• Aucun résultat trouvé

The Jahn-Teller theory of Mn3+ : Al2O3

N/A
N/A
Protected

Academic year: 2021

Partager "The Jahn-Teller theory of Mn3+ : Al2O3"

Copied!
6
0
0

Texte intégral

(1)

HAL Id: jpa-00208472

https://hal.archives-ouvertes.fr/jpa-00208472

Submitted on 1 Jan 1976

HAL

is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire

HAL, est

destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.

The Jahn-Teller theory of Mn3+ : Al2O3

C.A. Bates, P. Brauns, J.R. Fletcher, P.C. Jaussaud

To cite this version:

C.A. Bates, P. Brauns, J.R. Fletcher, P.C. Jaussaud. The Jahn-Teller theory of Mn3+ : Al2O3.

Journal de Physique, 1976, 37 (6), pp.763-767. �10.1051/jphys:01976003706076300�. �jpa-00208472�

(2)

THE JAHN-TELLER THEORY OF Mn3+ : Al2O3

C. A. BATES

(*),

P. BRAUNS

(t),

J. R. FLETCHER

(*)

and P. C. JAUSSAUD

(t)

(*) Department

of

Physics, University

of

Nottingham, University Park, Nottingham

NG7

2RD, England (t)

Service

B.T., C.E.N.G.,

B.P.

85,

Centre de

Tri,

38041

Grenoble,

France

(Reçu

le 8 décembre

1975,

révisé le

16 février 1976, accepté

le

17 février 1976)

Résumé. 2014 Une étude expérimentale des ions Mn3+ par

spectroscopie infrarouge

(Aurbach et

Richard)

dans

Al2O3

vient d’être publiée. Nous présentons ici une analyse

théorique

détaillée de ces

résultats, ainsi que de ceux que nous avions obtenus en résonance

paramagnétique acoustique.

Nous

montrons que tous les résultats

expérimentaux

connus à ce jour sur l’ion Mn3+ sont

expliqués

de

manière satisfaisante à l’aide d’un effet Jahn-Teller sur l’ion dont le fondamental est un 5E. On obtient ainsi les paramètres de réduction

Jahn-Teller q

= 0,50, r = 0,70. p est différent de zéro, mais ne peut être déterminé. Ces résultats sont tels que 2 q - p 1, et donc que le modèle de Cluster n’est pas valable.

Abstract. 2014

Following

the recent

publication

of infra-red measurements on crystals of

Al2O3 containing

Mn3+ ions, a detailed theoretical analysis of the system has been undertaken. From these results and our earlier APR measurements, it is shown by computer

analysis

that all the data for this 5E ion can be

satisfactorily explained

by a multimode lattice model of the Jahn-Teller effect. The

values q

= 0.50, r = 0.70 and p finite are found for the Jahn-Teller reduction factors, so that

(2 q 2014

p)

is less than unity.

Classification

Physics Abstracts

8.822 - 8.630 - 8.512

1. Introduction. -

(3d)4

ions in octahedral or near-

octahedral sites are well known to exhibit

strong

Jahn-Teller effects as

they

have

’E (S

= 2 T =

2)

ground

states. The

A’203

lattice is a

particularly important

host and

many experimental

measurements and theoretical calculations have been made. The results obtained have been summarized in three papers

[1-3]

and in references contained therein.

Experiments using

far infra-red

spectroscopic

methods

have

subsequently

been carried out on

crystals

of

Mn3+ : A1203. Preliminary

measurements were

reported by Stoelinga et

al.

[4]

and

by

Aurbach et

al.

[5]

and more detailed results in the

frequency

range 3 to 30 cm -1 and with

applied magnetic

fields up to 55 kG have

just

been

given by

Aurbach and Richards

[6].

While a set of values for the parameters for the

Cr2 + : A’203

system were deduced in

[3]

from the APR

data,

the lack of accurate information on the

Mn3 + : A’203

system

precluded

a detailed

analysis

for that

system. However,

the new results

[6]

have indicated that the

problem

of

Mn3 + : A’203

should be re-

opened

and it is the purpose of this note to

give

the

results obtained.

2.

Theory.

- Consider a

single 3d4

ion

coupled

to

E-type displacements

of its

surroundings.

The Hamil-

tonian for this

system

can be written as

where YE

is the orbit-lattice

coupling

constant for

E-type displacements (Q±

=

Qo

±

iQe)

of the nearest

neighbours, 3Çmagnetic is

the

magnetic

Hamiltonian as

given

in eq.

(3)

of

[3]

and

Jelattice

is the lattice Hamilto- nian. An

important

contribution to

JClattice

is the

anharmonic

term -1 B(Q 3

+

Q 3 ).

We

simplify

our discussion

by taking

the

magnetic

field H

parallel

to the

crystal

c-axis so that the effects of randon internai strains and

electric fields

can be

neglected [1].

That

is,

we take

Q + = Q - = 0 in Ilmagnetic

We also

neglect

the

trigonal

distortions in

JClattice

and assume therefore that the Jahn-Teller

coupling

is

diagonalized by

the

unitary

transformation

exp(iOT3) given

in

[3]

where 0 = tan -1

(QeIQo).

However, we assumed there that the

coupling

was very strong so that the cluster model limit was used for which the reduction factors had the values p =

0,

q = 0.485 and r =

2-q.

However, our recent ana-

Article published online by EDP Sciences and available at http://dx.doi.org/10.1051/jphys:01976003706076300

(3)

764

lysis [7]

of the Jahn-Teller

theory

of Ni 3 + ions in

A1203,

which have a 2E

ground

state, has shown that

a multimode lattice Jahn-Teller model and not a

cluster model is needed to

explain

the results. It

therefore seems

appropriate

to use a similar model for the

Mn3+ : A’203

system.

As a result of the

transformation,

the

angular part

of the effective vibronic Hamiltonian can be written as

where

O:t, 0 , 0+ and 0 +

are operators

transforming

as

Et

of the

C3

group. The 0 operators are linear in S and H and the 0 operators are

quadratic

in S as

follows

where the

signs

are taken in

pairs

as indicated. The parameters have been derived

by perturbation theory [3]

and may be

approximated

to the values

where

and

The relative

energies

of the excited levels

5T2(A), ST2(E)

and

3T 1 (3H)

are

Al, A 2

and E

respectively,

A

- 13 (A1

+

2 d2)

and the

trigonal splitting

v =

d 2 - A 1.

ba is the

off-diagonal trigonal

field

parameter. In eq.

(2),

p

(-0.1 cm -1)

is the

spin- spin

constant,

Ro

is the value

of Q + Q _ at

the mini-

mum in the

potential

energy versus

.JQ+ Q-

curve.

J e

is the momentum

conjugate

to 0 and M is the

ligand

mass..

As the first two

spin-independent

terms in

Jerot

are

much

larger

than the

remaining magnetic

terms,

they

give

a lower vibronic doublet

15E >

and an upper vibronic

singlet 1 5 A1 > (for

the

anharmonicity

cons-

tant B >

0) or 1 5A2 > (for B 0)

at a relative energy of b as before. We

neglect

all other

higher

states in the

analysis.

The

spin-dependent

terms in

remove the

spin-degeneracy

and

couple together

the

1 5E >

and

1 5A1 > (or SA2 >)

states. The fifteen states then

split

into three sets of five states, the sets

transforming

like

A, E+

and E- of the

C3

group

respectively.

The

additions and modifications to the model mentioned above

modify

the

original

matrix

appearing

in

table II of

[3]

to the three

(5

x

5)

matrices

given

here

in table I. The

problem

is to obtain values for the

parameters (A, f, Do, D’, K2, b,

g, p, q and

r)

which

account

satisfactorily

for all the

expérimental

data.

As the

NI3 + : A’203 system

was solved

by analytic methods, attempts

were made to use a similar method here.

Unfortunately,

it did not succeed as there were

many more terms

contributing

to each

expression

than in the case

of Ni3+ (as

the

spin degeneracy

is five

instead of

two)

and each term was of more or less the

same

magnitude making perturbation

calculations invalid. This means that numerical methods had to be used once

again

for the

3d4

system.

3.

Summary

of

expérimental

data and parameters obtained. - The data obtained from APR is

given

in table 1 of

[1].

For

example,

for an APR

frequency

v - 9.3

GHz,

the

edges

in the resonance lines occur

at

magnetic

fields H

of 4.8, 5.15, 10.8,

11.9 and 18.5

(all

±

0.2)

kG for the lines labelled

B, D, F,

E and G

respectively.

The new infra-red

data [6]

is

reproduced

in

figure 1 ;

the transformation

properties

of the

states, from which the levels are

derived,

are also

given. (The experimental points given

in the

figure

were those obtained in the first detailed

experiments

of

Aurbach and Richards and these results were used as a basis for

obtaining

the fit described here.

They

therefore do not

exactly

coincide with the

points finally

included in the

corresponding figure

3 of

Aurbach and Richards

[6].)

A

Telemechanique

T1600

computer

has been used to

diagonalize separately

the three 5 x 5 matrices

by

successive

approximations by simultaneously varying

the

parameters ap,f, Do, D’, K2, b, g, q

and r. The

program

stopped

when further corrections gave the

(4)

TABLE 1

The matrix

of

the

ground

rotational states

of jé

when

H //

c. The

(15

x

15)

matrix

splits

into three

(5

x

5)

matrices shown as blocks

A,

B and C which

transform

as

E+,

E- and A

respectively

under the

C3

group. The lower

half of

each matrix

only

is

given for simplicity.

The states

labelling

the columns and

rows are in the

form vibronic

state;

spin

state in

C3

symmetry

>.

The remaining parameters are defined in [3].

same results within the

experimental

error.

Many

different

starting points

were used for the

input

data to ensure that the fit obtained was

unique.

The

energies

of the 14 levels relative to the

ground

level

were

computed ;

the nine lowest levels are al*sô

6hpwn

in the

figure

and the

good

agreement with

thé

infra- red data is

readily displayed

with the

parameters :

From

[3],

this set of parameters

gives

the additional results

so that k =

0.06,

1 = 0.001 4.

Also, d

= 5.52

cm-1

and e = 0.13

cm-’.

If we label the levels from the lowest

upwards,

our

conclusions are in

agreement

with the new data

[6]

for the transitions

1 -+ 3, 1 -+ 4, 1 -+ 5

and 1 -+ 6.

However,

we

suggest

that the set of

experimental points

labelled 1 - 9 in

[6]

may be attributed to the transitions 2 -+ 7 and 2 -+ 8.

The other

pieces

of information

explained by

our

analysis

are as follows :

(i)

Five main APR lines

only

are seen

[1]

corres-

ponding

to the transitions between the levels labelled

2,

3 and 4

only.

(ii)

As v decreases

[ 1 ],

the

magnetic

field at the

edge

of the APR lines decreases for the

B,

G and F

lines,

but it increases for the E and D lines.

(5)

766

FIG.1. - The energy level diagram for Mn3+ : Al203 for zero-strain sites as a function of H parallel to c. The infra-red data of Aurbach and Richards [6] is also included. The levels are labelled 1 to 9 and

have the C3 symmetry properties shown.

(iii)

In order to illustrate the closeness of the fit of the APR data

[1]

we have followed

[6]

and calculated the energy

separations

as a function of H between all

possible pairs

of levels labelled

2,

3 and 4 as deduced from our parameters. The results are shown in

figure

2.

On the

figure,

we have also

superimposed

the APR

data of

[1].

The closeness of this data to the calculated energy

separations clearly

demonstrates the

goodness

of the fit

given by

our parameters.

(iv)

The maxima in the

absorption

in the APR

experiments

occur at

(5.2

±

0.5)

K for all lines with the G-line

arising

from a

slightly higher

energy level than the rest

[1].

We need also

explain

the relative intensities of the infra-red lines. When H =

0,

the levels 3 and

4, transforming

as

E+

and E_

respectively,

are

dege-

nerate. If the

trigonal

field were also zero,

they

would

also be

degenerate

with level 2 and

together

.form

the

Tl triplet

in cubic

symmetry.

Levels 5 and 6 are also

degenerate

when H = 0 and transform as E in

FiG. 2. - The calculated energy separations between levels 2 - 3, 2 -+ 4 and 3 -+ 4 as a function of H. The APR data [1] is also

included.

both cubic and

trigonal

symmetry. Thus IR transitions from the A

ground

state appear to be allowed for both the A -+ E cubic

(and

thus for 1 ---> 5 and

1 -+ 6)

and

the A -+

T1

cubic

(i.e.

1 -+

2, 1

- 3 and

1 -+ 4)

transitions in contradiction to

experiment

if the

transition is induced

by

an operator of the form

E;r r. However,

it is well known from

thermally-

detected electric-field-induced EPR

[9]

that the

r f

electric field acts via an ionic effect in which the

neighbours

are

displaced

in

A’203 crystals.

We then

suppose that the infra-red transitions are

similarly

induced. The main contribution then arises from

E-type

ionic

displacements

and so the A - E tran-

sition is allowed whereas the A -+

Tl

is

forbidden,

in agreement with the observations

([6], Fig. 4).

The

transitions induced

by Eu,

via the ionic effects have the

same

properties

as acoustic transitions

[9].

Thus

in thermal

conductivity experiments, only

the A -+ E

transition is allowed and this is observed

[10].

As H

is increased from zero, the Zeeman term

(dominating

the

trigonal

field

mixing)

mixes the

E+

and E_ states

thus

enabling

the 1 - 3 and 1 - 4 transitions to be observed.

However,

the 1 --+ 2 transition remains almost forbidden and

gives only

a small amount of

absorption,

in agreement with observation.

In our

analysis

we have not used the infra-red data obtained for H

perpendicular

to c. For ions

strongly coupled

to their

surroundings,

the

signal peak

comes

from those

Mn3+

ions

having

a non-zero value of

strain

[1].

That

is, Q+

and

Q_

are non-zero. The

energy level

diagram

for such sites is different from that for sites with zero strain and

unfortunately

we

have insufficient information to relate them.

4. Discussion of results. -

Taking

 - 80 cm-1 1

gives

4 - 20 000

(see [3]

for

details). Assuming

the

off-diagonal trigonal

field parameter v’ to be 500

cm-1,

(6)

gives

ba N - 0.025 so

that, using

the value of

K2

calculated we get

vld - -

0.07. Therefore the dia-

gonal crystal

field parameter v is - 1 400

cm -1

and

Ki - -

0.012 so that b - 0.10 cm -1

and j -

0.001.

As far as the Jahn-Teller

theory

is

concerned,

the

important

parameter to obtain is p. If we use the value of

fp

deduced

by computer,

and the above value of ba

we

obtain p -

1. If ba is

increased, p

is reduced

by

the

same factor.

Alternatively,

if we use the

computed

value of ap and the formula

given

in

[3]

we find p is much smaller. This lack of agreement it not

surprising

as

omits fourth order terms in the

spin

operators and second order in H. As in the case of

Ni3+ [2],

the

magnitude

of a is

critically dependent

on

higher

order

terms but such corrections are very difficult to incor-

porate

for

(3d)4

ions.

Unfortunately therefore,

it

is

impossible

to deduce an accurate value

for p

but it is

undoubtedly

non-zero. This means that

(2 q - p)

is

less than

unity

but the difference cannot be’

reliably

found.

However,

we know from

experiment [3, 10]

that thete is a

large

amount of

phonon scattering

from

Mn3+

ions in

A’203

so that the Jahn-Teller

coupling

is strong. This indicates that while

(2 q - p)

is defi-

nitely

less than

unity, implying

that a multimode

full-lattice model of the Jahn-Teller effect should be

employed [8],

the difference from

unity

is

expected

to

be

quite

small.

Anharmonicity

effects are also

important

as

ô - 33

cm -1.

Transitions between the

ground

states

and excited states are thus

expected

in thermal

conductivity experiments

at

energies

between 30 and 50 cm -1.

Experimentally,

a mean

separation

of - 38

cm -1

is found

[10].

The

corresponding

inver-

sion

splitting

for

NI3 + : A’203

is found to be

57 ± 5 cm -1

[7] again indicating

that both the Jahn- Teller

coupling

constant and

anharmonicity

effects

are

larger

for

Mn3+

than for

NI3+.

In

addition, q

is found to be

positive

after

diagonalization implying

that B is

negative [3].

5. Conclusions. - The energy

level diagram

obtain-

ed for

Mn3+ adequately

accounts for both the IR data and the observation of

only

five APR lines.

As

only

five strong APR lines were also observed in all

Cr2 + : A’203 samples,

it is also our belief that the

Cr2 +

transitions must have an

origin

very similar to those for Mn3+ as the line

shapes

and

positions

had

an identical pattern. We thus suppose that the

Cr2 +

energy level

diagram

must be very similar to that of

figure 1, suitably scaled,

and not that shown as

figure

1 in

[3].

It has not yet been

possible

to do a

revised calculation for

Cr2+ using

a non-zero value for p, as at the moment

corresponding

IR data

(or

any other suitable

alternative)

is

lacking.

Acknowledgments.

- We wish to thank Pr. P. L. Ri-

chards and Dr. R. L. Aurbach for

bringing

their work

to our

attention,

for discussions and

correspondence,

and for

sending

us their results

prior

to

publication.

We

would also like to thank our

colleagues

for many useful discussions on the

problems

of

(3d)4

ions in

A1203.

We are

grateful

to the Commissariat à

l’Energie Atomique

and to the Science Research Council for financial

support

which enabled this

investigation

to

proceed

as a

joint

programme.

References [1] ANDERSON, R. S., BATES, C. A. and JAUSSAUD, P. C., J. Phys. C.

Solid State Phys. 5 (1972) 3397.

[2] ANDERSON, R. S., BATES, C. A., JAUSSAUD, P. C. and RAMPTON, V. W., J. Phys. C. Solid State Phys. 5 (1972) 3414.

[3] BATES, C. A., JAUSSAUD, P. C. and SMITH, W., J. Phys. C. Solid

State Phys. 6 (1973) 898.

[4] STOELINGA, J. H. M., WYDER, P., CHALLIS, L. J. and DE GÖER, A.-M., J. Phys. C. Solid State Phys. 6 (1973) L486.

[5] AURBACH, R. L., RICHARDS, P. L. and FORMAN, R. A., Bull. Am.

Phys. Soc. 18 (1973) 1572.

[6] AURBACH, R. L. and RICHARDS, P. L., Phys. Rev. B 12 (1975)

2588.

[7] ABOU-GHANTOUS, M., BATES, C. A., CLARK, I. A., FLETCHER, J. R., JAUSSAUD, P. C. and MOORE, W. S., J. Phys. C.

Solid State Phys., 7 (1974) 2707

and

ABOU-GHANTOUS, M., JAUSSAUD, P. C., BATES, C. A., FLET-

CHER, J. R. and MOORE, W. S., Phys. Rev. Lett. 33 (1974)

530 and

ABOU-GHANTOUS, M., BATES, C. A., FLETCHER, J. R. and JAUSSAUD, P. C., J. Phys. C. Solid State Phys. 8 (1975) 3641.

[8] HALPERIN, B. and ENGLMAN, R., Phys. Rev. Lett. 31 (1973)

1052.

[9] MOORE, W. S., BATES, C. A. and AL-SHARBATI, T. M., J. Phys.

C. Solid State Phys. 6 (1973) L209 and L277.

[10] RIVALLIN, J. and SALCE, B., Proc. 2nd International Conference

on Phonon Scattering in Solids, Nottingham (Plenum, London) 1975, p. 184.

Références

Documents relatifs

A medium Jahn-Teller coupling with an E vibrational mode can explain the Zeeman splitt- ings observed in high magnetic field.. However the theoretical intensities are

in the case of a 3T1 triplet coupled to Eg modes of vibration that an intermediate situation does exist between these well known cases, and that this

L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des

After the correction for noncolinearity of the magnetization and the external field H was made, the coefficients a,, b, in (2) were found by means of a

A comparison with available data (chapter 3) shows that all phases of one type have the same or very similar structures which are distinguished from

These topographies show that the substructure formed above 1 520 OC consists of rough dislocation sheets in ( 110 ) planes parallel to the compression axis..

The Jahn–Teller (JT) parameters for small, aromatic, organic radicals, C n H n (n ¼ 4–7), bis(cyclopentadienyl)cobalt(II) (cobaltocene), a sodium cluster (Na 3 ), a silver cluster (Ag

[r]