Full text of DOI:10.1063/1.114647

Gettering of Au to dislocations and cavities in silicon
J. WongLeung, E. Nygren, and J. S. Williams
Citation: Applied Physics Letters 67, 416 (1995); doi: 10.1063/1.114647
View online: http://dx.doi.org/10.1063/1.114647
View Table of Contents: http://scitation.aip.org/content/aip/journal/apl/67/3?ver=pdfcov
Published by the AIP Publishing
Articles you may be interested in
Selectivity of nanocavities and dislocations for gettering of Cu and Fe in silicon
Appl. Phys. Lett. 78, 2682 (2001); 10.1063/1.1363689
Gettering of nickel to cavities in silicon introduced by hydrogen implantation
Appl. Phys. Lett. 66, 1889 (1995); 10.1063/1.113311
Gettering of copper to hydrogeninduced cavities in silicon
Appl. Phys. Lett. 66, 1231 (1995); 10.1063/1.113246
Model for dislocation locking by oxygen gettering in silicon crystals
Appl. Phys. Lett. 58, 1842 (1991); 10.1063/1.105241
Simultaneous gettering of Au in silicon by phosphorus and dislocations
Appl. Phys. Lett. 33, 442 (1978); 10.1063/1.90371
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
131.252.200.90 On: Thu, 04 Dec 2014 04:29:38

Gettering of Au to dislocations and cavities in silicon
J. Wong-Leung,a) E. Nygren, and J. S. Williams
Department of Electronic Materials Engineering, Research School of Physical Sciences
and Engineering, The Australian National University, Canberra ACT 0200, Australia
͑Received 3 January 1995; accepted for publication 13 April 1995͒
The gettering of ion implanted Au to defects in Si has been studied using Rutherford backscattering
and channeling and transmission electron microscopy. Damage from a Si implant anneals into
dislocations which can efficiently trap diffusing Au. The damage introduced by a H implant evolves
during annealing into cavities which getter close to 100% of the Au, leaving very little Au in
solution. This process is driven by the diffusion of a supersaturated solid solution of Au to a
favorable sink. The internal surfaces of cavities are the most favorable sink, followed by
dislocations and then the Si surface. © 1995 American Institute of Physics.
Gettering is the process by which metallic impurities are
removed from the active regions of devices to favorable trap-
ping or precipitation sites. Effective gettering of Au can be
achieved either by phosphorus diffusion to produce regions
of very high doping ͑Ͼ1020 cmϪ3 n-type͒ which gives rise
to dramatically enhanced solubility for metal impurities,1,2 or
by selectively introducing damage by mechanical methods or
ion implantation.3–5 The gettering mechanisms for Au in the
two cases are different.6 The formation of Au-donor ͑phos-
phorus͒ complexes is believed to result in increased Au solu-
bility in the former case, while in the latter case, the dislo-
cations created by the introduced damage act as precipitation
sites or trapping centers for diffusing Au interstitials. Wong
et al.5 have used proximity gettering of Au to C-implanted
layers introduced very close to the active device regions.
More recently, Myers et al.7 and Wong-Leung et al.8 have
demonstrated the effective gettering of Cu to internal cavities
in Si produced by He and H implantation, respectively.
In this letter, we compare the gettering behavior of Au to
extended defects introduced by Si implantation and to inter-
nal nanocavities generated by H implantation. Our results
demonstrate the strong dependence of Au trapping and pre-
cipitation on the nature of the introduced defects.
Getter layers were introduced in two ways into n-type Si
wafers of ͑100͒ orientation. In the first set of samples, 250
keV Si ions were implanted to a dose of 1ϫ1015 cmϪ2 at
room temperature. This did not produce a continuous amor-
phous layer, as determined by Rutherford backscattering and
channeling ͑RBS-C͒ using 2 MeV Heϩ ions. This damage
was preannealed in some of the samples at either 650 or
850 °C for 1 h. All the anneals in this study were performed
using a conventional quartz tube furnace with flowing Ar.
In the second set of samples, a getter layer was intro-
duced by first implanting 97 keV H ions to a dose of 3
ϫ1016 cmϪ2 at room temperature, giving a projected range
of about 1 ␮m. Subsequent annealing above 600 °C formed a
sharp band of cavities close to the H range.8,9
tion to the use of RBS-C to determine the distribution of both
Au and gross disorder, the microstructure of the various
samples was examined using cross-sectional transmission
electron microscopy ͑XTEM͒.
Prior to annealing, samples implanted with both Si and
Au ͑with no anneal in between the two implants͒ contain a
deep damaged layer at a depth of 0.2–0.5 ␮m due to the Si
implant and a thin amorphous surface layer due to the Au
implant, as illustrated in Fig. 1͑a͒. When this sample was
subsequently annealed at 650 °C for 1 h, the random RBS
spectrum in Fig. 1͑b͒ shows that some of the Au ͑ϳ2%͒ is
trapped at a depth of ϳ0.2 ␮m ͑the front edge of the deeper
damage band͒ while most of the Au segregates to the surface
during solid phase epitaxial regrowth of the amorphous sur-
face layer. However, after a 1 h anneal at 850 °C, the situa-
tion is drastically different: a large amount of the implanted
Au ͑ϳ86%͒ has redistributed from the Si surface to the re-
gion of the Si implant damage at a depth of ϳ0.4 ␮m, while
some Au remains close to the surface. Figure 1͑a͒ also shows
the result of annealing on the Si implant damage in these
samples: the dechanneling step at depths of 0.2–0.4 ␮m sug-
gests the agglomeration of point defects into extended de-
fects, mainly dislocation loops, as confirmed in the XTEM
micrograph in Fig. 2. This micrograph clearly shows a well-
defined disordered band consisting of a dense network of
dislocation loops after an 850 °C anneal. There is no evi-
dence of discernable Au precipitates at the dislocation band
despite the large fraction of Au contained in this region ͓Fig.
1͑b͔͒. Indeed, the average Au concentration in the dislocation
band is ϳ3ϫ1019 cmϪ3, which is more than 4 orders of
magnitude above the solid solubility limit of Au in Si at
10
850 °C of ϳ2ϫ1015 cmϪ3
.
However, a few Au-rich pre-
cipitates are observed by TEM near the surface: in this case
the Au concentration within 100 Å of the surface is Ͼ1019
cmϪ3, again far exceeding the equilibrium Au solubility at
850 °C. The precipitation of Au following annealing at tem-
11
peratures above the eutectic temperature ͑363 °C͒ is inter-
Au was introduced into the various samples by a 95 keV
Au implantation to a nominal dose of 1ϫ1015 cmϪ2. The
samples were subjected to different annealing treatments at
650 or 850 °C for 1 h and the Au diffusion, solubility, get-
tering, and trapping effects at defects were studied. In addi-
esting. Presumably, Au first forms a eutectic molten phase
with Si, which solidifies to Au-rich crystallites on cooling.
Indeed, our previous TEM studies have given strong evi-
dence for such molten eutectic ‘‘precipitates’’ and their mi-
gration under temperature gradients.9
In contrast to strong gettering of Au to dislocations in the
above cases, Si-implanted samples which are preannealed at
a͒
Electronic mail: jwl109@rsphy1.anu.edu.au
416
Appl. Phys. Lett. 67 (3), 17 July 1995
0003-6951/95/67(3)/416/3/$6.00
© 1995 American Institute of Physics
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
131.252.200.90 On: Thu, 04 Dec 2014 04:29:38

FIG. 3. ͑a͒ RBS of samples with as-implanted H getter and near-surface
implanted Au showing the damage in the Si crystal as-implanted ͑᭺͒, after
1 h anneal at 650 °C ͑᭝͒ and 850 °C ͑᭿͒. ͑b͒ The profile of Au as-implanted
͑᭺͒, after 1 h anneal at 650 °C ͑᭝͒ and 850 °C ͑᭿͒.
FIG. 1. ͑a͒ RBS of samples with as-implanted Si getter and near-surface
implanted Au showing the damage in the Si crystal, as-implanted ͑᭺͒, after
1 h anneal at 650 °C ͑᭝͒, and 850 °C ͑᭿͒, respectively. ͓Note 650 °C depth
of damage ͑↓͒ and 850 °C damage ͑↓↓͒.͔ ͑b͒ The profile of Au as-implanted
͑᭺͒, after 1 h anneal at 650 °C ͑᭝͒ and 850 °C ͑᭿͒.
implant damage and the Au implant is favored over Au pre-
cipitation at the surface.
650 or 850 °C to form a stable band of dislocations prior to
Au implantation, did not exhibit such efficient gettering
when subsequently annealed at 850 °C. We interpret this dif-
ference as follows. Accumulation of Au at preformed dislo-
cations is not favored over Au precipitation near the Si sur-
face. However, defect-mediated Au diffusion to ‘‘evolving’’
dislocation networks during simultaneous annealing of Si-
For the H-implanted case, the Si portion of the RBS-C
spectra in Fig. 3͑a͒ shows that the main disorder from H
implantation is introduced at a depth of 1 ␮m, corresponding
closely to the H ion range.8 After annealing at 650 °C, the
high level of dechanneling at 1 ␮m is attributed to large
strain fields present around cavities as confirmed by TEM
͑not shown͒. Figure 3͑b͒ shows that most of the Au segre-
gates to the surface during 650 °C annealing: only a small
fraction of Au diffuses to and accumulates at the cavities.
However, when the annealing time at 650 °C is increased to
15 h, about 31% of the Au was found at the cavities. After
the 850 °C annealing ϳ96% of the implanted Au is located at
the H-induced cavities as shown in Fig. 3͑b͒. Indeed, for a
longer annealing time ͑e.g., 48 h͒ at 850 °C, essentially all
the implanted Au is driven from near-surface traps to the
cavities. Figure 3͑a͒ shows that, after the 850 °C anneal,
much of the initial H-induced damage has annealed, leaving
few strain centers surrounding cavities in good quality crys-
tal. The XTEM micrograph in Fig. 4 confirms this conclu-
sion. The cavities are more faceted, larger in size and fewer
in number compared to those formed after the 650 °C anneal.
FIG. 2. XTEM micrograph of sample with as-implanted Si-getter layers
after annealing at 850 °C for 1 h.
Appl. Phys. Lett., Vol. 67, No. 3, 17 July 1995
Wong-Leung et al.
417
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
131.252.200.90 On: Thu, 04 Dec 2014 04:29:38

sequently, when Au is released from this supersaturated near-
surface region, there is a strong driving force for Au to dif-
fuse away and precipitate or accumulate at favorable sinks.
Our data clearly indicate that the free surface and volume of
internal cavities, in relative close proximity to the Au im-
planted layer, are very favorable sinks for diffusing Au. On
cooling, the eutectic Au–Si molten phase, if it exists, freezes
to Au-rich crystallites, as observed in TEM. Although we
have not yet established the precise mechanism by which
supersaturated Au diffuses to the cavities, it is possible that
interstitial Au diffusion is mediated by Si point defects via,
for example, a ‘‘kick-out’’ mechanism.10,14
This behavior is consistent with our observations for Au
in samples with no cavities under similar annealing treat-
ments, where supersaturated Au forms precipitates at the sur-
face. For Au annealed at 850 °C, we have recently under-
taken in situ ͑hot͒ RBS analysis13 to demonstrate that the
rapid movement of Au to cavities occurs during annealing at
temperature and not during cooldown of the wafer. This sug-
gests that the efficient relocation of supersaturated Au from
the near-surface to cavities is a nonequilibrium first step in
the process of equilibration of Au in Si at temperature.
In conclusion, we have demonstrated a highly efficient
gettering process for Au in Si, in which close to 100% of a
supersaturated Au concentration can be relocated from weak
trapping sites at the surface to a band of nanocavities during
annealing. Our results clearly indicate that the extent of get-
tering of Au to implantation-induced defects strongly de-
pends on the nature of the defects. Our observations on Si
implant damage suggest that the migration and annealing of
defects assist the diffusion and trapping of Au to
implantation-induced dislocations. The very efficient getter-
ing of Au to cavities is driven by the diffusion and subse-
quent trapping/precipitation at suitable sites of a supersatu-
rated concentration of Au in Si. Indeed, nanocavities are
preferred gettering sites for Au trapping on cavity walls and
precipitation of a Au–Si eutectic melt, followed by a band of
dislocations and finally, Au precipitated close to the Si sur-
face.
FIG. 4. XTEM micrograph of sample with as-implanted H-getter layers
after annealing at 850 °C for 1 h. ͓Note the filled cavities ͑dark features͒ and
the faceted empty cavities.͔
They predominantly contain ͕111͖ facets, together with small
͕100͖ facets. Indeed, through-focal TEM shows a dark line at
the cavity wall for over and under-focus, consistent with a
high-Z layer at this interface.12 Further TEM observations of
some cavities ͓dark, poorly faceted features in Fig. ͑4͔͒ con-
firm that some of them are filled with Au-rich precipitates.
Our more detailed observations indicate that Au first appears
to bond to cavity walls ͑similar to Cu binding to cavities7,8
͒
prior to bulk phase formation. When cavity walls are satu-
rated with Au, the bulk phase is undoubtedly present as a
eutectic ͑Au–Si͒ melt at 850 °C but solidifies on cooling
through the eutectic temperature ͑363 °C͒, giving rise to the
Au-rich crystallites observed in some cavities by TEM.
H-implanted samples which were preannealed at 850 °C
prior to the Au implant exhibited much the same behavior
͑with 100% of the implanted Au at the cavities͒ as samples
which were not preannealed. This result suggests that the
migration and annealing of H-implantation-induced defects
is not the primary driving force for accumulation of Au at
cavities but, rather, the diffusion of supersaturated Au to a
preferred sink at the cavities. Furthermore, Au implanted
samples with no H-induced cavities retained essentially all
the Au at the surface following an hour anneal at 850 °C. In
such a case, supersaturated Au was found to precipitate lo-
cally at the near-surface in the absence of preferred cavity
sites.
Dave Eaglesham and Sam Myers are acknowledged for
fruitful discussions.
1 M. Nakamura, T. Kato, and N. Oi, Jpn. J. Appl. Phys. 7, 512 ͑1968͒.
2 D. Lecrosnier, J. Paugam, G. Pelous, F. Richou, and M. Salvi, J. Appl.
Phys. 52, 5090 ͑1981͒.
3 T. E. Seidel, R. L. Meek, and A. G. Cullis, J. Appl. Phys. 46, 600 ͑1975͒.
4 D. Jaworska, J. Sielanko, and E. Tarnowski, Appl. Phys. A 35, 119 ͑1984͒.
5 H. Wong, N. Cheung, and P. K. Chu, Appl. Phys. Lett. 52, 889 ͑1988͒.
6 J. S. Kang, and D. K. Schroder, J. Appl. Phys. 65, 2974 ͑1989͒.
7 S. M. Myers, D. M. Bishop, D. M. Follstaedt, H. J. Stein, and W. R.
Wampler, Mater. Res. Soc. Symp. Proc. 283, 549 ͑1993͒.
8 J. Wong-Leung, C. E. Ascheron, M. Petravic, R. G. Elliman, and J. S.
Williams, Appl. Phys. Lett. 66, 1231 ͑1995͒.
9 J. Wong-Leung, J. S. Williams, R. G. Elliman, E. Nygren, D. J. Eagle-
sham, D. C. Jacobson, and J. M. Poate, Nucl. Instrum. Methods B 96, 253
͑1995͒.
10 E. R. Weber, in EMIS Datareviews No. 4, ͑INSPEC, the Institution of
Electrical Engineers, London and New York, 1988͒.
11 T. B. Massalski, Binary Alloy Phase Diagrams ͑ASM International, Met-
als Park, OH, 1995͒.
We suggest the following behavior for Au in Si during
annealing. Recrystallization of the near-surface amorphous
layer segregates Au to the surface. The concentration of Au
in this region ͑Ͼ1020 cmϪ3 within 100 Å of the surface͒ far
exceeds the Au solubility limit in good crystalline Si. Con-
12 F. M. Ross and W. M. Stobbs, Surf. Interface Anal. 12, 35 ͑1988͒.
13 J. Wong-Leung, E. Nygren, J. S. Williams, and D. J. Eaglesham, Mater.
Res. Soc. Symp. Proc. ͑in press͒.
14
¨
U. Gosele, W. Frank, and A. Seeger, Appl. Phys. A 23, 361 ͑1980͒.
418
Appl. Phys. Lett., Vol. 67, No. 3, 17 July 1995
Wong-Leung et al.
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
131.252.200.90 On: Thu, 04 Dec 2014 04:29:38