2430
J. S.Choi, H. Y. Lee,and K. H. Kim
Electrical Conductivity of Nickel Oxide-Magnesium Oxide Single Crystals Jae Shi Choi,* Hoo Young Lee, and Keu Hong Kim Department of Chemistry, Yonsei University, Seoul, Korea (Received March 2, 1973) Publication costs assisted by Yonsei University
The electrical conductivities of MgO single crystals doped with 0.375, 0.75, 1, and 5% NiO were measured from 900 to 1200" under oxygen partial pressures, Po,, of 10-l to atm. As Po, is increased from lo-? atm, conductivities increase up to a maximum occurring in the region between and atm and then decrease as Po, is further increased higher than atm. The point of maximum conductivity can be explained as the point a t which there is a change in the current carrier type. It is suggested that the current carriers are 0 2 -a t Po, neater than that corresponding to maximum conductivity and Mg2+ at lower Po2's.
Introduction The conduction mechanism in doped metal oxides is complex, involving both ionic and electronic flow, and dependent on surface pressure. In general metal oxides fall into two groups on the basis of oxide density in relation to the density of the metal. For those with oxide density greater than that of the metal, the distance between metal ions in the oxide is less than that in the metal, and such oxides are truly ionic oxides-alkali metal oxides, aluminum oxide, and the alkaline earth metal oxides (magnesium oxide is a member of this group). Included in the other group, in which the oxide density is less than that of the metal, are the transition metal oxides such as NiO, FeO, COO, and MnO, in which electronic conductivity predominates. In this research a mixture of the two types was used, MgO single crystals doped with NiO. As indicated above, MgO is an ionic semiconductor and NiO is known as an electronic p-type semiconductor a t high temperatures. The system of NiO-MgO single crystals as a solid solution has the following characteristics. (1)Both components are cubic oxides. (2) Complete solid solubility exists in this system. (3) Replacement of Mg2+ by Nip+ in the oxide lattice does not introduce a strain in the lattice since the cation radii are similar ( r N i z + = 0.670 A, yMg2+ = 0.650 A). (4) Both oxides possess relatively good stability against decomposition in the temperature range 9001500". The most fundamental thing to find out about the conductivity of ionic compounds is whether the charge carriers are electrons or ions. In the case of magnesium oxide this point has not been settled. In fact, there are apparently conflicting observations and analyses of the type of charge carriers important a t various temperatures and oxygen pressures. Most analysesl-8 of the mechanism of conductivity have been made on the assumption that the charge carriers were electrons. Mittoff obtained experimental results indicating electronic conduction and analyzed the results of conductivity measurements made over a range of oxygen partial pressures on crystals of varying p ~ r i t y His . ~ major conclusions were that (1) impurities of variable valence dominated the conductivity and were responsible for changes in conductivity with changing gas pressures surrounding the crystals; (2) conductivity is electronic; and (3) the nature of the charge carriers in pure samples and in any samples a t higher temperatures is not known. The Journal of Physical Chemistry, Vol. 77, No. 20, 1973
On the other hand, SchmalzriedlO and Sambongi and Omorill reported ionic conduction from electrochemical solid cell measurements. By maintaining a different oxygen partial pressure on each side of the crystal. MittofP-14 determined the ionic transference number and obtained results indicating that conduction is ionic a t temperatures lower than 1300" and electronic above 1300". However, the ionic conduction mechanism proposed by Davies15 is the most progressive theory reported recently. According to him the conduction carriers in MgO are 0 2 and Mg2+ and their mobilities are dependent upon the oxygen partial pressures surrounding the crystal. Davies, however, used a poly-crystal of MgO as a sample, so there might be error due to grain boundaries in the sample, and there were no quantitative data on the dependence of conductivity on oxygen pressure. The samples used in this investigation were MgO-NiO single crystals and the electrical conductivities were measured at different oxygen partial pressures; for in order to determine the conduction mechanism of semiconductor metal oxides, it is helpful to consider the oxygen pressure dependence. This investigation has been carried out to find and explain the dependence of electrical conductivity on oxygen partial pressure and the point a t which the current carriers of MgO-NiO single crystals change.
Experimental Section The samples used in this experiment are MgO single crystals from Saxonberg Co. doped with 0.375, 0.75, 1, and 5, mol % NiO. Samples were cut into rectangular forms with dimensions of 0.592 x 0.408 x 0.369 cm, 0.75 x 0.404 x 0.161 cm, 0.652 x 0.429 x 0.201 cm, and 0.684 x 0.412 x 0.198 cm, respectively. A quartz sample container in a furnace was connected with a vacuum system through a glass joint, and four platinum probes provided contact with the sample. Figure 1 shows a four-probe contact model and the conductivity measurement circuitry. Conductivity may be computed according to Valdes' theory.16 When all the distances between adjacent probes are equal (SI= Sp = S3 = S) and the value of LIS is greater than 2, conductivity is calculated as 1 I IJ = __-
2irs v
Here u is the conductivity (ohm-I cm-I), I is the current through the sample, and V is the floating potential across the inner probes. B is a dc current source and current
2431
Electrical Conductivity of NiO-MgO Single Crystals
0
x
12oooc llOO°C 1wooc
0
9OO0C
A
Figure 1. Measurement circuit of conductivity and four-probe model.
through the sample is maintained from to l o w 4A; also the potential across the inner probes is maintained between 0.5 and 1.5 V. The potential difference V was measured by a Leeds and Northrup 7554 Type-K4 potentiometer connected to the inner probes, and the current through the sample was measured by a Keithley Instruments 610B electrometer. In the measurements, it was found to be essential to (1) test ohmic contact, (2) shield the entire circuit and ground the shielding at the electrometer, and (3) keep the entire circuit, including the current supply, well isolated from the ground. Therefore, the quartz sample container was covered by a grounded stainless steel pipe to provide shielding from the electric and magnetic fields from the furnace. Conductivity has been measured in the region of temperatures from 900 to 1200" and oxygen partial pressures from 10-1 to IO-? atm. The various oxygen partial pressures required were established using pure oxygen (research purity) obtained from Matheson gas products or a mixture of 0.001% oxygen in nitrogen from the Matheson Co. The sample container was evacuated to a pressure of mm and then the temperature of the sample 6.72 x was increased immediately to 850" a t this pressure. Pure oxygen or a mixture of oxygen-nitrogen was then allowed into the sample container and evacuated to the required oxygen partial pressures. The pressure in the evacuated sample container, pure oxygen gas pressure, and the gas pressure of oxygen-nitrogen mixtures were read on an ultrahigh-vacuum ionization gauge, a manometer, and a thermocouple gauge, respectively.
1
I
I
-7.0
-6.0
-5.0
I
I
-4.0
-3.0
\
I
-2.0
-1.0
log Po:,
Figure 2. Conductivity of 0.375% NiO-MgO single crystal as a function of oxygen partial pressure at constant temperatures.
0 l?OO°C b 1lOOOC -d.O
x looooc Q
900%
-4.5
E
4 4
*;
s o
,.5.0 r( 0
-5.5
R e dts To determine the electrical conductivity dependence on the oxygen partial pressure over NiO-MgO single crystals, doped with 0.375, 0.75, 1, and 5% NiO, conductivities were measured at different oxygen partial pressures in the region of 10-1 to atm a t constant temperature for 900, 1000, 1100, and 1200". The results of the electrical conductivity measurements are shown in Figures 2-5. The conductivity of the NiO-MgO system studied as a function of oxygen pressure a t 1200" is presented in Figure 6. Log u values of various samples are plotted against log Po2. Conductivities increased with increasing oxygen pressures for oxygen partial pressures up to a maximum in the pressure region between and 10-3 atm and then
0 I
-7.0
I
-6.0
I
I
-5.0
-4.0
I
-3.0
1
-2.0
I
-1.0
l o o Po
Figure 3. Conductivity of 0.75% NiO-MgO single crystal as a function of oxygen partial pressure at constant temperatures.
decreased as pressure was increased beyond that corresponding to the conductivity maximum. Conductivity dependence upon oxygen partial pressure is similar regardless of temperature or mol % of doped NiO. It presents a The Journal of Physical Chemistry, Vol. 77, No. 20, 1973
2432
J. S.Choi, H. Y. Lee, and K. H. Kim
0 5 % NIO-Ma0
Al% NlO-K00 XO.764: NiO-W.00
I
-?.O
I
-6.0
I
-5.0
I
I
-3.0
-4.0
1
-7.0
I
-1.0
log Poq l o g Po2
Figure 4. Conductivity of 1% NiO-MgO single crystal tion of partial pressure at constant temperatures.
as a func-
0 1200°c b liO0OC
x 1oooQc 0
L
-7.0
I
-c.o
I
-5.0
I
-4.0
I
-3.0
900'C
a, I
-2.0
I
-1.0
loo Po9
Figure 5. Conductivity of 5% NiO-MgO single crystal as a function of oxygen partial pressure at constant temperatures.
striking contrast to Mittoff's9 conductivity data for MgO single crystals containing small amounts of FeO impurity. It is a very interesting and remarkable result. The Journal of Physical Chemistry, Vol. 77, No. 20, 1973
Figure 6. Conductivity of NiO-MgO system as a function of oxygen partial pressure at 1200".
Discussion The conduction mechanism of ionic compounds may be thought of as (a) the motion of electrons in the conduction band, (b) the motion of electron holes in the valence band, (c) the diffusion of cations, or (d) the diffusion of anions.17 The first two mechanisms are classified as electronic conductivity and the last two as ionic conductivity. Electrons or electron holes generally diffuse rapidly through the crystal, if present in the conduction and valence bands respectively, and give rise to a high conductivity. Cations and anions will be attracted to the cathode and anode, respectively, upon application of an electric field, but the diffusion of these species is generally slow, since ionic diffusion depends upon the existence of atomic defects in the crystal. If cation interstitials are major defects in a crystal, cation diffusion will be relatively fast, since the interstitial ion is surrounded by empty interstitial sites in which to move. On the other hand, if the major type of defect is the cation or anion vacancy (or both) diffusion will be much slower, since, for example, in order for a cation to move toward the cathode, a cation vacancy must move into a nearest neighbor position. The possible types of conductivity are shown schematically in Figure 7 . Therefore, the total conductivity ut is C J ~= o, + C T ~where n1 and oe are ionic and electronic conductivity, respectively. The transference number t , for ions will then be t = UL e - u,
+
ut u > De The t, is a function of temperature, oxygen pressure, and mole per cent of impurities contained. The transfer of charge through an ionic compound may be reduced to the electrical equivalent circuit represented in Figure 8. In the
2433
Electrical Conductivity of NiO-MgO Single Crystals
I
X M+
M+ X-
d
XM+
Mt xX- Mt X- M'
XM+
M+ X- Mf X-
&-
Figure 7 . Illustration of possible mechanism of electrical con-
ductivity.
* 0. (1on 3
This is a very different result from our measurement. In our result for NiO-MgO single crystals, conductivity increases with increasing Po, up to a maximum in the region between and atm and decreases with increasing Po, in the pressure region higher than that of the maximum. This behavior cannot be explained by an electronic mechanism. Therefore we conclude that the mechanism of conduction in NiO-MgO single crystals must be ionic rather than electronic. According to Davies' report15 and self-diffusion data for 0 2 - and Mg2-, we suppose that a t oxygen pressures higher than that at the conductivity maximum, the conduction is via 0 2 - and a t Po, lower than that at the conductivity maximum, conduction is carried out by the Mg2+ ion. For higher oxygen partial pressures, oxygen is adsorbed on the crystal surface
SO, Nvvwyc
U-.(plpcJ
Figure 8. Conduction mechanism of ionic compounds.
equivalent circuit ue represents the conductivity path for electrons and u, represents the independent path for ions to transfer charge. As was mentioned in the Introduction, the ionic character of MgO was investigated by Schmalzriedlo and Sambongi and 0rnori.ll They measured the emf developed in an electrochemical solid cell of MgO and obtained a value similar to that for ZrOz, which is an ionic semiconductor. Parfittl8 found that the activation energies for self-diffusion of Mg2+ in the intrinsic and extrinsic regions are 3.5 and 0.92 eV, respectively. Oisch and Kingerylg obtained 2.7 eV for 0 2 -self-diffusion in the extrinsic region. Davies15 reported that the activation energies for the conduction of MgO are 5.2 eV for the intrinsic region and 2.7 eV for the extrinsic region at higher oxygen partial pressures and 3.5 eV for the intrinsic region and 0.92 eV for the extrinsic region a t lower oxygen partial pressures. Davies suggested that the current carriers in MgO are oxygen ions a t higher oxygen partial pressures and magnesium ions a t lower oxygen partial pressures. This is because the activation energy obtained from conductivity data at high oxygen pressures is equal to the value for 02self-diffusion, and a t low oxygen partial pressures is equal to the Mg2+ self-diffusion measurement. Davies' theory seems to be reasonable. However, his theory has a fault in that it cannot show the conductivity dependence upon oxygen pressure and does not discuss the conversion point of conduction carrier type. For data of conductivity dependence upon oxygen pressure we may introduce Mittoff's r e ~ u l t where ,~ the electrical conductivity of MgO single crystals containing small amounts of Fe in the temperature region of 1300" is observed to depend upon the partial pressure of oxygen surrounding the sample. The conductivity increases a t oxygen pressures both higher and lower than atm. At this pressure the conductivity is a minimum. He explains the dependence of conductivity on oxygen pressure by changes in stoichiometry and thus lattice defects in MgO and mentions an electronic mechanism rather than an ionic one.
+ V(-)
= V,(ads)
where V(-) is oxygen vacancy and Vo(ads) is adsorbed oxygen atom in a vacancy site (MgO has a structure of oxygen deficiency). The number of anion vacancies would then decrease, and the diffusion of 0 2 -ions through the crystal and the electrical conductivity would also decrease. In this experiment, we used a four-probe contact method and the conductivity is surface conductivity rather than bulk conductivity. Therefore, the conductivity was very sensitive to oxygen pressure. However, in the region of low oxygen pressure conduction is carried out by Mg2+, and this Mg2+ ion diffuses through cation vacancies. Ni has variable valences as Fe does. At low oxygen pressures Ni has a + 2 rather than a +3 oxidation state. Thus, a t oxygen pressures lower than about atm, that corresponding to the conductivity maximum, the cation vacancies decrease with decreasing Poz
+
+
+
2 ~ i 3 + 02V(+) = Wi2+ %02 where V(+) is cation vacancy and, as the Mg2+ diffuses through cation vacancies, the electrical conductivity increases with increasing Po2. Acknowledgment. The authors are grateful to Dr. Tae Sun Park, President, and the Graduate School of Yonsei University, and to Professor R. G. Sauer for his help. References and Notes (1) A. Lernpicki, Proc. Phys. SOC.,London, Sect. B, 66,281 (1953). (2) E. Yarnakaand K. Sawarnoto, Phys. Rev., 95,848 (1954). (3) R. Mansfield, Proc. Phys. SOC.,London, Sect. B, 66,614 (1953). (4) N. A. Surplice, Brit. J. Appi. Phys., 15 (6),639 (1964). (5) E. Yarnakaand K. Sawarnoto, J. Phys. SOC.Jap., IO, 176 (1956). (6) H. F. John, Phys. Rev., 91,822 (1953). (7) E. Yarnakaand K. Sawarnoto, Phys. Rev., 101,565 (1956). (8) T. J. Lewis and A. J. Wright, Brit. J. Appl. Phys., (2) 1 (4),441 (1968). (9) S.P. Mittoff, J. Chem. Phys., 31,1261 (1959). (10) H. Schrnalzried, J. Chem. Phys., 33,940 (1960). (11) K. Sarnbongi and Y. Ornori, Tohuko Univ. Sci, Rept., IIA, 244 (1959). (12) S.P. Mittoff, J. Chem. Phys., 33,941 (1960). (13) S.P. Mittoff, J. Chem. Phys., 36,1385 (1962). (14) S.P. Mittoff, J. Chem. Phys., 41,2561 (1964). (15) M. 0.Davies, J. Chem. Phys., 38,2047 (1963). (16) L. 8.Valdes, Proc. IRE. 42,420 (1954). 117) R. A. Swalin, "Thermodynamics of Solids," Wiley, New York, N. Y., 1961,pp 298-300. (18) R. Lindner and G. D. Partitt, J. CChem. Phys., 26, 182 (1957). (19) Y. Oisch and W. D. Kingery. J. Chem. Phys., 33,905(1960).
The Journal of Physicai Chemistry, Voi. 77, No. 20, 1973