Universidade de São Paulo

Instituto de Física de São Carlos

Departamento de Física e Ciência dos Materiais

"A Espectroscopia de Absorção de Raios-X

Aplicada ao Estudo da Estrutura
Atômica de Materiais Inorgânicos"

Valmor Roberto Mastelaro

Trabalho apresentado ao
Instituto de Física de São Carlos
Para o Concurso de Livre-Docência

São Carlos, S.P., Agosto de 2004

SERViÇO DE BIB~IOTECA
I F S C• U S P INFORMAÇAO
 Dedico este trabalho ....

Á meus Pais "in-memorian", responsáveis pela construção de um sonho

que se tomou realidade.

A minha esposa Solange e minhas filhas, Isabelle e Juliana por

participarem deste sonho e pelo amor, compreensão e carinho.

A todos os amigos que de alguma forma contribuíram na construção e

consolidação deste sonho.

11
Agradecimentos

.:. Aos alunos de Graduação e Pós-Graduação pela confiança que depositaram em

meu trabalho .

•:. Aos colegas do Laboratório de Materiais Vítreos do OEMa-UFSCar com os quais

convivi durante 5 anos e em especial ao Prof. Edgar Outra Zanotto por ter pro-
porcionado muito mais que conhecimentos científicos .

•:. Aos colegas do Grupo Crescimento de Cristais e Materiais Cerâmicos do IFSC-

USP e em especial ao Prof. Antonio Carlos Hernandes por ter me acolhido em
seu grupo de pesquisa .

•:. Ao Professor Aldo Felix Craievich pelo incentivo e por ter sido o responsável pe-
la realização de meu doutorado na França.

•:. Aos funcionários e técnicos do OEMa-UFSCar e do IFSC-USP pela prontidão e

qualidade na prestação de serviços .

•:. As agências de fomento FAPESP, CAPES e CNPq pelo suporte financeiro que
tornou possível a realização dos trabalhos descritos neste texto .

•:. Aos técnicos, secretárias e pesquisadores do Laboratório Nacional de Luz Sín-

crotron (LNLS) em Campinas onde realizei partes importantes de minhas pes-
quisas nestes últimos 7 anos .

•:. Aos colegas do OEMa-UFSCar, do IFSC-USP e de outras instituições pelo apoi-

o, colaboração, confiança, sugestões, respeito e amizade .

•:. A minha família pelo apoio e principalmente, por ter compreendido os momentos
O(

de ausência "física". A minha esposa Solange e as minhas filhas Isabelle e Juli-

ana meus eternos agradecimentos.

tu
 Resumo

o presente trabalho apresenta uma parte importante das pesquisas desenvolvi-

das pelo autor nos últimos 12 anos junto ao Laboratório de Materiais Vítreos do Depar-
tamento de Engenharia de Materiais da UFSCar e ao Grupo Crescimento de Cristais e
Materiais Cerâmicos do Instituto de Física de São Carlos, Universidade de São Paulo.
Inicialmente, foi realizada uma revisão dos aspectos teóricos e experimentais que en-
volvem a Espectroscopia de Absorção de Raios-X (XAS). Em seguida, foram apresen-
tados alguns dos trabalhos desenvolvidos utilizando a espectroscopia de absorção de
i
I raios-X como ferramenta de caracterização da ordem local em materiais vítreos, amor-
I fos e óxidos policristalinos. As publicações consideradas mais representativas em cada

I
,
I
um dos temas descritos foram anexadas ao texto como urna forma de sistematizar os
resultados alcançados.

IV
Abstract

This paper describes an important part of the research work carried out by the
author over the past 12 years in the Vitreous Materiais Laboratory of UFSCar's,
Departrnent of Materiais Engineering and in the Ceramic Materiais, and at the Crystal
Growth and Ceramic Group of the University of São Paulo's, São Carlos Institute of
Physics. I begin with a review of the fundamental aspects involving X-ray absorption
spectroscopy (XAS) experirnents and theory. This is followed bya description of some
of the researches in which the XAS technique was applied as a tool to characterize the
short-range order structure in glass, and in amorphous and polycrystalline oxide
materiais. The most representative papers on each of the research subjects discussed
in this text are atlached hereto.

v
 índice
Capitulo I
Introdução 01
Capitulo 11
Princípios Fundamentaisda Espectroscopiade Absorção de Raios-X 05
Breve Histórico sobre o Fenômenode Absorçãode Raios-X 06
O Espectro de Absorção de Raios-X 12
A Equação de EXAFS 16
A equação completa que descreve as oscilaçõesEXAFS 24
O fenômeno de espalhamentomúltiplo na análise do espectro EXAFS 25
Aná1isedas Estruturas Finas na borda de Absorção: XANES27
Obtenção de um espectro de absorção de raios-X 28
Métodosde medida do coeficiente de absorçãode raios-X 30

I
f
Análise do espectro EXAFS
Determinaçãodos dados estruturais a partir do espectro EXAFS
Principais vantagens e desvantagensda técnica de XAS
31
35
37
I Perspectivassobre o uso da técnica de XAS no Brasil 38
Referências 39
Capitulo 11I
A Espectroscopiade Absorção de Raios-X Aplicada no Estudo
Estrutural de Materiais Vítreos 42
Introdução 43
Relação entre a ordem local e a tendência à nucleação em vidros
silicatos 43
Estudo da ordem local em um vidro de composiçãoGe25Ga1OS65
que
apresenta o fenômeno de foto-expansão 44
Conclusão .• 46
Referências 47
Artigos: Caracterização Estrutural de Vidros silicatos 48
Artigos: Relação entre a ordem local em vidros de composição
Ge25Ga10S65
e o fenômeno de foto-expansão 48

VI

I F S ( • U S P SERViÇO DE 8IB~IOTECA
INFORMAÇAO
Capitulo IV
A Espectroscopia de Absorção de Raios-X Aplicada no Estudo
Estrutural de Materiais Óxidos 84
Introdução 85
Estudo da estrutura local em amostras óxidas de composição
AB03 (A=Sr, Pb; B=Ti) preparadas através de métodos químicos 85
Estudo da estrutura local de cerâmicas ferroelétricas 88
Estudo estrutural de cerâmicas ferroelétricas Pb1-xLaxTi03 (PLT) 89
Estudo Estrutural e o Mecanismo de Transição de fase em
amostras do sistema Pb1-xLaxTi03 91
Condusão 95
Referências 95
Artigos: Estudo do processo de cristalização em amostras de
composição AB03 98
Artigos: Estudo estrutural em cerâmicas ferroelétricas de composição
Pb 1-xLaxTi03 98
Capitulo V
Considerações Finais 146

••

V11
CAPITULO 1

INTRODUÇÃO

1
 o objetivo deste texto é de apresentar de forma sistematizada alguns dos trabalhos
que foram realizados na caracterização da estrutura atômica de materiais inorgânicos
através da utilização da Espectroscopia de Absorção de Raios-X (XAS). Essencialmen-
te, o trabalho aborda os resultados obtidos no estudo da estrutura local de materiais
vítreos, amorfos e policristalinos óxidos.
O conhecimento da estrutura a nível atômico de um determinado material, isto é, as
espécies de átomos presentes e sua organização estrutural, têm um papel fundamental
no desenvolvimento científico e tecnológico em diferentes campos da ciência, como na
Física, Química, Biologia e na Ciência dos Materiais. Com o advento da nanotecnologi-
a, a importância das técnicas de caracterização estrutural a nível atômico se tornou
ainda maior.
Quando nos referimos ao estudo estrutural de materiais sólidos inorgânicos, a fer-
ramenta mais comumente utilizada é a Difração de Raios-X (DRX). A técnica de DRX
permitiu, desde o início do século passado, a obtenção de informações estruturais pre-
cisas dos compostos ditos cristalinos. Com o passar do tempo, outras técnicas de ca-
racterização estrutural que utilizam nêutrons ou mesmo elétrons, vieram fazer parte de
um conjunto de técnicas utilizadas na caracterização estrutural desta classe de materi-
ais. Entretanto, a aplicação de técnicas de análise estrutural que necessitam da exis-
tência de ordem a longa distância em sólidos não cristalinos ou "mal cristalizados",
fornece somente informações qualitativas sobre sua estrutura cristalina. Como exemplo
desses materiais podemos citar os materiais amorfos, os quasi-cristais, os nanocristali-
nos e os materiais cristalinos, cuja estrutura tenha sido desordenada pela presença na
matriz de átomos estranhos introduzidos por difusão ou. na forma de impurezas. Este
aspecto de parcial ou não cristalinidade da estrutura torna difícil à tarefa de se obter
informações quantitativas da mesma como as distâncias interatômicas, o número de
primeiros vizinhos e da existência de desordem a nível local. Assim, surgiu a necessi-
dade da utilização de outras técnicas de caracterização que fossem capazes de sobre-
por estas dificuldades.
A partir da década de 70, uma técnica veio preencher em parte a lacuna existente
no estudo estrutural dos materiais "mal cristalizados" ou desordenados. Trata-se da
análise das estruturas finas de absorção de raios-X, conhecida pelo acrônimo XAFS
("X-Ray Absorption Fine Structure") ou XAS ("X-Ray Absorption Spectroscopy"). Devido
às suas características, a técnica de XAS é capaz de fornecer essas informações não

2
JS
acessíveis por outras técnicas, pois: é uma sonda local que não necessita da existência
de ordem a longa distância, permite o estudo de materiais em diversos estados físicos,
1-
é uma sonda seletiva do ponto de vista atômico e apresenta sensibilidade no estudo de
tis
sistemas diluídos.
Apesar da técnica de XAS ter somente apresentado um desenvolvimento significati-
:lS
vo nos últimos 30 anos, o fenômeno físico associado à técnica foi pela primeira vez
ai
observado por volta de 1912 por De Broglie e mais tarde por Fricke e por Kronig [1-3].
Ia
A técnica de XAS deixou de ser uma curiosidade científica para ser considerada como
1'- mais uma ferramenta de caracterização estrutural de materiais no início dos anos se-
tU
tenta com o surgimento de um modelo físico explicando a origem das estruturas finas
de absorção, bem como o surgimento de montagens experimentais mais eficazes liga-
t
r- I• das aos laboratórios de radiação síncrotron [4-6]. A utilização da técnica de XAS permi-
t
l(
m
•I tiu, então, a obtenção de respostas a certos problemas estruturais não completamente
~- ~
~ resolvidos.
i
l- I
~ Nos últimos 30 anos, um esforço muito grande foi feito com o objetivo de aperfeiço-
e
i-
I
I
ar a técnica tanto do ponto de vista experimental
tes de radiação síncrotron teve um papel fundamental
como teórico. O surgimento
neste sentido. Um grande avan-
das fon-

,- I ço também foi observado na teoria que descreve o fenômeno e no desenvolvimento de

novos softwares de simulação. A técnica de XAS tornou-se assim uma ferramenta efi-
caz na determinação da estrutura local de materiais seja ele orgânico ou inorgânico,
amorfo ou cristalino.
Como ocorre com outras técnicas, a técnica de XAS também apresenta algumas
limitações. A impossibilidade de se diferenciar dois átomos vizinhos na tabela periódica
r
(por exemplo os átomos de oxigênio e nitrogênio) e, principalmente, fornecer somente
I informações estruturais de curta distância, são as mais importantes.
Os trabalhos que serão apresentados nesse texto representam as principais etapas

I de nossa contribuição
••
ao estudo estrutural de materiais inorgânicos usando a técnica
de XAS. No Brasil, um grande avanço na utilização desta técnica foi observado a partir
do segundo semestre de 1997 após a entrada em funcionamento do Laboratório Na-
cional de Luz Síncrotron (LNLS), situado em Campinas, S.P. Iniciaremos fazendo uma
revisão detalhada dos aspectos teóricos e experimentais da técnica de XAS. Em segui-
da, apresentaremos os aspectos relativos à medida e os métodos de análise do espec-

3 I F S (. U S P SERViÇOINFORMAÇAO
Uti:ltti~IU r ECA
tro de absorção de raios-X que levam a obtenção de informações estruturais quantitati-
vas de materiais amorfos ou cristalinos.

Referências:

1- M. De Broglie, Comptes rendus 157, 924-926 (1913).

2- H. Fricke, Phys. Rev. 16,202 (1920).
3- R. de L. Kronig, Zeit. Phys. 70, 317 (1931), 75,191 (1932).
4- D. E. Sayers, E. A. Stern and F.W. Lytle, Adv. X-Ray Ana!. 13,248 (1970).
5- D. E. Sayers, E. A. Stern and F.W. Lytle, Phys. Rev. Lett. 30, 174 (1973).
6- F.W. Lytle, D. E. Sayers and E. A. Stern, Phys. Rev. B11, 4836 (1975).

"

4
3ti-

Capitulo 11

Princípios Fundamentais da
Espectroscopia de Absorção de R,"os-X /

01(

5
11.1-Breve Histórico sobre o Fenômeno de Absorção de Raios-X ou ali

f(n), (
A historia da absorção de raios-X pode ser dividida em períodos [1]:

ta co
- Período entre 1915-1933: A descoberta

Alguns artigos de revisão sobre a técnica de XAS fazem uma excelente descrição
sobre este período [2-6]. Apesar de concentrarmos a descrição histórica no período no essE
qual a técnica apresentou avanços mais quantitativos, ou seja, após o ano de 1971, res (
iremos citar as principais etapas do desenvolvimento da técnica, conforme o trabalho
publicado por R. Stumm von Bordwehr [3]: mos
1895: Rõntgen descobre os raios-X [7]; tern
1913: M. De Broglie observa o primeiro espectro de absorção; est»
1915: Publicação do primeiro espectro de absorção de raios-X por Wagner [8]; que
1918: Descoberta das estruturas finas de absorção na borda de absorção M do tório e Entl
do urânio por Stenstrõm; nan
1918: Descoberta das estruturas finas de absorção na borda do cromo e do fósforo por con
Fricke;
1920: De Broglie publica um artigo contendo todos as bordas K dos elementos do
magnésio ao urânio;
1931-1932: Kronig publica as teorias de absorção de raios-X para um material cristali-
no, teoria baseada na ordem a longa-distância (LRO) e para as moléculas, teoria de per
interferência de ondas eletrônicas baseada na ordem a curta distância (SRO) [9-12]. exi
Curiosamente, apesar da afirmação do próprio Kronig de que existia uma identidade ap
entre as duas teorias (a teoria de LRO seria uma extensão para uma quantidade maior vir
de centros ordenados da teoria de SRO), somente a teoria de LRO foi realmente aceita nO'
pela comunidade científica durante muitos anos, até a década de 70 quando uma nova
teoria foi então proposta. - F
1933: Pertersen, estudante de Kronig, desenvolve a teoria molecular e apresenta a se-
guinte expressão para as estruturas finas em moléculas diatômicas [13-15]:
1
de
X(E) -1 = ~ (21 + 1)(-li: se n(8, )sen(2kr + 8,)
'=0 (kr)- Er
D.

"6
 ou ainda, se definirmos o fator de difusão da onda eletrônica pelo átomo vizinho como
1 l~'r
f(n), com f(n)=- L(-1)2(21+1)sen(ol)exp(i81)=lf(n)lexp(i~), esta expressão pode ser escri-
k l~O

ta como:

X(E) -1 = _lIk,.-(~)I sen(2kr +~)

10
Como poderemos constatar mais adiante, a expressão acima contém os termos
10
essenciais para a compreensão da teoria de ordem a curta distância, a menos dos fato-
1, res de amortecimento e do livre caminho médio dos elétrons.
10
Em sua publicação Petersen descreve que '~ distância dos máximos e dos míni-
mos observados no espectro de absorção depende essencialmente das distâncias in-
temucleares de tal forma que seria possível determinar estas distâncias a partir dos
espectros de absorção de raios-X" [13-15]. Assim, no ano de 1933, foi a primeira vez
que o uso da técnica de absorção de raios-X como sonda estrutural foi considerada.
e Entretanto, somente com os trabalhos realizados na década de 70 e o início do funcio-
namento dos centros de radiação síncrotron, esta técnica passou a ser reconhecida
>r como uma nova técnica de caracterização estrutural da ordem a curta distância.

o - Período entre 1933-1969: Um período de estagnação

Nesta época, a maior parte dos físicos que se interessavam pela técnica neste
e período adotaram a teoria da ordem a longa distância (LRO), [3]. Apesar do fato de não
·1· existir um modelo teórico que explicasse a existência das estruturas finas de absorção,
a partir de 1963, F.W. Lytle demonstrava a firme convicção de que esta técnica poderia
r vir a ser uma nova ferramenta na caracterização estrutural de compostos não cristali-
nos [17-19].

- Período entre 1969-1975: Apresentação de um modelo

F.W. Lytle havia batizado já em 1965 as estruturas finas observadas após a borda
dos espectros de absorção de EXAFS (Extended X-ray Absorption Fine Structure) [20].
Em meados de 1970, F.W. Lytle se associou a outros dois pesquisadores americanos,
D.E. Sayers e E.A. Stern, para desenvolver as bases da teoria moderna que explica a

7
origem das oscilações EXAFS [21-25]. Esta teoria foi baseada no que hoje chamamos _I

de aproximação de ondas planas e difusão simples. Estes pesquisadores propuseram

um modelo que rejeitava a teoria da ordem a longa distância para explicar a origem das
estruturas finas de absorção. Segundo este modelo, as estruturas finas de absorção
o
estariam relacionadas à ordem a curta distância próxima ao átomo absorvedor. Além
P
disso, eles mostraram que a utilização da análise de Fourier dos espectros EXAFS po-
d
deria levar a obtenção de informações estruturais como a distância média, número de
d
coordenação e a flutuação nas distâncias. Devido à demonstração da praticidade da
d
técnica de EXAFS e por ter sido proposta uma teoria rigorosa para explicar o fenôme-
no, Lytle, Stern e Sayers são considerados hoje os artesãos da técnica de EXAFS.

E
- 1975: A técnica de EXAFS e os centros de Radiação Síncrotron
(

Um dos motivos, segundo Stum von Bordwer, para que a técnica de EXAFS tenha
ficado um longo período sem apresentar um desenvolvimento que pudesse levar a sua
utilização como ferramenta de caracterização estrutural de materiais, foi a dificuldade
na obtenção de dados experimentais [3]. A realização de medidas experimentais permi-
tiria um confronto entre experiência e teoria. A grande dificuldade na realização de ex-
perimentos advém do fato de que na obtenção de espectros de absorção de raios-X,
devemos utilizar radiação-X branca. A utilização dos tubos de raios-X convencionais
era pouco atrativa pelo fato de ser necessário um tempo de exposição muito longo. A
coleta de espectros de absorção de raios-X para uma grande variedade de compostos
somente ocorreu no início do ano de 1975 com o desenvolvimento de fontes de radia-
ção síncrotron; ainda como um subproduto dos laboratórios construídos para o estudo
da física de partículas.
Assim, o desenvolvimento da técnica de EXAFS a partir de 1975 está também
intimamente ligado a entraCta em funcionamento dos laboratórios de radiação síncro-
trono Os primeiros espectros de absorção de raios-X obtidos em um laboratório síncro-
tron foram obtidos por Kincad e Eisenberger em 1975 [26-28]. Estes dois pesquisado-
res obtiveram espectros de absorção na região dos raios-X "duros" em uma montagem
experimental do laboratório síncrotron de Stanford. As pesquisas desenvolvidas em
Stanford nesta época estavam ligadas ao estudo de materiais catalisadores, a física do
estado sólido e ao estudo de proteínas.

8
10S
- Período entre 1975-1985: Avanços na teoria
dm
fas
Um formalismo mais completo na descrição das oscilações EXAFS, levando em
.ão
consideração a aproximação de ondas esféricas e do efeito de difusão múltipla, foi pro-
ém
posto por C.A. Ashley e S. Doniach [29] e por P.A. Lee e J.B. Pendry [30] em meados
)0-
de 1975. Entretanto, pelo fato dos pesquisadores nesta época estarem mais preocupa-
de
dos em fazer com que a técnica fosse reconhecida no meio científico, esta teoria foi
da
deixada em segundo plano por aproximadamente 10 anos.
le-
i! Um fato importante para se colocar em prática a utilização da técnica de EXAFS
]
i foi à adoção do conceito de transferibilidade. Este conceito admite que os fatores de
J espalhamento atômico utilizado na fórmula padrão que descreve as oscilações EXAFS
I
I dependem somente da natureza dos átomos que foram considerados como isolados.

I
.~,
Assim, o fato dos átomos estarem ligados quimicamente não era levado em considera-
ha
I ção. Esta aproximação, fundamental na análise dos espectros EXAFS, levava segundo
ua
os trabalhos publicados nesta época, a um erro de 0,01A na determinação das distân-
de
cias atômicas. A proposição de uma fórmula simples aliada ao fato da simplicidade na
ni-
obtenção de dados experimentais levou inúmeros grupos de pesquisa ligados ao estu-
~x-
do de materiais a utilização da técnica de EXAFS.
X,
Paralelamente ao desenvolvimento prático da técnica que permitiu um maior
us
confronto entre os dados experimentais e as teorias existentes, ocorreu um grande a-
A
vanço no estudo das estruturas finas observadas na região da borda de absorção, re-
::>s
gião esta denominada de XANES pelo físico italiano A. Bianconi [31]. Nesta época, um
a-
grande debate sobre a origem das estruturas finas na região da borda de absorção foi
jo
iniciado e podemos dizer que a discussão continua até os dias de hoje. Nesse mesmo
período, teve início a organização de uma conferência que passou a ser realizada a
m
cada três anos e que de
<li
certa forma traduziu o desenvolvimento alcançado pela técnica
0-
nos últimos 24 anos [32].
0-

- A partir de 1985: uma melhor compreensão do fenômeno

m
m
Atualmente, a técnica de XAS é umatécnica de caracterização estrutural larga-
10
mente aplicada por um grande número pesquisadores no estudo de uma vasta gama
de materiais. Isto foi possível graças ao avanço observado na descrição teórica do fe-

9
nômeno de absorção de raios-X, bem como a construção de um número significativo teria
de centros de radiação síncrotron em todo o mundo. A caracterização da ordem local raict
através da técnica de XAS permitiu a solução de inúmeros problemas estruturais ante- urru
riormente impossíveis de serem resolvidos. No que tange a teoria envolvida na descri- estr
ção do fenômeno, atualmente, a fórmula de EXAFS, é descrita através de um modelo dos
de ondas esféricas tão simples como o modelo de ondas planas utilizado no início do par
desenvolvimento da técnica. O fenômeno de difusão múltipla que havia sido inicial- bra
mente discutido em 1975, é tratado hoje como parte integrante tanto na análise dos um
espectros EXAFS como dos espectros XANES. ma
Podemos dizer que o maior obstáculo na obtenção de informações estruturais é
a necessidade de um modelo teórico preciso ou de um espectro de um composto pa-
drão para ser utilizado na simulação dos espectros EXAFS. O uso de um modelo teóri-
co é hoje preterido pela maioria dos pesquisadores que atuam na área pelo motivo de
"
que estes modelos apresentam a mesma qualidade que um modelo experimental [33-
36]. Mais recentemente, a obtenção de informações estruturais a maiores distâncias (>
5 A) tornou-se possível graças ao desenvolvimento de uma teoria mais geral que es-
tendeu o intervalo de análise dos espectros EXAFS até 20-30 eV após a borda de ab-
sorção, ou seja, na região onde os fenômenos de difusão múltipla são importantes [37-
38].

- 1997: Início das experiências de absorção de raios-X no Brasil

As primeiras medidas de absorção de raios-X no Brasil foram realizadas no LN-

LS (Laboratório Nacional de luz Síncrotron) a partir de julho de 1997. A montagem da
primeira linha de XAS no LNLS nasceu de um esforço de muitos pesquisadores brasi-
leiros interessados na utilização da técnica como mais uma ferramenta de caracteriza-
ção estrutural. A construção tia primeira linha de XAS no LNLS foi possível graças a
um projeto temático concedido pela FAPESP cuja coordenação coube ao Prof. Osvaldo
L. Alves do Instituto de Química da UNICAMP. Além do suporte financeiro, imprescin-
dível para a construção da linha, a qualidade dos profissionais do LNLS envolvidos em
sua construção foi decisiva para que o objetivo fosse alcançado. Minhas primeiras ex-
periências de absorção de raios-X no LNLS foram realizadas já em 1997 quando sub-
metemos dois projetos propondo o estudo da estrutural local de materiais vítreos e ma-

10
ivo teriais óxidos cerâmicos. Hoje, o LNLS já conta com outras duas linhas de absorção de
581 raios-X, uma linha de absorção de raios-X "moles" (energias abaixo de 2,4 KeV) e com
te- uma linha de espectroscopia de absorção de raios-X dispersiva voltada para estudos
~ri- estruturais "in- situ". A alta demanda para o uso da linha de XAS do LNLS (400 perío-
slo dos de 12 horas foram solicitados em 2002), demonstra que a técnica de EXAFS já faz
do parte do conjunto de técnicas de caracterização experimental de muitos pesquisadores
al- brasileiros [39]. A construção de uma segunda linha de XAS pelo LNLS indica que em
os um futuro próximo o número de pesquisadores que a utilizam poderá crescer ainda
mais [40].

a-
ri-
~

3-
(>
s-
J-

I.:.
a
i-
[-

11 SERVI(,:ÜUtb[b~IOrCcl~
IFSC • USp INFORMAÇAO
11.2-O Espectro de Absorção de Raios-X

Em um processo de absorção, um fóton interage com a matéria seguindo diversos

processos como o efeito fotoelétrico, o efeito Compton e a produção de pares [41]. Na
região dos raios-X, aproximadamente de 300 eV a 30 KeV, o efeito com maior probabi-

lidade de ser observado é o efeito fotoelétrico, onde um fóton de energia E=hv é total-

mente absorvido por um átomo e um elétron é emitido com uma certa energia cinética
[41]. A esse elétron emitido damos o nome de fotoelétron. A energia cinética do fotoelé-
tron (Ec) é dada por
Ec =E-E o (1)

onde Eo é a energia mínima necessária para arrancar um elétron do átomo. No caso

dos raios-X, essa energia é suficiente para arrancar um elétron de níveis profundos.
A principal característica de um espectro de absorção de raios-X, em função da
energia, é a existência de descontinuidades no coeficiente de absorção. Estas descon-
tinuidades, chamadas de bordas de absorção, são características de cada elemento e
ocorrem quando a energia de um fóton é suficiente para promover um elétron de um
nível profundo do átomo para um nível não ocupado.
Quando um feixe de raios-X passa através da matéria, a intensidade incidente I

diminui de uma quantidade que é determinada pela absorção característica do material

que está sendo irradiado. Para um material de espessura dx, a diminuição de intensi-
dade dI é descrita como:
dI = -J1(E)Idx (2)

onde j.l(E) é o coeficiente de absorção linear que depende da energia do fóton inciden-

te. A integração da equação (2) sobre a espessura total x leva a equação conhecida
como Lei de Lambert:
It ;"Io «r"> (3)

À medida que a energia dos fótons de raios-X incidente aumenta, observa-se na

região da borda de absorção um aumento abrupto do coeficiente de absorção de raios-
X. Este aumento abrupto, que é seguido pela diminuição monotônica do coeficiente de
absorção, pode ou não conter oscilações-que são amortecidas à medida que a energia
aumenta. A Figura 1 apresenta a variação do coeficiente de absorção linear j.l em fun-

12
ção da energia dos fótons incidentes na borda K do átomo de cobre para uma amostra (1

de cobre metálico. íi

""E 0,6
XANES =EX=A=F~S~ _
,
,.o)
r
~
x 0,6
ch
o
.~
{l 0,4
o
~
~ 0,2
lU
Gl
'O

li 0,0 pré-borda
c -
'õ
s -0,2+---,.--......----,.-......----,.-......----,.-......-----,---1
""
o
8800 9000 9200 9400 9600 9800
Energia dos f6tons incidentes (eV)

Figura 1- Medida do coeficiente de absorção de raios-X na borda K do átomo de cobre

para uma amostra de cobre metálico.

Historicamente, na análise do espectro de absorção, podemos dividi-Io em três

regiões [1]:

a região (1) denominada de pré-borda, está localizada entre 2 e 10 eV abaixo da

borda de absorção. A região de pré-borda pode apresentar estruturas de absorção de
raios-X devido a transições eletrônicas de níveis internos (1s, 2s, etc) para níveis ex-'
ternos parcialmente preenchidos ou não preenchidos. A probabilidade de transição e a
intensidade da absorção observadas na região de pré-borda são determinadas em par-
te pela simetria do sítio do átomo absorvedor. A medida da posição e da intensidade da
absorção presente na região da pré-borda pode fornecer informações sobre o estado
de oxidação e sobre a simetria do sítio do átomo absorvedor.
a região (2), a qual pode se estender até aproximadamente 50 eV após a borda de
absorção, é denominada de XANES (X-ray Absorption Near-Edge Strutucture) ou NE-
XAFS (Near-Edge X-Ray Absorption Fine Structure). Esta região do espectro de absor-
ção é normalmente caracterizada pela presença de estruturas de absorção bem defini-
das quando comparadas às oscilações observadas na região denominada de EXAFS

13
stra (região 3). Estas estruturas têm como origem os fenômenos de espalhamento múltiplo
dos fotoelétrons nos átomos vizinhos do átomo absorvedor. A Figura 2 ilustra o proces-
so de espalhamento múltiplo que ocorre na região de energia do espectro XANES. Este
processo de espalhamento envolve centros espalhadores que estão localizados fora da
primeira esfera de coordenação. Devido a este fato, a região XANES é particularmente
sensível a mudanças no arranjo geométrico nos átomos vizinhos mais próximos do á-
tomo absorvedor, podendo ser utilizada na obtenção de informações sobre a simetria
do sítio e da estrutura a média distância.

bre

Figura 2- Ilustração esquemática do processo de espalhamento múltiplo observado na

região de XANES [2].
rês

A região (3), a qual pode se estender de aproximadamente 50 eVa 1000 eVapós a

borda de absorção, é denominada de EXAFS (Extended X-Ray Absorption Fine Struc-
da
ture). A região de EXAFS é caracterizada pela presença de oscilações de baixa inten-
de
sidade e baixa freqüência quando comparada a região de XANES. As oscilações E-
~x-
XAFS originam-se do processo de interferência construtiva e destrutiva entre o fotoelé-
~a
tron emitido pelo átomo absorvedor e o fotoelétron retro-espalhado pelos primeiros á-
:tr-
tomos vizinhos. Estas oscilações têm origem principalmente em processos de difusão
ja
simples (Figura 3). A análise do espectro EXAFS pode fornecer informações sobre a
jo Il

distância « 5A), o número e o tipo de átomos vizinhos localizados ao redor do átomo

absorvedor. Conforme ilustra a Figura 3, a freqüência das oscilações EXAFS depende
je
da distância média átomo absorvedor-átomo espalhador, enquanto a amplitude das
E-
oscilações depende do número de átomos espalhadores.

li-
S

14
 ~
c, x
t~
/

, .J
It
(bJ
t
~

....J
\ ;,

{c I
,
/* ,
~

-,
-

PHOíON ENERC,"(

(a) (b)
Figura 3- (a) Ilustração esquemática do processo de espalhamento simples observado
na região de EXAFS e (b) mudanças observadas na freqüência e na amplitude das os-
cilações EXAFS quando da variação da distância e do número de primeiros vizinhos
[41].

o grande mérito dos pesquisadores Stern, Sayers e Lytle foi ter obtido uma rela-
ção entre as oscilações observadas após a borda de absorção e a existência de uma
ordem local bem definida em torno do átomo absorvedor [21-25]. A hipótese feita por
estes pesquisadores foi confirmada quando o espectro EXAFS de uma amostra de gás
de neônio, onde somente espécies monoatômicas estão presentes, foi comparada com r

O de uma amostra de neônio no estado sólido, onde os átomos estão presentes em

posições bem definidas (Figura 4).

Figura 4- Espectro de Absor-

solid ção de raios- X de uma amos-
~gas
tra de neônio na forma gasosa
e na forma sólida cristalina
/~ [42].

_--' .l __ •• _ J .•• . •. A. ~

850 900 9~ 1000 E (eV)

15
 Na amostra na forma de um gás, nenhuma oscilação foi observada pois os áto-
mos encontram-se em posições aleatórias. Por outro lado, na amostra sólida cristalina,
as oscilações EXAFS foram então observadas uma vez que cada átomo apresenta ou-
tros átomos vizinhos em posições bem definidas.
A fórmula matemática que desenvolveremos a seguir, utilizada na obtenção de
informações estruturais de forma quantitativa das oscilações EXAFS, contém algumas
simplificações que serão introduzidas à medida que os cálculos forem sendo realiza-
dos. Uma versão mais completa da obtenção da equação de EXAFS pode ser encon-
trada na referência [43]. Entretanto, apesar destas simplificações, chegaremos a equa-
ção de EXAFS atualmente utilizada na grande maioria de programas de análise de da-
dos.
do

11.3- A Equação de EXAFS

10S

o processo da formação das oscilações observadas logo após a borda de ab-

sorção têm como origem um fenômeno de interferência entre a função de onda de um
la-
fotoelétron emitido de um átomo e as funções de onda retroespalhadas pelos átomos
na
vizinhos ao átomo que foi excitado (Figura 5).
>ar
tas

~m

Figura 5- Processo de interferência entre a onda associada ao fotoelétron emitido pelo

átomo central (C) e pelas ondas retroespalhadas pelos átomos vizinhos (V).

Na região de baixa energia (XANES), os fotoelétrons se propagam com um livre

percurso médio importante dando origem a efeitos de difusão múltipla (Figura 3). Por
outro lado, na região de energia característica das oscilações EXAFS, o livre caminho

16
médio dos fotoelétrons é limitado e somente os efeitos de espalhamento simples ou
espalhamento sobre um número limitado de átomos vizinhos apresentam uma contribu- [

o
ição significativa à amplitude do sinal.
Se 11 é o coeficiente de absorção experimental e J.1O o valor hipotético do coefici-
ente de absorção atômico, as oscilações EXAFS podem ser definidas como:

X = Jl- Jlo (4)

110

o coeficiente de absorção Jl pode ser calculado através da Regra de Ouro de

Fermi [44]:
c
u= :L1(iIHlft8(E -Er-hm) j (5)
f
r
2
Jl oc I(i IH 1/)1 n(E) (6)

onde n (E) é a densidade de estado do fotoelétron de energia E, H é o Hamiltoniano ~

que descreve o processo de absorção dipolar elétrica, (ij e Ir) são, respectivamente, o

estado inicial (1s no caso da borda de absorção K) e estado final do fotoelétron. f

Como a função de onda (il é localizada sobre o átomo central, a contribuição

principal da função de onda Ir) será próxima ao átomo central, ou seja, próxima a ori-

gem. Na ausência de átomos vizinhos, o estado final é dado pela função Ifo). A pre-

sença de átomos vizinhos induz uma perturbação laf). Neste caso, a função If) pode

ser escrita como:

If) = 110)+ laf) (7)

A absorção atômica que corresponde ao estado final sem a presença de átomos

vizinhos é dada por:

(8)

A absorção total é dada por:

(9)

Fazendo uma expansão de segunda ordem na equação (9) temos que:

(10)

17
ou
Se considerarmos Iai) como uma perturbação pequena, podemosdesconsiderar

o termo de segunda ordem. Então,

ci- (11 )

A partir destas equações podemos escrever a função X como:

X = J-l- J-lo = 2R ((iIH Iai) = 2R a (12)

J-lo e (i IH I10) e

de
Para determinar a expressão a qual permite simular os espectros EXAFS, é ne-
cessário calcular a na equação (12). Para isso, vamos seguir passo a passo os proces-
sos físicos que levam a modificação da função de onda do fotoelétron. Para simplificar
nosso procedimento, vamos considerar o processo de transição em uma borda de ab-
sorção K de uma molécula diatômica (um único átomo como vizinho). Iremos admitir
que para uma molécula mais complexa, o sinal EXAFS pode ser obtido através da adi-
no ção dos sinais individuais de cada átomo [45].
,o Inicialmente, o fotoelétron é emitido do átomo absorvedor (átomo central). Sua

função de onda é aquela que corresponde a absorção atômica If) =Ifo).

ão
R
)ri-
Átomo cen- Átomo vi-
re-
tral zinho
de O fotoelétron se propaga na direção do átomo vizinho. A função de onda do fotoe-
létron, no caso da borda de absorção K (níveI1s), é descrita através de uma função de
Hankel [46]:

10S R

, c
Atomo Átomo
central vizinho

(13)

A solução da equação de Schrodinger de uma partícula livre em coordenadas es-

féricas é dada pelo produto de um harmônico esférico por uma função de onda de
Hankel esférica [46]:

18
 ik(R-x)

I
h;(kR) = (ii ht(kR)Yz,m(9,<p)ai) =l/o)ht(kR) e _
R x
1(9,k) (14)

o fotoelétron é espalhado pelo potencial dos átomos vizinhos. A expressão

completa da função de onda após espalhamento contém, segundo a teoria de espa-
Ihamento por um potencial central, a função de onda incidente e um termo correspon-
dente a uma função de onda esférica emitida pelo átomo espalhador multiplicada por
um fator de difusão atômica f(B,k) [47]. Como somente os termos localizados sobre o
átomo central tem importância, vamos considerar somente o processo de difusão na
direção do átomo absorvedor, representada pelo termo de retroespalhamento comple-
xo f(1l,k).

Átomo Átomo
central vizinho
Se escolhermos então a origem no átomo central, temos que:
ik(R-x)

Iai) = lio)Jf(kR) e
R-x
1(1C,k) (15)

Quando consideramos que a onda retroespalhada atua somente na origem, ou

seja, nas vizinhanças do átomo absorvedor, temos que x=O, e a equação (15) pode
então ser escrita como:

Átomo Átomo
central vizinho

ileR

Iai) = lio)~+(kR)/(1C,k)~ R
(16)

A onda penetra no potencial do átomo central e sofre uma defasagem <51 (1= 1

pois se trata de uma onda representando o nível p).

(17)

Átomo Átomo
central 19 vizinho
 Para que o fenômeno seja completo, ou seja, para que ocorra a absorção, é ne-
cessário que um elétron seja realmente emitido: o processo é finalizado com uma se-
.são
gunda emissão do fotoelétron do átomo central gerando um segundo termo de defasa-
spa-
gem 51.
ion-
ikR

por laf) = Ifo)ht(kR)f(n,k) eR ei81ei81 (18)

re o
I na
R
ple-

Átomo Átomo
central vizinho

Se fizermos a hipótese de que nosso fóton não é polarisado, a função de Hankel

pode ser escrita como:
·kR
~+(kR)~~ (19)
2kR
Duas outras aproximações que são feitas é que na distância R onde o átomo
vizinho está localizado, a onda pode ser assumida como sendo uma onda plana se o
ou
diâmetro do átomo é pequeno comparado com R e o potencial do átomo central não se
xfe
estende até o átomo retroespalhador. Assim, iremos desconsiderar a curvatura da onda
que chega ao átomo vizinho, mesmo que ela seja na realidade esférica. A função de
onda emitida de momento angular 1=1 é assim substituída pela sua forma assintótica,
que é uma função de onda esférica de momento angular 1=0 [45].
Para uma onda plana incidente, a probabilidade de difusão em um determinado
ângulo () é dada pela função de difusão:
I 00

lJ(e,k)/ = - 2:(21 + l)exp(iol)sen(ol)~(cOSe) (20)

•• k 1=0
:=1
Como se trata de um processo de retroespalhamento, devemos fazer com que
e seja igual a n. Esta parte da onda será dada pelo modulo da função de retrodifusão
If(n,k)/. A defasagem causada pelo potencial do átomo retroespalhador será denotada

~ (onde ~ depende de k). Podemos então escrever a equação (20) da seguinte forma:
f(n,k) = If(n,k)/ei+<lt,k) (21)

20
onde f(n,k) é a defasagem total causada do átomo vizinho e lf(n,k)1 é a amplitude de c

retroespalhamento. (
A função de onda do fotoelétron após ser retroespalhada pelo átomo vizinho se-
rá dada então por:

Iai) = 110) 2:R2 /f(n,k)lei(2kR+2o/+~) (22)

c
Combinado às equações (18), (19) e (22), temos que: (

a = i/f(n,k)1 ei[2kR+2o/+<P(1t,k)]
2kR2 E
Assim, X(k) pode ser escrito como:

k
X(k) = If(n'2 )1 sen[2kR + 201 + 4>(n,k)] (23)
kR
A esta equação, alguns outros termos devem ser adicionados a fim de levar em
consideração alguns efeitos que ocorrem durante o processo de absorção em uma es- 1
trutura composta de um conjunto de átomos. t.
n
- Processo de amortecimento: perdas inelásticas e

A equação (23) foi desenvolvida supondo que o elétron emitido sofre processos ç
elásticos. Entretanto, quando da análise do caso real, podem ocorrer perdas devido a e
processos inelásticos. Os processos inelásticos podem ser classificados em duas cate- n
gorias [41]: s
(1) processos que contribuem com o aumento da intensidade da borda de absorção fi

sem que ocorra produção de um fotoelétron. Neste caso, trata-se de um processo mul- n
tieletrônico do tipo "shake-up" e shake-off" (48]. Uma vez que é realizada uma normali- q
zação por Ilo, todo fenômeno que contribui na intensidade da borda de absorção sem P
••
contribuir para o sinal EXAFS leva a uma diminuição do valor de X. Este efeito é levado d

em consideração através de um termo de redução S~ que é introduzido na equação

d
te
(23). Este termo vale aproximadamente 0,9 para todos os materiais (49J.
d
(2) Perdas inelásticas do fotoelétron "EXAFS" na matéria. A equação (23) supõe que
p
todos os fotoelétrons são retroespalhados sem perda de coerência. Para que o efeito
n

21
de de perdas elásticas seja levado em consideração, nos devemos multiplicar a equação
(23) por um termo relacionado ao livre percurso médio do fotoelétron:
se- 2R
exp[--] (24)
À(k)

o efeito do espalhamento inelástico é o amortecimento da onda do fotoelétron,

que tem como resultado um livre caminho médio relativamente pequeno do fotoelétron
(tipicamente da ordem de 6 A). No intervalo de energia das oscilações EXAFS, a vari-
ação de À é praticamente linear com a energia. Assim, podemos dizer que a técnica de
EXAFS é uma sonda local [50].
~i

Efeitos de desordem

I
bm Pelo fato do tempo médio de vida em uma transição de raios-X ser da ordem de
3S- 10-16S, podemos dizer que a técnica de EXAFS fornece uma "fotografia" quase que ins-
tantânea da estrutura local [41]. Este valor é pequeno quando comparado à vibração
molecular na ordem de 10-12S. A medida de EXAFS constitui então numa integração
espacial de posições instantâneas do átomo central (absorvedor) e dos átomos retro-
espalhadores (vizinhos). Esta integração inclui todos os fatores que afetam a distribui-
.os ção de distâncias interatômicas e podem ser discutidas em termos de uma desordem
)a estática e térmica. A desordem estática refere-se a distribuição de distâncias interatô-
te- micas dentro de uma dada camada, isto é, os primeiros vizinhos mais próximos. A de-
sordem estática é considerada grande se a diferença de distâncias for comparável a
ão resolução de EXAFS. Este conceito é importante uma vez que cada distância interatô-
ul- mica dará origem a uma freqüência EXAFS. Se as distâncias interatômicas ou as fre-
lIi- qüências apresentam valores que estão em um certo intervalo, uma diminuição de am-
~m plitude das oscilações EXAFS será observada. Se o intervalo de distâncias para um
O!

do determinado átomo for pequeno comparado com a resolução de EXAFS, será observa-

ão da uma adição das diferentes freqüências e assim, a análise das oscilações EXAFS
fornecerá uma informação sobre uma distância interatômica média. Se o intervalo de
distâncias é grande comparado com a resolução de EXAFS, as diferentes freqüências
Je
podem ser separadas e resolvidas dando origem a diferentes distâncias interatômicas
ito
na análise do espectro EXAFS.

22
 Se a distribuição de distâncias é Gaussiana, a freqüência média das oscilações
dará uma boa estimativa da distância interatômica média. Se a distribuição de distân-
cias varia significativamente de uma distribuição Gaussiana, as informações obtidas
n
poderão conter erros apreciáveis. n
Por outro lado, no caso da desordem térmica, na maioria dos casos, a distribui-
ção Gaussiana é um bom modelo, exceto para medidas feitas em altas temperaturas,
onde efeitos de não harmônicos podem ocorrer. Assim, na análise dos espectros E- (

XAFS, a distribuição de distâncias de uma determinada esfera de coordenação é des-

crita por uma função de distribuição Gaussiana.
(

O termo usual que considera os efeitos de desordem na amplitude do sinal E-

XAFS é o termo de Debye-Waller obtido em uma aproximação harmônica. Este termo
assume a existência de vibrações harmônicas e uma distribuição Gaussiana de distân-
cias. O termo similar utilizado na análise dos dados difração de raios-X descreve o
-}
desvio do átomo (considerado independente) de sua posição média enquanto que no
caso de EXAFS, o fator de Debye-Waller representa a distribuição de distâncias entre o
átomo absorvedor (átomo central) e o átomo espalhador (átomo vizinho) relativamente
ao átomo central.
O fator de Debye-Waller no caso de EXAFS, utilizado na simulação dos efeitos
de desordem, é constituído então por dois termos:
222
O' = O' estatico + (J' temüco (25)

Em princípio, é possível determinar a magnitude da contribuição do termo de-

pendente da temperatura através da realização de medidas em diferentes temperatu-
ras.
O termo dependente da desordem estática pode ser descrito por uma função
simétrica de distribuição de pares Gaussiana:

•.g(r) = cr-1(21tr1l2 exp [-(r -ro)2] (26)

20'2

onde (r-ro) corresponde ao desvio da distância média ro. A desordem estrutural estática
para cada ligação pode ser então escrita como:
n (2 2)
2 _'" rj -ro (27)
O'est. - L..J
j=1 n
O tempo característico das vibrações térmicas (t....10-13_10-14) é muito superior a
todos os outros tempos envolvidos na medida. Assim, como comentamos anteriormen- t v

23
te, o fotoelétron tem uma visão instantânea do arranjo atômico em torno do átomo ab-
sorvedor. A desordem térmica é vista então como um efeito de desordem estrutural, da
mesma maneira que a desordem estática. Se os átomos estão vibrando harmonica-
mente, essa contribuição do fator do Debye-Waller pode ser escrita como:
h hv
o
term.
= 8n m
2 v
coth-
2kT
(28)
r

onde rn, é a massa reduzida, T é a temperatura em graus Kelvin, k é a constante de

Boltzman e v a freqüência de vibração. Esta expressão mostra que com a diminuição
da temperatura ocorre uma diminuição exponencial do fator relacionado com as vibra-
ções térmicas. Em alguns casos pode ser interessante realizar medidas à baixa tempe-
ratura com a finalidade de diminuir a contribuição térmica do fator de Debye-Waller.
Para sistemas apresentando uma desordem considerada pequena (c <0,1A), é
possível expressar o termo de desordem pela expressão:

cr=exp(-2cr2e) (29)

Esta aproximação não é valida quando a desordem estática é maior que 0.1 A
devido à uma distribuição assimétrica de pares ou a presença de potenciais não har-
mônicos de vibração [18]. Neste caso, uma função assimétrica de distribuição de pares
deve ser utilizada. Uma discussão mais detalhada sobre este assunto bem como o uso
do modelo de Einstein de vibrações da rede pode ser encontrado na referência [43].

11.4-A equação completa que descreve as oscilações EXAFS

Finalmente, em uma amostra composta por mais que dois átomos, se admitir-
mos que os sinais EXAFS são aditivos e que somente os termos de difusão simples
são considerados, precisamos então fazer uma soma levando em consideração a pre-
sença de todos os vizinhos do átomo absorvedor [45]. Esta soma é feita sobre cada
esfera de coordenação j contendo Nj átomos caracterizada como:
- uma mesma espécie atômica;
- localizada a uma distância média Rj do átomo central;
- um mesmo fator de Debye-Waller;

- uma mesma amplitude de retrodifusão If(1t,k)l,

- uma mesma defasagem cpj(k).

24
 A equação que descreve as oscilações EXAFS pode ser então escrita como:

(30)

onde:
k é o vetor de onda;
X é a parte oscilatória normalizada do coeficiente de absorção após a borda de absor-
ção;

s; é o termo de redução da amplitude devido a efeitos multieletrônicos,

Nj é o número de vizinhos do tipo j situados a uma distância Rj

V(1t,k)1 é a amplitude de retroespalhamento dos átomos situados a uma distância j;

c/ é o fator de Debye-Waller em uma aproximação Gaussiana;

I(k) é o livre caminho médio do fotoelétron;
Õ1 é o termo de defasagem causado pelo átomo central;
Fi
~j(k) é o termo de defasagem causado pelo átomo vizinho a partir do fator de espalha-
mento atômico dos átomos vizinhos.
e

Apesar de que em princípio esta equação contém somente os termos de difusão

simples, é possível mostrar que ela pode também levar em consideração os efeitos de
ql
difusão múltipla. Na verdade, é esta a fórmula utilizada pelo programa FEFF a partir do • J

dE
qual espectros teóricos de EXAFS contendo efeitos de difusão simples e múltipla po-
aI
dem ser obtidos [38].
C(
di
11.5-O fenômeno de espalhamento múltiplo na análise do espectro EXAFS E.
P(
Um avanço importante na análise dos espectros EXAFS ocorrido na última dé-
P(
cada foi o desenvolvimento de uma teoria que tratava o fenômeno de difusão múltipla a
re
partir da análise dos espectros EXAFS [38]. Uma das limitações na teoria dos espec-
pl
tros EXAFS proposta inicialmente adivinha do fato de somente considerar os efeitos de
Pl
espalhamento simples, ou seja, o espalhamento devido aos átomos considerados co-
ul
mo primeiros vizinhos diretamente de volta para o átomo absorvedor. Apesar deste tipo
SI
de espalhamento ser usualmente o mais intenso, sabemos que o fotoelétron pode ser

25
espalhado por outros átomos antes de voltar ao átomo absorvedor, como ilustra a Figu-
ra 6.

(a)

a"
~S3 (c)
A ••

Figura 6- Exemplos de processos de espalhamento simples e múltiplos. (a) Espalha-

mento simples de átomos situados a diferentes distâncias do átomo absorvedor (A). (b)
e (c) diferentes configurações de espalhamento múltiplo a partir do átomo absorvedor.

A grande contribuição do trabalho de Sayers e colaboradores foi o de mostrar

que através do uso da transformada de Fourier era possível filtrar a partir do espectro
de absorção as contribuições das diferentes camadas atômicas. Assim, no caso de um
espalhamento simples, a fórmula proposta por Sayers e colaboradores é considerada
correta. Entretanto, para obtermos informações estruturais de átomos situados a longa
distância, os fenômenos de difusão múltipla tiveram que ser incluídos na equação de
EXAFS inicialmente proposta. A teoria deespalhamento múltiplo tem sido estudada
por um grande número de pesquisadores. Os avanços mais signficativos foram obtidos
••
pelo grupo do Prof. John Rehr da Universidade de Washington [35]. Este grupo foi o
responsável pela criação de um algorítimo denominado FEFF, o qual permite obter es-
pectros EXAFS teóricos onde todos os fenômenos de espalhamento múltiplo existentes
podem ser visualizados. Hoje, graças a estes avanços, a análise dos espectros EXAFS
utilizando funções de fase e amplitude teórica é feita rotineiramente. A equação (30) é
aquela utilizada na obtenção do espectro EXAFS pelo programa FEFF [35].

26
11.6- Análise das Estruturas Finas na borda de Absorção: XANES Cat
mat
Como foi citado anteriormente, o espectro da borda de absorção (XANES) pode
ser dividido em duas regiões. A análise da região da pré-borda pode fornecer informa- o fe
ções qualitativas importantes sobre o estado de oxidação e sobre a simetria do sítio do
átomo absorvedor, enquanto que a análise da região logo após a borda pode fornecer 11.7-
informações estruturais a curta e a média distância. Isso se deve ao fato de que os fe-
nômenos de espalhamento múltiplo, observados nesta faixa de energia, envolvem .a
participação dos átomos situados além da primeira esfera de coordenação. sua
A análise da região da pré-borda tem sido utilizada por muitos pesquisadores na terv
obtenção de informações estruturais e eletrônicas, principalmente no estudo de metais
de transição. A região de pré-borda de absorção K da maioria dos elementos de transi- de
.'.J
ção é caracterizada por apresentar estruturas de absorção as quais estão relacionadas tror
a transições para níveis internos do átomo. As estruturas observadas na região da pré- gru
borda nos elementos de transição dependem da geometria e das distâncias de ligação tub
com os átomos vizinhos, bem como do estado de oxidação do átomo absorvedor. Tem tes
sido observado que quanto maior o grau de desordem ao redor do átomo absorvedor, con
maior será a intensidade das estruturas na região da pré-borda. Para um mesmo átomo ces
com o mesmo número de oxidação, a intensidade e a posição da estrutura da pré- ene
borda pode variar com o número de coordenação do átomo absorvedor [51-52].
'J
Quando o estado de oxidação de um elemento aumenta, observa-se um desloca- 1- (
mento da borda de absorção para altas energias. Para uma diferença de urnaunidade 2- (
no estado de oxidação de um elemento em uma dada coordenação, um aumento de 3- J
5eV pode ser observado na posição da borda de absorção. Alguns dos fatores que 4- I
contribuem para este deslocamento são as distâncias interatômicas, as quais diminuem 5- J
com o aumento do estado de oxidação para átomos em uma mesma geometria, ou o
'"
estado eletrônico do átomo absorvedor.
A região de XANES, ou seja, aquela localizada logo após a borda de absorção çãe
tem sido alvo de inúmeros estudos [53]. A análise da região XANES do espectro de mo
absorção tornou-se mais acessível após 1991 com os avanços introduzidos no forma- mo
lismo de espalhamento múltiplo pelo grupo do Prof. Rehr. Hoje, muitos exemplos po- cor
rem 'ser"encõntrados na literatura sobre a obtenção de informações estruturais a curta p~f

~ a média distância a partir da simulação da região XANES do espectro de absorção. m;;:\

27
Cabe salientar entretanto que a obtenção do espectro XANES teórico está limitada a
materiais onde a estrutura cristalográfica, ou seja, seu grupo pontual, seja conhecido.
Uma revisão de toda a teoria envolvida no calculo do espectro XANES envolvendo
o fenômeno de difusão múltipla pode ser encontrado nas referências [53-56].

11.7-Obtenção de um espectro de absorção de raios-X

A obtenção de um espectro de absorção de raios-X de alta qualidade exige que

sua medida seja feita utilizando uma fonte de raios-X intensa e que cubra um largo in-
tervalo de energia [1].
A fonte de radiação que fornece raios-X de alta intensidade e em uma larga gama
de energias é aquela produzida por um anel de armazenamento de elétrons ou pósi-
trons conhecidos como fontes de radiação sincrotron. Apesar de existirem diversos
grupos de pesquisa que realizam suas medidas de XAS em laboratório utilizando um
tubo convencional, a grande maioria dos experimentos é atualmente realizada com fon-
tes de radiação síncrotron. Os principais fatores que limitam o uso de tubos de raios-X
convencionais nas medidas de XAS são a sua baixa intensidade e o longo tempo ne-
cessário para a coleta de dados de XAS de boa qualidade. Além disso, o espectro de
energia disponível em um tubo convencional é ainda muito limitado.
As principais características dos laboratórios de radiação síncrotron são [41]:
1- Grande intervalo espectral
2- Excelente coerência espacial
3- Polarização bem definida
4- Estabilidade
5- Alta intensidade

A rnonocrornatizaõão do feixe de raios-X policromático gerado pela fonte de radia-

ção que incidirá sobre a amostra é essencial na obtenção do espectro de XAS. Essa
monocromatização é obtida através do uso de um monocromador. O modelo básico de
monocromador utilizado em experiências de XAS utiliza dois cristais paralelos ambos
cortados na mesma direção cristalográfica ou planos hkl. O primeiro cristal é utilizado
para monocromatizar o feixe de raios-X enquanto o segundo cristal é utilizado para
manter o feixe de raios-X de saída paralelo ao feixe incidente. A escala de energia é

28
determinada pela rotação dos cristais. Um segundo modelo, denominado de monocro- 11.
mador tipo "Channel-Cut", é construido a partir de um bloco do material, ou seja, am-
bos os cristais formam uma só peça. A principal desvantagem deste segundo tipo de
monocromador é a presença de harmônicos para feixes de raios-x de altas energias ré

originadas nos processos de difração de planos cristalográficos de altas ordens, que n

podem afetar as medidas de XAS. Assim, a medida do possível, os monocromadores )(

formados por dois cristais independentes devem ser utilizados. Neste caso, os harmô- a
nicos podem ser eliminados fazendo com que o segundo cristal saia de sua posição de tI
maior intensidade. )

O esquema de uma montagem experimental convencional utilizada na medida ç

do espectro de absorção de raios-X é apresentado na Figura 7. ,
.~.

q;

~ C
FENDAS
MONOCROMAOOR
CÂMARAS DE IONIZAÇÂO ~.
r:

E1 D l (
JANELA DE

BE~L~ -1hI:L _J_

FENDA / \
l
[ r
_ ~ __ ~~ ...----,AMOST.-RA ---. ~
~I ~
~
~ (

/ (

PROTEÇÂO
RADIOLóGICA
/
PROTEÇÂO
RADIOLÓGICA I
Figura 7- Esquema de uma linha de XAS no modo de detecção por transmissão.

Após ser emitida pelo anel de armazenamento, a radiação-X passa por uma janela
de berílio que elimina a radiação ultravioleta oriunda do anel. Um conjunto de fendas
antes e após o monocromador permite ajustar as dimensões do feixe às dimensões da
amostra bem como obter uma melhor resolução em energia. Antes da amostra, temos
um detector que normalmente é uma câmara de ionização, cuja função é medir a inten-
sidade do feixe incidente. Após passar pela amostra, a medida da intensidade do feixe
transmitido (no caso de uma medida por transmissão) é feita por uma segunda câmara
de ionização. Em alguns casos, podemos ter uma terceira câmara de ionização que
pode ser utilizada na medida simultânea de uma amostra de referência. Toda a experi-
ência bem como o monocromador é controlado por um computador sem que seja ne-
cessária a interferência do usuário durante a realização da medida.

29
 11.8-Métodos de medida do coeficiente de absorção de raios-X

Diferentes métodos foram desenvolvidos para medir os espectros de absorção de

raios-X. O método mais simples, conhecido como método por transmissão, consiste na
medida simultânea da intensidade do feixe de raios-X incidente (10)e do feixe de raios-
X transmitido (I) que atravessa uma amostra de espessura x. A homogeneidade da
amostra é um fator fundamental nas medidas por transmissão pois pode afetar a ampli-
tude das oscilações EXAFS. Medidas por transmissão são indicadas para estudos por
XAS envolvendo elementos com número atômico maior que 16 e em altas concentra-

i ções (> 4%).

I
I
Uma outra maneira de se medir um espectro de XAS por transmissão é através
do método de medida dispersivo [57]. Neste método, um monocromador feito a partir
de um cristal curvo produz uma variação contínua do ângulo de Bragg (e assim da e-
nergia) em função da localização do fóton incidente sobre o cristal. Devido à geometria
da montagem, o feixe é focalizado na amostra permitindo o estudo de amostras de pe-
quena dimensão. Devido à rapidez na medida (10-2s), este tipo de montagem é indica-
do para estudos de transição de fase em função da temperatura, pressão, etc [57].
Dois outros métodos de medidas indiretas do coeficiente de absorção de raios-X
são também comumente utilizados: o método da medida por f1uorescência e o método
de medida por detecção de elétrons. Em ambos os casos, as medidas detectam um
processo secundário causado pela absorção em um processo primário. O fóton origi-
nado do segundo processo é diretamente proporcional ao processo de absorção primá-
rio.
A f1uorescência é um processo de desexcitação radioativa: o buraco gerado no
nível mais interno do átomo absorvedor (de energia E1)é preenchido por um elétron de
um nível eletrônico mais externo (de energia E2) provocando a emissão de um fóton
••
(de energia E1-E2). No caso da absorção por uma camada K, o processo de fluores-
cência é dominado pela produção de radiação Ka cuja energia é característica do ele-
mento absorvedor. A intensidade medida é proporcional então ao número de eventos
de absorção em amostras diluídas e assim, Ifllo será proporcional a u, O processo de
f1uorescência não é um processo muito eficiente fazendo com que um número razoável
de detectores seja utilizado na obtenção de espectros com uma boa razão sinal/ruldo:
O processo de desexcitação radioativa é dominante nos elementos com número atõml-

30
co maior que 40 para um processo de excitação da borda K. O método de detecção por
fluorescência é utilizado quando a quantidade do elemento a ser estudado é muito pe-
quena quando comparada aos outros elementos da amostra e quando a espessura da
mesma é relativamente grande.
O segundo método de medida indireta do coeficiente de absorção é através da
detecção total de elétrons, que é um processo de desexcitação não radioativa. As de-
sexcitações não radioativas correspondem a um processo de dois elétrons. O buraco
criado no nível mais interno (de energia E1) é preenchido por um elétron de uma cama-
da mais externa (de energia E2). Um outro elétron (de energia E3) é emitido então pelo
átomo. A energia cinética deste elétron (elétrons Auger) pode ser escrita como Ec=E1-
E2-E3. Os processos Auger são mais prováveis de serem observados do que as transi-
ções radioativas nos elementos de baixo número atômico. As medidas de detecção de
elétrons são indicadas no estudo de materiais na forma de filmes finos ou em materiais
na forma de "bulk". Este tipo de medida é indicado para elementos de baixo número
atômico e em altas concentrações.
Devemos observar que a profundidade de análise na amostra não é a mesma nos
três modos de medidas. O livre caminho médio dos fótons de f1uorescência é igual aos
fótons de raios-X enquanto o dos elétrons Auger é muito menor. No caso de deteção
de elétrons, a profundidade de análise é da ordem de 1000 a 5000 Á e varia de acordo
coma natureza do elemento analisado.
Outros dois modos de medida que não iremos apresentar em detalhes aqui são
os métodos de análise da estrutura da superfície denominados de SEXAFS e o ReflE-
XAFS. Ambos os métodos permitem obter informações estruturais a uma profundidade
variando entre 30 e 50 Á. Maiores detalhes sobre estes dois modos de medida podem
ser encontrados nas referências [49,58-60] .

••
11.9-Análise do espectro EXAFS

Como o sinal EXAFS representa apenas uma pequena porcentagem da absorção

total, a extração do sinal EXAFS deve ser feita de uma forma criteriosa e o mesmo pro-
cedimento deve ser seguido na análise de um conjunto de amostras cuja estrutura de-
seja-se comparar. Três suposições são feitas quando da análise dos espectros EXAFS:

31
(1) O sinal EXAFS é composto de uma soma de ondas originadas de diferentes átomos
vizinhos ao átomo absorvedor. Na primeira esfera de coordenação, na grande maioria
dos casos, o efeito de espalhamento múltiplo pode ser desconsiderado.
(2) A função de amplitude pode ser transferida de um composto modelo (experimental
ou teórico) para cada tipo de átomo da amostra de estrutura desconhecida.
(3) A função de fase também pode ser transferida de um composto modelo para cada
par de átomos A-B, onde A é o átomo absorvedor e B é o átomo espalhador.
O primeiro passo na análise dos dados de EXAFS consiste na extração do sinal X(k)
do coeficiente de absorção total seguindo alguns passos que podem ser visualizados
na Figura 8.

2,0,------------..., 2,0.,---------------,
1,5
(b)
1,5
1,0

0,5

>< 0,0 >< 1,0

:::l :::l
-o.s

-1,0 0,5

-1,5

4800 5000 5200 5400 5600 5800 6000 4850 4900 4950 5000 5050 5100

Energia (eV) Energia (eV)

2,0

1,5 (C) 0,6

(d)
1,0 0,4

0,5
0,2

0,0
X
::1. /Jl Q
.....•. 0,0
-0,5 X
-0,2
·1,0 .o:
-0,4
·1,5

·2,0 -0,6
4800 5000 5200 5400 5600 5110() 6000 o 6 10 12 14

Energia (eV) k(A')

Figura 8- Diferentes etapas da obtenção das oscilações EXAFS a partir do espectro de

absorção de raios-X. (a) remoção do background,(b) determinação da posição da bor-
da de absorção, (c) coeficiente de absorção atômica e (d) espectro EXAFS obtido após
normalização.

32
 Como foi definido anteriormente X pode ser obtido a partir da razão entre o coefi- to
ciente de absorção atômica (""'0), que corresponderia a um átomo isolado, e do dE
coeficiente de absorção devido aos processos de absorção de absorção pelos outros
átomos da amostra que pode ser obtido pela extrapolação da região da pré-borda n~
utilizando um polinômio geralmente de grau 1 (Figura 8a e 8c). Zé

Após a subtração do background e da normalização, a energia da borda de ab- ja:

sorção deve ser definida (Figura 8b). O valor da energia da borda é utilizado no calculo (1
do vetor de onda k dado pela equação:

k = [8~:m (hv + EO)r'2 (31)

ar
onde k é o vetor de onda, mé a massa do elétron e h é a constante de Planck. A esco-
(4
lha da energia Eo é arbitraria sendo feita geralmente uma escolha sobre o ponto de in-
f1exão da curva, isto é, o máximo da derivada primeira da borda de absorção.
O polinômio que representa o coeficiente de absorção atômico é obtido através
de um polinômio cuja forma depende da complexidade do sinal analisado. Normalmen- ar

te um polinômio de grau entre 5 e 7 é utilizado (Figura 8c). Após o processo de extra- er

ção de background e da obtenção do coeficiente de absorção atômica, o sinal X(k) po- se

"

ta
de então ser obtido. A Figura 8d mostra o espectro EXAFS obtido a partir da equação
(3). fi~

Sayers e colaboradores mostraram em 1971 que a transformada de Fourier de

se

X(k) tinha como resultado uma função de distribuição radial [21-25]. A transformada de
Fourier é definida como:

10
(32)

A função de distribuição radial é definida no espaço de distâncias R (distância á-

tomo central- átomo absorvedor). As distâncias encontradas através da transformação
_5
de Fourier são da ordem de 0,2 a 0,5 A menor que as distâncias reais. Isso se deve ao
processo de defasagem que o fotoelétron sofre ao se deslocar até o átomo vizinho. A -10

transformada de Fourier pode ser calculada com diferentes valores ou "pesos" de k

(multiplicação It). Esta opção pode ser útil na distinção entre átomos espalhadores le-
ves e pesados que formam a estrutura ao redor do átomo absorvedor. Um elemento Fi!

com uma pequena massa (oxigênio por exemplo), espalhará mais em baixos valores co
",
de k enquanto átomos pesados como por exemplo o chumbo, irão espalhar mais a al- tre
33
 tos valores de k. O cálculo da transformada de Fourier é feito entre dois valores de k,
definidos como kmin. e Kmax. A escolha do valor de Kmax. Depende de muitos fatores co-
mo a razão sinal/ruído enquanto o valor de Kmin. deverá ser escolhido de tal forma que
não inclua informações do espectro XANES. Para limitar esta região do espectro, utili-
za-se uma janela de corte W(k), já incluída na equação da transformada de Fourier. As
janelas de corte mais comumente utilizadas são:
(1) Janela de corte tipo Hanning:

W(k)=.!.{I-cos2n[ (k-kminJ (33)

2 (kmax - kmin)

onde W(k) é igual a zero para k=kmin. e k=kmax ..

(4) janela do tipo Kaiser-Bessel:

W (k) = i; {'t( 1-4(k

(kmax.
= kmin)2
kmin)
J1I2} / r,('t) (34)

onde Jo é uma função de Bessel de ordem zero. Normalmente, o valor de r se situa

entre 2 e 4. Como a transformada de Fourier é uma função complexa, quando da sua
solução, uma parte real e outra imaginária é obtida. A Figura 9a mostra a parte absolu-
ta e a parte imaginaria da transformada de Fourier do espectro EXAFS apresentado na
figura 8d. A parte absoluta é determinada pelo número de vizinhos e pelo fator de de-
sordem. A parte imaginária é muito útil na determinação das distâncias átomo central-
átomo absorvedor.

0.6
10
(b)
0.4
5

0.2.
g•..
o g
~ •.. 0.0

t
-5
-0,2

-10 -0,4

o 3 2 6 10 12 14

XAxisTitle k(A-')

Figura 9- (a) Transformada de Fourier mostrando a parte real e a parte imaginária bem
como as posições a partir do qual foi feita a transformada de Fourier inversa (b) espec-
tro EXAFS obtido a partir da transformada de Fourier inversa,

34
 Uma vez que o sinal foi decomposto em uma série de picos, é possível isolar ou
\

filtrar uma determinada esfera de coordenação através da realização de uma transfor- do

mada de Fourier inversa, a qual da origem a uma função X(k) de uma específica cama- tal
da de átomos vizinhos. Por exemplo, na figura 9a são apresentadas as posições de co
Rmin.E Rmax.quesão utilizadas neste caso para filtrar a primeira esfera de coordenação. SI!

O espectro EXAFS resultante deste processo é apresentado na Figura 9b. Camadas dE

situadas a maiores distâncias podem ser isoladas da mesma forma.

SI'

11.10-Determinação dos dados estruturais a partir do espectro EXAFS (I

Convencionalmente, a análise dos espectros EXAFS é feita através do Método dt

de Ajuste, o qual permite obter diretamente todos os parâmetros estruturais do sistema

(Nj, Rj, Oj) a partir dos parâmetros eletrônicos (lf(n,k)I, opq,(n,k», que podem ser obti-

dos a partir de uma amostra padrão de estrutura conhecida ou a partir de cálculos teó-
ricos [1]. rr

Para determinar os parâmetros estruturais, utilizamos programas de otimização o

pelo método dos mínimos quadrados que levam ao ajuste da função X(k) calculada. a
Para obter os parâmetros eletrônicos a partir de uma amostra padrão, é necessário
conhecer os parâmetros estruturais desta amostra. Conhecendo então a estrutura da ç
amostra padrão, podemos obter as defasagens e a amplitude de espalhamento e o li- c
vre percurso médio dos elétrons relativos a amostra padrão. Quanto mais próxima a
estrutura da amostra padrão for da amostra de estrutura desconhecida, mais próximos
(
serão os termos eletrônicos das duas amostras.
(
Entretanto, em certos casos, é impossível encontrar uma amostra padrão ade-
quada para o sistema estudado. Nesta situação, devemos fazer uso de funções de fase
e amplitude obtidas a pqrtir de programas que geram estas funções teoricamente.
O programa que atualmente vem sendo mais utilizado é o programa FEFF o qual
permite, além da obtenção das funções de fase e amplitude, obter os espectros EXAFS
e XANES teóricos [35-36]. Maiores informações sobre todos os parâmetros necessá-
rios no cálculo teórico das funções de fase e amplitude através do programa FEFF po-
dem ser encontradas nas referências [35-36].

35
J Além dos valores de Nj, Rj, Oj para cada camada, outros fatores são considera-
dos durante a simulação. O valor de Eo e o fator de correção de amplitude So podem
também ser ajustados durante o processo de simulação. Devido a existência de uma
~ correlação entre os fatores de amplitude (Nj, Se, Oj) e de fase (R, , Eo), o processo de
simulação deve ser realizado de uma maneira estratégica a fim de se evitar a obtenção
) de resultados duvidosos.
O número de parâmetros que podem ser ajustados durante a realização de uma
simulação depende do intervalo útil no espaço do vetor de onda do espectro EXAFS
(Ak) e do intervalo no espaço de distâncias onde foi feita a filtragem na transformada de

Fourier (AR). O número de parâmetros livres (n) durante a simulação pode ser ajustado
de acordo com o teorema de Nyquist [61]:

n= 2M.M (35)
1t

O número de parâmetros livres deverá sempre que possível ser n-2. Quanto
maior o número de camadas que serão simuladas, maior será AR e maior será também
o número de parâmetros a ser simulado. Em certos casos é possível se fazer algumas
aproximações com a finalidade de se reduzir o número de parâmetros livres.
Um outro fator importante que devemos analisar durante a realização da simula-
ção do espectro EXAFS é o limite de resolução em distâncias dado pela seguinte e-
quação:
1t
8R=-- (36)
2kmax.
onde Kmax. é valor máximo útil do espectro EXAFSexperimental. Em sistemas desor-
denados, onde a primeira esfera de coordenação é formada por uma certa distribuição
de distâncias, elas não poderão ser diferenciadas caso a diferença entre estas distân-
cias esteja abaixo do valor de 8R.
ti;

Finalmente, para quantificar se existe um bom acordo entre o espectro experi-

mental e o espectro obtido teoricamente, podemos calcular o fator de qualidade defini-
do como:

36
 L w(k)[)c(exp) -)c(theo)]2 v
R(%)=~/~·~=------------- (37)
a
LW(k)[x;(exp)]2
p
onde w(k) é um fator de ponderação e X são os espectros EXAFS experimental e teóri- p
co. E

s
11.11-Principais vantagens e desvantagens da técnica de XAS
c

A principal vantagem da técnica de XAS é que ela é uma técnica de estudo da

ordem local. Isto significa que através da técnica de XAS, podemos por exemplo, estu-
dar a estrutura de materiais não cristalinos ou comparar a estrutura de materiais na
fase líquida com a sua estrutura na fase sólida. Em circunstâncias ideais, a análise do (

espectro EXAFS pode fornecer informações estruturais com grande precisão chegando
a uma precisão maior que 0,01 A na determinação da distância interatômica.
No caso de materiais cristalinos, a técnica de XAS está em desvantagem com
relação à técnica de DRX. Entretanto, no estudo de materiais cristalinos apresentando
um certo grau de desordem local, ou no estudo de sistemas cristalinos contendo algu-
mas espécies atômicas de forma diluída, a técnica de XAS pode fornecer informações
estruturais que complementam as informações obtidas pela técnica de DRX. Finalmen-
te, a possibilidade de se obter informações sobre o estado de oxidação do elemento
analisado e da simetria do sitio do átomo absorvedor somente através da medida da
região da borda de absorção de raios-X (XANES), fazem desta técnica uma ferramenta
de caracterização estrutural muito poderosa.
Em comparação com outras técnicas espectroscópicas, a técnica de XAS apre-
senta a vantagem do sinal ser sempre detectado sem a necessidade de estados de
spin específicos ou de uma substituição isotópica. Pelo fato de cada elemento apresen-
tar uma borda de absorçãõ única, a técnica de XAS tem o poder de selecionar o ele-
mento a ser estudado mesmo em amostras que apresentam um grande número de es-
pécies químicas. Para elementos com número atômico acima do elemento fósforo, as
medidas de XAS são hoje feitas de maneira rotineira, principalmente nos centros mais
modernos de radiação síncrotron.
Como na maioria dos casos, a técnica de XAS apresenta algumas limitações ou
desvantagens. Como foi citado anteriormente, através da técnica de XAS não é possí-

37
 vel a identificação entre átomos espalhadores que se diferenciem por 2 ou 3 números
atômicos (C,O,N e F). Além disso, o valor finito de k faz com que a técnica de XAS a-
presente uma resolução limitada com relação a distâncias interatômicas. Finalmente,
'i-
por ser uma sonda local, na maioria dos casos, somente informações sobre a primeira
esfera de coordenação pode ser obtida. Entretanto, em casos onde a estrutura apre-
senta um certo grau de ordenação, informações estruturais das esferas situadas a mai-
ores distâncias podem ser obtidas.

a
11.12-Perspectivas sobre o uso da técnica de XAS no Brasil
1-

a
Certamente, o número de pesquisadores no Brasil que utilizam a técnica de XAS
como ferramenta de caracterização estrutural em seus projetos de pesquisa é ainda
muito pequeno se compararmos a outros países. Entre os principais fatores que podem
explicar este fato é o desconhecimento das potencial idades da técnica e a dificuldade
n
de realizar a medida uma vez que, no nosso caso, as medidas podem somente ser rea-
)
lizadas no Laboratório Nacional de Luz Síncrotron, situado em Campinas, S.P.. Apesar
deste número no contexto geral ser pequeno, a demanda pela utilização da linha de
XAS do LNLS tem sido maior que a oferta. Essa aparente contradição pode ser expli-
cada pelo fato de que um grande número de projetos vem sendo submetido por um
pequeno número de pesquisadores. Esta situação vem sendo alterada nos últimos a-
nos e a tendência da inclusão de novos usuários da técnica vem sendo observada.
Um fator que certamente deverá influenciar o aumento do número de usuários é
a construção de novas linhas como a de EXAFS dispersivo e a segunda linha de XAS.
Além disso, o [NLS vem a alguns anos financiando a participação de usuários localiza-
dos fora do estado de São Paulo, fazendo com que muitos pesquisadores localizados
fora do eixo Rio-São Paulo se tornem usuários desta técnica .
••
Ainda que até o presente momento na maioria de trabalhos de Pós-Graduação
(Mestrado e principalmente Doutorado) a técnica de XAS seja utilizada como uma das
técnicas de caracterização estrutural, o número de trabalhos onde a técnica é a ferra-
menta principal de caracterização estrutural vem crescendo dia-a-dia.
Assim, acredito que o uso da técnica de XAS no Brasil tende a crescer e em um
futuro não muito distante, venha fazer parte do conjunto de técnicas de caracterização
estrutural usuais a muitos pesquisadores, a exemplo das técnicas de DRX e RMN.

38
11.14- Referências

1- A. Michalowicz, Tese de Doutorado, Université Paris Vai de Marne, França (1990).

2- P. Eisenberger and B.M. Kincaid, Science 200, 1441 (1978).
3- R Stumm von Bordwehr, Ann. Phys. Fr., 14, 101 (1989).
4- B.K. Teo, "Inorganic Chemistry Concepts vol9. EXAFS: Basic Principies and Data
Analysis", (Springer-Verlag) (1986).
5- EA. Stern, Theory of EXAFS: X Ray Absorption Principies, Aplications and
Techniques, Chapter 1 (Eds. Konisberger and Prins, New York - 1988).
6- F.W. Lytle, J. Synchrotron Radiation 6, 123 (1999).
7- W.C. Rõntgen, Sitzung. Phyz. Mediz. Gesellschaft zu Würzburg 137, 132 (1895).
8- E Wagner, Ann. Phys. 46, 868 (1915).
9- R. de L. Kronig, Zeit Phys. 70, 317 (1931)
10- R de L. Kronig, Hand Phys. 2nd ed., Smekal ed. (Berlin:Springer), pg. 253 (1933)
11- R de L. Kronig, Zeit Phys. 75, 191 (1932).
12- R de L. Kronig, Zeit Phys 75, 468 (1932).
13- H. Petersen, Zeit Phys. 76, 768 (1932)
14- H. Petersen, Zeit Phys. 80, 258 (1933)
15- H. Petersen, Arch. Néercl. Sci. Exactes Nat. 14, 165 (1933).
16- RM. Levy, J. Chem. Phys. 43,1846 (1965).
17- F.W. Lytle, "Development in Applied Spectroscopy", J.R Ferraro and J.S. Ziomek
Eds. (New York, Plenum), 2, 285 (1963).
18- F.W. Lytle, Adv. X-Ray Analysis 9,398 (1966).
19- F.W. Lytle, Acta Cryst. 22, 321 (1967).
20- F.W. Lytle, Proc. Int. Conf. On Physics of Non-Cryst. Solids, Delft, pp 12-30 (1965).
21- D. E Sayers, E A. Stern and F.W. Lytle, Adv. X-Ray Anal. 13,248 (1970).
22- D. E Sayers, E A. Stern and F.W. Lytle, Phys. Rev. Lett. 27, 1204 (1971).
23- F.W. Lytle, D. E Sayers and E A. Stern, Phys. Rev. B11, 4825 (1975).
24- E. A. Stern, D.E Sayers, F.W. Lytle, Phys. Rev. B11, 4836, (1975).
25- D.E. Sayers, F.W. Lytle, EA. Stern, J. of. Non-Cryst. Solids 8-10, 401 (1972).
26- B.M. Kincaid and P. Eisenberger, Phys. Rev. Lett. 34, 1361 (1975).
27- P. Eisenberger and B.M. Kincaid, Chem. Phys. Lett. 36, 134 (1975).

39
 28- 8.M. Kinacaid, P. Eisenberger, K.O. Hodgson and S. Doniach, Proc. Nat. Acad. Sci.
USA 72, 2340 (1975).
29- C.A Ashley and S. Doniach, Phys. Rev. B11, 1279 (1975).
30- P.A Lee and J.B. Pendry, Phys. Rev. B11, 2795 (1975).
31- A Bianconi, Appl. Surface Science 6 (3-4),392-418 (1980).
32- Desde 1981 12 congressos sobre a técnica de XAS já foram organizados. Atual-
?: mente os congressos são realizados a cada 3 anos. O próximo congresso será realiza-
~

II do em Stanford em 2006.
33- N. Binsted, S.L. Cook, J. Evans, G. Greaves, RJ. price, J. Am. Chem. Soc. 109,
3369 (1987).
34- D.C. Koningsberger and R Prins, Eds, "X-Ray Absorption: Principies, Applications,
techniques of EXAFS, SEXAFS and XANES", (Wiley, New York) 1988.
35- J.J. Rehr, S.\. zabinsky, A. Ankudinov, RC. Albers, Physica B 208&209, 23 (1995).
36- S.1.Zabinsky, J.J. Rehr, A. Ankudinov, RC. Albers and M.J. Eller, Phys. Rev. B 52,
(1995).
37- AA Filipponi, A. Di Cicco, C.R Natolli, Phys. Rev. B 52, 15135 (1995).
38- J.J. Rehr, J. Mustre de Leon, S.\. zabinsky, RC. Albers, J. Am. Chem. Soc. 113,
5135 (1991).
39- Activity Report LNLS 2002.
40- Resumos da 14a Reunião Anual de Usuários, LNLS, 9-11 de fevereiro de 2004.
41- G.E. Brown, Jr, G. Calas, G.A Waychunas, J. Petiau, "X-ray absorption
Spectroscopy and its applications in mineralogy and geochemistry" in Review in
Mineralogyvol. 18, Chapter 11, pgs431-512 (1989).
42- P. Lagarde, LURE-Orsay, France. Communication privée (1982).
43- P. Eisenberger and G.S. Brown, Solid State Comm. 29,481-484 (1979).
44- J.E. Muller and J.W. Wilkins, Phys. Rev. B 29,4331 (1984) .
•
45- J.J. Rehr, XAFS V, Physica B 158, 1 (1989).
46- A. Messiah, Mecanique Quantique, t1, Ed. Dunod, 301 (1963).
47- A. Messiah, Mecanique Quantique, t1, Ed. Dunod, 315 (1963).
48- E.A Stern, B.A. Bunker and S.M. Heald, Phys. Rev. B (1979).
49- E.A. Stern and S.M. Heald, "Handbook on Synchrotron Radiation vol 1B, Chap. 10:
Basic principies and applications of EXAFS, p. 955-998", Eds. E.E. Koch, D.E. Eastman
and Y. Farge. North Holland Publishing Company (1983).

40
50- Nathalie Richard, Tese de Doutorado, Université Paris-VI, Paris, França (1996).
51- F. Farges, Journal of Non-Cryst. Solids 204, (1996) 53.
52- F. Farges, G.E. Brown Jr and J.J. Rehr, Geoch. et Cosmoch. Acta 60(16), 3023
(1996).
53- AL. Ankudinov, J.J. Rehr, J. Synchrotron Rad. 10: 366-368 Part 5 Sep. (2003).
54-AL. Ankudinov, C.E. Bouldin, J.J. Rehr, J. Sims, H. Hung, Phys. Rev. B 65 (10): Art.
No. 104107 MAR 1 (2002).
55- AL. Ankudinov , J.J. Rehr, J.J. Low, S.R. Bare, Topics in Cata!. 18 (1-2): 3-7
(2002).
56- H. Modrow, S. Bucher, J.J. Rehr, AL. Ankudinov , Physical Review B 67 (3): Art.
No. 035123 JAN 15 (2003).
57- G. Tourillon, E. Dartyge, A Fontaine, H. Tolentino, Journal of The Electrochemical
Society 135 (3): C150-C150 MAR (1988).
58- J. Stohr, Surface crystallography by rneans of SEXAFS and NEXAFS. In: Chemistry
and Physics of Solid Surfaces. V.R. Vanselow and R. Howe, eds., Springer Series in
Chem. Phys. Vol. 235, Springer-Verlag, New York, p. 231-275 (1984).
59- P.H. Citrin, J. Physique 47, C8, 437-472 (1986).
60- L. Bosio, R. Cortes and G. Folcher, J. Physique 47, C8, 113-116 (1982).
61- E.A Stern, Phys. Rev. B 48,9825 (1993) .

41
 Capítulo 11I

A Espectroscopia de Absorção de Raios-X Aplicada no Estudo

Estrutural de Materiais Vítreos

42
11I.1-Introdução o

Do ponto de vista estrutural, os materiais vítreos e os materiais amorfos se dife-

renciam dos materiais cristalinos por apresentarem somente ordem estrutural a curta íl
distância. Por outro lado, os materiais vítreos se diferenciam dos materiais amorfos por [.
apresentarem o fenômeno de transição vítrea. ç
Como destacamos anteriormente, a espectroscopia de absorção de raios-X é con- r
siderada apropriada para o estudo da estrutura de materiais vítreos e amorfos uma vez
que não necessita da existência de ordem estrutural a longa distância no material a ser
analisado.
Dentre os diversos problemas envolvendo o estudo estrutural de materiais vítreos
e amorfos que desenvolvemos utilizando a técnica de XAS, iremos aqui destacar dois
casos:
(a) Relação entre a ordem local e a tendência à nucleação em vidros silicatos;
(b) Estudo da ordem local em um vidro de composição Ge25Ga10S65
que apresenta o
fenômeno de foto-expansão

11I.2-Relação entre a ordem local e a tendência à nucleação em vidros silicatos

A motivação para o desenvolvimento deste estudo nasceu do fato de que diferen-

tes trabalhos na literatura procuraram estabelecer uma relação entre a estrutura a nível
molecular ou estrutura atômica em líquidos superresfriados e sua tendência em nuclear
no volume ou na superfície [1,2]. A principal suposição está no fato de que se a estrutu-
ra a nível local de um vidro e de um cristal de mesma composição são idênticas, so-
mente será necessário que ocorra um pequeno rearranjo na interface destas duas fa-
ses para que se de início o processo de nucleação. Assim, neste caso, mesmo que a
nucleação ocorra no volume da amostra, ela irá ocorrer mais facilmente. Se estes rear-
ranjas na interface forem muito importantes, o processo de nucleação irá somente o-
correr na superfície material/ar, onde existem ligações incompletas ou impurezas, ou
no volume, somente com o auxílio de agentes nucleantes.
Apesar do fato que a estrutura local dos Ions modificadores em alguns dos siste-
mas vítreos estudados já ter sido caracterizada por diferentes técnicas e mesmo por
EXAFS, não havia na literatura nenhum trabalho que fazia uma análise da proposição

43
de uma relação entre ordem local e tendência à nucleação para os sistemas que está-
vamos propondo a estudar.
Assim, realizamos através da técnica de EXAFS, o estudo da ordem local dos
fons modificadores em três sistemas vítreos silicatos CaSi03, Na2Ca2SbOge PbSi03
[Artigo-IILA]. Os dois primeiros sistemas, CaSi03 e Na2Ca2SbOg,apresentam nuclea-
ção volumétrica enquanto que o sistema PbSi03, a nucleação ocorre somente na su-
perfície. Estes sistemas vítreos foram escolhidos pelo fato de terem seus processos de
nucleação bem determinados.
Nossos resultados de EXAFS mostraram que em vidros que apresentam uma alta
tendência a nucleação volumétrica (sistemas CaSi03 e Na2Ca2SbOg),a estrutura local
dos cátions modificadores é similar à ordem local de suas fases cristalinas de mesma
oomposição. Por outro lado, a estrutura local do átomo de chumbo no vidro que apre-
senta uma baixa tendência a nucleação volumétrica (PbSi03) é completamente diferen-
te da ordem local observada em sua fase cristalina de mesma composição. Assim, ain-
da que o trabalho tenha sido realizado sobre um número limitado de sistemas, a pro-
posta feita por alguns autores sobre a existência de uma relação entre ordem local e
tendência à nucleação em materiais vítreos foi corroborada pelos resultados de EXAFS
obtidos neste trabalho.
Em um trabalho posterior, este estudo foi estendido a caracterização da ordem
local do átomo formador do sistema, neste caso, o átomo de silício [Artigo-IILB]. Ainda
que o estudo da ordem local dos átomos de silício pudesse ser feito através da técnica
de EXAFS, utilizamos neste caso a técnica de Ressonância Magnética Nuclear (RMN).
De acordo com os resultados de RMN, uma correlação entre a forma da distribuição
das unidades Qn (Qo, Q\ Q2, Q3 e Q4) e a tendência à nucleação foi observada. Deste
trabalho foi possível concluir que a similaridade na conectividade entre os tetraedros
Si04 na amostra vítrea e na sua fase cristalina de mesma composição tem um papel
fundamental na determinação da tendência a nucleação dos vidros metasilicatos carac-
terizados neste trabalho.

11I.3- Estudo da ordem local em um vidro de composição Ga10Ge25S65

que apre-
senta o fenômeno de foto-expansão

44
 Desde a descoberta do fenômeno de cristalização induzida pela luz por Feinleib e q
colaboradores os fenômenos foto-induzidos têm sido estudados em vidros calcogene- "d

tos {3]. Os vidros calcogenetos exibem diferentes mudanças quando expostos a luz que
tenha energia comparável ao seu "band-gap". Tais mudanças podem ser estruturais
(variações na densidade, foto-expansão), mecânicas (propriedades reológicas), ópticas
("photo-darkening") ou químicas (foto dissolução do metal). q
Estudos realizados em amostras de composição Ge25Ga10S65através das técnicas s
de AFM e perfilometría mostraram que a expansão superficial na amostra é da ordem
de algumas centenas de nanômetros [4,5]. Apesar dos inúmeros trabalhos experimen- n
tais nesta área, as explicações para o processo da foto-expansão são em grande parte a
especulativas [6-10]. ti
Deste modo, na busca de uma melhor compreensão deste fenômeno, realizamos
um estudo estrutural da ordem local através da técnica de XAS antes e após a geração
do referido fenômeno.
Inicialmente, realizamos o estudo da ordem local ao redor do átomo de germânio
em vidros de composição Ge25Ga10Ss5[Artigo-III.C]. O espectro EXAFS da amostra a
Ge25Ga10S65foi então medido em amostras antes e após a iluminação.
Sabe-se que a estrutura cristalina do GeS2 possui uma estrutura em camadas
construídas a partir de cadeias de tetraédros de GeS4 compartilhados pelo vértice. A
análise qualitativa dos espectros EXAFS mostrou uma diminuição na amplitude do sinal r
EXAFS a medida que o tempo de exposição aumenta. Para levar em conta esta dimi-
.
t
nuição, foi necessário assumir a existência de ligações Ge-S e Ge-O. Assim, a diminui- r
ção na intensidade do sinal de EXAFS pôde ser explicada pela interferência destrutiva
entre as ondas do fotoelétron retroespalhado pelos átomos de oxigênio e enxofre. c
Com a finalidade de verificar qual a origem do oxigênio detectado através da téc- t
nica de EXAFS após a iluminação, estudou-se a influência da atmosfera durante o pro- t
cesso de iluminação [Artigo-IU.O]. Para tanto, fixou-se o tempo de exposição, a densi-
t
dade de potência e comprimento de onda e procedeu-se a exposição das amostras em r
4 tipos de atmosferas.
Os resultados mostraram que a amostra iluminada na presença de oxigênio apre- (

senta uma foto-expansão maior que aquela iluminada em condições normais de labora-
tório. A observação mais surpreendente reside no fato de não ocorrer a foto-expansão

45
quando a amostra é mantida em atmosferas inertes como no vácuo, hélio o arqónlo
durante a iluminação.
Destas observações ficou evidente que o responsável pelas variações estrutu-
rais observadas nos vidros estudados é o oxigênio proveniente da atmosfera.
O fato da espessura da amostra iluminada na presença oxigênio ser maior do
que a amostra iluminada no ar levou-nos a investigar seu papel no efeito da fotoexpan-
são.
Recentemente, realizamos um estudo de espectroscopia de absorção de raios-X
na borda K do átomo de oxigênio e medidas de XPS de todos os constituintes nas duas
amostras. Os resultados deste estudo, que se encontram em fase final de análise, mos-
tram claramente a formação de ligações Ge-O na amostra após iluminação.

11I.4-Conclusão

- Relação estrutura local e tendência a nucleação em vidros silicatos

Podemos concluir que a utilização da técnica de XAS no estudo da estrutura

local dos átomos modificadores mostrou ser de grande utilidade uma vez que foi pos-
sível identificar uma certa relação entre a ordem local a curta distância e a tendência a
nucleação para um certo conjunto de sistemas vítreos. As informações obtidas por
RMN no estudo estrutural do átomo formador dos diferentes sistemas vítreos corrobo-
raram em certos casos com os resultados obtidos por XAS.
Este tema de pesquisa desperta ainda um interesse muito grande. A prova disso
que este é um dos temas de discussão dentro do "Basic Science Committee" da "In-
ternational Commission of Glass". Estudos atuais realizados por outros autores mos-
tram uma relação entre a tendência à nucleação e a ordem a média distância. Entre-
••
tanto, como foi o nosso caso, estes estudos foram realizados em um número peque-
no de sistemas o que impede no momento a generalização dos resultados.
Certamente, a extensão deste estudo a outros sistemas vítreos poderá nos dizer
o quanto dos resultados já obtidos poderá então ser generalizado.

- Estudo da ordem local em vidros e o fenômeno de foto-expansão

46
 Nossos estudos mostraram que para o sistema caracterizado o oxigênio tem
um papel fundamental no processo de fotoexpansão. Os resultados obtidos por XAS
e por XPS, ainda em fase final de análise, mostram que após a iluminação ocorre a
formação de ligações Ge-O. Os resultados de XAS mostram também que as interações
Ge-O são diferentes para átomos localizados na superfície e no volume da amostra.

11
I.V- Referências

1- E. Muller, K. Heide, E.D. Zanotto, J. Kristallogr. 200, 287 (1992).

2- E. Muller, K. Heide, E.D. Zanotto, J. Non-Crystalline Solids 155, 56 (1993).
3- J. Feinleib, J. deNeufville, S.C. Moss and S.R. Ovshinky. Appl. Physic Lett. 18, 254
(1971).
4- S.H. Messaddeq, V.R. Mastelaro, M. Siu Li, M. Tabackniks, D. Lezal, A. Ramos, Y.
Messaddeq, Applied Surface Science 205, 143-150 (2003).
5- S.H. Messaddeq, V.R. Mastelaro, A.Y. Ramos, M. Siu Li, D. Lezal, S.J.L. Ribeiro, Y.
Messaddeq, Journal of Non-Crystalline Solids 304, 160-166 (2002).
6- K. Tanaka, in Fundamental Physics of Amorphous Semiconductor edited by F.
Yonezawa - Springer Verlag, Berlim, p. 104 (1981).
7- H. Hamanaka, K. Tanaka and S. lijima, Solid St. Com., 23,63; ibid., 33, 355 (1980).
8- H. Hamanka, K. Tanaka, A. Matsuda and S. lijima, Solid St. Commun, 19, 499
(1976).
9- H. Hisakuni and K. Tanaka, Applid. Phys. Lett., 65, 2925 (1994).
10- K. Shimikawa, N. Yoshida, A. Ganjoo and Y. Kuzukawa, Phil Mag. Lett., 77, 153
(1998).

47
 Artigos

i
f

111. A Espectroscopia de Absorção de Raios-X Aplicada no Estudo

Estrutural de Materiais Vítreos

IlI.2. Relação entre a ordem local e a tendência à nucleação em vidros silicatos

Artigo-IlI.A- Relationship Between Short Range Order and ease of Nucleation in Na2Ca2Si309,
CaSi03, and PbSi03 Glasses

Artigo-IlI.B- 29Si MAS-NMR Studies of Qn Structural Units in Silicate Glasses and Their
Nucleation Ability.

IlI.3. Estudo da ordem local em um vidro de composição GalOGe25S65que apresenta o

fenômeno de foto-expansão

Artigo-IlIC.- Local Order around of gennanium atoms in GalOGe25S65

glass by EXAFS

Artigo-IlI.D- The influence of oxygen in the photoexpansion of GaGeS Glasses

48
 jQURNAL DF

NON-CRYSTALLINE SOLIDS
ELSEVIER lournal of Non-Crystalline Solids 262 (2000) 191-199
www.elsevier.comllocate/jnoncrysol

Relationship between short-range arder and ease of nucleation

in Na2Ca2Sh09, CaSi03 and PbSi03 glasses
Valmor Roberto Mastelaro Edgar Dutra Zanotto
a,*, b, Nicolas Lequeux c,
Robert Cortes d
a Institutode Física de São Carlos, Universidade de Sao Paulo. CiP. 369, CEP 13560-970. São Carlos, SP, Brazil
b Departamento de Engenharia de Materiais. UFSCar-13565-905, São Carlos, SP, Brazil
c Laboratoire de Ceramiques et Materiaux Mineraux, UA-CNRS, Ecole Supêrieure de Physique et de Chimie Industrielles de Paris,
75005 Paris, France
d Physique des Liquides et Electrochimie - Université Paris VI, 75005 Paris, France
Received 16 March 1999; received in revised forrn I1 October 1999

Abstract
A relationship between the short-range order around the modifier cations and the crystal nucleation tendency in
silicate glasses is demonstrated. New extended X-ray absorption fine spectroscopy (EXAFS) results on the local
structure around calcium and lead atoms were obtained and analyzed for both vitreous and crystalline samples. Three
different silicate systems were studied: wollastonite (CaSi03) and soda-lime-silica (Na2Ca2Si309), for which volume
nucleation is easily observed and lead metasilicate (PbSi03) for which nucleation occurs only on the sample surfaces in
typicallaboratory conditions. In the glasses that have a high nucleation tendency (Na2Ca2Si309and CaSi03), the local
structures of these modifier cations are similar to their short-range order in the isochemical crystalline phases, whereas
the local structure in the glass that presents a low nucleation tendency (PbSi03) is quite different from that of its
isochemical crystal phase. © 2000 Elsevier Science B.V. All rights reserved.

1. Introduction onstrated [I] that stoichiometric glass forming

Crystal nucleation is a key issue to under-
standing glass formation. It also plays a funda-
mental role in the development of advanced
glass-ceramics for novel applications.
ln the absence of catalyzing agents, most su-
percooled liquids crystaIlize heterogeneously from
the external surfaces when heated; only a few
systems crystaIlize in the bulk. lt has been dem-
«
• Corresponding author.
Ennail address: valmor@if.sc.usp.br (V.R. Mastelaro).
0022-3093/00/$ - see front matter © 2000 Elsevier Science B.V. AlI rights reserved.
PU: S O O 2 2 - 3 O 9 3 ( 9 9 ) O O 6 8 5 - 7
systems can be separated into two classes. The first
are able to crystallize in the volume (supposedIy by
homogeneous nucleation) and their temperatures
of maximum homogeneous nucIeation rate (Tmax)
are dose to the experimental glass transition tem-
peratures, Tg (here defined as the point where the
viscosity t7 = 1012 Pa s). The opposite behavior is
observed for the other type of glasses, which only
crystallize heterogeneously on the surfaces, i.e., the
(calculated) temperatures of maximum hornoge-
neous nucleation rate are substantially lower than
Tg (Tmax « Tg). For this second family of glasses,
the induction periods in the temperature range
where nucIeation could in principIe be detected are

49
192 V.R. Mastelaro et ai. I Journal
too longo Alternatively, the homogeneous nucle-
ation rates could be undetectably low.
Although the general trends described above
are now firmly established, a crucial question re-
mains [2,3]: what is the relationship, if any, between
lhe molecular structure of supercooled liquids and
their nucleation ability (volume ar surface nucle-
a/íon)? The main assumption is: if the local struc-
ture . of a glass and its isochemical crystal are
similar, only a few interfacial rearrangements will
be necessary for crystal nucleation, which then can
take place easily, even in the glass volume. If these
rearrangements are substantial, nucleation can
only occur on the external surfaces, assisted by
unsaturated bonds and solid impurities at the
surface, or in the volume, only with the aid of
nucleating agents.
The objective of this work is to confirm or to
refute the proposed relationship between the local
structure (the first coordination shell) of the mod-
ifier cations and the crystal nucleation tendency in
silicate glasses. The main backbone of silicate
glasses and crystals are composed of identical Si04
tetrahedra, thus the Si-O distances and coordina-
tion numbers are equal; however the type of link-
ages between these tetrahedra may be quite
distinct. Additionally, it has been proposed (1-3]
that for glasses that nucleate in the volume, the
local structures of the modifier cations are similar
to the structures of their isochemical crystal
phases. On the other hand, for glasses that only
nucleate on the surface, the local order is supposed
to be quite different from those of their isochemi-
cal crystal phases. A similar short-range order
around the modifier cations of a given glass and its
isochemical crystal phase is assumed to be a nec-
essary, but not a sufficient condition to assure ease
of volume nucleation, because only average values
of coordination numbers and atomic distances are
determined by different structural characterization
methods such as X-ray diffraction (XRD) and
extended X-ray absorption fine spectroscopy
(EXAFS).
In qualitative terms, tl!e degree of structural
similarity between parent glass and resulting
crystal should affect the main thermodynamic pa-
rameters controlling nucleation; liquid-crystal
surface energy (O') and thermodynamic driving
..
,
of Non-Crystalline Solids 262 (2000) 19/-/99
force (.'1G), thus affecting the nucleation ability of
the system. The homogeneous nucleation rate is
given by: I~K'/II' exp(-K"·(J)/T·.'1G~), where
K' ,K" are constants and 11 is the viscosity; hencc
the surface energy dominates due to its power-3
against power-2 for .'1G.
The exact relationships of the structural pa-
rameters of glass and crystal with surface energy
and driving force are not known, however, there
should be some optimum value of similarity. This
idea is clarified if one analyses two limiting cases:
(i) when the structures of glass and isochemical
crystal are toa different, then is bound to be
(J
large, but .'1G should also be large. On the other
hand, in the hypothetical case, (ii) when the
structures of glass and crystal tend to be almost
identical, O' should be very small, but .'1G should
also tend to be zero. In this case, nucleation should
not occur!
We present here new EXAFS results on the
local structures around calcium and lead, in both
vitreous and crystalline samples, in three different
systems with well defined nucleation tendencies:
soda-lime-silica glass (Na2Ca~SÍJ09) and wol-
lastonite (CaSiO), for which volume nucleation
is easily observed and lead metasilicate glass
(PbSi03) for which only surface nuc1eation is
observed [1].
2. Sample preparation, EXAFS measurements and
data analysis
The glasses were prepared by melting homoge- I
neous mixtures of analytical grade reagents I
(Na2CO), CaCO), Pb304 and Si02) in a platinum
crucible. The melting temperatures ranged frorn
1000 to 1550°C, with a hold time of 2 h. The melts
were then cast between two cold steel plates, with I
an estimated cooling rate of 400°C/s. The amor- •
phous nature of the splat-quenched glasses was
confirmed by XRD. To obtain fully crystallized
samples, the specimens were submitted to the [01-
lowing nucleation and subsequent development
treatments: theNa2Ca2Si309 glass sample was
heated at 600°C for 24 h and then heated at 690°C
for 20 min; the CaSiO) glass sample was heated at
725°C for 120 h and then heated at 885°C for 12 h
50
 VR. Mas/e/aro e/ alo / Journal
f and, finaIly, the PbSi03 glass sample was heated at
650°C for 6 h. The crystaIline phases were charac-
terized by XRD. Only one crystalline phase was
observed in each system: WoIlastonite-lA for Ca-
SiO) [4], Alamosite for PbSiOJ [5] and Combeite
for Na2Ca2SiJ09 [6]. The three silicate glasses
characterized in this work exhibit widely different
volume nuc1eation rates [I]. The maximum nucle-
ation rates, 'lma, are: Na2Ca2SiJ09 (~1012 m"
s"), CaSiOJ (~106 m" S-I) and PbSiOJ un-
detectably low « 30 m-J s"), Experimental val-
ues exist for the first two glasses. The limiting
value for PbSiOJ was readily caIculated because no
crystals were observed in the volume after treating
samples of I x 10 x 10 mrrr' at severaI tempera-
tures (T> Tg) for 100 h. If only one crystal was
observed, a nuc1eation rate of >- 30 m-J çl would
result.
X-ray absorption experiments (Ca K-edge
~ 4038 eV, Pb LlII-edge ~ 13055 eV) were carried
out on D21 and on D42 beam lines at LURE
(France), using a Si(lll) and Si(3 I 1) double-
crystal monochromator, respectively. Absorption
spectra were recorded in the transmission mode, at
10 K for the experiments made at the Ca K-edge
and at room temperature for the Pb Lm-edge ex-
periments. The rnaterials were finely ground, then
the powders were dispersed in ethanol and finalJy
were cast by drawing the solvent through a poly-
carbonate filter, to deposit a layer with the re-
quired thickness.
The EXAFS function, X(k), was extracted from
the measured X-ray absorption spectrum follow-
ing the usual procedure [7]: subtraction of a third-
order polynomial drawn through the EXAFS
,
I1 oscillations and normalization to the pre-edge re-
gion by fitting to a first-order polynomial extrap-
olated to the EXAFS region. Then, k2X(K) was
I
i
Fourier-transformed using a Kaiser window to
obtain the radial distribution function of back-
11 scattering atoms around the calcium and lead at-
oms. The contribution of a particular shelJ to the
1I EXAFS spectrum was determined by Fourier-fil-
J Fourier transform of Na2Ca2Si309 gIassy and
I tering the corresponding peak and back trans-
forming to the k space. The back-transformed data
I were fitted using the Round Midnight prograrn
developed by Michalowicz [8). The fitting param-
eters were: (i) the mean bond Iength between the
ofNon-Crystalline Solids 262 (2000) 191-199 I 'J.l
shell and the central atom, (ii) the number and
type of atom in each shell and (iii) the structural
disorder (mean square displacement between the
central and neighboring atoms) described by a
Debye-Waller term (a} The back-scattering am-
plitude and total phase shift used in the fitting
procedure were obtained in two ways: using a
reference material having a well-known structure
for lead (the red-PbO crystalline compound). As
we did not find any appropriate crystalline refer-
ence sample for calcium, we used the ab initio
functions caIculated by the FEFF5 XAS code,
developed by Rehr and co-workers [9,10). The er-
rors were estimated to be ±0.01 Á in the mean
bond length (R) and ±5% in the coordination
numbers (N) and in the Debye-Waller factor (a)
The quality of the fit is given by the reduced chi-
squared X;, defined in [11].
3. ResuIts
3.1. Glassy and crystalline CaSi03
Fig. l(a) and (b) show the EXAFS spectra and
their Fourier transform of CaSiOJ glassy and
crystalJized from the glass, respectively.
The first peak at the Fourier transforrn, situated
at approximately 1.8 Á, corresponding to the first
Ca-O nearest neighbors, was fitted using ab initio
calculated backscattering amplitudes and phase
shifts from the FEFF5 XAS code [9,10]. The re-
sults of our fittings are given in Table I together
with the resuIts of other authors. As an example,
the agreement between the filtered EXAFS spectra
and fitted curve for the CaSiOJ glass is shown in
Fig.2.
3.2. Glassy and crystalline Na2Ca2Si309
Fig. 3 shows the EXAFS raw spectra and their
crystallized from the glass, respectively. The first
peak at the Fourier transform, corresponding to
the first Ca-O bonds, presents equal position and
amplitude for both materiaIs. However, some dif-
ference is observed in the medium range order.
SERViÇO DE BIB~IOTECr~
51 I f S ( • USp INFORMAÇAO
194 v.R. Mas/e/aro et al. I Journal of Non-Crystalline Solids ]62 (]OOO) 191-199

A quantitative analysis of these samples has

been carried out using ab initio back-scattering
CaSiO, Glass
amplitudes and phase shifts calculated by the
0.0
FEFF5 XAS code [9,10]. The fitting results are
given in Table 2, together with the XRD results of
g
...• Ohsato and Takeuchi [12] for crystalline Com-
CaSiO, Crystallized beite.
-0.5

3.3. Glassy and crystalline PbSi03

4 6 8 10 12

(a) K(A") Fig. 4 shows the raw spectra and their Fourier
transform of glassy and crystallized PbSi03, re-
0.7 spectively.The first peak situated at approximately
--CaSiO, Crystallized
1.8 Á at the Fourier transform, corresponds the
----- CaSiO, Glass first Pb-O bonding and its intensity is smaller for
the crystalline sample. This difference in the first
peak intensity can be due to a smaller coordination
...:
...: number or to a higher distortion of the first co-
ordination shell of the crystalline sample. Quanti-
tative results for lead metasilicate glass and crystal
were obtained using the red PbO crystal as a model
(each Pb atom is coordinated by four oxygen ions
o.o.J-:.-----,--~...:=.L~.:::::::~~~~ situated at approximately 2.31 Á [13]). We have
o 2 4 6
chosen this compound as a reference because the
(b) R (A)
local order around the leadions is better than that
Fig. I. (a) Ca K-edge EXAFS raw spectra and (b) corre- in crystalline PbSi03 or in yellow PbO. Our fitting
sponding Fourier transform for glassy and crystallized CaSiO). results, assuming the existence of only one shell in

Table 1
Structural parameters obtained with different techniques for CaSiO) compounds'
Samples Nex-o Rc.-o (Á) (Jc.o Á (EXAFS) Technique References
(±0.3) (±0.01) Á (±0.01) Á
CaSiO) glass 5.6 2.49 0.24 EXAFS [14]
CaSiO) glass 2.44 XRD This work
CaSiO) wollastonite 6.33 2.39. XRD [15]
CaSiO) wollastonite 6.30 2.40 0.09 EXAFS [16]
CaSi03 wollastonite 5.9 2.37 0.08 EXAFS [17]
CaSiO) wollastonite 6.4 2.37 0.10 EXAFS [18]
CaSiO) wollastonite 6.33 2.39 Neutron
CaSi03 glass +3% AhO) 6.16 2.37 diffraction [9]
CaSiO) glass 6.0 2.43 XRD [20]
CaSi03 glass 6.5 2.40 Molecular [21]
dynamics
CaSiO) crystallized from 5.8 2.37 0.11 0.04 EXAFS This work
glass
CaSi03 glass 6.0 2.36 0.13 0.06 EXAFS
"In Ref. [14J the number of neighbors, Nc.-o, and the mean bond length Rca-o are the average values of the 3 different ca1cium sites.
Concerning the EXAFS data, N is the total Ca-O coordination number; R is the average Ca-O bond length, (J is lhe Debye-Waller
factor and X; is the best fit residual factor.

52
 --------------------------------------------------~ .. - .~

V.R. Alaste/aro et al. / Journal of Non-Crystalline So/ids 262 (2000) 19/-199 195

4. Discussion
03·
o 5imulation

,
J 0.2 --- Experimental
4.1. Glassy and crystalline CaSiOJ
e 0.1

e Table 1 shows that the results of differen t

0.0
f i.It studies, using different structural techniques, for
·0.1
the glassy and crystalline samples agree very well,
02 excepting the results of Taniguchi et aI. [14] and
·0.3
Yin et aI. [15]. Taniguchi et aI. reported a mean
bond length of 2.49 Á (obtained using EXAFS)
·04
2 4 6 8 10 12 and 2.44 Á (obtained using XRD) for the Ca-O
K(A")
pair. yin et aI. [15]reported a meanobond length af
.r ! 2.43 A (four pairs of Ca-O at 2.34 A and two pairs
:-
y
Ii 11i~.2. Fourier filtered experimental signal (line) and best fit
(doIs) for the CaSiO) glass.
of Ca-O at 2.54 Á). In both cases, the Ca-O mean
bond length is larger than all other results pre-
.e • sented in Table 1.
•r
st
i 0.5,--------------------, We present a local order study of a crystalline
CaSiO) sample, obtained on heat treating a glass
n for the first time. According to XRD experiments
Na,Ca,Si,O, Glass
)-
0.0 [16] in crystalline CaSiO), three distinct Ca sites
~- exist; six oxygens lie between 2.22 and 2.54 Á, with
lill
i a mean [Ca06] bond length of 2.38 Á, in each site.
61
, •• ·0.5 However, one of the three distinct Ca sites has a
Na,Ca,Si,O, Crystalline further oxygen neighbor at 2.64 Á. Assuming that
~~ an equal proportion of the three sites exist, we
te found an average coordination number of 6.33,
-1.0
~t with a mean bond length of 2.39 Á. The structural
19
I
4 6 10 12
parameters N and R determined by EXAFS are in
tn (I)
I good agreement with literature data of XRD [16],
EXAFS [17-19] and Neutron diffraction [20].
I
0.7~---------------. Regarding the glass structure, four different
--Na,Ca,Si,O, Crystallized
----- Na,Ca,Si,O, Glas$ studies should be mentioned; Yin et aI. [15] using
XRD, the detailed work of Gaskell et al, [20] using
neutron diffraction, Abramo et aI. [21] using mo-
lecu1ar dynamics and Taniguchi et aI. [14] using
EXAFS and XRD. Our results are in good
agreement with the results of neutron diffraction
(N = 6.15 and R = 2.37 Á, respectively) and with
the XRD results for the coordination number
(6.0). As mentioned before, the values of Ca-O
2 4 6
mean bond 1ength found by Taniguchi et aI. [14]
(b) RIA)
«
and by Yin et aI. [15] using EXAFS and XRD are
Ftg. 3. (a) Ca K-edge EXAFS raw spectra and (b) corre- larger than those of other studies.
sponding Fourier transforrn for glassy and crystallized The lower amplitude intensity observed in the
NalCalSiJ09. first peak ofthe glass Fourier transform (Fig. ljb)
is due to a higher disorder of the Ca-O neigh-
teso the first Fourier transform peak, are compared borhood (Debye- Waller factor of the glass sample
ller with literature values in Table 3. equal to 0.13 Á, Debye-Waller of the crystallized

53
196 v.R. Mastelaro et ai. I Journal of Non-Crystalline Solids 262 (2000) 191-199

Table 2
Structural parameters obtained for Na]Ca1Si309 samples'
Rc.-O (Á) O"c.o Á (EXAFS)
1
Samples Nc.-O X; Technique References
(±0.3) (±0.01) Á (±0.01 Á)
Na]Ca]Si309 mineral 5.8 2.40 XRD [12]
Crystallized from glass 5.5 2.34 O.ll 0.12 EXAFS This work

Glass 5.5 2.35 O.ll 0.15 EXAFS This work

aln Ref. [12] the number of neighbors, Nc._o, and the mean bond length, Rc.-O, are the average values of the 4 different calcium sites.
N is the total Ca-O coordination number; R is the average Ca-O bond length, O" is the Debye-Waller factor and X~ is the best fit
residual factor.

0.5,.-------------------, Summarizing, the EXAFS results show that the

CaSi03 glass and crystal have similar first shells
coordination around caJcium atoms (6.0 and 5.8,
».

respectively), similar Ca-O mean bond length

0.0
(2.36 and 2.37 Á, respectively) and that the glass
structure presents a higher local disorder when
compared to the crystalline one.
-0.5 Thus, if the assumption of Refs. [1-3) con-
cerning the nucleation tendency is valid, our
EXAFS results indicate that volume nucleation
-1.0+--......--~-__,--~-.,._-~-__,_-~ should be detected in CaSi03 glass. lndeed, the
4 10 work of Gránásy et al, [24] demonstrates that this
Ia) is the case.

8
-- PbSIO~ ClymUlne
- - •••• PbSIO Glass
~
The Na2Ca2Si309 crystalline sample presents a
complex crystallographic structure. According to
Ohsato and Takeuchi [12], the calcium ions are
distributed between four different sites Ca(l),
Ca(2), Ca(3) and Ca(4). Except for the Ca(3) site,
which has two oxygens at 2.39 Á, two at 2.31 A
o
o 2 6
and two at 2.33 Á (6 neighbors), for the three other
(b) RIA) sites, there are 7 oxygen at different distances,
varying between 2.28 and 2.58 Á for the Ca(l) site
Fig. 4. (a) Pb Lm-edge EXAFS ••raw spectra and (b) corre-
sponding Fourier transforrn for glassy and crystaJlized PbSiOJ• and 6 oxygen at different distances, varying be-
tween 2.31 and 2.51 A, for the Ca(2) site and 5
oxygen at different distances, varying between 2.21
sample equal to 0.11 Á). The value of the Debye- ' and 2.55 Á for the Ca(4) site. We did not take into
Waller factor obtained for the glass is in agreement account distances larger than 2.70 Á. Assuming
with thatfound by Li et al. [22] i.e., a = 0.14 Á for that an equal proportion of the three sites exist, we
a CaSi03 glass doped with 3% of A1203. Similar found an average coordination number of 5.8 ox-
va1ues were reported for other silicate glasses [23]. ygen, located at an average distance of 2.40 Á.

54
 V.R. Mastelaro el a/o I Journa/ of Non-Crystalline So/ids 262 (2000) 191-199 197

Table 3
Structural pararneters obtained from different analysis of PbSi03 sarnples"
Samples Npb-O Rpb_O (Á) opbOÁ (EXAFS) Technique References
(±0.3) (±0.01 Á) (±0.01 Á)
es
PbSi03 crystalline 3.0 2.35 XRD [25]
Alamosite
PbO-Si02 glass 3.0 or 4.0 ESR [26]
PbO-Si02 glass Pb03 XRD [27]
units
(PbOMSiOJ)I_x 4.0 2.30 EXAFS [28]
ites. glasses
t fit PbSi03 crystallized from 3.0 2.35 0.02 0.20 EXAFS This work
glass
PbSi03 glass 4.2 2.34 0.01 0.25 EXAFS This work
"ln Ref. [25) the number of neighbors, Npb-o, and the mean bond length, Rpb-o, are the average of the 3 different lead sites. N is the
total Pb-O coordination number; R is the average Pb-O bond length, ais the Debye-Waller factor and X~ is the best fit residual factor.

he
:lIs
.8, Table 2 shows that the glass and the crystal have
th similar Ca-O coordination shells. Our EXAFS
.ss values are in good agreement with the values for
~n Combeite, determined by XRD (12). The Debye-
Waller facto r (rr) of the glass is similar to that
1- found for the CaSi03 glass and for other silicate
ir glasses having complex structures, such as
n CaAhShOg (rr = 0.17) and CaMgSb06 (a = 0.10)
te [23].
IS In summary, the EXAFS analysis show that the
local structures around the Ca atoms in glass and
crystal are quite similar: N = 5.5 for glass and
crystal, the rnean bond lengths are quite similar
(2.34 and 2.35 A for glass and crystal, respectively)
and the Debye--Waller factors - that gives an idea
of the topological disorder - are also vsímilar
(a = 0.1I A). Thus, the Ca local structures of
glassy Na2Ca2Sh09 and its homologous crystal-
line phase are quite similar. This fact confirms the
suggestions of [1-3] concerning the nucleation
ability, because Na2Ca2Si309 cIearly crystallizes in
the bulk [1]. Our previous experience with this
glass and the treatments made for the EXAFS
experiments confirmed that volume crystallization
is easil y observed. « anions exist. In order to model their XRD data,
4.3. Glassy and crystalline PbSi03
According to the work of Boucher and Peacor
[25], there are three different lead sites in the
PbSi03 structure: Pb(I), Pb(2) and Pb(3). The
coordination numbers for Pb(l), Pb(2) and Pb(3)
sites are 3, 4 and 4, respectively. For each of
the three Pb atoms, the oxygen distances lie in the
range 2.2-2.3 A, with another one or two in the
range 2.45-2.60 A. If we consider only distances
below about 2.50 A and assuming that an equal
proportion of the three sites exists, an a verage
value of 3.0 oxygens as first neighbors for the lead
atoms, located at an average distance of 2.35 A, is
found. We believe that the first Pb-O shell ac-
cessed by EXAFS contains only Pb-O bonds be-
low 2.50 A. In fact, as can be observed in Table 3,
our EXAFS results for crystalline PbSi03 agree
well with this supposition.
Regarding the local order around lead in
PbSi03 glass, the ESR analysis of Hosono et aI.
[26] shows that the coordination number of lead
in PbSi03 glass is three or four. No quantitative
information about the local structure around lead
was given in that work. The structure of PbSi03
glass was also analyzed by Imaoka et al. using
XRD [27). They found that only Pb03 pyramidal
units, interconnected with several kinds of silicate
they assumed two non-equivalent Pb-O distances
in the Pb03 pyramidal units, as being equal to 2.25
and 2.45 A, respectively. Finally, the local struc-
ture around Pb in (PbO)r(Si02)I-x glasses, with
composition varying from 38.6 to 83.7 mol% PbO,

55
1% VR. Mas/e/aro e/ alo / Journal of Non-Crystalline Solids 262 (2000) 191-199

were investigated by Montenero et aI. [28] using Acknowledgements

EXAFS. They found that the local structure of
these glasses is similar to that of lead oxide, PbO The authors are grateful to CNPq, PRONEX
(the Pb ion is fourfold coordinated to oxygen). No and FAPESP (Proc. 96/12841-2 and 99/00871-2)
information concerning the Debye-Waller factor for the financial support of this research. The
was given. Therefore, our results for the glass are critical comments of Vladimir Fokin, Boris A.
in good agreement with those obtained by Mon- Shakmatkin, Natalia M. Vedishcheva, from St.
tenero et al. [28): lead ions in the glass sample are Petersburg and of Alain Ibanez from Grenoble
fourfold coordinated by oxygen at an average were fully appreciated.
distance of 2.34 A.
In fact, if we consider the experimental error
(±O.OI Á), the Pb-O mean bond lengths are equal References
(2.34 and 2.35 A, for crystal and glass, respec-
tively). The Debye-Waller factors for glass and [1] E.D. Zanotto. 1. Non-Cryst. Solids 89 (1987) 36l.
(2) E. Muller, K. Heide, E.D. Zanotto, Z. KristaIJogr. 200
crystal are also equal. The coordination numbers,
(1992) 287.
however, are significantly different (NpbO = 4 ± 0.3 [3] E. Muller, K. Heide, E.D. Zanotto, 1. Non-Cryst. Solids
for the glass, whereas it is equal to 3 ± 0.3 for the 155 (1993) 56.
crystalline sample). Thus, due to the different co- [4] Powder Diffraction File No. 27-88, ICPDS-ICDD 1983.
ordination numbers, the Pb environment in the [5) Powder Diffraction File No. 29-782 lCPDS-ICDD 1983.
[6) Powder Diffraction File No. 22-1455 lCPDS-ICDD 1983.
glass and crystalline phase are different.
[7] B.K. Teo, EXAFS: Basic Principies and Data Analysis,
Therefore, as suggested in [2,3J, this glass Inorganic Chernistry Concepts, vol. 9, Springer, Berlin,
should nucleate heterogeneously, as it indeed does. 1986.
Only surface nucIeation is observed in heat-treated [8] A. Michalowicz, Logiciels pour Ia Chimie: EXAFS pour le
PbSi03 glass [1] (see also Section 2). Our previous Mac, Societe Française de Chimie, 1991, p. 116.
experience with this glass and the treatments made [9] 1.1. Rehr, I. Mustre de Leon, S.I. Zabinsky, R.C. Albers, I.
Am. Chem. Soe, 113 (1991) 5135.
for the EXAFS experiments confirmed that only [10] 1. Mustre de Leon, 1.1. Rehr, S.l. Zabinsky, R.C. Albers,
surface crystallization occurs in this glass in typical Phys. Rev. B 44 (1991) 4146.
laboratory conditions. [11] F.W. Lytle, D.E. Sayers, E.A. Stern, Physica B 158 (1989)
701.
[12] H. Ohsato, Y. Takeuchi, Acta Crystallogr. C 42 (1986) 934.
[13] 1. Leciejewicz, Acta Crystallogr. 14 (1961) 80.
[14] T. Taniguchi, M. Okuno, T. Matsumoto, J. Non-Cryst.
5. Conclusions Solids 211 (1997) 56.
[15) C.D. Yin, M. Okuno, H. Morikawa, F. Maruno, T.
Our EXAFS results for the local structures of Yamanaka, 1. Non-Cryst, Solids 80 (1986) 167.
[16] K.F. Hesse, Z. KristaIlogr. 168 (1984) 93.
the modifier cations in glassy and crystalline Ca-
[17] R.G. Geere, P.H. Gaskel, G.N. Greaves, 1. Greengrass, N.
SiOJ and PbSiO) are in good agreement with Binstead, in: EXAFS and near Edge Structure, Series in
published structural EXAFS, XRD and Neutron Chemical Physics, vol. 27, Springer, Berlin, 1983, p. 256.
scattering data. An EXAFS analysis of the local [18] 1.M. Combes, G.E. Brown Jr., G.A. Waychunas, in:
structure of Ca in the Na2Ca2Si)09 system is Proceedings of the XAFS VI Conference, York, UK,
1990, p. 312.
presented for the first time.
[I9] N. Lequeux, A. Morau, S. Philippot, P. Boch, 1. Am.
The relationship between the local structure of Ceram. Soe, 82 (5) (1999) 1299.
the modifier cations and thf nucleation ability, [20] P.H. Gaskell, M.C. Eckersley, A.C. Barnes, P. Chieux,
previously proposed, was corroborated by our Nature 350 (1991) 675.
EXAFS analysis for the three systems studied. We [21] M.C. Abramo, C. Caceano, G. Pizzimenti, 1. Chem. Phys.
96 (12) (1992) 9083.
confirmed that glasses that easily nucleate in the
[22] Z.Y. Li, P.H. GaskeI, 1. Zhao, S.R. Daresbury Ann. Rep.
volume present similar local structures to their (199111992) 79.
isochemical crystaIline phases, whereas glasses that [23] N. Binstead, G.N. Greaves, C.M.B. Henderson, Contrib.
only nucleate on the surface do not, Mineral Petrol. 89 (1985) 103.

56
 V.R. Mastelaro et al. I Journal of Non-Crystalline Solids 262 (2000) 191-199 199

[24] L. Gránásy, T. Wang, P.F. James, J. Chem. Phys. 108 (17)
(1998) 7317.
[25] M.L. Boucher, D.R. Peaeor, Z. Kristallogr. Bd. 126 (1968)
98.
[26] H. Hosono, H. Kawazoe, T. Kanazawa, Yogyo-Kyokai-
Shi 90 (1982) 544.
[27] M. Imaoka, H. Hasegawa, r. Yasui, J. Non-Cryst. Solids
85 (1986) 393.
[28] A. Montenero, L. Dimesso, G. Antonioli, P.P. Lottiei, G.
Vlaie, in: Proeeedings of the 2nd European Conference on
Progress in X-ray Synchrotron Radiation Researeh, 1990,
p.667.

57
 ]OURNAL DF

NON·CRYSTALLINE SOLIDS
E1SEVIER Journal of Non-Crystalline Solids 273 (2000) 8-18
www.elsevier.comflocate/jnoncrysol

29SiMAS-NMR studies of Qn structural units in metasilicate

glasses and their nucleating ability
J_ Schneider a, V_R. Mastelaro a,*, H. Panepucci a, RD_ Zanotto b

• Instituto de Física de São Carlos, Universidade de Sao Paulo, CiP, 369, 13560-970 São Carlos, SP, Brazi!
b Departamento de Engenharia de Materiais, UFSCar 13565-905. São Carlos, SP. Brazi!

Abstract
The purpose of this work is to verify the possible existence of a relationship between the similarity of the local
structure of the network-forming cation Si4+ (Q" units and chemical shifts) in glasses and isochemical crystals and the
nucleating ability of these glasses. Four metasilicate glasses with widely different volume nucleation rates: Na2Ca2Si309
and Na4CaSh09 (very large), CaSiO) (intermediate) and CaMgSh06 (undetectably small) were chosen. We present
magic angle spinning nuclear magnetic resonance spectroscopy (MAS-NMR) data for Na2Ca2Sh09 and N~CaSi309
glasses and for their respective isochemical crystalline phases for the first time. Additionally, we repeat NMR rnea-
surements of glasses and crystals previously studied by other authors (CaSi03 and CaMgSb06) to test the consistency
of our experimental techniques and method of analysis. Different central chemical shifts of Q2 resonances in parent
glasses and their isochemical crystals were measured, indicating structural differences. The relative amount of Q" groups
in each glass was obtained from the deconvolution of the 29SiMAS-NMR spectra. The shape of the Q" distribution for
each system was considered as a measure of the similarity of the connectivities of Si04 tetrahedra in each glass with
respect to its isochemical crystal (which has only Q2 groups). A correlation was found between the shape of the Q"
distribution and the nucleation tendency of these glasses, indicating that similarities between the tetrahedra connec-
tivities in glass and isochemical crystal has a role in determining the internal nucleation tendency of the metasilicate
glasses studied. © 2000 Elsevier Science B.V. Ali rights reserved.

1. Introduction ature (Tg). For typical laboratory conditions, i.e.,

using centimeter size samples that are heat-treated
for a few hours above Tg, the second class of

I
One of the authors of this study [1] previously
proposed that there are two classes of stoichio- glasses only crystallize heterogeneously by surface
metric glass-forming systems: for the first, the nucleation or aided by nucleating agents. For this
volume (presumed to be homogeneous) nucleation second family of glasses, the expected tempera-
rates are experimentally dêtectable and the tem- tures of maximum homogeneous nucleation rates
peratures of maximum nucleation rate are the (calculated by the classical nucleation theory) are
same or greater than the glass transition temper- less than Tg• Thus, it is possible to classify the
crystal nucleating ability of different systems by
• Corresponding author, Tel.: +55-16 273 9755; fax: +55-16 this criterion .
271 3616. There are a number of questions concerning the
E-moi! address: valmor@if.sc.usp.br (V.R. Mastelaro), crystal nucleation kinetics in glasses. One of the

0022-3093/001$ - see front matter © 2000 Elsevier Science B.V. AlI rights reserved.
PU: S O O 2 2 - 3 O 9 3 ( O O ) O O 1 3 9 - 3

58
 J. Schneider et ai. I Journal ofNon-Crystalline
proposed hypotheses postulates that the nucleat-
ing tendency depends on the structural similarity
between the parent glass and its isochemical crys-
tal phase [2-7].
ln qualitative terms, the degree of structural
similarity between parent glass and resulting
crystal should affect the main thermodynamic pa-
rameters controlling nucleation; liquid-crystal
surface el!ergy (u) and thermodynamic driving
force (L\G), thus affecting the nucleation ability of
the system. The homogeneous nucIeation rate is
given by: I ~ (KljYJ)exp(-K"u3jTL\G2), where K',
K" are constants, YJis the viscosity and T is the
temperature. Hence, the surface energy dominates
due to its power-3 against power-2 for L\G. The
exact relationships of the structural parameters of
glass and crystal with surface energy and driving
force are not known; however, there should be
some optimum structural similarity. This idea is
cIarified if we analyze two limiting cases: (i) when
the structures of glass and isochemical crystal are
too different, then a should be too large, inhibiting
nucleation. On the other hand, in the hypothetical
case, (ii) when the structures of glass and crystal
are almost identical, a should vanish (which would
enhance nucleation), however, L\G should also
tend to zero, precluding nucleation.
To further test this hypotheses, in this article we
measure and compare the local structures of the
network-forming cation Si4+ (Qn groups) in four
metasilicate glasses that have different volume nu-
cleation rates, i.e.: Na2Ca2Si309 and Na4CaSi309
(largest), CaSi03 (intermediate) and CaMgSi206
(undetectable nucleation rates).
2. Literature review
We will sum up here some of the most relevant
studies that propose the existence of a structural
relationship between the parent glass and its iso-
chemical crystal and the nucleating ability of the
glass.
ln 1984, Schramm et al, [3) anaíyzed the devi-
trification of lithium silicate glasses using 29Si
magic angle spinning nuclear magnetic resonance
(MAS-NMR). Their studies concentrated on
glassy and crystalline Li20-2Si02 (LS2). They
Solids 273 (2000) 8-18 9
found that the most abundant species in LS2 glass
are the Q3 species (3 is the number of bridging
oxygen per Si04 tetrahedra). In fact, they obtained
the foIlowing distribution of Q" species: 22% Q2,
57% Q3, 14.6% Q4 and 6.4% QO+ Q'. As the LS2
crystal should have 100% Q3, the authors men-
tioned that the glass had a sufficiently large con-
centration of Q3 units to induce crystallization.
Moreover, the mean chemical shift p.'f. Q3 (-92
ppm) in lhe glass was similar to that observed in its
crystalline phase (-93 PPl11)' Based on the corre-
spondence of the chemicalshifts,they suggested
that the local silicon environments in both phases
are similar and concluded thatglasses of.the Li20-
Si02 system easily nucleate (which, in fact, they
do) because the local environment of silicon in the
glass is compatible with those in the isochemical
crystal. We will show later rhat this relationship
between the chemical shifts of glasses and crystals
that easily nucleate does not hold true for a
number of systems. .
Dickinson [4] compared the structure of glassy
and crystalline K2 TiSi309 and analyzed the hy-
pothesis that structural similarity/dissimilarity be-
tween the amorphous and crystalline phases could
affect nucIeation. Heat treatments of glassy
r
K2TiSi309 showed that only surface crystallization
I
took place. Using Raman spectroscopy and ex-
tended X-ray absortion fine structure (EXAFS), he
observed that there is a clear difference in the local
structures of crystalline and glassy K2 TiSi309•
Whereas the short-range structure of the crystal is
typical of metasilicates, with the intermediate
range order of ring silicates, the structure of the
glass does not have this type of local and inter-
mediate range structure. Instead, the glass con-
tains more polymerized species (three bridging
oxygens, BO), in addition to species with two BO.
Furthermore, he observed that the largest differ-
ence is in the Ti coordination, for which there is a
change in coordination from the regular octahe-
dron of the crystal to mixed 5- and 6-coordination
in the glass. Dickinson (4] attributed this large
difference in the structures of the crystalline and
glassy phases to the fact that this glass has an
undetectably small volume nucleation rate.
Looking at it from a more macroscopic struc-
~II
tural perspective, Zanotto and Muller [5]proposed
59
10 J. Schneider et ai. I Journal of Non-Crystalline Solids
a simple method to predict the nucleation tendency
in glass. Their approach was based on the follow-
ing argument: if the structures of both glass and
isochemical crystal phases are similar for compo-
sitions that nucleate homogeneousIy, then the mass
densities, p, of glass and crystaI must be similar.
On the other hand, Iarge differences between the
densities of glass and crystal should be expected for
compositions that nucleate heterogeneously. After
analyzing the densities of isochemical glassy and
crystalline phases for various stoichiometric sys-
tems belonging to both famiIies, they concluded
that if the density difference between a glass and its
crystalline phase is significant (> 10%), it will most
likely only nucleate heterogeneously. On the other
hand, if the densities are comparable, homoge-
neous crystal nucleation could occur. ln fact, they
found some exceptions to this rule because some
glasses that only nucleate heterogeneously also
have similar densities in the crystal phase. They
ascribed these exceptions to the fact that compa-
rable densities do not imply similar structures for
glass and crystal, but that the opposite is true, i.e.,
different densities denote distinct structures. Thus,
they concluded that similar densities for glass and
isochemical crystal is a necessary but not sufficient
condition for structural similarity and, inferen-
tially for homogeneous nucleation.
Using the classification proposed in Ref. [5], for
the metasilicate glasses considered in this study we
find that glassy and crystalline Na2Ca2Sh09,
Na4CaSi309 and CaSi03 have density differences
(Apf Pglass) of 1.8%, 3.8%, and 6.6%, respectively,
while this difference is 18.5% forCaMgSh06. In-
deed, volume nucleation is easily detected in the
first three systems, while the last only has surface
nucIeation on laboratory scale. However, as com-
parable densities for glass and crystal donot un-
equivocally imply similar structures, it would be
interesting to obtain quantitative structural data
for the above systems. We will show later that the
local structures of the network modifier cation, Ca,
have been determined in tw.ç>of these systems.
Hence, in this work, we extend the structural
analysis to the network-forming atoms, Si, for
these systems.
, Muller et al. [6] tested the hypotheses that
structural similarity at a molecular levei between
60
273 (2000) 8-18
glass and crystal should favor volume nucleation.
Literature data for the molecular structures ar
several silicate glasses that nucleate homoge-
neously and heterogeneously were critically ana-
lyzed. A comparison was made with the structures
oftheir equilibrium crystalline phases. Muller et ai
did, in fact, observe that for glasses in which ho-
mogeneous nucleation occurs, the structural ar-
rangements in glass and isochemical crystal appear
to be similar while for glasses displaying hetero-
geneous nucleation, the local structures of glass
and crystal is different. In short, they concluded
that it is possible to predict the nucleation ten-
dency by comparing the molecular structure of the
glass and the phase crystallized from such glass.
However, we emphasize that only available liter-
ature data resulting from a number of different
studies and often employing different structural
techniques, were used in the study of Muller et al.
[6).
Recently, aiming at more systematic results
regarding the short range structure around
the network modifiers in glasses and crystals,
Mastelaro et al. [7] undertook an EXAFS anal-
ysis of calcium and lead in three different silicate
glasses: CaSi03 and Na2Ca2Si309 (for which
volume nucleation is observed) and PbSi03 (for
which nucleation only occurs on samples surfac-
es). The results indicated that, for the two glasses
that nucleate internally, the local structure of Ca
was similar to its short-range order in the corre-
sponding crystalline phases. On lhe other hand,
the short-range order of the glass that only nu-
cleates heterogeneously differed from that of its
isochemical crystal. Hence, the proposed rela-
tionship between the local structure and the nu-
cleation mechanism was corroborated by the
EXAFS analysis.
As demonstrated in the summary of previous
research, there is growing evidence to support the
hypotheses that glasses that have detectable ho-
mogeneous nucleation rates have local structures
similar to the crystalline phases formed in them,
and vice versa. However, quantitative structural
information using the same experimental tech-
nique for glasses and isochemical crystals (syn-
thesized by devitrification), for both classes of
systems, were only obtained for the network
 J. Schneider et aI. I Journal of Non-Crystalline Solids 273 (2000) 8-18 II

100
modifiers. Therefore, the purpose of this work is to
make a quantitative analysis of the short-range --Na.,Ca.,S~09
_ POF 22-1455
structure around the network-forming cation Si4+ 80

(Q" groups). With this information, we hope to

verify the possible relationship of these groups in :; 60
~>,
glass and corresponding crystal with the nucle-
ation tendency. Four stoichiometric metasilicate ""C/)
c: 40

glasses having different volume nucleation rates ~

are considered. Quantitative 29Si MAS-NMR 20

spectra were obtained for these glasses and their

corresponding crystalline phases obtained by
15 20 25 30 35 40 45 50 55 60
devitrification. 2e
100

3. Experimental _ POF37-28
80
-- N,\CaS!Og
The glasses were prepared by melting homo- -..
::i 60
geneous mixtures of analytical grade reagents in ~
lOO-ml Pt crucibles, in electric furnaces. The melt- z-
.~ 40
ing temperatures ranged from 1450°C to 1550°C, CD
E
with a hold time of about 2 h. The liquids were then 20
cast between cold steel plates and manually pressed
at an estimated cooling rate between lOO°C/s and o
10 15 20 25 30 35 40 45 50 55 60
500°C/s. To obtain fully crystallized samples, the
specimens were subjected to nucleation and devel-
ze
opment treatments, which were chosen after a dif- Fig. I. XRD patterns of Na2Ca2Si)09 and N~CaSi)09
ferential scanning calorimetry (DSC) analysis. The crystalline phases.
Na2Ca2Sh09 glass sample was heated at 600°C for
24 h and then at 690°C for 20 min; the Na4CaSi)09
glass sample was heated at 505°C for 12 h and then High resolution 29Si-NMR spectra were ob-
at 596°C for 8 h; the CaSi03 glass sample was tained in magnetic field of9.4 T, at a frequency of
heated at 725°C for 120 h and then at 885°C for 12 79.45 MHz, with a spectrometer (Varian Unity
h and, finally, the CaMgSi206 glass sample was INOVA). Measurements were carried out under
heated at 874°C for 22 h and then at 877°C for 25 h. magic-angle sample spinning (MAS) of up to
The four metasilicate glasses measured in this work 5 kHz, using a 7 mm wide-body CP/MAS probe
have different internal nucleation rates: (from Varian) and 7 mm zirconia rotors, The
Na4CaSi309 (Imax ~ 1014 m ? S-I), Na2Ca2Si309 spectra were obtained from Bloch decay (BD)
(~lOl2 m" S-I), CaSi03 (~106 m-3 S-I) and signals after n/2 pulses of 4 us length. Relatively
CaMgSi206 « 103 m-3 S-I) [I]. long recycle times were used to avoid any possible
The crystalline phases obtained through devi- effects of differential relaxation across the inho-
trification were measured by X-ray diffraction mogeneously broadened NMR line. No saturation
(XRD). Diffractograms were obtained in an au- of the NMR signal was detected using recycle
tomatic difractometer (Rigaku Rgtafíex model times exceeding 250 s. Up to 600 free induction
R U200B), with nickel filtered CuKa radiation decay (FID) signals were collected and averaged,
(1.540 Á). Figs. 1 and 2 present the XRD patterns The resonance line of a. polycrystalline kaolinite
of the Na2Ca2Si309 and N~CaSi309 (Figs. I(a) sample was used as the secondary external stan-
and (b) and CaSi03 and CaMgSh06 (Figs. 2(a) dard for referencing chemical shifts (-91.2 ppm in
and (b». respect to tetrametilsilane, TMS).

61
12 J. Schneider et ai. I Journal of Non-Crystalline Solids 273 (2000) 8-18

100r-------------r-------------------,
•• POF 43-1460
80 --CaSiO, "Si MAS·NMR

20

~ 20 Z ~ • ~ a ~ ~ 80
.b
29 'ü;
c CaSi,O.
Q)
(Wollastonite)
HlO E
-CaMgSi,O,
c::::J
POF 38-466
80

:;; 60
.!!.
~
.,
'iij
c 40 Na2Ca,Si,o.
~
-75 -80 -85 -90 -95
20
Chemcal Shift [ppmfTMS]

Fig. 3. High resolution 29Si-NMR spectra of crystalline sarn-

o
1. 1& 20 60 pies. Arrows points to resonance !ines associated with impurity
28 phases present in the specirnens (see text for details).

Fig. 2. XRD pattems of CaSiO) and CaMgSh06 phases.

the experimental data. The attribution of NMR

4. Experimental resuJts
The XRD pattems presented in Figs. 1 and 2
were indexed according to the .JCPDS powder
diffraction files [8-11]. However, one or two crys-
tallographic peaks in the XRD could not be in-
dexed, indicating that the solids obtained after the
crystallization could contain minor quantities of
other crystalline phases. We used the crystalline
compounds as a reference for the interpretationof
the NMR spectra of the glasses.
Fig. 3 shows the high-resolution 29Si-NMR
spectra of the four crystaUtnesamples. A number
of resonance lines can be observed, with a full with
at half maximum (FWHM) typically less than
2 ppm. Table 1 shows the results for the isotropic
chemical shift and the integrated intensity ofeach
line, obtained from a multiple lorentzian fitting to
lines to Qn groups was done by considering the
stoichiometry and crystallographic data of each
material. Fig. 4 shows the 29Si-NMR spectra of
the corresponding samples. For each glass, a res-
onance line covers the chemical shift range of sil-
icon in several Qn groups with n = O, ... ,4. A
deconvolution of the spectra was performed to
obtain the Q" distribution, using the common as-
sumption of Gaussian distributions of isotropic
chemical shifts for each type of Q" unit [12-14].
Unfortunately, overlapping of Q" resonances fre-
quently occur, giving co-variances between best-fit
intensity parameters and, consequent1y, to uncer-
tainties regarding the reiative populations of the
Q" species. To minimize these problems, some
physical criteria were imposed on the fitting pro-
cedure to discard unrealistic numerical results.
Thus, FWHMs of individual Gaussians were re-

62
 1. Schneider et ai. I Journal of Non-Crystalline Solids 273 (2000) 8-18 13

Table I
Isotropic chernical shifts and line intensities of the 29Si-NMR
"SiMAS-NMR
lines of the crystalline sarnples (possible identification of minor
phases are discussed in the text)
Crystal Center (pprn) Area (%) CaMgSip.
(Oiopsite)
Na2Ca)Si309 Q2 -88.2 ± 0.3 -'
-90.4 ± 0.3 _li

-92.8 ± 0.3 _<I

üi" casr,o,
Q2 b -76.7 ±0.2 7± I ""c (Wolfaslonile)
:J
Q2 -85.0±0.3 -a >.

-87.0±0.3 _a Jg
:õ
~
Q2 c -83.5 ±0.4 8±1 ~
.,
U)
c
Wollastonite Q2 -87.8±0.2 28±5
-88.9 ±0.2 29±5 E
-89.5 ±0.2 35±5

Q2 d -82.0±0.5 5±2
Diopside Q2 -84.0±0.2 77±4
Q2 e -88.0 ±0.5 18±3
a Strongly overlapped.
b Impurity phase, possibly Na2Si03'
c Impurity phase, pseudo wollastonite. -60 -70 -60 -90 -100 -110 -120
d Irnpurity phase, orto-ensteatite.
Chemical Shift [ppmffMS]
e Impurity phase, wollastonite.
Fig. 4. High resolution 29Si-NMR spectra of glassy samples.

stricted to less than 20 ppm, based on the chemical
shifts observed in different crystalline silicates (15].
In addition, the centers of adjacent Gaussians
from different Q" units had to be separated by
more than 5 ppm. The number of Gaussians used
in each case was determined by the chemical shift
range spanned by the NMR spectrum and other
spectral features, such as partially resolved bumps
or asymmetries. With these criteria, up to four
partially overlapped Gaussian distributions were
needed to deconvolute these spectra. Equivalent
acceptable fittings were obtained from deconvo-
lutions with three and four Gaussian functions. To
choose the physically significant fittings, stoichio-
metric criteria and literature data of chemical
shifts were considered.
For glassy wollastonite and diopside, though
smaIl, there is an appreciable spectral intensity in
the chemical shift range of Q4 (-115 to - 100 ppm),
which is rarely overlapped with Q3 [15]. It is,
therefore, possible to ensure the existence of Q4
groups in these glasses. Conversely,. the NMR
spectra of glassy Na2CazSig09 and Na4CaSh09
samples indicate that there are no Q4 groups in
their networks. Additionally, the identification of
Q2 resonances was done assuming that the distri-
bution of Q" species in metasilicates must -satisfy
the stoichiometric condition QO+ Q I = QJ + Q4, so
that the average is always Q2. Thus, the remaining
resonances were readily assigned.
Table 2 shows the fitted parameters for the
chemical shift distributions of each Q" group in the
four samples. For wollastonite and diopside, four
Gaussian deconvolutions were performed because
of the presence of Q4 groups, while for
Na2Ca2Si309 and Na4CaSi309 samples, only three
Gaussian functions were used. If a fourth Gauss-
ian is added, its chemical shift cannot be assigned
to QOor Q4 species. Separations of approximate1y
7 ppm between the centers of Q" and Qn+1 distri-
butions were obtained in ali glasses, with FWHM
varying from 7 to 12 ppm. .h\.lthough the
QO+ QI = Q3 + Q4 constraint was not imposed in
the fittings, Table 2.shows that this condition was

63
 14 J Schneider et al. I Journal of Non-Crystalline So/ids 273 (2000) 8-/8

Table 2
Chemical shifts, line-widths and integrated intensities of the 29Si-NMR lines corresponding to different Q" units in the four glasscs:'
Composition Crystal Ó (pprn) Glass Chemical shift
~ FWHMb (ppm) ditference for Q'
Center (ppm) Area (%)
±0.5 pprn ±I pprn in crystal - glass
b-center (ppm)
Na2Ca2Si)09 QI -73.4 9 16±5
Q' -88.21-90.4/-92.8 -80.0 10 72±8 8.2110.4/12.8
Q3 -88.5 9 12±5

Na4CaSi)09 QI -70.0 7 14±5

Ql -85/-87 -77.5 8 67±8 7.5/9.5
Q3 -85.3 9 i9±7

Wollastonite QI -75.6 10 20±5

Q2 -87.8/-88.9/-89.5 -82.6 11 64±8 5.2/6.3/6.9
QJ -91.8 il 14±5
Q4 -i03.3 li 2±i

Diopside QI -77.5 11 28±8

Q2 -84 -84.0 il 43± IO O
QJ -92.7 i2 25±6
Q4 -103.0 i2 4±1
• For comparison, the values of isotropic chemical shift Ó for the main resonance lines in crystals and their difference respect to the Q2
central chemical shift in glasses are also shown,
b FWHM: fulI width at half maximum.

met for alI samples, within the uncertainty of the chemical shift seems rather shielded to correspond
numerical procedures. For clarity, we will describe to QI chain terminators generated by absent tet-
and compare the results for each glass and crystal rahedra. Furthermore, the existence of a number
separately. of non-bridging O-H groups can be disregarded
since no 29Si-NMR signal was detected after
4.1. casto, (wollastonite) IH_29Si cross-polarization experiments performed

on this sample. Based on the chemical shift, 4 pprn

This glass has a small volume nucleation rate.
At room temperature, the stable crystalline forms
of CaSi03 contain 'dreier' single chains of silicate
tetrahedra [16-18]. Thus, there are three different
Ql sites for silicon in the unit cell. The 29Si-NMR
Iess shielded than the rnain resonance lines, we
might suggest AI substitution for Si, giving rise to
a Q2(l AI) silicon resonance. Nevertheless, this
substitution is an unlikeIy possibility because no
27 Al-NMR signal was obtained from Bloch decay

spectrum of the crystalline wollastonite does, in-
deed, show three resolved resonance lines at -89.5,
-88.9 and -87.8 ppm associated with these sites,
with 93% of the total integrated intensity observed.
These chemical shifts agree with previous obser-
vations of Smith et aI. [19] and Mãgi et aI. [20],
who detected a broader line centered at around
-89 ppm. The remaining 7% of the silicon con-
tributing to the NMR spectrum corresponds to a
smaller resonance line at -83.5 ppm. Its isotropic
experiments, even after considerable signal aver-
aging. Then, the -83.5 ppm resonance probably
originates from a small fraction of another of the
known forms of wollastonite, differing in the ar-
rangement of the siIicate chains . .ln fact, Mãgi
et aI., reported for pseudo-wollastonite a reso-
nance line exactly at -83.5 ppm [20].
The following distribution was obtained from
the deconvolution of the NMR spectrum of the
glassy wollastonite sampIe: 20% QI, 64% Q2, 14%

64
 1. Schneider et al. I Journal
Q3 and 2% Q4. Thus, the largest amount of silicon
in glass and crystal samples corresponds to Q2
groups, It is useful to compare these fractions with
those obtained by Zhang et aI. [12,14]. They ap-
plied a 2D 29Si-NMR technique to correlate iso-
tropic (MAS) and anisotropie (off-MAS) speetra
of glassy wollastonite. They obtained more accu-
rate percentages for the Q" distribution with re-
speet to the usual method of fitting the ID MAS
spectrum, They obtained QO = 0.7 ± O.l%, QI =
19.3±0.3%, Q2=54.7±0.3, Q3=24.1±0.5% and
Q4 = 1.1 ± O.1%, which are comparable with our
results (the difference for the most abundant spe-
cies, Q2, is about 15%). Considering that the 2D
approaeh should be more precise, we assume that
the Q" determination method used in this work
presents a maximum discrepancy of 15%, based on
the differenee for the Q2 fraction. As for the center
of chemical shift distributions, the ones shown in
Table 2 compare with those reported by Zhang
et aI. [14]: -74.6 ppm for QI, -81.7 ppm for Q2 and
-90.4 ppm for Q3.
4.2. CaMgSi20ó (Diopside)
This sample has an undetectable homogeneous
nucleation rate. Crystallographic data of diopside
indicates that the silicate tetrahedra are arranged in
'infinite' linear chains, with only one non-equiva-
lent silicon site in the unit cell [21]. Fig. 3 actually
shows that the NMR spectrum of the crystal has
three resonance lines. The most intense line (77%),
centered at -84 ppm with 0.5 ppm FWHM, cor-
responds to Q2 units. Two other smaller and
broader lines are observed at -88 ppm (18%) and
-82 ppm (5%), which are ais o in the Q2 chemical
shift range. Previous NMR measurements made by
Smith et aI. [19] and Mâgi et aI. [20] have shown a
single resonance centered at -85 ppm, a shift
comparable to that obtained in our measurement
for the strongest resonance, but with 5 ppm
FWHM. 27 AI NMR experiments were als o per-
formed on our sample; however, no appreciable
signal was detected, indicating tha't AI, if present,
are in concentrations toa small to be detected by
NMR. On the other hand, effects of hydration as
those observed by Peck et aI. [22] can also be
disregarded in this case, because no signal was
of Non-Crvstalline Solids 273 (2000) 8-18 15
\
observed after IH_29Si cross polarization experi-
ments. Therefore, we associated the smaller NMR
lines to sarnple inhomogeneities, possibly distorted
wollastonite (-88 ppm line) and orthoenstatite,
Mg2Si206, (-82 ppm line), the last one according to
the measurements reported in Ref. [20].
For the glassy sample, the distribution of Qn
units was: 28% QI, 43% Q2, 25% QJ and 4% Q4. To
the best of our knowledge, only one MAS-NMR
experiment has so far been reported for diopside
I
by Murdoch et aI. [23]. No quantifieation of the Q"
speeies was made. The maximum of the speetrum
was loeated at -82.0 ppm and, by analogy with the
observed ehemical shift of Q2 units in the erystal,
Murdoeh et al. eoncIuded that the dominant con-
neetivity in the network is Q2. The deeonvolution
of our speetrum gives 43% Q2 with a chemieal shift
at around -83.0 ppm. Therefore, our results agree
with those reported in Ref. [23].
As mentioned earlier, this glass has the largest
homogeneous nucleation rate in the studied set of
samples. In crystaIline Na2Ca2Si309, the silicon is
organized in six-membered rings with a C2 sym-
metry [24]. Thus, there are three different Si sites
with Q2 connectivity in the unit cell. Accordingly,
(Fig. 3) the NMR spectrum is composed of three
partiaIly resolved resonance lines centered at
-88.2, -90.4 and -92.8 ppm. Due to the partialIy
disordered structure of the Na2Ca2Si309 crystal,
where Na/Ca sites with fractional occupation
number were determined from XRD [24], the res-
onance lines are broader than those from wollas-
tonite and diopside. For the glassy sample, the
most satisfactory fitting gives the following distri-
bution of species: 16% QI, 72% Q2 and 12% Q3.
On the other hand, as seen in Table 2, resonance
lines in the crystaIline state appear more shielded
with respect to the center and width of the fitted Q2
distribution in the glass (-80.0 ppm).
This glass has a similar1y large homogeneous
nucleation rate. As far as we know, crystallo-
graphic data are not available for this material.
65
lú 1. Schneider et aI. / Journal of Non-Crystalline Solids 273 (2000) 8-18

However, since it IS a metasilicate, according to
stoichiometric considerations only Q2 groups
should exist in the homogeneous crystal. As shown
lu Fig. 3, there are at least two overJapping lines in
the NMR spectra, at approximately -85 and -87
"pm, which are associated with Q2 species. The
total width of the group of lines is 4 ppm, a width
ldentical . to that observed in the spectrum of
crystalline Na2Ca2Si)09. Thus, the presence of
additional unresolved 29Si lines cannot be disre-
garded in Na4CaSi309• AIso, occupational disor-
der in the crystalline structure may be responsible
for the broadening of the resonance lines, as in
Na2Ca2Si309. On the other hand, the NMR
spectrum of this crystal has a resonance (0.35 ppm
FWHM) centered at -76.7 ppm. In different
crystal samples prepared from the same glass rod,
this signal appeared with varying intensity (from
5% to 8%), without changes in the shape of the
other lines. The smaller FWHM of this minor
signal indicates that silicon atoms are located in a
more ordered structure, as compared with those
silicons contributing to the resonances at -85/-87
ppm. Based on this fact, we could assign the -76.7
ppm resonance to an impurity silicate phase, per-
haps Na2Si03, which has a single ~ resonance at
-76.8 ppm [20]. For the glassy Na4CaSi309 sam-
pIe, the following distribution of Qn units resuIted:
14% Ql, 67% Q2, 19% Q3.
a function of the central chemical shift for Q"
resonance. Na2Ca2Si)Og, Na4CaSi309 and wo-
Ilastonite glasses have similar Q" distributions, Q2
being the most abundant species (72%, 67% and
64-55%, respectively). From the overall shape of
these distributions, we propose that the short-
range order around Si4+ in the glass is more similar
in respect to its crystal phase (where only Q2
connectivities exist) for the systems with sharper
distributions.
Due to the presence ofQ4 uníts, wollastonite has
a slightly wider distribution than Na-coritaining
glasses. Diopside has a substantially smaller con-
tent afQ2 species (43%) and, consequently, has the
larger Qn dístribution among the studied glasses.
Therefore, glasses having the greatest nucleating
ability have a considerable amount of tetrabedra
having the same connectivity as in the crystal phase
and a smaller width of the Q" distribution.
On the other hand, there are significant differ-
ences between the chemical shifts of the Q2 species
ín the glasses and isochemical crystals. Resonance
in crystalline Na2Ca2Sh09, Na4CaSi)09 and wo-
Ilastonite are systematically more shielded than the
corresponding Q2 resonance in glasses. Table 2
shows that the chemical shift differences are con-
siderableand are even comparable to the FWHM
of the Qn distributíons in the glasses. Therefore,

80 Q2

5. Discussion
.. A·· Na.Ca,S~O
..+
•
Na.CaS~O,

60
• • .•
•.
. wollastonl1e
diopslde
As we have briefly described before, Schramm ~
et alo [3] reasonably proposed that the LS2 glass ~
"
c

(we emphasize that this composition is no! a me-

B
.E
-e
40 •
..
Q3
tasilicate) has a high tendency to homogeneous ~ ir
crystallization owing to: B'"
.E

(a) Local structural similarity between the glass

20
.•. • •
and crystalline phases, inferred from the simi-
larity of the chemical shifts for Q3 in glassy o
• Q'
. ..~" .••
A Q4

-65 -70 -75 -80 -85 -90 -95 -100 -105

and crystalline LS2. • Chemical Shift [ppmfTMSI
(b) High concentration of Q3 in the glass (great-
er than 50%). Fig. 5. Integrated intensities for the Gaussian Q" distributions
In the set of metasilicate glasses studied in this of isotropic chemical shift for each glass. The chemical shifts
correspond to the centers of the Q" distributions. The points
work, only criterion (b) applies (for Q2 instead of with sticks in the lower part ofthe plot show the positions ofthe
Q3). The overall Q" distribution can be visualized maín Q2 resonance lines observed in the corresponding iso-
in Fig. 5, where integrated intensities are plotted as chemical crystals.

66
 J. Schneider et al. I Journal of Non-Crystalline Solids 273 (2000) 8-18
there are appreciable structural differences in the
environment of Q2-units in glass and in crystal,
despite the similarity in the dominant connectivity
(Q2) of the network-forming cations, although it is
difficult to determine the extent of these differ-
ences. A deshielding effect on the silicon resonance
in glass can be produced by geometrical distor-
tions, such as longer Si-O distances or more acute
Si-O-Si angles, and by weaker bonds between the
modifier cation and O atoms (15]. It should be
noted that, even in crystalline Na2Ca2Si309, there
is a 4.5 ppm difference between two non-equivalent
silicon atoms in the same 6-membered ring, a dif-
ference comparable to some differences observed
j between crystal and glass. Quite surprisingly to us
1 is that the chemical shifts for the Q2 groupings in
I crystalline and glassy diopside are in elose agree-
I ment. It is impossible, however, to infer structural
similarity from this coincidence because agreement
between chemical shifts is only a necessary condi-
tion for local sirnilarity.
Therefore, similarity in the overall degree of
connectivity of the forming cations in glass and
crystal, inferred from the shape of the Q" distri-
bution, seems to be more important to internal
nueleation than a elose structural similarity
around the Q2 silicon atoms, prabed by the
chemical shifts. It is interesting to note (Fig. 5 and
I
Table 1) that the centers of QR distributions in
Na4CaSi)09 glass are systematically shifted by
approximately 3 ppm in respect to the corre-
t sponding ones in Na2Ca2SiJ09 glass. The same
I shift is detected for the Q2 species in the crystal:
!, the group of tines are centered at around -90 ppm
in Na2Ca2Si309 and at -87 ppm in N~CaSi309,
spanning a 5 ppm range in the spectrum. Based on
these results we suggest a similar structural rela-
tionship between the glass and the crystal phase
for both compositions,
The silicon-oxygen tetrahedra in the crystalline
phases are organized in Q2 groups, forming rings
(in Na2Ca2Si309 and possibly in Na4CaSi309) or
chains (in wollastanite and, diopside). Though
high-resolution lD-NMR can prabe the distribu-
tion of Si in terms of Q" units, it shows very little
about the topological relationships among these
groups. The local range of the ID-NMR experi-
ment, restricted to a typical radius of 1 nm, gives
little information about tetrahedral neighbors
outside the first coordination sphere. An explora-
tion ofmedium range connectivities in glasses with
extreme nucleation and qualitatively different Q"
populations, such as those observed in
Na2Ca2Sh09 and diopside, would be very useful.
ln summary, the systems that have the largest
nucleation rates (Na2Ca2Sh09 and Na4CaSi3ü9)
have the largest percentage of Q2 units (72% and
67%, respectively) and the smallest Q" distribu-
tions. The only other groupings in these glasses are
Ql and QJ. Wollastonite, having a lower nucle-
ation rate than Na-containing glasses, has a
smaller Q2 content (64%), but some amount of Q4
is detected in this glass, widening its Qn distribu-
tion. Diopside, displaying undetectable homoge-
neous nucleation, has an even larger Q"
distribution. Only 43% of the network-forming
cations are organized in Q2 units and, conse-
quently, a larger fraction of Ql, Q3 and Q4
species are present in this glass than in the other
compositions.
Finally, we stress that while the correlation of
the Q" distribution and nucleatian tendency in the
metasilicate glasses studied here is interesting, the
possibitity of a fortuitous coincidence should
not be dismissed. Literature data for other sys-
tems, for instance, disilicate« Li20· 2Si02 (LS2),
Na20 . 2Si02 (NS2), K20· 2Si02 (KS2), show an
opposite trend. ln the case of these disilicates, for
which the crystal phases are 100% Q3, the crystal
nucleation rates decrease from LS2 to KS2 while
the amount of Q3 increase in that arder: 63-71%
Q3 in LS2, 79% Q3 in NS2 and 86% Q3 in KS2 [25].
These findings deserve further attention, We hope
that the present study encourages similar analyses
of a broader set of metasilicate glasses, in order to
check whether there is or not a general trend for
metasilicates.
6. Conclusions
The difíerences in the chemical shifts of the
predominant Q2 species in glass and crystal indi-
cate short range differences between the silicon
environments, even in those systems displaying the
greatest nucleating ability. On the other hand, the
67
18 J. Schneider et ai. / Journal oI Non-Crystalline
shape of the Q" distribution, indicating the degree
of similarity between silicate connectivity in glass
and isochemical crystal, shows a correlation with
the internal nucleation tendency of the four me-
tasilicate glasses studied.
We should stress, however, that while the cor-
relation of the Q/I distribution and nucleation
tendency in the metasilicate glasses studied here is
quite interesting, the possibility of a fortuitous
coincidence cannot be dismissed. Literature data
for other families of glasses, for instance, disili-
cates, do not show the same trend. Thus, to gen-
eralize the present findings for metasilicates, and to
seek plausible explanations for the phenomenon,
additional effort should be directed to verify the
existence of the correlation in other metasilicate
systems.
Acknowledgements
1.S. thanks FAPESP (Fundação de Amparo à
Pesquisa do Estado de São Paulo, Brazil) for a
post-doctoral fellowship. The critical comments of
Boris A. Shakmatkin and Natalia M. Vedishcheva
from the Institute of Silicate Chemistry, St. Pe-
tersburg, are fully appreciated. Funding by CNPq,
PRONEX and FAPESP (grant no. 99/0871-2) is
deeply appreciated.
References
[I] E.D. Zanotto, I. Non-Cryst. Solids 89 (1987) 361.
[2] B.H.W.S. de Jong, W.S. Keefer, G.E. Brown, Ch.M.
Taylor, Geochim. Cosmochim. Acta 45 (1981) 1291.
••
Solids 273 (2000) 8-18
[3] C.M. Schramm, B.H.W.S. de long, V.E. Parziale, J. Am.
Chem. Soc. 106 (1984) 4396.
[4] I.E. Dickinson, in: Proceedings of the XV International
Congress on Glass, Leningrad, vol. Ia, 1989, p. 192.
[5] E.D. Zanotto, E. Muller, I. Non-Cryst. Solids 130 (1991)
220.
[6] E. Muller, K. Heide, ED. Zanotto, I. Non-Cryst Solids
155 (1993) 56.
[7] V.R. Mastelaro, E.D. Zanotto, N. Lequeux, R. Cortes,
I. Non-Cryst. Solids 262 (2000) 191.
[8] Powder Diffraction File No. 22-1455, JCPDS-ICDD 1983.
[9J Powder Diffraction File No. 12-670, ICPDS-ICDD 1983.
[10] Powder Diffraction File No. 11-654, ICPDS-ICDD 1983.
[li] Powder Diffraction File No. 27-88, JCPDS-ICDD 1983.
[12] P. Zhang, C. Dunlap, P. Florian, P.J. Grandinetti, 1.
Farnan, I.F. Stebbins, I. Non-Cryst. Solids 204 (1996) 294.
[13] I. Mahler, A. Sebald, Solid State Nucl. Magn. Res. 5
(1995) 63.
[l4] P. Zhang, P. Grandinetti, I.F. Stebbins, 1. Phys. Chern. B
101 (1997) 4004.
[15] G. Engelhardt, D. Michel, High Resolution Solid-State
NMR of Silicates and Zeolites, Wiley, Norwich, 1987.
[16] Y. Ohashi, L. Finger, Carnegie Institution of Washington:
Yearbook (CIWYA 75), 1976, p. 746.
[17] K.F. Hesse, Zeit. Krist. 168 (1984) 93.
[18] K. Mamedov, N. Belov, Dokl. Akad. Nauk SSSR 107
(1956) 463.
[19] K.A. Smith, R.I. Kirkpatrick, E. Oldefield, D.M. Hender-
son, Arn. Mineral. 68 (1983) 1206.
[20] M. Mâgi, E. Lipmaa, A. Samoson, G. Engelhardt , A.
Grimmer, I. Phys. Chem. 88 (1984) 1518.
[21] B. Warren, W.L. Bragg, Z. Kristallogr. Kristall. Kristall-
phys. Kristallchem. 69 (1928) 168.
[22] I.A. Peck, I. Farnan, I.F. Stebbins, Geochim. Cosmochim.
Acta 52 (1988) 3017.
[23] 1.B. Murdoch, J.F. Stebbins, I.S.E. Carmichael, Am.
Mineralogist 70 (1985) 332.
[24] H. Ohsato, Y. Takeuchi, I. Maki, Acta Crystallogr. C 42
(1986) 934.
[25] H. Maekawa, T. Maekawa, K. Kawarnura, T. Yokokawa,
I. Non-Cryst. Solids 127 (1991) 53.
\
1
68
~
]OURNAL OF

~ NON-CRYSTALLINE SOLIDS
~
EUlEVIER JournaJ ofNon-Crystalline Solids 304 (2002) 160-166
www.elsevier-.comllocate/jnoncrysol

Section 3. Structure and metallic glasses

Local order around of germanium atoms in GalOGe25S65

glass by EXAFS
S.H. Messaddeq "". V.R. Mastelaro a, A.Y. Ramos h,e, M. Siu Li a, D. Lezal c,
S.l.L. Ribeiro d, Y. Messaddeq d
a Instituto de Fisica de São Carlos, Universidade de São Paulo, c.P. 369, CEP: 13560-970, São Carlos, SP, Brazil
b Laboratório National de Luz Sincrontron, CP 6192,13083-970 Campinas, Brazil

c Laboratory of Inorganic Materiais IlC ASCR and ICT, Pelleova 24, Prague 6, Czech Republic

d Instituto de Química, UNESP, Araraquara, C.P. 355, CEP: 14801-970, Araraquara, SP, Brazil
e Laboratoire de Minéralogie, Cristallographie UMR CNRS 7590, France

AbstTact

Irreversible photoexpansion effect has been observed in amorphous GalOGezSS6Sglasses when its surface was ex-
posed to light with energy greater than the band gap, 3.52 eV. A volume change of about 5% was reached in bulk
samples by controlling ilIumination time and the laser power density. To understand the atomic scale processes of the
photoexpansion effect, extended X-ray absorption fine structure (EXAFS) spectroscopy has been used as a local probe
of the germanium environment in the glass samples before and after illumination. Modifications are observed in the
average coordination shell around Ge atoms in the illuminated sample compared to the non-illuminated one. For the
non-illuminated sample, the Ge coordination shell is described by a distorted tetrahedron of sulfur atoms at around
2.20 À. After illumination, the EXAFS signal can be explained by introducing an additional contribution to this av-
erage environment. Based on an analysis of the EXAFS data we proposed a two-shell model of 0.5 oxygen atorns at
2.01 Á and 3.6 sulfur atoms at a 2.20 À. The existence of Ge--O bonds in the glass after illumination was confirrned by
infrared measurements. © 2002 Elsevier Science B.V. All rights reserved.

1. Introduction [2], rheological properties [3], etching, and, in

Photostructural changes have been subject of
research for several years because they not only
reveal interesting properties about the structureof
disordered solids but may have technological po-
tential as well [1]. In chalqpgenide glasses these
changes are induced by exposing a sample to near
band gap light. They include changes in the density
'Corresponding author. Te1.: +55-162339823; fax: +55-16
273 9811.
E-mai! address:sandrahm@if.sc.usp.br (S.R. Messaddeq).
some cases, photocrystallization [4]. Depending
on experimental conditions, these changes can be
irreversible, i.e. the changes are pennanent after
irradiation, or reversíble in which case the changes
can be removed by annealing to the glass transi-
tion temperature, Tg•
In our previous investigation of the photoin-
duced changes in GaGeS glass, we reported an
induced decay of refractive index associated with
a decrease in the density through dilation which
increased volume when samples were exposed
to laser light with ). = 351 nm and power density

0022-3093/02/$ - see front matter © 2002 Elsevier Science B.V. Ali rights reserved.
PII: S0022-3093(02)0 1 o 13-X

69
 SH Messaddeq et ai. I Journal of Non-Crystalline Solids 304 (2002) 160-166
of 5 W/cm2 [5]. Moreover, we found that this
photoexpansion effect was irreversible [6]. Several
mechanisms have been proposed for the irrevers-
ible changes, in which the interaction with light
induces bond breaking and structural changes.
lrreversible changes, which are most reported in
as-deposited films [7,8], have been explained in
tenns of bond breaking models [9]. ln this work,
exposing the GalOGe2SS65samples to light having
above band gap energy increased the thickness of
the illuminated region.
The main network of germanium based chalc-
ogenide glasses is described as Ge04 tetrahedral
units connected to each other by comers and edges
[10]. The volume change during the irradiation
should be correlated to structural changes in this
glassy network. The objective of the present study
was to investigate these changes, by using EXAFS
as a structural probe of the local environment
around the germanium atoms.
2. Experimental
Glass samples were prepared by melting ele-
ments (Ga, Ge, and S) in a quartz ampoule
evacuated to 10-3 Pa. All procedures including
synthesis, distillation and sample production were
carried out in a closed system. After melting at 900
°C during 6 h the ampoule was removed, quenched
in water and annealed at the glass transition tem-
perature, Tg, of the sample around 400 "C. Rods
about 60 mm in length and 10 mm in diameter
were obtained after cutting the quartz ampoule.
Slices of around 1 mm in thickness were cut and
polishing by a pad charged with polishing particles
(Ce02, mean diameter of 0.5 um).
Before the extended X-ray absorption fine
structure (EXAFS) measurements, the glass sur-
face area, 10 mm in diameter and 1 mm thick, was
exposed through a mask that was in contact with
the sample surface. For the photoexpansion ex-
periment, the sample was exposed to ultra-violet
(UV)-light (351 nm) from a Kr+ ion laser at room
temperature. After 5 and 10 h illumination, the
mask was removed from the samples and the ele-
vation of the exposed area was measured using a
3D perfilometer instrument (Tencor, TFl). An
161
1
I
!~
image of the exposed area was obtained using an
atomic force microscope (AFM - Digital Instru- 1
ments-Nanoscope 3 A). 1
!
EXAFS measurements at the Ge K-edge
(11104 eV) were perfonned at the XAS beam line
of at the LNLS facility (Campinas, Brazil) with an
average beam voltage of 150 mV [11]. The mono-
chromatization of the incident beam was made
using a double Si(lll) crystal [11].
Firstly, we measured the EXAFS spectra ofthe
reference compounds, crystalline GeS2 (C-GeS2)
and amorphous GeS2 (a-GeS2), in which the Ge
sulfur surroundings are quasi tetrahedral and the
local structure is well known. The reference com-
pounds were than ground to powder and sieved to
obtain powders having a grain size between 15 and
20 um, For the EXAFS measurements, the sieved
powders were than deposited on a polymeric mern-
brane. The data collection was then performed in
the transmission mode using ionization chambers
as detectors. Each sample was subjected to; at
least, five different scans, which were averaged to
increase the quality of the experimental data.
The EXAFS spectra of a GalOGe25S65 sample
(slices of 10 mm in diameter and 1 mm of thick-
ness) was then collected in samples before and
after iIlumination (exposed 5 and 10 h). The in-
tensity of the incident beam was measured by an
ionization chamber whereas the beam originated
by the sample was collected using the total electron
yield mode [11]. Each sample was subjected at
least, to five different scans, which were averaged
to increase the quality of the experimental data.
The EXAFS spectra were analyzed by a stan-
dard procedure of data reduction [12], using
WinXAS code [13]: extraction of the EXAFS sig-
nal, X(k), the oscillatory part of the absorption
coefficient, then Fourier transformed which was
obtained after a J2 EXAFS weighting and a BesseI
apodization function (coefficient of 4) over 2.8-
12.0 Á -I. The Fourier transforms give a radial
distribution function, modified by the phase shifts
due to the absorber and backscattered atoms. The
contributíon of each shell is then backtransformed
into k-space. Structural parameters, N (coordina-
tion number), R (interatomic distance), and (J
(Debye Waller term) are obtained from least
squares fitting in k-space, using theoretical phase
70
162 Si H. Messaddeq et a!. I Journal of Non-Crystalline Solids 304 (2002) 160··/66
and amplitude functions obtained from FEFF6.01
software [14].
The IR spectrum of the surface of GaGeS
samples (slices of 10 mm in diameter and 1 mm
of thickness) before and after ilIumination, was
measured in the refiection mode using a spectro-
photometer Perkin-Elmer 684.
3. Results
3.1. AFM measurements
Fig. 1 shows the result of three-dimensional
profile performed on the surface of a GalOGe25S65
sample after irradiation exposed 5 h. As can be
seen in Fig. 1, the exposed part of the glass surface
is higher than their surroundings that were not
illuminated by the laser. The elevation of the ex-
posed surface, compared to the unexposed region
of the same sample, was measured as 250 um.
3.2. EXAFS spectra of the reference samples
Fig. 2 shows norrnalized experimental EXAFS
spectra, X(K), ofthe c-GeS2 and a-GeS2 references,
Non-
Exposed
Exposed
!
!
750
~1I11
o
Fig. C Three-dimensional representation of profile measurement of GaGeS showing the effect of the photoexpansion
structure of glass exposed during 5 b.
71
together with that of the aI1I1í(J(l~~SÍ'ld'UI\\!lu ",
observe that EXAFS spectra (11' Ih~thh~!iIWIIII'I"
are similar.
The Fourier transforms (11' UII~ IrX t\ I ~~ "I i,,1 I,'
presented on Fig. 2 are shown li, filU; li •• I" ,
peak in the Fourier transforrn, Iljl.iíllh'd !lI 'lI'
proximately 1.98 Á, is duc to lho tlt~\ Ullil dílllilll "I
shell around gerrnanium. Wc should IWW 11\!l1 11 I'
Fourier transforms are not COI'H1"h.ltl IHI 101,,1
shift parameters, so that the p\ljt1.k 11I".itlfllllJ 'I!'
shifted to smaller distance compared to IIUé i!,1i
distances, The corrected distances IHtI 11I11H,1 ufh'f
the fitting procedure. The structurül flHHI!mlh'!"
for the first coordination shelI is prClIIOIl'~t' tu Tnhl.'
1, where N} represents the coordluauuu mm,III1
and R the Ge-S distance for the edge !lIIlUilllA (1!11i.j
tetrahedra. It is well known from X-r")' lhllHHH1t1l1
measurements [15] and spectroscoplc !iIUCtll'fl It f, I
that in c-GeS2, Ge atoms are bonded ~~)tvlll ~lfl
phurs located at the comer of a distort~d hHI íI
hedron with an average Ge-S bond IOllglh ul J O
Á. The crystalline structure is a layercd ,1II1wlllll'
built up from chains of GeS4 corncr-shuriug It't
rahedra, these chains being connectcd t hlllUldl
links of two GeSe4 edge-sharing tetrahcdru 11'/1
on lhe suríucc
SERViÇO DE BtB~IOTECi\,
IFSC • USP INFORMAÇAO
 SiH. Messaddeq et aI. I Journal of Non-Crystalline Solids 304 (2002) 160-166 163
\
Ge-5
I

(a)

(b)

(c)

2 4 10 12 o 1 2 3 4 5 6
R(A)

Fig. 2. Nonnalized k2-weighted EXAFS oscillations at germa- Fig. 3. Modulus of Fourier transfonn (radial distribution
nium K-edge for (a) c-GeS2, (b) a-GeS2 and (c) GalOG~SS6S function around the Ge atoms) for (a) C-GeS2, (b) a-GeSiz and
powder glass sample. (c) GalOGezsS6s powder glass samples.

Table 1
This structure is known to exist in c-GeS2, and its
Fit results concerning the S shell around Ge atoms
existence has also been demonstrated in an a-GeS2
Sample First coordination shell
phase [18]. The second neighbouring sphere of Ge
atoms is composed of a short Ge-Ge distance (2.5 N1 (±0.2) RI (±0.01)

Á) and a longer Ge-Ge distance (3.0 Á) forms the c-GeSz 4.0 2.20
third Ge neighbouring shell. a-GeS2 4.0 2.21
GaGeS powder 4.1 2.21

3.3. EXAFS measurements of GaJOGe25S65glass

before and a/ter illumination
EXAFS spectra of the GalOGeZSS65 samples are
shown in Fig. 4 as a function of exposure time. A
decrease of the amplitude 01 EXAFS signal is
observed as the exposure time increases.
'Fig. 5 shows the radial distribution function
obtained from Fourier transform of the normal-
ized signal EXAFS presented in Fig. 4. The mod-
ification in the amplitude of the EXAFS signal is
associated to a decrease of the amplitude of the
first peak in the FT, which indicates modifications
of the average coordination shell around germa-
nium atoms after the illumination.
During the fitting of the spectra of the non-
illuminated sample, the best fitting was achieved
when the first coordination shell around Ge atoms
was performed assuming the existence of four
sulfur atoms located at approximately 2.2 A.
Conceming the illuminated samples, to take into

72
164 SiH. Messaddeq et aI. I Journal of Non-Crystalline Solids 304 (2002) 160-166

.f",.
}',
n

e, ~
h

#1'
p
<\
" , .~",

.,.~
~
~
:!(

J
liJ
"
:J<
II
n
n

~.
o 1 2 3 4 5 6

2 4 10
Fig. 4. Nonnalized XAFS oscillations extracted at gennanium
K-edge for GalUGe2SS65samples as function of time glass illu-
mination: (O) non-iIIuminated, (O) iIIuminated 5 h and (A) il-
luminated 10 h.
account the amplitude decrease observed in the
Fourier transform first peak, the number of sulfur
scattering atoms and their average distance, R,
from the germanium central atom were allowed to
vary, as well as the width of the distribution of
these neighbours (Debye Waller term, o). Never-
theless, no fit was achieved with any single sheIl
model. Actually, for the iIluminated samples, the
only way to fit the EXAFS signal in a satisfactory
way is to introduce an additional backscattering
contribution. A fit to withg; ±1O% was obtained
assuming the existence of a contribution of O
atoms at a distance of around 2.0 Ã. Although in
the fitting procedure we assume the existence of
Ge-O and Ge-S bonds, only one peak is observed
in the Fourier transform first peak. This fact is due
to the intrinsic limitation in the EXAFS mea-
R(A)
12
Fig. 5. Radial distribution function around the Ge atoms for
GalOG~SS6Ssamples as function of time exposure: (O) non-il-
luminated, (O) ílluminated 5 h and (L!I) illuminated 10 h.
surements that reduces the resolution at such a
small distance resolution. However, in the EXAFS
fit (Fig. 6) the best fit was obtained only when the
contribution of the O atoms was included, The
structural parameters obtained for the illuminated
samples are also presented in Table 2, where N 1,
RI and Nz, Rz correspond to Ge-S and Ge-O
atomic pairs, respectively.
To confirm the presence of oxygen in the illu-
minated samples, infrared spectra of non-illumi-
nated and illuminated glasses was measured. The
results are shown in Fig. 7. The features in the
speetral region studied are the vibrational bands at
3500, 870 and 727 em-I. Comparing the infrared
spectra, the intensity ofthe peak at 870, whieh was
attributed to the Ge-O-Ge stretching [19], is much
smaller for the non-illuminated sample when com-
pared to the sample illuminated during 10 h. This
fact indicates that the number of Ge-O bonds in-
creases with illumination duration.

73
 S.H Messaddeq et al. I Journal of Non-Crystalline Solids 304 (2002) 160-166
t
I
165

l
t
\

-:j

-C'd

3 4 5 6 7 8 9 10 11 12 4000 3500 3000 2500 2000 1500 1000 500

1
k(A- ) Wavelength (em")
Fig. 6. Ji3 I (k) curves fit (solid line = experimental, dotted Fig. 7. Infrared spectra of Gal0G~SS6S samples non-illumi-
line = modelization) for GaIQGe2SS6Sbulk sample. nated (a) and illuminated glass during 10 h (b).

Table 2 hedral units with Ge-S distances of 2.22 A in GeS4

Fit results about the Ge-S and Ge-O first sheUs coordination tetrahedral units. The insertion of gallium in the
Sample Sulfur neighbours Oxygen neigh- structure may maintain the global structure, but
(Ge-S) bours (Ge-O) distorting it through edge or comer sharing linkage
NI RI N2 R2 of GaS4 tetrahedra substituted for GeS4 units [20].
(±0.2) (±0.01) (±0.2) (±O.OI)
Conceming the structure of GalOGe25S65glasses
Non-exposed 3.8 2.20 as function of the illumination time, an interesting
Exposed 5 h 3.7 2.21 0.4 2.01
0.5
change was noted. The amplitude decrease of the
Exposed 10 h 3.5 2.21 2.00
Ge fírst peak in the Fourier transform was a
function of exposure time. This decrease in the first
4. Discussion peak of the Fourier transforro was attributed to
presence of an additional contribution. The fitting
From the data of the reference samples, we procedure gives to a contribution of Ge-O atomic
conclude that in the GalOGe25~65sample before il- pairs at a distance on the order of 2.0 Á. This
lumination both the local and the medium range distance is in agreement with the usual Ge-O dis-
order around Ge are similar to that observed in the tances in oxides [21]. The decreases of the EXAFS
layered crystalline forro of C-GeS2' Loireau-Lo- signal can be then explained by destructive inter-
zac'h et al. [20] have reported that the incorporation ference between photoelectron waves backscat-
of Ga2S3 in GeS2-based glasses forros GaS4 tetra- tered by O and S.

74
166 SiH. Messaddeq et ai. I Journal of Non-Crystalline Solids 304 (2002) 160-166
The presence of oxygen in illuminated samples
was also confinned by infrared measurements
(Fig. 7). Murth et aI. [19] assigned the band at 870
cm:", observed in Ge02 gIass, as being due to the
Ge-O-Ge stretching. We assume that the vibra-
tional band for the ilIurninated sample due to the
Ge-O interactions could be a product of photo-
oxidation of the gIass surface. ln fact, the origin
of the vibrational band at 727 crrr ' observed in
the ilIurninated sample, was assigned by others
[22] to the Ge--O stretching mode in S3_xOGe-0--
GeOxS3_xclusters formed on the surface of a sam-
pIe. Similar results, i.e., dependence of the expo-
sure condition and appearance of absorption in the
region 80()....870 em-I (indicatingpresence of Ge-O
bonds), were also observed in glassy Ge30S70[22].
Thus, EXAFS spectra coIlected at the germa-
nium K-edge for GalOGe25S65 glass samples before
and after irradiation shows that for the non-illu-
minated sampIe, the first coordination sheIl is
fonned by four sulfur atorns at a mean bond
length of 2.21 Á, as observed in the a-GeS2' After
illumination, a change is observed in the amplitude
of the Fourier transfonn first peak, which is re-
Iated to lhe Ge first coordination shelJ.To account
for this change, an additional contribution is in-
troduced in the fitting procedure, This contribu-
tion is related to oxygen atoms situated at around
2.0 Á. Infrared measurement confinned the pres-
ence of Ge-O bond in the glass sampIes after il-
lumination.
5. Conclusions
These experimental results showed that glass
illumination leads to an introduction of oxygen in
the glass structure breaking Ge-S intermolecular
bonds. This structural study will now be extended
to the other two atoms (sulfur and gallium) to
have a complete overview of the photoexpansion
effect in the glass structure.
«
Acknowledgements
We are grateful to FAPESP for financial sup-
port (no. 99/04263-7). The authors would like to
·75
thank Professor Dr Luís Antônio Nunes (IFSC-
USP) for the infrared spectra measurement. This
work was partially supported by LNLS (no. 554/
99).
References
[1] E. Hajto, R.E. Belford, P.J.S. Ewen, AE. Owen, J. Non-
Cryst. Solids 115 (1989) 129.
[2] K Tanaka, J. Non-Cryst. Solids 35&36 (1980) 1023.
[3] J.S. Berkes, S.W. Ing, W.J. Hillegas, l. Appl. Phys. 42
(1971) 4908.
[4] V. Lyubin, M. KJebanov, M. Mitkova, J. Non-Cryst.
Solids 230 (1998) 739.
[5] S.H. Messaddeq, M. Siu Li, D. Lezal, S.l.L. Ribeiro, Y.
Messaddeq, l. Non-Cryst, Solids 284 (2001) 282.
[6] S.H. Messaddeq, M. Siu Li, D. Lezal, Y. Messaddeq,
S.l.L. Ribeiro, L.F.C. Oliveira, J.M.DA. Rollo, J. Opto-
eletron. Adv. Mater. 3 (2001) 295.
[7] K Tanaka, Y. Kasanuki, A Odajima, Thin Solid Films
117 (1984) 251.
[8] L. Tichy, H. Tica, P. Nagels, E. Sleeckx, Opt. Mater. 4
(1995) 771.
[9] S. Rajagopalan, KS. Harshavardhan, L.K. Malhotra,
K.L. Chopra, J. Non-Cryst. Solids 50 (1982) 29.
[10] P. Armand, A. Ibanez, H. Dexpert, E. Philipot, J. Non-
Cryst. SoIids 139 (1992) 137.
[11] H.C.N. Tolentino, A.Y. Ramos, M.C.M. Alves, R.A
Barrea., E. Tamura, J.C. Cezar, N. Watanabe, J. Synchro-
tron Radiat. 8 (2001) 1040.
[12] D.A Sayer, BA. Bunker, in: D.C. Konongsberfer, R.
Prins (Eds.), X-Ray Absorption: Principies, Applications,
Technique of EXAFS, SEXAFS and XANES, Wiley-
Interscience, New York, 1998, p. 211.
[13] T. Ressler, J. Phys. IV 7 (1997) C2.
[14] J.J. Rehr, J. Mustre de Leon, S.l. Zabinsky, R.C. Albers,
J. Am. Chem. Soco 113 (1998) 5135.
[15) S.e. Rowland, S. Narshirnhan, A. Bienenstock, J. Appl.
Phys. 43 (1972) 2741.
[16] G. Lucovsky, J.P. De NeufviUe, F.L. Galeener, Phys. Rev.
B 39 (1974) 1591.
[17] G.Lucovsky, F.L. GaJeener, R.C. Keezer, R.H. Geils,
H.A. Six, Phys. Rev. B 10 (1974) 5134.
[18] P. Boolchand, J. Grothaus, M. Tenhover, M.A Hazle,
R.K. Grasseli, Phys. Rev. B 33 (1986) 5421.
[19] M.K. Murthy, E.M. Kirby, J. Phys. Chem, Glasses 5
(1964) 144.
[20] A.M. Loireau-Lozac'h, F. Keller-Besrest, S. Benazeth,
J. Solid State Chem. 123 (1996) 60.
[21] S.J.L. Ribeiro, J. Dexpertghys, B. Pirou, V.R. Mastelaro,
J. Non-Cryst.Solids 159 (1993) 213.
[22] L. Tichy, H. Ticha, P. Nagels, E. Sleeckx, Opt. Mater. 4
(1995) 771.
Ji!b
m
~
applied
surfaee se iene e
ELSEVlER Applied Surface Science 205 (2003) 143-150
www.elsevier.com/locate/apsusc

The inftuence of oxygen in the photoexpansion of GaGeS glasses

S.R. Messaddeq":", VR. Mastelaro", M. Siu Lia, M. Tabackniks",
D. Lezal", A. Ramos", Y. Messaddeq"
"Instituto de Física de São Carlos, Universidade de São Paulo, CP. 369, CEP: 13560·970 São Carlos, SP. Brazil
bJIIS/i/u/o de Fisica, Universidade de São Paulo-U'Si; CP. 66318, 05315 São Paulo, Brazil
"Laboratory of Inorganic Materiais lIC ASCR and 1Cr. Pelléova 24, Prague 6, Czech Republic
dLabora/ório Nacional de Luz Sincrotron. LNLS, Campinas. SP. Brazil
"Instituto de Química. UNESP. CP. 355, CEP: 14801·970 Araraquara, SP. Brazil

Received 16 April 2002; received in revised fonn 16 April 2002; accepted 8 Septernber 2002

Abstract

Irreversible photoexpansion and photobleaching effect has been observed in amorphous GalOGe25S65 glass when its surface
was exposed to ligh t wi th energy greaterthan the band gap. 3.52 e V. The magnitude of the expansion of GaGeS glasses depends
on the exposure conditions. Extended X-ray absorption fine structure (EXAFS) spectroscopy and Rutherford backscattering
spectrornetry (RBS) have been used to identify the chemical nature of the glass samples before and after illumination. The
quantitative analysis of the EXAFS data leads to a two-shell model of 0.5 oxygen atoms at 2.01 Á and 3.6 sulfur atoms at a
2.20 Á. RBS technique demonstrated that chernical composition of the glass surface after irradiation is oxygen rich. The
existence of Ge-O bonds in the glass after illurnination was also confirmed by infrared measurements.
© 2002 Elsevier Science B. V. Ali rights reserved.

Keywords: Chalcogenide glasses; Photoexpansion; Oxygen; EXAFS; RBS

1. Introduction be removed by annealing in the case of a reversible

Arnorphous glasses of the type Geâ,
chalcogenide
(x = 1,2) undergo a variety of photoinduced structural
[1.2] and optical [3] changes induced by bandgap
radiation. The most important of these structural
changes is the photoexpansion phenomenon, in which
a positive volume change has been used to produce
convex microlens [4,5]. These phenomena are char-
acterized by a red (photodarkening) or a blue shift
(photobleaching) of the optical gap (Eg). This shift can
• Corresponding author, Tel.: 55-16-2739823;
fax: 55-16,2739811.
E-mail address:sandrahm@if.sc.usp.br (S.H. Messaddeq).
change, but is not removed by annealing in the case of
an irreversible change.
ln our previous investigation of the photoinduced
changes in GalOGe25S65 glass, we reported photo-
. bleaching effect associated with a decrease in the
. density through dilatation leading to an increase in
volume [6]. The thickness of the illuminated re gion
was increased after exposing the Ga IOGe25S65 bulk
glass with light having above band gap energy. More-
over, we found that this photoexpansion effect was
irreversible [7].
10 general, the photoexpansion and other photoin-
duced effects are attributed to the excitation of elec-
trons across the bandgap and consequent atom

0169-4332/02/$ - seefront matter © 2002 Elsevier Science B.V. All rights reserved.
MtS0169-4332(02)01013-9
144 S.H. Messaddeq et ai.! Applied
displacements. Recently other investigations indi-
cated photoinduced oxidation as underlying change
[8,9].
ln the case of the irreversible photoinduced
changes, the role of the oxygen has been studied.
Tanaka et al. [10] have observed that the magnitude
of the photoinduced shift in the absorption edge of the
GeS~ filrns is strongly dependent on the ambient
oxygen pressure. Similar results were also observed
in sulfur-rich films [11,12]. It showed using X-ray
diffraction, electron spin resonance (ESR) and elec-
tron spectroscopy for chemical analysis (ESCA) that
bleaching induced by illumination and by annealing of
Ge-S thin films results from photo- and therrnally
induced oxidation mainly proceeds at the surface of
the thin filmo
In the present paper, we will study the volume
change during the irradiation by using extended X-
ray absorption fine structure (EXAFS) as a structural
probe of the local environment around lhe germanium
atoms. RBS measurements will be performed to mea-
sure the amount of the element present in the sample
before and after exposure. The influence of exposure
condition on the thickness and the optical transnússion
in the infrared region is also studied.
2. Experimental
Polycrystalline germanium, gallium and sulfur were
weighted, mixed together and transferred to a quartz
ampoule. After vacuum sealing, a heating process up
to 950°C was started. At 950°C the liquid was
homogenized for 6 h through a continuous horizontal
rotation. Then the ampoule was withdrawn from the
fumace and air-cooled, An anneal at 400°C for sev-
eral hours followed.
For the photoexpansion experimento part of the
glass surface area was covered with a mask and were
exposed to ultra-violet (UV) light (351 nm),
J = 5 mW/cm2 from a Kr+ ion laser at room tempera-
ture. After illumination, the mask was removed from
the sample and the exposed area was m'hsured using a
profile an atornic force rnicroscope (AFM-Digital
Instrurnents-Nanoscope 3 A). Accordingly, for the
atmosphere influence experiments samples were held
in a vacuum cell having quartz windows, which was
pumped ovemight at pressure of 10-3 Torr, The cell
..,
Surface Science 205 (2003) 143-150
was then sealed, different gas flux was perforrned and
the cell was illuminated with UV light with an inten-
sity of 5 mW/crn2 on the sample for 3 h.
EXAFS measurements at the Ge K-edge
(11 104 eV) were perforrned at the XAS beam line
of ar the LNLS facility (Campinas, Brazil) with an
average beam current of 150 mV. The rnonochroma-
tization of the incident bearn was made using a double
Si (l 1 1) crystal. The intensity of this incident beam
was monitored by an air-filled ionization chamber.
The spectra of a Ga IOGe2SS65 sample before and after
illumination were collected in total detection electron
yield. The EXAFS spectra were analyzed by a stan-
dard procedure of data reduction [13] using WinXAS
code [14]. Structural parameters N (coordination num-
ber) and R (interatomic distance) are obtained from
least squares fitting in k-space. using theoretical phase
and amplitude functions obtained from FEFF6.0l
software [15].
The RBS experiments, used to determine the com-
position of the exposed and non-exposed area com-
positions, were perforrned atLAMFIlUSP. São Paulo.
using a He+ beam with an energy of E = 1.7 MeV,
charge Q = 30 mC, current I = 30 nA and detection
angle q = 170°. From the RBS data. the amount of Ga,
Ge, S. and O elements was determined as a function of
the depth profile.
A qualitative bonding analysis was perforrned by
means of Fourier transforrn infrared spectrometry
(FTIR) using a Nicollet 5PC FfIR spectrometer in
the 400-4000 cm - I range.
3. Results
3.1. EXAFS measurements ofGaJOGe2SS6S glass
before and after illumination
EXAFS spectra of the GalOGe25S65bulk glasses are
shown in Fig. 1 as a function of exposure time. A
decrease of the amplitude of EXAFS signal is
observed as the exposure time increases.
The Fig, 2 shows the radial distribution function
obtained from Fourier transforrn of the normalized
signal EXAFS presented in Fig. 2. The modification in
the amplitude of the EXAFS signal is associated with a
strong decrease of the amplitude of the first peak in
the FT. pointing out modifications of the average
77
 S.H. Messaddeq et a/./Applied
~
fi
~
2 4
Fig. I. Normalized XAFS oscillations extracted at germanium K-edge for GaIQGe2SS65samples as function of time glass illumination:
non-illuminated, (O) ilIuminated 5 h and C0J illuminated 10 h.
coordination shell around germanium atoms after the
glass illumination.
During the fitting of the non-illuminated sample, the
best fitting was achieved when the first coordination
shell around Ge atoms was performed assuming the
existence offour sulfur atoms located at approximately
2.2 Á. Concerning the illurninated samples, to take into
account the amplitude decreasing observed in lhe
Fourier transform first peak, the number of sulfur
scattering atoms and its average distance R from the
germanium central atom were allowed ..; to vary, as well
as the width of the distribution of these neighbors
(Debye Waller term o"). Nevertheless, no good fitting
was achieved with any single-shell model. Actually for
the illuminated samples, the only way that fits the
Surface Science 205 (2003) 143-150
1
b
\o!
~
a-Ga,oGe.sS.s Bulk
6 8 10 12
1
k(Ã- )
(O)
EXAFS signal in a satisfactory way, is to introduce
the contribution of an addítional backscattering con-
tribution. An excelIent fit was obtained assuming the
existence of a contribution of O atoms at a distance of
around 2.2 Á. Although in the fitting procedure we
suppose the existence of Ge-O and Ge-S bonds, anly
one peak is observed in the Fourier transform first peak.
This fact is due to the intrinsic lirnitation in the EXAFS
measurements that hinders the resolution at such low
leveI. However, in the EXAFS fit (Fig. 3) it is c1ear that
the best fit was obtained only when the contributian of
the O atoms was considered. The structural parameters
obtained for the illuminated samples are also presented
in Table 1, where N], R] and N2, R 2 correspond to Ge-S
and Ge-O atomic pairs, respectively.
78
",
146 S.H. Messaddeq et al.r Applied Surface Science 205 (2003) 143-150

ço
06
6b
I\
o o
1I \ o

01 ~~
10
oo '\I

•..o
E

-•..
(/)
c:
ca
....
•..eu
.~

1
::J
O
U.

o 1 2 3 4 5 6
R(Á)
Fig. 2. Radial distribution function around the Ge atoms for GalOGe2SS6S samples as function of time exposure: (O) non-illuminated, (O)
illuminated 5 h and <.6) illuminated 10 h.

3.2. IR measurements
Table 1
Fit results of the Ge-S and Ge-O first shells coordination ln arder to eontirm the presenee of oxygen in the
illuminated samples, infrared spectra of non-illumi-
Sample Sulfur neighbors Oxygen neighbors
(Ge-S) (Ge-O) nated and ilIuminated glasses were measured. The
c results are shown in Fig. 4. The features in the spectral
NI RI N2 R2
region studied are the vibrational bands at 3500, 870
(±0.2) (±O.Ol) (±0.2) (±O.Ol)
and 727 em -I. Cornparing the infrared spectra, the
Non-exposed 3.8 2.20 intensity of the peak at 870, which was attributed to
Exposed for 5 h 3.7 2.21 0.4 2.01
the Ge-O-Ge stretching, is mueh smaller for the non-
Exposed for 10 h 3.5 2.21 0.5 2.00
illuminated sarnple when eompared to the sample

79

.
.. .
 S.H. Messaddeq et al./Applied Surface Science 205 (2003) 143-150 147

o
o
~,.~
LO
<'l

t', ~
I I
r~, •

I· r i• 1 n
p
r t I + 7 + ,. :::I
1\ r!, I~ rJ,
r, , 7 t ~ ~
li ~i 1\ t~ A
41
U

, f 1 I 1 r i ,\\ f\ c:
ra
:!::

,\ ri rI ri' \ t \ f t \ . E
UI
c:
•..
ra
~ I i I i r' i f
l' t T,I \V' I-
I ~' I
I Ir' ! I ~
1\ T
i I
,
ti'
I r' i r ~
I
T
i 1
\'
i r \ il ~t '.J
1 í I
+1

,t;
.'~
./1

.'' ~ 1~~r V
~~~l ~ 4000 3500 3000 2500 2000 1500 1000 500
~!
, .. a-Gal0Ge2SS65
exposed 5hs
Wavelenght (em")

Fig. 4. Infrared spectra of GalOGe2SS65 samples non-illuminated

(a) and illurninated glass during 10 h (b).
345 9 10 11 12

the oxygen atoms are ineorporated near the surfaee

Fig. 3. eX(k) curves fi! (solid line: experimental; dotted line during the exposure processo
modelization) for Ga/OGe25S65 bulk sample.

3.4. lnfluence of the atmosphere conditions

illuminated during 10 h. This faet is an indieative that
the number of Ge-O bonds inereases after glass
illumination. We note that the vibrational band at
727 em -I is observed only when the sarnple was
exposed for 10 h in air.
3.3. RBS spectroscopy
The RBS data were analyzed by the RUMP routine
[16] and provi de the atomic eontent of the Ga, Ge, S
.,, and O. The results ofthe RBS speetrum of a Ga-Ge-S
before and after exposure are shown in Fig. 5. In
Fig. 5b the oxygen signal presents a non-statistieal
bum loeated at 0.86 Me V. Table 2 presents the ato mie
. ..
Fig. 6 shows the evolution of thickness in terrns of
different kinds of atmosphere conditions. It can be
observed that the photoexpansion phenomenon is infíu-
enced with the interaction of the surface of the glass
with the environrnent since the decrease of the value of
measurements the thickness with decrease of oxygen eontent is evi-
dent. The exclusion of atmosphere (P = 10 mTorr)
leads to any change in the surface of the glass. This
Table 2
Atomic content of Ga, Ge, S and O calculated from the RBS data
for Ga/OGe25S65 samples illuminated for 10 h

200nm 400nm 600nm 800nm

eontent ealeulated from the RBS data. It is observed
that illuminated sample has oxygen presenee with the Ga 6 7 7 7
oxygen content varying from 45 to 7% as the depth Ge 16 19 19 17
S 33 67 74 76
increases from 200 to 400 nm. The oxygen eontent
O 45 7
decreases with the depth of the profile, indicating that

80
148 S.H. Messaddeq et ai. / Applíed Surface Science 205 (2003) 143-150

E (KeV)

10000.---.-------.-----~-------.------~

8000

(a)
'"O
Q)
>- 6000
'"O
Q)
N
CO
E 4000 (b)

Z
)..-
O °2
2000

100 200 300 400 500

Channel
Fig.5. RBS spectrum of OalOOe25S65 samples non-illuminated (a) and ílIuminated glass duríng \O h (b). The positíon of the surface edges for
O is indicated.

expansion is irreversible, in the sense that annealing 4. Discussion

does not restore the initial morphology for vírgin bulk
[7]. Tichy et aI. [11] associate the nature of the irre-
versible photoinduced phenomenon in GeS2films (i) an
increase of the density of ~teronuclear bonds, and (ii)
incorporation of oxygen into the GeS matrix of the
glass. The effect of oxidation reaction at the surface of
the AS2S3 films has a1so been established by Beckers
et al. [17]. However, in that case, the exclusion of
oxygen precluded AS203 crystal formation.
We attempt to give a qua1itative explanation to the
light-induced volume expansion in air condition
based on a specific chernical bonding in chalcogen-
ides. Chemical bonds in chalcogenides comprise of
strong íntramolecular covalent bonds (prirnary bonds
which are due to binding s-p hybridized electrons of
chalcogen atom) and weak intermolecular Van der
Waals type bonds (secondary bonds which are due to

81

S.H. Messaddeq et al.r Applied Surface Science 205 (2003) 143-150
.-, 400
a
.5 300
'"
~ 200
Q
..:.::
.~ 100
E5 o
02
Fig. 6. Thickness measurernent Df GaGeS showing the effect of the photoexpansion
different kind of atmosphere.
lone-pair p-electrons of cha1cogen atom) (see, e.g.
[18,19)).
We suggest that a magnitude of the expansion is
restrieted by the strength of the interaction (degree of
interference [18]) between s-p hybridized and p-elec-
trons participating in the primary and secondary
bonds, respectively. Specifícally, in the case of our
glass system GaGeS, light-induced weakening in the
secondary bonds should result in strengthening of the
primary bonds as it is argued in [18], whereby lirniting
a magnitude of the glass network expansion.
Concerning the structure of GalOGe25S65 glasses as
funetion of the illurnination time, an interesting
ehange was noted. The amplitude deerease of the
Ge first peak in the Fourier transform was a function
of exposure time. This decrease in the first peak of the
Fourier transform was attributed to presence of an
additional eontribution. The fitting procedure gives to
a eontribution of Ge-O atomic pairs at a distance on
the order of 2.0 A. This distance is in agreement with
the usual Ge-O distances in oxides [20). The
decreases of the EXAFS signal can be then explained
by destructive interference between photoelectron
waves' baekscattered by O and S.
Thus, EXAFS spectra collected at the germanium
K-edge for GalOGe25S65 glass samples before and
after irradiation show that for the non-illuminated
sample, four sulfur atoms at a meáh bond length of
2.21 A forro the first eoordination shell, as observed in
the a-GeS2. After illurnination, a change is observed in
the amplitude of the Fourier transform first peak,
whieh is related to the Ge first coordination shell.
To account for this change, an additional eontribution
is introduced in the fitting procedure. This contribu-
t,
I 82
\
l
149
He Air Vacuum
Atmosphere
on lhe surface structure of glass exposed during 3 h in
tion is related to oxygen atoms situated at around
2.0 Ã.. Infrared measurement confirmed the presence
of Ge-O bond in the glass samples after illurnination.
The presence of oxygen in illuminated samples was
also confirmed by infrared measurements (Fig. 4).
Murth and Kirby [21] assigned the band at
870 em -I, observed in Ge02 glass, as being due to
the Ge-O-Ge stretehing. We assume that the vibra-
tional band for the illurninated sample due to the Ge-O
interactions could be a product of photooxidation of
the glass surfaee. Even the peak at 870 em -1 is also
observed in samples non-exposed and exposed in
vacuum, the intensity of this peak is more intense
than when the sample has been illuminated in air. In
fact, the illurninated sample shows a feature at
727 em -1, which was assigned by Tichy et aI. [11]
to the Ge-O stretching mode in S3_xOGe-O-
GeOxS3_x clusters formed on the surface of a sample.
They noted as x increases from O to 3, the "local
eletronegativity" of the Ge atom increases and this
causes a decrease in the bond strength of the central
Ge-O bond in the cluster causing a shift in the
vibrational frequeney to lower wavenumbers. Similar
results, i.e. dependence of the exposure eondition and
appearance of absorption in the region 800-870 em-1
(indicating presence of Ge-O bonds), were also
observed in glassy Ge30S70 [22].
Regarding our experimental data about enrichment
with oxygen in the exposed area, we note that weak-
ening of secondary bonds centered at sulfur atom can
result in the breaking of Ge-S bond and then oxidation
occurred producing Ge-O bonds. This behavior is
ascribed to a photoinduced reconstruction of the net-
work, resulting in a more chemically ordered structure.
150 S.H. Messaddeq et aI.! Applied Surface Science 205 (2003) 143-150

RBS analysis of the glasses after irradiation showed
that the amount of oxygen decreases with the increase
in the depth profile. An explanation for this behavior is
that during illumination in air the GeS2 molecules are
broken and the concentration of holes increase at the
surface. The interaction ofthe surface ofthe glass with
oxygen atoms from environment makes bonds with Ge
Ieading to an oxygen-rich surface layer. Another
important result comes from the absence of structural
changes under irradiation in vacuum. This result
shows the irnportance of the ambient atmosphere,
and suggests that during illumination the oxygen
are incorporated onto the glass surface inducing the
volume changes.
Recent AFM studies have shown the formation of
pyramid-like structures on the surface of AssoSeso film
when irradiated in air with laser [23]. These features
were not observed when films were irradiated in
vacuum. Apparently, the pyramids represent the pro-
duct of a reaction between the AssoSeso film and
ambient atmosphere.
5. Conclusion
The present work shows the structural investigation
of the photoinduced effect in the surface of GaGeS
glasses and demonstrated that the presence of oxygen
is a requirement for creating the volume expansion in
that glasses.
EXAFS measurements suggest that after glass illu-
mination, Ge-S intermolecular bonds are broken and
show an introduction of oxygen in the glass structure.
RBS analysis of the glasses after irradiation showed
that the amount of oxygen decreases with the increase
in the depth profile RBS analysis. The dependence of
the exposure conditions in IR measurement show the
. appearance of absorption in the region 800-870 cm-I
indicating presence of Ge-Q bonds.
Acknowledgements
We are grateful to FAPESP for financial support
(No. 99/04263-7). The authors would like to thank
Prof. Dr. Luis Antônio Nunes (IFSC-USP) for the
infrared spectra measurement. This work was partially
supported by LNLS (No. 554/99).
References
[I] N.A. Davydova, v.v. Tishchenko, J. Baran, M. Vleck, J. MoI. \
i
Struct. 450 (1998) 117.
[2] P. Loefller, T. Schwarz, H. Saulter, D. Lezal, J. Non-Cryst. l
Solids 232-234 (1998) 526.
[3] K. Petkov, P.J.S. Ewen, J. Non-Crysl. Solids 249 (1999) 150.
[4] H. Hisakuni, K. Tanaka, Appl, Phys. Lett. 65 (1994) 2925.
[5] K. Tanaka, Phys. Rev. B 57 (1998) 5163.
[6] S.H. Messaddeq, M. Siu Li, D. LezaJ, S.J.L. Ribeiro, Y.
Messaddeq, J. Non-Cryst. Solids 284 (2001) 282.
[7] S.H. Messaddeq, M. Siu Li, D. Lezal, Y. Messaddeq, S.J.L.
Ribeiro, L.F.c. Oliveira, 1.M.D.A. Rollo, 1. Optoelectr, Adv,
Mater. 3 (2001) 295.
[8] E. Marques, J.M. Gonzalez-Leal, R. Prieto-Acon, R.
Jimenez-Garay, M. Vleck, J. Phys. D: Appl. Phys. 32
(1999) 3128.
[9] J. Dikova, N. Starbov, K. Starbova, J. Non-Cryst, Solids 167
(1994) 50.
[10] K. Tanaka, Y. Kasakuni, A. Odajima, Thin Solids Films 117
(1984) 251.
[li] L. Tichy, A. Triska, H. Ticha, M. Frumar, Philos, Mag. B 54
(1986) 219.
[12] T. Kawaguchi, S. Maruno, Jpn. J. App!. Phys. 27 (1998)
2199.
[13] D.A. Sayer, B.A. Bunker, in: D.C. Konongsberfer, R. Prins
(Eds.), X-ray Absorption: PrincipIes, Applications, Technique
of EXAFS, SEXAFS and XANES, Wiley-Interscíence, NY,
1998, pp. 211.
[14] T. Ressler, J. Phys. IV 7 (1997) C2-269.
[15] 1.1. Rehr, J. Mustre de Leon, S.l. Zabinsky, R.C. Albers, J.
Am. Chem. Soe. 1J3 (1998) 5135.
[16] L.R. Doolitte, Nuc!. Instrum. Meth. B 9 (1995) 344.
[17] 1.S. Beckers, S.w. Ing, W.J. Hillegas, J. Appl. Phys. 42 (1971)
4908.
[18] R.M. Martin, G. Lucovsky, K. Helliwell, Phys. Rev. B 13
(1976) 1383.
[19] R. Zallen, M.L. Slade, A.T. Ward, Phys. Rev. B 3 (1971)
4257.
[20] S.J.L. Ribeiro, J. Dexpertghys, B. Pirou, V.R. Mastelaro, J.
Non-Cryst. Solíds 159 (1993) 213.
\
[21] M.K. Murthy, E.M. Kirby, 1. Phys. Chern. Glasses 5 (1964) {
144.
[22] L. Tichy, H. Ticha, P. Nagels, E. Sleeckx, Opt. Mater. 4
(1995) 771-779.
[23] J.T. Bloking, S. Kríshnaswami, H. Jain, M. Vleck,R.P. Vinci,
J. Opt. Mater. 17 (2001) 453-458.

83
 Capítulo IV

A Espectroscopia de Absorção de Raios-X Aplicada no Estudo

Estrutural de Materiais Óxidos

84
 T
,
i
~

IV.1- Introdução

Nestes últimos 12 anos, utilizamos a Espectroscopia de Absorção de Raios-X

(XAS) no estudo da estrutura local em uma série de materiais óxidos amorfos e
policristalinos que apresentam uma larga gama de propriedades físicas. De uma
maneira geral, o objetivo destes estudos estruturais foi o de verificar a existência de
uma relação entre estrutura local e propriedades físicas e químicas destes materiais.
Apesar do fato de que alguns destes materiais apresentam ordem a longa distância, a
utilização da técnica de XAS mostrou ser essencial para que esse objetivo fosse
alcançado.
Apresentamos neste capítulo dois estudos onde a técnica de XAS foi utilizada na
caracterização de materiais óxidos amorfos e policristalinos:

1- Estudo da estrutura local em amostras óxidas de composição AB03 (A=Sr, Pb;

B=Ti) preparadas através de métodos químicos;
2- Estudo estrutural em cerâmicas ferroelétricas de composição Pb1-xLaxTi03

IV.2- Estudo da estrutura local em amostras óxidas de composição AB03 (A=Sr,

Pb; B=Ti) preparadas através de métodos químicos

Materiais ferroelétricos de composição AB03 (A e B são cátions enquanto O é

oxigênio) na forma de cerâmicas ou filmes finos têm sido extensivamente estudados
devido ao interesse científico e tecnológico gerado a partir das propriedades físicas
apresentadas por esta classe de materiais [1-4]. Dentre os diferentes métodos de
preparação destes materiais, na forma de filmes finos ou na forma de pó, os métodos
de preparação através de uma rota química tem chamado a atenção dos
pesquisadores por apresentar certas vantagens quando comparados aos métodos
mais tradicionais de preparação de materiais cerâmicos e filmes finos [5].
Dentre os diferentes métodos de preparação de amostras na forma de pó ou na
forma de filmes finos que utilizam uma rota química de síntese~ temos utilizado o
método dos precursores poliméricos, baseado no processo de Pechini, que envolve a
formação de uma fase "vítrea" polimérica orgânica [6]. Este método tem sido utilizado
na produção de um grande número de materiais [7]. Devido à formação desta
85
 fase amorfa intermediária, o processo de cristalização dos filmes se asSertllJllIIR
processo de cristalização de um vidro inorgânico. Utilizando o modelo padrão t._",o",,>

nucleação em vidros, Schwartz e colaboradores propuseram um modelo no qual

grau de orientação e a microestrutura dos filmes finos preparados pela técnica <1ft
decomposição de metais orgânicos (MOD-Metalorganic Decomposition), são
correlacionados com a barreira termodinâmica de nucleação [8]. A partir deste
trabalho, Schwartz mostrou que a mudança na força motriz para cristalização tem
como resultado uma modificação microestrutural devido à variação que ocorre na
altura da barreira da interface filme/substrato e da nucleação superficial.
Devido à importância da fase intermediária no processo de obtenção destes
materiais, nos pareceu importante realizar um estudo com a finalidade de melhor
compreender a influencia desta fase no processo de cristalização destas amostras.
Inicialmente, a técnica de XAS foi utilizada no estudo do processo de cristalização
da fase SrTi03 na forma de filmes finos produzidas através do método dos precursores
poliméricos [Artigo-IV.A]. Estudos preliminares haviam mostrado que as amostras na
forma de filmes finos cristalizam a partir de uma fase amorfa inorgânica sem a
formação de uma fase intermediária [7d].
Os resultados obtidos no estudo da borda de absorção (XANES) mostraram que
esta fase amorfa é basicamente composta por pequenos agregados formados por
átomos de titânio coordenados por 5 átomos de oxigênio ([51TiO)04). Assim, durante o
c processo de transição entre a fase amorfa e a fase cristalina, ocorre uma mudança
significativa nestes agregados: após a cristalização, os átomos de titânio coordenados
por 5 átomos de oxigênio se transformaram em átomos de titânio coordenados por 6
átomos de oxigênio em uma simetria octaédrica. O filme apresenta-se em um estado
amorfo quando aquecido até 450°C. A formação de um filme policristalino acima de
450°C sugere que o processo de nucleação é superficial. Este processo de nucleação
c

irá promover a cristalização a partir da interface da superfície/ar na direção da

interface filme/substrato.
Em um segundo trabalho, estudamos o processo de cristalização do composto
PbTi03 (PT) a partir de uma fase amorfa também obtida através do método dos
precursores poliméricos [Artigo-IV.B].
A análise dos espectros XANES revelou a presença de espécies de titânio em

86
SERViÇO DE BIBLIOTECA
I FSC-USP INFORMAÇÃO
 I
i

diferentes coordenações sendo que havia uma predominância de átomos de titânio em

uma simetria octaédrica. A partir da simulação dos espectros EXAFS da amostra
amorfa, observou-se que os átomos de titânio são coordenados por 6 átomos de
oxigênio a uma distância media de 1.92 A. Esta distância média é menor que a
distância encontrada na fase cúbica. Esta diferença de distância seria uma
conseqüência da existência de diferentes espécies de titânio na amostra.
Como o processo de cristalização ocorre a uma temperatura muito baixa, foi
proposto que a cristalização deve envolver rearranjos à curta distância que tem como
efeito à formação de uma fase PbTi03 cúbica metaestável. Esta fase metaestável é
preferencialmente formada devido à similaridade da estrutura local da fase cúbica e da
fase amorfa. Assim, a cristalização da fase cúbica é mais favorável que da fase
tetragonal que apresenta um octaedro distorcido.
Mais recentemente, realizamos um estudo semelhante em amostras de
composição PbTi03 obtida através do Método dos Peróxidos Oxidantes (OPM)
[Artigo-IV.C].
Camargo e colaboradores desenvolveram uma nova rota de síntese chamada
de "Método dos Peróxidos Oxidantes" (OPM), que tem sido utilizada na preparação de
diversos sistemas de estrutura peroviskita em temperaturas relativamente baixas.
Além disso, estes materiais apresentam uma distribuição relativamente estreita no
tamanho das partículas em escala nanométrica [9-11].
Ainda que as vantagens inerentes do método OPM sejam atualmente
reconhecidas, pouco se sabe sobre a natureza do precursor amorfo e o mecanismo de
cristalização a partir desta fase precursora. Neste trabalho, estávamos interessados
em obter informações sobre a ordem local do material precursor amorfo obtido através
do método de OPM quando aplicado à síntese do composto PbTi03.
Os resultados da obtidos a partir da analise dos espectros XANES mostraram
que as espécies de titân/'b predominantes são coordenadas por 5 átomos de oxigênio,
em bom acordo com outros resultados encontrados na literatura. Isto significa na
verdade dizer que um dos átomos do octaedro estaria localizado a uma distância tal
que não poderia ser considerado como fazendo parte da primeira esfera de
coordenação.
A análise dos espectros EXAFS das três amostras consideradas amorfas indicou

87
 -- - - -- ----------------~~~~

que a ordem local ao redor do átomo de titânio é essencialmente a mesma. Uma vez
que a cristalização da fase pirocloro ocorre a partir da fase amorfa precipitada, foi
possível concluir que a estrutura que possui uma ordem local semelhante a da fase
precursora precipitada será preferencialmente cristalizada. A simulação dos espectros
EXAFS mostrou que a coordenação local dos átomos de titânio nos pós-calcinados a
uma temperatura de 450°C apresenta a mesma estrutura local da fase amorfa
precipitada e muito diferente da fase tetragonal.
Os resultados obtidos neste trabalho mostraram que o processo de cristalização
a partir do precursor amorfo favorece a cristalização de fases que apresentam uma
ordem local mais próxima a aquela da fase precursora, em acordo com nossos
estudos anteriores.

IV.3- Estudo da estrutura local de cerâmicas ferroelétricas

Os materiais cerâmicos de cornposrçao PbTi03 e PbZro,65

Tio,3503têm sido
exaustivamente caracterizados devido as suas propriedades ferroelétricas [12].
O titanato de chumbo, PbTi03 (PT), apresenta em temperaturas acima de
500°C uma estrutura tipo peroviskita com simetria cúbica, e quando na fase
ferroelétrica, apresenta uma simetria tetragonal [13]. Devido a esta grande distorção
estrutural, as cerâmicas de composição PT são extremamente frágeis. Contudo, é
possível obter amostras de PT mecanicamente resistentes se forem adicionados
elementos modificadores ou dopantes ao PT, em proporções adequadas [14]. Entre os
elementos mais utilizados neste caso podemos citar o lantânio, formando o composto
Pb1-xLaxTi03(PLT) e o bário formando o composto Pb1-xBaxZro,65Tio,3503
(PBZT). Além
de apresentar uma melhora das propriedades mecânicas, a substituição do átomo de
chumbo pelo átomo de lantânio provoca certas mudanças nas propriedades físicas
••
deste material como o aparecimento de uma Transição de Fase Difusa (TFD) isto é,
uma transição que não ocorre em uma temperatura definida Te, mas sim, em um
intervalo de temperaturas, conhecido como intervalo de Curie [14].
Além do interesse em se estudar as modificações estruturais causadas pela
substituição do átomo de chumbo por átomos de lantânio e bário, o interesse em se
estudar os materiais que apresentam o fenômeno de transição de fase difusa e

88
 T
1

comportamento relaxor advém do fato que estes materiais podem apresentar uma
melhor eficiência em suas propriedades como no efeito piezoelétrico induzido, nas
propriedades eletro-óticas quando da obtenção de cerâmicas transparentes e nas
suas propriedadeseletrostrictivas [15-18].

IV.3.1- Estudo estrutural de cerâmicas ferroelétricas Pb1.xLaxTi03 (PLT)

o interesse de se estudar as cerâmicas ferroelétricas de composição Pb1-

xLaxTi03 (PLT) advém do fato de que, as propriedades físicas acima citadas, as quais
sofrem mudanças significativas à medida que o átomo de chumbo é substituído pelo
lantânio, não serem completamente compreendidas.
Nas amostras onde a transição de fase ferro-paraelétrica é bem definida, ou
seja, para composições contendo entre O e 20% de lantânio, observou-se que a razão
c/a da cela unitária diminui de forma praticamente linear até 20% (moi) de La [14].
Neste intervalo de concentração, a estrutura cristalina pôde ser facilmente
caracterizada utilizando técnicas convencionais de caracterização estrutural de raios-X
[14]. Entretanto, nas amostras onde o fenômeno de TFD foi observado, a razão c/e
dos parâmetros de rede é praticamente constante para concentrações maiores que
25%. De acordo com a literatura, para concentrações acima de 20% de lantânio,
estaria ocorrendo uma transição de uma fase tetragonal para uma fase denominada
de "pseudo-cúbica" [14]_
Assim, as tentativas de se compreender o fenômeno de transição de fase difusa
nesta classe de materiais não teve Q sucesso desejado pelo fato de que a estrutura,
do ponto de vista cristalográfico, não poder ser definida com precisão.
Neste sentido, propusemos então a caracterização estrutural destas amostras
através da técnica de Espectroscopia de Absorção de Raios-X (XAS). Estes resultados
••
foram confrontados com os dados obtidos por Difração de Raios-X (DRX).
Os resultados obtidos na caracterização do sistema Pb1-xLaxTi03(PLTx) estão
condensados no Artigo-IV.D.
Os resultados obtidos por Difração de Raios-X (DRX), utilizando o método de
Rietveld, mostraram que o grau de tetragonalidade diminui gradualmente à medida
que a quantidade de lantânio aumenta. Uma transição de fase estrutural da fase

89
tetragonal para uma fase cúbica foi observada para amostras contendo entre 25 e
30 at. % de lantânio. Na verdade, para amostra com x variando entre O e 25 at. %,
a estrutura foi identificada como pertencendo ao grupo espacial P4mm, enquanto
que para a amostra contendo 30 at. % de lantânio, a estrutura foi identificada

como pertencente ao grupo espacial Pm3 m. Assim, os dados de DRX para a

amostra contendo 30 at. % de lantânio indicam que o átomo de titânio estaria em
uma posição ideal em um octaedro com 6 distâncias Ti-O idênticas.
Por outro lado, de acordo com os dados obtidos a partir da análise da borda
de absorção (espectros XANES), o átomo de titânio estaria em uma posição fora do
centro de simetria no octaedro TiOs para todas as amostras, incluindo a amostra
contendo 30 at. % de lantânio. No que tange a análise dos espectros EXAFS, a
presença da borda LIII situada logo após a borda K do átomo de titânio impediu que
obtivéssemos informações quantitativas a partir dos dados de EXAFS, principalmente
nas amostras ricas em lantânio. Apesar destas limitações, foi possível observar que,
apesar de ocorrer uma diminuição da desordem local à medida que a quantidade de
lantânio aumenta, ela não deixa de existir mesmo para a amostra contendo 30 at. %
de lantânio.
As diferenças estruturais observadas entre os resultados de XAS e DRX foram
explicadas pelo fato de que, quando da existência de ordem a Idnga distância, ainda
que a amostra apresente uma desordem na sua estrutural local, o padrão de difração
de Bragg é essencialmente formado por picos de difração. A informação sobre a
desordem pode ser obtida a partir de experimentos de espalhamento difuso mas não a
partir de padrões de difração e resultados obtidos a partir de refinamento de estrutura
através do método de Rietveld [19]. Para obter estas informações, que normalmente
são ignoradas, é necessário introduzir o efeito de desordem na simulação e realizar
medidas em intervalo de k suficientemente
I(
grande para que estas distorções possam
ser evidenciadas [20J. Como à existência destas distorções não são consideradas
quando da análise de dados de difração de raios-X em amostras na forma de pó, estes
experimentos fornecem somente informações sobre uma média da estrutura.
Este estudo está sendo ampliado para o sistema Pb1-xBaxZro,s5
Tio,3503 (PBZT) e
os resultados obtidos com as técnicas de XAS e difração de raios-X estão sendo
analisados e deverão estar concluídos no segundo semestre de 2004.

90
IV.3.2- Estudo Estrutural e o Mecanismo de Transição de fase em amostras do
sistema Pb1-xLaxTi03 pc
qL
Alguns dos materiais com estrutura peroviskita de composição do tipo AB03
e
apresentam uma transição de fase ferroelétrica-paraelétrica em temperaturas bem Er
determinadas, denominadas temperaturas de transição, Te [20]. As mudanças
pc
estruturais decorrentes do processo de transição de fase foram por longo tempo
at
estudadas por meio das técnicas de difração de raios-X e de nêutrons [21]. a(
Mais recentemente, a espectroscopia de absorção de raios-X tem sido 01
utilizada por um grande número de pesquisadores no estudo dos processos de
transição 'de fase em materiais de estrutura peroviskita [22). A utilização de uma
el
técnica que forneça informações sobre a estrutura local dos materiais é considerada
e:
como complementar aos métodos de difração. rj
O fenômeno de transição de fase ferroelétrica tem sido tradicionalmente
a
classificado como transição do tipo deslocamento ("displacive") ou ordem-desordem ("order-
T
disorder") [20]. Para entender as diferenças entre estes dois tipos de transição, vamos
e
considerar um modelo simples de potencial unidimensional e com uma transição de fase
P
estrutural contínua de segunda ordem [20).
1
Em uma transição do tipo "displacive", a energia potencial do sistema em
função do deslocamento coletivo dos átomos em todas as celas unitárias, varia com a
il
temperatura como indicado na Figura 1.

Te
~~
~\~T
o
j~
~ L---~~--~o--~==--~
collective otomic displocement
otomic displocement
••
Figura 1- Esquema da dependência com a temperatura da energia potencial para (a)
modelo "displacive" onde os deslocamentos atômicos são iguais e na mesma direção
em todas as celas unitárias; (b) modelo de ordem-desordem onde os deslocamentos
atômicos não são necessariamente correlacionados com de outras celas unitárias. As
linhas sólidas representam a probabilidade de ocupação em cada mínimo de energia
em uma cela unitária.
91
 No modelo displacivo, abaixo de Te, O potencial é um duplo poço e o 8itit(Jllul
permanece em um dos mínimos em uma posição fora do centro de simetria, A medída
que a temperatura aumenta, a energia potencial varia e a distância entre os dois poços
e a posição do átomo com relação ao centro de simetria diminui de maneira contínua.
Em T=Te, a barreira entre os dois poços deixa de existir. Acima de Te, a energia
potencial é descrita por um único poço e o sistema não apresenta distorção. Assim,
abaixo de Te todos os átomos estão deslocados em uma mesma direção enquanto
acima de Te, todos eles ocupam uma posição centro simétrica. Portanto, o sistema é
ordenado em todas as temperaturas.
No caso de transição do tipo ordem-desordem, a energia potencial de um átomo
em função de seu deslocamento é um poço duplo em todas as temperaturas e é
essencialmente independente da temperatura como indicado na Figura 1b. Além
o disso, as celas unitárias vizinhas interagem de tal modo que quando duas celas
Ite
apresentam uma distorção na mesma direção, a energia do sistema diminui. Acima de
~r-
Te. O sistema tem igual probabilidade de ocupar tanto o poço à direita como o poço à
)s
esquerda. Abaixo de Te. O sistema adquire uma distribuição preferencial de distorção.
Assim, o sistema é desordenado acima de Te e parcialmente desordenado abaixo de
Te.
A transição de fase estrutural nas amostras de estrutura peroviskita tem sido
interpretada como sendo do tipo "displacive". [27]. Entretanto. a existência de uma
desordem local em temperaturas bem acima de Te, levou muitos pesquisadores a
supor que além da transição de fase do tipo "displacive", poderia estar ocorrendo
simultaneamente uma transição do tipo ordem-desordem [26].
A técnica de espectroscopia de absorção de raios-X tem sido utilizada com
sucesso no estudo de transições de fase nesta classe de materiais pelo fato de ser
sensível a existência da "desordem" em materiais cristalinos. Para entender porque a
técnica de XAS apresênta uma maior facilidade de fornecer informações quando
comparadas a outras técnicas de caracterização estrutural como espalhamento
Raman, Ressonância Magnética Nuclear (RMN) e difração de raios-X, nós devemos
comparar a escala de tempo das medidas e o tipo de média estrutural que ocorre
durante a realização de uma medida. Se quisermos observar distorções que são
dinâmicas, ou seja. não são de origem topológica, a escala de tempo característica da

92
medida deve ser menor que o tempo característico no qual o sistema apresenta
alguma mudança. Por exemplo, o tempo característico de uma medida de ressonância
magnética nuclear e de espectroscopia Raman é da ordem de 10-10s [20]. Como as
distorções locais são dinâmicas e muito rápidas, em temperaturas acima de Te, estas
técnicas perdem sua habilidade em determinar e quantificar a existência de desordem
a nível local. Por outro lado, o tempo característico de uma medida de XAS é da
ordem de 10-15S.Este tempo é determinado pelo tempo de vida do buraco criado no
nível profundo dos átomos pelos raios-X [20]. Assim, a técnica de XAS permite a
determinação e observação da maioria dos problemas de distorção ou desordem local.
A medida por difração de raios-X é também muito rápida. Entretanto, a análise usual
dos resultados de difração assume que o modelo é periódico. Como a componente
periódica da estrutura acima de Te é simétrica, a existência de distorções locais não
será observada a menos que um outro modelo seja assumido [22].
Com a finalidade de verificar se os resultados obtidos na literatura quanto aos
mecanismos predominantes durante o processo de transição de fase em amostras de
estrutura peroviskita podem ser estendidos a sistemas mais complexos, realizamos
um estudo estrutural do sistema Pb1-xlaxTi03 (PLT) em função da temperatura para
algumas composições previamente selecionadas.
Os resultados obtidos quando do estudo do sistema Pb'-xLaxTiO, (PLT) em I,

função da temperatura encontram-se resumidos em um artigo [Artigo IV.E].

Medidas de (DRX) e (XAS) foram realizadas em duas amostras. Uma amostra
de composição Pbo,80Lao,20
Ti03 (denominada PLT20) e uma amostra de composição
Pbo,70Lao,30Ti03
(denominada PLT30). Estas duas amostras foram escolhidas pelo fato
de que a amostra PLT20 é representativa das amostras do sistema que apresentam
um comportamento ferroelétrico normal enquanto que a amostra PLT30, é
representativa de uma amostra do sistema que apresenta um comportamento
ferroelétrico relaxor. .•
As medidas de DRX foram realizadas em duas temperaturas abaixo e duas
temperaturas acima da temperatura de transição de fase Te. As medidas de XAS
foram realizadas para a amostra PLT20 em duas temperaturas acima e duas
temperaturas abaixo de Te, enquanto que para a amostra PLT30, foi somente
realizada uma medida abaixo e uma medida acima de Te.

93
 Para a amostra PLT20, observou-se uma transição de fase cristalográfica em
torno da temperatura de permissividade elétrica máxima, Te. Os dados de DRX
mostraram que abaixo dessa temperatura de transição (343K), a estrutura é tetragonal
(grupo P4mm), e que poucos graus acima (373K), ela já é cúbica (Pm-3m). Estes
resultados estão em bom acordo com a curva de permissividade elétrica desta
amostra [14]. Desse modo, os dados de DRX mostraram que para uma temperatura
logo acima de Te a estrutura é cúbica e o átomo de titânio, cuja ordem local estamos
caracterizando por XAS, estaria um octaedro perfeito com seis distâncias Ti-O iguais.
Medidas de DRX feitas em temperaturas mais elevadas não mostram modificações da
estrutura.
Para a amostra PLT30, cuja temperatura de transição é aproximadamente 223
K, os dados de DRX mostraram que a 93 K a estrutura é tetragonal (grupo P4mm) e
que a 293 K é cúbica (Pm-3m). Entretanto, neste caso, foi observado que para
temperaturas 30 K acima da Te, O difratograma de raios-X apresenta o mesmo padrão
de difração da medida feita a 93 K, ou seja, uma simetria tetragonal. Esse
comportamento pôde ser explicado pelo formato difuso da curva de permissividade
dessa amostra [14]. Assim, como para a amostra PLT20, os resultados de DRX para a
amostra PLT30 indicam que para temperaturas acima da transição de fase a estrutura
é do tipo cúbica e do ponto de vista da ordem local, o átomo de titânio está localizado
em um sítio regular com seis distâncias idênticas Ti-O.
)
Os resultados obtidos através da técnica de XAS na borda K do átomo de titânio
mostraram que para ambas as amostras, o átomo de titânio apresenta uma distorção
local com relação aos átomos de oxigênio da primeira esfera de coordenação quando
analisados em uma temperatura abaixo e acima de Te. Além disso, os dados de XAS
mostraram claramente que o grau de desordem não é alterado quando passamos de
uma temperatura bem abaixo de Te para uma temperatura bem acima de Te.
III

Em resumo, os resultados obtidos para o sistema PLT se assemelham aos

resultados obtidos para outros sistemas mistos e mesmo para amostras de
composição AB03. O fato de observarmos do ponto de vista da ordem local ainda
exista desordem acima de Te, indica que o mecanismo de transição de fase neste
sistema tem uma componente importante da transição do tipo ordem-desordem. Este
estudo deverá em breve ser estendido ao sistema PBZT.

94
IV.4- Conclusão

Em função dos resultados obtidos e da análise dos resultados que estão em fase
de análise, podemos dizer que a utilização da técnica de XAS como ferramenta de
caracterização estrutural foi muito útil.
No caso do estudo do processo de cristalização em amostras de composição
AB03, a utilização da técnica de XAS tornou possível estudar a etapa inicial do
processo de cristalização quando o material se encontra na forma amorfa. A
caracterização da estrutura da fase intermediária cristalina foi somente possível
através da técnica de XAS. A partir da identificação da estrutura local do átomo de
titânio, foi possível melhor compreender o mecanismo de cristalização destes
materiais.
Com relação ao estudo das amostras cerâmicas ferroelétricas de composição
Pb1-xLaxTi03 (PLT), o uso da técnica de XAS juntamente com a técnica de DRX
possibilitou uma melhor compreensão das mudanças estruturais à medida que a
quantidade de lantânio era variada. A extensão deste estudo a outros sistemas pode
levar a uma melhor compreensão do fenômeno de transição de fase difusa e
comportamento relaxor nestes materiais.
Os estudos de transição de fase ferroelétrica-paraelétrica utilizando as técnicas
de XAS e DRX, mostraram ser muito úteis na compreensão do fenômeno. Os
resultados obtidos para o sistema Pb1-xLaxTi03 mostraram que o mecanismo de
ordem-desordem é predominante. Os estudos de transição de fase serão estendidos a
outros sistemas com a finalidade de melhor compreender estes fenômenos e verificar
a qual ponto os resultados por nós obtidos possam ser generalizados.

IV.S- Referências
••

1- M. Tokoda, Y. Hamaji, K. Tomoro, D.A. Payne, Jpn. J. Appl. Phys.32, 4158 (1993).
2- V. Craciun, RK. Singh, Appl. Phys. Lett. 76, 1932 (2000).
3- RN. Schwartz, B.A. Wechsler, L. West, Appl. Phys. Lett. 67, 1352 (1995).
4- L.H. Parker, B.A. Tasch, IEEE Circuits Devices Mag. 6, 17 (1990).
5- (a) R Rovai, C.W. Lehmann, J.S. Brandley,Angw.Chem.lnt.38,2036 (1999); (b)

95
H.P. Baldus, M. Jansen,Angw.Chem. Int. 36, 603 (1997); (c) A.O. Gabriel, R.
Riedel, Angw.Chem. Int. 36, 384 (1997); (d) Y. Xu, C.H. Cheng, J.D. Mackenzie, J.
Non-Cryst. Solids 176, 1 (1994); J.D. Mackenzie, Y.Xu, J. Sol-Gel Sci.Technolog. 8,
673 (1997).
6- M.P.Pechini,U.S.Patentn3,330,697,(1967).
7- (a) E.R Leite, C.M.G. Souza, E.Longo, J.A. Varela, Ceramic. Inter. 21, 143
(1995).(b) M.A.L. Nobre, E. Longo, E.R Leite, J.A. Varella, Mater. Lett. 28, 215 (1996);
(c) S.M. Zanetti, E. Longo, J.A. Varela, E.R Leite, Mat. Lett. 31, 173 (1997).
8- RW. Schwartz, J.A. Voigt, B.A. Tutle, D.A. Payne, T.L. Reichert, RS. Da Salla, J.
Mater. Research 2, 444 (1993).
9- (a) E.R Camargo, M. Kakihana, Chem. Mater. 13, 1181 (2001). (b) E.R
Camargo, J. Frantti, M. kakihana, J. Mater. Chem. 11, 1875 (2001).
10- D. Fu, S. Hisao, K.S. Suzuki, Phys. Rev. B 62, 3125 (2000).
11- K. Yao, W. Zhu, X. Yao, L. Zhang, Mater. Sci Eng. B41, 322 (1996).
12- RC. Buchanan: Cer. Mater. for Elec., 2nd edn. (Marcel Dekker, NewYork 1991).
13- B. Jaffe, W.R Cook and H. Jaffe, "Piezoelectric Ceramics", Acad. Press, New
York, (1971).
14- E.N. Moreira, Tese de Doutorado, Departamento de Física - UFSCar (1996).
15- Abramov, Y.A.; Tsirelson, V.G.; Zavodnic, V.E.; Ivanov, S.A.; Brown, 1.0. Acta
Crystallographica. 1995, B51, 942.
16- (a) Tae-Yong, K.; Hyun, M.J. Applied Physics Letters 2000,77,3824; (b) Tavares,
E.C.S.; Pizani, P.S.; Eiras, J.A. Applied Physics Letters 1998, 72, 897.
17- (a) F. Jona, G. Shirane, Ferroelectrics, (Pergamon Press, Oxford) (1962); (b) A.D.
Bruce and RA. Cowley, Struc. PhaseTrans., (Taylor & Francis Ltd, London) (1981).
18- (a) S.A. Mabud, A.M. Glazer, Acta Cryst. 12, 49-53 (1979); (b) RJ. Nelmes, RO.
Pilitz, W.F~ Kuhs, Z. Tun, R. Restori, Ferroelectrics 108,165-170 (1990).
fi

19-A. Guinier, in "X-Ray Diffraction in Crystals, Imperfect Crystals and Amorphous

bodies", chap. 6, Dunod, Paris, 1956 and W. H. Freeman, San Francisco, 1963.
20- (a) N. Sicron, B. Ravel, Y. Yacoby, E.A. Stern, F. Dogan, J.J. Rehr, Phys. Rev. B
50, 13168 (1994); (b) E.A. Stern, Y. Yacoby, J. Phys. Chem. Solids 57(10), 1449
(1996); (c) T. Miyanaga, D. Diop, S./. Ikeda, H. Kon, Ferroelectrics 274, 41 (2002).
21- (a) M.E. Unes and A.M. Glass, Principies and Applications of Ferroelectrics and

96
Related Materiais, Clarendon Press, Oxford (1977); (b) S. Ghose, In: Solid to Solid
Phase Transformations (Ed. By W.C. Johnson, J.M. Howe, D.E. Laughlin and W.A.
Joffa) The minerais, Metais and Materiais Soe., Warrandale, PA, p. 719 (1994).
22- (a) RJ. Nelmes, W.F. Kuhs, Solid State Commum. 54, 721 (1985); (b) RJ.
Nelmes, RO. Piltz, W.F. Kuhs, Z. Tun, R Restori, Ferroelectrics 108, 165 (1990).

97
 Artigos

IV. A Espectroscopia de Absorção de Raios-X (XAS) Aplicada no Estudo

Estrutural de Materiais Óxidos

IV. 2. Estudo do processo de cristalização em amostras de composição AB03

Artigo-IV.A- Crystallization Study of SrTi03 Thin Films Prepared by Dip Coating

Artigo-IV.B- Topotatic-like phase transformation of amorphous lead titanate to cubic lead

titanate

Artigo-IV.C- Phase evolution oflead titanate from its amorphous precursor synthesized by the
OPM wet-chemical route

IV.3- Estudo da estrutura local de cerâmicas ferroelétricas

Artigo-IV.D - XAS and XRD Structural Characterization of Lanthanum Modified PbTi03

Ceramic Materiais

Artigo-IV.E - Short and Long-Range Order Structure in Pbl-xLaxTi03 Ceramics Belowand

Above the Ferroelectric Pbase Transition determined by XAS and XRD

98
Materiais Research, VaI. 2. No. 2. 93-97. 1999. © 1999

Crystallization Study of SrTi03 Thin Films

Prepared by Dip Coating

Edson R. Leitea, Valmor R. Mastelarob*, Sõnia M. Zanetti", Elson Longo"

aUniversidade Federal de São Carlos, Departamento de Química,
13560-905 São Carlos - S.P., Brazil
bUniversidade Federal de São Carlos, Departamento de Engenharia de Materiais,
13560-905 São Carlos - S.P., Brazil
*e-mail: valmorsêif.sc.usp.br

Received: February 27,1998; Revised: March 22,1999

ln this study, the crystallization process of SrTi03 thin films, prepared by a chemical method,
was characterized by Fourier Transformed Infra Red (FT-IR), Grazing Incident X-ray Diffraction
(GIXRD), Therrnal Analysis (TG) and X-ray Absorption Near Edge Structure (XANES). The results
showed that an amorphous inorganic phase is formed, consisting of clusters of oxygen five-coordi-
nate titanium. The amorphous phase begins crystallizing at temperatures above 450°C. No
intermediate crystalline phase and no preferential orientation was observed for films deposited on
MgO (100). FT-IR results suggest the presence of the carbonate group. However, the low thermal
stability of this group and the low crystallization temperature of the observed SrTi03 phase indicate
that this carbonate is adsorbed.

Keywords: thin films, SrTi03, characterization

1. Introduction standard glass nucleation approach, proposed a model in

Ferroelectric thin filrns have been intensively studied
due to the intense technological and scientific interest in
their potential applications. Arnong the severa! thin filrn
processing techniques, chernical rnethods such as sol-gel
and Metalorganic Decornposition (MoI5) have received
special attention due to the possibility of strict stoichiornet-
ric control, the low deposition costs involved (sirnple depo-
sition rnethods such as dip and spin coating can be used)
and the large substrate areathat can be covered. There are,
essentially, three different types of sol-gel: colloidal sol-
gel, inorganic polyrneric gel derived frorn organornetallic
cornpounds and gel routes involving the formation of or-
ganic polyrneric glass (polyrneric precursor rnethod). The
polyrneric precursor rnethod, based on the Pechini proc-
ess', has been successfully used to obtain cerarnic pow-
ders2-4 and cerarnic thin filrns5-7. •
During the crystallization process of thin filrns chemi-
cally prepared by the polyrneric precursor rnethod, an in-
terrnediate inorganic arnorphous phase is observed after the
pyrolysis step5-7. As a consequence, the thin filrn crystal-
lization process resernbles the inorganic glass crystal-
lization processo Recently, Schwartz et a/.8, using a
which the degree of orientation and rnicrostructure of thin
filrns prepared by MOD are correlated with the therrnody-
narnic barrier to nuc1eation. They showed that a change in
the crystallization driving force results in a rnicrostructural
rnodification due to the variation that occurs in the barrier
heights for interface (film/substrate) and surface nuclea-
tion. In another recent paper, Gust et al.9 showed that
BaTi03 thin filrns prepared by sol-gel present an interrne-
diate phase, presurnably BaTi02.C03, that is subsequently
transforrnéd into BaTi03. Random nucleation in the film
was observed for different substrates.
G. Braunstein et al.lo studied the crystallization of
SrTi03 thin filrns prepared by MODIO• They observed an
interrnediate crystalline carbonate phase and suggested that
this phase delays crystallization and prornotes a randorn
nuc1eation processo
The objective of this work was to study the crystal-
lization of SrTi03 thin filrns prepared by the polyrneric
precursor rnethod. Prelirninary studíes? have shown that the
SrTi03 phase crystallizes frorn an inorganic amorphous
phase, with no crystalline interrnediate phase, as observed
during the powder synthesís '. Thus, this perovskite oxide
rnight be a suitable material to investigate the crystal-

* Present address: Universidade de São Paulo, Instituto de Física de São

CarJos, BraziJ.

99
94 Leite et ai. Materiais Research

lization of thin films prepared by chernical methods. This

study may contribute to a better understanding of the crys-
tallization of thin films prepared by sol-gel.

2. Experimental Procedure
2.1. S'ynthesis and deposition

The precursor solution was prepared by the dissolution

of.strontium carbonate in a aqueous titanium citrate solu-
tion, prepared from titanium isopropoxide. The molar ratio
Mixture 11-+--------'
between titanium and strontium was 1.00 and the cítríc
acid/metal ratio was set at 1.28 (molar ratio). Ethylene
glycol was added to lhe citrate solution to promote polym- 1 CNMetal = 1.2-1.8:1
erization between the titanium citrate and strontium citrate. i; s-r-s"-O-IU-ti-O-n-'
The citric acid/ethylene glycol ratio was fixed at 40/60
. (mass ratio). Figure I presents a flow chart for thin film
preparation using the polymeric precursor method.
A polyrneric solution with a viscosity of 7 mPa s was
used for dip coating deposition. Prior to coating, the MgO
(100) substrate was cleaned by immersion in a sulfo-
chromic solution, followed by rinsing several times in
deionized water. The dip coating was conducted by ímmer-
sion of the cleaned MgO (100) substrate (1 em x 1 cm) in
the polymeric solution, followed by controlled withdrawal
at a speed of 0.7 cm/min. After deposition, substrates were
dried on a hot plate (-150 "C) and heat-treated at 325 °C
for 8 h in an oxygen flow (pyrolysis step). After pyrolysis,
the films were heat treated at different temperatures for 2 h
(crystallization step).

2.2. Characterization
Phase evolution was characterized by GIXRD, using
=
Cu Ka radiation with a grazing incident angle of a 2 and 0
LiF (100) as monochromator, and FT-IR in the transmis-
sion mode. The weight loss and pyrolysis behaviors of the
gel-derived powders were studied by TO. The microstruc-
ture was characterized by transmission electron micros-
copy (TEM).
Ti K-edge XANES spectra were collected at the LNLS
(Laboratório Nacional de Luz Sincrotron, Campinas.Bra-
zil) facility using the EXAFS (Extended X-Ray Absorption
Fine Structure) beam station. The LNLS storage ring was
operated at 1.36 GeV and 60 - 100 mA. Data were collected
at the Ti K-edge (4966 eV) in the total electron yield mode
for thin film samples, and in the transmission mode for
powder samples, The beam ;"as monochromatized using a
Si (111) channel-cut monochromator and the energy step
was equal to 0.5 eV. The energy calibration of the mono-
chromator was checked after each spectrum using a Ti
metal foil to provi de good energy reproducibility.
Crystalline powder samples were used as structural
references: r-Ti02 (rutile, a commercial powder standard).
Solid state reaction was used for ~-Ba2Ti04: BaC03 and
Ti02 were mixed in the stoichiometric amount required for
Crystalline thin film
Figure 1. Flow chart for lhe preparation of SrTi03 thin film using lhe
polymeric precursor method.
the desired phase and heat-treated at 1100 °C for 4h.
Ba2TiShOs (fresnoite) was obtained from a glassy sample,
SrTi03 were prepared by the polymeric precursor processo
The Ti atoms present coordination 6 in r-Ti02 and
coordination 4 in ~-Ba2Ti04, while they are coordinated by
5 oxygen atoms in Ba2TiShOg. The XANES experimental
data were analyzed using the (XAS) programs of Micha-
lowicz!'. The XANES spectra were normalized to an edge
jump of unity, after removal of lhe background absorption,
by subtraction of a linear function extrapolated frorn the
pre-edge.
3. Results and Discussion
3.1. XANES results
Figure 2 presents the Ti K-edge XANES spectra of
SrTi03 crystallized thin films compared to SrTi03 crystal-
lizedrphase spectra, while Fig. 3 shows lhe Ti K-edge

100
Vol. 2,No. 2,1999 SrTi03 Thin Films 95

XANES spectra of SrTi03 crystallized thin films compared

to SrTi03 amorphous thin films. The pre-edge feature of a
Ti K-edge XANES spectra located ar approximately 4970 .,o SrTi03
eV are usually attributed to energy leveI transitions from Ti Crystallized powder
fd
ls to the Ti3d/02p molecular orbitaI12•14• Dipole selection •..
.D

o
.- -'. ..- - _. .
rules forbid ls -7 3d transition (Laporte) but allow this '"01
.D
' .. '

-o . SrTi03
when p-d orbital mixing occurs, as is the case when Ti is .: Crystallized thin film
'"
.~
located in a Ti04 tetrahedron or in a ([5ITiO)04 site (i.e., (ij

without a center of symmetry) 14. The height and position §

of the pre-edge feature are direct functions of the degree of
p-d mixing, site distortion and oxidation stateI2.14.
Qualitative information can be obtained from XANES
spectra using model compounds. Farges'" made an experi-
mental study of the ls -7 3d transition as a function of Ti
coordination ([4)Ti, [51Tiand [6ITi). Based in the Is -7 3d
energy position and normalized height, Farges obtained
structural infonnation ofTi atoms in crystalline and glassy
fresnoites!". Using the infonnation obtained by Farges and
comparing the XANES spectra of SrTi03 compounds pre-
sented in Figs. 2 and 3, we were able to obtain qualitative
infonnation about Ti atoms in our samples.
The Is -7 3d energy position and Ti eoordination num-
ber for model compounds are presented in Table I. As can
be observed, the position and height ofthe Is -7 3d transi-
tion depends on the Ti coordination number.
We then compared the l s -7 3d energy position of
SrTi03 erystallized and amorphous powder samples and
the Is -7 3d energy position of erystallized and amorphous
thin films. The position of the Is -7 3d transition and the
Ti coordination number for these four samples are also
presented in Table 1. As ean be observed, the erystallized
thin film presents a similar Ti coordination when compared
to the SrTi03 erystallized powder phase. A mixture of Ti
atoms with coordination fi ve and six was observed for these
two samples. However, comparison of the erystallized thin
film with the amorphous one showed a significant change
o
Z
4950 4960 4970 4980 4990 5000 5010 5020
Energy / eV
Figure 3. Ti K-edge XANES spectra of SrTi03 crystallized thin film
compared to lhe SrTi03 arnorphous thio films.
in Ti coordination. In the amorphous thin films most of the
Ti atoms presented coordination five, with a small number
of Ti atoms in coordination four and six. These primary
results indicate that the eoordination of the Ti atoms in the
SrTi03 thin films depends on their physical state, i.e.,
whether they are crystalline or amorphous.
3.2. Phase evolution and microstructure
Figure 4 presents GIXRD pattems of SrTi03 thin films
on MgO (100) substrate he~t treated at different tempera-
tures for 2 h. At 325 "C a diffuse XRD pattem, related to
an amorphous phase, is observed. The film remained in the
amorphous state up to 450 "C, TO analysis up to 450 "C of
the polymeric precursor, heat-treated at 325 "C for 8 h in
an 02 flow, showed no weight loss. This result suggests that
the amorphous film, characterized by XANES and GIXRD,
consists of an inorganic amorphous phase. At 500 "C we
observed an XRD pattem of crystalline SrTi03 single
phase. No preferential orientation is reported between
500 "C and 800 "C, whieh characterizes a polyerystalline
thin filmo Fig. 5 presents a TEM micrograph of the film heat
treated at 600 "C, Well defined grain strueture and crack
free microstructure are observed. It is worth noting that no

• SrTiO)

::i
ai

~
o
U
c#= •

~.: : ~\
~
•
I\..
800·C

6OO·C

500·C

4S0·C

4950 4960 4970 4980 4990 5000 5010 5020 ----~-~~ 32S·C
Energy I eV
30 29 50
Figure 2. Ti Ksedge XANES spectra of SrTi03 crystallized thin film Figure 4. GIXRD pattems of SrTi03 thin film 00 MgO (loo) heat treated
compared 10 lhe SrTi03 crystallized powder sample. ai different temperatures,

101
96 Leite et ai. Materiais Research

Table 1. 1s ~ 3d transition position and Ti coordinaiion number for the reference compounds (*) and for the SrTi03 samples.

Samples Is ---? 3d Position (eV)

*Ba2Ti04 4969.2 100
*BalTiShOg 4970.5 100
*r-Ti02 4971.3 100
SrTi03 crystallized powder 4970.8 o <40 >60
SrTi03 amorphous powder 4970.1 ::::30 ::::40 ::::30
SrTi03 crystallized thin films 4971.0 O < 30 >60
SrTi03 amorphous thin films 4970.5 < 10 >90 <10

intermediate erystalline phase, sueh as a earbonate phase,

was deteeted by the GIXRD analysis, A earbonate interme-
diate phase was characterized by XRD during the SrTi03
powder synthesis using lhe polymerie precursor method''.
Figure 6 presents the FT-IR speetrum of lhe film heat-
treated at different temperatures. At 325°C, only vibrations
related to the earboxyl group (COO-) at 1620, 1402 and
1320 em-I are observed. At 400 and 450°C, the vibration
at 1620 em-I was not observed and the peak intensity at
1402 em-I appeared to deerease. A weak broad peak at
1460 em-I is observed at these temperatures. The peak at
1460 em-I might be related to a earbonate phase (C03-2
group), However, the earbonate phase identified by FT-IR
during the SrTi03 powder synthesis, using the same
method, showed a highly aeeentuated peak at 1450 em-I 3.
Using FT-IR, Braunstein et al. 10 observed a earbonate peak
at 1425 em-I for SrTi03 thin film prepared by MOD, for
samples heat treated at temperatures ranging frorn 500 to
600°C. They observed the erystallization of the SrTi03 Figure 5. TEM micrograph of the SrTi03 thin film heat treated at 600°C,
phase at temperatures above 700°C. In our work, the peak 2h.
at1460 em-I is not observed at temperatures above450 "C.
Hence, the broad peak observed in the amorphous thin film,
associated with 10w thermal stability, is likely related to a 80
non-structural C03-2 group, i.e., an adsorbed earbonate
group.

3.3. General discussion

The experimental results showed that, during the crys-,

tallization proeess of SrTi03 thin films prepared by the
polymerie precursor method, an intermediate inorganie
amorphous phase was formed tt"ter the pyrolysis proeess. 50
The XANES results showed that the amorphous phase is
basieally eomposed of smaJI clusters formed by oxygen 1700 1600 1500 1400 1300

five-coordinated Ti «[5ITiO)04). Thus, during the transi- Wavenumber/ em"

tion from amorphous to crystalline, a signifieant transfor- Figure 6. Fr-IR spectra of a thin film heat treated at dlfferent tempera-
mation oecurs in the Ti-O clusters, as indieated in reaetion tures,
(1):
As. deseribed in Eq. 1, the oxygen five-coordinated Ti
(f5lyiO)04 {Amorphous)T > 450 °C ~ in the amorphous stare will transform into oxygen six-co-
[61Ti06 {Crystal} (1) orillilDted Ti (Ti in oetahedral position) in the crystalline

102
Vol. 2. No. 2.1999 SrTi03 Thin Films
state. Crystallization around 500 °C is close to the crystal-
lization temperature of SrTi03 thin films amorphized by
ion implantation 10 or prepared by physical vapor deposition
(PVD)16. This low temperature crystallization is a strong
indication that crystallization occurs directly fram the
amorphous state, with no intermediate phase involved.
The film remained in the amorphous phase up to
450 "C. The formation of a polycrystalline thin film, for
films heat-treated at temperatures above 450 °C, suggests
a nucleation process controlled by surface nucleation. This
nucleation process will promote crystallization from the
surface/film interface towards the filmlsubstrate interface.
Thus, no epitaxial growth ís observed in the
SrTiO:YMg0(100) system.
4. Conclusion
These experimental results lead to the following con-
clusions: after the pyralysis step, an inorganic amorphous
phase is formed consisting of small clusters of oxygen
five-coordinated Ti (([51TiO)04); the amorphous phase be-
gins to crystallize at temperatures above 450 °C, with no
preferential orientation and with no intermediate crystal-
line phase; FI'-IR results suggest the presence of what is
prabably an adsorbed carbonate group.
Acknowledgments
The authors gratefully acknowledge the following Bra-
zilian research financing agencies: FAPESP, CNPq-
PADCf m, CAPES and PRONEXJFINEP. This research
was partially performed at LNLS National Synchrotron
Light Laboratory, Brazil.
References
1.Pechini, M.P. V.S. Patent n. 3,330,697, 1967.
11I
97
2. Kakihana, M. 1. sst-casa. Technol., n. 6, p. 7, 1996.
3.Leite, E.R.; Souza, C.M.O.; Longo, E.; Varela, I.A.
Ceramic Item., n. 21, p. 143, 1995.
4. Nobre, M.A.L.; Longo, E.; Leite, E.R.; Varela, 1. A.
Mater. Lett., n. 28, p. 215, 1996.
5. Bouquet, V.; Leite, E.R.; Longo, E.; Varela, I.A. Key
Engineering Materials, v. 132-136, p. 1143, Trans.
Tech. Publications, Switzerland, 1997
6. Liu, M.; Wang, D. J. Mater. Research, n. 10, p. 3210,
1995.
7. Zanetti, S.M.; Longo, E.; Varela, I.A.; Leite, E.R.
Mater. u«, n. 31, p. 173,1997.
8.Schwartz, R.W.; Voigt, I.A.; Tuttle, B.A.; Payne,
D.A.; Reichert, T.L.; DaSalla, R.S. J. Mater. Re-
search,n. 2, p. 444, 1993.
9.0ust, M.C.; Evans, N.D.; Momoda, L.A.; Mecartney,
M.L. J. Am. Ceram. Soe., n. 80, p. 2828,1997.
10. Braunstein, O.; Paz-Pujalt, a.R.; Mason, M.G. J.
Appl. Phys., n.73, p. 961,1993.
11.Michalowicz, A. Logiciels pour Ia Chimie, Société
Française de Chimie, Paris, p. 102, 1991.
12.Waychumas, a.A.Am. Mineral, n. 72, p. 89,1987.
13.Farges, F.; Brown Jr., a.E.; Rehr, 1.1. Geochim. Cos-
mochim Acta. V. 60, n. 16, p. 3023-3038, 1996.
14. Farges, F. Joumal of Non-Cryst. Solids, n. 204, p. 53,
1996.
15.Behrens, P.; Abrnann, S.; Felsche, J.; Vetter, S.;
Schulz-Ekloff, o.. Jeager, N.I.; Niemann, W. J. or
Chem. Soe. Commun., n. 1990, p. 678, 1990.
16.Hubert, T.; Beck, V.; Kleinke, H.J.Non-Cryst. Sol., n.
196, p. 150, 1996.
103

Journal
Topotatic-like Phase Transformation of Amorphous lead Titanate to
Cubic lead Titanate
Department of Chemistry, Universidade
Francesco Lanciotti Jr., Carlos E. M. Campos, and Paulo S. Pizani
Departamento de Física, Universidad
Instituto de Física de São Carlos, Universidade de São Paulo, São Carlos, Brazil
Instituto de Química, Universidade
The crystaUization precess of lead titanate (PT) prepared
using the polymeric precursor metbod was investigated using
X-ray diffractometry, Raman spectroscopy, electron micros-
copy, and X-ray absorption spectroscopy techniques. The
results showed that amorpbous PT was formed by an O-Ti-O
structure composed of fivefold and sixColdoxygen-coordinated
titanium. The local structure of the amorphous PT phase was
similar to that of the cubic PT pbase, Le., similar coordination
number and similar bond lengths, leading to a topotactic-like
transformation during the phase transformation from amor-
phous to eubíc perovskite PT. Because of the low crystaUiza-
tion temperature, every transformation observed during the
crystallization process was associated with a short-range re-
arrangement processo
I. Introduction
EAl) 11TANATE (pbTiOJ, PT) is a well-lmown fenoelectric
L material with a perovskite structure. During PT or léad
zirconate titanate (PZT) syntheses using chemicaI methods, an
intermediate pyrochlore or fluorite crystalline phase has been
reported before the fonnation of the perovskite phase; identifica-
tion of this phase strongly suggests phase segregation or stoichi-
ometric deviation during synthesis:·2
Leite et ae recently showed that the polym.eri.c precursor
method is a suitable synthetic route to obtain tbe PT phase. These
authors reported that tbe cubic perovskite PbTí03 (c-PT) phase
was formed from an inorganic amorphous precursor ata temper-
ature of 444°C. A gradual transition from c-PT to tetragonal
perovskite PbTi03 (t-PT) was observed witb increased calcination
time at this temperature. Hign-resolution transmission elecíron
microscopy (HRTEM) results showed that tbe size of the c-PT
partic1es was -5 nm. The identification of c-PT as an ÍDtennediate
phase supported the hypothesis that chemical ~omogeneity was
kept at tbe molecular levei during the syntbesis process, with no
B. A. lUllle-contributing editor
Manuscript No. 188333. Receíved August 9, 2000; approved April 23. 2002.
Supported by lhe Brazilian funding institutions FAPESP, CNPq (PADCT), and
FINEP.
*Member, American Ceramic Socicty.
2166
1. Anl. Ceram. Soc., 85 [9] 2166 70 (20m)
Edson R. Leite," Elaine C. Paris, and Elson Longo
Federal de São Carlos, São Carlos, São Pãulo 13565-905, Brazil
Federal de São Carlos, São Carlos, São Pãulo 13565-905, Brazil
Valmor Mastellaro
Carlos A. Paskocimas and Jose Arana Varela"
Estadual Paulista, Araraquara, São Pãulo 14800, Brazil
cation segregation. However, it was Dot clear why the c-PT phase
was formed at low temperatures. Crystallization was observed at a
temperature at whieh long-range diffusion had to be constrained,
and the equilibrium configuration was kinetically suppressed, 4.5
The crystallization oeeurred at a low homologous temperature
(T/Tm, wbere Tm is the mehing temperature). The TITm value
observed was 0,345.
This article reports on an investígatíon of the phases formed
during tbe synthesis ofPT processed using the polymerie precursor
method. Pbase evolution has been characterized using X-ray
diffractometry (XRD), X-ray absorption near-edge structure
(XANES), extended X-ray absorption fine structure (EXAFS),
HRTEM, and Raman spectroscopy. Thís report demonstrares that
the crystallízation process occurs by short-range diffusion because
ofthe structural similarity between the amorphous PbTiOJ (a-PT)
and the c-PT. Crystallizatíon in this proeess oceurs at Iow
homologous temperature, where the driving force ís large but the
mobility is negligible, i.e., sufficient to promote only tbe short-
range rearrangements neeessary for crystal fonnation. The long-
range distribution of the species (titanium, oxygen, and lead) must
not oecur.
IL Experimental Procedure
A precursor solution containing metallic cations was prepared
with a molar ratio of Pb(II):Ti(TV) = I: 1. An aqueous solution of
titanium citrate was prepared from títanium isopropoxíde (99% +
purity, Aldricb Chemical Co., Milwaukee, WI) with a concentra-
tion of2.10 wt"1o ofTi02• Ethylene glycol (EG; 99.9%, E. Merck,
Dannstadt, Germany) was added to the titanium citrate solution
with a mass proportion of 40% ofEG to 60% of citric acid (99.9%,
E. Merck). Lead acetate trihydrate (Pb(CH3COO)2'3H20, 99%+
purity, Aldrieb) was dissolved in the aqueous solution oftitanium
citrate to obtain the stoichiometric Pb(II):Ti(IV) = 1:I solution.
The solution was then polymerized at 250°C for 3 h, followed by
a 16 h pxepyrolysis treatment at 300°C in a tube furnace with air
flow. The powder tbus obtained after prepyrolysís is hereinafter
referred to as the amorpbous precursor. The amorphous precursor
was then calcined at various times and temperatures.
The pnase evohnion was followed using XRD (Model D-5000,
Siemens, Karlsmhe, Gennany), with CuKa radiation and a graph-
ite monochromator, and by Raman spectroscopy. Tbe Raman
measurements were taken using a double monoehromator (Model
UIOOO, Jobin-Yvon, Longjum.eau, France) coupled to a cooled
GaÀtl.~otomultiplier and a conventional photon-counting system.
104
3eptember 2002 Topotatic-Like Phase Transformation of Amorphous PbTi03 to Cubic PhTi03 2167

• Tetragonal Phase
(b)
• Cubic Phase

(a)
I
960mln.

460 mm.

240 mio

eOmio.

30mln

20 rtun.

5mln

Precursor

200 400 600 800

Ramso Sbift (em" )

Fig. 1. (8) XRD pattem ofthe amorphous precursor calcined at 444°C for varíous Jengths of time. (b) Raman spectra of the amorphous precursor calcined
at 444°C for various lengths of time.

TheA88.0 nm exciting wavelength of an argon-ion Iaser was used,
with maxímum output power of the laser kept within 200 m W. AlI
the measurements were taken at room temperature.
Titanium K-edge XANES and EXAFS spectra were obtained of
the amorphous precursor powder and the crystalline powder
calcined at 500°C for 8 h at the LNLS (National Laboratory of
Synchrotron Light) facilities in Campinas, São Pãulo, Brazil. The
LNLS storage ring was operated at 1.36 GeV and 60-100 mA.
Data were collected at the titanium K-edge (4966 eV) in the
transmission mode for powder. To monitor the energy calibration,
XAS spectra of a metallic titanium foil was recorded at the same
time as XAS spectra of the samples, using a third ion chamber. The
X-ray beam was monochromatized using a silicon (111) channel
cut monochromator with an energy step equal to 0.5 eV. Crystal-
line r.-Ba2Ti04, B~TiSi208 (fresnoite), and r-Ti02 (rutile) pow-
ders were used as structural references in the analysis of the
XANES spectra of the samples, The titanium atoms in these
structures were coordinated by four, five, and six oxygen atoms,
respectively.
. AIl the EXAFS spectra were treated according to the classic
plane-wave single-scattering approximation,6.7 per the procedure
described in Ref, 8. The EXAFS analysis was conducted using a
microcomputer, using a program set written by Michalowicz,"
according to the recommended procedures described by the Inter-
national Workshop on Standards and Criteria in XAFS.IO After
atomic absorption remova! and nonnalization, the PX(k) weighted
EXAFS signaI was Fourier-transfonned to R distance space in the
3.8-12.5 A -I limits. The energy threshold was selected arbitrarily
at the inflection point ofthe absorption spectra, The contribution of
the fírst neighboring shell was extracted by a back Fourier-
transform in R space and then fitted using theoretical phase and
amplitude functions.
Theoretical amplitude and phase functions were used to model ,
the Ti-O pair, as reported in the MacKale et al,I I tables. The
theoretica1 amplitude and phase functions were first checked on
the crystalline t-PT sample (RTi-<l = J.75 A (X I), RTi-<l = 1.98
A (X 4), and RTi-<l = 2.42 A (X 1».'2
In every fit, the nurnber of free parameters (NpwJ was kept
suspension was deposited on a carbon-covered nickel grid. HR-
TEM facilities were provided by LNLS-Campinas, São Pãulo,
Brazil, and TEM faciJities were provided by the Institute of
Chemistry, UNESP, Araraquara, São Pãulo, Brazíl.
lU. Results and Discussion
(1) Phase Evolution
Figure 1 presents the phase formatÍon characterized by XRD
(Fig. 1(3» and Raman spectroscopy (Fíg. l(b) for the amorphous
precursor calcined at 444°C for various times. This temperature
was considered the probable crystallization temperature, based on
previous differential scanning calorimetry (DSC) analysis.ê From
the XRD analysis (Fig. leal), a broad continuum pattern was
observed for the amorphous precursor and the precursor calcined
for 5 min, consistent with an amorphous structure (no long-range
order). A c-PT crystaIline phase was detected after a 20 min
ca1cination time. A gradual transition ftom c-PT to t-PT was
observed when the calcination time was increased. The analysis of
the R.aman spectra showed no characteristic Rarnan peaks for the
amotphous precursor. However, for the precursor calcined for 5
min, Rarnan features could be observed in the regions of -1 00 and
750 em - J. The shoulder centered at 750 em -I usually has been
related to longitudinal optical phonona'? typical ofthe perovskite
PT • Tetragonal
~
:i
~
c
o
~ a-PT
o
~
'O
Q)

smaller than the number of independent points (Nind), defmed as ~

iã
Nind = 21f.R!l.khr, where tlR is the width of the R-space filter
windows and âk the actual interval of the fit in the k space.? For É
o
the fit made at the titanium edge, the error was estimated to be z
) approximateJy :!::0.01 A in distance and :!::5% in coordination
numbers.
4950 4960 ·4970 4980 4990 5000 5010 5020 5030 5040
The amorphous precursor powder was characterized using
HRTEM (300 kV, ModelJEM 3010 ARP, IEOL, Tokyo, Japan) Energy(eV)
and TEM (200 kV, Model CM200, Philips, Eindhoven, The
Netherlands). For the HRTEMITEM study, a drop ofthe powder Fig.2. XANES spectra oftbe amorphous precursor and erystallíne t-PT.

105
..l168 Journal of lhe American Ceramic Society ..=Leite et a/.. Vol.. 85, No .. 9

0,8
(a) (b) ~ PbTiO. Amorphous
0,3
-- PbTIO. Crystalllzed
.g 0,7
:J
~r::: 0,6
0,0 CI
~ 0,5
g -o.s .eÊ 0,4
Nl'<!

"" ~
-o.e s•.. 0,3

-ê 0,2
::l
o
IL 0,1
-{l,9

0,0
3 4 5 6 7 8 9 10 11 12
o 2 4
K(A"') R(A)

Fig, 3. (a) EX.AFS spectra of the amorphous precursor and crystalline t-PT .. (b) Fourier-transform magnitude of the amorphous precursor and (-PT.

structure. Increased calcination time produced a well-defined
Raman spectrum, typical of the t-PT structure. The XRD pattem
for the amorphous precursor calcined for 5 mín showed no
diffraction peak.. However, the same powder analyzed using
Raman spectroscopy showed a broad peak. This result was a strong
indication that there was short- or medium-range order, such as
Ti-O octahedra or cJusters of Ti-O octahedra. The peak centered
at 750 cm-I was not observed in the amorphous precursor.
However, a typical photoluminescence (PL) signal of a_PTI4 was
observed for this sample. The intensity of PL may have hidden the
Raman scattering associated with the Ti-O octahedron,
(2) X-r4y Absorption Study
Figure 2 presente the XANES spectra of the a-PT and erystal-
line r-PT phases. The spectra display a pre-edge absorption feature
characteristic of titanium compounds.15,16 We have followed the
notation of PilIep et ai. 16 to indicare the main features in our
spectra, which are the pre-edge transitions called A2 and A4.
According to the literature, 16,17 features A2 and A4 are caused by
dipole Is ---7 Ti3d/02p transitions. The splitting ofthe dipole Is _
Ti3d!02p transitíons into features A2 and A4 is caused by .the
energy splitting of the Ti3d orbitals in an octahedral field into t2.g
(dAY' d%Z, d...,J and eg (fi; - dj.,d;) orbitals, respectively.l" These
dipole transitions are forbidden in a centrosymmetric environment
(Laporte's rule), such as in a regular octahedron. Their presence in
the PT spectrum is due to the out-of-center distortiori of the
títaníum atoms, According to this rule, the intensities of the
pre-edge peaks generally decrease with decreasing distortioD.16
Figure 2, which shows the XANES spectra af a-PT and
crystalline l-PT phases, reveals that the A2 and A4 feamres are
more intense for the crystalline l-PT sample, The crystalline l-PT
phase presents a very distorted first-shell structure.12 Thus, from
the XANES results, we expect that the first coordination sheU
around titanium atams in the a-PT phase is less distorted than in
the crystalline t-PT phase ..
The pre-edge feature of our a-PT sample resembles that
obtained for a titanium silicate sample by Behrens et aI..18 Based
on the larger widtb of tbe pre-edge features in their samples,
Behrens et ai. have assumed the coexistence of various titanium
atam environments with octahedral titanium dominating. The
Behrens et ai.. study has demonstrated that their amorphous
compounds were formed basically through fourfold oxygen-
titanium coordination (TiO" tetrahedra), sixfold oxygen-títanium
coordination (Ti06 octahedra), and fivefold oxygen-titanium co-
ordination (TiOs square-based pyramids). We believe that the
same supposition can be made for our amorphous sample, i.e., the
amorphous precursor is formed basically by sixfold oxygen-
titanium coordination (Ti06 octahedra) and fivefold oxygen-
titanium coordination (TiOs square-based pyramids).
Conceming the EXAFS ana1ysis, only the first shell of coordi-
nation (first Ti-O aeighbors) has been taken into account. The
structural determination for the second sheIl of titanium is com-
plex, and the simulation does not fit well the experimental data.
Figure 3(a) shows the EXAFS spectra of lhe amorphous
precursor and crystalline t-PT phases. The Fourier-transformed
magnitude ofthe amorphous precursor and (-PT are shown in Fig.
3(b). The first coordination shell, re1ative to the Ti-O bond, is
well-defined, and, as observed in Fig. 3(b), the amorphous precur-
sor presents a more intense Íll'st peak than does the crystalline
l-PT. Moreover, ta1áng into account the crystalline material, we
observe lhat the peak oí the amorphous material related to the first
Ti-O bonds shifts to higher distances.
Table I presents the fitting results for the first shell (Ti-O) for
the crystalline t-PT sample. In the first shell-fitting procedure,
interatomic distances (R), Debye-Waller factors (o), and energy
shift (AE) were allowed to vary, while the coordination number
(N) was fíxed. Under these conditions, it was possible to obtain a
"good fit. In agreem.ent with the tetragonal structure, 12 three
subshells were used to fit the first shell of the crystalline sample:
one oxygen atam located at 1.76 A, four oxygen atoms Iocated at

Table I. Fitting Results for the First Shell, Showing the Oxygen Coordination Number,
Bond Lengths, and Debye Factors around the Titanium Atom
Coordíaation Meanbondlength, Dcbye-Wallcr Energy shift,
Sample number.Nn-O RTi-O (A)t factor. crTI-O (A)i 4E (eV)

PbTi03, crystallíne l.ot 1.76 0.05 5.0

4.0t 1.96 0.01 2.3
l.ot 2.40 0.12 5.2
PbTiOJ, amorphous 1.0 1.74 0.06 11.0
5.0 1.94. 0.08 9.0
tValuesk:eplconslanlduringlhe fillÍDgprocedwe.t:tO.OI A for lhe first shell. §:tO.02 Á.

106
Scptember 2002 Topotatic-Like Phase Transforn:ation of Amorphous PI;Ti03 to Cubic PhTi03 2169

0,15 --Simulation 0,15

-- Experimental
(a) o Experimental (b)
0,10 0,10
o Simulation

0;05
~ 0,05

i 0,00 0,00
g
-0,05 ,..
~ ~,05

-0,10
~,10
-0,15
~,15
~,2D
~.20
~,25

• 6 8 10 12 6 8 10 12

K(A"')

Flg, 4_ (a) Comparison between experimental and theoretical curves for the crystalline sample (t-PT). (o) Comparison between experimental and theoretical
curves for lhe amorphous precursor, Best simulation was obtained using a two-shell model.

1.98 A, and one oxygen atom locatedat 239 A. To test the validity
of our simulations, we then allowed lhe coordination number to
vary, Under these conditions, either lhe Debye-Waller factor or
lhe coordination number changed appreciably. A comparison
between experimental and Iheoretical curves for lhe crystalline
sample is presented ín Fig, 4(a). Severa! simulatíons of Ti-D
configurations were made for lhe amorphous precursor sample,
The best simulation was obtained with a two-shell model, pre-
sented in Table L A comparison of the experimental and tbeoret-
ical curves for the amorphous sample is illustrated in Fig, 4(b).
This result was in agreement with our XANES results and the
EXAFS results published by Sengupta et aI. 19 for a-PT prepared
using the sol-gel method,
(3) HRTEM Study
HRTEM and selected area electron diffraction (SAED) analyses
have been conducted as complementary structural analyses. The
HRTEM image of the amorphous precursor (Fig, 5) shows a
structure with no long-range order, characterizing an amorphous
structure. The SAED patterns ofthe amorphous precursor (ínset in
Fig. 5) reveal an inner diffuse ring and a second diffuse ring. The
diffraction pattem is related to the Fourier transform of the radial
distribution function, p(r), which gives the probability that an atom
is surrounded by other atoms at a distance r. Therefore, a diffuse
diffraction pattern suggests that p(r} is not periodic, The presence
of diffuse rings suggests only the development of a short-range
order (amorphous structure).
Figure 6 shows a HRTEM image of the amorphous precursor
after a short exposed time to an electron beam. We can observe the
first crystalline particles, with a size of -5 nm. The SAED partem
(inset in Fig, 6) of these small particles has been classified as
perovskite c-PT. In this case, the crystallization is promoted by
local heating induced by the electron beam (in situ crystallization).
These results support the XRD and Raman data and show that the
c-PT particles are in the range of 5 nm, which must contribute to
stabilization of the cubic structure,
(4) General Dlscussion
The XANES experimental results revealed in the a-PT precur-
sor the presence of various titanium-atom environments, with
octahedral titanium dominating, The mean value of the Ti-O bond

1<lg. 6. HRTEM image of lhe amorphous precursor after in situ crystal-

Fig. 5_ HRTEM image of the PT amorphous precursor. íization promoted by local heating, induced by the electron beam.

107
2.170 Journal of the American Ceramic Society=-Leite et al.
length found by EXAFS was 1.92 :!: 0.02 A. The fírst crystalline
phase observed was perovskite c-PT phase, with N = 6 and Ti-O
bond length of 1.98 A?O The shorter Ti-O bond Iength value
observed for the amorphous precursor compared with the c-PT
may have been a consequence oftwo Ti-O structures, Calculations
of ab initio quantum mechanics'" suggested that the chemical
bond between titanium and oxygen (Ti-O) was highly covalent in
character, whereas the bond between the lead cation and oxygen
was basically ionic, This result, together with the EXAFS and
XANES resulrs, suggested that a-PT is formed from a O-Ti-O
nerwork. The charge of the lead cation must have been compen-
sated by negatively charged nonbridging oxygens (NBO). In tbis
model, the lead cation acted as a network modífier, The XANES
and EXAFS analyses of tbe a-PT were based on the well-know
structure of t-PT. The lower symmetry of the I-PT phase in relation
to the c-PT phase was used as a reference to identify amorphous
structures with symmetry higher than that displayed by the t-PT.
Based on the experimental data, it is possible to postulate the
following qualitative model to describe the crystalJization process
of PT processed using the polymeric precursor method
Short-range rearrangements of tbe lead ion must occur. The
lead-ion rearrangement drives tbe transformation of the Ti-O
square-based pyramids (N = 5) to Ti-O octahedra (N = 6). During
the lead-ion motion, short-range rearrangements of the oxygen
ions must occur simultaneousJy to transform N = 5 into N = 6 and
to modify the Ti-O bond length from 1.92 to 1.98 A. Thus, at the
end of the process, a perovskíte c-PT is formed,
The crystallization process must ínvolve shorter-ranged re-
arrangements because of the low homologous temperature in
which the crystallization takes place (TITm .., 0.354). As conse-
quence of the short-range rearrangement process, the equilibrium
configuration must be kinetically suppressed, resulting in a meta-
stable phase (c-PT). The c-PT phase ís formed preferably, because
the amorphous phase is similar to the cubic crystalline phase, with
similar N value and Ti-O bond length. Thus, for lhe crystallization
process, it is postulated that the phase transformation from a-PT to
c-PT is topotactic,
The crystallization of perovskite c-PT is kinetieally more
favorable than is perovskite t-PT. A similar approach has been
used to explain the fluorite phase stability during PZT synthesis
using the 801- gel method. S
There is a relationship between the partiele síze and the ela ratio
for ferroelectric materials.22.23 SmaIler particle sizes lead to a ela
ratio dose or equal to 1.00 (c/a = 1.00 implies a cubic structure).
These previous works do not contradict our results. We show that
the formation of c-PT is kinetically more favorable than is the
formation of t-PT because ofthe low mobility ofthe species at the
crystallízation temperature. The small partic1e size may favor the
cubie structure because of the short-range rearrangement process
involved in the crystallization process of the cubie phase. The
transformation from c-PT to r-PT occurs only by grain growth. The
c-PT is a eonsequence of the topotatic-like phase transfozmation
from an amorphous to a crystalline phase and not aconsequenee of
the partic1e size.
Recent1y Muralt et al.24 have shown that the incorporation of
lead excess into the perovskíte lattice of Pb(Zr,Ti)03 (PZT)
promotes a volume increase of the lattice, modifying the lattice
strueture of the PZT. This result suggests that a chemical hetero-
geneity during the crystallization process ean result in various
lattice struetures for perovskite materials, such as PT. However,
this effect ean be negleeted in this study because of the high
chemical homogeneity of the method of synthesis Used.3•2S
IV. Conc:lusions
The following conclusions can be made from the experimental
data.
Vol. 85. No. 9
(I) a-PT is formed of a mixture of two Ti-O structures, i.e ..
fivefold and sixfold oxygen-coordinated titaniurn.
(2) The amorphous strueture of the a-PT phase is similar to
the c-PT phase, leading to a topotactic-like phase transformation,
(3) Each transformation observed during the crystallization
process must be associated with the short-range rearrangement
process because of the low crystallization temperature.
References
'c. D. E. Lakeman, Z. Xu, and D. A. Payne, "On lhe Evolution of Structure and
Composition in Scl-Gel-Derived Lead Zirconate Titanate Thin Leycrs," J Meter,
Re .e., 10 [81 2042-51 (1995).
28. A. Tuttle and R. W. Schwartz, "Solutioo Deposition of Ferroelectric Thin
Films," MRS Bu/l .• II [6] 49-54 (1996).
'E. R. Leite, E. C. Paris, E. Longo, and J. A. Varela, "Direct Amorphous to Cubic
Perovskite Pbase Transformation for Lead Títanate," J. Am. Ceram. Soc., 83 [6]
1539-41 (2000).
4c. G. Levi, "Metastability and MicrostruclUre Evohitian in the Synthesis of
lnorganics from Precursors," Acta Maler., 46 [3] 787-800 (1998).
5A. D. Polli, F. F. Lange, and C. G. Levi, "Metastability ofthe Fluente, Pyrochlcre.
and Perovskíte Structure in lhe PbO-ZIOz-Ti02 System," J. Am. Cerom. Soc., 83 (5)
873 (20(0).
60. E. Sayers, E. A. Stem, and F. W. Litle, "New Technique for Invesrigaticg
Noncrystalline Structures--Fourier AnaJysis of Extended Xcray Absorptiou Fine
Strucmre," Phy. .•. Rev. Leu., 27, 1204 (1971).
'a. K. Teu. EXAFS: Baste Principies and Daia Analysis: pp. 58-72. Springer
Verlag, Berlín, Gennany. 1986.
·V. Mastelam, S. Benazeth, H. Dexpert, A. Ibanez, and R. Olütrault-Ficher,
"Structure of the Ag-As-Se Cltalcogeoide Glasses: The AsSe-Ag,Se Line,"
J. Non-Cryst. Solid», 151, I (1992).
9 A. MichaJowicz, "EXAFS ponr Ic Mac"; p. 102 in Logiciels pour Ia Chímie.
Sociêté Française de Chimic, Paris, France, 1991.
tO"R.eport ou lhe IntemationaJ Wmbhops OU Standards and Criteria in XAFS"; p.
752 in X'N/)' Ahrorplion Fine Siruaure: Proceedingz of lhe Vi lsuerruuional
Confer"nr;e 011 X-ray Ab!lOrplloll Fine Smsctures. Edited by S. S. Hasuain. Ellis
Horwood, New YOIk. 1991.
liA. G. MacKale, a. W. Paalíkas, A. P. Shauw, and G. J. Knapp, "lmproved
Ab-lnitio Calculatioas of Amplitude and Phase Functions for Extended X·ray
Absorptío» Fine StmctuIe Spectroscopy," J. .ím. Chem. Soe., no, 3763 (1988).
12 A. M. GI""", and S. A. Mabnd, "Powde.- Profile Refinement of Lead Zirconate
Titanate at Seveml Temperatures li Pure PbTiO"u Acta Crystallogr .. Sect, A: Cry .•1.
Pby«. DiDi-., Theor. Gen: Cr)'Sla/logr., 34, 1065 (1978).
13 J. O. Freíre and R. S. Katiyar. "LaUice Dypamics of Crystals with Tetragonal
BaTiO. SúudUre,u Phys. Re». B: CondelU. Mauer, 37. 2074 (19&8).
'4p. S. Pizani, E. R. Leite, F. M. PODtcs.1l. C. Paris, J. H. Raagel, E. Lee, E. Longo.
P. Delega, and J. A. Varela, "Pbotolumincscence of Disordered ABO, Perovskite,"
Appl. Phy», Leu., 71, 824 (2000).
ISF. F"'llcs, "CoordinatiOll of Ti in C!ystaJline and Glassy Fresnoites: A High-
Resolntion XANES Spect:roscopy Sludy at lhe li K Edge," J. Non-Cryst. Soiid.s, 204,
S3 (1996).
"'B. PiUep. M. Froba, M. L. F. Phillips, J. Wong, G. D. Stucky, and P. Behrens,
"XANES Spectros<:opic Sludy oflhe Elec:Irooic S1ruct1Ire of Ti in K TiPO. and Some
ofUs lsomorphous Couq>ounds,u Salid S/ale Commun., 103 [3] 203 (1997).
"T. Uozumi. 1(. Obda, A. Kotani, O. Dunneyer, J. P. Kappler, E. Beaurepaire,
aad J. C. Parlebas, "Experimental and TheoreticaJ lnvestigation ofthe Pre-edge Peaks
ai the Ti K Edge AbsOlptioo SpecIra in Ti~," Euraplry s. Leu., 18, 85 (1992).
tOp. Bchrens.J. Felsehe, S. Vetk:r, G. Schulz-Eldoff, N.l. Jaeger, and W. Niemann,
"AXANES and EXAFS lnvestigalioo ofTitanium Silicate," J. Chem. Soe. Commun .•
678, (1991).
'9S. S. Sengupla, L Ma. D. L. Adler, anel D. A. Payne, "Extended X -ray
Absorption Fine Structure Dctenninatioo of Local Slntcture in Sol-Gel-Derived Lead
Titan~, Lead.Zitt.anate, and Lead.Zirconate Titanatc," J. Mater. Res., 10 [6) 1354
(1995).
2Opowde.- Diffraction File No 40-99. Intemational Centre for Diffraction Data.
Newtown Square, PA, 1992.
21G. Brito, "Quantum Mcchanic Sludy Df Ferroelcct:ric Systems" (in Pcrtuguese);
M.S. Thesís, Federal University of São Caríos, Brazil, 1999.
22K. Uchino. E. Sadanaga, K. ÜOIIÍshi, T. Morohashi, and H. Yamamura,
"ParticlelGrain Síze [)epeodeoce of Fcnoelcctridty"; pp, 107-15 in Ceramíc Trans-
actions, Vol, 8, Ceramic Dlelectrics: Composuion, Proce ss ing, and Properties.
Edited by H. C. Ling and M. F. Yan. American Ceramic Socíety, Westerville, OH,
1990.
23M. H. Frey anel O. A Paync, "Graín-Síze Effcct ou Structure and Pbase
Transfonnation for Barium Titaaate," Ph)'s. Rev. B. Conden s. Mauer, 54 [5]3158-68
(1996).
24p. Muralt, S. Hiboux, C. MucJler, T. Mae<Iaer, L. Sagalowicz, T. Egami, and N.
Setter, "Bxcess Lead in lhe Perovskite Lattice of PZT Thin Films Made by ín-Situ
2sM. Kakihana, T. Okuho, M. Alima, °
Reactive Sputtecing," Jlllegr. Ferroelectr., 36 [1-4]53-62 (2001).
Uchiyama, M. Yashirna. and M.
Y oshimnra, "Polymeric Complcx Synthesis of Perovskite Lead Titanate at R.educed
Temperature: Possible Formalion of Hcterometallic {Pb,Ti)-Citric Acid Complex."
Chem. Maler .• 9,451-56 (1997). O
108
 Available online at www.sciencedirect.com
}OURNALOf
SCIENCE@DIRECT.
SOLlD STATE
CHEMISTRY
Journal of Solid State Chemistry 177 (2004) 1994-2001
htlp:!lcbt!vi&:r.c-()IJvl(Jc(tll·/.ij~C

Phase evolution of lead titanate from its amorphous precursor

synthesized by the OPM wet-chemical route
Emerson R. Camargo,":" Elson Longo," Edson R. Leite," and Valmor R. Mastelaro"
allEC-Deparlamellla de Química. UFSCar. Universidade Federal de São Cartas. Rod. Washington Luiz km 235. São Carlos SP. Caixa Pasta I 676.
13565-905 Stio Carlos, SP. Brazil
b /FSC-/tLuitlllo de Fisica de São Cartas USP. São Cartas SP. CP 369. 13560-970. Brazil
Received 19 September 2003; received in revised forro 25 December 2003; accepted 23 January 2004

Abstract

Lead titanate was synthesized by the OPM wet-chemical rouie by the dissolution of Ti meial in H202 followed by the addition of
Pb2+ at high pH, resulting in a reactive and amorphous precipitate with (pb:Ti= 1:1) mole ratio, which was heat treated between
400°C and 700'C. The amorphous precipitate was characterized by DSC, and ali of the powders were characterízed by X-ray
diffraction, Raman and XAS (EXAFS and XANES) spectroscopy at the Ti K edge. A metastable, stoichiometric and cubic
pyrochlore phase (pb2Ti206, Fá3m) was identified by XRD and Raman spectroscopy up to approx. 450°C. Only tetragonal PbTi03
was identified at higher temperatures. XAS spectra showed that the local structure around the absorbing Ti atom ofthe intermediate
pyrochlore phase is similar to that observed in the amorphous precursor. This fact indicates that the metastable intermediate
pyrochlore (Pb,Ti206) is kinetically favored to be formed because of íts similarity to the amorphous precipitate, instead of the
slightly different and thermodynamically favored tetragonal (pbTi03, P4lmmm) perovskíte structure that is only formed at higher
temperatures, after the crystallization of the metastable intermediate pyrochlore.
© 2004 Elsevier Inc. Ali rights reserved.
Keywords: Lead titanate; Wet-chemical synthesis; OPM route; Metastable intennediate

1. Introductíou other chemical impurities, what is a remarkab1e

Alternative synthetic routes have been developed for
the synthesis of advanced materials, Most of these
materials present some unique properties, which is a
resuIt of the nanometric size of the particles obtained
using some special processing conditions. [1] One of us
developed recently a new synthetic route called "the
oxidant peroxo method (OPM)" used to prepare several
lead perovskites (e.g., PbTi03 and some compositions of
PZT) at relatively low temperature with a sharp particle
size distribution of nanoparticles [1-5] and a high final
yield of approx. 100% [2]. This,.method can be well
described as a molecular level oxy-reduction reaction
between lead ions and some metal-peroxo complexes, at
high pH, that fonns an amorphous, chemically homo-
geneous and reactive intermediate precipitate. This
precipitate is completely free from carbon, halídes or
*Corresponding author, Fax: + 55-16-260-8350.
Eonail address:camargo@liec.ufscar.br (E.R. Camargo).
characteristic of the OPM route when compared to
others conventional wet-chemical methods of synthesis,
such as sol-gel, [6.7] hydrolysis of alkoxides, partial
oxalate [8] and the polymerizable complex methods (also
known as the Pechini method) [9,10]. Moreover, the
OPM technique uses water as solvent and a relatively
simple experimental apparatus, without the necessity of
dry atmosphere or toxic compounds.
Although the inherent advantage of the OPM route,
the nature of the amorphous precursor and the crystal-
lization mechanism is still unknown. Considering that
most of the final properties of a material synthesized by
wet-chemical methods resuIt from the properties of its
initial precursor, we were interested in to describe the
local structure of the amorphous precursor obtained by
the OPM route when applied for the synthesis of lead-
based perovskites. In this context, X-ray absorption fine
structure (XAS) spectroscopy is a powerful character-
ization tool to probe the local structure around a specific
element. [11] In this paper, we describe the structure at

0022-4596/$ - see front matter ít) 2004 E1sevier Inc. AlI rights reserved.
doi: 10.1 016/j.jssc.2004.01.020

109
 ER Camarqo et tiL I Journal oISolid State Chemistry 177 (2004) 1994-2001
the short range order around Ti atoms in the amorphous
precipitate (with mole ratio of Pb:Ti = 1:1) and their
heat-treated powders by the use ofX.A,.Stechnique at the
titanium K-edge. Moreover, additional structural and
chernical information were also obtained by means of
Fourier transform Raman spectroscopy, differential
scanning calorimetry (DSC) and X-ray powder diffrac-
tometry (XRD).
2. Experimental section
2.1. Synthesis
Amorphous precipitate with mole ratio (Pb:Ti = 1:1)
was synthesized as described by Camargo and Kakihana
[2] through the dissolution of 0.250 g of titanium metal
powder (98% Wako Pure Chemical Ind Ltd, Japan) in
an aqueous solution of 80 mL of H202 (30% Mal-
linckrodt, Brazil) and 20mL of arnmonia aqueous
solution (30%, Mallinckrodt, Brazil). After approx.
5 h, a yellow transparent solution of peroxytitanate
complexes was obtained, and a second aqueous solution
of 1.729g of Pb(N03)2 (99.99%, Aldrich, USA) was
slowly dropped into the peroxytitanate solution, form-
ing an orange precipitate that was filtered and washed
with diluted ammonia solution to eliminate all of the
nitrate ions. The washed precipitate was dried at 50°C
for 5 h, ground and calcined at several temperatures
between 200°C and 700°C (from 5 to 60min).
2.2. DSC, Raman and XRD characterization
A small amount of the ground and dry amorphous
precipitate (lO.4mg) was characterized by differential
scanning calorimetry (DSC 404C controIled by TASC
424j3A, Netzsch, Germany) between 25°C and 12OO°C
using a platinum crucible. It was used a constant
heatingjcooling rate of lO°C/min and a fíux of
0.50cm3jmin of air, AlI of the powders (amorphous
and heat-treated) were characterized at room tempera-
ture by Fourier transform Raman spectroscopy (FT
Raman-Bruker RFS 100/S) using the 1063um line of a
YAG Iaser, and by XRD using the eu Ksx radiation
(Rigaku D/MAX 200, with a rotary anode operating at
150kV and 40mA) in the 28 range from 10 to 80° with
step scan of 0.02°.
2.3. XAS characterization 11I
The titanium K-edge (4966eV) X-ray absorption
spectra were collected at the Laboratório Nacional de
Luz Síncrotrom (LNLS, Campinas SP, Brazil) facility
using the D04B-XASl beam line [12]. The LNLS
storage ring was operated at 1.36GeV and 100-
160mA. XAS data of the amorphous and heat-treated
1995
powders were collected in the transmission mode at
room temperature using a Si(lll) channel-cut mono-
chromator, Ionization chambers were used to detect tlie
incident and transmitted fluxo X-ray absorption near
edge structure (XANES) at the Ti K-edge were recorded
for each sample between 4910 and 5200 eV using energy
steps of 0.5 eV. Extended X-ray absorption fine struc-
ture (EXAFS) measurements at the Ti K-edge (using
steps of 2 eV) were carried out for each sample between
4840 and 5900eV. To provide good energy reproduci-
bility during the XANES data collection, the energy
calibration of the monochromator was checked after
each spectrum using a Ti metal foil.
The EXAFS analysis was carried out by using the
software suite written by Michalowicz [I3J according to
the recommended procedures described by the Interna-
tional Workshop on Standards and Criteria in XAS [14J.
After atomic absorption remova 1and normalization, the
k31.(k) weighted EXAFS signal was F?urier transformed
to R distance space in the 2.0-11.0 A -I k range. Each
speetrum was Fourier transformed using a Kaiser
apodisation window with T = 2.5. For comparison
purpose between different samples, alI XANES spectra
obtained at the Ti K-edge were background removed
and normalized using as unity the fírst EXAFS
oscillation.
3. Results aod díscussions
Studying the synthesis of lead titanate (PbTi03,
thereafter denoted PT) by means of the Pechini method,
Leite et alo [15} suggested, based 011 XRD and DSC
analysis, that a cubic-like intermediate phase crystallizes
directly from the amorphous precursor, and that the PT
tetragonal perovskite structure (referred to as tetrago-
nal) is formed from this cubíc-like phase at 444°C when
the temperature is raised. However, it is possible that the
energy released from the burning of the organic matrix,
as observed by presence of exothermic peaks in the DSC
curve, interfered in the crystallization processo 011 the
other hand, Fu et alo [16] reported that nanocrystalline
PT (-7 um) synthesized by sol-ge1 route is orthorhom-
bic at room temperature, changing to tetragonal phase
at 166°C. Although both of the studies show the
presence of different structures, they agree with the
presence of a crystalline phase other than the tetragonal
phase in nanocrystals of PT at 10w temperature.
Fig. 1 shows DSC curves between lOO°Cand 700°C of
the amorphous precipitate during the (a) heating and (b)
cooling steps. The crystalline state ofthe material during
the DSC experiment is indicated on the top of the
curves. The peak centered at 490°C in the heating curve
(Fig. Ia) can be assigned to the phase transition from the
tetragonal to the high-temperature PT cubic paraelectric
perovskite phase (referred to as cubic). The broad
110
 1996 E.R Camarqo et ai. I Journal of Soiid State Chemistry 177 (2004) 1994-2001

*
••. L.U.t •.••. .t.J
100 200 300 400 500 600 700 11 •• 11 ",,', 11.11" 11" •••• """ 'I!, li .1, ", ••••• Ir li li "",,"'U .H

o- (a) «o'e
@. Heating
--- 490'C

amorphOU~
Cubic

DiJ
~.I----~-
470 475 480
Cublc

(b)
Cooling
- Telragonal

100 200 300 400 500 600 700

Temperature ('C)

Fig. 1. DSC curves of the amorphous precipitate obtained during (a)

(c)
the heating step, and (b) during the cooling processo The inset in (b)
shows the endothermic peak of the phase transition from the cubic PT (d]
to the tetra.gonal PT. Observe that there is not any other phase ~1.miiimllnil~ii.milnlin'li~ii~iiwnim
••mIiPliniiim"mii~llnih~i,mlltth~iiimli~ii"
•• n'iimiimhn
•• tt
••mi"~iinii=iii,"

transition at temperatures below 475°C. 900 800 700 600 500 400 300 200 100

Raman Shift ( CM-1)

exothermic peak between 200°C and 470°C with
maximum at 440°C can be assigned to the crystallization
process from the amorphous precipitate. The small
endothermic peak at 476°C in the cooling curve (Fig. Ib)
can be assigned to the transition from the cubic to the
tetragonal phase.
lt is well known that Raman spectroscopy is a suitable
structural analytical technique for identifi.cation of
crystalline phases in nanocrystals when XRD has not
enough resolution [17]. Fig. 2 shows Raman spectra
collected at room temperature of the amorphous
precipitate and of three heat-treated samples. The
spectrum (a) of the powder calcined at 700°C for 1h
(referred to as PT700) shows the typical spectra of the
tetragonal phase [2]. Spectra (b) and (c), obtained from
the powders ca1cined at 450°C for 40min (pT450-40)
and for 15min (pT450-15) respectively, show comple-
tely different spectra that cannot be assigned neither as
tetragonal nor as cubic PT phases, Yao et al, [18]
reported the formation of a metastable stoichiometric
cubíc pyrochlore (Pb2Th06) phase at 10w temperature
(identified only by XRD) prior to the crystallization of
the tetragonal PT phase during the synthesis of a glass-
ceramic materiais by a sol-gel rqute. Moreover, they
reported the presence of a sharp exothermic peak at
450°C in some DSC curves, assigned as the crystal-
lization of the tetragonal phase (essentially the same
temperature observed in the heating curve of the Fig,
Ia). The process described by Yao et al, [1&]follows
Eq. (1):
(1)
Fig. 2. FT-Raman spectra or (a) PT700, (b) PT450-40, (c) PT 450--15,
and (d) of the amorphous precursor.
Although similar spectra to curves (b) and (c) of Fig. 2
were already published (16,19,20], there is not a
consensus regarding the structure of the lead titanate
formed at low temperature. However, the general shape
of the spectrum of Fig. 2b is similar to that observed for
the pyrochlore 1eadpyroniobate (pb2Nb207), with main
peaks at 145, 231, 285, 334, and 741 em -I, [20-22]
indicating that some correspondence between these two
structures (pb2Nb207 and the stoichiometric Pb2 Th06)
can be considered. [20] The spectrum of Fig. 2d shows
the amorphous nature of the precipitate. It is evident
that the presence of three different Raman spectra in
Fig. 2 represents three different structural organization
of the material, which can be attributed to lhe crystal-
lization of the stoichiometric cubic pyrochlore structure
(Figs. 2b and c) from the amorphous precipitate (Fig.
2d), followed by a phase transition to the tetragonal PT
phase (Fig. 2a) at approx. 440°C. This sequence agrees
with the previously reported results of the synthesis of
lead titanate by wet-cbemical methods [15,18-20].
The XRD pattem (a) shown in Fig. 3 was obtained
from the powder PT700 (for the sake of clarity, the
highest peak (101)was eut at 3% of its height). All of the
diffracted peaks were identified as the single tetragonal
PT phase, in agreement with the Raman spectrum af
Fig. 2a and DSC analysis in Fig. 1. In spite the fact that
tetragonal PT could be identified in pattern (b) of the
powder calcined at 500°C for 30min (PT500-30), it is

) 111
 E:R Catuarqo ('I <11. I Journal o/ Solid State Chemistry 177 (2004) 1994-2001 1997

t.J.J,.,.lJ,t...J""uJ I, I .1. ".1.«

---
lI' 1" III li' "'" 11., 1"'" ••

(o) :5 Cl ~ :: Cl ::
~ Cl
N N -;-
(a)
M ~
<=
o.
o
N
Cl
o. R (b)
@
..a (c)
(11
(d)
"O
4)
.!:>!
(b) (ij
E
z
o

4950 4960 4970 4980 4990 5000 5010 5020 5030 5040 5050
Energy(eV)
(c) /\
Fig. 4. Titanium K-edge XANES spectra of (a) PTIOO, (b) PT45D-
40min, (e) PT450-15, and (d) amorphous precipitate sample.

------~ ~---------~-----
_:.d)__ ~.~_. ~~~~
.11 ""li' 1"'" 'li 'I' ali •• 1
10 20 30
29 (deg)/CuKcr:
Fig.3. XRD pattems of(a) PT700, (b) PT500-30. (e) PT 450-15, and
(d) of the amorphous precursor.
possible to observe small and broad peaks at 29° and 35°
(marked with full circles) indicating that a secondary
phase is present. Specifically, the peak at approximately
29° reproduces the same broad peak observed in the
XRD patterns (c) and (d) regarding the powder PT450-
15 and the amorphous precipitate, respectively. If the
stoichiometric cubic pyroehlore Ph:!Ti206 phase is
considered as the secondary phase in Fig. 3b, one can
assign, based on the Raman evidences of Fig, 2, that
these two peaks are the (222) (100%, centered at 29.55°)
and the (400) (55%, at 34.33°) diffraction planes ofthe
Pb2 Ti206 pyrochlore phase (PDF card number 26-
0142). Therefore, the precipitate seems to crystallize first
in the stoichiometric cubic pyrochlore Pb2 Ti206 phase,
instead of the thermodynamically expected tetragonal
PT.
According to the literature, XANES spectra of
transition metal oxides are stroagly dependent on the
symmetry of the metal site, and titanium (and other
transition metals as well) possess unoccupied electron
states of d character, These states are inaccessible to the
dipole excitation of a s electron, and the displacement of
the transition metal from its centrosymmetric site drives
the mixing of p character of the transition metal atom.
Therefore, the presence of large peaks in a XANES
.. a'~'~~I'~,~~~~ spectrum (e.g., 4965-4975eV in the titanium case)
before the main rise in the spectrum is the signature of
"I li. ali •••• " •••• " "
a large displacement of the transition atom from its
50 60
centrosymmetric site [23J.
The XANES spectrum obtained at Ti K-edge for
PT700 shown in Fig. 4a is characteristic of the PT
tetragonal phase [23J, in agreement with the XRD and
Raman results previously discussed (Fig. :2and 3). The
XANES spectra of the samples PT450-40, PT450-15,
and ofthe amorphous precipitate sample (curves (b)-(d)
in Fig, 4) exhibit identical edge structure, indicating the
same local symmetry for titanium atoms in these three
samples. However, these three spectra present a different
pre- and post-edge structure when compared with the
PT700 tetragonal XANES spectrum (Fig. 4a). The
transitions situated at the pre-edge region are less
intense, principal1y ín the amorphous precipitate
sample (Fig. 4d). The decreasing in the intensity of
the transition labeled as A in curves (a)-(d) can be
attributed to a decreasing in the distortion of the Ti06
octahedra. According to Veddrinskií [24], the electroníc
transition assigned as B in the XANES spectra does not
depend strongly on small displacements of the atoms in
the Ti06 octahedra, but it is mainly affected by
structural changes in the vicinity of the absorbing
titanium atom.
lu the post-edge region of the XANES spectra of Fig.
4, it is observed the presence of large transitions in
spectra (b}--(d) in comparison with spectrum (a) of PT
tetragonal, where sharp transitions are easily observed.
The post-edge transitions in the XANES spectra are
mainly due to contributions from multiple scattering
effects caused by atoms located in the vicinity of the X-
ray absorbing atom, and are used to probe the medium
range order around the absorber atom. The broadening
in the transitions observed in the post-edge region of

112
1998 E.R Camarqo et ai. I Journal ai Solid
spectra (b)-( d) can be explained by an existence of
relative disorder in the medium range structure around
the Ti absorbing atom. If these three XANES spectra
are compared to results published in the literature, one
can see that the pre-edge features are characteristic of
pentacoordinated Ti atoms, however, a rnixture of
fourfold and sixfold coordination cannot be excluded
[25,26].ln fact, the XANES spectra (b)-(d) of Fig. 4 are
very similar to the XANES spectra of amorphous
precursors of Pb(Zr,Ti)03 ceramics studied by Malic
et al. [27]. According to their work, Ti atoms in samples
presenting different Zr to Ti ratio are coordinated by
five oxygen atoms. Two local symmetry are possible for
the Ti pentacoordinating site, the centrosymmetric D3h
trigonal bipyramid, and the C4L• pyramid with the Ti
atam positioned at long of the C4 axis. [28] Since the
centrosymmetric symmetry is not compatible with the
XANES spectra, the most probable structure is the C4L••
lndeed, to say that titanium atoms are pentacoordinated
means, at a first approximation, that the sixth oxygen of
the octahedra is located sufficiently far from the
titanium atom, which allow us to consider it not
coordinated to the titanium, resulting in a highly
distorted octahedra.
Fig. 5 presents the EXAFS spectra of PT700, PT450-
40, PT450-15 samples, as well the EXAFS spectrum of
the amorphous precipitate powder, whilst Fig, 6
illustrates the Fourier transform moduli (FT) of these
EXAFS spectra. A qualitative analysis of the EXAFS
spectra and of their respective FT curves leads to the
same conclusion obtained previously from the analysis
of the XANES spectra, i.e., the EXAFS and the FT
curves of the PT450-40, PT450-15 and of the amor-
phous precipitate look very similar, but they differ
essentially from those of the PT700. This difference is
better noted at high values of k (Fig. 5), where the
presence of the EXAFS oscillations are not observed in
O.5~--------------------------------,
0.0
-0.5
2-
~
~ -1.0
-1.5
-2.0;-"'-'r-r--'-'-~"--'-'--r-'-r'""~-'--r-...-,,.......-I
2 3 4 5 6 7 8 9 10 11 12
K (A-I)
Fig. 5. EXAFS spectra of (a) PT700, (b) PT450-40min, (c) PT45O--15,
and (d) amorphous precipitate sample.
State Chemistry 177 (2004) 1994-2001
6~------------------------------.
--(a) PT700
o (b) PT450-40
O (c) PT450-15
• (d) precipitate
2 3 4 5 6 7 8
R (Â)
Fig. 6. EXAFS spectra (k-weighted) and corresponding Fourier
transforms moduJi for: (a) PT700, (b) PT450-40min, (c) PT450-15,
and (d) of the amorphous precipitate sample.
curves (b)-(d), indicating the existence of a high degree
of disorder in these samples (confirmed by the absence
of structural peaks for distances longer than 3.0 Á in
theirrespective FTcurves (b)-(d) in Fig. 6). lt should be
stressed that other differences are observed between the
FT curve ofthe PT700 and the other FT curves in Fig. 6.
The first main difference is the shift toward higher
distance ofthe first Fourier transfonn peak of curve (a)
of PT700 when compared with curves (b)-(d), which is
related to the first shell around titanium atoms (Ti-O
bonds). An increase in the Ti-O mean bond length is
thus expected for the PT tetragonal (PT700) sample. A
second significant difference is the increasing in the
amplitude of this FT peak of curve (a).
As expected, alI of the results (OSC, XRD, Raman
and XAS) regarding the we1l-crystallizedPT700 sample
confirm that the PT tetragonal phase is formed,
However, the most interesting result is obtained when
the XAS spectra (EXAFS and XANES) of samples
PT450-40 and PT450-15 are compared to their respec-
tive XRD and Raman curves and to PT700 results. It is
clearly demonstrated by DSC, Raman, XRO and XAS
that the crystalline phase obtained after the heat
treatment up to 450°C cannot be assigned as the PT
tetragonal, but the probable structure seems to be the
stoichiometric cubic pyrochlore Pb2 Ti206 with distorted
octahedra Ti sites. Comparing the XAS spectra of
PT450-40 and PT450-15 samples to the spectrum of the
amorphous precursor, one can infer that the local
structure around the titanium is essentially the same.
Therefore, it is reasonable to conclude that the local
order of the amorphous precipitate drives the crystal-
lization into the most favorable structure, that is the
Pb2Ti206• In other words, the structure that posses a
similar local organization to the amorphous precursor ís
preferentialIy crystallized.
113
 CUllldfylJ r.1 111I Journal of Solid State Chemistry 177 (2004) 1994-2001 1999

(i'/Ir/ll'.l'Is 0/ EXAFS spectra
II quantitative result from the EXAFS, the
lIpoCltUm of the crystaUized PT tetragonal compouud
(PT700) was first1y fitted, To do that, the coutributiou
of the first coordination sheIl of the FT (presented in
Fig. 6) was extracted by a back Fourier transform in R
space in a range between 0.8 and 2.4Â, and then fitted
using theoretical phase and amplitude functions. To
model the Ti--O pair, theoretical amplitude and phase
fuuctions were calculated using the FEFF 8.2 program
[2.9J.lu ali of the fits, the number of free parameters was
kept smaller than the number of independent points
(defined as Niud = 211R ãkln where I1R is the width of the
R-space fílter windows, and I1k is the actual interval of
the fit in the K space) [14]. The Ti--O oxygen first shell
for the crystalline tetragonal PbTi03 compound was
fitted using the three shells model according to the XRD
data-Bü] (e.g., the irregular octahedral Ti site with three
different Ti--O bonds length). During the fitting
procedure, the coordination number (IV) was fíxed and
the interatomic distance (R) and the Debye-Waller
factor (O") were allowed to vary, A single value of O" was
applied for all of the three types of Ti-O bound
considered, reducing the number of free parameters
during the fitting. Table I presents the fitting results and
Fig. 7 shows the comparison between the fíltered
EXAFS spectra (11; = rr/2I1k). Since the useful !'.k
range is about 9.0 A (from 2.5 to ] 1.5Â, see Fig. 7
and 8) !'.r-values under 0.17 Â cannot be unambiguously
resolved. According to the work of Malic et al. [26]
titanium atoms in an amorphous lead titanium gel,
which was obtained from the hydrolysis of all alkoxide-
derived heterometallic complexes, are coordinated by 5
oxygeu atoms distributed in two sets of distances
(around 1.79 and 1.94Â). Despite the fact that the
difference between these two distances is lower than the
resolution determined from the useful àk: range, these
two shells model as well the one shell model, were tested
for the EXAFS spectra ofPT450-40, PT450-]5, and for
the spectrum of the amorphous precipita te. The results
obtained for the fittings of these two models were lhe
same, indicating that the best fitting was achieved when
it was assumed that the first shell is formed by 5 oxygens
located at 1.94Â around the titaniurn atoms (see
Table 1). However, this result should be carefully under-
stood. As already discussed, the calculated distance
between the oxygens and the titaniurn atom is affected
0.4
--Rtting
0.3
0.2
• Experiment

experimental EXAFS spectra of PTIOOand the theore-

tical curve for the crystalline tetragonal PbTi03_ As can 0.1
be seen in this table, the fitting of the EXAFS spectra
agrees with the data found by Glazer PO} using the -
2'
N
.::L
0.0

XRD technique, what validates the procedure used. -0.1

Before doing the analysis of the other three samples
(PT450-40, PT450-15, and amorphous precipitate) the -0.2
distance resolution (I1r) that can be achíeved during the
-0.3
fitting of these EXAFS spectra was detennined as a 3 4 5 6 7 8 9 10 11
function of the useful I1k range of the filtered EXAFS
K (A-I)
spectra. The resolutíon of the features in EXAFS is
limited by the finite range of the data (31], and is Fig. 7. Fitting and back Fourier filtered signal of the Ti--G first shell
determined by the useful I1Ie range of the filtered for the PbTi03 tetragonal (PT700) sample using a three-shell model,

Table 1
Fitting results obtained by FEFF 8.2 suit for the tetragonal perovskite sample PTIOO, and for PT450-40, PT450--15 and for the amorphous
precipitate

Sample NI RI (Á) 0"1 N2 R,. (Á) 0"2 N3 R3 (Á) 0"3

Tetragonal PbTi03 by XRD [30J 4.0 1.98 1.0 1.76 1.0 2.40
PT700 4.0 1.99 0.07 1.0 1.79 0.07 1.0 2.41 0.07
(±O.OI) (±0.01) (±0.03) (±0.01) (±0.O3) (±0.01)
PT 450-40 5.0 •• 1.93 0.102
(±0.01) (±0.OO5)
PT 450-15 5.0 1. 93 0.1J5
(±0.01) (±o.o05)
Amorphous 5.0 1.94 0.125
Precipitate (±0.01) (±0.OO5)

The results obtaíned by Glazer et al. [30J are also included. In the table, "N' is related to the number of coordínating oxygens around the Ti atom,
"R" is the distance between the oxygen and the Ti atom, and "O""is the Debye-W.a11erfactor,

114
2000 E.R Camarqo et ul. I Journal of Soiid Stute Chemistry 177 (2004) 1994-2001
0.4,.-------------------,
--Fitting
0.3
• Experiment
0.2
g 0.1
N
~ 0.0
-0.1
-0.2
3 4 5 9 10
Fig. 8. Fitting and back Fourier filtered signal of lhe Ti-O first shell
for sample PT45ú-40 using lhe one-shell fitting mode!.
by the useful Me range used for the fitting. Therefore,
oxygens located at lightly different distances from the
titanium are indistinguishable, resulting that the calcu-
lated distance can be an average of distances that differ
less than 0.17 Â from each other. The fitting results are
presented in Table 1, and Fig, 8 shows the comparison
between the filtered experimental EXAFS spectrum of
the PT450-40 and its theoretical curve, when the one
shell-fitting model was used.
3.2. General discussion and considerations
When the fi.tting rosults of the PT450-40 (Fig. fi and
Tablc 1) are interpreted considering the XANES
analysis regarding the titanium displacement from its
centrosymmetric position, and the fact that Ti atoms are
located inside of octahedral sitesin the stoichiometric
cubic pyrochlore Pb2Th06, one can consider that the
five coordinating oxygens around the absorbing Ti atom
are organized as in the point group C4 o- In this spatial
configuration, the oxygens are positioned at the five
vertices of a square-based pyramid and the titanium
atom is located in some point in the C4-axiS.A distorted
octahedra can be visualized with a sixth oxygen atom
located at a distance higher than 2.4Â (therefore,
outside of the first shell considered in the fitting
procedure). This assumption of distorted octahedra
agrees with the XRD, Raman and XAS results.
Moreover, the results indicate that the spatial
arrangement of the oxygen anã titanium elements in
the C40 syrnmetry point group is formed immediately
during the precipitation of the amorphous material, that
is during the oxy-reduction reaction between Ti-peroxo
complexes and the Pb2+ ions. The similarity of the
EXAFS and XANES spectra between the amorphous
precipitate, PT400-15 and PT400-40 indicates that these
samples present similar titanium local structures. This
observation can explain the reason why the pyrochlore
is formed prior to tetragonal perovskite, and why this
stoichiometric pyrochlore has been systematically ob-
served during the synthesis of lead titanate by means of
several wet-chemieal routes. These results indicate that it
is easier for the amorphous material to erystallize in the
most similar strueture (Pb2 Th06), which is kinetically
favored, instead of a slightly different strueture (in this
case, the tetragonal PT perovskite), although thermo-
dynamically favored.
Evidently, this metastab1e pyrochlore structure is not
obtained by loss of lead, as usually reported, since
tetragonal PT with mole ratio of (Ti.Pb = 1:1) is
crystallized further, at higher temperature. If the
11
structure is the pyrochlore-lead deficient type, it should
be stable even at heat treatment at high temperatures,
and thus easily identified by means of XRD or Raman.
However, this lead-deficient phase was not identified
after the phase transition to lhe tetragonal PT. There-
fore, it is more plausible that this pyrochlore phase is
stabilized by the well-known effect on nanosized
particIes [32], what turns the identification of this
intermediate metastable pyrochlore phase quite diffi-
culty to be characterized. It should be observed that
only the phase transition from the stoichiometrie
pyrochlore (Pb2 Th06) to the tetragonal PT phase could
be identified, without any signal of the reverse phase
transition. On the other hand, the transition between
tetragonal PT and cubic PT phases was easily identified.
It means that is possible to bypass this intermediate
phase by heat treatment at temperature higher than
450°C, as indicated by the DSC curves. A general
equation for the lead titanate system synthesized by wet-
chemical routes can be proposed as
2[PbTi](amoq>hous)
-+ P~ Th06(pyrochlore)
-+ 2PbTi03(tretragonal)
+- -+ 2PbTi03(cUbic), (2)
where the first term at left side represents the amorphous
precursor, with mole ratio of (pb:Ti = 1:1), which ean be
obtained by several wet-chemical methods (e.g. sol-gel,
Pechini, OPM, ete). The unidirectional arrows means
that the phase transition can occur only in the indicated
direction and consequently, the bi-directional arrow
means that the phase transition can occurs in both
directions.
4. Conclusions
Despite .the fact that single phase tetragonal lead
titanate (PbTi03) perovskite can be obtained by the
"oxidant peroxo method-OPM" at relatively low
temperature, it was observed that an intermediate cubic
and stoichiometric pyrochlore (Pbj Ti206) phase is
formed prior to the expected tetragonal PbTi03, as
supported by the XRD and Raman analysis. Moreover,
115
 }:Jt CtIf/tllr!lO ct ai.
powders ca1cinedat temperatures up
that the coordination sites around the
iUtlt atem are essentially the same of the amorphous
precipita te, but different from that observed in the
tetragonal PbTi03. This result indicates that the crystal-
lization process from the amorphous precursor favors
structures that have their local structure around the
titanium atom as that originally found in the amorphous
precursor. Considering ali of the previously reported
studies of the synthesis of PbTi03 by any wet-chemical
routes, it seems that the formation of the pyrochlore
Pb2 Ti206 is a general path to obtain pure lead titanate.
AIl of the results lead 10 same conclusion, which
tetragonal PbTi03 pure tetragonal PT is obtained after
the phase transition from the metastable Pb2 Ti206,
however the reverse transition from the tetragonaJ PT to
the pyrochlore Pb2 Ti206 does not occur.
Acknowledgments
Acknowledges to LNLS-NationaJ Synchrotron Light
Laboratory-e-Brazil, to MCf-Ministry of Science and
Technology of Brazil, and to CNPq. This work was
supported by FAPESP.
References
[1) A.P. Alivisatos, J. Phys. Chem. 100 (1996) 1322tH3139.
[2) E.R. Camargo, M. Kakihana, Chem. Matec. 13 (2001) 1181-1184.
[3) E.R. Camargo, J. Frantti, M. Kakihana, J. Mater. Chem. 11
(2001) 1875-1879.
[4) E.R. Camargo, M. Popa, J. Frantti, M. Kakíhana, Chem. Mater,
13 (2001) 3943-3948.
[5) E.R. Camargo, M. Kakihana, J. Am. Cecam. Soe. 85 (2002)
2107-2109.
[6) C.D.E. Lakeman, DA. Payne, Mater. Chem. Phys. 38 (1994)
305-324.
[7) L.L. Heneh, J.K. West, Chem. Rev. 90 (1990) 33-72.
[8) E.R. Camargo, M. Kakihana, E.R. Longo, E.R. Leite, I. Alloys
Compounds 314 (2001) 140-146.
[9] M. Kakíhana, J. Sol-<3e1 Sei. TechnoI. 6 (1996) 7-55.
O(
I Journal of Solid State Chemistry 177 (2004) 1994-2001 2001
[10] E.R. Camargo, E. Longo, E.R. Leite, J. Sol-Gel Sei. Technol. 17
(2000) II 1-12L
[li] I.I. Rehr, C.A Albers, Rev, Mod. Phys. 72 (2000) 621-654.
[12] H.CN. Tolentino, A.Y. Ramos, M.CM. Alves, R.A. Barrea, E.
Tamura, I.C Cezar, N.I. Watanabe, J. Synehrotron Rad. 8 (2001)
1040-1046.
[13] A. Miehalowicz, J. Phys. IV 7 (1997) 235-236.
[14] S.S. Hasnain, Report 00 the International Workshops 011
Standarts and Criteria in XAS, in: X-ray Absorption Fine
Strueture: Proceedings of the VI International Conference
on X-cay Absocption Fine Structures, Ellis Horwood, New Y ork,
1991.
[15] E.R. Leite, E.C Paris, E. Longo, J.A. Varela, J. Am. Ceram. Soe.
83 (2000) 1539-154L
[16] D. Fu, S. Hisao, K-S. Suzuki, Phys. Rev. B 62 (2000) 3125-3129.
[17] M. Kakihana, M. Yashima, M. Yoshimura, L Borjesson, M.
KalI, Trends Appl. Spectrose. I (1993) 261-311.
[18] K. Yao, W. Zhu, X. Yao, L. Zhang, Mater. Sei. Eng. B 41 (1996)
322-328.
[191 E.R. Leite, E.C Paris, E. Longo, F. Lanciotti, E.CM. Campos,
P.S. Pizani, V. Mastelaro, CA. Paskocimas, 1.A. VareJa, J. Am.
Ceram. Soe, 85 (2002) 2166-2170.
[20] D. Bersani, P.P. Lottici, A. Montenero, S. Pigoni, G. Gnappi,
I. Non-Cryst. Solids 193 (1995) 490-493.
[21] A. Jayaraman, G.A. Kourouklis, A.S. Cooper, G.P. Espinosa,
J. Phys. Chem, 94 (1990) 1091-1094.
[22] N. Wakiya, J. Shiihara, K. Shinozaki, N. Mizutani, J. Sol. State
Chem. 143 (1999) 344-348.
[23] B. Ravel, E.A. Stem, R.I. Vedriuskü, V. Kraizman, Ferroelectrics
206 (1998) 407-430.
[24] R.V. Vedrinskii, V.L. Kraizman, AA. Novakovieh, P.V. Deme-
khin, S.V. Urazhdin, J. Phys.: Condens. Matter 10 (1998)
9561-9580.
[25] Z. Liu, R.J. Davis, J. Chem. Phys. 98 (1994) 1253-1261.
[26) B. Malie, I. Arcon, M. Kosec, A. Kodre, I. Mater. Res. I2 (1997)
2602-2611.
[27) B. Malie, I. Arcon, A. Kodre, M. Kosec, J. Sol-Gel Sei. Technol.
16 (1999) 135-141.
[28] F.A. Cotton, ChemicaI Applications of Group Theory, 2nd
Edition, WiJey-Interscience, New York, 1971.
[29} A.L. Ankudinov, B. Ravel, U. Rehr, S.D. Conradson, Phys. Rev.
B 58 (1998) 7565-7576.
[30] A.M. Glazer, S.A. Mabud, Acta Ccystallogr. B34 (1978)
1065-1078.
[31] D.e. Koningsbergec, R. Prins, X-ray Absorption, PrincipIes,
Applications, Techniques of EXAFS, SEXAFS and XANES,
Wlley, NewYork, 1988.
[32} C.C. Chen, A.B. Herhold, CS. Johnson, A.P. Alivisatos, Seienee
276 (1997) 398-401.
116
 Abstract

XAS and XRD Structural
Characterization of
Lanthanum Modified
PbTi03 Ceramic
Materials
Person P. Neves, t Antonio C. Doriguetto, *.t
Valmor R. Mastelaro, t Luiz P. Lopes, Yvonne P.
Mascarenhas, t Alain Michalowicz, t José A.
Eirasf
Instituto de Física de São Carlos, Universidade
de São Paulo, CP. 369 - São Carlos-SP, Brazil,
Groupe de Physique de Milieux Denses,
Université Paris xn -Val de Mame, 94010,
Creteil Cedex, France, and Departamento de
Física, Universidade Federal de São Carlos, São
Carlos - SP, Brazil
* dorigue@ifsc.usp.br
RECEIVED DATE (to be automatically
inserted after your manuscript is accepted if
required according to the journal that you
are submitting your papel"to)
For the first time combined X-ray absorption
spectroscopy (XAS) and X-ray diffraction
(XRD) experiments were carried out to probe
the short and long-range order in lanthanum
modified PbTi03 cerarnic materials (Pb1-
°
xLaxTi03) for x ranging from to 30 at. % of
La. XRD results show that a tetragonal structure
fitted well to ali samples, except to the one with
30 at. % of La for which XRD results
highlighted a cubic unit cell. XRD refinements
also confirmed the existence of only A-site
vacancies. X-ray Absorption Near Edge
Spectroscopy (XANES) measurements indicate
that the local structure around Ti atoms has a
different compositional dependence from the one
obtained from XRD, mainly for the sample with
30 at. % ofLa According to XANES data, for
the sampIe with 30 at. % ofLa, a local distortion
around Ti atoms persist. The analysis of
Extended X-Ray Absorption Fine Structure
(EXAFS) spectra shows a decrease in the local
disorder around Ti atoms as the content of La
increases. Structural details, in terms of short
and long-range order, are presented and the
eorrelation between the XRD and
XANESIEXAFS data is discussed.
1. Introduction

* To whom correspondence should be Lanthanum-doped lead titanate cerarnics (Pb..

addressed: Departamento de Física e xLaxTi03 or PLT) have been studied in detail
Informática, Instituto de Física de São Carlos, due to their interesting physical properties.' The
Universidade de São Paulo, c.r. 369, CEP isomorphic substitution of lead by lanthanum
13560-970, São Carlos - SP, Brazil, TeI: 55-16- atoms induees some interesting changes in the
2739887, Fax: 55-16-2739881, E-mail: physieal properties of PbTi03 material. When
dorigueCâljfSC.lW. br, the lanthanum eontent is higher than 25 at. %, a
diffuse eharacter of the ferroeleetrie-paraelectric
t Instituto de Física de São Carlos, phase transition (DPT) is observed.' The
rnodifieation in the phase transition character is
t Université Paris Xll,
c
followed by a linear deerease in the Curie
§ Universidade Federal de São CarIos. temperature (Te) to room temperature for x~30
at. %.1.2.3 It has been suggested that the
lanthanum indueed modification coneeming
Aceito para publicação no joumal of Physics PbTi03 results in structural changes that ean be
and Chemistry B (2004) direetly related to the nature of the phase
transition.4•5,6,7
The PbTi03 erystal strueture has been
'>:JChaustively studied. 8,9, 10 Its room temperature

117
srrucrure is tetragonal, P4mm (a = 3.902(3) and bands that appeared in temperatures above the
(: = 4.156(3) Â), with four atoms in the phase transition in lead titanate ceramics
asymmetric unit: Pb at (0,0,0); Ti at (Y:!, Y:!, containing 30 at. % of La. They explain the
0.5377(4)); 0(1) at (Yí, Yí, 0.1118(3)); and 0(2) observation of these Raman-forbidden bands at
at (O, Y:!, 0.6174(3)).11 Assuming the perovskite temperatures substantially higher than the phase
structure notation, the sites occupied by Pb and transition temperature by the existence of short-
Ti, are labeled asA and B, respectively. ranged clusters with tetragonal 4mm (C4V)
In La3T-modified PbTi03 materials, Ll+ symmetry distributed over the globally cubic
replaces Pb2+ rather than Ti4+ in Pb-based m 3 m (Os) symmetry matrix.
perovskites.Í/ In order to keep the neutrality of In spite of the scientific-technological
the charge, since that replacement is aliovalente, impo rtanc e of this system, conclusive
site vacancies are created.12 investigations are scarce considering its
Considering that the PbO-La20r Ti02 solid crystalline structure, mainly regarding phase
solution is lying within a temary diagram formed transition induced by a cation replacement. In
by PbTi03, and the two meta-stable defect this paper, we present a short and long-range
cornpounds La2J30ll3Ti03 and La[Ti3/401l4]03, structural study of the Pbl-xLaxTi03 ceramic
these compositions have been described by the system. The XAS technique was used to probe
general defect equation: 13 the local structure around Ti atoms whereas the
3(1-av) L 3y XRD technique was used to obtain information
Pb . a ---=---
3 + y(1.5 - a) 3 + y(1.5 - a) conceming the crystallographic structure of the
samples, as well as the distribution of vacancies.
y(2a -15) x [Ti 3 To the best of our knowledge, it is the first time
3 + y(1.5 - a) 3 + y(1.5 - a) that the short and long-range order structure of
--=-y..>...( 1_.5_-_a...::..)_]
the PLT ceramic system is simultaneo usly
03 >
studied using XAS and XRD techniques.
3 + y(1.5-a)
where O is a site vacancy and a. is the Pb 2. Experimental Section
elimination factor. The a. factor can adopt values
ranging from 1.5 (A-site vacancies only) to 0.75 Seven powdered PL T samples, the nominal
(E-site vacancies only). There is a consensus in composition Pb1-xLaxTi03 with x = 0.0; 5; 10;
the literature that lanthanum exclusively 15; 20; 25; 30 ato % (abbreviated as PLTx, from
occuples . t he A site.
. 1415 . PLTOO to PLT30), were prepared by the
Raman scattering and X-ray diffraction conventional mixed oxide method. The oxides,
techniques have been respectively used to probe PbO, La203, and Ti02, weighed according to
the short and long-range order in the Pb1- stoichiometry, were mixed by ball milling in
'O .
x L ~x T 1 3 ceramic system,
141516T
>. avares et al ., 16 distilled water for 10 hours. The slurry was dried
using the Raman scattering technique probed the and calcined in a covered alurnina crucible at
short-range structure for x ranging from 0.0 to 1100 0
e for 3 hours. Ceramic bodies were then
30 at. %. According to this work, in highly formed by isostatic pressure, and fired at
doped sampIes (x ~ 27 ato %) and for -1 175°C for 3 hours. The densities of alI fired
temperatures above the phase transition, X-ray ceramics were found to be around 95% of the
diffraction measurements indicate a cubic theoretical one defined by the Archimedes
structure. However, Raman scattering method. Surface analyses, by scanning electronic
measurements show a residual ~ort-range microscopy (SEM), show that the grain-size of
structural disorder in this cubic phase. These all fired samples has about 1.5J.!m. The PLT05
authors suggested that the existence of this and PLT30 sample compositions were examined
short-range structural disorder could be point-to-point using an electron microscope
correlated to the relaxor behavior in highly equipped with an electron microprobe
doped samples. spectrometer (Wavelength Dispersive Spectrum
Recently, Tae-Yong Kim et al.14 also (wnS) accessory).
observed the presence of Raman-forbidden

118
 Disk-like ceramic bodies were polished to
ensure parallel surfaces, and silver eIectrodes
were then deposited onto surfaces by sputtering.
The compIex dieIectric permittivity was
measured from -180 to 500°C using an
HP4192A impedance anaIyzer at 1 kHz.
The X-ray powder diffraction pattems
were measured at room temperature on a
Rigaku Denki powder diffractometer with
geometry 6-26; a rotating ano de X-ray source
(Cu Ka radiation, À = 1.542 A), and a
scintillation detecto r. The measurements were
carried out using a current of 100 mA and
tension of 50 kY. Each pattem consists of 5,851
steps ranging from 3° to 120026. The data were
colIected with a step size of 0.02°. The count
time was 5 sec per step. The refinement of the
structure was carried out using the Rietveld
method.17 The program used in the refinements
was the General Structure Analysis System
(GSAS) package.Í''
The least-square refinements of all samples
were started considering the PbTi03 structure
published by Nelmes & Kuhs in the space group
P4mm.ll Since P4mm is a quiral space group,
the unit cell origin was de:6ned by fixing the Pb
atom at O, O, O position. In this way, all other
atorns in the unit cell are described by only three
positional parameters, oZn, 020(1),and o~), the
shift along the c axis from the ideally cubic
perovskite structure: 1/2, 1/2, 1/2+OZTi;
1/2, 1/2, 0+020(1); and O, 1/2, 1/2+0.7..0(2)for Ti,
0(1), and 0(2) atorns, respectively. However, in
each case, the refinement was started by setting
OZo(I) and oZ0(2) equal to zero, whereas oZn was
given a small positive value to define the positive
direction of the c axis. Scattering factors of
ionized atomswere used in alI refinements. The
02- scattering factors were taken from Rez et
al.19 Refinements of the anisotropic temperature .
factors were tried, but the trial results were not
physically significant. Therefore, the temperature
factors were kept isotropiê (U) during the
refinements.
The titaniurn K-edge X-Ray absorption
spectra were collected at the LNLS (National
Synchrotron Light Laboratory) facility using the
D04B-XASl beam line.20 The LNLS storage
ring was operated at 1.36 GeV and 100-160
mA. The sample pellets, obtained afier sintering,
were grounded for XAS measurements. XAS
data were collected at the Ti K-edge (4966 eV)
in transmission mode at room temperature using
a Si(l11) channeI-cut mono chro mato r.
Ionization chambers were used to detect the
incident and the transmitted flux. XANES
spectra at the Ti K-edge were recorded for each
sample between 4,910 and 5,200 eV using
energy steps of 0.5 eY. Due to the presence of
the Ianthanum Lm edge at 5,483 eV, EXAFS
measurements at the Ti K-edge (using steps of 2
eV) were carried out for each sampIe between
4,840 and 5,400 eY. Four acquisitions were
recorded for each sample to improve the signal
to noise ratio. To provide good energy
reproducibility during the XANES data
collection, the energy calibration of the
monochromator was checked during the
collection of the sample data using a Ti metal
foil.
The EXAFS analysis was carried out by
using the program sei written by Michalowicz "
according to the recommended procedures
described by the Intemational Workshop on
Standards and Criteria in XAFS.22 After atornic
absorption removal and normalization, the
K3X(k) weighed EXAFS signal was Fourier
transformed to R distance space in the 2.0-10 A-
I K range. Each spectra was Fourier transformed
using a Kaiser apodisation window with .=2.5.
A qualitative interpretation of XANES spectra
obtained at the Ti K-edge was performed using
the software package developed by Michalowicz
and Noinville.23 For comparison purposes
among different samples, all spectra were
background removed and normalized using as
unity the fust EXAFS oscillation.
3- Results and Discussion
a. Relative Dielectric Permittivity. Figure 1
shows the temperature dependence of the
relative dielectric permittivity of PLTx samples,
As the content of La increases, we observed an
increase in the dielectric diffuseness foIlowed by
a gradual decrease in the temperature of
maximurn dielectric perrnittivity (Te). As can be
seen in Figure 1, depending on the La content,
the ferroelectric behavior can be normal or
relaxo r: the relative dielectric permittivity curve
of the PLT20 sample (Te RI 80°C) presents a
narrow peak representative of a normal
119
ferroelectric material, whereas the PLT30 one
(Te ~ -50°C) presents a broad peak
representative of a relaxo r ferroelectric material.
ln the first case, for temperatures higher than Te,
a centrosymmetric cubic structure (Pm3 m) is
expected since the material undergoes a phase
transition to a paraelectric material. In the
second case, for temperatures higher than Te, a
residual short-range structural disorder has been
found'" not to be compatible with the
centrosymmetric cubic structure usually
observed for the paraelectric phase.
b. X-ray Diffraction. The tetragonal
structural model fitted well to alI XRD data,
except for the PLT30. Conceming the PLT30,
the a and c celI parameters converged to very
similar values and the shift parameters are ozTi =
8z0(1) = 8z0(2) = o. This resuIt highlights a
cubic unit ceIl. The overIap of the three
tetragonal reflections (102), (201) and (210),
confirms the tetragonal-to-cubic phase transition
(see Figure 2). Until PLT25, these three
reflections appear individually, but for PLT30
they coalesce according to the cubic muItiplicity
of (210). Therefore, a new Rietveld refinement
was conducted for PLT3 O using as initial model
the cubic structure obtained by Glazer &
MabudlO for PbTi03 at 5500 C. In this structure,
the asymmetric unit is defined by three
independent atoms, Pb (O O O), Ti (Y:z Y:z Y:z), and
O (Y:z Y:z O), and no positionaI parameter needs to
be refined. The fitting obtained in the cubic
perovskite phase Pm3 m is significantly better
than in P4mm. The experimental and calculated
XRD pattems of the two extremes of the
compositional series studied here, PLTOO and
PLT30, are shown in Figures 3 and 4,
respectively. Figure 5 is an ORTEP_324 view of
the PLTOO structure, which is tetragonal P4mm.
ln PLTOO, the Ti, 0(1), and 0(2) are positively
shifted along the c direction, whereas in the
PLT30 structure (Pm3 m), these atoms occupy
exactIy the comers (A site), the face centers
(oxygens sites), and the body center (B site) of
the unit cell, which means the ideal perovskite
cubic structure. The final structural parameters
of all samples are shown in Table I.
For all La-containing PbTi03 samples, the A
site occupancy was shared between Pb and La.
Considering A -site vacancies only, the
refinements of the sample containing La were
started with the defect equation Pb(l-
l.5y)LayDo.5yTi03, meaning that the a value is
equal to 1.5. Several models considering B-site
vacancies only (a == 0.74) and intermediate a
values (0.74 :s a :s 1.5) were carried out,
however these refinements present unrealistic
values for the atornic occupancies and worse
statistical compatible parameters, even for the
samples with high La content. The A -site
occupation is also given in Table I. The Pb and
La content obtained from XRD data present
very similar values to the nominal compositions.
Nevertheless, it is important to emphasize that
realistic occupancy results depend on the atomic
number di:fference among the atoms involved in
the site occupancy refinements. In this case, the
Pb (82) and La (57) atomic numbers are
significantIy di:fferent, which usually permits a
suitable occupancy refinement. However, false
minima and other problems remain such as the
correlation between occupancy and the thermal
displacement of the atoms. Therefore, in order
to confirm the nominal values, the chemical
composition oftwo samples, PLT05 and PLT30,
were quantified by WDS measurements. The
obtained chemical formula for PLT05,
Pbo.8~ao.049Th.o21Do.o4603, and PLT30,
PbO•61J-ao.285Tio.9790o.12203,are compatible with
the nominal and XRD compositions. Therefore,
it is expected that other samples also present
very similar compositions to the nominal one.
Figure 6 shows the La content dependence of
the a and c cell parameters and the c/a ratio. The
a dimension remains almost constant from
PLTOO to PLT30, whereas c dimension
decreases gradually in the increase of La
content. Consequently, the ela ratio decreases
very fast reaching values close to 1 starting from
the PLT20 sample. For the PLT30, ela is exactly
equaI .to 1 due to cubic symmetry constraint.
These results highlight a structural phase
transition from terragonal-to-cubic occurring
between PLT25 and PLT30 samples.
ln Table I the shift from the ideaIly cubic
perovskite structure is also shown: OZTi, 020(1),
and 020(2). The evolution of these parameters
with lanthanum content, multiplied by their
respective c axis unit cell parameter, is presented
in Figure 7. For PLTOO, a large difference
Ml!tween the oxygen and titanium shifts is
120
 observed: OZ0(2) = OZo{l) = 0.48 A and OZn = 0.15
A. Apart from the increase of La content, allof
them decrease almost linearly until reaching zero
for PLT30. These changes directly affect the
geometric structural parameters, which are also
shown in Table n. In Figure 8 the lanthanum
dependence of the Ti-O and Pb-O distances is .
presented. It can be observed that Ti-0(2) and
Pb-O(l) separations (see also Figure 5) remain
constant throughout the whole series. This
behavior was expected since Ti-0(2) and Pb-
O( 1) lie on the perpendicular plane to the ferro-
distortive axis, which is unaffected by the c-axis
disto rtio n. On the other hand, it is observed that
increasing La content, Ti-O(I) also increases,
i
whereas, as expected, Ti-O( 1 decreases due to
translational symmetry. In fact, Ti-O(l), Ti-
O( 1)', and Ti-0(2) distances are statistically
different from PLTOO to PLT20. For PLT25 the
three distances present very similar values and
for PLT30 they are obviously equal due to the
cubic symmetry constraint. In contrast to what
has been observed with Ti-O, the Pb-O(I), Pb-
0(2), and Pbi-0(2) distances remain statistically
different along the whole curve, except for
PLT30. When increasing La content, the Pb-
0(2) distance decreases, whereas, as expected,
Pbi-0(2) increases due to symmetry dependence
until they reach the constant curve of Pb-O(l )
taking place just iJ,1the PLT30 sample.
c. X-ray Absorption Near Edge Structure
(XANES). Figure 9 shows the XANES spectra
obtained from the Ti K-edge for PLT samples.
The pre-edge region of the K-edge XANES
spectra of some transition metal oxides are
characterized by a pronounced feature, several
volts before the main rising edge.25,26,27 In
transition metal oxides that crystallize in
centrosymmetric structures, this pre-edge
feature is very small or absent; in
noncentrosymmetric structures it can be quite.
1arge. 25' 26,27 A ccor din g to t h e I'iterarure, th e
physical origin of the pre-ed~e feature, labeled A
in Figure 9, is due to the transition of the
metallic ls electron to an unfilled d orbital. 25
This dipole forbidden electronic transition is
normally allowed by the mixture of p character
from surrounding ox:ygen atoms into the unfilled
d states. On the other hand, the pre-edge feature
labeled B in Figure 9 was shown to be caused by
Ti l s electron transition to the unoccupied 3d-
originated eg-type molecular orbital of Ti06
polyhedra neighboring the absorbing Ti atoms,
which are weakly affected by the core hole
potential " The area of the B peak does not
depend strongly on small displacements of the
atoms from their sites in cubic crystal lattice, but
it changes significantly when 4d atoms appear in
the vicinity of the absorbing Ti atorn, for
instance, Zr atoms in the PZT solid solutioo.27
As ean be seen in Figure 9, the intensity
of the A feature for the PLT05 sample is quite
similar to the PLTOO one. The increase of
lanthanum content to 10 at. % (PLTlO) leads to
a significant deerease in the intensity of the A
feature. A further increase oflanthanurn to 3O at
% (PLT30) promotes a slight variation in the A
feature intensity. This behavior indicates a small
dependence of the A feature intensity in
lanthanum concentration, at least for samples
containing more than 10 at. % ofLa.
Figure 10 shows the XANES spectrum ofthe
SrTi03 sample, in which Ti atoms are practically
located at the Ti06 octahedron center. As was
previously observed," it results in a very small A
feature. The comparison of the A feature in
SrTi03 sample with that one observed in PLT
samples indicates that the Ti atoms are moved
off-center in Ti06 octahedron.
Figure 11 compares the intensity of the
peak A in the XANES spectra with the
magnitude of the Ti off-center position obtained
from XRD data as a function of lanthanurn
contento The intensity of peak A in the XANES
spectra was obtained using a single Gaussian
. function in the interval from 4,965 to 4,980 eV.
Aceording to the literature, the intensity of A
peak is proportional to the amplitude of the Ti
atom distortion.26,27,28 From XRD data, the Ti
off-center magnitude is defined as (Ti-Oí l ) - Ti-
0(1)' )12 (Ti-O distances compiled from Table
11). From O to 10 at. % of La, the off-center
magnitude obtained from the XANES spectra
decreases as the content of La increases,
compatible with XRD. For the samples
containing more than 10 at. % of La, XANES
results show that the magnitude of Ti off-center
position remains almost constant. On the other
hand, XRD data suggest that the magnitude of
Ti off-center position decreases as the La
content increases reaching an idealized cubic
perovskite structure for the PLT30.
121
I F S ( • U S P SERv IÇO LH: t)IB~IOTECA
INFORMAÇAO
 For the PLTOO sample, the magnitude of the
off-center position can also be obtained from
EXAFS data. According to the data presented
on Table Ifl, it was found that the distortion is
equal to 0.31Á, very elose to the values obtained
from XRD and XANES data and by others
authors using XAS and XRD techniques.16,17,18,19
With regard to the analysis of the B pre-edge
feature, since we found no significant variation
in its shape and amplitude when the content of
La was increased, it is reasonable to assume that
La atom replaces preferably the Pb atom rather
than the Ti one. This assumption is in
accordance with our XRD resuIts.
d. Extended X-Ray Absorption Fine
Structure (EXAFS). The EXAFS signals and
the corresponding Fourier transforms (FTs) of
the PLT samples are compared in Figures 12a
and 12b respectively. The presence of the
lanthanum Lm edge at 5,483 eV just after the Ti
K-edge (4,920 eV) limits the analysis of the Ti
K-edge EXAFS spectra and due to this fact the
.1.k range that is available in the simulation is
limited to approximately 8 A-I.
As can be seen in Figure 12, some
differences appear in the FTs when lanthanum is
added to the PbTi03 perovskite structure. The
fust difference is a shift towards smaller
distances of the Fourier transform fírst peak,
which is correlated to the fírst coordination shell
around titanium atoms (Ti-O bonds). Another
significant difference is an increase in the
amplitude of this Fourier transform first peak,
mainly for the 30 at. % of La sample (PLT30).
At longer distances, the apparition of a peak
around 3.5 A for alI Ianthanum-containing
PbTi03 samples was observed. However, its
shape and amplitude did not vary significant1y as
the La increases from 10 to 30 ato %.
The theoretical EXAFS spectra for the
PLTOO and PLTIO samples were calculated
using the FEFF8.2 software package/" in order
to visualize the effect of the presence of
lanthanum in the short-range ordef structure
around the Ti atom. The FEFF program
applying a curved-wave multiple-scattering
theory is a powerful computational techni;bue to
calculate the theoretical EXAFS spectra. The
calculations were based on the crystallographic
structure of the PLTOO and PLTIO samples
obtained from our Rietveld refinements. As the
effect of thermal and static disorder represented
by the Debye-Waller factor was not introduced
in the theoretical calculation, a large difference
in the amplitude of experimental and theoretical
Fourier transform curves is expected.
Figure 13 shows the modulus of the
Fourier transform curves obtained from the
theoretical EXAFS spectra. As can be seen, the
fust peak of the Fourier transform, which is
correlated to the first Ti-O neighbors, shifts
towards lower distances for the PLTlO sample
and presents a higher intensity. As the number of
first Ti-O neighbors is the same in both samples,
this increase of intensity indicates a decrease of
the Ti off-center distortion for the PLTIO
sample. Similar to the experimental Fourier
transform curve shown in Figure 12, a more
intense peak located between 3 and 4 A is also
observed in the theoretical Fourier transform
modulus of the Ti K EXAFS spectrum of the
PLTlO sample (Fig. 13). According to the
calculation, this peak is mainly formed by
contributions from single (Ti-Pb, Ti-La, and Ti-
Ti) and from multiple-scattering paths (Ti-O- Ti,
Ti-O-O, O-Ti-O-Ti, Ti-O-Pb, and Ti-O-La). The
increase in the intensity of this peak observed in
the PLTI0 Fourier transform curve could be
attributed to the decrease of the structural
distortion at the medium range-order structure
around the Ti atoms.
In order to obtain quantitative structural data
concerning the Ti-O first coordination shell, the
EXAFS spectra of the Fourier transform first
peak (between 0.8 and 2.2 A) presented in
Figure 12 were extracted using a back Fourier
transform and then fitted using theoretical phase
and amplitude functions, To model the Ti-O pair
we used theoretical amplitude and phase
functions calculated using a FEFF 8.2
program." In all fíts, the number of free
parameters was kept smaller than the number of
independent points de:fined as Nnd=2LlRA.Kln,
where L1R is the width of the R-space filter
windows and L1K is the actual interval of the fit
in the K space." The reliability of the fit was
deterroined by a quality factor.22
According to the XRD results (see Figure 5
and Table TI), for PLT samples from O to 25 at.
% of La, the fírst coordination shell around the
Ti atom forms a distorted octahedron with three
different Ti-Q bond lengths (three-shell): two of
122
 them along the axial c axis and the third one
referring to the four equatorial bonds, which
present the same length due to the structure
symmetry. On the other hand, for the PLT30
sample the first coordination shell around the Ti
atom forros a perfect octahedron, for which the
six Ti bonds are equivaleut (single-shell) due to
the Ti m3 m (Oh) point symmetry. Therefore,
during the fitting procedures of the EXAFS
spectra using the FEFF 8.2 program, 29 initially
applied to the PLTOOsample, two models were
tested: a single-shell and a three-shell. ln both
models, the coordination number (NTi-o) was
fixed while the interatomic distance (RTi-d and
the Debye-Waller factor (ano) were free to
VaI)'. In the three-shell model,a single value of
ar-o was applied to the three types of Ti-O
pairs reducing the number of free parameters
used in the fitting, Table m presents the fitting
results. As observed, the three-shell model is
significant1y better than the single-shell one for
the PLTOO. Moreover, the three bond lengths
obtained from the EXAFS three-shell fitting are
in accordance with those ones obtained from our
XRD refinement (Table lI) and also from
previous XAS work, where a three shell fitting
was also use d.'28 31 F'igure 14 shows the
comparison between the filtered experimental
EXAFS spectrum and the one-shell and the
three-shell theoretical models for the PLTOO.
For La-containing samples, we followed the
same procedure adopted to fit the EXAFS
spectra of the PLTOO'sample. First1y,we carried
out the fitting ofthe first Ti-O coordination shell
of the PLTlO sample. As can be seen on Table
Ill, the fitting quality factors of both models
(single and three-shell) are comparable.
lndistinguishable fitting quality factors were also
observed for the other PLTx ceramic samples.
This fact was expected because the distance
resolution (AR=n/2~k) achieved during the
fitting is determined by the us~l ~k data range,
which in this case, was limited by the presence
of the La Lnredge Iocated just after the Ti K-
edge. Since the useful ôk range was about 8.0 A
for alI sampIes, AR. was found to be equal to
0.19 A This means that Ti-O bond lengths that
differ less than 0.19 A cannot be resolved. Tius
explains why the three-shell or single-shell
mode1s are statistically indistinguishable ando
l
..
provide unrealistic structural results for rich
lanthanumsamples (10 to 30 at. %).
Thus EXAFS and XANES results are in
accordance with previous studies using Raman
spectroscopy in the PLT samples.14,15,16
According to the work of Tavares et al.16, the
presence of extra Raman active modes in a 27
at. % La sample can be explained by the
existence of a short-range structural disorder.
On the other hand, Kim et al. 15,16 argue that the
extra Raman active modes observed to a 30 at.
% La sample measured at 300 K arise from
short-ranged c1usterswith tetragonal 4mm (C4V)
symmetry, which are distributed over the
globally cubic m3 m (00 symmetry matrix.
Although our XANES and EXAFS results show
a local disorder concerning the PLT30 sample,
any conclusion couId be drawn regarding the
existence or not of tetragonal clusters, as
suggested by Kim and co-workers.l":"
Since there is a certain discrepancy between
XANESIEXAFS and XRD structural results
concerning the PLT30 sample, it is important to .
understand the XRD results in the context of our
XAS results.
AlI XRD and XAS measurements presented in
this paper have been taken at room temperature,
~25°C. Therefore, from PLTOO to PLT25
samples, for which the respective ferroelectric-
paraeIectric transition temperature (see Figure 1)
is above the experimental one, and which
present a ferroelectric-normal behavior below
Te, a tetragonal structure is expected. This
expectation was coufirmed forboth techniques,
XRDandXAS.
Concerning the PLT30 sample for which T is
around -50°C, XRD and XAS measurements
have been taken above the ferroelectric-
paraeIectric transition temperature (Tc).
According to previous Raman spectroscopy
measurernents at room temperature, the
structure of this sample should present a residual
short-range structural disorder in a global cubic
. 141516
matnx. ' , . Here, the XAS measurements
show that the Ti atom remains moved off-center
in its octahedron. On the other hand, the XRD
analysis shows that the Ti atom is in its ideal
position (Y2, Y2, 7'2).
It is important to emphasize that XRD is a
Iong-range probe technique. However, using
high-precision X-ray diffraction single crystal
123
data, Abramov et aI.32 have perfonned a charge
density study, which allowed them to conclude
that the cubic SrTi03 structure is subject to two
types of distortion with short-range order. The
Ti atom lies dynamically off-center in the
octahedron of its ligand O atorns, and the
oetahedra thernselves vibrate. Abramovet alo 32
argued that depending on the ato mie site
symmetry in the high-symmetry phase, several
subsidiary minima will occur equidistant from
the ideal position, either lying on the
circurnference of a circle (for the Ü atorns in
perovskite) or the surface of a sphere (for the A
and B atoms), The mechanism of reorientation
between these mínima will involve motion
around the circle or sphere rather than through
the ideal position, since the potential barriers on
this surface will be lower than the potential
barrier in the center. Thus, for a high enough
degree of anharmonicity, the atom may move
around the circle or sphere without ever crossing
the ideal position. Even at this extreme, the
crystal will retain its higher symmetry, Pm3 m.
A similar phenomenon quite probably takes
place for the PLT30 sample as pointed out by
our XANES/EXAFS results, which are
techniques more sensible to probe sort-range
order than the XRD one. For the other samples,
the atoms should remain in one minimum long
enough for the surrounding structure to relax
leading to lowercrystal symmetry, P4mm. ln
this case, XANESIEXAFS and XRD techniques
are compatible with each other.
A similar discrepancy between XAS31 and
XRD10 results was commented by Sicron et al.31
when they studied the nature of the ferroelectric
phase transition in the PbTiü3 compound. They
observed a discrepancy between XAS and XRD
results conceming the temperature dependence
of Ti distortion below and above the structural
ferroelectric transition. Whereas XAS results
showed an off-center position of the Ti atorns
above the transition temperature, XRD data
suggested that Ti atoms go back to" the ideal
position.
Teslic and Egarni also found a discrepancy
between structural results obtained by XRD and
neutron scattering data when they studied the
local atomic structure of PbZrl-xTixÜ3 (PZT),
Pb(Mg1l3Nb2l3)03 ~PMN) and Pbl-xLaxZryTil_yÜ3
(PLZT) systerns. ,34 Their structural studies
yielded evidence of slight local deviations frorn
crystallographic lattice periodicity and local
chemical ordering.
Summarizing, the discrepancy observed in the
XAS and XRD resuIts conceming the PLT30
sample is basically due to the fact that X-Ray
absorption spectroscopy probed the short-range
order around the Ti atorns, whereas X-ray
diffraction provides structural infonnation about
the average structure.26,27,28,32,33,34
Conclusions
The structure of the Pb1_xLaxTi03
ceramic samples was examined by XRD-
Rietveld refinements and the XAS technique.
According ta XRD-RietveId refinements, the
tetraganality of the structure gradually
diminishes with the increase afLa content until a
structural phase transitian fiam tetragonal-to-
cubic structure, which takes place in the range
between the 25 and 30 at. % af La samples. ln
fact, the samples with x ranging fiam O ta 25 aí.
% crystallize in the P4mm space group, whereas I
the sample with 30 ato % ofLa crystallizes ín the t
Pm3 m space group. The refinement of XRD
data also confirmed the existence of only A -site
vacancies, which are introduced to the
perovskite structure by the alio valente
replacement ofPb2+ by La3+ cations,
According to the XANES data, the Ti atom is
in an off-center position io the Ti06 octahedra
(Ti point symmetry = 4mm) for all samples,
c.
inc1uding the PLT30 samples for which XRD
highlighted a cubic structure (Ti point symmetry
= m3 m). However, the incompatibility between
the XAS and XRD techniques conceming
PLT30 is only apparent. Recent1y, it has been
showrr" that, even in an ideal perovskite cubic
structure, where the crystal retains its higher
symmetry (Pm3 m), atoms may move around its
ideal position. This structural short-range order
phenomenon could be probed by the XAS
technique, but not by a conventional XRD one.
The presence of the lanthanum Lm-edge after
the titanium K-edge did not allow us to obtain
quantitative structural information from the
EXAFS spectra of the La-rich PbTiü3 samples. fi
In this way, except for the PLTOO (sample
without La), any attempt to fit the first shell
using single or three-shell models leads to
124
unrealistic structural models. However, it was
possible to infer that the disorder in the first shell
around titanium atoms persists, in accordance
with the XANES and XRD results. PLT30
Finally, our XANES and EXAFS results show
that the Ti atoms are in an off-center position for PLT25
any lanthanum content, as mentioned in previous
""
studies using the Raman spectroscopy in PL Tx :::J
.i PLT20
samples.14,15,16 ~
.u;
c
PLT15
Acknowledgment. Research partially carried ~
out at LNLS - National Laboratory of
PLT10
Synchrotron Light, Brazil. We are also grateful
to FAPESP (00/09722-9), CNPq (Brazil) and
USP-COFECUB for their financial support. The
authors wish to thank Dra. Ducinei Garcia (DF-
UFSCar) for her help during the sample 50 51 52 53
preparation. We also thank José A. L. da Rocha Degree ze I Ka. eu
and J. G. Catarino for the important support
during the XRD measurements. One of us,
Figure 2. Evolution of the (102), (201), and
ACD, thanks FAPESP (01106993-4) for a
postdoctoral feIlowship. YPM, VRM, JAE aIso
(210) reflections on the La content doping
thank CNPq for a researcher fellowship.
mcrease

11000 -..----- __ .----------, J.;a: -

(\PlT20
10000
II
11
>-
1=
9000 li
1\ 3.]-: -
~
!:::
8000
Ii PLT15
::iE
ffi 7000
Ii
• I
to.
jI
PLT10
•
Q.
!\ ;\ 1\
ºg: 6000 !,
i \
j \
:.;
J \
i \
-.Y: -

D-: -

.."'" iI \ j I\\/ I\\

~ 5000

o
/]!)
4000

:::
11):0. 11<1

PLT30 <,
,I I f }\ \ PLT05
~
/ ,/ / i
..
t>
,
\.

/-._./
'
1000
Figure 3. Experimental (+) and calculated (-)
o~~~~~~~~~~~~~
-200 -100 o 100 200 300 400 500
XRD pattems of the PLTOO. The difference
TEMPERA TURE fC)
between the two pattems is presented below the
Figure 1. Temperature dependence of the
relative dielectric permittivity ofPLTx samples pattems.

125
 r ~ ~<"T'"-"~~~ ,
;:' 4.15~""~\--------------' 1.06

""
-, I UO 1~
1 I -O-a

·\~o ::::~/.
1".:' 1
«
'.0WuGWL,JuULA
i I ~i,
1.04
1 1 ~
fi 4.05 o
.1!!
••E '§
1.03
: -Á-,. • • (1]

I ,".' '" ..:....., . , f! 4.00

1.02 o

, I '"
n,

I ~ 3.95 e~o--o
....JI' UlJj~ A _D __ O-e~~~o
3.90 '--.:;:.o_.-_..,.o_-_-.-_-r-_-.-_--.-_--r-e-J
1.01

1.00
!>}I) 40,' ".6 ..., ].'.'J
~-lHETA
&-jJi J[;.V
••• .14. PLTOO
PLT05PLT10PLT15PLT20PLT25PLT30
Sample

Figure 4. Experimental (+) and calculated (-)

XRD pattems of the PLT30. The difference

Figure 6. Dependence of the cell parameters and
between the two pattems is presented below the
the c/a ratio with the La content.
pattems.

O.5~---------------,

°'
4 ~,
O~
l-O-l<7,
-O-~,)

-; 0,3 ~,,~-/)"-1lZ0(2)
.~
N

i::: o--o- __ o~

--0-,,\
0,0 L..,.----.----r--,---,---,---r'
PLTOO PLT05 PLT10 PLT15 PLT20 PLT25 PLT30
(a) (b) Sample

Figure 5. ORTEP-3 drawing showing (a) lhe unit

Figure 7. Dependence of the shift from the
cell, atom designation, and the Ti (site A) octahedral

coordination environment and (b) the Pb,La (site B) ideally cubic perovskite structure (ozn, OZo(l)

••Displacement
12-fold coordination environment. and oZQ(2», multiplied by their respective c unit

ellipsoids are drawn at the 50% levei considering an cell parameter, with the La contento
idealized isotropic thermal parameter of 0.025 N
(Symmetry codes: j x, y, z+ 1; jj -y+ 1, x, z; iii z; y, z-

1).

126
 f.
3.2 4----.
Pb-O(2) -------..-------..,

:: :.: ~b-O(l) • ~

~
., 2.6 ~b'-O(2) .-. __ ~

-'
"O
2.4 .~
r:
dl 2.2
T•..O(1)·~
Ti-O(2) ------. _

2.0 .---4-4---4~~
1.8 ~~ -----

PLTOO PLT05 PLT10 PLT15 PLT20 PLT25 PLT30

Sample

4960 4980 5000 5020 5040

Energy (eV)
Figure 8. Dependence of the Ti-O and Pb-O
Figure 10. Ti K-edge normalized XANES
bond lengths with the La
spectra of SrTi03 powder samples.
contento

0.35 0.35

0.30
•o o XANES
0.30
• • XRD ~
(\)
~0.25 o 0.25 :;)

~'"
'~ª
c
~ 0.20 9-
0.20
• o
o u
(\)
j50.15 o o 0.15
~
~ • o '"
»
ri) 0.10 XRD 0.10 :;
i= x
0.05 • 0.05 »
z
m
(f)
0.00 • • 0.00

PLTOO PLT05 PLT10 PLT15 PLT20 PLT25 PLT30

Sample

4960 4980 5000 5020 5040

Energy(eV) Figure 11. Variation of the intensity of peak A

in XANES spectra compared with Ti site

Figure 9. Ti K-edge normalized XANES spectra distortion from the center of the oxygen

ofPLTx powder samples. octahedron obtained from XRD data. (Ti atom

distortion = (Ti-O(l) - Ti-O(l)i ) /2, where the

Ti-O bond lengths were compiled from Table

lI).

127
 PLTOO
o
12
PLT10 '.
PLT15 10
~c: -1
:>
.e PLT20 ê
.!!!. ~ 8
PLT25 c
g -2 f!
... PLT30 ':::
6
Gl
'C
::s
&4
2 3 4 5 6 7 8 9 10
K(A')
2

O~~~-r~~~~~~.-~~~
o 234 5 6
(a) R (A)

10~------------------------,
8

6
Figure 13. Fourier transform modulus of Ti K
E
~ 4 EXAFS spectra of the PLTOO and PLT 1O
c
f!
I- 2
.~ calculated by FEFF8.2
:; o
~
·2 programo

23456
R (A) 0.6 r---------------------,
- Three-shell Fitting
0.4 o Experimental
(b) o One-shell fitting

0.2
Figure 12. (a) Experimental KX(k) versus k

EXAFS spectra of PLTx samples and (b)

Fourier Transform of K3X(k) EXAFS spectra

presented in Figure 3. For the sake of clarity,

each curve was shifted downward by a constant

Figure 14. Fitting and back Fourier filtered
in relation to the preceding one.
experimental signal of the Ti-O first shell for the

• PLTOOsample using a three-shell model,

128
 Table 1. Cell parameters, fractional coordinates, cation distribution, and powder and reflection data statistics.

Sample A c OZTi* ÕZo(l) OZ0(2) Composition T wRp t R(F2)

PLTOO 3.90197(4) 4.1373(1) 0.036(1) 0.115(2) 0.115(2) Pbl.ooLao.ooTh.ooDo.oOO3 0.097 26.8 0.035

PLT05 3.90809(4) 4.0733(1) 0.029(1) 0.101(2) 0.110(2) Pbo.91Lao.06

Th.ooDo.02403 0.078 16.5 0.033

PLTlO 3. 91502( 4) 4.0181(1) 0.030(1) 0.077(2) 0.096(1) Pho.8sLao.OS

Th.ooDo.oSOO3 0.083 14.2 0.044

PLT15 3.91687(4) 3.97939(4) 0.023(1) 0.057(2) 0.077(1) PbO.81Lao.13

Til.ooDo.o703 0.061 10.2 0.040 0'1
~
•....•
PLT20 3.92357(2) 3.94878(2) 0.017(1) 0.034(3) 0.053(1) PbO.7sLao.17Til.ooDo.lOO3 0.042 4.4 0.044

PLT25 3.93367(2) 3.95371(3) 0.021(1) 0.022(5) 0.042(2) PbO.67Lao.22

TÍt.ooDo.1303 0.049 4.9 0.063

PLT30 3.91690(1) 3.91690(1) O O O PbO.6~ao.2STÍt.ooDo.1303 0.052 6.1 0.058

• The atomic fractional coordinates are: Pb,La at O, O, O; Ti at 1/2, 1/2, 1/2+ÕZTi; 0(1) at 1/2, 1/2, O+ÕZü(l); and 0(2) at O, 1/2, 1/2+õZo(2).
Therefore, ÕZTi,OZQ(I),and õZQ(2)stand for the shift from the ideally cubic perovskite structure along the (001) direction. t The Pb and La content
were obtained from A site occupancy refinement .

..." .••;,:.":-:;,", ..:;"'"",,,,.,,,::~ ··"""ú';i"~\", .•.. ;.'i;,,".' ,;,:·-~··:;-",.'itR:!\'~:'''';';'''''·''''''''':''~/"J..':,~,. -·;:'-"';";":'''<:.''"_i;':,_,c:i._.,,,

' •.·."i·"·.''','.,'"''';,,.;:..;·i'''"'-''.·:"'ú,,·;''''''-"'''''"~~''"'~·~~oiit't -/')«4 r'm _o, 2 r e:liIIIiIIiIiI;I t 5 '.$&'\' •••
"-- .""""
....
,. "'~~.i,:·_\'·-,"':,',..;t;;
~:iií."Wt~:;>&'lijJil;2Ú:::~~~~:~\~~~;t;;lQZ;1;.;;j~'L;;;:~L m'
..~;i:J;;";;;~ii.;,L:;,;,'~;J~:;m:M."Mli;"!iiitÚ~'''''{I"'ja",i\iI'~r''~*llít~f't'tll''fti.'tctY'MtMtWféa' l
f@ ftwW:1n"iii:i1*àtltt==tar t tt(i"&Ü' te n t t rfr

Table 2. Selected geometric parameters (Á).

PLTOO PLT05 PLTI0 PLTl5 PLT20 PLT25 PLT30

Pb-Tí 3.541(3) 3.503(3) 3.493(2) 3.464(3) 3.444 (3) 3.465(3) 3.39213(1)

Pb-Tiili 3.360(3) 3.365(3) 3.351(2) 3.358(3) 3.367(2) 3.365(3) 3.39213(1)

Ti-O(l) 2.39(1) 2.329(9) 2.199(8) 2.12(1 ) 2.04(1) 1.98(2) 1.95845(1)

Ti-O(I)i 1.74(1) 1.744(9) 1.819(8) 1.86(1) 1.91(1) 1.97(2) 1.95845(1)

Ti-O(2) 1.978(1) 1.982(1) 1.9755(8) 1.9704(8) 1.9669(5) 1.9685(4) 1.95845(1)

Pb-O(2) 3.203(6) 3.162(4) 3.094(4) 3.019(4) 2.935(4) 2.908(7) 2.76967(1)

pb-O(2)ili

Pb-O(1)
2.519(4)

2.800(2)
2.517(3)

2.794(1)
2.542(3)

2.7858(7)
2.581(4)

2.7788(7)
2.64:0(4)

2.7775(5)
2.674(6)

2.7829(6)
2.76967(1)

2.76967(1)
-
o
M

0(1)-0(2) 2.844 2.852 2.860 2.851 2.838 2.845 2.770

O(li-O(2) 2.843 2.793 2.751 2.733 2.730 2.734 2.770

0(2)-0 (2)Ü 2.759 2.763 2.768 2.770 2.774 2.782 2.770

••
Symmetry codes: x, y, z+ 1; -y+ I, x, z;
1 U W x, y, z-l ). The symmetry codes as in Figure 4.
Table 3. Structural parameters ofPLTx samples obtained from EXAFS analysis.

Single-shell fitting Three-shell fitting

Samples

N=6.0 NI = 4.0 Nz = 1.0 N3 = 1.0

R = 1.97(1) Â R J= 1. 98(1) Â s, = 1.78(1) Â R3 = 2.39(3) A
PbTi03
a= 0.116(5) Â oi = 0.075(6) Â az = 0.075(6) Â as= 0.075(6) Â

Qllality factor

N=6.0
= 5.2 Quality factor

NI = 4.0
= 1.O
N: = 1.0 N3 = 1.0
--
M

R= 1.92(1) Â RI= 1.94(1) Â Ri= 1.81(1) Â R3= 2.41(5) A

PLTIO
a= 0.125(4) Â o i = 0.10(1) A , az = 0.10(1) Â as= 0.10(1) A

Qllality factor = 1.1 Qllality factor = 1.3

•
References

1 Keizer, K.; Lansink, G.J.; Burggraaf AJ. J. Phys. Chem Solids 1978,39,59.
2 Garcia, D.~Eiras, I.A Ferroelectrics 1991, 123,51.
3 Bhaskar, S.; Majumder, S.B.; Katiyar, R.S. Applied Physics Letters 2002,80(21),3997.

4 Dai, x.. Xu, Z.; Viehland, D. 1. AppI. Phys. 1996,51,4356.

5 Hennings, D. Mat. Res. Bull. 1971,6,329.

6 Kaizer, K.; Bouwama, J.; Burgraaf, AI. Phys. Status Solid 1976, A 35, 28l.

7 Kuwabara, M.; Goda, K; Oshima, K Phys. Rev. B 1990,42, 10012.

8 Shirane, G.; Pepinsky, R; Frazer, BrC, Physical Review 1955,97, 1179.

9 Shirane, G.; Pepinsky, R; Frazer, B.C. Acta Crystallographica 1956, 9, 131.

10 Glazer, AM.; Mabud, S.A; ActaCrystallographica 1978, B34, 1065.

11 Nehnes, RI.; Kuhs, W.F. Solid State Communications 1985, 54, 721.
12 Shannon, RD. Acta Crystallographica 1976, A32, 75l.
13 Hennings, D.; Rosenste, G. Materials Research Bulletin 1972, 7, 1505.

14 Tae-Yong, K; Hyun, M1. Applied Physics Letters 2000, 77, 3824.

15 Kim, T.Y.; Jang, H.M.; Cho, S.M. Journal Of Applied Physics 2002,91,336.
16 Tavares, E.C.S.; Pizani, P.S.; Eiras, J.A Applied Physics Letters 1998, 72, 897.

17 Rietveld, H.M. Joumal of Applied Crystallography 1969, 2, 65.

18 Larson, A.C.; Von Dreele, RB. GSAS. 1996 Generalized Structure Analysis System, Los Alamos
National Laboratory.

19 Rez, D.; Rez, P.; Grant, 1. Acta Crystallographica 1994, A50, 481.
20 Tolentino, H.C.N.; Ramos, AY.; Alves, M.C.M; Barrea, RA; Tarnura, E.; Cezar, I.C.; Watanabe, t
N. J. Synchrotron Rad. 2001, 8, 1040.
•
21 Michalowicz, A I. Phys. IV 1997, 7, C2-235.
I
!
f

22 Report on the Intemational Workshops on Standards and Criteria in XAFS, In X-Ray Absorption
Fine Structure: Proceedings of the VI Intemational Conference on X-Ray Absorption Fine Structures,
ed. S.S. Hasnain. Ellis Horwood, New York, 1991, 752.
23 Michalowicz, A; Noinville, V. In Gallad 2.0 code, LURE, Orsay, France 1992.
I
I
132
I r
t

I
I
;

24 Farrugia, L.J. I Appl. Cryst 1997, 30, 565.

25 Ravel, B.; Stern, E.A. Physica B 1995,208-209,316.

~ 26 Ravel, B.; Stern, E.A.; Vedrinskii, RI.; Kraizman, V Ferroelectrics 1998,206-207,407.

27 Veddrinskii, RV; Kraizman, VL.; Novakovich, AA; Demekhin, Ph.V; Urazhdin, S.Y. I Phys.:
Condensed Matter 1998, 10,9561.

28 Miyanaga, T.; Diop, D.; Ikeda, S.I.; Kon, H. Indian Geotechnical Conference 2002, in press.

29 Ankudinov, AL.; Ravel, B.; Conradson, S.D.; Rehr, J.J. Phys. Rev B 1998, 58, 7565.

30 Rehr, J.I; Albers, RC. Review ofModern Physics 2000, 72, 621.

31 Sicron, N.; Ravel, B.; Yacoby, Y; Stern, E.A; Dogan, F.; Rehr, II Physical Rev. B 1994, 50,
l3168.

32 Abramov, YA; Tsirelson, V.G.; Zavodnic, VE.; Ivanov, S.A; Brown, I.D. Acta
Crystallographica. 1995, B51, 942.

33 Teslic, S.; Egami, T.; Viehland, D. I Phys. and Chem ofSolids 1996,57, 1537.

34 Egami, T.; Teslic, S.; Dmowski, W.; Viehland, D.; Vakhrushev, S. Ferroelectrics 1997, 199, 103.

133
Short and Long-Range Order Structure in the PLT20 (Te=70°C) and PLT30 (Te=-50°C)
Pb1-xLax
Ti03 Ceramics Below and Above samples were refined as tetragonal and
the Ferroelectric Phase Transition cubic respectively. Thus, XRD results clearly
determined by XAS and XRD shown a tetragonal to cubic phase transitíon
above their respective ferroelectric-
P. P. Neves", V. R. Mastelaro\ A. C. paraelectric phase transition temperature,
Doriquetto", Y. P. Mascarenhas", Te. On the other hand, XAS measurement in
A. Mlchalowicz.é J. A. Eiras" both samples shows that the local order
around Ti atoms below and above the phase
'Instituto de Física de São Carlos, Universidade transition temperatures are practically the
de São Paulo, C.P. 369 - São Carlos-SP, same. XAS results suggest then that the
Brazil, local structure about titanium site is only
2Groupe de Physique de Milieux Oenses,
weakly dependent on crystallographic
Université Paris XII-Vai de Mame, 94010,
phase. In conclusion, in PLT samples,
Creteil Cedex, France, 30epartamento de
although the displacive nature of phase
Física, Universidade Federal de São Carlos,
transition cannot be ignored, these results
São Carlos - SP, Brazil
are consistent with the order-disorder

Abstract character of its ferroelectric transition.

In the present work, in-situ X-ray absorption

spectroscopy (XAS) and X-ray Diffraction
(XRD) techniques were applied to the
structural analysis of Pb1-xLaxTi03ceramics
(PLTX, X=20 and 30) below and above the
ferroelectric phase transition (Te). XAS at Versão preliminar 26-07-2004

the Ti K-edge and XRD measurements were

carried out on two different samples: the
PLT20 sample which psesents a
ferroelectric-normal behaviour for which Te
is around 70°C and the PLT30 sample, for
which Te is around -50°C. According to
XRD-Rietveld refinements made below and
above the phase transition temperatures,

134
I, Introduction technique to the diffraction methods, which
are based on the long-range order structure
The peroviskite-type compounds [8-12].
presenting ferroelectric properties has XAS studies have provide interesting
attracted the attention of many researches information concerning the short-range
due their technological properties. The order structure and the mechanism of phase
peroviskita cornpounds based on the AB03 transition of AB03 peroviskite compounds
formula undergo a structural phase [8-12]. It has been observed that the local
transition at one specific temperature structure of a number of ferroelectric
denominate Te. peroviskites was essentially disordered at
The fundamental nature of the temperatures above the phase transition [8-
ferroelectric phase transition in AB03 12]. Based on these results, it was argued
perovskite materiais has been the subject of that the mechanism of phase transition on
a great deal of research and some this class of materiais was not purely
controversy [1]. One the main question is displacive but shows that an essential
whether tne transition is essentially element of order-disorder is present at
displacive or order-disorder [2]. Peroviskite temperatures above Te[8-11].
ferroelectrics as well as crystals belonging The existence of a disardered local
to other crystal families do have a Curie- structure well above the phase transition
Weiss constant typical of displacive temperature was also observed by XAS on
ferroelectrics [3]. Some of these crystals mixed crystals with impurity in the B site
also display underdamped soft modes [1,13]. Hanske-Petitierre et aI. studied the
above Te. While these properties indicate a KTa1-xNbx03system by XAS and showed
displacive like behaviour, others that Nb atoms are displaced to off center
experiments show qualitative evidence for positions far above the temperature ot the
order-disorder like behavior [4,5]. phase transition where the long-range
Structural changes in the peroviskite structure was classified as being cubic [1].
materiais have been for long time studied by Shuvaeva et aI. studied the local structure in
•
X-ray diffraction and neutron methods [6,7]. Ba1_XSrTi03system by XAS and observed
Recently, X-ray absorption that the Ba off-center displacement
spectroscopy (XAS) has become an disappears abruptly at x=O.3, which is the
important technique to study the local boundary point between tetragonal and
structure of this class of materiais and has cubic phases [13]. On the other hand, Ti off-
been considered as a complementary center displacement decreases gradually

135
with Sr concentration and approaches zero measurements indicate that the local
at about x=0.7. They concluded that the structure around Ti atoms had a different
change of macroscopic symmetry does not compositional dependence from that
cause any significant change of Ti obtained from XRD mainly for the 30 at. %
displacement and that Ti atoms are sample. According to the XANES data, for
disordered in a wide range of Sr the 30 at. % sample, a local distortion
concentration [13]. around Ti atoms persist.
Lanthanum-doped lead titanate In the present work, in-situ X-ray
ceramics (Pb1-xLaxTi03 or PLT) have been absorption spectroscopy (XAS) and X-Ray
studied in detail due to their interesting Diffraction techniques where applied to the
physical properties [14,15]. The isomorphic structural analysis of Pb1_xLaxTi03ceramics
substitution of lead by lanthanum atoms (PLTX, X=20 and 30) below and above the
induces some interesting changes in the ferroelectric phase transition (Te). XAS and
physical properties of PbTi03 material. XRD measurements were carried out on two
When the lanthanum content is higher than different samples: the PLT20 sample which
25 at. %, a diffuse character of the presents a ferroelectric-normal behaviour
ferroelectric-paraelectric phase transition where Te is around 343 K and the PLT30
(DPT) is observed [15]. The modification in sample, which presents a ferroelectric-
the phase transition character is followed by relaxor behavoir where Te is around 223 K.
a linear decrease of the Curie temperature
(Te) to room temperature for x-30 at. %. It 11. Experimental
has been suggested that the lanthanum
induced modification on PbTi03 results in The powdered PLT samples, nominal
structural changes that can be directly composition Pb1-xLaxTi03 with x = 20 and 30
related to the nature of the phase transition. at. % (abbreviated PLTx, PLT20 to PLT30),
The short and long-range order were prepared by conventional mixed oxide
structure of Pb1-xLaxTi03 ceramics with x method. The oxides, PbO, and La203, Ti02,
varying from 0.0 to 0.30 has been studied weighed according to the stoichiometry,
11I

using XAS and XRD techniques [16]. XRD were mixed by ball milling in distilled water
results show that a tetragonal structure fitted for 10h. The slurry was dried and calcined in
well to ali samples, except to that one with a covered alumina crucible at 11OO°Cfor 3h.
30 at. % of La for which XRD results pointed Ceramics bodies were then formed by
out to a cubic unit cel/. X-Ray Absorption . isostatic pressure, and fired at -1175°C for
Near Edge Spectroscopy (XANES) 3h. The densities of ali fired ceramics were

136
found to be around 95% of the theoretical were recorded at 293, 328, 383 and 433 K
one by Archimedes method. (the phase transition temperature of PLT20
The titanium K-edge X-Ray sample is 343 K).
absorption spectra were collected at the The EXAFS analysis was carried out
LNLS (National Synchrotron Light by using the program set written by
Laboratory) facility using the D04B-XAS1 Michalowicz [17], according to the
beam line. lhe LNLS storage ring was recommended procedures described by the
operated at 1.36 GeV and 100-160 mA The International Workshop on Standards and
sample pellets obtained after sintering were Criteria in XAFS [18]. After atomic
grounded for XAS measurements. XAS data absorption removal and normalization, the
were collected at the Ti K-edge (4966 eV) in K3x(k) weighted EXAFS signal was Fourier
transmission mode at the different transformed to R distance space in the 2.0-
temperatures using a Si(111) channel-cut 10 A-1 K range. Each spectra was Fourier
monochromator. lonization chambers were transformed using a Kaiser apodisation
used to detect the incident and transmitted window with t=2.5. A qualitative
flux. XANES spectra at the Ti K-edge were interpretation of XANES spectra obtained at
recorded for each sample between 4910 the Ti K-edge was performed using the
and 5200 eV using energy steps of 0.5 eV. software package developed by Michalowicz
Due to the presence of the lanthanum LIII and Noinville. For comparison purpose
edge at 5483 eV, EXAFS measurements at between difterent samples, ali spectra were
the Ti K-edge (using steps of 2 eV) were background removed and normalized using
carried out for each sample between 4840 as unity the first EXAFS oscillation [19].
and 5400 eV. Four acquisitions were The X-ray powder diffraction patterns
recorded for each sample to improve the were measured at room temperature on a
signal to noise ratio. To provide good energy Rigaku Denki powder diffractometer with
reproducibility during the XANES data geometry 8-28, rotating anode X-ray source
collection, the energy calibration of the (eu Ka radiation, Â = 1.542 A), and
monochromator was checked during the
scintillation detector. The measurernents
collection of the sample data using a Ti
were carried out on difterent temperatures
metal foiL The XAS spectra of PLT30
before and after the phase transition
sample were recorded at 100 K and 293 K, temperature using a current of 100 mA and
which are respectively below and above the tension of 50 kV. Each pattem consists of
phase transition temperature of PLT30
5851 steps ranging trom 30 to 120028. The
sample. The XAS spectra of PLT20 sample

137
data were collected with a step size of 0.02°. edge as function of th. tl"np4Imtl.(~.{rfr.mf'I';~\
The count time was 5 sec per step. The pre-edge region of the
refinement of the structure was carried out spectra of some transition motal o)(llfllj;ctlrlJ~~j,t·)
using the Rietveld method [20]. The characterized by a pronounced
program used in the refinements was the several volts before the main rising edgt
General Structure Analysis System (GSAS) [22-24]. In transition metal oxides that
package[21].' crystallize in centrosymmetric structures,
this pre-edge feature is very small or absent;
111. Results and Discussion in noncentrosymmetric structures it can be
quite large. According to the literature, the
The experimental XRD patterns of the physical origin of the pre-edge feature
PLT20 and PLT 30 compositions as a labeled A on Figure 3 is the transition of the
function of the temperature are shown in metallic 1s electron to an unfilled d state.
Figures 1 and 2, respectively. We only This dipole forbidden electronic transition is
shown on the Figures the part of X-ray normally allowed by the mixture of p
patterns where some changes concerning character from surrounding oxygen atoms
the phase transition is more easily into the unfilled d states.
observed. As can be seen, for both samples, the
According to XRD-Rietveld size of pre-edge features does not shown
refinements made below transition any dependence with the temperature,
temperatures, the PLT20 (Tc=343K) and meaning that below and above the phase
PLT30 (Te=223K) samples were refined as transition temperature the local order around
tetragonal. On the other hand, the titanium atoms are practically the same.
refinements made above the phase None of the samples have a pre-edge
transition temperature indicates that the similar to the undistorted EuTi03 compound
samples presents a cubic structure. Table I which structure presents a regular Ti06
shows the XRD parameters obtained atter ·octahedra. Thus, these results suggest that
the refinement. Thus, XRD results clearly the local distortions have different
shown a tetragonal to cubic phase'transition temperature dependence from that
above their respective ferroelectric- observed from the crystallographic structure.
paraelectric phase transition temperature, Although at the temperatures above the
Te. phase transition temperature of PLT20 and
Figure 3 shows the XANES spectra of PLT30 samples the structure is
PLT20 and PLT30 samples obtained at Ti K- crystallographically cubic, the local

138
distortions that determine the pre-edge a cubic structure with a regular Ti06
structure must persist into the high- octahedra, XAS measurements shows for
temperature phase. both samples that the local distortions
Figure 4 shows the EXAFS spectra of persist, even at very high temperatures
PLT20 and PLT30 obtained at the above the phase transition temperature. The
temperatures below and above the phase apparent contradiction between XAS and
transition. For the PLT20 sample, the XRD results are fact that XAS probe short-
EXAFS spectra were collected at 293 and range order while X-ray diffraction data
433K whereas for the PLT30 sample, they provide information about the average
are collected at 293 and 193K. structure.
As can be observed on Figure 4, for
both sarnples, the EXAFS spectra collected IV. Conclusions
above and below the phase transition According to XRD-Rietveld
temperature are quite similar. This result refinements made below and above the
confirms the analysis of the XANES spectra: phase transition temperatures, the PLT20
the local order around Ti atoms is not (Te=70°C) and PLT30 (Te=-50°C) samples
significantly affected by the temperature. were refined as tetragonal and cubic
Figure 5 shows the Fourier Transform respectively. Thus, XRD results clearly
curves obtained from the EXAFS spectra shown a tetragonal to cubic phase transition
presented on Figure 4. As can be seen, for above their respective ferroelectric-
the PLT20 sample, it can be observed a paraelectric phase transition temperature,
smaller increasing on the intensity of the Te.
Fourier transform first peak that ís related to On the other hand, XAS
the first Ti-O coordination shell. Although a measurement in both samples shows that
. decreasing in the intensity of the Fourier the local order around Ti atoms below and
transform ís expected when we increase the above the phase transition temperatures are
temperature due to the Debye-Waller factor, practicatly the same. XAS results suggest
in this case, we did not observe this effect then that the local structure about titanium
probably due to the partial reductfon of the site is only weakly dependent on
disorder caused by the structural phase crystallographic phase. In conclusion, in
transition from tetragonal to cubic. PLT sarnples, although the displacive nature
Thus, although XRD measurements of phase transition cannot be ignored, these
clearly indicates that at temperatures above
the phase transition both samples presents

139

I
t
I
results are consistent with the order-disorder [8] N. Sicran, B. Ravel, Y. Yacoby, E.A
character of its ferroelectric transition. .Stem, F. Dogan, J.J. Rehr, Phys, Rev. B 50,
13168 (1994)
Acknowledgments [9] 8. Ravel, N. Sicron, Y. Yacoby, E.A
Research partially performed at LNLS - Stern, F. Dogan, J.J. Rehr, Ferroelectrics
National Laboratory of Synchrotran Light, 164,265 (1995).
Brazil. We are also grateful to FAPESP [11] E.A. Stem, Y. Yacoby, J. Phys. Chem.
(00/09722-9), CNPq (Brazil) and USP- Solids 57(10),1449 (1996).
COFECUB for their financial support. One of [12] T. Miyanaga, D. Diop, S.1. Ikeda, H.
us, ACD, thanks FAPESP (01/06993-4) for a Kon, Ferroelectrics 274, 41 (2002).
postdoctoral fellowship. YPM, VRM, JAE [13] V. Shuvaeva, Y. Azuma, K. Yagi, H.
also thanks CNPq for a researcher Terauchi, R. Vedrinski, V. Komorov, H.
fellowship. Kasatani, Phys. Rev. B 62(5),2969 (2000).
[14] J.F. Meng, G.T. Zou, J.P. Li, Q.L. Cui,

References X.H. Wang, Z.C. Wang, M.Y. Zhao, Solid

State Communications 90 (10): 643-645,
(1] O. Hanske-Petitpierre, Y. Yacoby, J.
Mustre de Leon, E.A. Stem, J.J. Rehr, Phys (1994).
[15] L.A. Bassora, J.A.Eiras, Ferroelectrics
Rev. B 44(13) 6700 (1991).
[2] N. Sicron, 8. Ravel, Y. Yacoby, E.A. 223 (1-4): 285-292 (1999).
[16] Person P. Neves, Antonio C.
Stem, F. Dogan, J.J. Rehr, Phys. Rev. B
Doriguetto,Valmor R. Mastelaro, Luiz P.
50(18),13168 (1994).
[3] M.E. Lines, AM. Glass,: Principies and Lopes, Yvonne P. Mascarenhas, Alain
Michalowicz, José A. Eiras, Submitted to
applications of ferraelectrics and related
Journal of Physical Chemistry B (2003) .
.materials. Oxford: Clarendon Press (1977).
[4] R. Comes, M. Lambert, A. Guiniér.; Acta [17] A. Michalowicz, J. Phys. IV 1997, 7, C2-

Cryst. A 26, 244 (1970). 235.

. [18] Report on the Intemational Workshops
[5] Y. Yacoby, Z. Phys. B 31,275 (1978).
[6] S.A. Mabud, AM. Glazer, Acta Cryst. 12, on Standarts and Criteria in XAFS, In X-Ray

49-53 (1979). Absorption Fine Structure: Proceedings of the

[7] R.J. Nelmes, R.O. Pilitz, W.F. Kuhs, Z. VI lnternational Conference on X-Ray

Tun, R. Restori, Ferraelectrics 108, 165-170 Absorption Fine Structures, ed. S.S. Hasnain.

(1990). Ellis Horwood, New York, 1991, 752.

[19] A. Michalowicz, V. Noinville. In Gallad
2.0 cade, LURE, Orsay, France 1992.

140
(20] H.M. Rietveld, Journel ot Applied Figure Captions
Crystallagraphy 1969, 2, 65.
(21] A.C. Larson, R.B. Von Dreele, GSAS. Figure 1- XRD patterns of PL T20 sample
1996 Generalized Structure Analysis
obtained at different temperatures.
System, Los Alamos National Laboratory.
[221 B. Ravel, B.; E.A. Stern, Physica B 1995,
Figure 2- XRD patterns of PL T30 sample
208-209,316. '.
obtained at different temperatures.
[23] B. Ravel, E.A. Stern, R.1. Vedrinskii, V.
Kraizman, V. Ferroelecfrics 1998, 206-207,
407. Figure 3- Ti K-edge normalized XANES
spectra of PLT20 and PLT 30 powder
[24] R.V. Veddrinskii, V.L. Kraizman, A.A. samples at two different temperatures.
Novakovich, Ph. V. Oemekhin, S.V. Urazhdin,
J. Phys.: Condensed Matter 1998,10,9561.
Figure 4 Experimental KX(k) versus k EXAFS
spectra of PLT20 and PLT30 samples.

Figure 5- Fourier transform moduJus of Ti K

EXAFS spectra of PLT20 and PL T30
samples.

141
 PLT20

Pm3m

65 70 75 80 85 90
Degree 26 I CuKa

Figure 1

PLT30 Sample

183K

~ ~ 00 ~ 00 ~ ro ~ 00 ~
29

Figure 2·

142
 PLT20 -- 293K
433K PLT30 --293K
o 173K
:;j
.!!!..
~
c
~
.Q
c
ã. o
~
(;
~ I/l
~ ~
"O
Q) "O
A
.~ .~
m
E
(;
\ m
Eo
z z
4960 , 4980 5000 5020 5040 4950 4965 4980 4995 5010 5025 5040

Energy(eV) Energy(eV)

Figure 3

0,6
--PLT30293K
--PLT2D 293K 0,6
--PLT30173K
--PLT2D433K
0,4
0,4

0,2
0,2
oL e
5' 0,0 a 0,0

-<),2
-<),2

-<),4

-<),4
2 3 4 5 6 7 8 9 10 11 2 3 4 5 6 7 8 9 10 11

k(A')
•• k(A")

Figure 4

143
 6
~- PL T30293K
-~ PLT20 293K __ PLT30193K
5 __ PLT20433K 5

2 3 5 6 2 3 4 5 6

R(A) R (A)

Figure 5

144
I

Capitulo V

Considerações Finais

145
Considerações Finais

Acredito que os trabalhos reunidos neste texto dão uma idéia da potencialidade
do uso da espectroscopia de absorção de raios-X como uma ferramenta de
caracterização estrutural de materiais, sejam eles amorfos ou cristalinos .
. Na maioria dos trabalhos os quais desenvolvemos, o uso de mais de uma
técnica de caracterização se mostrou necessário a fim de obtermos uma visão mais
ampla da estrutura do material e sua evolução durante os diferentes processos por
nós analisados. A complementaridade das diferentes técnicas utilizadas ficou muito
clara no desenvolvimento destes trabalhos.
O Inicio do funcionamento do Laboratório Nacional de Luz Sincrotron em
Campinas a partir de 1997 foi sem dúvida um marco importante no desenvolvimento
de nosso trabalho de Pesquisa. A partir desta data, minha contribuição na
disseminação da técnica de XAS no Brasil se deu pela participação nos projetos
envolvendo a construção e o desenvolvimento da linha de XAS no LNLS, na
realização de um certo número de atividades envolvendo a formação de alunos e
pesquisadores no uso da técnica de XAS e uma divulgação da técnica através de
cursos, colóquios e seminários.
Hoje, alguns dos alunos com os quais tive o prazer de trabalhar e onde a
técnica de XAS foi utilizada no desenvolvimento de seus estudos, se .tornararn
usuários independentes do LNLS e já estão participando da formação de outros
pesquisadores e alunos.
Com relação aos trabalhos que desenvolvemos durante estes últimos anos e
que em parte estão descritos neste texto, acredito que trouxeram uma significativa
contribuição aos temas por nós abordados.
A evolução das montagens experimentais e dos programas de análise e de
calculo teórico dos espectros XANES e EXAFS possibilitou a realização de estudos de
problemas dinâmicos como por exemplo o estudo das transições de fase. É nosso
desejo nestes próximos anos é de aprofundar nossos estudos nesta direção, seja
através da realização de novos experimentos, seja através da utilização dos novos
métodos de análise. A implantação de uma escola sobre a técnica de XAS a ser
realizada anualmente está também entre nossos objetivos futuros.

146