• Aucun résultat trouvé

EFFECT OF Y ON CATION AND ANION GRAIN BOUNDARY DIFFUSION IN CHROMIA

N/A
N/A
Protected

Academic year: 2021

Partager "EFFECT OF Y ON CATION AND ANION GRAIN BOUNDARY DIFFUSION IN CHROMIA"

Copied!
7
0
0

Texte intégral

(1)

HAL Id: jpa-00230354

https://hal.archives-ouvertes.fr/jpa-00230354

Submitted on 1 Jan 1990

HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.

EFFECT OF Y ON CATION AND ANION GRAIN BOUNDARY DIFFUSION IN CHROMIA

W. King, J. H . Park

To cite this version:

W. King, J. H . Park. EFFECT OF Y ON CATION AND ANION GRAIN BOUNDARY DIFFUSION IN CHROMIA. Journal de Physique Colloques, 1990, 51 (C1), pp.C1-551-C1-556.

�10.1051/jphyscol:1990186�. �jpa-00230354�

(2)

COLLOQUE DE PHYSIQUE

Colloque Cl, supplement au n o l , Tome 51, janvier 1990

EFFECT OF Y ON CATION AND ANION GRAIN BOUNDARY DIFFUSION IN CHROMIA

W.E. K I N G and J . H . PARK*

Chemistry and Materials Science Department, Lawrence Livermore National

;aboratory, Livermore, CA 94550, U.S.A.

Materials Science Division, Argonne National Laboratory, Argonne, IL 60439, U.S.A.

Abstract - Cation and anion diffusivities in the lattice and along grain boundaries have been measured on sintered polycrystals of Cr203 and Cr203-0.09 weight percent Y,03 at llOO°C and at the oxygen paartial pressurecorresponding to that of the Cr/Cr203 equilibrium at that temperature. Results show the effect of Y is to enhance cation grain boundary diffusion. It was also observed that under the same conditions, anion diffusion is significantly faster than cation diffusion. The results are discussed in terms of their implications for the reactive element effect on high temperature oxidation of chromia forming alloys.

1 - INTRODUCTION

In 1937, Pfeil I11 discovered that trace additions of oxygen active elements like Y, La, and Ce to alloys that form chromium or aluminum oxide layers upon exposure at high temperatures have a very strong beneficial effect on the oxidation behavior of these materials. Specifi- cally in the case of chromia formers, it was found that additions of these so-called "reactive-elements" reduced the growth rate of the oxide layer and enhanced the adhesion of the oxide layer to the base alloy in comparison to addition free alloys. This work addresses the mecha- nism by which Y affects the growth rate of chromium oxide.

Unlike many other phenomena, the rate controlling step in the formation of thermally grown oxide layers is the transport of the fastest moving spiecies across the oxide layer. Results of marker experiments lead one to believe that the formation of thermally grown chromium oxides proceeds by outward diffusion of cations, and therefore to conclude that in chromia, anions diffuse orders of magnitude slower than cations. It was also believed that oxidation was controlled by bulk diffusion. This hypothesis was supported for chromia by the cation diffusion measurements of Hagel /2/. More recently, Hoshiio and Peterson and Atkinson and Taylor /3,4/ demonstrated concIusively that bulk diffusion could not account for observed oxide growth rates, and short circuit paths, like grain boundaries, must be the dominant path for mass transport.

The effect of reactive-element additions on the growth rate of chromium oxide has been well documented. Three mechanisms have been proposed to explain the observed reduction in oxide growth rate: (a) doping effects on diffusion in the oxide (b) blocking of diffusion paths by the reactive element which has been incorporated into the growing oxide layer and (c) short-circuit diffusion models 151. In some cases, the effect was so large that the rate-controlling process had apparently changed from outward diffusion of cations to inward diffusion of anions /6,7,8,9/.

This evidence leads one to suspect that if anions diffuse orders of magnitude slower than cations in chromia and growth of chromia is controlled by the outward diffusion of cations, then the effect of the reactive element addition is to effectively block the outward diffusion of cations in some cases leading to a reversal of the growth mechanism. In an attempt to understand the effects of reactive element additions on transport in chromium oxide have selected model systems, sintered chromia polycrystals and sintered chromia polycrystals doped with a reactive element, in this case yttria, and have studied the transport of cations and anions using tracer diffusion techniques. We present measured values of cation and anion bulk and grain boundary diiffusion in these model systems. The results are compared with earlier cation diiffusion D, 3,4,101 and anion diffusion measurements D, 1 l/. We then investigate the relation of these results with results from thermal oxidation experiments

2 - EXPERIMENTAL TECHNIOUES

S a m ~ l e Preuaration -Transport of cation and anions was studied using tracer diffusion techniques. In the case of cations, the radioactive tracer S'Cr was used. To study anion diffusion 'Q was used coupled with secondary ion mass spectromemy (SIMS) depth profiling. For this investigation, model systems , sintered Cr203 and Cr203-Y203 alloys were employed. The ceramics were prepared using the citrate process and spray drying. Pellets were sintered at the Cr/Cr,O, equilibrium at 1550°C followed by a pre-diffusion anneal at 110O0C to set the stoichiomefq. The cation or anion tracer was deposited and diffusion anneals were carried out at 1100°C at the Cr/Cr,03 equilibrium. The details of this process is described in Ref. 110, 121.

Denth ~ r o f i l i n ~ -The procedure used for the cation diffusion experiments can be found in Ref. 1101. After annealing, the samples intended for anion diffusion experiments were removed from the stainless steel tubes and loaded into a VG SIMSLAB secondary ion mass specuome- ter (SIMS) for analysis. A 30 keV Ga' ion source was used for sputter depth profiling and analytical ion imaging. Samples were initially surveyed in the ion induced secondary electron image mode to identify a region that was smooth and free of surface defects. Depth profiling was initiated using a 20 or 200 nA Ga'beam current at 30 keV. An electronic gate was employed to minimize crater edge effects. The '60-

Article published online by EDP Sciences and available at http://dx.doi.org/10.1051/jphyscol:1990186

(3)

Cl-552 COLLOQUE DE PHYSIQUE

and the '80- signals were recorded as a function of time. Typical sputtering times ranged from 2.41 X 10% to 8.82 X 10's. The raster was (2.1 X 1W2) X (2.1 X 10-7 cm2. A depth p d l e from such an area includes effects from both bulk and grain boundary diffusion. At least two depth profiles were acquired from each sample. Seven depth profiles were analyzed for the pure Cr20, samples and four depth profiles were analyzed for the Cr,O,-Y,O, samples.

After depth profiling, sedondary ion images were recorded from the bottom of the depth profile crater to establish the spatial distribution of '80 relative to 'Q. Topographical contrast was removed from the secondary ion images by normalization to the sum of the '60- and the '80- images.

3 - DATA ANALYSIS AND RESULTS M

-

-

Our analysis follows the p~oneering work of Atkinson and Taylor /4/ and is the same as employed in an earlier work 1101. We consider the transport of tracer to occur along three paths: (1) Diffusion in the volume of the grains (concentration CV, diffusion coefficient D). This component produces the steeply falling part of the depth profile near the surface. (2) Diffusion along fast grain boundaries followed by leakage into the volume (concentration C diffusion coefficient D'). This component produces the long "tail" on the depth profile. (3) Diffusion along slow grain boundaries follfowed by leakage into the volume (concentration C,, diffusion coefficient D,). The motivation for introducing the concept of slow and fast boundaries instead of the interpretation used in the past/lO, 131 will be discussed later. The conaibution from slow boundaries fills out the junction between bulk diffusion and fast boundaries.

In general, D << Ds < D'.

The roncentration of tracer is the sum of the three contributions

C, a function of depth y, and time t, is given by 1141

for cations and

for anions. The A's are fitting parameters, with A, = I12(Df)'n, and A,and A, are related to the parameters for short-circuit diffusion by 1151:

and

The 6 s are the widths of the respective short circuiting paths.

Equations 1-2 were fitted to the depth profile data using the MINPACK-I unconstrained, non-linear least squares fitting algorithm 1161. The diffusion parameters were obtained from eqs. 3 and 4. The result of a typical fit to the experimental data is shown in Fig. 1 for cation diffusion in pure chromia and Fig. 2 for anion diffusion in pure chromia. The total fit, i.e. the sum of the three contributions, is plotted along with the contributions from bulk, slow, and fast boundaries.

For both cations /10/ and anions /12/, the bulk diffusivity was determined by non-linear least squares fitting of the time dependence of the parameter A,. Similarly, grain boundary diffusivities were determined by non-linear least squares fitting of the time dependence of the parameter A,.

Validitv of the data

The validity of this analysis depends on the tail in the depth profile being due entirely to grain boundary diffusion, and not to some artifact such as cracks or porosity that allow flow of the tracer deep into the sample, or to uneven removal of material in the sputtering process. To verify this, we have carried out analytical ion microscopy (AIM) at the end of depth profiling. Figure 3a shows a secondary electron image of the region under investigation. A single Cr20, grain is visible at center. Figure 3b is the 'Q-/C60-+1803 image and Figure 3c is the 180-/

('60-+"0-) image. Clearly the fastest transport route in this figure is the triple point at right center. It is also evident that the grain boundary surrounding the grain is rich in '80. From other AIM images that were acquired, we observed that all grain boundaries do not exhibit the same diffusivity. In fact a spectrum of diffusion coefficients is to be expected. In light of these AIM results, we may consider our results L0

represent grain boundary diffusion. In addition, using the SIMS, it was possible to obtain depth profiles from the interior of a single grain.

Such depth profiles exhibited only a contribution from bulk diffusion. These result verify that the tail is due to grain boundary diffusion.

(4)

0 Experimental

- T-l Fit

- -

BUL

-.- SlolBoUndarh

-

-

Fad Boundaries

Figure I .

Figure 2 .

~ e p f h (pm)

Cation penetration plot for pure chronua annealed at 1100°C.

Anion penetration plot for pure chromia annealed at IIOO°C.

Figure 3. %'O (at left), %"0 (at center), and ion induced secondary electron image from bottom of depth profile crater in pure chromia. In the SIMS images, concentration increases with intensity.

(5)

COLLOQUE DE PHYSIQUE

The results from the experiments on the sintered samples are presented in Table I for cation diffusion and Table I1 for anion diffusion.

Table I. Cation d~%fusivities measured at llOO°C for bulk~low, and fast grain boundaries. *from ref 141.

Pure chromia 3.7 X 1G-l8 3.1 X 1@z2 (8.6rt2.3) X ~ o - ~ O

9.6 X 11318

*

3.4 X l@22

*

Chromia doped with yttria 2.7 11318 3.0 X 1 f 1 2 (2.3S.7) X l0-l9

Table II. Anion dz%f1(~1'vilies measured a! llOO°C for bulk~low, and fasf grain boundaries.

D (cm2/s) D'S (cm3/s) 4 6 , (cm3/s)

Pure chromia 8.3 X l@14 2.1 l@17

Chromia doped with yttria 1.5 10-12 2.2 X 1@18 7.2 X l@17

4 -DISCUSSION

Relation of the oresent results to transuort in m w i n e oxides.

If oxidation proceeds by diffusion along short circuit paths like grain boundaries in the oxide layer, the equation that describes the usual diffusion controlled oxidation kinetics.

where x is oxide thickness, t is time, and kp is the so-called parabolic rate constant, still holds, but the expression fork, must be modzed to reflect the fact that grain-boundary diffusion is the dominant mechanism of mass transport in the oxide/l7/:

In Eq. 6, D' is the grain boundary diffusion coefficient, 6 i s the effective grain boundary width, and g is the average grain diameter. If cation diffusion is the rate controlling mechanism, the constant in Eq. 6 is of the order 50. the exact value depends on the oxygen partial pressure dependence of D'& The parabolic rate constant is a function of both grain-boundary diffusivity and oxide grain size.

To compare the present results from Table Il and literature data/lO, 12,18,19,20,21,22/ for growth of Cr203 /18,19,20,21,22/ with cation and anion diffusion measurements /10,12,23/, we convert the experimental values of kp to values of D'6using the procedure described by Park et a1./10/ In that work, the assumption was made that the contribution of anion diffusion to the growth of C r p , films was negligible.

Because the oxygen partial pressure dependence of anion diffusion in Cr,O, is unknown, it is not possible to convert values of kp to D'6 based on the relative magnitudes of anion and cation diffusivities described above. Therefore we calculate values of D'S assuming that cation diffusion is the operative mechanism and compare the results with values of D'dmeasured from cation and anion diffusion experi- ments.

Figure 4 shows a plot of D'6versus inverse temperature for the available literature data. Also included in Fig. 4 are the D'6values from cation /10/ and anion 112,231 tracer diffusion studies. It is important to note that the values for D' 6deduced fmm k, measurements are spread over 2 orders of magnitude. Also, all values for D'6, except those deduced from the ion implantation experiments of King and Grabowski /22/ and Pivin et alJ19/, are higher than the values measured for cation grain boundary transport in Cr,O, and Cr$, doped with Yz03 at 1 100°C. The data of Cotell et al. demonstrate the the effect of Y addition to Cr is to reduce the oxide growth rate, consistent with anion diffusion. In previous work, we observed that reactive element additions had the effect of modifying the oxide mocrosuucture.

Specifically, we observed that oxides grown on addition-free Cr was almost uniformly porous as revealed by rapid (nearly infinite diffusiv- ity) penetration of ='Cr through microns of oxide after only 15 min. anneal at 500°C. Conversely, in Cr with Y additions, the oxide was observed to be dense and behave like sintered material. Cation grain boundary diffusion is far too slow to explain the observed oxide growth rates in unimplanted samples. However, the D'6values deduced from oxidation experiments on unimplanted samples /18,20,21/ are consistent with the current anion grain boundary diffusion data We would propose therefore that the effect of the reactive element at low concentrations, like those obtained in alloys, the effect of the reactive element is to modify the oxide microstructure such that oxidation is ratc controlled bv nrain boundarv difffusion of anions.

(6)

Figure 4 . Grain boundary cation and anion diffusivities in chromia measured using tracer techniques and deducedfrom oxidation experiments.

The oxidation rates of the implanted samples 119,221 are consistent with rate control by cation grain boundary diffusion. The addition of a reactive element, either by alloying of by implantation slows the oxide growth rate. Figure 4 would suggest that the consequence of adding the reactive element is to change the relative contributions of cation and anion diffusion to oxide growth, the limits apparently being rate controll by anions when no reactive element is present and rate controU by anions when the largest reduction in growth rate is observed.

In the past, we have analyzed the results of oxidation experiments by Caplan and Sproulej24l in terms of cation grain boundary diffusion 110, 12.22.251. Because Caplan and Sproule measured the growth rate of the slowest growing oxide grains on their samples, they speculated that growth of these grains was controlled by bulk diffusion. In light of ?he current anion diffusion results, we have reanalyzed the Caplan and Sproule data assuming that growth of these grains is controlled by bulk diffusion of anions. This result along with other measured anion buUc diffusivities are plotted in Fig. 5 11 1.12.231. Note the result of Caplan and Sproule at 1 100°C lies between the measured anion diffusivities of Hagel 1231 and of King and Park I l Y . The temperature dependence of the diffusivities deduced by Grahm et al. are in remarkable agreement with the diffusivities deduced from the data of Caplan and Sproule. Our analysis of Caplan and Spmule's data assumes that anion diffusion has the same oxygen partial pressure depencence as cation diffusion. The cation diffusion results of Park et al./10/ and Atkinson and Taylor/4/are also shown in Fig. 5. Comparison of these results indicate that in the bulk, as in the grain boundaries, cations diffuse orders of magnitude slower than anions.

From this work we conclude that anions diffuse orders of magnitude faster that cations in chromia. There is a strong possibility that growth of chromia is controlled by inward diffusion of anions along grain boundaries. Typical scales on addition free alloys are porous and grow at rates faster than predicted from anion diffusion. At low reactive-element concentrations, like those obtained in alloys, the effect of the reactive element is to modify the oxide microstructure such that oxidation is rate controlled by grain boundary difffusion of anions. At higher concentrations, like those obtained in the ion implantation work of Pivan et al. and King and Grabowski 119,221, the effect is to change the relative contribution of cation and anion diffusion to the limit where growth is controlled by cation grain boundary diffusion.

Finially, grain boundary transport data has been analyzed assuming the o b s e ~ e d penetration plots to be the sum of three conmbutions:

transport along bulk, slow, and fast grain boundaries. Although this model is sufticient to analyze the data, analytical ion microscopy reveals that aiple points in chromia are the sites of the most rapid transport, i.e. fast boundaries, and other boundaries exhibit a spectrum of diffusivi- ties.

5 - ACKNOWLEDGEMENTS

The analytical ion microscopy and SIMS depth profiling were canied out at Argonne National Laboratory. The authors are grateful to C.L.

Wiley for help in the depth profiling, and to S. J. Rothman for helpful discussions. Work supported under the auspices of the U. S. Depan- ment of Energy by the Lawrence Livenore National Laboratory under conaaci number W-7405-Eng48

(7)

COLLOQUE DE PHYSlQUE

Figure 4 . Bulk cation and anion diffusivities in chromia measured using tracer techniques and deducedfrom oxidation experiments.

Pfeil, L. B., Report Patent Number 459,848 (1937).

Hagel, W. C. and U. Seybold, J. Electrochem. Soc. m ( 1 9 6 1 ) 1146.

Hoshino, K. and N. L. Peterson, J. Amer. Ceram. Soc. & (1983) C202.

Atkinson, A. and R. I. Taylor, NATO AS1 Series B, v01 129, eds G. Simkovich and V. S. Stubican 1984) 285.

Whittle, D. P. and J. Stringer, Phil. Trans. R. Soc. Lond. (1980) 309.

Goncel, 0. T., D. P. Whittle and J. Stringer, Oxid. MeL fi(1981) 287.

Wright, I. G. and J. Stringer, Metallography 6 (1973) 65.

Wright, I. G. and B. A. Wilcox, Met Trans. 5 (1974) 283.

Wright, I. G. and B. A. Wilcox, Oxid. Met 8 (1974) 283.

Park, J. H., W. E. King and S. J. Rothman, J. Amer. Ceram. Soc.

212

(1987) 880.

Grahm, M. J., J. I. Eldridge and D. F. Mitchell, Fiftv Years of theReactive Element Effect, v01 39, eds W. E. King (Trans Tech 1989)

King, W. E. and J. H. Park, Mater. Res. Soc,, v01 122, eds M. H. Ym, W. A. T. Clark and C. L. Briant (Materials Research Society Reno, NV 1989) 193.

Atkinson, A. and R. I. Taylor, Phil. Mag. (1981) 979.

Crank, J. The Mathematics of Diffusion, (Clmndon Press Oxford, England 1956).

LeClaire, A. D., Brit. J. Appl. Phys. u ( 1 9 6 3 ) 352.

Mor6, J. J., B. S. Garbow and K. E. Hillstrom, User Guide for MINPACK-1, Argonne National Laboratory Report ANL-80-74 (1980).

Matsunaga, S. and T. Homma, Oxid. Met. Q (1976) 361.

Lillemd, K. P. and P. Kofstad, Oxid. Met. (1982) 127.

Pivin, J. C., D. Delaunay, C. Roques-Carmes, A. M. Huntz and P. Lacombe, Corros. Sci. a ( 1 9 8 0 ) 351.

Cotell, C. M., K. Pnybylski and G. J. Yurek, Fundamental As~ects of Hieh Temuerature Corrosion-11, eds D. A. Shores and G. J.

Yurek (The Elecaochemical Society Pennington, NJ 1986) 103.

Hindam, H. and D. P. Whittle, Oxid. Met. 18 (1982) 245.

King, W. E. and K. S. Grat~wski, Environmental demadation of ion and laser beam treated surfaces 111: Hiph temoerature oxidation and hot corrosion, eds G. S. Was and K. S. Grabowski (TMS Warrendale, PA 1988) 277.

Hagel, W. C., J. Amer. Ceram. Soc. 48 (1965) 70.

Caplan, D. and I. Sproule, Oxid. Met. 5 (1975) 459.

King, W. E., K. S. Grabowski and P. M. ~ h d o , Norman L. Peterson Memorial Svmwsium - Oxidatron of Metals and Associated Mass Trans~ort, eds M. A. Dayananda, S. J. Rothman and W. E. King (The Metallurgical Society, Inc. Warrendale, PA 1987)

Références

Documents relatifs

and Ag in Pb/17/ which are two typical fast diffuser systems. Contrary t o grain boundary self diffusion which proceeds by vacancy jump process, it is attractive to assume

L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des

L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des

penetration along grain boundaries. The grain boundary diffusion coefficients deduced from the two plots differ by less than one order of magnitude. This expression is valid

Particularly, segregation effeccs which - - lay - - an important role in diffusion kine- tics can be used to describe a dynamic fractal dimension of grain boundaries at

— Measurements in this laboratory of cation diffusion in bicrystals, of silver injection in the grain boundary region, and of light-scattering and electron microscope observations

L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des

More precisely, the absence of enhanced intergranular diffusion under these experimental conditions means that grain boundary diffusion coefficients of cobalt and nickel