Abstract-We studied three sodium phosphate glasses (1 - x)NaPO, + Fe,O, (x = 0.05, 0.1, 0.15) pre- pared in air in the same conditions using âFe Mossbatter ...
1. Phys. Chem. Solids Vol. 56, No. 6, pp. 877-881, 1995 Ekvier Science Ltd Printed in Great Britain 0022-3697/95 $9.50 + 0.00
MijSSBAUER
SPECTROSCOPIC INVESTIGATION IN SODIUM PHOSPHATE GLASSES
G. CONCAS,t
F. CONGIU,?
C. MUNTONIt
OF IRON
and G. PINNAt
tDipartimento di Scienze Fisiche, Universita di Cagliari, Via Ospedale 72, I-09124 Cagliari, Italy SDipartimento di Scienze Chimiche, Universita di Cagliari, Via Ospedale 72, I-09124 Cagliari, Italy (Received 25 October 1994; accepted 7 December 1994)
Abstract-We studied three sodium phosphate glasses (1 - x)NaPO, + Fe,O, (x = 0.05,0.1, 0.15) prepared in air in the same conditions using “Fe Mossbatter spectroscopy in order to investigate the effect of a growing Fe,O, addition on the features of iron sites. High spin ferric and ferrous iron in octahedral coordination is present in all the samples. We found that the distortion, compared to a regular octahedron, of the ferric and ferrous iron sites increases from the low to the high iron content samples. Keywords: A. amorphous
materials, A. glasses, A. oxides, C. Mossbauer spectroscopy, D. Mossbauer
effect.
1. INTRODUCTION The practical
applications
of phosphate
glasses were
owing to their low chemical durability. Some years ago it was discovered that the addition of iron oxide in lead or sodium metaphosphate results in a hard, stable and chemically durable glass [l-S]. Recently, iron doped phosphate glasses have been studied using “Fe Mossbauer spectroscopy [l, 6111, but the available data are often hardly comparable, owing to differences especially in preparation techniques. In these glasses, iron has been found in two states of oxidation, Fe’+ and Fe3+. The ratio FeZ+/Fe3+ is related to the particular atmosphere in which the samples have been prepared (air or inert gases) and to the preparation time [6]. It has been proposed by some authors [l, 6, 12, 131 that ferric irons can occupy two sites, but the features of these sites are still controversial. In order to clarify the characteristics of iron sites in sodium phosphate glasses, we studied three glasses with increasing iron content (1 - x)NaPO, + Fe,O, (x = 0.05,0.1,0.15) using Miissbauer spectroscopy. The experimental data have been fitted with a computer program that accounts for the proper disorder of a glass structure, and computes a probability distribution of the Mossbauer parameters. primarily
limited
2. EXPERIMENTAL The samples were prepared using NaH,PH, . and Fe,O, materials. These compounds were melted in air, at 1100°C for 1 h and
poured into stainless steel moulds cooled by cold water [5]. They were then annealed at 300°C for 1 h. All samples were subjected to chemical analysis and X-ray diffraction, as described elsewhere [S], and found to be in the amorphous state. The three samples we examined contain different amounts of iron oxide, 5, 10 and 15 mol%, and they will therefore be indicated from now on as Fe5, Fe10 and Fe15, respectively. composition of the samples as obtained by chemical analysis. The spectra were obtained in a standard transmission geometry, using a source of s7Co in rhodium (37 MBq). Calibration
on powder 40 mg/cm2. The
samples
with
absorber
temperature thickness
spectra have been tentatively
Lorentzian line shapes, but the quality of the fit was very poor. These trial fits showed that it was necessary to fit the spectra using a method proposed by Hesse and Rubartsch [14] and improved by Wivel and Msrup [15]. Amorphous materials do not show simple line shapes, owing to of bond lengths of iron atoms in the glass network. This method makes use of a set of doublets with fixed width and and computes the contribution of each curve to the spectra by a least square fitting allowing for the empirical linear 877
G. CONCAS
878
et al.
Table 1. Density, nominal and experimental compositions (mol%) of the glasses-investigated d Na, 0 Sample (g/cm3) nom. exp.
Fe5 Fe10 Fe15
2.62(l) 2.76(l) 2.84(l)
P@,
nom.
exp.
47.5 48.7(2) 47.5 46.9(l) 45.0 47.3(2) 45.0 43.5 (1) 42.5 44.3(2) 42.5 42.6(l)
Fez03
nom.
exp.
5 10 15
4.4(l) 9.2 (1) 13.1 (1)
W I
I
I
I
I
I
I
I
I
I
I
(a) Fe5
I
I
I
-
I
I
I
relation between isomer shift 6 and quadrupole splitting A of each doublet [14, 151
S={+qA.
(1)
(b)
-
a,
The program gives the best fit values of the free parameters r and r~ and the probability distribution of the doublets versus quadrupole splitting. Following Wivel and Mnrrup [15] minimisation is performed using the Lagrangean multipliers y and p’s, which control the smoothing and the end point behaviour of the distribution profiles, respectively.
8-
Fe10
-
(cl
-
J
z
3. RESULTS
The absorption transmission spectra for the three samples we examined are qualitatively similar. In Fig. 1 we show the spectra for Fe5, Fe10 and Fe15. Maximum resonant absorption relative to background changes from 3.5% for Fe5 to 14% for Fe15. The absolute errors on the absorption data are 0.06, 0.05 and 0.08 per cent for Fe5, Fe10 and Fel5, respectively. The spectra can be interpreted as the superposition of two components; the more intense can be assigned to ferric iron and the less intense to ferrous iron. We used a set of 50 Lorentzian doublets, 25 to fit the contribution to the absorption spectra due to Fe’+ and 25 for Fe2+. The Lorentzian half width has been fixed to 0.12 mm s-i, typical of a spectrum of thin cc-iron foil [15]. For the three samples quadrupole splitting ranges from 0 to 2 mm s-i for Fe3+ and from 0 to 4 mm s-i for Fe2+. Some trial fits led to the use of the values y = 1 for Fe3+, y = 10 for Fe’+ and b’s = 10 for all fits. In Table 2 we present the results of the fit of the absorption spectra. The fit parameters (r and q) of experimental data and the reduced chi-square (x2) are shown; all chi-square values are about one. The mean value of quadrupole splitting (A), the mean value of isomer shift (6) and the standard deviation (Us) of the probability distribution of the splitting are shown. The isomer shift values are referred to a-iron and computed by means of eqn (1). The contribution of each valence state to the total absorption area is also shown. The errors are equal to one unit on the last digit.
8
12
Fe15
tm
II
-4
10
18
II
-2 Velocity
I
0
I
II
I * I
2
_
I
181 4
(mm/s)
Fig. I. Transmission absorption spectrum of the Fe5 (a), Fe10 (b) and Fe15 (c) samples. The experimental data are reported as dots. The full line shows the fit.
In Fig. 2(a) we show the probability distribution of the 25 Lorentzian doublets versus quadrupole splitting for Fe 3+ in Fe5, which gives the contribution of each doublet to the total resonant absorption. This distribution exhibits a sharp peak with a small shoulder, and can be fitted by two Gaussian curves with very different amplitudes: this Gaussian shape has already been observed in phosphate glasses [6]. For high values of quadrupole splitting, small oscillations due to the numerical calculation procedure can be observed [6, 15, 161. The probability distribution of the 25 Lorentzian doublets versus quadrupole splitting for Fe2+ is shown in Fig. 2(b). Also this distribution shows two Gaussian contributions, one much more intense than the other.
Investigation
of iron in sodium
phosphate
glasses
879
Table 2. Mossbauer parameters as obtained by fitting the absorption spectra. Table shows the free parameters (/s
@y
0.26 0.40 0.56 0.35 0.22 0.31 0.50 0.70 0.22 0.34 0.44 0.58
76 I5 7 2 48 39 7 6 42 45 7 6
regular octahedron [20]. In this case, therefore, the addition of Fe,O, increases the distortion of the ferric ion site. The quadrupole splitting of Fe’+ decreases from 2.23 mm s-’ in Fe5 to 2.06 mm s-’ in Fe15 The high spin ferrous ion has a non spherical electron shell and, in octahedral co-ordination, it has been found associated to a quadrupole splitting that decreases on rising distortion of the site [21]. Therefore in our samples the addition of iron also causes an increase in the distortion of the ferrous site. The probability distributions of the quadrupole splitting show the presence of more than one Gaussian component, as can be observed from Figs 24. This composite Gaussian shape has already been observed in phosphate glasses, and each Gaussian component has been interpreted as the contribution of a particular iron site to the distribution [6]. The position of the maximum of the Gaussian curves A,,, represents the most probable value of the quadrupole splitting of the site. The half width I-o is strictly related to the variation of the distortion of the octahedron among the units in the glass, so we can state that I-o is a measure of the level of disorder of the iron site [14, 151. The presence of two ferric iron sites in our samples is supported by the fact that the preliminary trial fit by Lorentzian line shapes of the absorption spectra gets better if two doublets, instead of one, are used to fit the ferric contribution. Following this interpretation, we can state that in Fe5 the ferric iron occupies mainly one site, called I in Table 2, while a small fraction is in another site, called III, with very different values of A,,, and To. This additional site III is much more distorted than site I, as shown by the value of the quadrupole splitting. In Fe10 and Fe15 the Fe3+ ions are distributed in two sites, characterised by markedly different values of quadrupole splitting. The first of them, called I, has about the same Mossbauer parameters of site I of Fe5 and can be identified with it. The second one, named site II, presents similar Mijssbauer parameters in Fe10 and Fe15 but it is very different from site III in Fe.5. Site II is more distorted and more disordered that site I, as indicated by the values of Amaxand l-o. These two sites are occupied by a comparable number of ferric iron atoms. The ferrous iron shows a similar behaviour. Also in this case three non-equivalent sites, named site I, II and III, can be identified by their different Mossbauer parameters. Site I is present in all the three samples; in Fe5 the majority of iron atoms occupy this site, and only a small fraction can be found in site III. In Fe10 and Fe15 the ferrous atoms are distributed between site I and site II with similar
Investigation
weights.
The quadrupole
site III is very distorted. that
splitting
values show that
A,,, and To values indicate
site II is characterised
distortion
of iron in sodium phosphate glasses
and disorder
by a higher
level of
than site I.
5. CONCLUSIONS We found in all samples octahedral Fe*+/Fe’+
ferric and ferrous iron in
co-ordination. ratio is about
In
all
samples
the
l/9. The level of distortion
compared to a regular octahedron of the ferric iron environment increases with the iron content. In fact, while in Fe5 almost all the ferric atoms are in one site (site I), another in Fe10 disordered
and
highly occupied Fe15.
site (site II) is present
Site II is more
distorted
and
that site I, so this fact causes the increase
of the mean distortion Fe5. Fe** exhibits
in Fe10 and Fe15 compared
to
881
4. Musinu A., Piccaluga G. and Pinna G. J. Non-Crysf. Solidr 122, 52 (1990). 5. Medda M. P., Piccaluga G. and Pinna G. Proc. 4th Meet. Synthesis and Methodologies in Inorganic Chem istry, Biessanone, Italy (1993).6. Brooks J. S., Williams G. L., Allen D. W. and de Grave E., Phys. Chem. Glasses 33, 167 (1992). 7. Brooks J. S., Williams G. L. and Allen D. W., Phys. Chem. Glasses 33, 171 (1992). 8. Kurkjian C. R. and Sig&y E. A., Phy.r. Chem. Glasses 9, 73 (1968). 9. Lewis Jr G. K. and Drickamer H. G., J. Chem. Phys.
49, 3785 (1968). 10. Dyar M. D., Am. Miner. 70, 304 (1985). II. Wong J. and Angel1 C. A., Glass-Structure by Spectroscopy, p. 97. Marcell Dekker, New York (1976). 12. Armelao L., Bettinelli M., Rizzi G. A. and Russo U.. J. Mater. Chem. 1, 805 (1991). 13. Music S., Gotic M., Popovic S. and Grzeta B., J. Radianal. Nucl. Chem. 116, 141 (1987). 14. Hesse J. and Rubartsch A., J. Phys. E 7, 526 (1974). 15. Viwel C. and Msrup S., J. Phys. E 14, 605 (198 1). 16. Le Caer G. and Dubois J. M., J. Phys. E 12, 1083
a similar behaviour.
(1979). 17. Gutlich P., Link R. and Trautwein A., Mtissbauer Spectroscopy and Transition Metal Chemistry, p. 56.
REFERENCES
Springer-Verlag. Berlin (1978). 18. Dyar M. D., J. Am. Ceram. Sot. 69, C-160 (1986). 19. Coey J. M. D., J. Physique C-6, 89 (1974). 20. Dyar M. D., Am. Miner. 69, 1127 (1984). 21. Ingalls R., Van Der Woude F. and Sawatzky G. A., Mtissbauer Isomer Shifrs (Edited by G. K. Shenoy and F. E. Wagner), p. 361. North-Holland, Amsterdam (1978).
1. Sales B. C., Abraham M. M., Bates J. B. and Boatner L. A., J. Non-CrJ>st. Solids 71, 103 (1985). 2. Sales B. C., Ramsey R. S., Bates J. B. and Boatner L. A., J. Non-Cryst. Solids 87, 137 (1986). 3. Sales B. C. and Boatner L. A., J. Non-Cryst. Solids 79, 83 (1986).
