PHYSICAL REVIEW B VOLUME 60, NUMBER 5 1 AUGUST 1999-I

Multiphonon tunneling conduction in vanadium-cobalt-tellurite glasses

H. Sakata and K. Sega
Department of Applied Chemistry, Tokai University, 1117, Kitakaname, Hiratsuka, Kanagawa 259-1292, Japan

B. K. Chaudhuri
Indian Association for the Cultivation of Science, Jadavpur, Calcutta 700 032, India
共Received 20 August 1998; revised manuscript received 4 February 1999兲
Ternary vanadium oxide glasses in the system V2O5-CoO-TeO2 are fabricated by press quenching of glass
melts, and dc conductivities 共␴兲 are investigated at temperatures from 330 to 475 K for different glass com-
positions. From the study of the phase diagram, the glass formation region is found to be in the range of
V2O5⫽0 – 85 mol %, CoO⫽0 – 35 mol %, and TeO2⫽25– 100 mol %. These glasses contain microcrystalline
clusters dispersed in the glass matrix. In the high-temperature regime above ⌰ D /2 共⌰ D is the Debye tempera-
ture兲, the small polaron hopping model is found to be applicable. In the low temperature 共below ⌰ D /2兲 regime,
however, both Mott’s variable-range hopping and the Greaves’ intermediate range hopping models are found
to be not applicable. The most probable transport for the entire range of temperature and compositions is
concluded to be due to multiphonon tunneling of large polarons between the microclusters, supporting the
model proposed by Shimakawa. 关S0163-1829共99兲03029-5兴

I. INTRODUCTION glasses with single transition metal ion 共TMI兲, conductivities

of relatively small number of oxide glasses containing two
The dc conductivities of transition metal oxide 共TMO兲 TMI’s have so far been thoroughly investigated.7,9,18,19 The
glasses has been targeted for extensive studies1–9 because of effect of the second transition metal ion on the overall con-
their interesting semiconducting properties as well as for ductivity is found to be different in different systems. That
their probable technological applications. The conduction is, it decreases18 or increases20 with the addition of the sec-
mechanism in these glasses was understood by the small ond TMI. For Fe2O3-V2O5-P2O5 glasses,20 a SPH transport
polaron hopping 共SPH兲 model10,11 based on strong electron- between Fe and V ions in addition to the SPH between V
lattice interaction. The experimental results of conductivity ions was suggested.
and other transport properties of many binary1–3,4,6 and In the present work, we report the conductivity of a
ternary5,7–9 vanadate glasses supported the SPH model. V2O5-CoO-TeO2 type glassy system prepared by adding
At low temperatures 共below ⌰ D /2, ⌰ D is the Debye tem- CoO to the V2O5-TeO2 oxide mixtures. Interestingly, these
perature兲 where polaron binding energy is less than kT 共k is glasses are found to contain microcrystalline clusters dis-
the Boltzmann constant and T is the absolute temperature兲, persed in the glass matrix. They also show some interesting
the three-dimensional 共3D兲 variable-range hopping 共VRH兲 features along with decreasing effect on conductivity with
共Ref. 12兲 with T 1/4 dependence of conductivity 共␴兲 takes increase of Co content. The main objective of this work is to
place. The VRH was reported for V2O5-TeO2 共Ref. 3兲 or explain the accurately measured conductivity data of these
Bi2O3 共Ref. 6兲 glasses and V2O5-Bi2O3-BaTiO3 共Ref. 13兲 and glassy materials with two TM ions with an appropriate the-
similar other vanadate glasses. We also reported VRH in oretical model.
V2O5-SnO-TeO2 glasses.14 It should be mentioned that the
above two models were based, in common, on a single pho- II. EXPERIMENT
non approach.
Recently Shimakawa,15 assuming microclusters in a glass Reagent grade V2O5 共99.99%兲, CoO 共99.99%兲, and TeO2
network, revealed that the dc and ac conduction of 共99.99%兲 were used as raw materials. After mixing in air a
V2O5-P2O5 共Refs. 1,16兲 and V2O5-TeO2 共Ref. 17兲 glasses batch of 6 g with prescribed compositions, the mixed mass of
could be interpreted by multiphonon tunneling of large po- each glass composition was melted in alumina crucible for 1
larons between microclusters in the glass. The conductivity h at 1023 K in an electric furnace. The melt was then poured
in this model is given by ␴ ⬀(T/T 0 ) n , where n is a constant on a thick copper block and immediately quenched by press-
depending on glass composition15 and T is temperature. This ing with another similar copper block. Following this proce-
multiphonon tunneling model of Shimakawa15 considering dure we obtained bulk glass of 2⫻2 cm2 size and about 1
polaron hopping between micloclusters in the glass network mm in thickness.
has not been well investigated. The multicomponent glasses The glass forming region was determined by X-ray dif-
that we have prepared are found to contain microcrystalline fraction analysis 共Philips, X’pert System PW 3020兲. The
phases embedded in the glass matrix. So there is a good glass transition temperature (T g ) was determined by differ-
chance of verifying the Shimakawa’s model15 with samples ential thermogravimetric analysis 共DTA兲 共Rigaku, DSC8230/
of our present investigation. TAS300兲 of the glass samples at a heating rate of
Here it should be pointed out that compared to the TMO 10 °C min⫺1. The density 共d兲 of glasses were determined by

0163-1829/99/60共5兲/3230共7兲/$15.00 PRB 60 3230 ©1999 The American Physical Society

PRB 60 MULTIPHONON TUNNELING CONDUCTION IN . . . 3231

TABLE I. Chemical composition and physical properties of

glasses.

Glass composition 共mol %兲a

dc T gd ␴0 Wb R
⫺3 ⫺1
V2O5 CoO TeO2 共g cm 兲 共K兲 共S cm K兲 共eV兲 共nm兲

70 10 20 3.095 513 1.72⫻105 0.561 0.400

60 10 30 3.463 520 3.79⫻104 0.538 0.404
50 10 40 3.832 529 1.83⫻104 0.540 0.413
40 10 50 4.200 540 7.71⫻102 0.511 0.429
a
Nominal composition.
b
FIG. 1. Transmission electron micrograph of the 475–435 K.
60V2O5-10CoO-30TeO2共mol %兲 glass 共scale: 0.1 ␮m兲.
c
Accuracy in the density measurement, ⫾1%.
d
Accuracy in the measurement of T g , ⫾2 K.
the Gay-Lussac method using toluene as an immersion liq-
uid. 0⬉CoO⬉35 mol %, and 25⬉TeO2⬉100 mol %. The density
The dc conductivity 共␴兲 of the as-quenched glasses was 共d兲 and T g data are given in Table I.
measured at temperatures between 330 and 475 K using the The Seebeck coefficients 共Q兲 for these glasses with differ-
four-point probe technique, a constant dc current of 0.1 ␮A ent compositions were measured. As shown in Table II, the
being applied between the electrodes using silver paste with negative values indicate n-type semiconducting behavior of
a spacing of 2 mm. The Seebeck coefficient 共Q兲 for the these glasses. No temperature dependence of Q was ob-
glasses was determined by measuring the thermoelectric served, which is similar to many other vanadate glasses.22,23
power of the glass samples with a temperature difference of No dc polarization was observed.
10 K between the two electrodes as reported earlier.21 To Figure 3 shows the Arrhenius plot of ln(␴T) between 475
study the presence of microclusters in the glass, a transmis- and 395 K. Deviation from a linear curve occurs around
sion electron microscopic study of the fine glass powder on ⌰ D /2 共⬃240–190 K, depending concentration兲, where ⌰ D is
carbon-grids was made using an electron microscope 共Hita- the Debye temperature. Figure 3 indicates a temperature de-
chi: Model H6000兲. pendent activation energy. The experimental conductivity
data in such a situation is well described with an activation
III. RESULTS energy for conduction 共W兲 given by the Mott formula10,11
The x-ray diffraction 共XRD兲 pattern of these glasses 共with
Cu K ␣ radiation兲 indicated homogeneous glassy character ␴ ⫽ 共 ␴ 0 /T 兲 exp共 ⫺W/kT 兲 , 共1兲
without showing any crystalline peak. However, a transmis-
sion electron microscopic study indicated the presence of where ␴ 0 is a temperature independent parameter as dis-
microcrystalline grains 共20 to 40 nm in size兲 共Fig. 1兲 uni- cussed below. W values obtained from fitting of the linear
formly distributed in the glass matrix. Their concentration is part of the curves in Fig. 3 共high-temperature regime, 475–
very small not detected from the XRD patterns of the 435 K兲 are given in Table I together with the mean spacing
glasses. So the glasses of our investigation are actually glass- 共R兲 for the V-O-V chains calculated from glass density 共d兲
nanocrystal composites. Similar behavior for all the using R⫽(1/N) 1/3 共where N is the V-ion density兲. At tem-
V2O5-CoO-TeO2 glassy system is observed. From an EPMA peratures lower than 395 K the linearity between ln(␴T) and
observation of V, Co, and Te elements in the glass, we T ⫺1 deviated appreciably as seen from Fig. 3.
found a homogeneous distribution of each constituent Figures 4 and 5 present the effects of V2O5 and CoO
element and no phase separation was observed. Figure 2 contents on ␴. At 425 K, ␴ was found to be 9.4
shows the glass formation region: 0⬉V 2 O 5 ⬉85 mol %, ⫻10⫺5 – 2.3⫻10⫺6 S cm⫺1 for CoO⫽0 – 30 mol %. The
conductivity increased with increasing V2O5 content for a
fixed CoO content 共Fig. 4兲, while it decreased with an in-
crease of CoO content 共Fig. 5兲.

TABLE II. Seebeck coefficient of V2O5-CoO-TeO2 glasses.

Temperature 共K兲
Glass composition 共mol %兲a 405 430 445 455
V2O5 CoO TeO2 ⫺Seebeck coefficient 共␮ V K⫺1兲b

60 0 40 556 547 546 547

60 10 30 596 583 600 605
60 20 20 635 638 611 634
a
FIG. 2. Glass formation region of V2O5-CoO-TeO2 glasses: 䊊, Nominal composition.
glass; 䊏, crystallized; 䉱, not melted. b
Accuracy in the measurement, ⫾10 ␮ V K⫺1.
3232 H. SAKATA, K. SEGA, AND B. K. CHAUDHURI PRB 60

FIG. 5. Effect of CoO content on dc conductivity of

60V2O5 xCoO (40-x)TeO2 共mol %兲 at 405 K.

observed in TiO2-V2O5-P2O5 共Ref. 18兲 and V2O5-MnO-TeO2

共Ref. 21兲 glasses.
The logarithm of the conductivity 共Fig. 3兲 shows a linear
temperature dependence up to a critical temperature T D
(⬃⌰ D /2) and then the slope changes with deviation from
FIG. 3. Temperature dependence of dc conductivity for linearity, and the activation energy is temperature dependent.
xV2O5 10CoO (90-x)TeO2 共mol %兲 glasses: 共a兲 x⫽40; 共b兲 x⫽50;
Such a behavior is a feature of small polaron hopping.1 So
共c兲 x⫽60; 共d兲 x⫽70. The scale for ␴ T is log10 ( ␴ T).
we first discuss the thermal variation of conductivity assum-
ing the SPH model10,11 based on a strong coupling of elec-
IV. DISCUSSION tron with the lattice by a single phonon. This model gives ␴
in the nonadiabatic regime for TM oxide glasses as given by
The glass formation region of the glasses of the present
Eq. 共1兲 viz.
investigation 共Fig. 2兲 is similar to that of the
V2O5-MnO-TeO2 glasses.21 The area of the region generally ␴ ⫽ 共 ␴ 0 /T 兲 exp共 ⫺W/kT 兲 . 共2兲
depends on ⌬T⫽T c ⫺T g 共T c is crystallization temperature兲
and the values of ⌬T are similar to those of the The activation energy W can be written as
V2O5-MnO-TeO2 glasses 共e.g., ⌬T⫽60 K兲 studied earlier.21
With an increase of V2O5 content in the glass, the con- W⫽W H ⫹W D /2 共 for T⬎⌰ D /2兲 , 共3a兲
ductivity increased 共Fig. 4兲. This means a decrease in the
V-O-V spacing R as indicated in Table I, which raises the W⫽W D 共 for T⬍⌰ D /4兲 . 共3b兲
hopping probability, providing that the activation energy 共W兲 The temperature independent conductivity in Eq. 共1兲 is given
and the fraction of reduced transition metal ion 共C兲 are al- by
most unchanged for varying compositions, as seen later in
the SPH model. In the V2O5-CoO-TeO2 glasses of our ␴ 0 ⫽ ␯ 0 Ne 2 R 2 C 共 1⫺C 兲 exp共 ⫺2 ␣ R 兲 /k, 共4兲
present investigation, the CoO addition lowered the conduc-
tivity 共Fig. 5兲. This means, Co ion in the glass hindered the where ␯ 0 is the optical phonon frequency (⬃1⫻1013 Hz)
carrier transport. Because CoO is not a glass network former, obtained from the Debye temperature given in Table III, R is
the Co ions are isolated in the glass network, which causes the mean spacing between transition metal ions given by R
obstruction in the hopping of electrons due to the lack of ⫽N ⫺1/3, ␣ is the tunneling factor, W H is the polaron hopping
oxygen bonds. A similar lowering of conductivity was also energy, W D is the disorder energy, and C is the fraction of
reduced transition metal ion. The values of W and ␴ 0 shown
in Table I are estimated from Fig. 3. The values of W in-
creases from 0.511 to 0.561 eV for different glass composi-
tions indicating an increase in conductivity 共Fig. 3兲. Al-
though the exact formula of C for TMO glasses with two
TMI is not yet known, we assume C
⫽0.35,0.315,0.285,0.25 for the V2O5⫽40– 70 mol % 共Table
I兲, referring to the data from the Fe2O3-V2O5-P2O5 glasses.20
These values of C and other parameters shown in Table III
were fitted to the linear part of the conductivity data in the
high temperature regime 共above ⌰ D /2兲. The values of ␣ thus
estimated from the fitting are given in Table III. These values
are somewhat larger than those of the binary glasses3,6 with
single transition metal ions. This may be due to the effect of
FIG. 4. Effect of V2O5 content on dc conductivity of heterogeneity of the present glasses containing microclusters
xV2O5 10CoO (90-x)TeO2 共mol %兲 glasses at 405 K. in the glass matrix.
PRB 60 MULTIPHONON TUNNELING CONDUCTION IN . . . 3233

TABLE III. Parameters for small polaron hopping conduction.

Glass composition 共mol %兲a

N ␣ ␯0 ␴ 0c ␣⬘ ␴ 0 共cryst兲
V2O5 CoO TeO2 (10 cm⫺3)
22
C b
共nm⫺1兲 13
(10 Hz) (10⫺18) 共nm⫺1兲 共S cm⫺1 K兲

70 10 20 1.650 0.250 64.1 1.70 7.7 0.25 1.21⫻105

60 10 30 1.516 0.285 64.8 1.65 2.9 1.6 4.24⫻104
50 10 40 1.419 0.315 65.0 1.67 0.78 1.9 3.37⫻104
40 10 50 1.266 0.350 66.5 1.68 0.028 6.0 9.62⫻102
a
Nominal composition.
b
From Ref. 20.
c
In S cm⫺1 K calculated from Eq. 共4兲.

We examined the pre-exponential factor ␴ 0 in Eq. 共4兲 glasses we, therefore, attempted to apply both the VRH mod-
with the experimental data. Estimation using Eq. 共4兲 gives els proposed by Mott10,12 and Greaves25 which is valid for
␴ 0 ⬃10⫺18 to 10⫺20 S cm⫺1 K 共Table III兲, a large difference the intermediate range of temperature. The expression for the
between the experimental ␴ 0 values 共Table I兲 extrapolated conduction by the VRH model10,12 is based on a single opti-
from the data in Fig. 3. Such a large difference between the cal phonon approach. In this model ␴ is given by10,12
experimentally and theoretically observed values of the tem-
perature independent conductivity ratio is considered to be a ␴ ⫽B exp共 ⫺A/T 1/4兲 , 共8兲
consequence of the presence of microcrystalline clusters in
where
these glasses 共Fig. 1兲. We then assume the conduction in
microcrystallites to be due to also SPH as well as in the A⫽4 关 2 ␣ 3 /9␲ kN 共 E F 兲兴 1/4, 共9兲
glassy matrix, and conduction is then expressed as

␴ ⫽ ␴ 共 glass兲 ⫹ ␴ 共 cryst兲 , 共5兲 B⫽ 关 e 2 /2共 8 ␲ 兲 1/2兴 ␯ 0 关 N 共 E F 兲 / ␣ kT 兴 1/2, 共10兲

where ␴共glass兲 is ␴ in Eq. 共2兲, ␴共cryst兲 is the conductivity of N(E F ) is the density of states at the Fermi level. A and B are
microcrystals and is given by obtained from the slopes of the ln ␴ vs T ⫺1/4 共Fig. 6兲. Then
the mean hopping distance in VRH R VRH and the hopping
␴ 共 cryst兲 ⫽ 关 ␴ 0 共 cryst兲 /T 兴 exp共 ⫺W ⬘ /kT 兲 , 共6兲 site energy (W 0 ⫽W D ) are evaluated from Eqs. 共9兲 and 共10兲
共Ref. 12兲,
where W ⬘ is the activation energy of conduction. ␴ 0 共cryst兲 is
described as R VRH⫽9 1/4/ 兵 8 ␲ N 共 E F 兲 ␣ kT 其 1/4, 共11兲

␴ 0 共 cryst兲 ⫽ ␯ ⬘0 N ⬘ e 2 R ⬘ 2 C ⬘ 共 1⫺C ⬘ 兲 exp共 ⫺2 ␣ ⬘ R ⬘ 兲 /k. 共7兲 W 0 ⫽3/兵 4 ␲ R VRH

3
N共 E F 兲其. 共12兲
The parameters with primes are defined for microcrystals as The values of A and B are given in Table IV, which is of the
the same as those in Eq. 共4兲.
order of 1024 eV⫺1 cm⫺3. This value of N(E F ) is found to be
If composition of the microcrystals is similar to that of the
large compared with those of the TMO glasses with single
glass matrix, then W ⬘ ⱌW. Assuming ␯ 0⬘ ⱌ ␯ 0 , C ⬘ ⱌC,
N ⬘ ⱌN, and hence R ⬘ ⱌR, we estimate ␴ 0 共cryst兲 for differ-
ent ␣ ⬘ from Eq. 共7兲. Estimation of gives ␴ 0 共cryst兲 ⬃102 to
105 S cm⫺1 K 共Table III兲, which is the same order of the ␴ 0
values 共Table I兲 obtained from the conductivity data 共Fig. 3兲.
Thus we conclude that SPH model is valid in the high tem-
perature regime for these glasses containing microcrystals
providing SPH in both glass matrix and microcrystalline
clusters.
In Fig. 3 we notice that the temperature dependence of
conductivity deviated from the linearity for temperatures less
than 435 K 共above which the SPH law is valid兲. We then
attempted to apply variable-range hopping 共VRH兲 共Refs. 10
and 12兲 as reported for binary or ternary vanadate
glasses.3,6,14 However, the validity of such a high tempera-
ture range is not beyond question. But it has been pointed
out24 that depending on the strength of Coulomb interaction
the expression for the density of states at the Fermi level is
modified and the VRH 共Refs. 10 and 12兲 may be applied
even at high temperatures ⬃300 K and above, though the
VRH should actually be applicable in the low temperature FIG. 6. ␴ vs T ⫺1/4 for xV2O5 10CoO (90-x)TeO2 共mol %兲
regime 共below ⌰ D /4兲 which is below 100 K. For these glasses: 共a兲 x⫽40; 共b兲 x⫽50; 共c兲 x⫽60; 共d兲 x⫽70. The scale for
conductivity is log10␴.
3234 H. SAKATA, K. SEGA, AND B. K. CHAUDHURI PRB 60

TABLE IV. Mott parameters for variable-range hopping con- TABLE V. Parameters for Greaves’ variable-range hopping
duction. conduction.

Glass composition 共mol %兲a Glass composition 共mol %兲a

B A N(E F ) b
A B N(E F ) b R VRHb W D b
⫺1 ⫺1 ⫺3 V2O5 CoO TeO2 共K 兲 共S cm K 兲 共eV⫺1 cm⫺3兲
1/4 ⫺1 1/2
V2O5 CoO TeO2 共K 兲 共S cm 兲 共eV cm 兲 共nm兲 共meV兲
1/4

70 10 20 45.6 2.62⫻1020 4.1⫻1023

70 10 20 38.5 5.43⫻1016 7.6⫻1023 0.17 65
60 10 30 47.6 1.11⫻1021 3.5⫻1023
60 10 30 42.5 3.11⫻1018 5.1⫻1023 0.19 72
7.90⫻1017 50 10 40 45.3 3.46⫻1019 4.2⫻1023
50 10 40 41.2 5.8⫻1023 0.18 69
40 10 50 42.0 6.82⫻1016 5.8⫻1023
40 10 50 32.2 9.83⫻1013 1.5⫻1024 0.14 54
a
a Nominal composition.
Nominal composition. b
Values calculated assuming ␣ ⫽20 nm⫺1.
b
Values calculated assuming ␣ ⫽20 nm⫺1.

TMI for which N(E F ) is of the order of we conclude that appearance of the VRH at T⫽330– 475 K
1019 – 1021 eV⫺1 cm⫺3 共Refs. 3, 6, and 14兲. is rather unreasonable for the present glasses. So an alterna-
We shall now apply the Greaves law25 of VRH which is tive approach has been made to explain the experimental
valid for the intermediate range of temperature 共below conductivity data for the entire range of temperature of our
⌰ D /2兲. According to this model,25 the expression for the con- investigation.
ductivity can be written as Shimakawa,15 assuming microclusters of the order of 8
nm in the glass network, explained dc and ac conductivities
␴ T 1/2⫽A exp共 ⫺B/T 1/4兲 , 共13兲 of V2O5-P2O5 共Refs. 1 and 16兲 and V2O5-TeO2 共Ref. 17兲
glasses by a multiphonon tunneling model of large polarons
where A and B are constants. The slope B of ln(␴T ) vs 1/2
共weak-coupled electron兲26–28 between microclusters. The
T ⫺1/4 共Fig. 7兲 is given by conductivity of the present glasses presented T 1/4 depen-
dence 共Fig. 6兲, and the structure may be similar to that for
B⫽2.1关 ␣ 3 /kN 共 E F 兲兴 1/4. 共14兲 V2O5-TeO2 glasses. Hence we discuss the conductivity data
共Fig. 3兲 using the multiphonon tunneling approach.15
Figure 7 shows the relationship of ln(␴T1/2) against T ⫺1/4
The dc hopping conductivity is generally given by26
drawn by rearranging the data from Fig. 3. The linear rela-
tionship confirms the Greaves VRH 共Ref. 25兲 in the interme- ␴ ⫽N c 共 eR 兲 2 ⌫/6kT, 共15兲
diate temperature range. The values of the parameters A and
B obtained from these curves are given in Table V. The where N c is the number of localized electron, R the hopping
N(E F ) values were estimated from Eq. 共14兲 assuming ␣ distance, and ⌫ is the hopping rate. The hopping rate in
⫽20 nm⫺1. The values of the density of states at the Fermi multiphonon tunneling of localized electrons with weak
level calculated from the parameter B of the Greaves model25 electron-phonon interaction is described by26–28
given in Table V (⬃1023 eV⫺1 cm⫺3) are also found to be
⌫⫽ 关 C exp共 ⫺ ␥ p 兲兴关 1⫺exp共 ⫺h ␯ 0 /kT 兲兴 ⫺p , 共16兲
very large, compared to the usual semiconducting oxide
glasses. So none of these two VRH models is found suitable where Cⱌ ␯ 0 , p⫽⌬/h ␯ 0 , ␯ 0 the acoustical phonon fre-
to explain the low temperature 共below ⌰ D /2兲 conductivity quency, ⌬ the difference of site energy 共W D in VRH兲, and ␥
data of these multicomponent glass nanocomposites. Thus is a constant and a measure of electron-phonon coupling.
The electron overlapping term exp(⫺2 ␣ R) is implicitly in-
corporated in C 共Ref. 15兲. For h ␯ 0 ⬍kT, Eq. 共16兲 is approxi-
mated by

⌫⫽ 关 C exp共 ⫺ ␥ p 兲兴共 T/T 0 兲 p , 共17兲

with T 0 ⫽h ␯ 0 /k. Since N c must be N(E F )kT, 29 we have
from Eqs. 共15兲 and 共17兲

␴ ⬀ 共 T/T 0 兲 p , 共18兲
where p is integral number but becomes nonintegral number,
providing distribution of hopping site distance is taken into
consideration.29
Figures 8 and 9 represent the relationship between ln ␴
and ln T for different glass compositions. The linear relation-
ship between these two quantities is clearly seen. According
to the Shimakawa model15 the experimental relationship be-
tween ␴ and T is expressed as ␴ ⫽ ␴ 0⬘ T n using a constant ␴ 0⬘
FIG. 7. Plots of ␴ T 1/2 vs T ⫺1/4 for depending on glass composition and n, being nonintegral
xV2O5 10CoO (90-x)TeO2 共mol %兲 glasses: 共a兲 x⫽40; 共b兲 x⫽50; number. The best fit of the data gave n values for the glasses
共c兲 x⫽60; 共d兲 x⫽70. The scale for ␴ T 1/2 is log10(␴T1/2). viz. n⫽11.7– 13.6 (CoO⫽10 mol %) and n⫽12.2– 14.5
PRB 60 MULTIPHONON TUNNELING CONDUCTION IN . . . 3235

TABLE VI. Physical parameters for multiphonon tunneling con-

duction. 共␯ 0 ⫽1.2⫻1012 Hz and T 0 ⫽57.6 K are used for calcula-
tion兲.

Glass composition 共mol %兲

⌬
V2O5 CoO TeO2 n ⫺log10 ␴⬘0 ␥ G 共meV兲

70 10 20 13.00 38.2 2.76 1.5 64

60 10 30 13.65 40.3 2.79 1.5 68
50 10 40 12.97 38.9 2.89 1.3 64
40 10 50 11.72 36.6 3.19 0.9 58
70 20 10 12.29 36.9 2.92 1.2 61
60 20 20 14.53 43.1 2.81 1.6 72
50 20 30 13.69 41.6 2.99 1.3 68
40 20 40 12.98 40.4 3.16 1.0 64

⫽57.6 K. A single-phonon process such as VRH cannot take

FIG. 8. Relationship between ␴ and T for place under the small ␯ 0 , provided the site energy ⌬ is com-
xV2O5 10CoO (90-x)TeO2 共mol %兲 glasses: 共a兲 x⫽40; 共b兲 x⫽50; parable or larger than h ␯ 0 , and this is possible because
共c兲 x⫽60; 共d兲 x⫽70. The scales for conductivity and temperature ⌬/h ␯ 0 ⫽n⫽11– 14 from the data in Table VI. Using the ex-
are log10␴ and log10T. perimental p values combined with Eqs. 共15兲 and 共17兲,
we estimated ␥ and ⌬ for the present glasses 关assuming
(CoO⫽20 mol %) as shown in Table VI. These values of n N(E F )⫽2.9⫻1020 eV⫺1 cm⫺3 共Ref. 14兲兴. For a
are comparable with those obtained by Shimakawa for 50V2O5-10CoO-40TeO2(mol %) glass, for example,
V2O5-TeO2 and V2O5-P2O5 glasses.15 we obtained ␥ ⫽2.87, ⌬⫽64 meV, and ⌫⫽7.706
If we assume percolating microclusters in the glasses with ⫻10⫺5 (T/57.6) 12.97 (Hz).
mean size l c 共say, ⬃4 nm兲, and ␣ ⫺1 共the Bohr radius兲 which The condition of occuring multiphonon tunneling process
is larger than the mean V-O-V spacing (aⱌ0.4 nm⫽R, is ␥ ⫽2 – 3 共Ref. 26兲. In the weak electron-lattice interaction,
Table I兲, electron transfer occurs easily in the cluster. As- ␥ ⫽ln(⌬/EM)⫺1, where E M is the measure of the electron-
suming a mean separation of each cluster (R c ⱌl c ) and a lattice coupling strength.15 The weak-coupling condition is
difference of site energy ⌬, hopping of weakly coupled elec- that G⫽(E M /h ␯ 0 )(kT/h ␯ 0 ), nearly less than 1 共Ref. 31兲.
tron 共large polaron兲 by tunneling is possible to occur be- From ␥, we have ⌬/E M ⫽ ph ␯ 0 /E M , which gives E M /h ␯ 0
tween microclusters. Since ␯ 0 is given by (a/ ␣ ⫺1 ) ␯ D 共Ref. with p(⫽n) values. The values for ⌬, ␥, and G thus esti-
30兲 where ␯ D is the maximum phonon frequency 共Debye mated are presented in Table VI. These ␥ and G values sat-
frequency兲 and ␯ D ⫽k⌰ D /h, ␯ 0 becomes smaller than nor- isfy the condition ␥ ⫽2 – 3 共Ref. 26兲 or G⬍1 共Ref. 31兲. Ac-
mal ␯ D ⫽1.2⫻1013 Hz. cordingly, we conclude that the multiphonon tunneling
Assuming R c ⱌlcⱌ ␣ ⫺1 ⫽4 nm and a⫽0.4 nm, we have conduction of weakly coupled electrons is the most probable
␯ 0 ⫽1.2⫻1012 Hz with ⌰ D ⱌ600 K and T 0 ⫽h ␯ 0 /k carrier transport in the present glasses.

V. CONCLUSION

Semiconducting V2O5-CoO-TeO2 glasses were prepared

by the press-quenching technique from the melts and the dc
conduction mechanism was investigated in terms of different
physical models. The glass formation region was found
to be 0⬉V2O5⬉85 mol %, 0⬉CoO⬉35 mol % and
25⬉TeO2⬉100 mol %. The x-ray diffraction patterns of the
glasses exhibited amorphous character, but transmission
electron microscopic study of the glass together with elec-
tron diffraction revealed microcrystallites in the glass matrix.
Glasses were found to be n-type semiconducting. The in-
verse temperature dependence of ln(␴T) in the range 330–
475 K gave linearity but deviated from linearity for tempera-
tures less than about 435 K.
The high temperature 共above ⌰ D /2兲 dependence of con-
ductivity could be qualitatively explained by the small po-
laron hopping model. Moreover, the low temperature 共below
FIG. 9. Relationship between ␴ and T for ⌰ D /2兲 conductivity data could not be explained either by
xV2O5 20CoO (80-x)TeO2 共mol %兲 glasses: 共a兲 x⫽40; 共b兲 x⫽50; Mott’s or Greaves’ variable-range hopping model giving rise
共c兲 x⫽60; 共d兲 x⫽70. The scales for conductivity and temperature to unusually large values of the density of states at the Fermi
are log10␴ and log10T.
3236 H. SAKATA, K. SEGA, AND B. K. CHAUDHURI PRB 60

level compared to those of transition metal oxide glasses. crystalline TiO2 phase. The ac conductivity is also similarly
These are special features observed in the multicomponent effected since ac conductivity and the imaginary part of the
semiconducting glasses with two transition metal ions and dielectric constant are related to each other.
containing microcrystalline clusters. Presence of such a mi- Finally, considering microclusters present in these
crocrystalline or nanocrystalline phase in the glass matrix glasses, the multiphonon tunneling conduction between mi-
also effects the frequency dependent conductivity and dielec- croclusters by large polarons 共weekly coupled electron兲 was
tric constant as in the case of V2O5-Bi2O3 共Ref. 32兲 and found to be the possible mechanism of conduction in the
V2O5-PbO 共Ref. 33兲 glasses containing a BaTiO3 phase. V2O5-CoO-TeO2 glasses with two transition metal ions.
These glasses show very high dielectric permittivity. Re- Similar approach could be extended to many other multi-
cently we have also observed 共unpublished兲 a very large di- component TMO glasses containing microcrystalline or
electric constant in V2O5-P2O5 glasses containing a nano- nanocrystalline clusters.

1
M. Sayer and A. Mansingh, Phys. Rev. B 6, 4629 共1972兲. 17
V. K. Dhawan and M. Mansingh, J. Non-Cryst. Solids 51, 87
2
L. Murawski, C. H. Chung, and J. D. Mackenzie, J. Non-Cryst. 共1982兲.
Solids 32, 91 共1979兲. 18
H. Hirashima, K. Nishii, and T. Yoshida, J. Am. Ceram. Soc. 66,
3
V. K. Dhawan, A. Mansingh, and M. Sayer, J. Non-Cryst. Solids 704 共1983兲.
51, 87 共1982兲. 19
R. Singh and K. Sethupati, J. Phys. D 22, L709 共1989兲.
4
M. Sayer and A. Mansingh, J. Non-Cryst. Solids 58, 91 共1983兲. 20
H. Hirashima, M. Mitsuhashi, and T. Yoshida, J. Ceram. Soc.
5
H. Hirashima, D. Arai, and T. Yoshida, J. Am. Ceram. Soc. 68, Jpn. 90, 411 共1982兲.
486 共1985兲. 21
K. Sega, Y. Kuroda, and H. Sakata, J. Mater. Sci. 33, 1303
6
A. Ghosh and B. K. Chaudhuri, J. Non-Cryst. Solids 83, 151 共1998兲.
共1986兲. 22
A. Ghosh, J. Appl. Phys. 65, 227 共1989兲.
7
N. Lebrun, M. Lévy, and J. L. Souquet, Solid State Ionics 40/41, 23
H. Mori and H. Sakata, J. Mater. Sci. 31, 1621 共1996兲.
718 共1990兲. 24
M. Mostefa, S. Boutiche, and M. Khodja, Solid State Commun.
8
H. Mori, T. Kitami, and H. Sakata, J. Non-Cryst. Solids 168, 157 82, 697 共1992兲.
共1994兲. 25
G. N. Greaves, J. Non-Cryst. Solids 11, 427 共1973兲.
9 26
H. H. Qiu, M. Kudo, and H. Sakata, Mater. Chem. Phys. 51, 233 N. F. Mott and E. A. Davis, Electronic Processes in Non-
共1997兲. Crystalline Materials, 2nd ed. 共Clarendon, Oxford, 1979兲, p. 65.
10
N. F. Mott, J. Non-Cryst. Solids 1, 1 共1968兲. 27
D. Emin, Adv. Phys. 24, 305 共1975兲.
11
I. G. Austin and N. F. Mott, Adv. Phys. 18, 41 共1969兲. 28
N. Robertson and L. Friedman, Philos. Mag. 33, 753 共1976兲; ibid.
12
N. F. Mott, Philos. Mag. 19, 835 共1969兲. 36, 1013 共1977兲.
13
S. Chakraborty, M. Sadhukahn, D. K. Modak, and B. K. 29
K. Shimakawa and K. Miyake, Phys. Rev. Lett. 61, 994 共1988兲.
Chaudhuri, J. Mater. Sci. 30, 5139 共1995兲. 30
D. Emin, Phys. Rev. Lett. 32, 303 共1974兲.
14
H. Mori, K. Gotoh, and H. Sakata, J. Non-Cryst. Solids 183, 1228 31
R. Englman and J. Jortner, Mol. Phys. 18, 145 共1970兲.
共1995兲. 32
S. Chakraborty, A. K. Bera, S. Mollah, and B. K. Chaudhuri, J.
15
K. Shimakawa, Philos. Mag. B 60, 377 共1989兲. Mater. Res. 9, 1932 共1994兲.
16 33
M. Sayer, A. Mansingh, J. M. Reyes, and G. Rosenblatt, J. Appl. M. Sadhukhan, D. K. Modak, and B. K. Chaudhuri, J. Appl. Phys.
Phys. 42, 2857 共1971兲. 85, 3477 共1999兲.