MICROSTRUCTURE EVOLUTION DURING NORMAL GRAIN GROWTH UNDER HIGH PRESSURE IN 2-D ALUMINIUM FOILS

We investigated the effect of high hydrostatic pressure on the normal grain growth in 2-D aluminium foils. The time dependence of the mean grain area was obtained. It was shown that normal grain growth takes place both at atmospheric pressure and under high hydrostatic pressure. The grain growth rate decreases by a factor 1.3 under high pressure. The activation volume for grain growth was 0.13 of the atomic volume. It was shown that high pressure strongly influences the ratio oflow angle boundaries and general boundaries at the early stages of secondary recrystallization.


INTRODUCTION
The investigation of the hydrostatic pressure effect on the kinetics of a thermally activated process permits to assess the activation volume V*.
The activation volume enters as follows in the equation for the acti- vation free energy G*: G*=Q*-TS*+pV*, (1) where Q*, S* are the energy and entropy of activation, respectively, T is the temperature and p is the pressure, V* is the volume expansion of the crystal related to atomic rearrangements during the thermally activated process.
Knowledge of V* permits therefore to draw conclusions concerning the mechanism of the process investigated.V* can be obtained from the equation V* -RT ln(v)/p, (2) where R is the gas constant, and v is the rate of the process investigated.
There are a number of papers where the influence of high pressure on diffusion was studied (Curtin et al., 1965; Buescher et al., 1973;  Kedves and Erdelyi, 1989; Kohler et al., 1992; Mehrer, 1996).It was shown that the diffusion processes are suppressed by pressure and V* varies in range from 0 to 0.9 f (f is the atomic volume).V* values close to 0 f may indicate an interstitial diffusion mechanism whereas V* values close to f are usually attributed to a vacancy diffusion mechanism.
Much less is known about the pressure effect on grain boundary (GB) diffusion (Martin et al., 1967; Erdelyi et al., 1987; Vieregge et al., 1991;  Lojkowski et al., 1995; Lojkowski, 1996).Activation volumes in the range 0-1 9t have been found indicating that both interstitial and vacancy GB diffusion mechanisms are active.The question arises as to what information about the GB migration mechanism can be drawn from high pressure studies.There are only a few papers where influence of high pressure on the grain growth (Hahn and Gleiter, 1979;  Lojkowski et al., 1995) and migration of individual grain boundaries (Molodov et al., 1984; 1994) was studied.It was found, that the acti- vation energy and activation volume for grain boundary migration in aluminium bicrystals are larger than that for grain growth in aluminium polycrystal (Lojkowski et al., 1995; Molodov et al., 1984; 1994) and depend on the GB crystallography.The latter result is particularly important when grain growth is investigated, since during grain growth migrating boundaries replace the shrinking ones and the crystal- lography of GBs continuously changes.This effect has not been taken into account during the above studies of grain growth under high pressure.The purpose of the present study is to investigate the effect of pressure on grain growth in aluminium using the recently developed experimental methods that permit to analyse the fraction of special and low energy GBs in the polycrystal during grain growth.

EXPERIMENTAL 1. Material Preparation
The experiments were carried out using foils of thickness in the range 0.08-0.1 mm.The 2-D (columnar) structure was obtained by rolling of 99.999% purity aluminium to 90-95% reduction in thickness and subsequent annealing at 833 K for 20 min.After annealing, a structure containing grains grown through the foil was obtained.The.grain size was larger than the thickness of the foil and the grain boundaries were perpendicular to its surface.The mean grain area was 0.025 mm2.To investigate the pressure effect on grain growth, the foils were annealed at 773 K for 0.5, 1, 1.5, 3, 4 and 6 h under atmospheric pressure and under the pressure of 1.2 GPa.

The High Pressure Experiments
The high pressure annealing was carried out under 99.999 wt.% argon atmosphere.The pressure was controlled by a managing gauge with the accuracy of 2%.The temperature was controlled using three Pt 6wt.%Rh-Pt 30wt.%Rhthermocouples with an accuracy of K.A three-zone high pressure furnace permitted to maintain a constant temperature over a plateau of 40mm length, so that the specimens were in a uniform temperature field.The high accuracy of the temperature control was crucial to obtain meaningful results.According to Molodov et al. (1994) the displacement of a GB during annealing under high pressure is vt tvo exp(-Q/RT) exp(-pV*/RT), (3) where v is rate of GB motion, is the annealing time, v0 is the pre- exponential factor.If 6T is the range of temperature fluctuations, 6p is the range of pressures applied, and 6t is the accuracy of time measurement, the contribution of different constituents at constant temperature to the error in GB displacement is Al/l= (Q/RTZ)6T+ (V*/RT)6p + 6t/t. (4) If Q 10kJ/mol, T 103K, V* 10cm3mo1-1 and 6T,, K, the influence of the fluctuations in temperature on the GB displacement is in the order of 15% of the pressure effect for a pressure on the order 0.1 GPa.It is impossible to hold the temperature constant inside the high pressure cell with a higher accuracy than K. Therefore, to obtain meaningful results for the pressure effect, the pressure applied must exceed GPa.These conditions were fulfilled in the present experi- ments.It has to be pointed out that in the present experiment thin aluminium foils were annealed under pressure.This is possible without deformation only under high gas pressure.

Structure Investigation Techniques
The mean grain size for the initial state of the specimens was measured using optical metallography.GBs etch to a different degree depending on their energy.During metallographic investigations in the optical microscope in bright field conditions the low energy GBs produce a weak contrast and it is practically impossible to reveal them.However, we performed an etching and observation procedure that permitted to reveal the low energy GBs using dark field conditions, an inclined beam and a large aperture.The angle of incidence of the light beam on the specimen surface was varied by rotating the microscope stage.This caused the variation in intensity of the grain colouring and permitted to obtain data on thegrain size distribution in the foil including the low energy GBs.The technique for grain size determination that we used is described in the paper of Stoyan et al. (1987).
The obtained data allowed to determine mean grain size as well as to trace the evolution of the topological characteristics of the grain structure in the foils.(The topological class of the grain is the number of triple junctions belong to this grain.)One of the most important indications of the normal grain growth is the linearity of mean grain area vs. annealing time (Fradkov and Udler, 1994).An additional confirmation of normal grain growth is the time independent topo- logical class distribution and linearity of the mean topological class vs. relative grain area.
Determination of grain misorientations was performed using the Scanning Electron Microscope and the Electron Channelling Pattern (ECP) technique (Sursaeva and Shvindlerman, 1995).

RESULTS
Figure shows the time dependence of the mean grain area for annealing at atmospheric pressure and under the pressure of 1.2 GPa.It is seen that the grain growth rate decreases under high pressure by a factor 1.3.According to Eq. ( 2) the value for the activation volume, V* 1.3 10 -6 m 3 mol-or V* 0.13 f was obtained.Annealing time, hours FIGURE Time dependence of mean grain area increment at T= 773 K (o) at the hydrostatic pressure 1.2 Gpa; (m) at the atmospheric pressure.(/) t=3h; (r) t=4h; (.) t=6h at the hydrostatic pressure 1.2Gpa; (/x) t= 1.5h; (x) 3 h; (o) 6 h at the atmospheric pressure.
Figure 3 shows the linear relationship between mean topological class and the relative grain area.
Figure 4 shows the time dependence of the fraction of low energy GBs in the specimen.Up to 4 h of annealing time there is no difference in the fraction of low energy GBs for specimens annealed under high and low pressure.The fraction of low energy GBs increases up to about 50% of the total number of GBs.However, upon further annealing, the behaviour of the specimens annealed under high and low pressure is different.For the specimens annealed at atmospheric pressure the fraction of low energy GBs decreases to the level of about 22% (Sursaeva et al., 1997) while it slightly increases above the 50% level for the specimens annealed at the pressure of 1.2 GPa.

DISCUSSION
So far all the papers known to us concerning grain growth have been dealing with three dimensional (3-D) specimens (Knyazev and Titorov,  1985; Rybin et al., 1984).X-ray methods were used to obtain statistical information about the distribution of GB misorientations.The data obtained in this way provide some averaged information.In contrast, the study of 2-D polycrystals permits to obtain the crystallographic data of each grain in the sample and permits a more straightforward identification of the mechanisms of grain growth (Shvindlerman et al.,  1994a,b,c).
It is known that the process of recrystallization includes three dif- ferent stages" primary recrystallization, normal grain growth and sec- ondary recrystallization.
The changes ofthe mean grain area vs. annealing time are governed by a specific law at each stage.At the stage of normal grain growth there is a linear mean grain area dependence on time (Fradkov and Udler, 1994).
The linear grain growth rate (Fig. 1) and constancy of the grains mean topological class vs. relative area (Fig. 3) indicates that in our samples the primary recrystallization was accomplished before the high pressure annealing.In that respect, the grain growth process during the present experiments takes place in the same way as in the previous papers concerning this system (Shvindlerman et al., 1994a,b,c).Appli- cation of high pressure in the present study of grain growth permitted to obtain two new results: the activation volume for normal grain growth in 2-D aluminium foils and the pressure effect on the fraction of low- energy GBs.As far as the activation volume is concerned, the value obtained, V*=0.13ft is significantly less than that obtained for migration of individual GBs in aluminium bicrystals (Molodov et al.,  1994).The V* values obtained by Molodov et al. were in the range 1-3 f, the exact value depending on the GB crystallography.The GBs studied were of symmetrical tilt type, with tilt axis (0 0 1), (1 1) and (0 1).The highest values of the activation volumes were obtained for the (0 1) tilt GBs.The above results were explained drawing attention to the fact that the (0 1) tilt GBs have the most tightly packed structure.
On the other hand, the present value ofthe activation volume matches in part with the results of Lojkowski et al. (1995) obtained for grain growth in A1 4N.Lojkowski et al. (1995) studied the effect of tem- perature and material purity on the activation volume for grain growth in polycrystalline aluminium.They have found that the activation volume for grain growth was 0.2 f for A1 5N annealed at 560 K.The same value was found for grain growth in A1 4N at the temperature above 700 K. On the other hand, the activation volume for grain growth at temperatures below 580 K was 0.7 f.Lojkowski et al. (1995) con- cluded that the high activation volume for grain growth in A14N at low temperatures is caused by interaction of GBs with impurities.They postulated that at high temperatures impurities desegregate from GBs and they take an open structure permitting a vacancy-less migration mechanism.Gleiter and Lissowski (1971) proposed such a mechanism, where GB migrates owing to movement of the contacts between atoms, involving very small movements of atoms.This mechanism can be called "atomic bonds migration mechanism" for GB migration.On the other hand, they postulated that for the GBs with segregated impurities transfer of atoms perpendicular to the GB takes place with the co- operation of vacancies.It seems that this mechanism is active at low temperatures, where impurities segregate to GBs.Interaction with segregated impurities could also explain the high activation volumes found by Molodov et al. (1994).
When comparing GB migration in polycrystals and bicrystals, one has to take into account that the coverage of the GB by impurities is much higher in the latter case, since for a constant number of impurity atoms in a bicrystal there is only one GB.Furthermore, this GB sweeps all the volume of the bicrystal and is able to approach all the impurity atoms.Another possibility to explain the discrepancy of activation volumes measured for bicrystals and polycrystals is that in the case of polycrystals GBs with arbitrary axes and less close packed structures dominated the average grain growth rate, while the less mobile GBs with simple misorientation axis could not contribute to the average activation volume.In conclusion, it seems that the prevailing GB migration mechanism in polycrystals is the bond migration one.The vacancy mechanism becomes active presumably in GBs with densely packed structure that may result from impurity segregation or from good atomic matching of the adjoining crystals.
Let us discuss in detail the observed effect of pressure on the fraction of low energy GBs (Fig. 4).associate this effect with a stronger decrease of the low-angle boundary mobility than that of the high-angle ones.2. The average activation volume for grain growth was 0.13 of the atomic volume and is compatible with grain boundary migration by movement of inter-atomic bonds.

Figure 2
Figure2shows the topological class distribution as a function of time.It is seen that the topological class distribution is time independent

FIGURE 3 FIGURE 4
FIGURE 3The mean topological class fi(S) of the grains versus relative area S/.