Dielectric polarization of thermochemically treated MgO : Li+ crystals

(1)

HAL Id: jpa-00220006

https://hal.archives-ouvertes.fr/jpa-00220006

Submitted on 1 Jan 1980

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.

Dielectric polarization of thermochemically treated

MgO : Li+ crystals

J. Crawford, Jr. Eisenberg, D. Eisenberg

To cite this version:

(2)

JOURNAL DE PHYSIQUE _Colloque_C6,_{supplement au no 7 , Tome}_41,_Juillet_1980,_page_C6-394

Dielectric polarization of thermochemically treated

MgO

:

Li+ crystals

(*)

J. H. Crawford, Jr. a n d D. J . Eisenberg

Department of Physics and Astronomy, University of North Carolina at Chapel Hill, Chapel Hill, N.C. 27514, USA

RCsurnC.

-

On a utilise les mesures de la perte diklectrique et la dtpolarisation thermiquement stimulee en cristaux de MgO : Li, traitts thermochimiquement i 1 400 K dans les atmospheres oxydantes, afin d'examiner 1'Cvolution des rnicrogalaxies, ou les inclusions semiconductrices avec grande teneur des centres [LiI0. Ces regions ont une forme filamenteuse, se developpant evidemment le long des dislocations-coins. L'analyse de ces donnees produit la polarisation saturee, le rapport des axes majeurs et mineurs des inclusions (traitkes comme sphero'ides allongi.~), et la fraction volumCtrique occupee par eux. On a comparC ces rbultats, a la fois avec la bande d'absorption optique de [ ~ i ] ' i 1,83 eV, pour les cristaux chauffes en air statique et en oxygene pur et coulant. La grande difference entre les effets des deux atmospheres indique que la pression partielle effective de I'oxygene joue un r61e important durant l'kolution des inclusions. On suggkre que l'introduction trks rapide de ces inclusions dans I'Cchantillon est causee par la diffusion d'oxygkne (ou des lacunes cationiques) le long des canaux des dislocations-coins, et de la diffusion rapide de I'interstitiel Li+.

Abstract. - Dielectric loss (DL) and thermally stimulated depolarization measurements (TSD) on MgO : Li crystals, thermochemically treated at 1 400 K in oxidizing atmospheres, have been used to investigate the evolution of microgalaxies or semiconducting inclusions with a high [LiI0 center content. The inclusions are filamentary in natbre, evidently forming along edge dislocations. Analysis of DL and TSD data yield the saturation polarization, the ratio of major-to-minor axes of the inclusions treated as prolate spheroids and the volume fraction occupied by the inclusions. These together with the 1.83 eV [ ~ i ] ' optical absorption band are compared for crystals heated in static air and in pure flowing oxygen. The large difference between the two atmospheres indicates that the effective oxygen partial pressure plays an important role in the time evolution of the inclusions. It is suggested that the very rapid introduction of semiconducting inclusions throughout the bulk is due to pipe diffusion of oxygen (or cation vacancies) along edge dislocations and rapid diffusion of interstitial Li'

.

Annealing of M g O : Li+ crystals in a n oxidizing atmosphere above 1 150 K followed by quenching t o room temperature introduces stable [LiI0 centers [I]. This center is a charge-compensated ([o=-Li+- 0 - I ) ,

<

100

>

axial complex which yields a hole upon thermal o r optical ionization. Hence t h e [LiI0 is a n acceptor introduced by the thermochemical treatment. It was previously identified in M g O : L i + crystals exposed t o X- o r y-rays by ESR [2] o r E N D O R [3] and, more recently, magnetic circular dichroism measurements [4] o n annealed crystals confirm t h e previous suggestion [5] that the 1.83 eV optical b a n d , half-width

-

0.5 eV, is associated with [LiI0. The radiation-induced centers become thermally unstable losing holes near 220 K which then recombine with trapped electrons. From both isochronal anneals a n d glow,-peak analysis, the ionization energy is estimated t o be

-

0.6 eV.

Thermally stimulated depolarization (TDS) a n d

(*) This work supported by U.S. Department of Energy under contract No. DE-AS05-78-ER05866.

dielectric loss (DL) measurements o n crystals contain- ing stable [LiI0 reveal a n exceptionally large bulk polarization which indicates a Maxwell-Wagner relaxation within isolated conducting inclusions rather than a Debye-like dipole reorientation 161. T h e magnitude of the loss tangent a t t h e peak (tan 6

>

1) is t o o large t o arise from spherical inclusions a n d suggests a filamentary structure. T h e Sillars- Van Beek [7] analysis of Maxwell-Wagner relaxation for prolate spheroids shows the major-to-minor axis ratio R is typically

-

100 for moderate annealing times. Such a n elongated structure suggests t h a t dislocations a r e somehow involved. Recently, Orera

et al. [8] have found both positive a n d negative dislocation-decoration effects depending u p o n whether MgO : Li+ crystals were deformed before o r after therrnochemical treatment.

Both T S D curves and D L vs. frequency curves [6] a r e consistent with a temperature dependent relaxation time, T = TO exp[E/kT], in which E 0.6 eV

a n d TO

-

10-l3 s. Since T is inversely proportional t o

(3)

DIELECTRIC POLARIZATION OF THERMOCHEMICALLY TREATED MgO : Li+ CRYSTALS C6-395

conductivity a of the inclusions and 0.6 eV corres- ponds to the ionization energy of radiation-induced [ ~ i ] ~ , this temperature dependence is apparently mainly due to [LiI0 ionization. Another significant development is the observation of a non-linear current-voltage characteristic for thermochemically treated MgO : Li+ which indicates that field-assisted tunneling between one semiconducting filament and another occurs when a sufficiently large voltage is applied.

Although first formed as filamentary structures, for long annealing times the conducting inclusions would be expected to spread out and eventually be distributed homogeneously as a result of diffusion. To explore such a possibility and to examine the effects of different atmospheres, we investigated the effect of annealing time in both static air (SA) and pure flowing oxygen (FO) upon several parameters : (a) saturation polarization as indicated by the area

A under the TSD peak, (b) the axial ratio R, ( c ) the

volume fraction V; and (d) the absorption coefficient a . The results are shown in figure 1.

32

k

\, f i ANNEALED IN STATIC AIR 4 9 2 0 4 4 0 4 4 8

ANNEALING T l M E ( S E C I a> 2 7 - 180 <.- -- 2 1 -- 1 4 0 .... 3 6 0 - 1 8 - I 2 0 ANNEALED I N FLOWING OXYGEN AT 1400 K _{- 1 5 - 1 0 0}

-

ABSORPTION COEFFICIENT - - - ---- SATURATION POLARIZATION AXIAL RATIO ... VOLUME FRACTION I I I I I

l o Z lo3 lo4 lo5

ANNEALING TlME ISEC I

b)

Fig. 1. - The influence of annealing at 1 400 K upon optical absorption coefficient or , volume fraction V . . .

.

, axial ratio R - - -, and saturation polarization A

-.-.

Panel (a) refers to anneal in static air and Panel (b) refers to pure flowing oxy- gen.

The behavior of the various parameters for the SA case is shown in figure la. Log time is chosen as the abscissa to show the initial behavior. The axial ratio R declines from the first point of 600 s annealing time as does the saturation polarization. The value of a passes through a maximum near

2 000 s and then declines showing recovery late in the anneal, whereas the volume fraction V occupied by the inclusions increases gradually throughout. For FO annealing the same general trends are observ- ed but the changes are much less drastic and the time scale is shifted. Comparison of the effect of annealing under SA vs. FO conditions suggests that in the former, though the polarizable inclusions are rapidly established, there is insufficient oxygen pre- sent to supply the holes necessary to sustain the [LiJo center density and inclusion shape necessary to maintain the high initial polarization. Evidently, holes are lost to slowly developing, more stable structures Femote from the dislocations so that the filamentary inclusions either shorten or fragment, drastically reducing R at long annealing times.

FO annealing on the other hand results in a much more stable inclusion structure and only at long annealing times d o the values of R and A begin to

decline. The gradual increase in both A and V suggests that the filaments are indeed growing fatter. The final consequence of such a growth would be merging of inclusions with loss of polarization. However, it is too early to say whether such an, effect is responsible for the slight loss in A a t the longest anneals. The variation in a may well be due

to extinction effects because of the high local [LiI0 concentrations in the inclusions. These results are in general agreement with those of Chen et al. who

used different oxygen partial pressures in a flowing gas mixture.

The most striking aspect of these results is the rapid initial formation of the semiconducting inclusions throughout the bulk of the crystal. As can be seen from figure Ib the polarizable inclusions are well established after only a 90 s FO anneal. To account for this behavior we propose the following sequence of events : (1) Li20 precipitate particles begin to dissolve at and above

-

1 200 K. We pos- tulate that interstitial Li', Li:, migrates rapidly but dissolves substitutionally in Mg2+ sites. In effect the Li: is the agent maintaining equilibrium between the Li20 and the Mg2+ vacancies with which it inte- racts. This behavior is analogous to the migration and dissolution of copper and gold in germanium and s licon except for the additional requirement of charge balance. Boldu et al. [9] have shown that charge balance is provided by oxygen absorbed on or taken into the crystal.

(4)

C6-396 J. H. CRAWFORD A N D D . J. EISENBERG gen. A peculiar kind of dislocation climb can be

induced by oxygen atoms in the dislocation core, each releasing two holes and a Mg2+ vacancy. The latter will not migrate far from the dislocation before capturing a I& and a hole to form a [ ~ i ] ~ . The remaining hole can migrate to the Li20 precipitate restoring charge balance so that the process can be repeated. The driving force is the availability of oxygen in the dislocation core, and hence the partial pressure of oxygen in the gas, since the energy required to form a charge compensated [LiI0 center is much smaller than that necessary to form an uncompensated Mg2+ vacancy. Another mechanism which would have the same chemical consequence as oxygen pipe diffusion is the surface absorption of oxygen followed by the release of two holes and a cation vacancy which enters the crystal along the dislocation network. In the first case chemical climb is the pump producing Mg2+ vacancies and holes,

whereas in the second the surface growth provides the driving force. The results would be equivalent provided in the second case cation vacancies are pumped into the crystal along edge dislocation segments.

This chemical model for the formation of filamentary semiconducting inclusions is of course spe- culative and needs to be tested experimentally. Obviously, more studies of the effect of dislocation density on the kinetics of formation (and remo- val) of inclusions are needed. We believe such studies will be richly rewarded since they should give insights into the general problem of oxidation reac- tions in refractory oxides.

We wish to thank Dr. Y. Chen for supplying the crystals used in this work, for his keen interest and encouragement and for helpful discussions. We also are grateful to Drs. L. M. Slifkin and J. L. Bold6 for valuable discussions and insights.

DISCUSSION

Question. - A. S. NOWICK.

Do the precipitates remain stable (so that the solubility limit continues to be exceeded) up to the highest heat treatment temperatures that you use ? Also, what is the reason for the maximum in the LiO absorption and the conductivity as a function of heat treatment temperature as shown in the paper by Chen ?

Reply. - J. H . CRAWFORD.

Our work was confined to anneals at 1 400 K. However, Dr. Chen and coworkers find an enhance- ment of field-assisted tunneling conductivity and optical absorption up to 1 600 K indicating an increas- ed (Li)' concentration within the inclusions and possibly a change in inclusion size. The reason for the retrogression above 1 600 K is not understood but it may be due to the demand for Li+ ions in the region of the dissolving precipitate particles.

Question. - R. CAPELLETTI.

Just a comment. We have seen analogous depolarization peaks related to dislocations decorated by impurities and vacancies in alkali halides. The peaks were deeply affected in amplitude and position by deformation; similar behaviour should be expected for your peak in MgO : Li.

cations can obviously be important for pipe diffusion from the surface but the dislocation density would have to be

-

1010/cm2 to play a role in dispersion of Li from the precipitates.

Reply.

-

J. H. CRAWFORD.

We believe that the cylindrical semiconducting inclusions are formed remote from the precipitate particles by rapidly migrating interstitial Li' ions. Therefore, there is no direct connection between dislocation density required and the concentration of precipitate particles in our model.

Question. - W. J. FREDERICKS.

Does magnesium nitride form during the 1 300 OC anneal of MgO in air ? (when Mg is burned in air a substantial amount of Mg nitride is formed).

Reply.

-

J. H. CRAWFORD.

No, we have not looked for nitrides. Your question suggests perhaps we should.

Question. - J. S. DRYDEN.

In a Maxwell-Wagner system such as you suggest the conducting particles will most likely vary in size and shape leading to a distribution of relaxation times. Why do you obtain a single relaxation time ?

Reply. - J . H. CRAWFORD.

Reply. - J . H. CRAWFORD. We expect that dislocation edge segments are Such studies are planned for our system. involved and that the lengths cover a narrow range. When we measure either the TSD or dielectric loss

(5)

DIELECTRIC POLARIZATION OF THERMOCHEMICALLY TREATED MgO : Li+ CRYSTALS C6-397

Question. - A. KESSLER.

1) Did you observe any shift of the maximum temperature Tmax of the TSD-peak with a changing annealing temperature ?

2) By annealing you change the number of charge carriers. There exist in some ionic crystals first order kinetics peaks due to a crystal-electrode interface polarization of the Maxwell-Wagner type by blocked mobile defects Tma, which shifts with a changing defect concentration (e.g. KRUZE, I., MULLER, P.,

Phys. Status

Solidi

(a) 13, 197 (p. 72) ; KESSLER, A.,

Contribution A67). Can you point out what makes the difference in your peak position ?

Reply. - J. H. CRAWFORD.

1) Yes, but only in the static air anneals. The peak broadens appreciably and shifts to lower temperature. 2) We believe that the shift in the peak is associated with a fragmentation of the cylindrical inclusions along dislocations in the situation where there is insufficient oxygen to sustain them.

References CHEN, Y., TOHVER, H. T., NARAYAN, J. and ABRAHAM, M. M.,

Phys. Rev. B 16 (1977) 5535.

SCHIRMER, 0. F., J. Phys. Chem. Solids 32 (1971) 499.

ABRAHAM, M. M., UNRUH, W. P. and CHEN, Y., Phys. Rev. B

10 (1974) 3540.

(41 MODINE, F. A., Solid State Commun. 20 (1976) 1097.

[5] CHEN, Y. and ABRAHAM, M. M., J. Am. Ceram. Soc. 59 (1976) 101.

[6] EISENBERG, D. J., CAIN, L. S., LEE, K. H. and CRAWFORD, J. H., Appl. Phys. Lett. 33 (1978) 479.

171 VAN BEEK, L. K. H., Physica 26 (1960) 66.

[8] ORERA, V. M., CHEN, Y. and ABRAHAM, M. M., Phys. Rev. B (to be published).

[9] BOLDU, J. L., ABRAHAM, M. M. and CHEN, Y., Phys. Rev. B

19 (1979) 4421.

See also :