Oxidation Kinetics of Copper

An Experiment in Solid State Chemistry

Y. Ebisuzaki and W. B. Sanborn North Carolina State University, Raleigh. NC 27695 Considerable research interest has been shown recently in solid state chemistry due to its technological importance; thus, i t seems appropriate to introduce some solid state chemistry in the underg;aduate rurrirulum. The presrnt article dgscrihes the oxidation kinetics in metals and illustrates the mle defects play in diffusion-controlled reactions. Copper is probably the most suitable metal for student study since the "xidation proceeds at muderately high t e n ~ ~ e r a t ~ and l r e the mechanism has herome the prototype for other metsls (1-6). The ex~)eriment is suitahle for iuniors and seniors in uhvsiral " or inorganic chemistry laboratory. tant. The critical film thickness for the transition to the diffusion process depends upon the metal and the model employed in describing thin film oxidation. In copper the transition is assumed to take place a t around lo2to 103A, thus an oxide film of cm is considered thick (2,8). The reactions occurring in the Cu-Cuz0-02 system consists of Cu+ ions and electrons entering the oxide a t the oxide-metal internhase 1Cu Cu+ + e-) and diffusine out to the oxide-eas 2e' + 11~0;'hi&dary where oxidation pn,ceeds (2&+ Cu.10).The empirical oxidation rate is aiven hv dx/d/ = k l x . where k (cm2s") is the parabolic r a t e c o n s t a t and r is the oxide film thickness. This rate equation is in agreement with the general observation that, as the oxide becomes thicker, the rate of reaction hecomes slower. Integration of the rate equation results in a paraholic equation or the parabolic law (Wagner) ( 4 4 )

Theory Let us first consider the nature of defects in c u ~ r o u oxide s and the role they play in the diffusion of Cu+ ions in theoxidation Drocess. T h r oxide is nonstoichiometric, riolatinpl the law of definite proportions with the excess oxygen content heing of the order of 0.1'70, thus giving a formula of (7). Each excess oxygen removes two electrons Cut.99901.000 from the electronic band of the solid and the resulting anion is incorporated in a regular lattice site (Fig. 1) ( I ). The deficiency of Cu+ ions give rise to vacant cation lattice sites (0) and these are shown schematically in Figure 1 (5, 6 ) . The vacancies play an important role in the oxidation process since an adiacent Cu+ ion can move into a vacant site and then another ion will move into the evacuated position and thus allow the cuprous ions to diffuse across the oxide layer. During the oxidation process in Cu-Cu20-Ozsystem, a concentration gradient of cation vacancies is established with the vacancies heing created a t the oxide-gas interphase and the concentraa minimum at the metal-oxide houndarv. The tion d r o ~ ~ ito ne rate of oxid; growth is controlled by the diffusion of Cu+ across the oxide layer due to the concentration gradient of vacancies and this mechanism is supported by the proposed experiment. The oxidation process of Cu will now be considered. When a clean surface of Cu is exposed to oxygen, the gas molecules are rhemisorhed and a mckolayr of oxide is f o k e d . A rapid oxidation process follows in which the movement of Cu+ ions across the oxide film is controlled by the strength of the electric field between the metal and the chemisorhed oxygen (2,8). As the film thickness grows, the effect of the electric field diminishes as its gradient decreases and a slower, diffusion dependent process of the parabolic law becomes impor-

Equation (1) may be expressed in terms of weight increase (Aw) MCUSA m z = 2kt (2) Mo PA where p is the density of CuzO and A is the area' of the growing oxide film. A plot of Awz against t (time) maybe employed to calculate the rate constant. The relationship between the paraholic rate constant and the diffusion coeffiwill now he established. cient (D) Derivation of the Rate Law Only the positive ions and the electrons are appreciably mobile in Cu9O and need he included in the derivation oft he rate law (I, 2:4,6). A current of charged particles per unit area oer unit time or the flux. i. consists of two terms. one due to r diffusion along the con&tration gradient ( d ~ f d x with ) beina measured from the metal surface and the other due to the Gotion in an electric field, E:
j0 ---0
D

dCe -dl.

ECepe

(electron current)

(3)

. (4) &I (ionic current) dx where p is the mobility or velocity per unit field. The following steps will lead to a simple flux equation:

dCj j . = -D.-+EC.

(1) Eliminate E using eqns. (3) and (4); substitute for p from Einstein relation (Dlu = k'Tle) and assume a stead" state whew j, = I , = J ( I ) , where e is electronrr charge and k ' is
(2) We may set r,o

Rollzmann constant. =PC.@,,t , a = P C , ~I., , + t , and r, I. where o is conductivity and t is transference numhpr. (31 Finally [he concentration of mobile electrons (C,) isequal to the concrntration of cation vacancies (C,): C, = C, = C.

Figure 1. A schematic representation 01 nonstoichiometric cuprous oxMe. The excess 0'- in Cu.0 lead to cation vacancies which are designated as 0 .

When a thin copper sheet with a fairly large surface area is oxidized on the surface,the change in the area between that of the metal and the oxide is quite small, although the thickness of the film increases owing to the outward diffusionof the cuprous ions. A s a first approximation, me area (A) may be assumed to be that of the metal sheet and that it remains constant during the reaction in which an oxide scale forms. Volume 62 Number 4 April 1965
341

'

be left in the furnace for long duration. I t will be necessary to form a new fused thermocouple tip after many hours of use a t high temperatures. Results and Discussion When the square of the weight increase (Aw2) is plotted against time, a straight line is obtained, confirming the parabolic law as reauired by ean. (2). The results a t three temperatures are pr&ented.in Figure 2. The paraholic rate constants ( k ) were calculated and thp results were found to be reprod"cible. The rate constants at various temperatures from 687 to lO0OoCare summarized in the table. A straight line is obtained when the logarithm of the rate constants (loglo k) are plotted against the reciprocal of absolute temperature (Fig. 3). The temperature dependence of the rate constant is summarized by 3 4 . 5 0 . 9 keal mol-' (cm2s-') k = (6.0f 0.2) X 10V2exp -

Figwe 2. The square of lhe weight increase ( A s ) is ploned agsinst time (0 fw oxidation at 807V (smallest slope), 913'C and 1000'C (highest dope).

RT

Now the following flux equation is obtained:

Equation ( 5 ) may be rewritten (1,2) as

When the substitution is made for the concentration gradient in eqn. (6),we obtain j = 2D(C1 - Co)/x, where Cl and Co are the concentration of vacancies a t the metal-oxide and oxidegas interphase, respectively. The oxide growth rate, d d d t , is equal to the flux multiplied by the volume of oxide (V). That is,

The relation between the measured rate constant and the diffusion coefficient is k = 2DV(C1 - Co). The dependence of k on the concentration of vacancies (C) may be tested by changing the oxygen pressure (4,6). The rate law for the diffusion process and the effect of temperature on the rate constant in air can he measured in a laboratory experiment. The temperature effect is given by the familiar Arrhenius' equation k = koexp(-EJRT), where E. is the activation energy. A convenient temperature range for copper is 500 to 1000C. Experlmenlal A small copper sheet (-0.1 X 1 X 3 cm) is cleaned (dil. HN03 and HzO) and the area measured with a micrometer. The samole is s u s ~ e n d e d hv a fine wire in a furnace and allowed toixidize. 6 u r apparatus consisted of a modified double-oan anahtical balance sitting on a cement laboratory tahle topkith avertical furnace below one of the balance pans. As a precaution, considerable separation (12 in.) was provided between the furnace and the table top.2 The sample was suspended from the balance pan and hung freely in the furnace. The oxidation rate of Cu in air or the weight increase was obtained in reference to the apparent weight of the sample a t the commencement of the experiment. As an alternative to the above procedure, the individual samples were also removed from the oxidation furnace and the increase in weieht measured a t room temperature. A tube furnace (1 in. iyd.) heated with nickel-chromium (Nichrome) wire with uDner temperature of approximately ' 1 0 0 0 ~ was ~ employed.A~he e thermocou~leti^ was laced directlv below the s a m ~ l and the temper&e.was f&owed with astudent pocentiheter. If a chromel-alum~l thermocouple isemployed, it should not Caution: Organic solvents should not be used near the hot furnace. 342 Journal of Chemical Education

The parabolic law was followed in the present temperature range, however, a more complex relationship is reported to exist a t low temperatures (several hundred OC), especially in the earlv . staees .. of oxidation (8) Several points should he made concerning the manner in which the oxidation experiment was performed. The literature reports various furnace-hnlancearrangements in an attempt to minimize the effect of heat on the balance operation (9). We have found i t is sufficient to employ a large spacing between the furnace and the balance along with an insulating screen above the furnace. The buoyancy effect concerned us, thus the rate constants were measured with the balance arrangement discussed and also individual samples were removed from the furnace and weighed. The rate constants obtained by both methods are shown in Figure 3. Within experimental precision, we could not distinguish the two types of results. The oresent observation is in aereement with the ~ublished information that the buoyancy corrections are quire small in this rate constants (@) areshown exoeriment ( 9 ) .The ~ublished

Parabolic Rate Constants for Cu (k)

Temperature (OC)

k (cm2C1)

687 696 742 742 864

8.85 X 1 0 P 0 9.67 X 2.24 X lo-' 1.94 X lo-* 1.19 X

2.10 X

892
900 949 949 1000

1.98 X
4.02 X

5.15 X 10-a 6.37 X

(it K-')) The open cwcles (0) repesenl tne rate constants determined in mis
~ ~

Fbure 3. me rate eonstants liw.. as a function of temosrature . k ) are oioRBd , are litaratws values 19).

.~~~

repon and We closed c rcles 1.1

Acknowledgment

One of the authors (YE) expresses her appreciation to the students who have successfully carried out the experiment and made this report possible.
Literature Clted
(I) Moare, W.J.. "Seven Solid States," Benjamin, New York, 1967,Chapter 5.

(2) ~ o t tN , F . . and G U ~ ~ R Y ,w., ' " E I p ~m ~ c in ~~ h~ i ec ~r y ~ s~ ~D,O "V ~ N ~ .W York, 1964,CbaptPrVIII. B~.~W B., , Brsttain, W.H., and S ~ ~ ~ W.,J I W them , phys.. 14,714 (1916). (4) Kubaxhemki. 0.. and HopLms. B. E., "Oxidation of Metals and AUoya," Buttewonhs. London.LPs2. ( 5 ) M W ~W. , J., J. m ~ ~ou . c .~ , 2 3 (1961). 2 (6) Hauffe, K.."Oxidation olMetals,"Pienum. New York, 1965,Chapter 111. (7) O'Keeffe, M.,and Mmm, W. J.,J. Chem. Phys., 36.3W (1962). (8) ~ ~ N. t t F., , and F ~ ~ F P., I "oxidation ~ ~ ~ a[ ,M ~ ~and ~ AII~~~." I S A ~sac. ~M ~ ~~ A . ~ , Metals Park, OH, 1970, pp. 37-60. (9) Tylecote,R. F.. il Inst Met., 78.327 (1950).

(s

Volume 62

Number 4

April 1985

343