- Email: [email protected]
The amorphous-to-gamma transformation in ion implanted Al2O3
102
Nuclear
THE AMORPHOUS-TO-GAMMA P.S. SKLAD
“, J.C. McCALLUM
” :Melult rr,,d (‘rrumrt.r
D/rurw?
und -” Sol/d
In~trumcnt\
TRANSFORMATION ‘I. C.J. McHARGUE Sue
D/~~r.tron. Oah RI&
and Methods
I” Physics
Research
IN ION IMPLANTED ‘) and C.W.
~Vrrr,oncrl Lhrcrroy,
WHITE P. 0.
046 (1990) 102L106 North-Holland
AI,O, *
‘) 60s
_7OOH.Ouk Rd,q:
TV 37831 -h_<76, C’SA
Amorphous surface layers, free of implanted impurities, were produced on single crystal a-AlzO, substrates by Implantation at layers to post-lmplanta- 1 X5 o C with 2 x IO” Al/cm’ at 90 keV and 3 x IO” O/cm’ at 55 keV. The response of these amorphous elrctron microscopy (TEM) and Con annealing in the temperature range from 700 to X00 o C was investigated by in situ transmission conventIonal TEM of specimens annealed in bulk form. It was found that the amorphous Al?O, transformed 10 the cubic transitional y phase along an irregular front. The recrystallized material is columnar with the Individual columns or domains being irregular 1” size and shape. The y-Al,O, is composed of two varlantb which are twin related and have the following orientation relationship with the substrate: (llO),~~(lOlO),, and { Ii1 },~~(OOOl),,. The temperature-dependent velocity of the amorphous-to-y transformation front was measured by In situ TEM. Assummg a thermally activated process. the activation energy has been determined from an Arrhenius plot of velocity versus Tm ‘. These results are discussed In terms of possible rate controlling processes of the transformation.
1. Introduction a number of investigations have conRecently. centrated on the effect of ion implantation on ceramic materials [1.2]. Current research is focussed on the use of ion implantation to tailor the near-surface optical. electrical. magnetic. or mechanical properties of crystalline oxides. In many cases the implanted material must be subjected to post-implantation annealing treatments in order to achieve the desired property modification. Although it is possible to measure the changes in surface properties independently. the development of reliable processing techniques requires an understanding of the microstructural development which takes place as a result of implantation and/or post-implantation annealing. A number of electron microscopy techniques have been applied to the characterization of these materials. In an earlier investigation, TEM and Rutherford back scattering (RBS) were used to study the recrystallization behavior of amorphous surface layers on single crystal n-AlzO, [3]. The microstructures produced by annealing bulk specimens in the temperature range from 800 to 1200°C were characterized by conventional imaging and diffraction techniques. It was found that the amorphous Al,O, first transforms to y-Al *O,. a cubic transitional form, which in turn transforms to n-Al,Oi. The y-to-a transformation begins at the original crystallinePamorphous interface and proceeds to-
* Research sponsored in part by the Division of Materials Science. US Department of Energy, under contract DEAC05-840R21400 with Martin Marietta Energy Systems. Inc. 016%583X/90/$03.50 (North-Holland)
” Elsevier Science Puhllshers
B.V.
ward the specimen surface. By measuring the position of the y-cu interface as a function of annealing temperature and time it was possible to determine the temperature-dependent velocity of the y-n transformation front. Assuming a thermally activated process. the data are consistent with an activation energy of 0.36 eV for the transformation. Since no amorphous material remained after even the shortest anneals at X00” C. it was not possible to measure the kinetics of the amorphous Al LO3 to y-Al ,O, transformation. In the present work the results of a series of in situ TEM experiments. carried out to investigate the transformation of amorphous AllO, to y-AIIO,, are reported. Such in situ annealing experiments are quite useful in that they allow the progress of the transformation to be studied dynamically. Details of the experimental technique and analysis are also presented.
2. Experimental
details
High purity Al*O, single crystals with (0001) normal to the specimen surface were given an optical grade polish followed by a 120 h anneal at 1450°C to remove residual damage. The specimens were implanted at - 185°C with 2 X IO’” Al/cm* at 90 keV and 3 X 10’” O/cm’ at 55 keV. These ion energies were chosen to give the same projected range for both ion species. In addition, the crystals were implanted with the ion beam several degrees off-axis to minimize channeling effects. Characterization of the microstructural development was primarily accomplished with the use of transmission electron microscopy. Specimens prepared in cross section following established techniques [4] were an-
P.S. Sklad et al. / Amorphou.r-to-aammu
nealed in situ in a Philips CM12 operating at 120 kV with a Gatan model 628 single tilt heating holder. Annealing temperatures ranged from 700 to 800” C. The temperatures reported for the in situ experiments are the specimen cup temperatures as measured by a Pt/Pt-13% Rh thermocouple spot welded to the cup. Progress of the transformations was monitored with a Gatan model 622 video system. In addition, conventional imaging and diffraction techniques were used to characterize the microstructures produced by the in situ anneals and by selected anneals on bulk specimens. Results of in situ experiments are complicated by the physical constraints imposed by the experimental conditions employed. The thin-film geometry of the specimen. required for electron transparency, results in a situation in which solid state reactions take place in close proximity to the specimen surface where stress relaxation and surface energy effects may have a strong influence on the results. The geometry of the specimen, the design of the specimen furnace, and the location of the entire assembly in the pole piece of the objective lens of the electron microscope make adequate control and monitoring of the specimen temperature difficult. Despite the fact that the microscope column is maintained at a pressure of 1OK’ Torr, care must be taken to minimize contamination of the specimen surface. In addition, localized heating or radiation effects produced by the electron beam may also influence the outcome of the experiment. Although a detailed discussion of these factors is outside the scope of this paper, it should be noted that a number of precautionary measures were taken to minimize adverse effects. In particular, the
a Fig. 1. TEM
i
b
trun.@wtutron
in tntplmtted AI,O,
103
electron beam was spread to a diameter of approximately 10 urn in order to avoid subjecting any area of the specimen to high electron flux. In addition. except for one or two minutes necessary to adequately record the progress of the transformation for each time interval. the area of interest was moved 50%100 urn from the electron beam. Control experiments have established that continuous beam exposure can result in a factor of two increase in the velocity of the transformation front at 700 o C, but minimal effects at XOO”C. The results presented in this paper reflect only those experiments in which beam effects were minimized.
3. Results An example of the microstructural development which takes place during in situ annealing is shown in fig. 1, which compares the as-implanted microstructure to the microstructure observed during annealing at 725°C. The as-implanted specimen (fig. la) exhibits a featureless region characteristic of amorphous material which extends from the implanted surface to a depth of = 170 nm. The amorphous nature of the implanted region has been confirmed by electron diffraction as well as RBS. Figs. 1b and lc. which show the same region of the specimen observed during annealing at 725 o C. illustrate a number of important features of the transformation from the amorphous phase to the crystalline y phase. The recrystallized region has a columnar appearance with the individual columns or domains being irregular in size and shape. The domains
C
a-Al ,O, s pecime of cross-sectioned implanted with 2 x 10 ” Al /cm2 and 3x10” O/cm’ 6-k nplanted. (b) and (c) during in situ annealing at 725 o C. I = imp1la mted layer, S = substrate. II. CRYSTALLINE
OXIDES
at
(a)
,’ CERAMICS
104
P.S. Skid
et ul. / Amorphous-to-gammcr
are characterized by a high density of defects. These observations are consistent with results of previous investigations of specimens annealed in bulk [5]. The transformation front is irregular indicating local variations in the growth rate. This behavior is in contrast to the more planar growth of the cu-Al,O, front through y-Al,O, which was observed in previous experiments [2,5]. Convergent beam electron diffraction (CBED) was used to demonstrate that the orientation relationship between the recrystallized regions and the single crystal is (llO),]](lO~O), and {ill),]] cu-Al 203 substrate {OOOl},. A more detailed examination of the domain structure in a specimen annealed in situ at 700 o C reveals that there are two variants of y-AlzO,. The CBED pattern shown in fig. 2a corresponds to an orientation near the (IOiO) zone axis of the cu-AlzO, substrate whereas the patterns in figs. 2b and c represent orientations near a (110) zone axis of y-Al,O,. The arrows in each pattern correspond to the specimen normal. These results indicate that the two y-Al,O, variants are twin related across an invariant 111 plane. Similar domains have been observed during the growth of other cubic materials on hexagonal substrates and have been shown to result from double positioning [6,7]. During the growth of y-Al,O, the oxygen atoms can choose between two equally favorable sets of sites on the a-Al z0, substrate which are rotated 60” with respect to each other. As the individual domains of each orientation grow and impinge on each other a twin boundary results. In addition to the oriented domains, irregularly shaped cavities were observed along the original amorphous-crystalline boundary and between some of the domains. The formation of the cavities during in situ annealing is attributed to the accommodation of the volume difference between the less dense amorphous
trunsformutron
111tmplunrrd
AI,O,
Similar microstructural development occurred in specimens annealed in situ at all of the temperatures investigated in this study. The velocity of the interface has been measured from the video recordings of the transformation at temperatures of 700, 725, 750. 775. 785, and 800 o C. Measurement of the growth velocity as a function of specimen thickness revealed a trend toward higher growth velocity in thicker regions. In the more extreme cases where the thickness variation was large over a distance of l-2 km along the edge of the specimen, the growth rate varied by more than 30%. Specimen surface effects acting to retard the transformation could provide an explanation for the variation in rate with thickness. However, similar effects could be the result of temperature variations within the specimen. The higher growth rate in the thicker regions could be attributed to slightly higher temperatures which could arise as a result of a complex interaction between radiative heating and cooling effects and bulk conduction effects. For these reasons only results from the thicker regions were analyzed since it was assumed that they would be more representative of the bulk response of the material. In fig. 3 the growth rate of the y phase, as measured from the video recordings, is plotted on a log scale as a function of inverse temperature. The velocities fall into two distinct groups, one for temperatures 750 o C and below and another for temperatures above 750 o C. Assuming that all the observed growth is thermally activated, activation energies of = 1.6 eV and 7.8 eV can be extracted for the low and high temperature regions respectively. These values do not compare favorably with other reported values for thermally activated processes in Al,O, [3] (e.g., the activation energy for bulk diffusion of 0 or Al in a-Al,O, are reported at 6.6 eV and 5.0 eV respectively). Neither do they compare well with the activation energy measured
Fig. 2. Convergent beam electron diffraction patterns from a specimen annealed in situ at 700 o C. (a) (lOi0) zone axis of 2: a-AlzO, substrate. (b) and (c) (110) zone axis patterns of twin-related variants of y-Al zO, recrystallized regions. Arrows correspond to specimen normal.
P.S. Sklad et al. / Amorphous-to-Ronlmu
101
o.ol
(
750
725
I
I
LwL+
9.2
9.4
9.6
9.6
700 I
i 10.0
10.2
10.4
104/T ("K-'l
Fig.
3. Temperature dependence of the velocity amorphous/y-Al 20, interface.
in implanted
Al,O.,
105
proximately a factor of 3 higher than the rate observed at 75O’C in situ. One possible explanation is that the temperature of the electron transparent area of interest in the specimen during in situ annealing may have been = 30 o C lower than the temperature measured by the thermocouple attached to the specimen cup. In view of the complex geometry of the cross-sectioned TEM specimens which were mounted on a graphite washer to provide mechanical stability, it is possible that the heat conduction between the specimen cup (furnace) and the thin edge of the TEM specimens is reduced by the contact resistance introduced by the presence of multiple interfaces. The observed difference in growth velocity may also be due to a difference in the annealing atmosphere. a pressure of = lo-’ Torr in the microscope column compared to 1 atm of argon in the case of bulk anneal. Investigations of the annealing behavior of other oxide materials [S] have established that, in some cases, the presence of water vapor may increase the crystallization rate by more than an order of magnitude. Investigations are in progress to determine whether small amounts of water vapor or other impurities present in the argon gas may result in enhanced growth kinetics.
T (“‘2 775
600
transformcrfion
of
the
previously for the transformation from y-Al,O, to c1Al,O,. i.e, 3.6 eV. At this time it is not clear whether two distinct mechanisms of crystallization exist or whether the behavior observed is due to a continuous change in transformation kinetics. It is possible to speculate that the two different activation energies are associated with the crystallization of different forms of transitional alumina, the simple cubic y form in the lower temperature regime and the more highly ordered tetragonal 6 form at the higher temperatures. However, there is no electron diffraction evidence for the presence of the 6 phase. The underlying mechanisms controlling the crystallization behavior are currently the subject of more detailed investigations. Growth velocities observed in the in situ experiments were used to estimate an annealing time which would result in partial recrystallization of the y phase during annealing of a bulk specimen in flowing argon at 750 o C. However, since subsequent TEM observations indicated that all of the amorphous Al,O, had been transformed to y-Al,O,, this data point represents only an estimate of the minimum growth rate. The data point is also plotted in fig. 3. This estimated growth rate is ap-
4. Summary In situ transmission electron microscopy and conventional electron microscopy were used to investigate the response of amorphous Al,O, to post-implantation annealing. The amorphous layers were produced on single crystal cu-Al,O, substrates by implantation at - 185°C with 2 X lOI Al/cm2 at 90 keV and 3 X 10h O/cm2 at 55 keV. The amorphous material transformed to y-Al,O, along an irregular front composed of individual columns or domains. The y domains correspond to two twin-related variants which have a distinct orientation relationship with the substrate. The temperaturedependent velocity of the transformation was measured in an attempt to determine the activation energy. The mechanisms responsible for the observed behavior are the subject of continued investigation. The author would like to acknowledge C.P. Haltom, A.M. Williams, and A.T. Fisher for their assistance in specimen preparation, and A.R. McDonald for preparing this manuscript.
References [l] See, for example: Ion Beam Modification of Insulators, eds. P. Mazzoldi and G.W. Arnold (Elsevier. Amsterdam. 1987). II. CRYSTALLINE
OXIDES
/ C‘ERAMICS
106
P.S. Sklud et ~1. / Amorphous-to-gummu
[2] C.W. White. C.J. McHargue, P.S. Sklad. L.A. Boatner and G.C. Farlow. Mater. Sci. Rep. 4 (1989) 41-146. [3] C.W. White. L.A. Boatner. P.S. Sklad. C.J. McHargue. J. Rankin, G.C. Farlow and M.J. Aziz. Nucl. Instr. and Meth. B32 (1988) 11. [4] P.S. Sklad, Proc. Ann. EMSA Meet. 43 (1985) 276.
trcm.&rmulron
(51 [6] [7] [X]
m lmplunted
Al,O?
P.S. Sklad. Proc. Ann. EMSA Meet 44 (1986) 508. D.W. Pashley and M.J. Stowell. Philos. Mag. X (1963) 1605. K.L. More. Proc. Ann. EMSA Meet. 46 (19X8) 56X. J. Rankin. L.A. Boatner, C.W. White and L.W. Hobbs, Mater. Res. Sot. Symp. Proc. 100 (19X8) 453.
Nuclear
THE AMORPHOUS-TO-GAMMA P.S. SKLAD
“, J.C. McCALLUM
” :Melult rr,,d (‘rrumrt.r
D/rurw?
und -” Sol/d
In~trumcnt\
TRANSFORMATION ‘I. C.J. McHARGUE Sue
D/~~r.tron. Oah RI&
and Methods
I” Physics
Research
IN ION IMPLANTED ‘) and C.W.
~Vrrr,oncrl Lhrcrroy,
WHITE P. 0.
046 (1990) 102L106 North-Holland
AI,O, *
‘) 60s
_7OOH.Ouk Rd,q:
TV 37831 -h_<76, C’SA
Amorphous surface layers, free of implanted impurities, were produced on single crystal a-AlzO, substrates by Implantation at layers to post-lmplanta- 1 X5 o C with 2 x IO” Al/cm’ at 90 keV and 3 x IO” O/cm’ at 55 keV. The response of these amorphous elrctron microscopy (TEM) and Con annealing in the temperature range from 700 to X00 o C was investigated by in situ transmission conventIonal TEM of specimens annealed in bulk form. It was found that the amorphous Al?O, transformed 10 the cubic transitional y phase along an irregular front. The recrystallized material is columnar with the Individual columns or domains being irregular 1” size and shape. The y-Al,O, is composed of two varlantb which are twin related and have the following orientation relationship with the substrate: (llO),~~(lOlO),, and { Ii1 },~~(OOOl),,. The temperature-dependent velocity of the amorphous-to-y transformation front was measured by In situ TEM. Assummg a thermally activated process. the activation energy has been determined from an Arrhenius plot of velocity versus Tm ‘. These results are discussed In terms of possible rate controlling processes of the transformation.
1. Introduction a number of investigations have conRecently. centrated on the effect of ion implantation on ceramic materials [1.2]. Current research is focussed on the use of ion implantation to tailor the near-surface optical. electrical. magnetic. or mechanical properties of crystalline oxides. In many cases the implanted material must be subjected to post-implantation annealing treatments in order to achieve the desired property modification. Although it is possible to measure the changes in surface properties independently. the development of reliable processing techniques requires an understanding of the microstructural development which takes place as a result of implantation and/or post-implantation annealing. A number of electron microscopy techniques have been applied to the characterization of these materials. In an earlier investigation, TEM and Rutherford back scattering (RBS) were used to study the recrystallization behavior of amorphous surface layers on single crystal n-AlzO, [3]. The microstructures produced by annealing bulk specimens in the temperature range from 800 to 1200°C were characterized by conventional imaging and diffraction techniques. It was found that the amorphous Al,O, first transforms to y-Al *O,. a cubic transitional form, which in turn transforms to n-Al,Oi. The y-to-a transformation begins at the original crystallinePamorphous interface and proceeds to-
* Research sponsored in part by the Division of Materials Science. US Department of Energy, under contract DEAC05-840R21400 with Martin Marietta Energy Systems. Inc. 016%583X/90/$03.50 (North-Holland)
” Elsevier Science Puhllshers
B.V.
ward the specimen surface. By measuring the position of the y-cu interface as a function of annealing temperature and time it was possible to determine the temperature-dependent velocity of the y-n transformation front. Assuming a thermally activated process. the data are consistent with an activation energy of 0.36 eV for the transformation. Since no amorphous material remained after even the shortest anneals at X00” C. it was not possible to measure the kinetics of the amorphous Al LO3 to y-Al ,O, transformation. In the present work the results of a series of in situ TEM experiments. carried out to investigate the transformation of amorphous AllO, to y-AIIO,, are reported. Such in situ annealing experiments are quite useful in that they allow the progress of the transformation to be studied dynamically. Details of the experimental technique and analysis are also presented.
2. Experimental
details
High purity Al*O, single crystals with (0001) normal to the specimen surface were given an optical grade polish followed by a 120 h anneal at 1450°C to remove residual damage. The specimens were implanted at - 185°C with 2 X IO’” Al/cm* at 90 keV and 3 X 10’” O/cm’ at 55 keV. These ion energies were chosen to give the same projected range for both ion species. In addition, the crystals were implanted with the ion beam several degrees off-axis to minimize channeling effects. Characterization of the microstructural development was primarily accomplished with the use of transmission electron microscopy. Specimens prepared in cross section following established techniques [4] were an-
P.S. Sklad et al. / Amorphou.r-to-aammu
nealed in situ in a Philips CM12 operating at 120 kV with a Gatan model 628 single tilt heating holder. Annealing temperatures ranged from 700 to 800” C. The temperatures reported for the in situ experiments are the specimen cup temperatures as measured by a Pt/Pt-13% Rh thermocouple spot welded to the cup. Progress of the transformations was monitored with a Gatan model 622 video system. In addition, conventional imaging and diffraction techniques were used to characterize the microstructures produced by the in situ anneals and by selected anneals on bulk specimens. Results of in situ experiments are complicated by the physical constraints imposed by the experimental conditions employed. The thin-film geometry of the specimen. required for electron transparency, results in a situation in which solid state reactions take place in close proximity to the specimen surface where stress relaxation and surface energy effects may have a strong influence on the results. The geometry of the specimen, the design of the specimen furnace, and the location of the entire assembly in the pole piece of the objective lens of the electron microscope make adequate control and monitoring of the specimen temperature difficult. Despite the fact that the microscope column is maintained at a pressure of 1OK’ Torr, care must be taken to minimize contamination of the specimen surface. In addition, localized heating or radiation effects produced by the electron beam may also influence the outcome of the experiment. Although a detailed discussion of these factors is outside the scope of this paper, it should be noted that a number of precautionary measures were taken to minimize adverse effects. In particular, the
a Fig. 1. TEM
i
b
trun.@wtutron
in tntplmtted AI,O,
103
electron beam was spread to a diameter of approximately 10 urn in order to avoid subjecting any area of the specimen to high electron flux. In addition. except for one or two minutes necessary to adequately record the progress of the transformation for each time interval. the area of interest was moved 50%100 urn from the electron beam. Control experiments have established that continuous beam exposure can result in a factor of two increase in the velocity of the transformation front at 700 o C, but minimal effects at XOO”C. The results presented in this paper reflect only those experiments in which beam effects were minimized.
3. Results An example of the microstructural development which takes place during in situ annealing is shown in fig. 1, which compares the as-implanted microstructure to the microstructure observed during annealing at 725°C. The as-implanted specimen (fig. la) exhibits a featureless region characteristic of amorphous material which extends from the implanted surface to a depth of = 170 nm. The amorphous nature of the implanted region has been confirmed by electron diffraction as well as RBS. Figs. 1b and lc. which show the same region of the specimen observed during annealing at 725 o C. illustrate a number of important features of the transformation from the amorphous phase to the crystalline y phase. The recrystallized region has a columnar appearance with the individual columns or domains being irregular in size and shape. The domains
C
a-Al ,O, s pecime of cross-sectioned implanted with 2 x 10 ” Al /cm2 and 3x10” O/cm’ 6-k nplanted. (b) and (c) during in situ annealing at 725 o C. I = imp1la mted layer, S = substrate. II. CRYSTALLINE
OXIDES
at
(a)
,’ CERAMICS
104
P.S. Skid
et ul. / Amorphous-to-gammcr
are characterized by a high density of defects. These observations are consistent with results of previous investigations of specimens annealed in bulk [5]. The transformation front is irregular indicating local variations in the growth rate. This behavior is in contrast to the more planar growth of the cu-Al,O, front through y-Al,O, which was observed in previous experiments [2,5]. Convergent beam electron diffraction (CBED) was used to demonstrate that the orientation relationship between the recrystallized regions and the single crystal is (llO),]](lO~O), and {ill),]] cu-Al 203 substrate {OOOl},. A more detailed examination of the domain structure in a specimen annealed in situ at 700 o C reveals that there are two variants of y-AlzO,. The CBED pattern shown in fig. 2a corresponds to an orientation near the (IOiO) zone axis of the cu-AlzO, substrate whereas the patterns in figs. 2b and c represent orientations near a (110) zone axis of y-Al,O,. The arrows in each pattern correspond to the specimen normal. These results indicate that the two y-Al,O, variants are twin related across an invariant 111 plane. Similar domains have been observed during the growth of other cubic materials on hexagonal substrates and have been shown to result from double positioning [6,7]. During the growth of y-Al,O, the oxygen atoms can choose between two equally favorable sets of sites on the a-Al z0, substrate which are rotated 60” with respect to each other. As the individual domains of each orientation grow and impinge on each other a twin boundary results. In addition to the oriented domains, irregularly shaped cavities were observed along the original amorphous-crystalline boundary and between some of the domains. The formation of the cavities during in situ annealing is attributed to the accommodation of the volume difference between the less dense amorphous
trunsformutron
111tmplunrrd
AI,O,
Similar microstructural development occurred in specimens annealed in situ at all of the temperatures investigated in this study. The velocity of the interface has been measured from the video recordings of the transformation at temperatures of 700, 725, 750. 775. 785, and 800 o C. Measurement of the growth velocity as a function of specimen thickness revealed a trend toward higher growth velocity in thicker regions. In the more extreme cases where the thickness variation was large over a distance of l-2 km along the edge of the specimen, the growth rate varied by more than 30%. Specimen surface effects acting to retard the transformation could provide an explanation for the variation in rate with thickness. However, similar effects could be the result of temperature variations within the specimen. The higher growth rate in the thicker regions could be attributed to slightly higher temperatures which could arise as a result of a complex interaction between radiative heating and cooling effects and bulk conduction effects. For these reasons only results from the thicker regions were analyzed since it was assumed that they would be more representative of the bulk response of the material. In fig. 3 the growth rate of the y phase, as measured from the video recordings, is plotted on a log scale as a function of inverse temperature. The velocities fall into two distinct groups, one for temperatures 750 o C and below and another for temperatures above 750 o C. Assuming that all the observed growth is thermally activated, activation energies of = 1.6 eV and 7.8 eV can be extracted for the low and high temperature regions respectively. These values do not compare favorably with other reported values for thermally activated processes in Al,O, [3] (e.g., the activation energy for bulk diffusion of 0 or Al in a-Al,O, are reported at 6.6 eV and 5.0 eV respectively). Neither do they compare well with the activation energy measured
Fig. 2. Convergent beam electron diffraction patterns from a specimen annealed in situ at 700 o C. (a) (lOi0) zone axis of 2: a-AlzO, substrate. (b) and (c) (110) zone axis patterns of twin-related variants of y-Al zO, recrystallized regions. Arrows correspond to specimen normal.
P.S. Sklad et al. / Amorphous-to-Ronlmu
101
o.ol
(
750
725
I
I
LwL+
9.2
9.4
9.6
9.6
700 I
i 10.0
10.2
10.4
104/T ("K-'l
Fig.
3. Temperature dependence of the velocity amorphous/y-Al 20, interface.
in implanted
Al,O.,
105
proximately a factor of 3 higher than the rate observed at 75O’C in situ. One possible explanation is that the temperature of the electron transparent area of interest in the specimen during in situ annealing may have been = 30 o C lower than the temperature measured by the thermocouple attached to the specimen cup. In view of the complex geometry of the cross-sectioned TEM specimens which were mounted on a graphite washer to provide mechanical stability, it is possible that the heat conduction between the specimen cup (furnace) and the thin edge of the TEM specimens is reduced by the contact resistance introduced by the presence of multiple interfaces. The observed difference in growth velocity may also be due to a difference in the annealing atmosphere. a pressure of = lo-’ Torr in the microscope column compared to 1 atm of argon in the case of bulk anneal. Investigations of the annealing behavior of other oxide materials [S] have established that, in some cases, the presence of water vapor may increase the crystallization rate by more than an order of magnitude. Investigations are in progress to determine whether small amounts of water vapor or other impurities present in the argon gas may result in enhanced growth kinetics.
T (“‘2 775
600
transformcrfion
of
the
previously for the transformation from y-Al,O, to c1Al,O,. i.e, 3.6 eV. At this time it is not clear whether two distinct mechanisms of crystallization exist or whether the behavior observed is due to a continuous change in transformation kinetics. It is possible to speculate that the two different activation energies are associated with the crystallization of different forms of transitional alumina, the simple cubic y form in the lower temperature regime and the more highly ordered tetragonal 6 form at the higher temperatures. However, there is no electron diffraction evidence for the presence of the 6 phase. The underlying mechanisms controlling the crystallization behavior are currently the subject of more detailed investigations. Growth velocities observed in the in situ experiments were used to estimate an annealing time which would result in partial recrystallization of the y phase during annealing of a bulk specimen in flowing argon at 750 o C. However, since subsequent TEM observations indicated that all of the amorphous Al,O, had been transformed to y-Al,O,, this data point represents only an estimate of the minimum growth rate. The data point is also plotted in fig. 3. This estimated growth rate is ap-
4. Summary In situ transmission electron microscopy and conventional electron microscopy were used to investigate the response of amorphous Al,O, to post-implantation annealing. The amorphous layers were produced on single crystal cu-Al,O, substrates by implantation at - 185°C with 2 X lOI Al/cm2 at 90 keV and 3 X 10h O/cm2 at 55 keV. The amorphous material transformed to y-Al,O, along an irregular front composed of individual columns or domains. The y domains correspond to two twin-related variants which have a distinct orientation relationship with the substrate. The temperaturedependent velocity of the transformation was measured in an attempt to determine the activation energy. The mechanisms responsible for the observed behavior are the subject of continued investigation. The author would like to acknowledge C.P. Haltom, A.M. Williams, and A.T. Fisher for their assistance in specimen preparation, and A.R. McDonald for preparing this manuscript.
References [l] See, for example: Ion Beam Modification of Insulators, eds. P. Mazzoldi and G.W. Arnold (Elsevier. Amsterdam. 1987). II. CRYSTALLINE
OXIDES
/ C‘ERAMICS
106
P.S. Sklud et ~1. / Amorphous-to-gummu
[2] C.W. White. C.J. McHargue, P.S. Sklad. L.A. Boatner and G.C. Farlow. Mater. Sci. Rep. 4 (1989) 41-146. [3] C.W. White. L.A. Boatner. P.S. Sklad. C.J. McHargue. J. Rankin, G.C. Farlow and M.J. Aziz. Nucl. Instr. and Meth. B32 (1988) 11. [4] P.S. Sklad, Proc. Ann. EMSA Meet. 43 (1985) 276.
trcm.&rmulron
(51 [6] [7] [X]
m lmplunted
Al,O?
P.S. Sklad. Proc. Ann. EMSA Meet 44 (1986) 508. D.W. Pashley and M.J. Stowell. Philos. Mag. X (1963) 1605. K.L. More. Proc. Ann. EMSA Meet. 46 (19X8) 56X. J. Rankin. L.A. Boatner, C.W. White and L.W. Hobbs, Mater. Res. Sot. Symp. Proc. 100 (19X8) 453.