Conversion electron mössbauer analysis of glass surfaces

zyxwvu zyx zyxw zyxwv zyxw zy zyxw zyxwvu J. SAWICKI et al. : Conversion Electron Mossbauer Analysis of Glass 173 phys. stat. sol. (a) 41, 173 (1977) Subject classification: 5; 2; 22.8.1 Institute of Physics, Jagiellonian University, Cracow (a), Institute of Nuclear Physics, Cracow ( b ) , and Institute of Physics, Technical University, Gdarisk (c) Conversion Electron Mossbauer Analysis of Glass Surfaces BY J. SAWICKI (a), B. SAWICKA (b), and 0. GZOWSKI(c) Mossbauer spectra of 57Fe in iron-phosphate glass are studied both in transmission (powder samples) and conversion electron reemission mode (pellets). Octahedrally coordinated ferrous and is found to be strongly reduced near the surface ferric ions are determined. The ratio FeZ+/Fe3+ of air-heated pellets, and reestablishes to the value about 0.4 after heat treatment of the samples in vacuum. Es werden MoBbauerspektren von 57Fein Eisenphosphatglasern sowohl in Transmission (Pulverproben) als auch mit der Konversionselektronen-Reemissionsmode(Pellets) untersucht. Oktaedrisch koordinierte Ferro- und Ferriionen werden bestimmt. Es wird gefunden, daB sich das Verhaltnis Fe2+/FeS+in der Nahe der Oberflliche der an Luft erhitzten Pellets stark verringert, wobei nach Temperung im Vakuum die Proben wieder den Wert von etwa 014 annehmen. 1. Introduction Mossbauer spectroscopy has been used t o identify the oxidation state and location of iron in glasses [l]. The behaviour of iron in oxide glasses has been discussed by Kurkjian [ 2 ] , and the studies of semiconducting glasses were reviewed by Taneja et al. [3]. Usually transmission-type Mossbauer experiments for powder samples have been performed ; therefore, the average properties of iron in glass were only studied. However, the phenomena associated with the superficial layers of the glass samples (e.g. light reflection, conductivity, etc.) are of considerable interest. I n the present work, in addition t o the transmission Mossbauer experiments, the Mossbauer conversion electron reemission method was applied to glass surface studies. 2. Experimental Method The principles of the conversion electron Mossbauer spectroscopy and its applications t o 67Penuclei can be found, e.g., in [4 to 71. I n the method electrons instead of y-rays are resonant sensors. After the resonant Mossbauer absorption of 14.4 keV y-rays 57Fe nuclei deexcitate with the emission of 7.3 keV conversion electrons or 5.6 keV Auger electrons. As the penetration of so low-energy electrons in solids is about 2000 to 3000 A, the reemission and thereby the Mossbauer spectrum is observed only for iron atoms within the superficial layer of this thickness. It makes possible to study the surface properties of bulk samples. The electron counter has been modified to accomodate in i t pellets of = 2 cm diameter and 0.4 t o 0.5 cm thick. The counter was filled with a 90% He-10% CH, gas mixture and operated in the proportionality range. An anode voltage around 1000 V was applied. The count rate was x 200 counts/s. Other details of the electron counting technique are presented elsewhere [5, 7 , 81 . The Mossbauer spectra were measured with the constant acceleration spectrometer working in the time mode. A W o source in Pd host of x 10 mCi was used. The velocity zyxw zyxwvutsr zyxwvuts zyxwvu zyxwvuts zyxw J. SAWICKI, B. SAWICKA, and 0. GZOWSKI 174 scale of the spectrometer was calibrated in respect to a metallic iron R'BS standard foil. Lorentzian line fits with the least-squares method were performed for all spectra measured. The transmission spectra of 14.4 keV y-rays were measured for powdered samples of =r 45 t o 50 mg/cm2 ( sz 10 mg/cm2 Fe). The grain size of the powder was about 0.1 t o 0.3 mm. The measurements a t liquid nitrogen temperature were carried out in EL vacuum dewar with thin mica windows. The glass samples were prepared by the standard method from the oxide mixture. The basic glass producing substance was P205having 50 mol% concentration. The remaining part consisted of iron oxides FeO, Fe20, (40mol%) and a network modifier MgO (10 mole/,). Because of the large hygroscopicity of P20,,diammonium phosphate (NH,)HPO, was used as the starting material. Melting was done in air a t a temperature of 1300 "C. After 2 h of melting the glass was poured onto a metal plate heated to a temperature of 250 "C. I n order t o eliminate internal stresses the obtained glass samples were annealed a t a temperature of 300 "C for several hours. The amorphous structure of all glass samples was shown b y X-ray analysis. The Mossbauer spectra were measured for three different kinds of samples (treated after preparation in different ways): 1. fresh (only annealed), 2. heated in air (1 h a t 560 "C), 3. heated in vacuum Torr, 1h a t 560 "C). 3. Results and Discussion Examples of the transmission Mossbauer spectra are shown in Fig. 1 and 2. The spectra of powders heated in air (Fig. l a ) or in vacuum (Fig. l a ) are identical. The spectra can be decomposed on three lines (dashed crosses indicate relevant amplitudes and linewidths) or two doublets (indicated by solid crosses). According t o the general interpretation of the Mossbauer spectra in iron containing glasses the doublets correspond to iron in the ferric state Fe3+ and the ferrous state Fez+, respectively. The values of the isomer shift IS (central shift of the doublet) and the quadrupole splitting QS (distance between lines in the doublet) assign valency and co-ordination number of iron ions. The following values were obtained in our room temperature measurements: for the Fe3+doublet IS = 0.4 mm/s and QS = 0.8 mm/s, for the Fe2+ doublet IS = 1.2 mmjs andQS =: 2.1 mm/s. b - zyxwvutsr zyxwvu zyxwvu &&rnrn/s~ I 1 I I I Fig. 2. Transmission spectrum at liquid nitrogen temperature: glass powder heated in vacuum 7 2 3 ve/oci& (mmis) ---+ Fig. 1. Transmission Mossbauer spectra measured at room temperature: a) glass powder heated in air; b) glass powder heated in vacuum - 2 - ? 0 zyxwv zyx 175 Conversion Electron Mossbauer Analysis of Glass Surfaces Data a t 80 K showed no significant difference (Fig. 2) in comparison with the room temperature spectra. The larger positive central shift observed for both doublets is caused by the relativistic second-order Doppler effect (temperature shift). The magnitude of the shift between room and liquid nitrogen temperature is typical for iron. Also, as usually for Fe3+in the high-spin state, the quadrupole splitting is almost constant, while for the high-spin Fe2+i t increases a t lower temperatures. I n the case of Fe3+ the quadrupole splitting is associated mainly with the electric field gradient of neighbouring ions. For Fe2+the electric field gradient of its own 3d electron shell is much stronger and temperature-dependent, The measurable broadening of Fe2+!ines and the higher intensity of the Feat doublet area a t liquid nitrogen temperature can indicate some kind of electron transfer process between Fea+ and Fes+ ions. Magnetic hyperfine structure observed in iron phosphate glasses earlier “31, and possible short-range magnetic ordering, were not seen in our samples. The studies a t liquid helium temperature, being in progress now, can elucidate this problem. The conversion electron Mossbauer spectra of pellet samples are presented in Fig. 3. The spectrum of the fresh sample (Fig. 3a) shows the Fe3+ quadrupole doublet only. The same spectrum was also observed for the sample heated in air. The sample heated in vacuum shows a superposition of two quadrupoles (Fig. 3b) similar t o those observed in all transmission spectra taken for powder samples (i.e. from both Fe3+ and Fe2+). The line positions, linewidths, and line amplitudes obtained by the Lorentzian line least-squares fitting for all spectra measured are collected in Fig. 4. It is seen that the positions of lines for all room temperature spectra, as well as linewidths, are constant in the error limits. The linewidths being ahout two times larger than the natural width indicate a highly random structure of the glass studied. The IS and QS data measured by us fit very well to the correlation between IS and QS established for various glasses [ 1, 21, presented in Fig. 5. Our data are indicated by black points. On this basis highly ionic Fe2+and Fe3+, both in octahedrally coordinated positions, can be unambiguously assigned. Therefore, iron ions are modifiers in the glass investigated. It seems very interesting to note that in conversion electron Mossbauer spectra of fresh and air-heated pellets the Fe2+doublet was not observed. Therefore, in these samples only Fe3+ions are present in depth about 2000 to 3000 A (or even deeper), while the whole sample contains the usual amount of Fe3+ and Fez+, as it is seen in transmission spectra. Vacuum heating recovers the Fe2+state and the Fe2+/Fe3+ratio of about 0.4 typical for the bulk state is observed also in the surface layer. Very natural is the assumption that the outer layer of fresh or air-heated samples is oxidized. We mention that the surface oxidation cannot a t all be observed by transmission Mossbauer spectroscopy, as the relative volume of the superficial layer in the glass zy zyxwv zyx zyxwvu zyx zy ... Fig. 3. Conversion electron Mossbauer spectra: a) fresh glass pellet; b) glass pellet heated in vacuum . I -2 I -7 . I I I I 0 7 2 3 velociij.(mmisi ---+ 176 zy zy zyxwvutsrq zyxwvutsrq J. SAWICKIet al.: Conversion Electron Mossbauer Analysis of Glass Fhi- I +a zyxwv zyxwvutsrq Fig. 4. Collection of the data: line positions, widths, and amplitudes for the samples studied. Error limits in positions are indicated. (a)Fe2+/Fe3+ = 0.376 & 0.024, (b) 0.350 & 0.033, (c) 0.535k & 0.126, (d) 0.509 & 0.084, ( e ) 0.369 & 0.167, (f) 0.413 & 0.159 Fig. 5 . Correlation between isomer shifts (IS) and quadrupole splitting5 (QS) for various glasses. The data obtained in this work are represented by black points, other data taken from [l]. Isomer shifts are given relative to a metallic iron source. x Alkali silicates; alkaline - earth silicates; A alkali borates; o phosphates; 0 our data; + other glasses powder is very small. It is also to be noticed that the electronic states of Fe3+and Xe2f observed t o a depth of 2000 to 3000 A are the same as in the bulk sample. It is to be emphasized that the conversion electron S7Fe Mossbauer spectroscopy proved to be a very useful technique for studying the chemical state of iron close to the glass surface. As the state of iron sensitively depends on the glass preparation conditions (oxidizing or reducing atmosphere) and on the heat treatment, the glass technological processes can be controlled and possibly improved with the help of the method of analysis presented above. Acknowledgements The assistance of Mrs. K. Rusnak, Dr. J. Stanek, and Mr. J. Kowalski in the experiment is kindly acknowledged. References zyx zyx [l] J. M. D. COEY, J. Physique (Suppl. 12) 35, C6-89 (1974). [2] C. R. KURKJIAN, J. non-crystall. Solids 3, 157 (1970). [3] S. P. TANEJA,C. W. KIMBALL,and J. C. SCHAFFER, in: Mossbauer Effect Methodology, Vol. 8, Ed. I. J. GRUVERMAN, Plenum Press, New York 1973 (p. 41). [4] K. R. SWANSON and J. J. SPIJKERMAN, J. appl. Phys. 41, 3155 (1970). [5] B. SAWICKA, J. SAWICKI,and J. STANEK, Nukleonika (Warszawa) '21, 949 (1976). [6] M. J. TRICKER,to be published. [7] J. STANEK, J. SAWICKI,and B. SAWICKA, Nuclear Instrum. and Methods 130,613 (1975). and J. STANEK,Nuclear Instrum. and Methods 138, 565 (1976). [S] J. SAWICKI,B. SAWICKA, [9] J. G. VAUOHAN,L. K. WILSON,and D. L. KINSER,Amer. Ceram. SOC.Bull. 62, 384 (1973). (Received February 8, 1977)