The loss in the sample weight which corresponds to the degree of reduction was r e corded up to 11 0 0 oc . W i t h heat i n ^g up to 11 0 0 OC i

n

^a II 2 f 1 ow , Fe, Co, and Ni ions are reduced into the metallic states, but only Mn into the divalent state. Thus, reduction of cation­

substituted barium hexaaluminate is expressed as

BaMAl

1

1019 -a +

(1

.

5

- n -a

)

ll

2

^--+

M(h

+

I3aAlll<17.5

+

(1

.

5

- n-a

)

l

Q

O

(1)

where n = 0 for M = Fe, Co, and Ni and n ⁼ 1.0 for

Mn.

The ni­

tial oxidation state of

M

is

calculated

from th w ight loss nd the final oxidation states after reduction from the fol owing equation,

Oxidation state

2

n +

2L1W(714.12

⁺

Mw

⁺

16n)/16W(1-L1W/W) (2)

where L1W is the weight loss, W, the initial sample weight, and Mw, atomic weight of M. Oxidation state of

Mn

species in Sr1-x­

LaxMnA111019-a was also determined in a similar way.

5.3 Preparative Strategy for High Temperature Catalysts

Perovskite type oxide, Lao.6Sro.4Mn03, was impregnated onto BaAl12019· Surface area and activity for methane combustion over the supported catalyst are listed in Table 5.1. When the loading

- 83 -

Chapter 5

Table 5.1 Surface area and combustion activity of Lao.6Sro.4Mn03 supported on BaAl12019

Loading /wt%

0 10 20 30 100

Surface areaa

;m2g-1

6.0 4.8 4.2 3.4 0.3

695 610 555 405

755 740 790 710

a BaAl12019 was prepared by calcining BaC03/7-Al203 mix­

tures at 1450 °C. Supported and neat Lao.6Sro.4Mn03 was calcined at 1300 �.

b Temperatures at which conversion levels are 10% and 90%.

Reaction condition: CH4 1 vol%, air 99 vol%, S.V.=48000 h-1.

amount of the oxide increased, the specific surface area de-

creased more drastically as compared to that estimated from the change of specific gravity of supported caLalysts. The initiation temperature (Tlo%) of catalytic combustion was reduced with an increase of loading amount, whereas the temperature of the high

conversion level (T9o%) was significantly lncreas�d. Moreover, the catalytic activities are much lower than that of the unsup­

ported Lao.sSro.4Mn03 catalyst. These results mean that loading of oxide on support and subsequent calcination induced the deac- tivation due to a partial _solid slate reaction _{between perov­}

skite and hexaaluminate. Thus, the large surface area and the high catalytic activity can not be achieved simultaneously by the direct impregnation method.15

To solve such problems in conventional impregnated cata­

lysts. Misono et a1.l6 reported ^a new preparation method for

- 81 -

MATERIAL DESIGN OF COMBUSTION CATALYST BY STRUCTURAL MODIFICATION OF IIEXAALUMINATE

supported perovskite catalysts, in whjch the solid state reaction is prevented by precoatjng the support with T.a203. For the excel­

lent thermal stability, however, catalysts for hJgh temperature processes is preferred to be consist of a single equilibrjum phase to give the sufficient thermal stability. r-rom this view- point, introducing active component through partial substitution was noted, because some transition metals can be situate at Al3+

site in the hexaaluminate lattice.S-11

Cation-substituted hexaaluminates were prepared by employing

the sol-gel technique. After calcination at 1300 �. cation-sub- stituted samples (BaMAl11019-al were composed of hexaaluminate phase, but the unsubstituted single oxides were hardly detected by X-ray diffraction.15 Since transition metal ions occupy Al sites in the hexaaluminate structure, diffraction lines shifted to lower angles as compared to those of BaAl12019· Surface areas of BaMAl11019-a• which were lowered by the substitution, wer still large, ranging between 10 and 15 m2/g (Table 5.2). The high heat resistance of barium hexaaluminate is retained in the cation-substituted samples because their hexaalumlnate structure is essential for the large surface area.

Figure 5.2a shows TEM photographs of BaMnAlJ1019-a after calcination at 1300 °C. The sample consisted of highly dispersed planar crystallites 100-200 nm in diameter and ca. 20 nm thick.

The crystal orientation was determined by selected area diffrac­

tion (SAD) as shown by ^x in Fig. 5.2b .. This planar facets have orientations parallel to the (001) plane of the hexaaluminate stru c t ure. The crystal habit is the same as unsubstituted BaAl12019• in which the crystal growth is strongly suppressed along the ^c axis as was mentioned in Chapter 4. Mn-substitution in the hexaaluminate lattice was evidenced by the EDS spectrum

- 85 -

Chapter 5

Table 5.2 Surface area and combustion activity of BaMAl11019-a

M Surface areaa T1o%b T9o%b

lm2g-1

I oc I oc

Al 15.3 710 730

Cr 15.7 700 770

Mn 13.7 540 740

Fe 1₁._{1 560 780}

Co 15.2 690 720

Ni 11^.1 _{710 770}

Thermal reactjonc 810 860

a After calcination at 1300

oc.

b Temperatures at which conversion levels are 10% and 90%.

c An empty reactor packed with alumina beads.

Reaction condition: CH4 1 vol%, air 99 vol%, S.V.=48000 h-1.

taken from one of the crystals (Fig. 5.2c). The chemical composi­

tion calculated from integration or the spectrum agreed with that of the starting material. Hjgh dispersion of active species in

the hexaalumJnate matrix is expected to elimjnate the catalyst deactivation due to agglomeration or evaporatjon of active spe- cies.

5.4 Catalytic Properties of Catio11-substituted Hexaaluminate 5.4.1 Catalytic Activities for Methane Combustion

Combustion activities of the cation-substituted samples are summarized as T1o% and T9o% in Table 5.2.17 When an empty reac­

tor packed w ith alumina beads was submitted to the reaction, combustion appears to be initiated by the radical formation at the surface of alumina beads and to progress through chain reac­

tions in the gas phase, as can be characterized by the high initiation temperature (Tlo%) and the steep rise in CH4 conver-

- 86 -

a

MATERIAL DESIGN OF cmABUSTION CATALYST IW STRUCTURAL MODIFICATION OF I!EXAALUMINATE

c

/.1

50nm

b

r,,

IU.J. 11

l'l'l"rl 1'1''1 r ., 1111 r1 '11"1'1

Fig. 5.2. a ) TEM micrograph, b ) SAD pattern, and c ) EDS _spectrum of BaMnAl11019-a after calcination at 1300 �. _{A Cu} peak is from a microscope grid.

sion. Since such a combustion behavior was observed over urJsubstj- tuted and Co-, Ni-, or Cr-substJtuted samples, the catalytic activities of these transition metal ions appears to be quite small in the hax_aa l_umi n_ate lattice. _However, the in1tiation _tern-

perature was significantly lowered by employing the catalyst with the large surface area whjch is effective in promot1ng the radi­

cal reaction. Tl1e Fe- and Mn-substi tuted samples catalytically initiated the surface reaction at lower temperatures (TlO% ⁼ 540- 560

�).

The lowest T1o% value (540 �) was attained by BaMn­

Al11019-a • which also showed the relatively high activity at a

Chapter 5

Table 5.3 Relative catalytic activity of BaMnAl11019-a and LaCo03 at 500 aC

Reaction rate per unit surface area

/mol m-2 h-1

LaCo03 0.351

per unit mass /mol g^-1 h-1

0.1 6 9 0.105

Reaction condition: CH4 1 vol%, S.V.=48000 h-1.

air 99 vol%,

high conversion level (T9o%). Table 5.2 indicates that the low Tgo% value is not always obtained for the samples with h igh catalytic activity. This is because the overall reaction over active catalysts (Mn, Fe) was J imited by lhe mass transfer at the high conversion level. On the other hand, combustion completes immediately once the homogeneous reaction initiates over the sample with a low cataJytic activity.

The catalytic activity of Mn-substltuted hexaaluminate is compared to that of perovskite-type oxide, LaCo03, which is known as one of active combustion catalysts. Table 5 .3 shows the reaction rates of methane oxj da tj on at 500 oc whi ell are expressed per unlt surface area and p r catalyst mass. The activity ^per

unit surface area is notably low for BaMnAl11019-a · whereas the activities of the two catalysts are comparable per unit mass.

This means that high catalytic activity of BaMnAl11019-a re- sults from the large surface area. High catalytic activity is

essential to the ignition at the lowest possible temperature. At high temperatures, however, t he surface reaction proceeds so

MATERIAL DESIGN OF COMBUSTION CATALYST BY STRUCTURAL MODIFICATION OF IIEXAALUMINATE

rapidly that the overall reactjon is determined by the ^mass transfer of reactants to the cataJyst surface from the gas phase.1,Z Maintaining the large surface area of combustion

catalysts is important in promoting the reaction under the mass transfer limitation.

These results conclude that the large surface area plays an important role in catalytic and homogeneous combustion. Both the catalytic activity and the large surface area are indispensable to the high temperature catalytic combustion, but no materials reported so far satisfy these requirements in their use at high temperatures. The Mn-substituted hexaaluminate is the first

promising material which attains the catalytic activity and heat resistance at combustion temperature above 1200 �.

5.4.2 Oxidation State of Substituent Cations

S ince the catalytic activity of metal oxides ge nerally depends on their reduction-oxidation behavior. the difference 1n the catalytic activity of cation-substitut d hexaaluminat s should be closely related to the oxidation state of substituents in the lattice. To determine the oxidation stale of M in BaM- Al12o19_a , TG measurement in a Hz flow was carried out. The average oxidation number of M calculated are listed in Table 5.4.

Although BaCrA111o19^_a is too stable Lo be reduced by Hz up to 1100 oc' the oxidation state of Cr is thought to be +3.0 as reported for LaMgAl11_xCrx01g by Vjvien et al.18 Cobalt and nickel are both in the divalent state, but Fe is in the trivalent state. The average oxidation number of Mn (+2.4) indicates ^the mixed valence state of Mn2+ and Mn3+ in the hexaaluminate lat- tice.

Determining oxidation states of transition elements, M, from

- 89 -

Chapter 5

Table 5.4

2

'+-

0

<lJ

-t..J -t..J cU U1

c 0 -t..J u rU 0 X

Average oxidation state of M in BaMAl11019-a M

Cr Mn Fe Co Ni

3.0

2.0

Average oxidation state

3.0 2.4 3.0 2.1 2.0

At

--;..

Cr

0 Mn

+-

0 -50 -100 -150 -200 6H t(M01.5)-6 Ht(MO )/ kJ·mol-1

Fig. 5.3. Relation between oxidation state of M in BaMAl11019-a and (�H

f

(M01.5)-�H

¥

(MO)).

- 90 -

MATERIAL DESIGN OF COMBUSTION Ci\TALYST BY STRUCTURAL MODIFIC/\TION OF IIEX/\ALUMINATE

in situ TG measurement was the first quantitative analysis of the present system, although those in the LaMAl11o19 single crystal have already been studied qualitat1vely.l1,19 Table 5.4 shows that all the transition elements were in di- or trivalent states.

The sequence of the oxidation states appears to be related to the bonding strength between metal and oxygen, which is roughly est imated as the standard h e a t of fo rmat i o n of oxides,

�H

r

(MOn). The heat of MOx formation monotonously decreases with the number of 3d-electrons. J.n the metal cat lons of interest, only divalent, trivalent, or mixed valent states are considered to be possible. Figure 5.3 shows the relation between the average oxidation state of M and (�1If(Mo1.5)-�H

r

(MO)). In this case,

�H

:f

(Co01.5) is calculated from the Born-Haber cycle by assuming that the lattice energy of Coz03 is equal to that of Fe2o3. It is apparent that Co and Ni with low (111I

f

(M01.5)-111l

r

(MO)) prefer the divalent state, but Fe and Cr wlth high (�H

[

(MOJ.5)-�II

[

(MO)) prefer the trivalent state in the hexaaJumlnate. Manganese ions are in the intermediate state betw en 2.0 and 3.0 as shown ln Table 5.3. The trivalent Mn ions were also observed in single­

crystal La-hexaaluminate by absorption spectra when the samp e prepared in Hz atmosphere was subsequently annealed in air.19 This result is consistent with the quantitative analysis in the present study.

5.4.3 Relation between Catalytic Activity and Oxidation State Figure 5.4 shows TPD profiles of oxygen from DaMAl11019-a·

The rate of oxygen desorption was plotted as a function of cata- lyst temperature. Oxygen desorption was observed for cation- substituted samples, but not for BaAl12019 up to 1000 �. The cation-substituted catalysts can be classified into two groups

- 91 -

Ch.apter 5

I :::!::

I (JI 0 N

0 I

E :::1.

-;u OJ L 0 c

o_ L 0 1./) 0 OJ

0.4 Al 0.2

0 0.4 Co 0.2

0 ^�

O.L ' Ni 0.2

0 0.4 0.2 0 0.4 0.2 0 0.4 0.2 0 0

Cr

Fe

�

Mn

�

�

200 400 600 800

Temperature I ·c

Fig. 5.4. TPD profiles of oxygen from BaMAl11019-a·

from the desorption patterns. Oxygen desorption from the first group, with M = Cr, Co and Ni, was extremely small� Samples in

this group showed the low catalytic activity for methane com­

bustion (Table 5. 2). In the second grou p, i.e., Mn- and Fe­

substituted systems, with relatively high activity for CH4 com­

bustion, a large amount of oxygen was desorbed at high tempera-

t u r e s ( >50 0 oc ) . The desorption started at

about 600 � and the peak of desorption rate located above 900 �­

The desorption from BaMnAl11019-a sample was rather small but located at a low temperature (ca. 850 �).

Obviously,the catalytic activity of BaMAl11019-a was re-

- 92 -

MATERIAL DESIGN OF COMBUSTION CATALYST BY STRUCTURAL MODIFICATION OF HEXAALUMINATE

fleeted by their TPD profiles (Fig. 5.4). Oxygen species, which were desorbed from the Mil- and Fe-substitutcd samples above 50 0 OC , a p p ear to b e a c t i _v e f o ^r me L h an e comb us L i on . The oxygen des_orption accompanies the reduc tion of transltJon elements, M, in the hexaaluminate lattice be_caus_e no desorption peaks appeared without substitution. The oxygen desorption from BaMAl11019-a

(M=Mn, Fe) corresponds to the partial reduction of M3+ to M2+.

Since reduced Mn or Fe species (divalent state) are reoxidized easily in 02 flow, an exothermic peak was observed when the

catalyst was heated in air after TPD measurement. As for systems of M=Cr, Co, and Ni, amounts of oxygen desorption were as small as that from unsubstituted barjum hexaaluminate . This result indicates that the oxidation states of these ions (cr3+, co2+, and Ni2+) are too stable in a hexaaluminate lattice to contribute to the reduction/oxidation cycle.

A sequence of catalytic activity (Tlo%) of BaMAl11019-a (Table 5.2) is apparently different from that of the single oxJde of M. A correlation between catalytic oxidation activity and the heat of formation of metal oxides has been proposed by several researchers.20-23 Here, we assume that the catalytic oxidation

of methane proceeds via the following reduction/oxidation cycle of M between di- and trivalent in the hexaaluminate lattice.

MO +

1 -0 4 2

MO

( 3)

+ + ( 4)

Thus, the reversibility of the reduction/oxidation cycle between MO and Mo1.5 can be estimated by (�Hr(MOl.s)-�Hf(MO)). The catalytic activity (Tlo%) of BaMAl11019-a is plotted against

- 93 -

Chapter 5

500

�---�

ou -

600

o-0

..-:=-

0

700

Al-4

0 -50 -100 -150 -200 6Ht(M01_5)-6Hf(MO)/ ^kJmol-1

Fig. 5.5. Relation between catalytic activity of BaMA111o19^_a and (�Hf(M01.5)-�Hr(MO)).

this parameter in F1g. 5.5. It is apparent that the activity is roughly summarized by a volcano-like relation with (�H;(M01.5)-

�Hf(MO)). o The relation indicates that the reduction of M01.5

into MO is the slowest step in methane oxidation over the cata- lysts located on the right arm in Fig. 5.5. Thus, catalytic activity decreases with the increasing stability of M01.5 states.

In contrast, the oxidation of MO into M01.5 is rather difficult and the catalytic activity decrease with the increasing stability

of MO states in the left arm ln Fig. 5.5. _B

a

sed on this volcano­

like relation. manganese ions play a key role ln the reduction/

oxidation cycle with the lowest energy, and this will probably be effective in methane oxidation.

- 94 -

MATERIAL DESIGN OF COMI3USTI ON CATALYST BY STRUCTURAL MODIFICATION OF HEXAALUMINATE

5.5 Structural Modification of Hexaalumlnate Catalyst 5.5.1 Catalytic Properties of BaMnxA112^_xo19^_a

For the practical application to catalytic combustors, the catalytic activity of the Mn-substituted hexaalumlnate needs to be further improved by structural modifications. In th]s section the hexaaluminate with various Mn-contents was examined as the structural modification.

The crystal structure of Mn-substituted system was investi­

gated by X-ray diffraction as a function of atomic fraction of Mn in BaMnxA11₂-x0¹9-a ^· The lattice constant of hexaaluminate structure increased with the substitution of Mn ions for Al3+

sites at x < 3.0 _(Fig. 5.6). Further addition of Mn at x > 3.0

did not change the lattice constant, but a second phase similar to BaAl204 appeared. As a result, samples at x ^< 3.0 _{are homo-} geneous solid solutions of hexaaluminate but th _{- ose a} t _{x >} 3 . 0 are mixtures of hexaaluminate and BaAl2o4-llke phase.

Figure 5.7 shows the change in surface areas of BaMnxAJ12_x- 019-a calc1'ned at 1300 °C. _{Alth - oug h tl -�le} surface area decreased unvaryingly with x, the decrease was gradual at x < 2.0. At this composition, the sample consisted of the single phase of hexaalu­

minate. In contrast, the surface area was steeply reduced as the second phase precipitated at x > 2.0. Although the hexaalumlnate structure is effective in retaining the large surface area, the BaAl204 phase does not contribute to thJs effect (Chapter 3).

The cation substitution in barium hexaaluminate does not lead to any serious reduction of surface area only when the single phase of hexaaluminate is retained.

The catalytic activity for methane combustion over BaMnx­

All2-x019^-a was tested as a function of x in Fig. 5.8. At x <

3.0, the catalytic activity increased with Mn-content because of

- 95 -

Chapter 5

0.568 -

Q

E

c

-⁰

0.564-

r1:l

0.560 _Q

0 1.0 2.0 3.0 4.0 5.0

Fig. 5.6. Lattice constants of 8aMnxAl1z-x019-a after calci­

nation at 1300 �-

I(J)

N

E

-ru

10

ClJ L

rU

<lJ u

"+-rU

5

L -

::::J c.J)

0 0 1.0 2.0 3.0 L+.O 5.0

X

In BaMnxAl12-x019-0

Fig. 5.7. Surface area of the BaMnxAl1z-x019-a system after calcination at 1300 �-

- 96 -

MATERI AL DESIGN OF COMBUSTION CATALYST BY STRUCTURAL MOD!FICAT!ON OF IIEXAALUMINATE

100

� 0

80

-

I---t u

60

"+-

0 0 c

.U)

_L

L.O

(1} >

c 0

20

u

400 500 600 700 800

Reaction temperature/ OC

Fig. 5.8. Catalytic combustion of methane over BaMnxAl1z-x019_a·

x Surface areajm2/g

0 () 1. 0

• _2.0 0 _4.0

15.3 13.7 10.4 0.9

Reaction condition: CH4 1 _vol%, air 99 vol%, S.V.=48000 h-1.

the increase in population of the active sites, which contribute to the reduction/oxidation reaction. However, the catalytic

activity was appreciably lowered by the steep decrease in the surface area at x ^> 3.0, where the second phase with large crys­

tallite size appeared. Since the relative activity per unit surface area is rather low in Mn-substituted hexaaluminate as was revealed in Table 5.3, the Mn content, x, should be low enough to retain the large surface area.

- 97 -