Ga. Waychunas et al., SHORT-RANGE ORDER MEASUREMENTS IN MGO-FEO AND MGO-LIFEO2 SOLID-SOLUTIONS BY DLS SIMULATION-ASSISTED EXAFS ANALYSIS, The American mineralogist, 79(3-4), 1994, pp. 274-288
EXAFS analysis is here used to characterize short-range order (SRO) an
d local structure in oxide solid solutions by direct measurement of th
e identity and number of next nearest neighbor (NNN) cations. Precisio
n is improved through the use of constraints, including simulation-der
ived interatomic distances. EXAFS Fe-K edge measurements were made on
seven MgO-FeO solid solution samples rapidly quenched from 1140-degree
s-C. The number of Fe and Mg cations in the NNN shell about an average
Fe atom were found to deviate from a random distribution by only 1.4%
of total NNN atoms, on average. EXAFS-derived first shell distances a
gree with simulation results. Fourier deconvolution of Mossbauer spect
ra of these samples resolves the fine structure, which further support
s a random cation distribution. Computer simulations of alpha-LiFeO2 a
nd solid solutions of alpha-LiFeO2 and MgO were carried out for random
distributions and for locally ordered distributions in which local ch
arge-balancing cation interchange (simulated diffusion) was permitted.
Such interchanges decrease Li-Li and Fe-Fe NNN contacts relative to t
he number of Li-Fe contacts but do not affect Mg distribution. Fe-K ed
ge EXAFS measurements of two samples of alpha-LiFeO2 prepared at 1000-
degrees-C gave 4.6 and 4.8 NNN Fe atoms, in agreement with the 4.89 pr
edicted by a locally ordered simulation. Precise fitting of EXAFS spec
tra of 12 MgO-alpha-LiFeO2 solid solution samples (also quenched from
1000-degrees-C) required the additional assumption of no Mg atom clust
ering. EXAFS measurement of the remaining Li,Fe NNN atoms around probe
Fe atoms also corresponded closely (within 2.6% of total NNN atoms) t
o a random distribution. The change from locally ordered to random app
eared between 100 and 80% LiFeO2. Several samples with very low Fe con
tent gave erratic results attributable, we believe, to sample heteroge
neity.