Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://www.issp.ac.ru/libm/papers/1999-30.pdf Äàòà èçìåíåíèÿ: Sun Jun 25 20:23:06 2006 Äàòà èíäåêñèðîâàíèÿ: Tue Oct 2 04:30:01 2012 Êîäèðîâêà:

PII:

Acta mater. Vol. 47, No. 2, pp. 385±395, 1999 # 1999 Acta Metallurgica Inc. Published by Elsevier Science Ltd. All rights reserved Printed in Great Britain S1359-6454(98)00381-4 1359-6454/99 $19.00 + 0.00

TRACER DIFFUSION OF Au AND Cu IN A SERIES OF NEAR ô 5 310001 SYMMETRICAL Cu TILT GRAIN BOUNDARIES
E. BUDKE1, T. SURHOLT1, S. I. PROKOFJEV2, L. S. SHVINDLERMAN2 and CHR. HERZIG1{
1 Institut fur Metallforschung, Universitat Munster, Wilhelm-Klemm-Str. 10, D-48149 Munster, õ õ õ õ Germany and 2Institute of Solid State Physics, Russian Academy of Sciences, 142432 Chernogolovka, Moscow Distr., Russia

(Received 15 June 1998; accepted 29 October 1998) AbstractïGrain boundary diusion of Au and Cu was measured in a series of Cu bicrystals with symmetrical near ô 5, á 36X98 310001 CSL tilt grain boundaries (GBs) using the radiotracer and the serial sectioning technique. The orientations of the bicrystals were very precisely determined with the Kossel technique where all three macroscopic parameters describing the orientations of the grains in the bicrystal were evaluated. The tilt angles ranged from 33.218 to 39.268. The GB diusion of the radiotracers 195 Au and 64Cu was measured as a function of tilt angle and temperature. In the investigated temperature range 1030±661 K the orientation dependence of both radiotracers shows a characteristic cusp not exactly at but slightly below the ideal ô 5 CSL GB. The Arrhenius parameters, activation enthalpy and frequency factor, determined from lower temperature data adopt a maximum, again slightly before the ideal ô 5 CSL GB. These features are discussed with respect to the accidental small twist and second tilt orientations and the corresponding dislocation network inherent in the investigated real GBs. With increasing temperature a negative deviation from a straight Arrhenius behaviour is observed. This result indicates a certain change in the GB structure in the temperature range above 800 K. # 1999 Acta Metallurgica Inc. Published by Elsevier Science Ltd. All rights reserved.

1. INTRODUCTION

A number of investigations were carried out in the last few decades to relate the behaviour of grain boundary (GB) diusion to the structure of GBs. The results of these works and the given interpretations are to some extent contradictory. On the one hand sharp narrow cusps of GB diffusivity and peaks of Arrhenius parameters are reported at certain GBs of high coincidence [1±5]. Grain boundaries of high coincidence and therefore of highly ordered atomic structure are expected to yield minima in diusivity and maxima in Arrhenius parameters due to the larger mean vacancy formation and migration enthalpies in comparison to GBs of low coincidence and of lower degree of perfection [6]. Unfortunately, the experiments supporting the CSL model were performed using solute GB diusion in combination with electron microprobe analysis. Due to the rather large required solute concentrations in the GBs in such an analysis, these studies were carried out, in fact, under conditions of chemical diusion. This sets serious obstacles to the interpretation of the deduced GB diusion parameters and the observed minima and maxima in sdDGB . It was argued [7]
{To whom all correspondence should be addressed. 385

that the observed minima might be related to the structural dependence of solute GB segregation which is included in the GB diusion parameters in type-B kinetics of solute diusion. On the other hand a possible continuous relationship between GB diusion and the macroscopic grain orientation of symmetrical [001] tilt GBs is supposed. Sommer et al. [8] investigated self-diusion along bonded symmetrical [001] Ag tilt GBs and found a continuous variation of dDGB with the tilt angle Y. The tilt angle Y describes the grain orientation of the two adjacent grains with respect to the [001] tilt axis. Ma and Ballu [9] studied Ag diusion in a series of [001] Au tilt GBs in thin ®lms at low temperatures using a surface accumulation technique in combination with Auger spectroscopy. Again, a continuous variation of dDGB was observed. These results were interpreted in terms of the structural unit model for diusion [10, 11]. This model considers the core of symmetrical [001] tilt GBs as binary arrangements of structural units. These structural units are derived from two nearby delimiting boundaries which are made up of one unique type of structural unit. Variations in the tilt angle Y lead to a continuous change in the fractional arrangement of both types of structural units of the delimiting boundaries. Therefore, GB diusion is supposed to change con-


386

BUDKE et al.:

TRACER DIFFUSION OF Au AND Cu IN Cu GBs

tinuously with the variation of Y. It should be pointed out that the distance in the tilt angle Y between the experimental points in the investigations of Sommer et al. [8] as well as of Ma and Ballu [9] were comparably large (5±108). These studies therefore yield no proof in favour of the structural unit model of GB diusion, if possible cusps turn out to be rather narrow. The aim of the present study is to overcome the previous obstacles for a clear decision between the two con¯icting considerations about the relation of diusion in oriented GBs and their individual atomistic structure. Therefore, using the radiotracer 195 Au, the diusion of Au was studied in the present paper along a series of symmetrical [001] tilt GBs of Cu with tilt angles Y in the range 33.21± 39.268 in the vicinity of the ideal S 5, Y 36X98 310001 tilt boundary. The precise radiotracer technique in combination with the serial sectioning method of the diusion zone enables us to obtain accurate GB diusion pro®les in a single GB. The application of carrier free 195Au radiotracer with very high speci®c activity and a high-eciency detector system allows very small concentrations of the diusant in the GBs. Thus any disturbance of the GB structure is almost excluded. Furthermore, the Au±Cu phase diagram [12] suggests weak chemical interactions between Au and Cu atoms. Eects of Au segregation therefore are expected to be rather small. Nevertheless, in order to check the reliability of our measured orientation dependence of Au GB diusion in Cu, the GB self-diusion was also studied at one temperature with the 64Cu radiotracer of short half-life, which impedes a full systematic study with this Cu radiotracer.

2. EXPERIMENTAL

The Cu bicrystals were grown in a graphite mould in high purity argon atmosphere by directional crystallization using Cu single crystals of high purity (nominal 99.995%) Cu. It should be pointed out that the purity of the bicrystals used for the diusion measurements was higher than the nominal purity of the initial Cu material since the technique of growing bicrystals acts like a zone re®nement of the Cu material. Bicrystals of 16 á 3 mm2 in cross-section and with lengths up to 200 mm were grown. The orientations of the grains of each bicrystal were determined by Kossel-technique measurements. The advantage of this technique is that it can measure all three macroscopic parameters describing the orientation of the adjacent grains with very high precision. Table 1 contains the resulting tilt angles Y with respect to the [001] axis, the twist angles F with respect to the [310] axis and the second tilt angles C with respect to the 130 axis. The orientation of the GB plane was carefully

adjusted during the bicrystal growth process but was not explicitly remeasured. Quadratic diusion samples (6 á 6 mm2) were cut from the bicrystals by spark erosion and cleaned by chemical etching in nitric acid. One front (001) surface of each specimen was mechanically polished by standard metallographic procedures and ®nished by electrolytical polishing. The electrolytical polishing procedure removes the dislocation enriched surface layer resulting from the mechanical polishing process. The samples were sealed in quartz ampoules under puri®ed argon atmosphere (I102 Pa) and were annealed ®rst at 1173 K for two days and furthermore under the conditions (temperature T, time tïat least, however, for two weeks) of the intended diusion anneals in order to achieve quasi-equilibrium GB segregation of spurious impurities and to prove the stability of the GB. The GB quality was checked by optical microscopy. Slightly curved GBs indicate changing inclinations along the GB leading to dierent crystallographic orientations of the GBs which is accompanied by a change in the GB structure. Such specimens were discarded. The carrier-free radiotracer 195Au (half-life 183 days) was used in this work in the form of a HCl solution with an initial speci®c activity of about 63 á 106 Bq/mg. For the GB self-diusion measurements, the radiotracer 64Cu (half-life 12.7 h) was produced from the nuclear reaction 63 Cu(n,g)64Cu with initial speci®c activity of 0.26 á 106 Bq/mg by means of exposing copper with natural isotope composition to neutron radiation at a reactor (GKSS Forschungszentrum, Geesthacht, Germany). Immediately after radiation the radioactive Cu was delivered within a few hours to our laboratory in Munster and was dissolved in diluted õ nitric acid. Both 195Au and 64Cu radiotracers were dropped onto the polished surface of the Cu speciTable 1. Orientations of the investigated Cu bicrystals with orientations near the S = 5, Y = 36.98 310001 CSL GB, determined by Kossel-technique measurements. Y: tilt angle with respect to the [001] axis, F: twist angle with respect to the [310] axis, C: tilt angle with respect to the 130 axis No. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 Y (8 ) 33.2 33.6 35.1 35.1 35.8 36.0 36.2 36.2 36.5 36.6 36.9 37.2 37.4 37.5 37.7 38.1 39.2 1 0 0 9 8 0 6 6 3 7 7 0 5 7 4 2 6 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 0.18 0.08 0.04 0.03 0.20 0.29 0.37 0.37 0.09 0.17 0.02 0.29 0.03 0.10 0.08 0.30 0.26 F (8) 0.80 0.60 0.05 0.22 0.95 0.25 0.05 0.55 0.10 1.85 1.27 0.60 1.27 1.30 1.80 0.30 0.50 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 0.15 0.15 0.01 0.04 0.10 0.10 0.10 0.25 0.15 0.20 0.06 0.20 0.01 0.10 0.05 0.35 0.45 C (8) 0.47 0.20 0.75 1.72 2.03 2.50 1.15 0.90 1.10 1.12 0.42 0.11 1.14 0.11 0.15 0.76 0.30 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 0.15 0.13 0.02 0.03 0.16 0.02 0.17 0.19 0.12 0.23 0.02 0.15 0.03 0.02 0.03 0.16 0.13


BUDKE et al.:

TRACER DIFFUSION OF Au AND Cu IN Cu GBs

387

mens and dried. The initial surface activity of each sample ranged from 2 á 104 to 3 á 104 Bq for the Au radiotracer and from 105 to 106 Bq for the Cu radiotracer, depending in the latter case on the chosen diusion annealing time of the sample. Usually several bicrystals were annealed in the same ampoule under puri®ed argon atmosphere for the actual diusion annealings. The Au diusion measurements were performed at six temperatures in the temperature range 1030±661 K and the selfdiusion measurements at the temperature 919 K. The temperatures were measured with Ni/NiCr thermocouples and controlled within 21 K. After the diusion annealings the specimens were reduced in diameter to eliminate the eect of radial diusion and were sectioned on a microtome. The thickness of each section was determined from its weight, diameter and the density of Cu. The intensity of the decays of 195Au and 64Cu in each section was detected with a liquid scintillation counter of very low background and automatic sample changer.
3. RESULTS

and (b) and 64Cu GB diusion pro®les in Cu bicrystals are shown in Fig. 2. The pro®les were plotted as ln " vs y6/5. The GB diusion related tails were c ®tted by straight lines. From the slopes d ln "ad y6a5 c the values of GB diusivity sdDGB (s: solute GB segregation factor; s 1 for GB self-diusion, d: GB width, DGB: GB diusion coecient) were calculated using the Suzuoka solution of the GB diusion equation, which implies instantaneous source conditions d ln " þp c p r þu sdDGB 2q D t þ 6a5 1 dy where p 1X6807, q 0X740, r 0X4916 and u 0X5084 in typical experimental conditions when the parameter 102 `b sdDGB a2D3a2 t1a2 `104 [13, 14]. These parameters change slightly for other ranges of b [13]. According to the B regime, the conditions b b 10 and a sda2Dt1a2 ( 1 are ful®lled, see Tables 2±7. The lattice diusion coecient D of Au in Cu follows the Arrhenius relations DAu T 8X0 á 10þ6 191X1 kJamol m2 as à exp þ RT in the temperature range 982 KrTr633 K and

The diusion time t for all studied temperatures was chosen so that all measurements were carried out in the type-B kinetic regime [13, 14]. Typical penetration pro®les of 195Au are shown in Figs 1(a)

2

Fig. 1. Tracer penetration pro®les of 195Au in symmetrical near S =5, Y = 36.98 310001 Cu tilt GBs after diusion annealing at T = 1030 K for t = 15 120 s (a) and at T = 780 K for t = 247 620 s (b).


388

BUDKE et al.:

TRACER DIFFUSION OF Au AND Cu IN Cu GBs

used for evaluating the GB self-diusion in Cu has been determined in the temperature range 905 KrTr574 K [16]. The resulting GB diusion parameters sdDGB of 195Au and 64Cu in Cu are listed together with other relevant experimental parameters in Tables 2±7. The high quality of the measured diusion pro®les in Figs 1(a) and (b) and Fig. 2 shows as in our study on Al bicrystals [17] that the underlying mathematical conditions of the Suzuokas solution can be exactly realized in GB diusion experiments on bicrystals. The uncertainty in the slope of the GB diusion pro®les is less than 1% in most cases, i.e. the GB diusion parameter is determined extremely accurately for this set of bicrystals. Such good linearity over several orders of magnitude in concentration is usually not observed in polycrystalline samples due to the presence of dierently behaving GBs [18] and due to the occurrence of some slight GB motion [19, 20]. 3.1. Orientation dependence of the grain boundary diusion parameter sdDGB The dependence of the GB diusion parameter sdDGB on the tilt angle Y in the temperature range 1030±661 K is shown for 195Au in Figs 3(a) and (b) for four temperatures. At smaller tilt angles sdDGB Y increases monotonously with increasing tilt angle Y. At Y 36X538 a characteristic minimum of sdDGB is observed, slightly before the ideal S 5, Y 36X98 310001 GB. In the immediate vicinity of the ideal S 5 CSL GB the GB diusivity seems to increase drastically within only a few tenths of a degree in the tilt angle variation. Behind the ideal S 5 GB, sdDGB seems to decrease with increasing tilt angle Y. The variation of sdDGB Y is not symmetrical to the S 5 GB. The observed cusp at Y 36X538 becomes more pronounced with decreasing temperature. Qualitatively the same characteristic orientation dependence of dDGB Y is also observed for the GB self-diusion at T 919 K, see Fig. 4. Again, dDGB Y reveals a cusp not exactly at but slightly below the ideal S 5 CSL GB.
Table 3. Results and experimental parameters of 195Au diusion along [001] tilt GBs with dierent tilt angles Y in symmetrical near S = 5, Y = 36.98 310001 GBs. T = 978 K, t = 20 430 s, D = 5.03 á 10þ16 m2/s, a = sà7.80 á 10þ5 No. 1 4 5 6 7 9 10 14 15 16 17 Y (8) 33.21 35.19 35.88 36.00 36.26 36.53 36.67 37.57 37.74 38.12 39.26 2 2 2 2 2 2 2 2 2 2 2 0.18 0.03 0.20 0.29 0.37 0.09 0.17 0.10 0.08 0.30 0.26 sdD 1.80 1.65 1.44 2.03 1.88 1.31 2.18 2.23 2.77 2.42 1.87 á á á á á á á á á á á
GB

Fig. 2. Tracer penetration pro®les of 64Cu in symmetrical near S = 5, Y = 36.98 310001 Cu tilt GBs after diusion annealing at T = 919 K for t = 34 560 s.

DAu T 5X4 á 10þ5 205X6 kJamol m2 as à exp þ RT

3

in the high temperature range 1350 KrTr993 K [15]. The Arrhenius relation DCu T 1X0 á 10þ5 196X8 kJamol m2 as à exp þ RT

4

Table 2. Results and experimental parameters of 195Au diusion along [001] tilt GBs with dierent tilt angles Y in symmetrical near S = 5, Y = 36.98 310001 GBs. T = 1030 K, t = 15 120 s, D = 2.01 á 10þ15 m2/s, a = sà4.53 á 10þ5 No. 1 3 4 5 6 6 7 9 10 14 15 16 17 Y (8 ) 33.21 35.10 35.19 35.88 36.00 36.00 36.26 36.53 36.67 37.57 37.74 38.12 39.26 2 2 2 2 2 2 2 2 2 2 2 2 2 0.18 0.04 0.03 0.20 0.29 0.29 0.37 0.09 0.17 0.10 0.08 0.30 0.26 sdD 2.45 2.37 2.52 2.18 2.65 2.65 2.40 1.86 3.24 3.62 3.42 3.38 2.72 á á á á á á á á á á á á á
GB

(m2/s)

b 11 11 11 9.8 12 12 11 8.4 14.6 16 15 15 12

10 21.93% 10þ1921.42% 10þ1921.39% 10þ1922.17% 10þ1921.41% 10þ1921.78% 10þ1920.96% 10þ1921.03% 10þ1920.85% 10þ1921.85% 10þ1921.05% 10þ1921.19% 10þ1920.84%

þ19

(m3/s) 2.0 3.1 3.7 2.7 1.6 1.2 3.5 4.2 6.1 4.0 3.6

b á á á á á á á á á á á 10 10 10 10 10 10 10 10 10 10 10
4 4 4 4 4 4 4 4 4 4 4

10 20.98% 10þ1921.77% 10þ1922.37% 10þ1920.84% 10þ1921.27% 10þ1921.40% 10þ1920.95% 10þ1921.07% 10þ1922.32% 10þ1921.49% 10þ1921.80%

þ19


BUDKE et al.:

TRACER DIFFUSION OF Au AND Cu IN Cu GBs

389

Table 4. Results and experimental parameters of 195Au diusion along [001] tilt GBs with dierent tilt angles Y in symmetrical near S = 5, Y = 36.98 310001 GBs. T1=872 K, T2=868 K, t = 67 392 s, D1=2.89 á 10þ17 m2/s, a1=sà1.79 á 10þ4, D2=2.56 á 10þ17 m2/s, a2=sà1.90 á 10þ4 No. 1 2 3 4 5 6 6 7 8 9 10 12 14 16 17 T (K) T T T T T T T T T T T T T T T
1 1 2 2 1 2 2 2 1 1 1 2 2 1 2

Table 6. Results and experimental parameters of 195Au diusion along [001] tilt GBs with dierent tilt angles Y in symmetrical near S = 5, Y = 36.98 310001 GBs. T = 661 K, t1=5 673 360 s, t2=5 680 595 s, D = 6.42 á 10þ21 m2/s, a = sà1.31 á 10þ3 No. 1 3 4 5 6 7 7 9 10 14 15 16 17 Y (8 ) 33.21 35.10 35.19 35.88 36.00 36.26 36.26 36.53 36.67 37.57 37.74 38.12 39.26 2 2 2 2 2 2 2 2 2 2 2 2 2 0.18 0.04 0.03 0.20 0.29 0.37 0.37 0.09 0.17 0.10 0.08 0.30 0.26 sdDGB (m3/s) 3.54 5.83 8.52 7.65 6.12 3.02 2.64 1.51 9.50 1.98 2.53 1.28 7.71 á á á á á á á á á á á á á 10þ2120.58% 10þ2120.84% 10þ2120.60% 10þ2121.02% 10þ2120.50% 10þ2120.66% 10þ2121.48% 10þ2120.49% 10þ2120.70% 10þ2021.28% 10þ2021.09% 10þ2020.87% 10þ2120.57% t (s) t t t t t t t t t t t t t
1 2 2 1 1 1 2 1 1 2 2 1 1

Y (8) 33.2 33.6 35.1 35.1 35.8 36.0 36.0 36.2 36.2 36.5 36.6 37.2 37.5 38.1 39.2 1 0 0 9 8 0 0 6 6 3 7 0 7 2 6 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 0.18 0.08 0.04 0.03 0.20 0.29 0.29 0.37 0.37 0.09 0.17 0.29 0.10 0.30 0.26

sdD 8.61 7.28 7.09 8.03 1.23 8.40 8.51 8.44 1.15 7.01 1.46 8.50 1.17 1.41 9.38 á á á á á á á á á á á á á á á

GB

(m3/s)

b 1067 903 1054 1195 1519 1248 1266 1255 1468 461 1809 1264 1732 1747 1395

b 1.4 2.4 3.5 3.1 2.5 1.2 1.1 6.2 3.9 8.1 1.0 5.6 3.2 á á á á á á á á á á á á á 10 10 10 10 10 10 10 10 10 10 10 10 10
6 6 6 6 6 6 6 5 6 6 7 6 6

10þ2021.59% 10þ2021.52% 10þ2021.48% 10þ2020.83% 10þ1920.93% 10þ2020.60% 10þ2020.62% 10þ2020.61% 10þ1922.63% 10þ2020.65% 10þ1920.73% 10þ2021.33% 10þ1920.77% 10þ1920.45% 10þ2020.78%

3.2. Orientation dependence of the grain boundary diusion Arrhenius parameters The temperature dependence of 195Au GB diusion of all bicrystals is shown in Figs 5(a)±(d) in Arrhenius coordinates. In the whole investigated temperature range the measured sdDGB values of the bicrystals, with the exception of bicrystal No. 14, do not follow the linear Arrhenius relation QGB sdDGB T sdDGB 0 à exp þ 5 RT (sdDGB 0 : pre-exponential factor, QGB: eective activation enthalpy of GB diusion). A rather concave curvature is observed at high temperatures. Therefore, the experimentally determined points were ®tted by Arrhenius lines only in the lower temperature range, see Figs 5(a)±(d). Due to the high precision of sdDGB these ®ts are regarded as meanTable 5. Results and experimental parameters of 195Au diusion along [001] tilt GBs with dierent tilt angles Y in symmetrical near S = 5, Y = 36.98 310001 GBs. T = 780 K, t = 247 620 s, D = 1.29 á 10þ18 m2/s, a = sà4.42 á 10þ4 No. 1 3 4 5 6 6 7 7 9 9 10 11 12 12 13 14 15 16 16 17 Y (8 ) 33.21 35.10 35.19 35.88 36.00 36.00 36.26 36.26 36.53 36.53 36.67 36.97 37.20 37.20 37.45 37.57 37.74 38.12 38.12 39.26 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 0.18 0.04 0.03 0.20 0.29 0.29 0.37 0.37 0.09 0.09 0.17 0.02 0.29 0.29 0.03 0.10 0.08 0.30 0.30 0.26 sdD 2.96 4.52 4.53 5.44 3.99 3.53 2.39 3.01 1.71 2.21 5.13 6.08 5.34 7.77 7.33 6.19 8.93 5.85 6.51 5.18 á á á á á á á á á á á á á á á á á á á á
GB

ingful. The resulting pre-exponential factor sdDGB 0 and the eective activation enthalpy QGB for each orientation Y are listed together with the corresponding temperature range of these ®ts in Table 8. The dependence of QGB and sdDGB 0 on the tilt angle Y of the bicrystals is shown in Figs 6(a) and (b), respectively. Consistent with the sdDGB results, the Arrhenius parameters decrease at smaller tilt angles with increasing tilt angle. A sharp maximum is adopted at Y 36X538 slightly before the ideal S 5 CSL GB and in the immediate vicinity of the S 5 GB the Arrhenius parameters decrease drastically (bicrystal No. 10, Y 36X678). Behind the high coincidence GB, QGB and sdDGB 0 increase with increasing tilt angle Y. In our opinion this particular behaviour of QGB and sdDGB 0 re¯ects information of general validity on GB diusion in bicrystals but also some special features, which depend on the particular properties of the presently investigated set of bicrystals, as will be discussed below.
4. DISCUSSION

(m /s) 2.0 3.1 3.1 3.7 2.7 2.4 1.6 2.0 1.2 1.5 3.5 4.2 3.7 5.3 5.0 4.2 6.1 4.0 4.5 3.6

3

b á á á á á á á á á á á á á á á á á á á á 104 104 104 104 104 104 104 104 104 104 104 104 104 104 104 104 104 104 104 104

4.1. Estimation of the structure dependence of Au GB segregation in Cu bicrystals Since the Au GB diusion in Cu bicrystals was studied in the type-B kinetic regime, the orientation dependence of Au GB segregation on the measured GB diusion parameters sdDGB and the determined Arrhenius parameters sdDGB 0 and QGB
Table 7. Results and experimental parameters of 64Cu diusion along [001] tilt GBs with dierent tilt angles Y in symmetrical near S = 5, Y = 36.98 310001 GBs. T = 919 K, t = 34 560 s, D = 6.49 á 10þ17 m2/s, a1=1.67 á 10þ4 No. 6 7 9 10 14 Y (8) 36.00 36.26 36.53 36.67 37.57 2 2 2 2 2 0.29 0.37 0.09 0.17 0.10 sdD
GB

10 20.49% 10þ2020.80% 10þ2020.72% 10þ2020.56% 10þ2021.28% 10þ2020.69% 10þ2020.72% 10þ2021.57% 10þ2020.52% 10þ2021.23% 10þ2020.68% 10þ2020.82% 10þ2020.64% 10þ2020.61% 10þ2020.60% 10þ2020.54% 10þ2020.65% 10þ2020.64% 10þ2020.96% 10þ2020.71%

þ20

(m3/s) 3.4 2.9 3.0 3.6 3.7

b á á á á á 10 10 10 10 10
2 2 2 2 2

6.61 á 10 22.3% 5.68 á 10þ2020.83% 5.85 á 10þ2020.68% 7.01 á 10þ2020.58% 7.25 á 10þ2020.70%

þ20


390

BUDKE et al.:

TRACER DIFFUSION OF Au AND Cu IN Cu GBs

Fig. 3. Dependence of the diusion parameter sdDGB of 195Au diusion along near S =5, Y = 36.98 310001 Cu tilt GBs on the tilt angle Y at T = 1030 and 872 K (a), T = 780 and 661 K (b).

HGB Hs (HGB: activation enthalpy of GB diusion, Hs: solute GB segregation enthalpy) have to be considered. Comparing the orientation dependence of the Au GB diusion parameters sdDGB in Cu, Figs 3(a) and (b), qualitatively with the orientation dependence of GB self-diusion, Fig. 4, it is evident that the GB diusion parameter show the same characteristic orientation dependence for GB self- as well as for GB solute diusion of Au in Cu. This pro-

Fig. 4. Dependence of the diusion parameter dDGB of 64 Cu self-diusion along near S =5, Y = 36.98 310001 Cu tilt GBs on the tilt angle Y at T = 919 K.

vides convincing evidence that the observed sharp variation of sdDGB must be related to the orientation dependence of the GB diusion coecient DGB. Furthermore, the GB segregation of Au in Cu polycrystals was determined in the temperature range 526±450 K by measuring GB diusion in type-B and type-C kinetic regimes [14]. The temperature dependence of the segregation factor of Au in Cu leads to the Arrhenius parameters s0 0X88 and Hs þ9X7kJamol on the condition of Henrytype segregation behaviour. Hofmann et al. [21] studied the orientation dependence of GB segregation enthalpies of Si, P and C in symmetrical [001] Fe± 3.5% Si tilt GBs. The smallest values of Hs are observed at symmetrical GBs of high coincidence. An increasing deviation from the high coincidence orientation reveals increasing values of Hs; maximum values are obtained at general large angle GBs. Taking now into account that the determined Hs value of Au GB segregation in Cu polycrystals can be related to general large angle GBs [14], the variation of Hs with the tilt angle Y in the vicinity of the ideal S 5 CSL GB should not exceed 9.7 kJ/mol. As a consequence, the measured variation of QGB Y of at least 60 kJ/mol, Fig. 6(a), cannot be explained by segregation eects alone. A considerable part of this variation must be attributed to the structure dependence of GB diusion.


BUDKE et al.:

TRACER DIFFUSION OF Au AND Cu IN Cu GBs

391

Fig. 5. Temperature dependence of

195

Au diusion along near S =5, Y = 36.98 310001 Cu tilt GBs in Arrhenius coordinates.

4.2. Structure dependence of Au and Cu GB diusion in Cu bicrystals The orientation dependence of GB diusion in the tilt angle range 33X218Y36X538 indicates a cusp of sdDGB and maxima of Arrhenius parameters, as determined from the lower temperature

data, in the vicinity of the S 5 CSL GB, see Figs 3 and 6. This behaviour can be well explained in terms of the CSL model. Grain boundaries of high coincidence and therefore of highly ordered atomic structure are expected to yield minima in diusivity and maxima in Arrhenius parameters due


392

BUDKE et al.:

TRACER DIFFUSION OF Au AND Cu IN Cu GBs

Fig. 6. Dependence of the eective activation enthalpy QGB (a) and the pre-exponential factor (sdDGB)0 (b) of 195 Au diusion along near S =5, Y = 36.98 310001 Cu tilt GBs on the tilt angle Y. The Arrhenius parameters are determined from lower temperature diusion data.

to the larger mean vacancy formation and migration enthalpies in comparison to GBs of low coincidence and of lower degree of perfection [6]. Small deviations in Y from the ideal CSL GB are compensated by additionally inserted GB dislocations [22]. Despite small deviations in the tilt angle, the GB core almost keeps the highly ordered atomic structure with the exception, however, of localized additionally inserted GB dislocations. These intrinsic dislocations obviously oer ``easy''
Table 8. Arrhenius parameters of 195Au dius with dierent tilt angles near symmetrical 310001 GBs at suciently low temperatures. not given if only two points of data were taken Arrhenius ®t No. 1 3 4 5 6 7 9 10 14 15 16 17 T (K) 661±872 661±780 661±780 661±872 661±868 661±872 661±872 661±872 661±1030 661±780 661±872 661±780 (sdD
GB)0

ion along Cu GBs S = 5, Y = 36.98 Errors of QGB are into account in the Q (kJ/mol)

(m3/s)
þ15

GB

2.1 3.9 4.9 9.0 1.3 6.9 1.2 7.3 5.0 9.6 2.5 2.0

á á á á á á á á á á á á

10 10þ15 10þ16 10þ16 10þ15 10þ15 10þ14 10þ16 10þ17 10þ17 10þ16 10þ15

72.9 2 2.9 73.8 60.2 63.9 2 5.4 67.5 80.3 2 7.4 87.3 2 0.1 61.9 2 1.0 43.2 2 1.9 45.3 54.3 2 0.1 68.5

paths for rapid atomic motion. With increasing tilt deviation DY from the ideal S 5 GB, the density of the additionally inserted GB dislocations increases and therefore the diusion ¯ux along the GBs also increases. The CSL model of diusion was supported by GB diusion measurements of Ge in oriented Al [111] tilt GBs with orientations near the ideal S 7, Y 38X28 123111 CSL GB [17], where a distinct cusp of sdDGB was clearly observed at all investigated temperatures and maxima of the Arrhenius parameters were obtained at the ideal S 7 GB. In the present study, however, the GB diusion parameter increases drastically within a few tenths of a degree of Y in the immediate vicinity of the ideal S 5 CSL GB, at Y b 36X538. This enhancement of sdDGB at Y 36X678 of Au is observed at all investigated temperatures, Fig. 3, and also for the GB self-diusion, Fig. 4. Furthermore, the orientation dependence of the Arrhenius parameters reveals a drastic decrease slightly before the ideal S 5 CSL GB. It has to be noted that all individual samples of a given orientation were cut from the corresponding same as-grown bicrystal. A critical look at the determined orientation parameters of the used bicrystals, Table 1, reveals that sdDGB is enhanced and QGB and sdDGB 0 are small for those bicrystals with comparatively large twist angles F. Especially the twist angle F of bicrystals No. 10 (Y 36X678, F 1X858) is the largest one in the presently used set of Cu bicrystals. But also the twist components of bicrystals Nos. 11 and 13±15, just behind the ideal CSL GB, are comparatively large. On the other hand, no signi®cant in¯uence on the GB diusion is observed from small deviations in the second tilt angle C of the bicrystals, Table 1. The dependence of sdDGB on the orientation angle Y appears rather asymmetric with respect to the ideal S 5, Y 36X98 orientation. In our opinion this fact results from the presently available set of bicrystals, which accidentally reveal somewhat larger twist angle contributions just for the orientations 36X678`Y`38X128. The features measured in the course of sdDGB and QGB can be tentatively explained by additionally inserted dislocation structures into the GBs due to the twist and second tilt components of the bicrystals. Vystavel et al. [23] studied the dislocation structures of Cu tilt GBs with small deviations in Y, F and C from the ideal S 5, Y 36X98 310001 tilt GB applying transmission electron microscopy. The investigated Cu bicrystals stem from the same origin as the Cu bicrystals of this study. Figure 7 shows schematically the observed GB dislocation pattern in the boundary plane of these bicrystals. The dislocation structure consists of an array of parallel edge dislocations lE and of a hexagonal dislocation net. The spacing between the parallel edge dislocations amount to a few 10 nm. The hexagonal dislocation net is composed of three


BUDKE et al.:

TRACER DIFFUSION OF Au AND Cu IN Cu GBs

393

Fig. 7. Schematic illustration of arrays of dislocation lines in the plane of a near S =5, Y = 36.98 310001 Cu tilt GB with small deviations in Y, F and C from the ideal CSL GB [23]. lE represents the line of edge dislocations forming an array of parallel dislocations; the hexagonal shaped dislocation net consists of the screw dislocation lS and of lD1 and lD2 of mixed screw and edge character. The orientation of the [001] axis, which is the diusional direction, is indicated.

CSL GB, on the other hand by additionally inserted ``easy'' diusion paths formed due to the twist angle deviations of individual bicrystals, especially in the immediate vicinity of the S 5 CSL GB. Both features superimpose to the presently observed orientation dependence of GB diusion which re¯ects the special ``®ne structure'' of the presently used set of bicrystals. This short discussion clearly demonstrates that results of GB diusion experiments in real bicrystals can be interpreted in a satisfying manner only, if as much as possible parameters describing the GB structure are accurately known. Furthermore, it is evident that previous diffusion investigations on less well de®ned bicrystals and GB diusion pro®les require a critical revision. 4.3. Temperature dependence of Au GB diusion in Cu bicrystals In Section 3.2 it was already pointed out that in the investigated temperature range the GB diusion parameter sdDGB of Au in Cu does not follow straight lines in the Arrhenius plots, Fig. 5, but reveals a downward curvature at temperatures T b 800 K. At suciently low temperatures we have approximated the temperature dependence of sdDGB by Arrhenius lines and the resulting Arrhenius parameters show a characteristic orientation dependence consistent with the orientation dependence of sdDGB . At higher temperatures, however, the temperature dependence of sdDGB is ambiguous. Due to the lack of a sucient number of experimental points it is not possible to decide between a straight Arrhenius behaviour or a continuous curvature of sdDGB . A continuous curvature in the Arrhenius plot would indicate a decrease of structural order in the GBs with increasing temperature. Molecular dynamic studies [25] support the idea of temperature induced disorder due to an increasing point defect concentration in the GBs. Despite of this, the GB retains the crystalline character even at higher temperatures. That study is not in con¯ict with the present experimental results. The observed deviations of sdDGB at higher temperatures indicate a smaller activation enthalpy of GB diusion which might be linked to larger defect densities in the GBs. The characteristic orientation dependence of sdDGB was observed even at high temperatures [T 1030 K, Fig. 3(a)] which still reveals a crystalline order of the GBs at these temperatures. On the other hand, the experimental points may be approximated by straight but dierent Arrhenius lines at high temperatures as well as at low temperatures, respectively. Such an approximation also leads to smaller activation enthalpies of GB diusion at high temperatures in comparison to the low temperature range, however, a sharp bend or a jump in the diusivity is expected at a certain temperature. This interpretation supports the idea of a phase transition-like change of the GB structure.

dierent dislocations: lS represents dislocations of almost screw character, lD1 and lD2 are dislocations of mixed screw and edge character. Vystavel et al. identi®ed the array of parallel dislocation to be responsible for the second tilt deviation C from the ideal S 5 CSL GB. The hexagonal dislocation net stems from the twist deviation F of the bicrystal from the ideal high coincidence GB. Since the authors determined the directions of the dislocation lines, it is possible to compare them to the diusion direction of the present GB diusion experiments, see Fig. 7. Assuming now that dislocation lines, which are aligned especially parallel to the [001] diusion direction, contribute to ``easy'' paths of rapid diusion, the array of edge dislocations lE, directed almost perpendicular to the [001] diusion direction, is expected to be rather ineective for the GB diusion. Since these dislocations are formed due to the second tilt deviation C of the GBs, no in¯uence of C on GB diusion is expected, in agreement with the present results. The dislocation lines lD1 and lD2 of the hexagonal net are oriented in such a way, that they can represent such paths of rapid diusion along the [001] direction in the GBs. These paths may be characterized by a comparatively smaller migration enthalpy and/or lower correlation of sequences of atomic jumps [24]. Since the hexagonal dislocation net is formed due to the twist deviation F of the bicrystals, the observed in¯uence of the twist angle on GB diusion can be qualitatively understood. The observed orientation dependence of Au and Cu GB diusion is based on two features: on the one hand by a cusp in sdDGB and maxima in the Arrhenius parameters exactly at the S 5 CSL GB due to the highly ordered atomic structure of the


394

BUDKE et al.:

TRACER DIFFUSION OF Au AND Cu IN Cu GBs

Several investigations on thermodynamic and kinetic properties of GBs indicate a ``special to general'' GB phase transition at a certain temperature Tc and only below Tc the GBs exhibit special properties [6, 26, 27]. The temperature Tc itself depends on the magnitude of the deviation in orientation from the special high coincidence GB. With increasing deviation from the ideal CSL GB Tc is expected to decrease. Furthermore, recent molecular dynamics simulations of self-diusion in GBs in f.c.c. metals indicate a transition from ``solid-like'' to ``liquid-like'' diusion at elevated temperatures [28]. At high temperatures the GB diusion is characterized by the same universal atomic mobility and by a low activation enthalpy, in comparison to lower temperatures, where the GB diusion is ``solid-like'' [28]. The temperature, where this transition occurs, depends on the GB energy of the CSL GB. In CSL GBs with fairly large GB energies, such as the presently studied S 5 CSL GB, the transition ``solidlike'' to ``liquid-like'' diusion is predicted at roughly 0.65Tm (Tm: melting temperature). In GBs with comparatively lower energies, e.g. S 7, Y 38X28 123111 CSL GB, the crystalline-like diusion is expected to occur up to I0.9Tm [28]. This prediction is in agreement with our recent diusion results of Ge in near S 7 [111] tilt GBs of Al [17] where indeed no curvature in the Arrhenius dependencies was observed in the investigated temperature range 0.56±0.73Tm. On the basis of the present experimental results it is dicult to give a consistent explanation of the observed non-Arrhenius behaviour of sdDGB data.
5. SUMMARY AND CONCLUSIONS

. The particular orientation behaviour of GB diusion in the present study can be explained by the superposition of two features: on the one hand, GB diusion is generally in¯uenced by the highly ordered atomic structure of the S 5 high coincidence CSL GB, where cusps in sdDGB and peaks in the Arrhenius parameters exactly at the ideal S 5 GB are expected. On the other hand, the additionally inserted net of GB dislocations formed due to the small twist deviations F from the ideal S 5 GB in our real bicrystals provides additional paths of ``easy'' diusion. Especially, the GBs in the immediate vicinity of the S 5 GB reveal (as an accidental result of the production process) relatively large twist angles. Therefore, the cusps in sdDGB Y and peaks in the Arrhenius parameters expected at Y 36X98 are masked to some extent and are only apparently shifted to a lower value in Y. . The deviation of sdDGB T from the straight Arrhenius dependence at higher temperature indicates a temperature induced change of the GB structure. Within the scope of the limited number of experimental results, it is not possible to decide whether a continuous curvature, a sharp bend or a jump occurs at a certain temperature. Therefore, the nature of the temperature induced change of the GB structure requires further detailed considerations.

. The parameter sdDGB of 195Au diusion and dDGB of 64Cu diusion along symmetrical S 5, Y 36X98 310001 Cu GBs strongly depends on the tilt angle Y. The GB diusion of solute and of solvent atoms show characteristic cusps not exactly at but slightly below the ideal S 5 CSL GB. In the immediate vicinity of the S 5 CSL GB the GB diusion parameter increases. This behaviour is reproducibly observed for all investigated temperatures. . The temperature dependence of sdDGB of Au GB diusion was approximated by Arrhenius relations at lower temperatures. The resulting Arrhenius parameters show the opposite behaviour in comparison to sdDGB Y. A characteristic peak in QGB and sdDGB 0 is observed slightly below the tilt orientation of the ideal S 5 CSL GB. At higher temperatures, however, a downward curved deviation from the linear Arrhenius relation is observed for almost all investigated GBs.

Acknowledgements ïThis work was supported by the German Ministry of Research and Technology (BMBF) under contract No. 03HE3MUE. The authors would like to thank S. S. Khasanov for the determination of the orientations of the bicrystals. One of the authors (S. Prokofjev) thanks the Russian Foundation for Fundamental Researches for support (contract No. 96-0248840). The production of the 64Cu isotope was performed with support of the reactor service group of the Research Center Geesthacht, Germany. REFERENCES 1. Aleshin, A. N., Bokshtein, B. S. and Shvindlerman, L. S., Soviet Phys. Solid St., 1977, 19, 2051. 2. Aleshin, A. N., Aristov, V. Y., Bokshtein, B. S. and Shvindlerman, L. S., Physica status solidi (a), 1978, 359, 1256. 3. Aleshin, A. N., Prokofjev, S. I. and Shvindlerman, L. S., Scripta metall., 1985, 19, 1135. 4. Aleshin, A. N., Prokofjev, S. I. and Shvindlerman, L. S., Defect Diusion Forum, 1989, 66±69, 861. 5. Li, X. M. and Chou, Y. T., Acta mater., 1996, 44, 3535. 6. Straumal, B. B., Klinger, L. M. and Shvindlerman, L. S., Acta metall., 1984, 32, 1355. 7. Ma, Q. and Ballu, R. W., Mater. Res. Soc. Symp., 1991, 209, 33. 8. Sommer, J., Herzig, Chr., Mayer, S. and Gust, W., Defect Diusion Forum, 1989, 66±69, 843. 9. Ma, Q. and Ballu, R. W., Acta metall. mater., 1993, 41, 133. 10. Ballu, R. W. and Brokman, A., Scripta metall., 1983, 17, 1027.


BUDKE et al.:

TRACER DIFFUSION OF Au AND Cu IN Cu GBs

395

11. Ballu, R. W., in Diusion in Crystalline Solids, ed. G. E. Murch and A. S. Nowick. Academic Press, New York, 1984. 12. Massalski, T. B. (ed.), Binary Alloy Phase Diagrams. Am. Soc. Metals, Metals Park, Ohio, 1986. 13. Kaur, I., Mishin, Y. and Gust, W., Fundamentals of Grain and Interphase Boundary Diusion. Wiley, Chichester, 1995. 14. Surholt, T., Mishin, Y. M. and Herzig, Chr., Phys. Rev. B, 1994, 50, 3577. 15. Fujikawa, S., Werner, M., Mehrer, H. and Seeger, A., Mater. Sci. Forum, 1987, 15±18, 431. 16. Maier, K., Physica status solidi (a), 1977, 44, 567. 17. Surholt, T., Molodov, D. A. and Herzig, Chr., Acta mater., 1998, 46, 5345. 18. Surholt, T., Minkwitz, C. and Herzig, Chr., Acta mater., 1998, 46, 1849. 19. Gutho, F., Mishin, Y. and Herzig, Chr., Z. Metallk., õ 1993, 84, 584.

20. Koppers, M., Mishin, Y. and Herzig, Chr., Acta õ metall. mater., 1994, 42, 2859. 21. Hofmann, S., Lejceck, P. and Adamek, J., Surf. Interface Anal., 1992, 19, 601. 22. Schober, T. and Ballu, R. W., Phil. Mag., 1970, 21, 109. 23. Vystavel, T., Paidar, V., Gemperle, A. and Gemperlova, J., Interface Sci., 1997, 5, 215. 24. Mishin, Y., Phil. Mag. A, 1995, 72, 1589. 25. Ciccoti, G., Guillope, M. and Pontikis, V., Phys. Rev. B, 1983, 27, 5576. 26. Maksimova, E. L., Shvindlerman, L. S. and Straumal, B. B., Acta metall., 1988, 36, 1573. 27. Maksimova, E. L., Rabkin, E. I., Shvindlerman, L. S. and Straumal, B. B., Acta metall., 1989, 37, 1995. 28. Keblinski, P., Phillpot, S. R., Wolf, D. and Gleiter, H., Phys. Rev. Lett., 1996, 77, 2965 and Private communication.