Vous êtes sur la page 1sur 6

Microelectronic Engineering 92 (2012) 53–58

Contents lists available at ScienceDirect

Microelectronic Engineering
journal homepage: www.elsevier.com/locate/mee

Influence of thermal cycles on the silylation process for recovering k-value


and chemical structure of plasma damaged ultra-low-k materials
T. Fischer a,⇑, N. Ahner a, S. Zimmermann b, M. Schaller c, S.E. Schulz b
a
Chemnitz University of Technology, Chemnitz, Germany
b
Fraunhofer ENAS, Chemnitz, Germany
c
Globalfoundries Dresden Module Two GmbH & Co. KG, Dresden, Germany

a r t i c l e i n f o a b s t r a c t

Article history: The integration of porous ultra low-k (ULK) materials within the interconnect system of integrated cir-
Available online 20 May 2011 cuits is one of the most promising approaches to reduce RC-delay or crosstalk. The application of plasma
processes for patterning, cleaning and photoresist removal is known to degrade the electrical parameters
Keywords: of the ULK by carbon depletion, densification and introduction of silanol species. In this study we inves-
k-Recovery tigate the influence of different thermally assisted immersion processes on the effectiveness of a silyla-
Plasma damage tion based k-restore process of plasma damaged porous low-k dielectrics. The results show that network
Porous low-k
recovery effects like methyl incorporation are different at the surface and in the depth of the damaged
Silylation
area addicted to precursor temperature and curing conditions. The most promising process sequence
was shown to consist of an immersion at temperatures of 100–150 °C with subsequent UV anneal
depending on the precursor used for recovery process.
Ó 2011 Elsevier B.V. All rights reserved.

1. Introduction
In this study we compare the effects of different thermally as-
Integration of porous ultra-low-k materials replacing SiO2 or sisted wet chemical silylation processes, thermal pre-treatments
dense low-k materials to reduce crosstalk and RC-delay introduces and UV curing processes on the effectiveness of the recovery of
new challenges to technology development [1]. Besides cleaning structural properties of a plasma damaged ULK. The structural
and chemical mechanical polishing, the application of plasma pro- changes of the dielectric are analyzed by FTIR in transmission
cesses for ULK dielectric patterning and photoresist removal are mode and using surface reflection. Additionally the surface condi-
critical processes, which may damage the porous materials [2]. tions before and after silylation are investigated by contact angle
Plasma processes are known to cause carbon depletion and measurement.
densification of the ULK’s surface region, which leads to an increase
in k-value due to the formation of silanol groups and moisture
absorption [3]. The replacement of oxidizing by reducing plasma 2. Experimental
chemistries (e.g. H2, He) is one opportunity to reduce the degrada-
tion of the ULK [4]. Another or additional approach is the recovery All experiments were done using a porous CVD SiCOH material
of the ULK’s electrical and structural properties by using silylation with a film thickness of 500 nm deposited on 300 mm Si substrates
processes [5–7]. A silylation reaction using chemicals like HMDS or provided by GLOBALFOUNDRIES Inc. The samples were treated by
OMCTS is able to replace hydrophilic polar silanol groups by hydro- a He/H2-ashing process to produce a damaged layer at the surface.
phobic non-polar Si(CH3)x compounds by the following reaction For recovering the chemical structure of the ULK a restore-solution
sequence [8]. (APRS) and rinsing set provided by Air Products and Chemicals Inc.,
as well as Hexamethyldisilazane (HMDS) diluted in Isopropanol
ðCH3 Þ3 Si—N—SiðCH3 Þ3 þ HO—Si— have been used. All samples were immersed in the restoration li-
! ðCH3 Þ3 —Si—O—Si— þ ðCH3 Þ3 —SiNH2 ð1Þ quid inside of a reaction tube at room temperature. After immer-
sion the reaction tube was put onto a hotplate to heat up the
ðCH3 Þ3 —SiNH2 þ HO—Si— ! ðCH3 Þ3 —Si—O—Si— þ NH3 silylation precursor. By this method the samples on the bottom
of the reaction tube as well as the precursor can nearly reach the
temperature of the hotplate underneath. The heated precursor
⇑ Corresponding author. could condensate in upper liquid to realize a closed system. All
E-mail address: tobias.fischer@zfm.tu-chemnitz.de (T. Fischer). experiments, including the transfer between the process steps,

0167-9317/$ - see front matter Ó 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.mee.2011.04.021
54 T. Fischer et al. / Microelectronic Engineering 92 (2012) 53–58

Immersion (20°C) Immersion (20°C)


Immersion (50°C) Immersion (50°C)
Immersion (100°C) Immersion (100°C)
0.025 Immersion (150°C) 0.025 Immersion (150°C)
Immersion (200°C) Immersion (200°C)
Reference damaged Reference damaged
Absorption

Absorption
0.020 0.020

0.015 0.015

0.010 0.010
700 750 800 850 700 750 800 850
-1 -1
Wavenumber [cm ] Wavenumber [cm ]
(a) Immersion with APRS (b ) Immersion with HMDS
Fig. 1. Reflection FTIR spectra of samples immersed at different temperatures.

were done in nitrogen atmosphere to prevent an influence by at higher temperatures, shown in Fig. 1, can be observed. This indi-
ambient conditions. Structural changes were compared by trans- cates that diffusion of molecules increased with increasing temper-
mission FTIR measurement as well as surface reflection FTIR mea- ature, but silylation reaction decreases.
surement at 80° incident angle. All measurements were done using
the Bruker Tensor 27 FTIR spectrometer.
3.2. Immersion followed by UV curing

3. Results and discussion Fig. 3 shows the reflection FTIR spectra of both, APRS and HMDS
treated samples with subsequent UV radiation at 254 nm for
As reported earlier, FTIR spectra in the spectral range from 770 2 min. Compared to the reflection spectra of the immersed only
to 860 cm1 show the most characteristic peaks indicating destruc- samples, peak intensity of SiMe1 and SiMe2 increased at the surface
tion or recovery of Si–Methyl (SiMe) bonds within the low-k-mate- by using UV curing, but show nearly no temperature dependency.
rial [9]. Comparing the transmission and reflection spectra, effects Fig. 3(a) shows an increase of the CH2 peak at 735 cm1. APRS- as
like diffusion into the bulk material and surface reactions can be well as HMDS-molecules do not have CH2 fragments in their struc-
described and differentiated. ture. This indicates that UV radiation also causes hydrogen separa-
tion from the methyl groups and is able to support a bonding of
3.1. Immersion without additional treatments unstable precursor fragments to the dielectric network. An in-
crease of this effect with increasing temperature suggests that
Fig. 1 shows the reflection FTIR spectra for samples immersed in the thermal energy can also support the separation of hydrogen
APRS and HMDS. A comparison of these spectra indicates a stron- from methyl compounds. UV radiation also causes crosslinking in
ger surface silylation reaction for the HMDS treated samples than Si–methyl compounds, seen in an increase of SiMe1 and SiMe2.
for the samples treated with APRS. A split of the SiMe2 peak, seen Comparing the reflection data of APRS treated samples with the
in Fig. 1(b), indicates a change of the bonding angle of the HMDS transmission data, seen in Fig. 4a, an UV anneal for samples im-
treated samples. This change in bonding angle causes a change in mersed at low temperatures leads to a higher repair effect at the
water bonding angle which is an indicator of hydrophobicity of surface, but causes a decrease of repair effect in the depth. This
the sample surface. For APRS treated samples the contact angle in- indicates an incomplete silylation reaction in deeper regions of
creased from 25° at the damaged surface up to 45° after silylation, the dielectric. UV treatment may lead to desorption of unstable
whereas the contact angle of the HMDS treated samples shows an precursor fragments from the deeper regions of the damaged area.
increase up to 85°. This indicates that HMDS as precursor for sily- Transmission spectra in Fig. 4 show the same temperature depen-
lation process causes higher hydrophobicity than APRS. Looking at dency like the immersed only samples, but for the APRS samples a
the transmission spectra in Fig. 2 for both immersion processes a smaller repair effect in the depth can be seen. Both spectra show a
weak temperature dependency can be observed. Immersion at decrease in repair effect at 200 °C showing that silylation reaction
50 °C shows the best results regarding the restoration of SiMe2, may be hindered at this temperature. HMDS shows a stronger tem-
whereas the peak of SiMe1 increases with increasing temperature. perature dependency concerning silylation reaction in the deeper
Comparing the transmission data of APRS and HMDS, APRS treated area of the damaged region with only a slight increase in SiMe2
samples show higher depth diffusion, seen by an increase of both groups, whereas the characteristic peak for SiMe1 groups show a
methyl peaks. This effect may be due to an increased reactant dif- strong temperature dependency. An increase of SiMe1 groups with
fusion at higher temperatures. There may also occur a stronger increasing temperatures up to 150 °C indicates that thermal treat-
silylation reaction due to higher reaction energy provided by ther- ment stabilizes bonding of SiMe1 groups. This effect also suggests
mal heating. Above 50 °C the repair molecule’s structure can be- that SiMe1 compounds can diffuse better through the material be-
come instable and methyl groups can be removed from the cause of smaller molecule size. On the one hand a decrease of
molecule itself. This would also lead to better diffusion within SiMe1 at 200 °C can be explained by desorption of OH groups at
the porous structure because of smaller molecule fragment size. 170 °C [10], which hinders the silylation reaction due to the fact
For both samples a decrease of the repair effect at the surface layer that OH is needed as an reaction partner for the silylation agent.
T. Fischer et al. / Microelectronic Engineering 92 (2012) 53–58 55

0.160 0.160

Immersion (20°C) Immersion (20°C)


0.155 0.155
Immersion (50°C) Immersion (50°C)
Immersion (100°C) Immersion (100°C)

Absorption normalized
Absorption normalized

0.150 Immersion (150°C) 0.150 Immersion (150°C)


Immersion (200°C) Immersion (200°C)
Reference damaged Reference damaged
0.145 0.145

0.140 0.140

0.135 0.135

0.130 0.130
770 780 790 800 780 800
-1 -1
Wavenumber [cm ] Wavenumber [cm ]

(a) Immersion with APRS (b) Immersion with HMDS


Fig. 2. Transmission FTIR spectra of samples immersed at different temperatures.

Immersion (20°C) + UV (2 Min) Immersion (20°C) + UV (2 Min)


Immersion (50°C) + UV (2 Min) Immersion (50°C) + UV (2 Min)
Immersion (100°C) + UV (2 Min) Immersion (100°C) + UV (2 Min)
Immersion (150°C) + UV (2 Min) Immersion (150°C) + UV (2 Min)
0.025 0.025
Immersion (200°C) + UV (2 Min) Immersion (200°C) + UV (2 Min)
Reference damaged Reference damaged
Absorption

Absorption

0.020 0.020

0.015 0.015

0.010 0.010
700 750 800 850 700 750 800 850
-1 -1
Wavenumber [cm ] Wavenumber [cm ]

(a) Immersion with APRS (b ) Immersion with HMDS


Fig. 3. Reflection FTIR spectra of samples immersed at different temperatures with subsequent UV at 254 nm for 2 min.

0.160 0.160

Immersion (20°C) + UV (2 Min)


0.155 Immersion (50°C) + UV (2 Min) 0.155 Immersion (20°C) + UV (2 Min)
Immersion (100°C) + UV (2 Min) Immersion (50°C) + UV (2 Min)
Immersion (150°C) + UV (2 Min) Immersion (100°C) + UV (2 Min)
Absorption normalized
Absorption normalized

0.150 Immersion (200°C) + UV (2 Min) 0.150 Immersion (150°C) + UV (2 Min)


Reference damaged Immersion (200°C) + UV (2 Min)
Reference damaged
0.145 0.145

0.140 0.140

0.135 0.135

0.130 0.130
770 780 790 800 770 780 790 800
-1 -1
Wavenumber [cm ] Wavenumber [cm ]

(a) Immersion with APRS (b) Immersion with HMDS


Fig. 4. Transmission FTIR spectra of samples immersed at different temperatures with subsequent UV at 254 nm for 2 min.
56 T. Fischer et al. / Microelectronic Engineering 92 (2012) 53–58

TPT + Immersion (20°C) TPT + Immersion (20°C)


TPT + Immersion (50°C) TPT + Immersion (50°C)
TPT + Immersion (100°C) TPT + Immersion (100°C)
0.025 TPT + Immersion (150°C) 0.025 TPT + Immersion (150°C)
TPT + Immersion (200°C) TPT + Immersion (200°C)
Reference damaged Reference damaged
Absorption

0.020

Absorption
0.020

0.015 0.015

0.010 0.010
700 750 800 850 700 750 800 850
-1
Wavenumber [cm ] Wavenumber [cm ]
-1

(a) Immersion with APRS (b) Immersion with HMDS


Fig. 5. Reflection FTIR spectra of samples immersed at different temperatures with thermal pre-treatment at 300 °C.

0.160 0.160

TPT + Immersion (20°C)


0.155 TPT + Immersion (50°C) 0.155 TPT + Immersion (20°C)
TPT + Immersion (100°C) TPT + Immersion (50°C)
Absorption normalized

Absorption normalized

TPT + Immersion (150°C) TPT + Immersion (100°C)


0.150 0.150
TPT + Immersion (200°C) TPT + Immersion (150°C)
Reference damaged TPT + Immersion (200°C)
0.145 0.145
Reference damaged

0.140 0.140

0.135 0.135

0.130 0.130
770 780 790 800 780 800
-1
Wavenumber [cm ] Wavenumber [cm ]
-1

(a) Immersion with APRS (b) Immersion with HMDS


Fig. 6. Transmission FTIR spectra of samples immersed at different temperatures with thermal pre-treatment at 300 °C.

TPT + Immersion (20°C) + UV (2 Min) TPT + Immersion (20°C) + UV (2 Min)


TPT + Immersion (50°C) + UV (2 Min) TPT + Immersion (50°C) + UV (2 Min)
TPT + Immersion (100°C) + UV (2 Min) TPT + Immersion (100°C) + UV (2 Min)
0.025 TPT + Immersion (150°C) + UV (2 Min) 0.025 TPT + Immersion (150°C) + UV (2 Min)
TPT + Immersion (200°C) + UV (2 Min) TPT + Immersion (200°C) + UV (2 Min)
Reference damaged Reference damaged
Absorption

0.020
Absorption

0.020

0.015
0.015

0.010
700 750 800 850 0.010
700 750 800 850
-1
Wavenumber [cm ] -1
Wavenumber [cm ]

(a) Immersion with APRS (b ) Immersion with HMDS


Fig. 7. Reflection FTIR spectra of samples immersed at different temperatures with thermal pre-treatment at 300 °C and subsequent UV at 254 nm for 2 min.
T. Fischer et al. / Microelectronic Engineering 92 (2012) 53–58 57

PC + Immersion (20°C) + UV (2 Min) TPT + Immersion (20°C) + UV (2 Min)


0.160 TPT + Immersion (50°C) + UV (2 Min) 0.160 TPT + Immersion (50°C) + UV (2 Min)
TPT + Immersion (100°C) + UV (2 Min) TPT + Immersion (100°C) + UV (2 Min)
TPT + Immersion (150°C) + UV (2 Min) TPT + Immersion (150°C) + UV (2 Min)
0.155 0.155 TPT + Immersion (200°C) + UV (2 Min)
TPT + Immersion (200°C) + UV (2 Min)
Reference damaged Reference damaged
Absorption normalized

Absorption normalized
0.150 0.150

0.145 0.145

0.140 0.140

0.135 0.135

0.130 0.130
770 780 790 800 770 780 790 800
-1 -1
Wavenumber [cm ] Wavenumber [cm ]
(a) Immersion with APRS (b ) Immersion with HMDS
Fig. 8. Transmission FTIR spectra of samples immersed at different temperatures with thermal pre-treatment at 300 °C and subsequent UV at 254 nm for 2 min.

On the other hand desorption of precursor molecules at this tem- all peaks for APRS samples treated at 200 °C. Also the HMDS sam-
perature may happen. A change in contact angle compared to the ples show a decrease of the repair effect at this temperature. This
immersed only samples could not be observed. indicates that samples treated at this temperature have a lot of un-
bound groups compared to the samples without thermal pretreat-
3.3. Thermal pre-treatment followed by immersion ment (Fig. 5). Comparing the reflection and transmission data to
the spectra of samples without thermal pre-treatment (Figs. 5
In Figs. 5 and 6 the reflection and transmission spectra of sam- and 6) a decrease in repair effect can be observed. This is caused
ples immersed at different temperatures with thermal pre-treat- by missing silanol groups removed by thermal treatment, which
ment at 300 °C for 10 min can be seen. The spectra show a slight are need for a complete silylation reaction. These effects also indi-
increase of SiMe1 and SiMe2 absorption for temperatures of 50 °C cate that methyl groups observed in spectra 5 and 6 are unlinked
and higher, associated with an increase in the peak intensity of compounds of the precursor desorb due to UV radiation. A restore
these bonds. This indicates that silylation reaction is consolidated process at 200 °C increases this effect, because additional silanol
because of the formation of silanol dangling bonds. The tempera- compounds are removed. For these samples a change in bonding
ture dependency of the APRS treated samples is stronger. Up to angles compared to the rest of the samples could not be observed
50 °C an increase in SiMe1 and SiMe2 can be seen. For higher tem- (see Fig. 8).
peratures a continuous decrease of these peaks can be observed.
Clark and Schwab [10] stated, that at temperatures higher than
170 °C silanol groups will desorb from the network. With missing 4. Conclusion
silanol groups a silylation process is hindered and unbonded pre-
cursor fragments reside in the network. The decrease at higher We compared the influence of different thermally assisted li-
temperatures is explainable by a desorption of the molecules at quid phase k-recovery processes supported by different curing
these temperatures. These effects are smaller for APRS. Comparison steps. It was shown that there is a strong temperature dependency
of reflection data to transmission data shows a strong temperature of the liquid phase k-recovery process concerning the recovery of
dependency concerning the depth diffusion of SiMe compounds. network conditions at the surface regions and in the bulk of the
The decrease of SiMe2 (Fig. 6) compared to the data of pure immer- dielectric. It was shown that additional thermal energy can in-
sion (Fig. 2) also indicates a decreased silylation reaction caused by crease precursor penetration into the bulk of the dielectric and pre-
the loss of silanol groups. An increase of the SiMe1 peak for temper- cursor molecule fragmentation. We observed that improved
atures up to 150 °C can also lead to a fragmentation of the mole- hydrophobicity is caused mainly by changes in bonding angle of
cules associated with different diffusion effects because of methyl groups to silicon and less by their amount. UV treatment
thermal pre-treatment. Comparing the reflection spectra and causes a stronger repair effect concerning recovery of network at
transmission spectra shows that temperatures up to 150 °C lead the surface, but causes damage in the depth of the dielectric. For
to stronger diffusion, but cause a smaller network arrangement immersion and immersion with subsequent UV treatment, the
of precursor fragments. A strong decrease for both SiMe1 and SiMe2 experiments showed that at temperatures up to 50 °C reactant
species seen in the transmission spectra suggests an inchoate diffusion without fragmentation of the repair species occurs. A
desorption of precursor groups at this temperature. The samples fragmentation of the silylation agents starts at 100 °C and a tem-
showed no difference in contact angle compared to the immersed perature increase of up to 200 °C finally stops the silylation due
only samples. to missing silanol groups and the incidence of desorption effects.
With thermal pre-treatment silanol groups were removed and a
3.4. Thermal pre-treatment followed by immersion and UV curing silylation reaction was hindered. An additional UV curing step
shows that with thermal pre-treatment, accompanied with miss-
As described before UV anneal can cause crosslinking and ing silanol compounds for reaction, only unstable or unlinked mol-
desorption of unstable precursor fragments. Looking at the results ecule fragments have formed. Considering the investigated process
in Fig. 7, it can be seen that UV anneal causes a strong decrease in parameters and sequences, the most promising process parameters
58 T. Fischer et al. / Microelectronic Engineering 92 (2012) 53–58

for recovering the network structure of a plasma damaged ultra References


low-k material seem to be an immersion process at 100–150 °C
depending on the precursor with subsequent UV curing. [1] International Technology Roadmap for Semiconductors 2009. Available
from: <http://www.itrs.net/Links/2009ITRS/2009Chapters_2009Tables/2009_
Interconnect.pdf>.
[2] E. Ryan et al., Mater. Res. Soc. Symp. Proc. 766 (2003).
[3] D.L. Moore et al., J. Electrochem. Soc. 152 (7) (2005) G528–G533.
Acknowledgments [4] T. Chang et al., J. Vac. Sci. Technol. B26 (2008) 219.
[5] T. Chang et al., J. Electrochem. Soc. 149 (2002) F81.
The project described in this publication has been funded in line [6] K. Kohmura et al., Thin Solid Films 515 (2007) 5019.
[7] Y. Kayaba et al., Jpn. J. Appl. Phys. 47 (2008) 8364.
with the technology funding for regional development (ERDF) of
[8] T. Rajagopalan et al., Appl. Surf. Sci. 252 (2006) 6323.
the European Union and by funds of the Free State of Saxony. All [9] N. Ahner et al., AMC Conference Proceedings (2009).
graphics have been generated by Fraunhofer ENAS. [10] P.G. Clark, B.D. Schwab, Semicond. Int. 26 (2003) 46.

Vous aimerez peut-être aussi