Structural relaxation and crystallization of semiconducting vanadate glass accompanying a jump of the electrical conductivity Tetsuaki Nishida,*" Junichi Kubota," Yonezo Maeda," Fusao Ichikawab and Takafumi Aomineb aDepartmentof Chemistry, Faculty of Science, Kyushu University, Hakozaki, Higashiku, Fukuoka 812, Japan bDepartment of Physics, Faculty of Science, Kyushu University, Hakozaki, Higashiku, Fukuoka 812, Japan * 10Fe203 glasses (20 <x d30) consists of a quadrupole doublet peak. The linewidth and the quadrupole splitting decreased gradually after heat treatment at the crystallization temperature, reflecting an increased uniformity of Fe- 0 bonds and a decreased distortion of FeO, tetrahedra, respectively. FTIR spectra of heat-treated vanadate glasses revealed an evident peak separation between VO, and FeO, tetrahedra. The FeO, units were not observed in non-treated glass because they occupy substitutional sites of VO,.An X-ray diffraction study of heat-treated glasses established the formation of a KV308 phase composed of VO, pyramids. Additional diffraction peaks due to a K3V5014 glasses plot of the differential thermal analysis (DTA) data yielded activation energies of 2.0-2.9 (d0.3) eV, indicating that cleavage of Fe-0 bonds with bond energies of ca. 2.6 eV triggers the crystallization. Magnetic susceptibility measurements of 25K20* 65V,05 * 10Fe203 glass revealed an antiferromagnetic behaviour with Curie-Weiss constants of -13, -93 and -108 K after the heat treatment at 340 "C for 0,2100 and 5000 min, respectively.A jump of the electrical conductivity from 6.3 x lo-' S cm-' to 4.3 x lo-, S cm-' and this can be ascribed to an increased probability of electron hopping from V4+ to V5+ which was brought about as a result of a decreased distortion of VO,, FeO, and V05 units. Electrical conducting glasses have attracted much interest in The Mossbauer effect is a very effective tool for investigating the field of solid-state chemistry and materials science because the local structures of inorganic glasses, in which small amounts they can be used as an electrode, solid-state electrolyte, sensor, of Mossbauer ions such as Fe3+ and Sn4+ are doped as a etc. Several vanadate (V,O,-based) glasses show a semiconduc- 'probe'. Information can be obtained from the Mossbauer ions tivity with the electrical conductivity (B) of the order of about on the chemical bond and the symmetry of the oxygen lo-' to Scrn-',' which is known to be due to the polyhedra of the network former (NWF), e.g., VO,, SiO,, electron hopping from V4+ to V5+ in the 'skeleton' structure. AlO,, GaO, and BO,, because the Mossbauer ions occupy The mechanism of semiconductivity was explained by Mott2 NWF sites in several oxide gla~ses."-'~ Nishida discovered and by Austin and Mott3 with a small polaron hopping theory, three general for determining the structural role of in which the activation energy, W, was estimated to be equal Mossbauer ions in inorganic glasses. (1) The Debye tempera- to the summation of the electron-hopping energy, W', and ture, OD, obtained from low-temperature Mossbauer measure- half of the lattice disorder energy, W,.In the case of ments, is >280K when iron and tin occupy NWF sites, Sb2O3-Sr0-V2O5 glasses,' W was reported to be 0.41-0.61 eV whereas it is <270 K when they occupy network modifier and W, <0.1 eV. (NWM) sites, as is usual for alkali-metal ions. (2) y-Ray The electrical conductivity of insulating oxide glasses is irradiation of oxide glass results in the reduction of Fe3+ to generally less than lo-'' S cm-l and the alkali-metal ions are Fe2+ when Fe3+ occupies NWF sites, while oxidation of Fe2+ known to be representative 'carriers'. Crystallization of oxide to Fe3+ takes place when iron occupies NWM sites. (3) The glass generally results in a decrease in the electrical conduc- glass-transition temperature, q,of oxide glasses has a linear tivity.By contrast, if a conducting crystalline phase like relationship with the quadrupole splitting, A, of Fe3+ obtained 'vanadium bronze', R,V205, precipitates in the matrix glass, from the Mossbauer measurements. The linear relationship, the electrical conductivities of glass ceramics will become named the 'G-d rule', gives a large slope of 680 "C (mm s-')-l higher than those of non-treated glass. In the case of potassium when Fe3+ occupies tetrahedral NWF sites, while the slope vanadate glass, 25K20-65V205 10Fe,03, an insulating becomes 260 "C (mm sF1)-l when Fe3+ occupies octahedral KV30E phase4 and a K3V5014 phase' are expected to precipi- NWF sites." By contrast, the slope becomes only tate upon heat treatment, and hence it is expected that the 35 "C (mms-')-' when Fe3+ occupies NWM sites.These electrical conductivity of the glass ceramic will decrease. findings made it possible to deduce the structural role of Vanadium pentoxide, V205, is known to have a layered Mossbauer ions in oxide and non-oxide glasses, i.e. whether structure composed of VO, pyramids.6 Structural studies of NWF or NWM. several glasses have been carried out by NMR,7 IRE and FTIR spectroscopy is also a powerful tool for studying the Mossba~er~'~~ Intechniques. As a result, the 'skeleton' proved to local structure and crystallization of inorganic glasse~.~~-~~ be composed of distorted VO, tetrahedra and distorted VO, -5Fe203 glass," tetragonal pyramids.It is generally considered that the fraction for example, precipitation of a mayenite (12Ca0. 7A1,03) of distorted VO, tetrahedra will increase with increasing alkali- phase was confirmed by a step-by-step peak shift of the A1-0 metal oxide content. A 57Fe Mossbauer study of alkali-metal stretching band (v3 mode) of distorted AlO, tetrahedra from vanadate glasses containing iron revealed that the fraction of 782 to 837 cm-l. The peak position is consistent with the non-bridging oxygen (NBO) in the distorted VO, tetrahedra characteristic IR frequency range elucidated by Tarte,20 i.e. increases with increasing alkali-metal oxide c~ntent.~"~ tetrahedral A13+ in a condensed material is observed in the J.Mater. Chem., 1996,6( 12), 1889-1896 1889 700-900 cm-l range, while octahedral A13+ bands are found in the region 500-650cm-'. In glass ceramics including the mayenite phase, phase separation of distorted FeO, tetrahedra, i.e. precipitation of an iron-rich phase in the glass phase, was elucidated from the new peak which appeared at 570cm-', which was assigned to the v3 mode of FeO, tetrahedra." Since the stretching band of octahedral Fe3+ in oxide glasses is generally observed at ca. 540 cm -,15,18,20 discrimination between tetrahedral and octahedral Fe3 + becomes possible from the IR measurements. Tarte2' elucidated that tetrahedral Fe3+ in a condensed material is observed at 550-700 cm-', while octahedral Fe3+ is found at 400-550 cm-l.An 'FTIR transmission method' combined with a conventional Johnson- Mehl-Avrami (JMA) equation is effective for the kinetic study of the crystallization of IR-transmitting gla~ses.'~.~~ In the present study, however, the FTIR transmission method has not been applied because vanadate glasses in general are not optically transparent in the IR region. Differential thermal analysis (DTA) is also effective for the kinetic study of the crystallization of glasses, and a Kissinger plot2' is often used for estimating the activation energy, E,. The Kissinger equation is expressed by In (x2/a)=E,/RT, +C (1) in which T, and a are the crystallization peak temperature (in K) and the heating rate of the sample, respectively; R is the gas constant and C is a constant.From the Kissinger plots of 38Na20 61 W03 * 57Fe203 and 38Na20 62W0, gla~ses,'~ E, values of 2.6 and 4.7 (k0.3) eV, respectively, were obtained. The former value can be assigned to the Fe-0 bond energy, although it has not been reported in the literature to our knowledge. The latter value is the same as the W-0 bond energy of 4.5 eV,22 within experimental error. These results indicate that cleavage of Fe-0 and W-0 bonds trigger the crystallization of iron-containing and iron-free tungstate glasses, respectively. In the crystallization of iron tellurite glass, 95Teo2-5Fe203, the Kissinger plot yielded an E, of 2.3 (k0.3) eV, which is comparable to the E, obtained from the JMA plot in the FTIR transmission method: i.e.2.9 (k0.3) eV.17 These E, values are comparable to the Fe-0 bond energy of 2.6 eV described above or to the Te-0 bond energy of 3.0 eV.22 This study was carried out in order to elucidate the effect of heat treatment on the structure of potassium vanadate glasses containing 10 mol% Fe203. 57Fe Mossbauer effect, FTIR, DTA and X-ray diffraction (XRD) studies were used for this purpose. The changes of the electrical conductivity, 0, and of the molar magnetic susceptibility, xm, of 25K20. 65V205-10Fe203 glass were investigated at several stages of the heat treatment in order to elucidate the relationship between the structural change and the physical properties of the material. Experimental Potassium vanadate glasses, xK20 (90 -x)V205 10Fe203, were prepared by melting weighed amounts of a reagent mixture composed of K2C03,V205 and Fe203 of guaranteed reagent grade (Ishizu Pharmaceutical Co.Ltd). The reagent mixture placed in a platinum crucible was melted at 1050°C for 1h in an electric muffle furnace, and the melt was quenched with ice-cold water. Homogeneous black glass samples were prepared in the compositional range 5<x <30. 35K20* 65V205 glass was prepared similarly after melting the reagent mixture at 900 "C. 57Fe Mossbauer spectra of heat-treated glass samples were measured at room temperature with a 57Co( Pd) source of 10 mCi (3.7 x 10' Bq). A sheet of metallic iron foil (a-Fe) enriched with 57Fe was used as the reference for the isomer shift, 6, and for calibrating the velocity scale (abscissae of the spectra).The spectral analysis was made by a least- 1890 J. Muter. Chern., 1996, 6(12), 1889-1896 squares method. FTIR spectra were measured by the conven- tional KBr disk method. The values of Tg were determined by conventional DTA conducted at a heating rate of 5 "C min-l. a-A1203 powder was used for temperature calibration. The XRD pattern was recorded at a scanning rate of 2 degrees min -'with Cu-Ka X-rays. The electrical conductivity, 0,was measured at room temperature for rectangular samples (1.0 x 5.0 x 0.8 mm3) to which lead wires were attached with an indium solder. In this paper are shown the mean values of the electrical conductivities obtained by dividing the current density (A cm-2) by the electric field (V cm-l).Each current density was obtained after 15 s had passed under constant voltage, because the system consisting of the glass sample and the contacts has a capacitance. The voltage dependence of CJ was investigated by the dc two-probe method. In order to confirm that the electrical conductivity obtained by the dc two-probe method reflects the steady-state response from the glass sample, the ac four-probe method was also used with a frequency of 1.0 Hz and a current of ca. 0.6 PA. Magnetic susceptibility measurements were performed using the Faraday method in the temperature range 80-300 K with HgCo(NCS), as the standard. Results 57Fe Mossbauer spectra of 25K20*65V205 10Fe203 glasses heat-treated at 340 "C for 0 to 2100 min are shown in Fig.1. Each spectrum is comprised of a doublet with an isomer shift, 6, of 0.38+0.01 mm s-', which indicates that iron is present in a distorted tetrahedral environment, Fe3+( Td),9-14 in both the glass and glass ceramics. The doublet peak narrows markedly after prolonged heat treatment, due to increased uniformity of FeO, tetrahedra. The decrease in the linewidth (FWHM), I-', reflects an increased uniformity of Fe-0 bond lengths and 0-Fe -0 angles in the partially crystallized glasses (i.e. glass ceramics). The decrease in r is illustrated in Fig. 2 together with the other Mossbauer parameters. The quadrupole splitting, A, shows a gradual decrease with heat treatment time, indicating an increase in the symmetry of the FeO, tetrahedra, i.e.a decrease in the distortion. Mossbauer spectra of 20K20 7OV2O5 * 10Fe203 and 30K20 60V205 10Fe203 glasses heat-treated at 370 "C for 0 and 5000min are shown in Fig. 3. The Mossbauer spectrum of non-treated 20K20 7ov20, 10Fe203 glass [Fig. 3(a)] has the -4 -3 -2 -1 0 1 2 3 4 velocity/mm s-1 Fig. 1 57Fe Mossbauer spectra of 25K20-65V20,-1OFe2O3 glass heat- treated at 340 "Cfor (a) 0, (b) 10, (c)60, (d) 600 and (e) 2100 min 0.7 -0.6 E E2$ 0.5 EE 0.4 0.3 0.2 Fig. 2 Change of the Mossbauer parameters of 25K,0*65Vz0,-10FezO, glass heat-treated at 340 "C for 0 to 5000 min. 6, isomer shift with respect to metallic iron; A, quadrupole splitting; r,line-width (FWHM).-4 -3 -2 -1 0 1 2 3 4 velocity/mm s-1 Fig. 3 Mossbauer spectra of 20Kz0.70V,0, * lOFe,O, glass heat- treated at 370 "C for (a) 0 and (b) 5000 min. Massbauer spectra of 30K,0-60V,05 lOFe,O, glass heat-treated at 370 "C for (c) 0 and (d) 5000 rnin are shown for comparison. following parameters: 6 =0.40 mm s -', d =0.62 mm s -' and r=0.49 mm s-'. The 6 value reflects tetrahedral Fe3 ions.+ The Mossbauer parameters of an intense doublet observed in the sample heat-treated at 370 "C for 5000 rnin [Fig. 3(b)] are: 6=0.40 mm s-', d=0.46 mm s-' and r=0.42 mm s-'. A weak doublet (outer one) in this sample has the parameters: 6=0.34 mm s-', d= 1.11 mm s-' and r=0.26 mm s-'. 30K20-60V205-10Fe203 glass showed a homogeneous crys- tallization which is essentially the same as the 25K20* 65V205 10Fe203 glass (Fig.1). XRD patterns of 25K20*65V205-10Fe203 glass and glass ceramics are shown in Fig. 4. Some diffraction peaks due to crystalline phase(s) appeared after the heat treatment at 340 "C for 150 rnin and the peaks became more intense with increasing heat treatment time. Almost all the diffraction peaks shown in Fig. 4(d)-(f) are assigned to the KV30, phase having a K20/V205 ratio of 0.33, which is the same as the K20/( V205 +Fe203) ratio in 25K20 65V2059 10Fe203 glass. It is known that the KV308 phase is composed of distorted 15 20 25 30 35 40 45 50 55 28/deg re es Fig. 4 XRD patterns of 25K,0.65V,05 -lOFe,O, glass heat-treated at 340°C for (a) 0, (b) 150, (c) 600, (d) 1200, (e) 2100 and (f) 5000min.Diffraction peaks marked with solid circles refer to the KV,08 phase. V05 pyramids.23 In the case of 20K20 70V205 10Fe203 glass heat-treated at 370 "C for 5000 min, the XRD study revealed a few weak diffraction peaks assigned to a K4V10027phase24 [Fig. 5(a)], in addition to several peaks due to the KV308 phase observed in 25K20-65V20510Fe203 glass ceramic [Fig. 4, 5(b)]. The XRD patterns of 30K20 60V205* 10Fe203 glass ceramic [Fig. 5(c)] and 35K20 *65V205 glass ceramic [Fig. 5(d)] prepared by heat-treatment at 370 "C for 5000 rnin showed some diffraction peaks assigned to the K3V5014 phase, in addition to several peaks due to KV308. The K3V5014 phase having a K20/V205 ratio of 0.60 is known to be composed of V04 tetrahedra and V05 pyramids.25 A higher K20 content will be associated with the appearance of the K3V5014 phase in the glass ceramics 30K20-60V205 10Fe203 and 35K20.65V2O5.The formation of V04 tetrahedra in xK20*(90-x) V205 10Fe203 glasses can be deduced from the FTIR spectra shown in Fig. 6. The stretching vibrations of V-026 and V=027 bonds in V05 pyramids were observed in the regions 850-860 and 960-1000 cm-', respectively. The stretching vibration of the V-0 bond in V04 tetrahedra28 was observed at 750-780 cm-'. Fig. 6 evidently shows that the peak intensity of the V-0 stretching mode in V04 tetrahedra increases with increasing K20 content, at the expense of V-0 bands in VO, 10 15 20 25 30 35 40 45 50 55 28/degrees Fig.5 XRD patterns of (a) 20K,O*70Vz05 -10Fe,03, (b) 25K,O. 65V205 lOFe,O,, (c) 30Kz0-60Vz05 * lOFe,O, and (d) 35K20-65V20, glass samples heat-treated at 370 "C for 5000 rnin J. Mater. Chem., 1996,6( 12), 1889-1896 1891 A IIIIIII 1000 500 1000 500 v1crn-l Fig. 6 FTIR spectra of xK20.(90-x)V,O, * 1OFe2O3 glasses (a) x = 5, (b) x=10, (c) x=15, (d) x=20, (e) x=25 and (f) x=30 pyramids This means that the fraction of VO, tetrahedra increases in the 'skeleton' structure Crystallization of 25K20*65V205-10Fe20, glass can be deduced from the FTIR spectra shown in Fig 7, because the linewidth became smaller after the heat treatment and the Fe-0 stretching band of FeO, tetrahedra'* 2o appeared at 580 cm-I Note that this peak was not observed in the iron-free potassium vanadate glass, 35K,O *65V,05, even after heat treatment at 370 "C for 5000 min The appearance of the Fe-0 stretching band in the heat-treated samples is proof that some Fe3+ ions in the glass ceramics occupy site(s) other than those of the vanadium ions, although they occupy V5+or V4+ sites in non-treated vanadate glasses The DTA curve of 20K20* 7OV2O5 10Fe203 glass illustrated in Fig 8 is comprised of three exothermic peaks due to the crystallization, designated as T,(l), z(2) and T,(3) The T,(1) peak shifted from 250 "C to 257, 262 and 267 "C, and z(2) from 343 "C to 356, 369 and 374 "C, when the heating rate was increased from 2 "C min-' to 5, 10 and 15 "C min-', respect-ively Under the same conditions, T,(3) shifted from 403 "C to vFe-O(Td) d 1000 500 1000 500 v1crn-l Fig.7 FTIR spectra of 25K20-65V20,-1OFe2O3 glass heat-treated at 340°C for (a) 0, (b) 10, (c) 60, (d) 300, (e) 900, (f) 1500, (8) 3000 and (h) 5000 min 1892 J Muter Chem, 1996,6(12), 1889-1896 100 200 300 400 500 TI'C Fig. 8 DTA curves of 20K,0*70V205-10Fe20, glass recorded at heating rates of (a) 2, (b) 5, (c) 10 and (d) 15"C min ' and T, indicate the glass-transition temperature and the crystallization peak temperature, respectively 412,425 and 429 "C The Kissinger plot shown in Fig 9 yielded three straight lines for T,(l), z(2) and T,(3), with E, values of 2 9, 2 1 and 2 9 (fO 3) eV, respectively The DTA curve of 25K,0*65V,05.10Fe203glass is shown in Fig 10, in which two crystallization peaks of q(1)and q(2) were observed The former peak shifted from 260°C to 272, 280 and 284"C, and the latter from 319 "C to 332, 340 and 344"C, when the heating rate was increased from 2"Cmin-' to 5, 10 and 15"C min-l, respectively The Kissinger plot shown in Fig 11 yielded two straight lines for z(1) and T,(2), from which E, values of 2 0 and 2 4 (+O 3) eV, respectively, were obtained The DTA curve of 30K,0.60V205-10Fe203glass shown in Fig 12 is comprised of two crystallization peaks of T,(1) and c(2) The former peak was observed at 283, 294, 300 and 304"C, and the latter at 320, 331, 338 and 342"C, when the heating rates were 2, 5, 10 and 15 "C min-', respectively The Kissinger plot shown in Fig 13 yielded E, values of 2 6 and 2 8 (fO 3) eV for T,(1) and T,(2),respectively Fig 8, 10 and 12 indicate that the exothermic peak of K(2) is commonly observed in xK20-(90-x)V205-10Fe203 glasses and its inten- sity increases with increasing K20 content The E, for the z(2) peak was increased from 2 1eV to 2 4 and 2 8 eV when 211 c -11 10 10 9 J1 3 15 17 19 21 103 KIT, Fig.9 Kissinger plot of 20K20-7oV2o5-1OFe2O3 glass, in which T,(1 ), T,(2)and T,(3)were obtained from Fig 8 E, is the activation energy for the crystallization -eT :ndo J I I I I I I ,'I 100 200 300 400 500 T/"C Fig. 10 DTA curves of 25K20. 65V205 10Fe203 glass recorded at heating rates of (a) 2, (b) 5, (c) 10 and (d) 15"Cmin-' 1.5 1.6 1.7 1.8 1.9 2.0 103 KIT, Fig.11 Kissinger plot of 25K20-65V205* 10Fe203 glass, in which T,(1) and c(2)were obtained from Fig. 10 the K20 content was increased from 20 mol% to 25 and 30 mol%, respectively. The third crystallization peak, T,( 3), was not observed in 25K20 * 65V205* 10Fe203 and 30K20 9 60V205-10Fe203 glasses because these glasses melted in this temperature range, e.g. at 430 and 380°C in Fig. 10(b) and 12(c), respectively. Glass-transition temperatures, Tg, of 220-222, 213-217 and 204-206 "C were obtained for 20K20*70V205-10Fe203 (Fig. 8), 25K20*65Vz0,-10Fe203 (Fig. 10) and 30K20*60V205 10Fe203 (Fig. 12) glasses, respectively. The decreased gradually with increasing K20 content, as is generally observed in alkali-metal oxide glasses due to a decrease in the coordination number (CN) and/or the formation of NB0.11-14 In the case of xK20.(90-x) V205.10Fe203 glasses, the decrease in Tg is ascribed to an increase in the formation of VO, tetrahedra at the expense of V05 pyramids and to the formation of NBO in V04 and FeO, tetrahedra.The results of the magnetic susceptibility measurements are shown in Fig. 14 and 15. Using the well known eqn. (2), peff =(8xm (2) the effective magnetic moments, peff, of 25K20 *65V205 -10Fe203 glass were estimated to be 5.8 and 5.5 pB at 300 and 80 K, respectively. In eqn. (2), xm and T are the molar magnetic exo 100 200 300 400 500 TI"C Fig. 12 DTA curves of 30K,0*60V205 *10Fe203 glass recorded at heating rates of (a) 2, (b) 5, (c) 10 and (d) 15 "Cmin-' 12.5 12.0 h 11.5 N.g11.0 C-10.5 10.0 1.55 1.60 1.65 1.70 1.75 1.80 1.85 103 WT, Fig.13 Kissinger plot of 30K20.60V20, 10Fe203 glass, in which T,(1) and T,(2)were obtained from Fig. 12 =t 4.5 4.0 3.550 100 150 200 250 300 350 TIK Fig. 14 Effective magnetic moment, pen, us. temperature, T, of 25K,0*65V205 10Fe203 glass obtained at low temperature (0);the curves for the same glass after heat-treatment at 340 "C for 2100 min (0)and 5000 min (+ ) are shown for comparison susceptibility and the temperature (in K), respectively. These values are representative of high-spin Fe3 , which has five + unpaired electrons in five 3d orbitals.The value of peff is consistent with the Mossbauer measurements which revealed the absorption peak due to Fe3+ with a tetrahedral symmetry: Fe3'(G). As shown in Fig. 14, the peffvalues decreased after heat treatment at 340 "C for 2100 min, and they were 5.4 and 4.0~~300 and 80 K, respectively. The peff values of at J. Muter. Chem., 1996, 6(12), 1889-1896 1893 5 ~'~l*''~l'~~~l~'~~l'~~~l"~'Iheat treatment at 380 "C for 2100 min were determined only .d Fig. 15 Reciprocal of the magnetic susceptibility, x,,, ', us temperature T,of 25K@. 65V205 * 10Fe~0, glass obtained at low temperature (0) Xm for a sample after heat-treatment at 340 OC for 2100 min (0)and 5000 min (+ ) are shown for comparison 25K20* 65V205* 10Fe203 glass ceramic prepared by heat treat- ment at 340 "C for 5000 min were 5 2 and 3 9 pB at 300 and 80 K, respectively Since the distinct decrease in the magnetic moment seems to be related to an antiferromagnetic inter- action, the reciprocal of the molar magnetic susceptibility, xrn-', was plotted us temperature (Fig 15) The intercept of the straight line obtained for the non-treated glass sample yielded a Curie-Weiss constant, 8, of -13 (& 5) K, which indicates a weak antiferromagnetic interaction in the paramag- netic phase By contrast, 25K2O-65V2O5 10Fe203 glass cer- amics prepared by heat treatment at 340°C for 2100 and 5000 min yielded 8 values of -93 and -108 (+_5)K, respect-ively Fig 15 shows that the plots of the sample treated for 2100 min deviate from a straight line in the temperature region below 130 K, probably because of the precipitation of small antiferromagnetic particles Because of this, the 8 value at -93 K was estimated from the 'linear' parts of the plots obtained in the temperature region above 130 K A sharp increase of the electrical conductlvity, Q,was observed in 25K20-65V205 10Fe203 glass after heat treat- ment at 380°C for lOmin, as illustrated in Fig 16, which indicates that the values of 0 obtained by the dc two-probe method are almost the same as those obtained by the ac four- probe methods Since the four-probe method can be applied effectively to samples with relatively high conductivities, the 0 values of non-treated glass and the glass ceramic prepared by r7 lo4 .-10-5 cn2 10-6 10-8 Fig.16 Electrical conductivitles, 6, us time, t, of 25K20-65V,O, lOFe,O, glass obtained after the heat treatment at 380 "C 0 and 0 refer to the results obtained by the dc two-probe and ac four- probe methods, respectively The error bars indicate the region of t~ values obtained by the dc two-probe method under a voltage of 0 to & 10 V, as shown in Fig 17 1894J Mater Chem, 1996, 6(12), 1889-1896 by the dc two-probe method The concordant values of obtained by the dc two-probe and ac four-probe methods indicate that the 'contact resistance' between the indium elec- trode and the surface of the glass sample can be ignored, as can be the 'polarization effect' around the electrode In order to confirm the absence of 'contact resistance', additional con- ductivity measurements were carried out by the dc two-probe method for non-treated 25K20*65V20510Fe203 glass, on which a silver film of thickness ca 100nm was deposited as the electrode by a radiofrequency (rf) sputtering method The electrical conductivity proved to be almost the same as that obtained for the samples on which the indium electrode was attached Fig 17 shows the relationship between the electrical conduc- tivity obtained by the dc two-probe method and the voltage The values of Qobtained for non-treated 25K20*65V205 -10Fe203 glass are constant irrespective of the magnitude of the voltage, 2e the glass shows ohmic behavlour By contrast, the electrical conductivities obtained for the samples heat treated at 380°C increased with increasing voltage This non- ohmic behaviour becomes pronounced with increaslng heat- treatment time The error bars given in Fig 16 represent the region of the electrical conductivities obtained from the dc two-probe method under a voltage of 0 to flOV Fig 17 shows that heat treatment of the vanadate glass at 380°C for 2100min yielded Qvalues less than the detection limit of 5 x lop8S cm-l when voltages of 2 to -3 V were applied Because of this, the electrical conductivity of the glass ceramic prepared by heat treatment at 380 "C for 5000 min could not be estimated These results indicate that the formation of an insulating phase (KV308phase) in the glass matrix results in a gradual decrease in the electrical conductivity Discussion As shown in Fig 1 and 2, the Mossbauer spectra of 25K20 65V205-10Fe20, glass showed a distinct decrease in the linewidth (FWHM) from 049 to O28mms-' and a decrease in the quadrupole splitting (A) from 061 to 0 51 mm s-' when heat treated at 340 "C These Mossbauer results indicate that a structural relaxation and a rearrange- ment of several ions took place in the distorted FeO, and VO, tetrahedra Distortion of the structural units of vanadate glasses will be reduced as a result of the 'cooperative move- ment'29of the fragments, like a cluster composed of VO,, FeO, and V05 units The fragments will migrate from one site with 10-2p"""! c J EN Fig.17 Variation of the electncal conductivity, 6 of 25K,0.65V20, * lOFe,O, glass with voltage, E, at different times, t, investigated by the dc two-probe method under a voltage, E, of 0 to +1OV Note that only the (r values exceeding the detection limit of 5x 10 * S cm-' are given and hence the central part of the results obtained after the heat treatment at 380 "C for 2100 min is depleted t=lOmin (+), 30mm (V), 60min (A), 150mm (n),600min (0), 2100 min (0),untreated glass (0) a higher energy to another site with a lower energy in the supercooled liquid state which is defined as the temperature region between the glass-transition temperature (T,)and the melting point (T,).Hence, it is expected that brief heat treatment brings about a structural relaxation of the glass matrix.By contrast, precipitation of crystalline phase(s) like KV308 becomes significant when the glass sample is heat treated at 340 "C for more than 150 min, as could be confirmed from the intense XRD peaks after heat treatment for 1200-5000 min. Since no diffraction peak was observed in the XRD study of the glasses heat treated for 10, 30 and 60 min, it is concluded that the jump of the electrical conductivity shown in Fig. 16 is due to the structural relaxation of the 'skeleton' structure, which was initially composed of distorted VO,, FeO, and VO, units. Note that a jump of the electrical conductivity was observed in the semiconducting glass before the insulating KV308 phase precipitated. As described above, the electrical conductivity of semiconducting vanadate glass is known to be due to a step-by-step hopping of the valence (3d) electrons from V4+ to V5+.Hence, the increase in the electrical conduc- tivity can be ascribed to an increasing probability of the electron hopping, which is brought about as the result of a decreasing distortion of VO,, FeO, and VO, units. Note that the mean electrical conductivity of vanadate glass was increased distinctly upon brief heat treatment before the KV308 phase precipitated. In the DTA study of 25K20.65V205.10Fe203 glass (Fig. lo), two exothermic peaks of z(1)and z(2) were observed in the regions of 260-284 "C and 319-344 "C, respect- ively. In the Kissinger plot shown in Fig. 11, two straight lines obtained for z(1) and c(2) yielded E, values of 2.0 and 2.4 (k0.3) eV, respectively.These activation energies are much smaller than the V-0 bond energy of V04 tetrahedra (4.9 eV22) and VO, pyramids (3.9 eV22). This result suggests that a cleavage of Fe-0 bonds triggers the crystallization, as was observed recently in the DTA study (Kissinger plot) of 38Na20-61 W03 57Fe203 glass," in which the Fe- 0 bond energy was estimated to be 2.6 (k0.3) eV. By contrast, the DTA study of iron-free 38Na20*62W03 glass', yielded an E, value of 4.7 (f 0.3) eV which is, within experimental error, equal to the W-0 bond energy of 4.5 eV.22 E, values of 2.0-2.4 (&0.3) eV indicate that cleavage of Fe- 0 bonds causes the rearrangement of V4+-O-V5+ bonds in the VO, and VO, units, which is directly associated with the semiconductiv- ity (electron hopping) of vanadate glass.Hence, it is considered that the electrical conductivity due to the electron hopping from V4+ to V5+ is enhanced in the less distorted skeleton structure before crystalline phases precipitated. In the non-treated vanadate glass, it is considered that FeO, units substitute VO, and hence they are identical from the structural point of view. This can be deduced from the absence of Fe-0 bands in the FTIR spectra as shown in Fig. 6 and 7(a). The FTIR study of heat-treated samples revealed a distinct peak separation between VO, and FeO, tetrahedra due to the structural relaxation [Fig. 7(b)] and to the partial crystallization [Fig.7(c)-(h)]. A decrease in the distortion of VO, tetrahedra and VO, pyramids is also deduced from the FTIR spectra. The FTIR spectra of 25K20- 65V20, -10Fe203 glass ceramics showed the absorption bands due to FeO, tetrahedra at 580 cm-', reflecting a phase separa- tion of FeO, tetrahedra from VO, tetrahedra. As a result of the precipitation of the KV308 phase composed of VO, pyramids, it is considered that the fraction of VO, and FeO, tetrahedra increases in the glass matrix because it was originally composed of VO,, FeO, and VO, units. A' systematic decrease in the effective magnetic moments, shown in Fig. 14, is ascribed to an antiferromagnetic interaction due to Fe3+ with five unpaired electrons in the 3d orbitals. This is consistent with the appearance of Fe-0 bands in the FTIR spectra of heat-treated 25K20.65V205 10Fe203 glass [Fig. 7(c)-( h)]. The linear relationship obtained for the plot of xml us. Tand the Curie-Weiss constants, 8, obtained from the intercepts of the straight lines (Fig. 15) revealed that Fe3+ ions were enriched in the paramagnetic glassy phase, not in the crystalline KV308 phase. Taking account of the XRD study (Fig. 4) that no crystalline phase was associated with the iron, the size of the antiferromagnetic 'nanocluster' or 'nanocrystal' precipitated in the glassy phase, if present, will be too small to be detected by the Mossbauer measurements because superparamagnetism is generally observed in small particles of diameter <10 nm.30 The non-ohmic behaviour of 0 observed in the heat-treated sample can be ascribed to the crystalline KV308 phase and the antiferromagnetic nanocluster or nanocrystal, because the homogeneous single phase of the vanadate glass becomes heterogeneous when the crystalline phases (particles) precipitated.Conclusions The 57Fe Mossbauer spectra of xK20.(90- x)V205 * 10Fe203 glasses (20 <x,<30) heat-treated at around the crystallization temperature (340 "C) show a simultaneous decrease of the linewidth and the quadrupole splitting, reflecting a structural relaxation and crystallization. The symmetry of FeO, tetra-hedra constituting the network structure becomes higher upon this relaxation and crystallization. Well resolved FTIR spectra are observed in the heat-treated samples due to the relaxation and crystallization.An XRD study of the heat-treated samples revealed precipi- tation of the KV308 phase with a small amount of K4V10027 or K3V5014 phases. The K20/V205 ratio (0.33) of the KV308 phase is equivalent to the K20/(V205+Fe203) ratio of non- treated 25K20. 65V20, * 10Fe203 glass. A Kissinger plot of the DTA data yielded E, values of 2.0-2.9 (f 0.3) eV for the crystallization of xK,O * (90-x)V205-10Fe203 glasses (20 <x<30), indicating that cleavage of Fe-0 bonds with a single bond energy of 2.6 eV triggers the crystallization, as was observed recently in 38Na20-6 1 W03 57Fe203 glass. The magnetic moment of 25K20-65V205 * 10Fe203 glass was decreased after heat treatment at 340°C for 2100 and 5000 min, due to an antiferromagnetic interaction of Fe3+ ions.The reciprocal of the magnetic susceptibility, zrn-', yielded Curie-Weiss constants, 8, of -13 K for non-treated glass and -93 and -108 K for 2100 min and 5000min treated samples, respectively. Upon brief heat treatment at around the crystallization temperature (380 "C), a jump of the electrical conductivity, 0, was observed in 25K20-65V205 1OFe2O3 glass from 6.3 x lo-' S cm-' to 4.3 x lop4S cm-l, which is ascribed to an increase in the probability of electron hopping from V4+ to v5+. Precipitation of the insulating KV308 phase resulted in a gradual decrease in the electrical conductivity, and it became comparable to or less than that of the non-treated glass.It is concluded that the electrical conductivity is associated primar- ily with the semiconductivity of the glassy phase at the early stages of the heat treatment, whereas it is representative of the insulating crystalline phases in the case of glass ceramics. This paper was financially supported by Mitsubishi Heavy Industries, Ltd, Daiichi Radioisotope Laboratory, and a Grant-in-Aid for Scientific Research (C). References 1 H. Sakata, M. Amano, T. Ishiguro and T. Hirayama, J. Ceram. SOC.Jpn., 1992,100,1398(in Japanese). J. Muter. Chem., 1996,6(12), 1889-1896 1895 9 10 11 12 13 14 15 16 17 18 N F Mott, Ado Phys, 1967,16,49 I G Austin and N F Mott, Ado Phys, 1969,18,41 H T Evans and S Block, Inorg Chem , 1966,7, 1808 A M Bystrom and H T Evans, Acta Chem Scand, 1959,13,377 K Jansen and G Sperlich, Phys Status Solidi B, 1973,55495 S L Segel and R B Creel, Can J Phys ,1970,48,2673 Y Dimitriev, V Dimitrov, M Arnaudov and D Topalov, J Non-Cryst Solids, 1983,57, 147 T Nishida, J Non-Cryst Solids, 1989,108,87 T Nishida and Y Takashima, Bull Chem SOC Jpn ,1987,60,941 T Nishida, J Non-Cryst Solids, 1994,177,257 T Nishida, J Radioanal Nucl Chem ,1994,182,451 T Nishida, Hyper-ne Interact, 1995,95,23 T Nishida, in Mossbauer Spectroscopy of Sophisticated Oxides, Akademiai Kiado, Budapest, 1996,ch 2 T Nishida, M Suzuki, S Kubuki, M Katada and Y Maeda, J Non-Cryst Solids, 1996,19423 T Nishida, J Muter Chem, 1992,2,733 T Nishida, S Inoue and Y Takashima, Bull Chem SOC Jpn ,1992, 65,1927 T Nishida, H Ide, T Shinmyozu, Y Takashima and Y Matsumoto, Jpn J Appl Phys, 1990,29,1293 19 T Nishida, J Radioanal Nucl Chem ,1995,190,381 20 P Tarte, in Physics of Non-Crystalline Solids, ed J A Pnns, North- Holland, Amsterdam, 1965, pp 549-565 21 H E Kissinger, Anal Chem ,1957,29, 1702 22 D Chakravorty, in Modern Aspect of Solid State Chemistry, Plenum, New York, 1980,pp 391-423 23 H T Evans and S Block, Inorg Chem ,1966,5,1808 24 V V Illanonov, R P Ozerov and E V Killdisheva, J Inorg Chem USSR, 1956,1,777 25 A M Bystrom and H T Evans, Acta Chem Scand, 1959,13,377 26 H Hirashima, T Koyama and T Yoshida, Yogyo-kyokai-shi,1985, 93,554 (in Japanese) 27 Y Dimitriev, M Arnaudov and V Dimitrov, J Solid State Chem , 1981,38,55 28 M Kurzawa, J Mater Sci Lett, 1992,11,976 29 G Adams and J H Gibbs, J Chem Phys, 1965,43,139 30 N N Greenwood and T C Gibb, in Mossbauer Spectroscopy, Chapman and Hall, London, 1971, ch 10 Paper 6/03778J, Received 30th May, 1996 1896 J Mater Chem , 1996, 6(12), 1889-1896