The Oxidation of Titanium Monoxide at High Temperatures - The

D. E. Poland, A. K. Kuriakose, J. L. Margrave. J. Phys. Chem. , 1965, 69 (1), pp 158–160. DOI: 10.1021/j100885a024. Publication Date: January 1965...
0 downloads 0 Views 285KB Size
D. E. POLAND, A. K. KURIAKOSE, AND J. L. MARGRAVE

158

water-free thorium oxide surface. Weight loss data for the remaining three safiiples indicate the same general trend although the uncertainty in the data does not permit any quantitative conclusions to be drawn.

Further elucidation of the energetics and mechanism of the interaction of water with the surface of thorium oxide must await precise and accurate water adsorption and weight loss data for these and additional samples. These experiments are currently in progress.

The Oxidation of Titanium Monoxide at High Temperatures

by D. E. Poland, A. K. Kuriakose, and J. L. Margrave’ Department of Chemistry, University of Wisconsin, Madison, Wisconsin

(Received JuEy 3,1964)

The rates of oxidation and the nature of the scales formed on single crystal samples of Ti012 have been studied a t 1 atm. oxygen pressure, between 802 and 924’. The reaction follows a parabolic rate law with an activation energy of 45 kcal./mole, after an initial apparently linear rate law region. X-Ray diffraction studies show that the product layer consists of rutile, although formed in two distinctive layers. The oxidation of both Ti metal and TiOl seems to have a common mechanism of scale growth since the two processes have nearly equal activation energies in the same temperature range.

There have been nearly a hundred studies of the kinetics and mechanism of the oxidation of titanium metal cited in the various recent books and review articles2-6 on the subject. Complete agreement among the various authors is not found, partly due to the variations in subtle experimental factors and partly due to the real complexity of the process, when examined closely. There is general agreement in that the main product is rutile with the innermost layer consisting of a solution of oxygen in the metal lattice. X-Ray investigations by Kofstad and co-workers’ show that at 900” this solution tends toward a limiting coniposition of TiOo 35, although Hurlens finds that its composition corresponds to Ti60 a t 700’. Hurlen’s study shows that below 550’ the oxidation of titanium follows a logarithmic rate law while above 600’ it follows a parabolic rate expression. Kofstad, et nl.,’ report that above 80Oo, the oxidation initially obeys a parabolic rate law which eventually gives way to a linear rate. X study of the oxidation kinetics of titanium monoxide was undertaken in the present investigation in The Journa: of Physical Chemistry

order to complement the data on the oxidation of titanium metal and to allow comparison of the kinetics and energetics of the two processes.

Experimental The samples of titanium monoxide used were thin slabs of nearly elliptical cross section, cut from a single crystal boule obtained from the Linde Company. The stoichiometry was established by reaction with fluorine as TiO1.* and the structure was nTaC1-cubic. (1) Department of Chemistry, Rice University, Houston 1, Texas. (2) (a) A. D. McQuillan and M. K. McQuillan, “Titanium,” Butterworth and Co., Ltd., London, 1956; (b) U. R. Evans, “The Corrosion and Oxidation of Metals.” Edward Arnold, London, 1960. (3) 0.Kubaschewski and B. E. Hopkins, “The Oxidation of Metals and Alloys,” Butterworth and Co., Ltd., London, 1953. (4) K . Hauffe, “Oxidation von Metallen und Metallegierungen,” Springer-Verlag, Berlin, 1956. (5) P. Kofstad, K. Hauffe, and H. Kjollesdal, Acta Chem. Scand., 12, 239 (1958). (6) P. Kofstad and K. Hauffe, Werkstofe Korroswn, 7, 642 (1956). (7) P. Kofstad, P. B. Anderson, and 0. J. Krudtaa, J . Less-Common Met&, 3, 89 (1961). (8) T. Hurlen, J . Inst. Metals, 89, 128 (1960).

OXIDATION OF TITANIUM IONO OXIDE

AT

HIGHTEMPERATURES

Trace impurities of aluminum, iron, and lead were present to the extent of 1-10 p.p.m. An emission spectroscopic analysis indicated the presence of trace amounts of aluininuni, calcium, magnesium, copper, and silicon. The specimens were ground flat with 400-grit alumina paper. polished with crocus cloth, washed in a sequence of distilled water, trichloroethylene, acetone, and methanol, and dried. They had a brass-like appearance and their surface areas were calculated from their nieasured geometrical dimensions. The technique used for the kinetic study was similar to the one for the oxidation kinetics of ZrC and ZrBz reported earlier. The cleaned samples were suspended from a calibrated quartz helical spring enclosed in a glass tube, into the hot zone of a Vycor-tube furnace, down which was passed a stream of dried, commercial tank oxygen at a flow rate of 60 ml./niin. A tinier was started inimediately and the weight gain of the samples with time was recorded by noting the extension of the spring using a cathetometer, at suitable intervals. The temperature of the furnace was measured with a calibrated P t us. Pt-107, Rh thermocouple, and was maintained within 2-3' of the reported value throughout the run. Correctioris vere applied for the slight furnace temperature gradient. The atmospheric pressure was measured prior to each experiment and it ranged between 737 and 743 torr over the entire set of experiments. After oxidation of the sample, an X-ray diffraction pattern of the oxidized surface was taken and then it mas fractured for microscopic examination of the layers.

159

tinued for a long time (180 min. a t 844') and the run at 924', where the product layer was yellowish white. This double-layered structure of rutile coatings has also been observed by other workers. Kofstad, el al.,' attribute one layer to a recrystallization of the rutile, though in their work double-layer formation was observed only above 900'. In this work, slight differences in the X-ray diffraction patterns were noticed between the two layers which could indicate some recrystallization. Plots of the weight gain of the T i 0 samples per unit area with time were apparently linear during the initial stages of the reaction, after which the lines tended toward a parabolic behavior. I n general, parabolic plots of the data (cf. Figures 1 and 2 ) were more consistent after an initial period of erratic behavior.

40

30 20 10

0

10

20

M

40

50

60

70

8 0 90 100 TIME- MINIJTES

110

120

I30

140

150

16rl

170

Figure 1. Oxidation of Ti0 (parabolic plot).

Results and Discussion Between 300 and t500° an initial change of color of the specimen from brassy yellow to violet and blueblack, indicating the formation of higher oxides,'O was observed, although without any measurable weight gain in these short times. X-Ray analysis of the various colored surfaces indicated nothing but T i 0 until the slate-gray TiOz layer was formed. Probably these early layers were very thin and epitaxially formed on the surface of the T i 0 single crystal, whereas the gray TiOz layers were sufficiently crystalline to yield good rutile powder patterns. The weighable oxide layer fornied on all of the samples during the ltirietic studies was found by Xray analysis to be only rutile, although it consisted of two distinctive layers-a marble-white inner and a silvery-white outer-when examined after fracturing the specinlens. The outer surface of the specimens had a slate-gray color indicative of a rioIistoichioiiietric coiiiposition, except in the case where the run was con-

Figure 2. Oxidation of Ti0 (parabolic plot). (9) A. K. Kuriakose and J. L. Margrave, J . Electrochem. Soc., 111, 827 (1964).

(IO) P. Ehrlich, Z . Elektrochem., 45, 362 (1939).

Volume 69, Number 1

January 1965

D. E. POLAND, A. K. KURIAKOSE, AND J. L. MARGRAVE

160

40

Table I : Kinetic D a t a for the Oxidation of TiOL.S

I

Temp.,

.Ot

I

L

,O‘I 1

Figure 3.

.4rrhenius plot of the parabolic rate constants.

Hence, the reaction rate constants were calculated based on a parabolic rate law. The initial deviation could be attributed to either a short-time linear reaction process or to rionequilibriuin temperature conditions. The break in the plot of the data a t 924’ may be due to a change to linear reaction kinetics as reported by Kofstad, et u L . , ~ to occiir after a region of parabolic behavior in the oxidation of titanium metal. A least-squares tit of the data in the parabolic region gives the rate constants presented in Table I a t the various tmiperatures. Figure 3 is an Arrhenius plot of the rate constants. The points for 855 and 892’ involve greater uncertainties in the oxidation temperatures thaii the others, since these were two of the initial experiments. An activation energy of 45.3 f 0.6 kcal./mole is obtained by a least-squares method after eliminating the two erratic points (49 f 4 kcal./mole if they are included), and this value is comparable to the values 50 and 31 kcal./niole obtained, respectively, by Hurlen8 and Iiofstad, et u L , ~for the parabolic region of the oxidation of titanium metal. Since the activation energies are about the same, it may be assumed that the rate-limiting process is the formation of TiOz for the Oxidation of both titanium metal and titanium monoxide There has been a controversy in the literature regarding the identity of th(. diffusing ionic species in the

Thr Journal of Physical Chemistry

802 844 855 886 892 924

O C .

Parabolic r a t e constant, mg.’/om.‘

0 0 0 0 1 1

20 44 32 91 36 77

min. - 1

f0 i0 f0 f0

01 01 02 01 f 0 03 f 0 03

oxide scales. Kofstad, et u L . , ~ interpret their work as if oxygen ions were the mobile species, while Kinna and Knorr,ll Hurlen,8 and Gulbrarisen and AndrewI2 favor Ti ion transport. This discrepancy has been , ~being the result of explained by Kofstad, et ~ l . as “plastic flow” of the layers during the crystallization process. Then experimental techniques and temperatures may determine the position of markers after oxidation, independent of which ion is more mobile, One of the ways to predict the more mobile ionic species is to compare the diffusion coefficients of oxygen and titanium in rutile but comparable data seem to be unavailable. Haul and Dunibgeri13 recently reported the results of oxygen isotope exchange with single crystal rutile. For the temperature range 710 to 950’ they found that the exchange rate is controlled not only by the rate of oxygen diffusion in the solid, but also by a phase boundary reaction involving activation energies of 75 and 61 kcal./inole, respectively. Therefore if oxygen diffusion alone were the rate-determining step in the oxidation of Ti and TiO, then the activation energy should be nearer to 7 5 kcal. Since the observed value is much smaller, it seems reasonable to assume that a titanium ion is the predominantly diffusing species in the oxidation rather than the oxygen ions. Carriahan and Brittairil4 have recently reported internal friction studies of nonstoichiometric TiOz crystals and suggested that the dominant nonstoichionietric defect in rutile is a Ti interstitial, in agreement with this conclusion.

Acknowledgiizents. The authors are pleased to acknowledge the support of this work by the United States Atomic Energy Commission. (11) W. Kinna and W. Knorr, 2. Metallkunde, 4 7 , 594 (1956). (12) E. A. Gulbransen and K. F. Andrew, Trans. ATME, 185, 741 (1949). (13) R. Haul and G. Dumbgen 2 . Elektrochem., 66, 636 (1962). (14) R. D. Carnahan and J. Brittaln. J . A p p l . P h y s . , 34, 3095 (1963).