Materials Chemistry and Physics 82 (2003) 295–300

AC impedance studies of copper doped silica glass

Sung-Ping Szu∗ , Chung-Yi Lin
Department of Physics, National Chungshing University, Taichung, Taiwan, ROC
Received 11 October 2002; accepted 5 November 2002

Abstract
Copper doped silica glass with nominal composition 10Cu (mol%)–90SiO2 (mol%) is prepared using the sol–gel method. The dc
conductivities are Arrhenius temperature dependent with activation energy 1.0 eV. The conducting species is attributed to the un-reduced
Cu+ ions. The Kohlrausch–Williams–Watts function, φ(t) = exp[−(t/τ0 )β ], and the coupling model are utilized for analyzing electric
modulus at various temperatures. The analyses indicate that the ionic relaxation is coupled with the structural relaxation, the decreasing
of β at 550 ◦ C is due to the approaching the glass transition temperature of the sample. The dielectric constants at high frequency limit
increase with temperature initially and then level off at temperatures above 550 ◦ C; while, the dissipation factors increase exponentially with
temperatures. The increase of dielectric constants and dissipation factors with temperature changes are explained by the net polarization
in the medium at high frequency.
© 2003 Elsevier Science B.V. All rights reserved.
Keywords: AC impedance; Sol–gel; Copper; Silica glass

1. Introduction and diluted HCl was added as a catalyst. The pH value of

Sol–gel technique has been used extensively to prepare
novel ceramics or glass materials. One of the applications of
this technique is to fabricate glass–metal nano-composites
[1]. The metal cluster doped gel has nonlinear optical prop-
erties [2], and its UV-Vis absorption spectrum shows sur-
face plasma absorption peak [3]. The dc resistivity of the
nano-particle doped gels has also been studied [4,5].
In this paper, copper doped silica gel with nominal com-
position 10Cu (mol%)–90SiO2 (mol%) was prepared using
the sol–gel technique. The ac impedance technique was em-
ployed to study the conductivities and the dielectric relax-
ations of the gel at various temperatures.
2. Experimental
The sample was prepared according to the procedure de-
scribed later. Two solutions A and B were initially prepared.
Solution A was prepared by mixing 10 ml tetraethylorthosil-
icate (Si(OC2 H5 )4 ) with the same volume of methanol.
Solution B was prepared by dissolving copper nitrate
(Cu(NO3 )2 ·3H2 O) in 3.2 ml water. After 1 h of mixing at
room temperature, the two solutions were mixed together
∗ Corresponding author. Tel.: +011886-422840427x399.
E-mail address: spszu@phys.nchu.edu.tw (S.-P. Szu).
the sol was 1.8. The mixed solution was stirred in an Erlen-
meyer flask cover with plastic wrap at 60 ◦ C for 1 h. The sol
was distributed for every 20 ml in small Teflon dishes and
then covered the dishes tightly. They gelled within a week
at room temperature. However, gels were kept in room tem-
perature for another 6 weeks for complete drying. The color
of dried gel is blue and its size is about 2 cm in diameter and
1 mm thick. The gel was heat treated stepwise from room
temperature to 700 ◦ C in flowing oxygen with a heating rate
of 0.5 ◦ C/min. During the heating process, the sample was
kept at 250, 450 and 700 ◦ C for 10 h at a time. In the end,
the flowing oxygen was switched to 10%H2 –90%N2 mix-
ture gas in order to reduce Cu2+ ions into metallic Cu. The
temperature was still kept at 700 ◦ C for 10 h in flowing mix-
ture gas. Finally, the furnace as cooled down with a cooling
rate of 0.5 ◦ C/min. After such slow heating and cooling
processes, a sample with about size 1 cm × 1 cm × 0.08 cm
was obtained for ac impedance measurement.
The ac-complex impedance was measured using an
HP4194A impedance analyzer. Contacts were made by
coating the surfaces of the sample with silver paste. A
100–15 MHz frequency range was employed. Measure-
ments were made in flowing nitrogen gas over the temper-
ature range 20–680 ◦ C with a heating rate of 1 ◦ C/min. The
complex impedance Z –Z plot was fitted by a submerged
semi-circle and the intercept of the semi-circle with the Z
axis was the dc resistance. The complex electric modulus

0254-0584/$ – see front matter © 2003 Elsevier Science B.V. All rights reserved.
doi:10.1016/S0254-0584(03)00220-7
296 S.-P. Szu, C.-Y. Lin / Materials Chemistry and Physics 82 (2003) 295–300

M∗ and dielectric constant ε∗ are related to the complex

impedance by M ∗ = (ε∗ )−1 = jωC0 Z∗ , where C0 is the
geometric capacitance of the specimen. In the frequency
domain, the complex
electric modulus is described
 by
∞
M ∗ (ω) = (1/εs ) 1 − 0 dt exp(−jωt)(−dϕ/dt) , where
ϕ(t) is the relaxation function and εs is the dielectric con-
stant at high frequency limit. It is known that the relaxation
of the electric field in most ionically conducting glasses
follow the Kohlrausch–Williams–Watts (KWW) function
[6], ϕ(t) = exp[−(t/τ0 )β ]. Thus, we employ the KWW
function to simulate the electric modulus by taking εs , τ 0
and β as fitting parameters.

3. Results and discussion

A typical complex impedance Z –Z plot is shown in

Fig. 1. The data is fitted by a submerged semi-circle. The
departure of the semi-circle in the low frequency part could
be due to an electro-gel interface [7]. Fig. 2 is the Arrhe-
nius plot of the dc conductivities. The activation energy of
conductivity is 1.04 ± 0.01 eV. This value is larger than the
activation energies of most ionic oxide glasses [8].
Fig. 3 shows a typical plot of the real and imaginary parts
of the electric modulus as function of frequency. The solids
curves are simulated using the KWW function. Table 1 sum-
marizes the fitting parameters τ 0 , β, εs , and the submerged
angle θ for sample at various temperatures. The tempera-
ture dependencies of these parameters are also displayed in
Figs. 4–7.
Fig. 2. The Arrhenius plot of conductivities vs. inverse of temperatures.
The Arrhenius plot of τ 0 versus temperature is shown
in Fig. 4. The activation energy obtained from the slope is
1.04 ± 01 eV, which is very close to that of the dc conduc-
tivity. We may say that both the dc conduction and its re-
laxation process are a thermal activated hopping process.
The temperature dependence of KWW exponent β
is shown in Fig. 5. The exponents are around 0.89 for

Fig. 1. The complex impedance Z –Z plot for sample at T = 660 ◦ C. Fig. 3. The real and imaginary parts of the electric modulus as function
The solid curve is a submerged semicircle, where its origin is 6.9◦ below of frequencies for sample at T = 660 ◦ C. Solid curves are fitted with
the Z axis. parameters τ0 = 1.46 × 10−5 s, β = 0.82 and εs = 9.45, respectively.
 S.-P. Szu, C.-Y. Lin / Materials Chemistry and Physics 82 (2003) 295–300 297

Table 1
The fitting parameters of the KWW function τ 0 , β, εs , and the submerged angles θ at various temperatures
Temperature (◦ C)

660 640 620 600 580 560 540 520 500 480 460 440 420 400

τ 0 (s) 1.46 × 1.95 × 2.67 × 3.62 × 4.99 × 6.87 × 9.51 × 1.34 × 1.94 × 2.94 × 4.7 × 7.64 × 1.3 × 2.4 ×
10−5 10−5 10−5 10−5 10−5 10−5 10−5 10−4 10−4 10−4 10−4 10−4 10−3 10−3
β 0.82 0.83 0.84 0.85 0.87 0.88 0.89 0.89 0.89 0.89 0.89 0.89 0.89 0.95
εs 9.45 9.45 9.42 9.42 9.43 9.42 9.42 9.40 9.39 9.37 9.31 9.19 8.96 8.18
θ (◦ ) 6.9 6.6 6.5 6.2 6.3 5.9 3.5 3.3 3.7 3.6 3.4 3.8 1.3 0

temperatures between 420 and 540 ◦ C, and then decrease

Fig. 4. The Arrhenius plot of τ 0 vs. inverse of temperatures.
with increasing temperature. Fig. 6 shows the submerged
angles θ as a function of temperatures. Initially, the angle
increases with temperature, then levels off to about 3.5◦ at
temperatures between 460 and 550 ◦ C, and increases again
at higher temperatures.
Fig. 7a displays the temperature dependency of dielec-
tric constant at high frequency limit, εs . Fig. 7b is the
dissipation factor, tan δ(= ε /ε ), at 1 MHz as function of
temperatures. As the temperature increases, the values of
εs increases rapidly to about 9.4, then levels off at temper-
atures above 450 ◦ C, while the dissipation factors are very
small at low temperatures, and increase exponentially at
temperatures above 400 ◦ C.
According to the Debye model for dipolar relaxation be-
havior, the trace of Z –Z plot should be a semicircle. How-
ever, the Debye relaxation behavior does not occur very of-
ten. Either the Z –Z plot shows a submerged semicircle
where the center of the semicircular arc is depressed below
the real axis, or the plot is a screwed arc [9]. The non-Debye
relaxation behavior can also be identified in electric modulus

Fig. 5. The values of β as function of temperatures. Fig. 6. The submerged angles θ as function of temperatures.
298 S.-P. Szu, C.-Y. Lin / Materials Chemistry and Physics 82 (2003) 295–300

Fig. 7. Temperature dependences of high frequency limit dielectric constant, εs (a), and dissipation factor at 1 MHz (b).
 S.-P. Szu, C.-Y. Lin / Materials Chemistry and Physics 82 (2003) 295–300
formalism. For non-Debye relaxation, the KWW exponent
β is less than one, while β equals to 1 for Debye relaxation.
As shown in Table 1, both θ and β indicate non-Debye relax-
ation behavior, and it is more distinct at higher temperatures.
The conductivity of 10%Na2 O–90%SiO2 glass [10] is
∼10−5 ( cm)−1 at 400 ◦ C. The dc conductivity of the sam-
ple is about 10−9 ( cm)−1 at 400 ◦ C, it is much smaller
than that of ordinary alkali silicate glasses. An interesting
question concerns the conducting species in the sample. Af-
ter a long heat treatment at high temperatures in an oxygen
atmosphere, the possible positive and negative mobile ions,
such as Cl− , NO3 − , H+ , OH− are expected to be driven off
by the heat applied to the sample. Therefore, it is improba-
ble that the conductivity is due to those mobile ions.
Another possibility causing the dc conductivity is the elec-
tron thermionic emission between particulate Cu. The nom-
inal composition of the sample is 10Cu (mol%)–90SiO2
(mol%). If all copper ions are reduced to copper metal, using
8.93 and 2.0 g/cm3 , respectively, for density of copper and
silica gel heated at 700 ◦ C the volume fraction of the par-
ticulate copper is about 2.4%. This small fraction of metal-
lic copper in the composite prohibits thermionic emission
of electrons between copper particles. Therefore, a possible
conducting species is un-reduced Cu+ ions. Although sam-
ple has been heated at 700 ◦ C in a reduced atmosphere for
10 h, the presence of un-reduced copper Cu+ ions inside the
bulk sample is still possible.
The temperature dependence of β, shown in Fig. 5,
indicates a transition at 550 ◦ C. The β is a constant be-
tween 420 and 550 ◦ C, and starts to decrease as the
temperatures increase above 550 ◦ C. The values of β in
0.4Ca(NO3 )2 –0.6K(NO3 ) ionic glass [11] display the same
behavior as the glass approaches its glass transition tem-
perature (Tg ). According to the Nagi’s coupling model
[12], β is related to the coupling strength between mobile
ions or coupling between mobile ion and the structural
relaxation. Moynihan [13] defines the ratio of mechanical
relaxation time to the ionic relaxation time as the decou-
pling index, R␶ . If R␶ is small at Tg , such as R␶ ∼ 106 for
0.4Ca(NO3 )2 –0.6K(NO3 ) ionic glass, the ionic relaxation
is coupled with the structural relaxation. Under those cir-
cumstances, β will decrease with temperature as the glass
approaches its Tg . However, if coupling between mobile
ions is the dominant effect, such as most of the melted oxide
glasses, β will increases with temperature, and R␶ is ∼1012
at Tg . An empirical equation for R␶ at Tg is log Rτ (Tg ) =
14.3 + log σdc (Tg ) [14]. Using this equation, the decoupling
index R␶ is about 107 at 550 ◦ C for this sol–gel prepared
porous material. The value of R␶ justified that decreasing of
β at 550 ◦ C is due to the approaching its Tg and the ionic
relaxation is coupled with the structural relaxation.
As shown in Fig. 7a, initially, εs , which is the dielectric
constant at a high frequency limit, increases with tempera-
ture, and then levels off at temperatures above 500 ◦ C. Sim-
ilar trends for εs are observed in other fast ion conducting
material [15]. The reason is given later. There will normally
299
be variation in free energy barriers from site to site in glass.
At high temperatures, the jump frequency of the charge car-
rier becomes large and comparable with the frequency of
the applied field. Charge carriers hop easily out of the sites
with low free energy barriers in the electric field direction,
and tend to accumulate at sites with high free energy barri-
ers. This leads to a net polarization of the medium and gives
an extra contribution to ε . At low temperatures, the jump
frequency of the charge carrier becomes smaller than the
frequency of the applied field. The periodic reversal of the
applied field takes place so rapidly that there are no excess
charge carries jumping in the field direction, and the polar-
ization due to charge piling up at high free energy barrier
sites disappears and there is no static contribution. Hence
the values of ε decrease.
Fig. 7b shows an exponential increase of the dissipation
factor with temperature. This is typical behavior [16,17], and
can be understood for the same reason given later. At low
temperatures, there will be no charge piling up at sites with
high free energy barriers. The electric field in the sample
will respond to the applied field simultaneously. Hence, the
dissipation factors are small. However, as the temperature
increases, charges will pile up at sites with high free energy
barriers. The electric field in the sample cannot respond to
the applied field simultaneously at high frequency. There-
fore, the dissipation factors increase with temperatures.
4. Conclusion
A silica glass doped with nano-sized copper metal is pre-
pared using the sol–gel method. However, not all copper ions
are reduced to metal after long heat treatment in a reduc-
ing atmosphere. These un-reduced Cu+ ions are attributed
to the conductivity in the ac impedance measurements. The
KWW function is used for analysis in the electric modu-
lus. The analysis indicates a non-Debye relaxation behavior
of the conducting ions. All three fitting parameters, β, εs ,
τ 0 , are temperature dependent. The change of β indicates
that the glass transition temperature of the gelled prepared
glass is about 550 ◦ C, and the coupling strength between the
conducting ion and the structural relaxation is stronger than
that between conducting ions. The same activation energy,
which is 1.0 eV, of τ 0 and σ indicates that both dc conduc-
tion and its relaxation process are thermal activated hopping
processes. Finally, the increase of εs and the dissipation fac-
tor with temperature changes are due to the net polarization
in the medium at high frequency.
References
[1] S. Szu, C.Y. Lin, C.H. Lin, J. Sol-Gel Sci. Technol. 2 (1994) 881.
[2] J.D. Mackenzie, J. Sol-Gel Sci. Technol. 1 (1993) 7.
[3] D. Kundu, I. Honma, T. Osawa, H. Komyjama, J. Am. Ceram. Soc.
77 (1994) 1110.
[4] T.K. Kundu, D. Chakravorty, J. Mater. Res. 11 (1996) 200.
[5] A. Chatterjee, D. Chakravorty, J. Mater. Sci. 27 (1992) 4115.
300 S.-P. Szu, C.-Y. Lin / Materials Chemistry and Physics 82 (2003) 295–300
[6] C.T. Moyniham, L.P. Boesch, N.L. Laberge, Phys. Chem. Glasses
14 (1973) 122.
[7] M.A. Villegas, L.R. Jurado, J.M. Fernandez Navarro, J. Mater. Sci.
24 (1989) 2884.
[8] A. Feltz, in: D.R. Uhlmann, N.J. Kreidl (Eds.), Glass Science
and Technology: Structure, Microstructure, and Properties, vol. 4A,
Academic Press, San Diego, 1990, pp. 132–139.
[9] S. Elliott, The Physics and Chemistry of Solids, Wiley, Chichester,
England, 1998, p. 550.
[10] E. Seddon, E.J. Tippett, W.E.S. Turner, J. Soc. Glass Technol. 16
(1932) 450.
[11] F.S. Howell, R.A. Bose, P.B. Maccdo, C.T. Moynihan, J. Phys. Chem.
78 (1974) 639.
[12] K.L. Ngai, R.W. Rendell, Phys. Rev. B 30 (1984) 2133.
[13] C.T. Moynihan, N. Balitactacr, L. Boone, T.A. Litovitz, J. Chem.
Phys. 55 (1971) 3013.
[14] C.A. Angell, Chem. Rev. 90 (1990) 523.
[15] D. Ming, J.M. Ravez, J. Giate, P. Hagenmuller, J. Solid State Chem.
116 (1995) 185.
[16] M.J. Strutt, Arch. Electrotech. 25 (1931) 715.
[17] H. Moore, R.C. DeSilva, J. Soc. Glass Technol. 36 (1952) 5.