Room Temperature Electroplated Copper Recrystallization:
In-Situ Mapping on 200/300 mm Patterned Wafers
K. J. Kozaczek1, R. I. Martin, L-Y. Huang, D. S. Kurtz, P.R. Moran
Hypernex, Inc,. 3006 Research Drive, State College, PA 16801
Abstract. The recrystallization kinetics of electroplated copper at room temperature was studied with a novel X-ray diffraction
(XRD) metrology tool. The relaxation of short-range strains, grain growth, and the evolution of crystallographic texture
(preferred orientation of grains) may be monitored in-situ across a 200 or 300 mm patterned wafer with a spatial resolution
ranging from 50 \im to 4 mm. The instrument provides for rapid and automated studies of the effects of deposition parameters of
liners and copper seed layers (gas flow, pressure, temperature), liner type, and electroplating conditions (current density,
rotational speed, bath chemistry) on recrystallization kinetics of sputtered and electroplated copper. The results are useful in
developing and optimizing the deposition and annealing processes. A particular example includes the study of room temperature
recrystallization kinetics of electroplated copper on of Ta and TaN/Ta type liners.
INTRODUCTION
The resistance of the damascene copper
interconnects to stress voiding and electromigration
depends, aside from line geometry, on microstructure
and interfaces. The microstructure can be optimized by
the selection of proper barrier layer materials and
optimizing process parameters such as annealing
temperature, which controls the copper grain size
distribution, grain boundary character distribution, and
crystallographic texture [1]. Recrystallization kinetics
of electroplated copper, at either room or elevated
temperature, is dependent on numerous factors.
Retarded recrystallization has been attributed to the
incorporation of plating additives in electroplated
copper, which aside from bath chemistry, is dependent
on rotation speed, flow rate and current density [2-5].
Higher recrystallization rates in thicker films have been
observed experimentally [6,7] and a model of grain
growth mechanisms has been proposed [8]. Other
reported factors that may affect the recrystallization rate
is the texture of the copper seed layer, which is
controlled by the barrier layer material [9] and the
deposition temperature [10]. The recrystallization rate
of electroplated copper deposited on non-texture seed
was higher than that of a film deposited on (111)
textured seed.
In thin films, recrystallization is usually
accompanied by the relaxation of short-range strains,
grain growth, and the evolution of crystallographic
texture (preferred orientation of grains). All these
phenomena can be monitored by x-ray diffraction
methods. The abnormal grain growth in Cu-Co and Cu
films is coupled to a transition of texture from a (111)
fiber to a (100) fiber [11]. A significant weakening
during recrystallization of the strong as-plated (111)
fiber texture was attributed to multiple twinning of
copper grains [12]. Higher temperature annealing of
copper blanket films leads to development of (511)
orientation (twin to (111)) and transition to (100) fiber
and its twin orientation (122) [13]. The transition from
(111) to its twin orientation (511) at room and elevated
temperature was reported in [14]. Neutron diffraction
has been used to study the recrystallization kinetics in
rolled copper through in situ texture measurements
[15]. The relief of microstrains and grain growth were
measured by the analysis of the diffraction line profile.
The decrease in width of the diffraction peak was used
to study the recrystallization of electroplated copper
[2,5].
The methods used to study the recrystallization
kinetics of copper films include the sheet resistance
measurement (indirect, volumetric), Focused Ion Beam
imaging (direct, surface), Transmission Electron
Microscopy (direct, volumetric), Electron Back
Scattered Diffraction (direct, surface), and X-ray
Diffraction (indirect, volumetric). Good correlation has
been shown between the sheet resistance and XRD
methods [5,16].
We have developed an XRD
instrument and a method for determining the
recrystallization kinetics of sputtered or electroplated
copper in-situ on 200/300 mm patterned wafers
containing fine features (ranging from 50 jtim to 4 mm)
[13,16].
Microstructure
evolution
during
recrystallization is monitored by following film texture
change as a function of time. This is accomplished in a
quantitative manner by monitoring the volume fractions
of the major texture components. Three partial pole
figures are collected simultaneously and used for
CP683, Characterization and Metrology for VLSI Technology: 2003 International Conference,
edited by D. G. Seiler, A. C. Diebold, T. J. Shaffner, R. McDonald, S. Zollner, R. P. Khosla, and E. M. Secula
© 2003 American Institute of Physics 0-7354-0152-7/03/$20.00
485
determining the orientation distribution function (ODF).
From the ODF the volume fraction of relevant texture
components (such as (111) and its first and second
generation twins (511) and (57 13), respectively) are
calculated and monitored during the recrystallization
period. In addition, grain growth and microstrain
relaxation are monitored by tracking the changes in the
diffraction line profile. The width of the diffraction line
profile is defined by the full width at half maximum
(FWHM).
The FWHM decreases during
recrystallization as the crystallites become larger and
the density of defects (such as vacancies and
dislocations) decreases [17].
This method was used to study in-situ the
recrystallization kinetics of electroplated copper
deposited on a and (3-Ta liners.
EXPERIMENT
Electroplated copper was deposited on 200
mm oxidized Si wafers containing arrays of parallel
lines 0.2 micron wide, with aspect ratio of 1 and pitch
of 0.3 micron. The thickness of copper varied across
the wafer from thinner at the center to thicker at the
edge. Two wafers were prepared by sputtering on two
different liners: 100ATaN/250Act-Ta/1200ACu, and
300A(3-Ta/1200ACu. The sputtering and electroplating
conditions were the same for both wafers. The
microstructure of copper overburden was monitored
immediately after the deposition in the clean room for
140 hours at 18 locations spaced uniformly over the
entire wafer. The XRD data collection time was 40
seconds/point and each point was tested at a frequency
of 15 minutes and a spatial resolution of 1x2.5 mm2.
The local film thickness was mapped using the XRD
technique described in more detail in [16].
RESULTS
The monitored parameters of copper
overburden microstructure were the volume fraction of
(111) fiber texture and the FWHM of the (111)
reflection. Figure 1 shows an example of the evolution
of these parameters during room temperature
recrystallization over a period of 140 hours. The
average correlation factor between the (111) fiber plots
and the FWHM plots for each wafer was 0.95
indicating the same sensitivity to changes in copper
microstructure. The data was collected at multiple
locations on a wafer during one recrystallization
experiment. Since there usually are variations in film
thickness (electroplated films intentionally had a
concave shape and are thicker at the edge), barrier layer
composition, and plating additive concentration across
the wafer, one microstructure map provides statistical
486
information for separating the effects of these variables
on the recrystallization kinetics. The film thickness
varied from 530 nm (center) to 750 nm (edge) for the
TaN/Ta/Cu wafer and from 480 nm to 800 nm for the
Ta/Cu wafer. The Ta/Cu wafer had a pure p-Ta liner
whereas the TaN/Ta/Cu had a pure oc-Ta liner in the
central regions with a radially increasing content of PTa reaching 80% at the wafer's edge, on average. The
electroplated copper films deposited on TaN/Ta/Cu
seed liners had a weaker (111) fiber texture (8% by
volume average across the wafer) right after deposition
as compared to 12% of (111) fiber in films deposited on
Ta/Cu seed liners. They also had a slightly higher
FWHM of (111) copper reflection that indicates a
combination of higher density of lattice defects and
smaller grain size. The x-ray diffraction transient
curves (as shown in Figure 1) were fitted with the
Avrami [18] curve even though the recrystallization of
the overburden above the trenches may not satisfy the
Avrami's assumption of random nucleation [20].
= /.-(/.-/ 0 )exp
(1)
where I0 and !«> are the quantities measured at time zero
and after long times (infinity), t0.5 is the time for 50%
recrystallization, and P is a constant that is proportional
to the product of the nucleation rate and grain growth
rate. Figure 2 shows the effects of the barrier layer
composition, overburden thickness, and additive
concentration on the recrystallization rate. Thickness
and phase composition of the liner were measured
simultaneously during the microstructure evolution.
For each type of liner, the local film thickness has a
predominant effect on recrystallization rate. Thickest
regions, located mostly close to the wafer's edge
recrystallized 2.5 times faster than the thinnest close to
wafer's center. The exponential character of this
relationship is in agreement with other reports [6-8].
Those regions may as well have a different
concentration of impurities that was not measured
directly. However, from Figure 2 it is evident that for a
given thickness there is a large scatter of
recrystallization times, which very likely is due to the
different impurity concentrations.
The impurity
concentration depends on the bath agitation (for a fixed
bath chemistry) and therefore has a radial dependence
across the wafer. From the data collected during
experiments, one can select the locations that have the
same film thickness but are located at different radii
and therefore have a different impurity concentration.
Figure 3 shows the dependence of the recrystallization
time on the radial location (impurity concentration).
The effect is of the same order of magnitude as the
effect of film thickness for this particular range of
thickness and deposition processes.
0.49 -
25
«= 0.09 -& -*''* ••'•^ ''^^''^'^^
£ 0.08 -
•| o.oe -
§ 0.45 -
\
06
?
*"-x^.
•§
Fast recrystallization
003
0.43
-
0.41
-
%
\
g
slow recrystallization
^'^^.^
'%V^^N
\
\
*S^VW
*5 0.39
v
o °- "
>
§ 0.47 ~»-***^ ^ ""^^*>^>v^
Slow recrystallization
''^>-*.
Fast recrystallization
0.37 -
. —————,_____ ,————— ,————— ,————— ,————— ,————— li. 0.35
0
20
40
60
80
100
120
140
- ——————————————————————— .————— •———— ,——
0
20
40
60
80
100
120
time (hours)
140
time (hours)
Figure 1. (a) Evolution of crystallographic texture (quantified as a volume fraction of (111) fiber), (b) grain growth and
microstrain relaxation as measured by the width of the diffraction f>eak (full width at half maximum, FWHM).
1000
-i
900
•=- 800-
- p-Ta liner
o 700-
§
"E 600
a-Ta liner
500
400
20
40
60
80
120
100
Time to 50% recrystallization (hours)
Figure 2. Room temperature recrystallization rates of EP Cu overburden deposited over arrays of 0.2 jam lines with a 0.3
pitch.
p-Ta liner
70
a-Ta liner
0>
20
80
85
distance from center (mm)
Figure 3. Recrystallization time dependence on radial distance from wafer's center. One is electroplated copper film with a
thickness of 630nm deposited on Ta/Cu seed stack (P~Ta liner), the other, 650 nm thick, was deposited on TaN/Ta/Cu seed stack
(a-Ta liner).
487
Ta assumes a heteroepitaxial relationship with TaNx
such as the (110) Ta planes are parallel to (111) TaN
and as a result, large tensile misfit stresses develop in
Ta epilayer. The presence of high density of defects
such as vacancies and dislocations may affect the
surface morphology of the interface between the
PVD/electroplated copper. The smaller grain size in
electroplated copper caused by pinning at impurities,
vacancies, and dislocations provides the driving force
for the abnormal grain growth at room temperature
[11]. The incubation time of room temperature
recrystallization was approximately 2 hours for the ocTa liner and 10-12 hours for P~Ta liner, on average.
Using this data and following the model proposed in
[11] the diffusion distance would be up to 2.5 times
smaller in copper deposited on a-Ta liners, indicating a
higher density of pinning sites. Also, the relatively
weak (111) texture and absence of (100) texture of
electroplated copper indicate that the surface energy
and strain energy minimizations were not the primary
driving forces during deposition and that these energies
may be available during the recrystallization process.
The average Avrami exponents P (see equation 1) were
determined to be 3 (on average) for both liners
indicating two-dimensional grain growth with a
constant nucleation rate [20]. Only the central parts of
both wafers had the P exponent equal to 2.5, indicating
two-dimensional growth with some site saturation.
This is consistent with the assumption that the center of
the wafer has more impurities incorporated in the
copper, which would slow down the grain growth.
The linear dependence on the radial distance from the
center may suggest that the impurity distribution is a
function of the linear velocity of the wafer surface with
respect to the bath. The regions close to the wafer's
center experience a low bath agitation and higher rates
of incorporating the additives into copper, and have
therefore the higher recrystallization times than the
wafer's edge with lower impurity levels. Such a
scenario is consistent with the arguments and data
presented in [2]. Similarly, by examining locations
positioned on the same radius but differing in film
thickness we may estimate the thickness effects on the
recrystallization kinetics.
Figure 4 shows such
dependence for both liners. The higher scatter of data
in the case of "ot-Ta" liner is very likely due to the fact
that the liner had different contents of P-Ta ranging
from 70% to 90%; higher p-Ta content locations had
longer recrystallization times.
The data obtained in the in-situ
recrystallization experiments shows that the type of the
liner had a dominant effect on the recrystallization
kinetics, more pronounced than the effects of the film
thickness (in ranges typical of product wafers) and
impurity incorporation. The liner may effect the
electroplated copper recrystallization indirectly,
through the quality of the copper seed layer. The XRD
measurements on numerous samples deposited under
various conditions showed a consistent trend in copper
liner microstructure. The PVD copper films deposited
on TaN/ct-Ta have smaller grain size and higher level
of microstrains, as evidenced by FWHM analysis. The
crystallographic texture of copper was predominantly
(111) fiber that was weaker and has a broader
distribution in films deposited on oc-Ta liners. The
recrystallization rate of electroplated copper deposited
on non-texture seed was higher than that of a film
deposited on (111) textured seed [10]. The XRD
studies of ot-Ta films deposited on 5- TaNx showed that
SUMMARY
The room temperature recrystallization kinetics of
electroplated copper has been studied with a novel Xray diffraction (XRD) metrology tool. The relaxation
55-
P-Ta liner
50
1?
45
'
3
J2 40*
J35-
oc-Ta liner
30*
25
20
500
550
600
650
700
750
800
850
900
Film thickness (nm)
Figure 4. Effect of film thickness on the rectystallization kinetics for two different liners.
488
950
10. Jiang, Q-T., et al., "Influence of Cu Seed Deposition
Temperature on Electroplated Cu Texture Formation in
Damascene Structures," in Mat. Res. Soc. Symp. Proc.,
Vol 562, MRS, Warrendale, 1999, pp. 229-234.
11. Harper, J.M.E., et al., "Crystallographic Texture Change
During Abnormal grain Growth in Cu-Co Thin Films,"
Appl Phys. Lett. 65 (2), 177-179 (1994).
12. Lingk, C., Gross, M.E., Brown, W.L., "Texture
Development of Blanket Electroplated Copper Films,"
J. Appl Phys. 87 (5) 2232-2236 (2000).
13. K.J. Kozaczek, et al., "Methodology of Quantitative
Texture Analysis in Thin Films and Interconnects," in
Advanced Metallization Conference 2000, ed. D.
Edelstein et al. MRS, Warrendale, PA, 2000, pp. 193197.
14. Okabayashi, H., et al., "Microstructure in Electroplated
Copper Films," in Advanced Metallization Conference,
edited by M. Gross et al., MRS, Warrendale, PA, 2000,
pp. 93-99.
15. Hansen, N., Leffers,T., Kjems, J.K., "Recrystallization
Kinetics in Copper Investigated by In Situ Texture
Measurements by Neutron Diffraction," Ada Metall 29,
pp. 1523-1533(1981).
16. K.J. Kozaczek, et al., "Crystallographic Texture and
Phase Metrology During Damascene Copper
Processing," in Materials Research Society Symp. Proc.
Vol. 721 MRS, Warrendale, PA, 2002, pp. 5-16.
17. Klug, H.P., Alexander, L.E., X-Ray Diffraction
Procedures, John Wiley & Sons, New York, 1974, pp.
618-708.
18. Avrami M., J. Chem. Phys. 8, 212 (1940).
19. Lingk, C., Gross, M.E., "Recrystallization Kinetics of
Electroplated Cu in Damascene trenches at Room
Temperature," J. Appl. Phys. 84 (10) 5547-5553.
20. Christian, J.W., Theory of Transformations in Metals
and Alloys, Pergamon, Oxford, England, 1975 p. 525.
of short-range strains, grain growth, and the evolution
of crystallographic texture (preferred orientation of
grains) were monitored in-situ across 200 mm
patterned wafers with two distinctly different liners.
The wafer mapping capabilities coupled with Avrami
type analysis allowed us to extract the effects of
different factors on the recrystallization kinetics. The
liner type had a predominant effect on recrystallization
rate followed by film thickness and impurity
incorporation in electroplated copper. The observed
grain growth was two-dimensional with a constant
nucleation rate. Electroplated copper recrystallized
two times faster on a ot-Ta liner as on a (3-Ta liner. The
impurity distribution and the thickness variation caused
the recrystallization rates across each wafer to differ by
a factor of 2.5.
REFERENCES
1.
2.
3.
4.
5.
6.
7.
8.
9.
Zschech, E., Slum, W., Zienert, L, Besser, P.R., Z
Metallkd. 92, 803-809 (2001).
Malhotra, S. G., et al., " Copper Room Temperature
Resistance Transients as a Function of Electroplating
Parameters," in Advanced Metallization Conference,
edited by M. Gross et al., MRS, Warrendale, PA, 2000,
pp. 77-83.
Lee, H., et al., " Effect of Plating Current Density and
Solution Additive on the Microstructure and
recrystallization Rate of Electroplated Copper Films," in
Advanced Metallization Conference, edited by M. Gross
et al., MRS, Warrendale, PA, 2000, pp. 63-68.
Gross, M.E.,et al., "Mechanistic Studies of Room
Temperature Recrystallization
of Electroplated
Damascene Copper and Sputter-Deposited Copper," in
Advanced Metallization Conference, edited by M. Gross
et al., MRS, Warrendale, PA, 2000, pp. 85-91.
Stafford, G.R., et al., "The Influence of Additives on the
Room Temperature Recrystallization of Electroplated
Copper," in Advanced Metallization Conference, edited
by M. Gross et al., MRS, Warrendale, PA, 2000, pp.
109-115.
Walther,
D.,
et
al.,
"Room
Temperature
Recrystallization of Electroplated Copper Thin Films:
Methods and Mechanisms," in Materials, Technology
and Reliability for Advanced Interconnects and Low-k
Dielectrics," edited by Oerlein, G.S., et al., MRS,
Warrendale, PA, 2001, pp. D10.1.1-D10.1.10.
Jiang, Q.T., et al., "Electroplated Copper
recrystallization in Damascene Structures at Elevated
Temperatures," in Mat. Res. Soc. Symp. Proc., Vol 564,
MRS, Warrendale, 1999, pp. 429-434.
Harper, J.M.E., et al. "Mechanisms for Microstructure
Evolution in Electroplated Copper Thin Films at Near
Room Temperature," J. Appl Physics 86 (5), 2516-2525
(1999).
Ueno, K., Ritzdorf, T., Grace, S., "Seed Effect on SelfAnnealing of Electroplated Copper Films," in Advanced
Metallization Conference edited by Sandhu, G.S., et al.,
MRS, Warrendale, PA, 1999, pp. 95-101.
489