ISSN 19950780, Nanotechnologies in Russia, 2010, Vol. 5, Nos. 9–10, pp. 656–664. © Pleiades Publishing, Ltd., 2010. Original Russian Text © A.V. Bagazeev, Yu.A. Kotov, A.I. Medvedev, E.I. Azarkevich, T.M. Demina, A.M. Murzakaev, O.R. Timoshenkova, 2010, published in Rossiiskie nanotekhnologii, 2010, Vol. 5, Nos. 9–10.
ARTICLES
Characteristics of ZrO2 Nanopowders Produced by Electrical Explosion of Wire A. V. Bagazeev, Yu. A. Kotov, A. I. Medvedev, E. I. Azarkevich, T. M. Demina, A. M. Murzakaev, and O. R. Timoshenkova Institute of Electrophysics, Ural Division, Russian Academy of Sciences, ul. Amundsena 106, Yekaterinburg, 620016 Russia email:
[email protected] Received March 12, 2010; in final form, June 9, 2010
Abstract—A device and an experiment for obtaining zirconium dioxide nanopowders using the electrical explosion of a wire (EEW) with aerodynamical separation are described. The physicochemical and techno logical properties of powders depending on the conditions under which they are obtained are studied. It is shown that, like for other metals with high oxidation heat, there is an additional dispersion of powders due to combustion. The results are discussed from the point of view of current concepts of electrical explosion and the combustion of metal powders. The optimum technological conditions according to the criterion of the maximum nanofraction yield were found. It is shown that this method allows us to obtain weakly aggregated ZrO2 nanopowders with a spherical particle shape (the specific surface reaching 70 m2/g (average particle size 10 μm) are also formed. As the explosion products slow down, the coarse drops mechanically leave them behind, pass to the ambient gas, and burn as single particles. In the combustion of particles of metals with a high heat of oxidation, there is intensive evaporation followed by vapor condensation and oxidation. Because the vapor concentration around drops is less than in the column of the explosion products, the particles of condensed oxide forming due to combustion on average are smaller than for particles forming immediately during the condensation of explosion products and form the most finely disperse fraction of the powder. Unburnt residua form the coarse fraction. Additional dispersion due to combustion also occurs at a higher overheating up to K ≈ 1 when the average size of particles is 100–200 nm. That corre lates with work [9], where it is found that, during the combustion of a dredge of submicron particles of Al (200 nm) in the gas flow, the average size of particles of the combustion products is less than the average size of the starting particles by a factor of 2–3. The theory of combustion of metal drops from units to a hundred micrometers in size was developed in [10]. According to this theory, combustion proceeds in two regions. At the drop surface, the heterogeneous oxidation of metal and evaporation of metal and oxide occurs, possibly with the decomposition of oxide. The oxygen supply to the surface and the withdrawal of evaporation products occur by way of diffusion; heat abstraction occurs by heat conduction. At a certain distance from the surface within the vaporphase of combustion, afteroxidation and the condensation of vapors with the heat release occurs. Processes at the surface can be approximately com puted without the effect of the vaporphase region. The procedure of such a computation is described in [11] by the example of aluminum. In this article it is applied to a computation of the combustion of zirco nium drops. The required thermodynamic data were taken from [12]. We preset the type of indifferent gas, pressure P of the gas medium, volume concentration kO2 of oxygen, and the temperature T∞ far off the drop 2010
658
BAGAZEEV et al.
Table 1. Output computation of the combustion of a zirconium drop Gas
Tb, K
αev
Weight fraction of residuum
Q, kJ/mol
Ar
3625
0.72
0.38
239
Zr(c) + 0.621O2 = 0.057Zr + 0.643ZrO + 0.017ZrO2 + 0.283ZrO2(c) N2
3580
0.59
0.55
362
Zr(c) + 0.687O2 = 0.047Zr + 0.532ZrO + 0.013ZrO2 + 0.407ZrO2(c) N2
3874
0.45
0.66
76
Zr(c) + 0.125O2 + 0.319N2 + 0.087ZrO2(c) = 0.037Zr + 0.398ZrO + 0.013ZrO2 + 0.638ZrN(c) Note: Second substrings are the total equation of reaction. The symbol “c” designates condensed substances.
as conditions. As a result of our computation, we determine the drop temperature Tb on combustion (further called the combustion temperature), degree αev of metal evaporation, composition of the gas medium at the surface, equation of the total reaction at the surface, its heat effect Q (by one mole of Zr), and a number of other values. These results are indepen dent of the drop radius. The combustion time is pro portional to square of the drop radius. Table 1 lists computation data for the conditions, specifically, P = 1 atm, T∞ = 300 K, kO2 = 21%, argon, and nitrogen. As is evident from Table 1, with the use of argon as the actuation gas, the combustion temperature of the zirconium drop is found to be lower than the boiling temperature of Zr and ZrO2 taken separately (4613 and 4573 K, respectively [13]) and close to the boiling temperature of their mixture (3944 K by the thermo dynamic computation). The main product of drop evaporation is monoxide ZrO; vapors of Zr and ZrO2 are less by a factor of 1–2 order; and the oxygen con centration at the drop surface is vanishingly small (~10–4). The fraction of the evaporated metal is suffi ciently large (αev = 0.72). This metal takes part in forming nanoparticles of ZrO2. The nonevaporated 28% zirconium turns into a residual, possibly frac tured, particle of dioxide with a weight of 0.38 of the starting drop and belonging to the coarse fraction of the powder. In the case of nitrogen, one can either introduce or not introduce the formation of condensed nitride into the computation. This allows us to imitate the extreme cases of kinetics of a heterogeneous reaction when the reaction rate is higher or lower than the rate of the dif fusion supply of the reagent. If nitride is not formed, the process is similar to combustion within argon. Because of the greater heat conductivity of nitrogen, in its atmosphere the combustion temperature and the evaporation degree decrease and the weight of the residual particle of oxide increases. Gaseous com pounds of nitrogen (ZrN, NO, NO2) are formed in small amounts (in total ~10–3) and not included into
the total equation of the reaction. Generally, consider ing reactions of nitrogen in the gas phase changes computational data only by fractions of a percent when compared assuming the inactivity of nitrogen. If the formation of condensed nitride ZrN (c) is assumed in the computation, only nitride is formed and there is no oxide, in spite of the fact that variation in the free energy and the reaction heat during oxida tion are greater than on nitriding by a factor of 3. A similar fact was also established in [11] for aluminum. Oxide, on the other hand, is consumed if it was ini tially at the drop surface (see Table 1, line 3). Nitride can burn and oxidize, turning into oxide. A fraction of it can remain in the powder. RESULTS AND DISCUSSION. SPECIFIC SURFACE AND POWDER YIELD In the first set of experiments, we study the depen dences that the abovementioned values have on over heating and the oxygen concentration during the EEW in the nitrogen–oxygen mixture. Zirconium is related to a group of metals for which overheating is limited by the development of a shunt arc discharge along the surface of the wire, which is induced by the thermoemission of electrons [8]. This is illustrated by the oscillogram of the current in Fig. 2a at which we observe a current transfer from the wire in the arc discharge. The energy of the arc dis charge takes no part in the dispersion of metal; there fore the introduced energy is counted up only to the moment when the arc begins. Figure 2b presents the dependence that the efficiency of the energy supply from a battery into the wire has on overheating. It is evident that, at K ≈ 1.2, the energy supply in the wire is practically stopped. The specific surface as a function of overheating (Fig. 3) has a maximum at K = 0.5 and, further, decreases as the overheating grows. At K = 0.5, i.e., under the conditions of the combustion of coarse drops, ZrO2 powders are formed with the size dBET = 15–20 nm similarly to the way Al2O3 powders are obtained [5, 16].
NANOTECHNOLOGIES IN RUSSIA
Vol. 5
Nos. 9–10
2010
CHARACTERISTICS OF ZrO2 NANOPOWDERS
659
S, m2/g 70
U, V 8 arc
60
6
50
(a) 4
40 30
2
1
20 0 −2
5
10
Efficiency, % 100
15
20
25 t, μs
(b)
0.6
0.8
1.0
4 1.2
1.4 K
In the second series of experiments, we studied the effect that the oxygen concentration and the type of the actuation gas have. The conditions were studied on a combustion of drops at K = 0.5 at which the maxi mum specific surface of the powder was obtained.
80 l = 226 mm 103 mm 70 mm
50 40 0.2
0 0.4
Fig. 3. Specific surface of powders within the filter (1, 3) and cyclone (2, 4) as a function of overheating for the EEW in the N2–O2 mixture at kO2 = 21% (1, 2) and 10% (3, 4).
90
60
2
10
−4
70
3
0.4
0.6
0.8
1.0
1.2 K
Fig. 2. (a) Typical oscillogram of current on the EEW of Zr in air at K = 1.1 (U0 = 30 kV, l = 103 mm); (b) coefficient of the efficiency of the energy supply in a wire as a function of overheating.
The decrease in the specific surface as the overheat ing grows is probably explained by the variation in the condensation conditions. Small drops burn not in ambient gas, but in the volume of explosion products where the vapor concentration is relatively high. Therefore, particles in the course of the condensation growth achieve large sizes. At K < 0.5 (curves 1, 2), the specific surface is reduced due to the formation of coarse drops with diameters of tens of micrometers. They reach the walls of the explosive chamber, having no time to burn out. Therefore, the weight fraction of the condensate formed from these drops and, consequently, its contri bution to the specific surface decrease. In this case a “crust” of incompletely burntout particles grows on the chamber walls. At K < 0.44, the wire is destroyed by fragments (the socalled threshold of explosion) and no visible yield of the powder is observed. The dependence of the specific surface upon the oxygen concentration is discussed below. NANOTECHNOLOGIES IN RUSSIA
Vol. 5
Nos. 9–10
It is evident (Fig. 4) that, in the nitrogen–oxygen mixture, the specific surface of both fractions of the product has a maximum in an oxygen concentration range of 15–20%. With the use of argon, the specific surface increases monotonically as kO2 decreases to 8%. There is a correlation between S in the filter and the cyclone (correlation coefficient r = 0.80). The nonmonotonic dependence of the specific sur face upon the oxygen concentration can be explained by the fact that there are factors affecting S in the opposite way. As computations by the combustion model show, as kO2 increases, the evaporation degree and, consequently, the quantity of the finedispersed condensate rise; this should increase S. On the other hand, at the same time the partial pressure of vapors around the drop increases, resulting in a more inten sive growth of particles and an increase in their average size; this acts in the direction of a decrease in S. There fore, there is an optimum concentration of oxygen where there is a maximum of S. In argon, due to a lesser heat transfer, the equal degree of evaporation corresponds to the lower con centration of oxygen; therefore, curves for argon should be moved in the direction of the lower kO2 in respect to curves for nitrogen. In Fig. 4, they are indeed moved to the left. It is possible that, at an even greater kO2, curves for argon show a maximum of S. The precipitation of the oxide condensate on the drop is lesser in argon, which is favorable for the growth of the specific surface of the fine fraction. Therefore, the minimum S in argon increases up to 73 m2/g against 64 m2/g in nitrogen. 2010
660
BAGAZEEV et al.
S, m2/g 80 3 70 60 50 40 4 30 20 10 0 5 10
1 2
15
20
25
30 35 kO2, vol.%
Fig. 4. Specific surface of powders within the filter (1, 3) and cyclone (2, 4) at K = 0.5 as a function of oxygen con centration in nitrogen (1, 2) and argon (3, 4).
One experiment was performed with a variation in pressure; it is found that, as the pressure decreases by a factor of two, the specific surface in the filter increases by 25%; in the cyclone, it increases by 10%. The pressure effect can be explained with the attrac tion of the combustion model. According to our com putation, as the pressure is lowered by half, the degree of evaporation somewhat increases, but at the same time the partial pressures of vapor components decrease almost in half, which results in the condensa tion growth of the nanoparticles and the growth in S being limited. An analysis of the yield of all the powder showed (Table 2) that, as the overheating and oxygen concen tration increase, the yield increases from 75 to ~90% due to a more complete oxidation of metal and dBET increases simultaneously (Figs. 3, 4). To analyze data, in addition to graphs, regression models were constructed in the form of polynomials of the first and second orders in normalized factors. They show that the yield into the filter and cyclone also increases as the overheating and oxygen concentration grow in argon when compared with nitrogen. At the Table 2. Yield of the powder and the productivity of the device Explosion conditions Branch K
kO2, % l, mm
Y, %
P, g/h
same time, the yield into the cyclone is approximately a factor of two larger than into the filter, and the spe cific surface in them is lower than in the filter by a fac tor of two. Thus, the selection of the optimum conditions is mainly determined by the required size of particles. The yield of the powder and productivity can be obtained close enough in both argon and nitrogen. By means of sedimentation, a weight fraction of particles of
