enhanced chemical vapor deposition
5-5-1 Motivation
As minimum device features shrink below 180 nm, the increase in propagation delay, the resistance and capacitance delay (RC) of the interconnect has become a limiting factor in
182
ultra-large scale Integration (ULSI) device performance. Since RC delay is a product of the resistance in the metal interconnect (R) and the capacitance between the metal line (C), incorporating copper (Cu) wiring and low-k dielectric to replace the conventional AlCu/SiO2
into interconnect technology can effectively reduce the RC delay [61, 99, 121].
Various low dielectric constant materials have been proposed to decrease the time delay caused by capacitance. Recently, organosilicate glass (OSG), deposited by PE-CVD method using various organo-precursors, such as methylsilane (MS), Tetramethylsilanetetrasiloxane (TOMCATS) and Diethoxymethylsilane (DEMS), is the most promising low-k dielectric candidate [65, 121, 137, 141]. Among these precursors, DEMS precursor is a strong candidate based on the excellent film properties. Diethoxymethylsilan (DEMS ; H-Si(CH3) (OC2H5)2 )-based low-k films not only have a lower dielectric constant (k=2.8-3.0) but also have a greater hardness (higher cross link) as it contains an O:Si ratio of 2:1 in the precursor produced optimum films [136].
However, during the interconnect fabrication process, thermal cycle and photoresist stripping are the indispensable steps. Therefore, the physical (thickness and refractive index) and electrical properties (dielectric constant and leakage current) of the low-k interconnect dielectric are needed to be resistance against heat, moisture and chemical stress in order to prevent the degradation during the interconnect fabrication process. Furthermore, to ensure the high reliability performance of the high-speed ULSI, the moisture resistance is essential to avert moisture penetrating from the outside the package[66, 138, 160].
In this work, the stability of the low-k films deposited using a DEMS precursor against heat, moisture and chemical stresses is clarified. The low-k films prepared using DEMS or DEMS/O2 reactant gas are submitted to reliability tests to distinguish the stability divergence
183
for the effect of the addition of O2. Furthermore, the electrical measurements and material analyses have also been used to evaluate the low-k film before and after the reliability tests.
5-5-3 Experimental Procedures
Material prepared- All thin film deposition was performed on an Applied Materials Producer system with a 200 mm Producer chamber. The thin films were deposited on p-type (100) silicon substrates by radio frequency (13.56 MHz) PE-CVD with Diethoxymethylsilan (DEMS, CVD precursor) carried to the reaction chamber in the vapor phase by inert helium (He) gas. The chamber pressure and RF power were maintained at 6 Torr and 700 W, respectively, throughout the deposition process. The deposition temperature and He flow were kept 400oC, and 150 sccm, respectively. The DEMS flow rate was fixed at 1500 mgm and oxygen (O2) flow was varied from 0 to 250 sccm (herein the O2/DEMS ratio was 0~0.175).
Reliability Test- To determine the thermal stability, the films were annealed for 1h in a
nitrogen ambient at temperatures ranging from 400 to 800oC. Moreover, to mimic the thermal stresses encountered during Cu interconnects fabrication process; thermal annealing at 425oC in nitrogen ambient for 1 h was performed 7 times. For the humidity test, a pressure cooker test (PCT) was carried out at 120 oC, 100% relative humidity, and 2 atmosphere pressure for 168 h. To test the impact of O2 plasma on the film properties, the as-deposited DEMS-based films were exposed to O2 plasma environment in a cathode-coupled rf asher. The pressure and RF power were 10 mTorr and 200 W, respectively, and the process time was set at 60 s.
Analysis Method- The low-k films were analyzed for thickness and refractive index (RI,
184
at 633 nm) by reflectometer (SCI FilmTek) and/or ellipsometer (Nano-SpecR9100) before and after stressing. The thickness change is defined herein as
%
where THKstressing and THKAs.dep. represent the measured thickness after stressing and as-deposition.
Transmission FT-IR spectra were measured the chemical bonding of the film using Bio-Rad spectrometer at 4 cm-1 resolution. All spectra are an average of 32 scans and background corrected to a silicon reference. The dielectric constant (k) and leakage current density were measured by an SSM mercury probe cyclic voltammeter (CV) system (SSM 495) at 1 MHz frequency. The k value was obtained from the average of 9 sites measurement.
5-5-3 Results and Discussions
The dependence of the stress behavior of the as-deposited DEMS-based low-k films as a
function of O2/DEM ratios is shown in Figure 5-5-1. The intrinsic stress of the DEMS-based
low-k films becomes more tensile as the O2 flow rate is increased. Furthermore, thermal
stability had been evaluated by stress/temperature analysis. Figure 5-5-2 shows the stress
hysterisis curves of the low-k films with polymerization of DEMS and oxidation of DEMS
and O2. It reveals that the stress shift between the first thermal cycle and the second cycle was
suppressed for the low-k films prepared by DEMS and O2. Minimal change in stress is
observed for low-k films deposited using DEMS and O2. The shift magnitude is 2.0E07
dyne/cm2, significantly smaller than that with pure DEMS deposition (3.0E09 dyne/cm2).
185
This result implies that the low-k films deposited using DEMS and O2 have better thermal
stability.
In the Cu integration processes, there are at least 7-8 layers interconnect dielectric
deposition with the deposition temperature of around 400oC. The dependence of the heating
cycles on the degradation of the thickness and dielectric constant of DEMS-based low-k films
with various DEMS/O2 ratios are compared in Figure 5-5-3 and 5-5-4, respectively. For
DEMS-based low-k films, the thickness remains constant after performing 425oC annealing
cycling independence of DEMS/O2 ratios shown in Figure 5-5-3. This implies that low-k film
deposited using DEMS precursor has a higher bonding strength when subjected to the heating
tests related to other OSG films prepared by other precursors [107, 130, 131]. On the other
hand, it can be seen from Figure 5-5-5 that the change of the dielectric constant of low-k
films with DEMS/O2 was found to behave differently to that with DEMS only. The dielectric
constant of the low-k films deposited only using DEMS gradually increase with increasing
the 425oC heating cycles. In contrast, as O2 was incorporated into the reaction, the dielectric
constant of DEMS-based low-k films almost retains the same value, even after the seven
cycles of the 425 oC heating test. To further investigate the heating resistance of DEMS-based
low-k films, different heating temperatures, ranging from 400 to 800 oC, were performed. The
influence of the heating temperatures on the change of thickness and the dielectric constant of
DEMS-based low-k films for different DEMS/O2 ratios are shown in Figure 5-5-5 and 5-5-6,
186
respectively. The behavior is similar to other OSG low-k films using other precursors [130,
131], the dielectric constant of the low-k films deposited with DEMS or DEMS/O2 maintain a
stable value (k=2.8-3.2) at temperatures up to 600oC. Moreover, the dielectric constant of
low-k films deposited only using DEMS degrade to 3.6 as the heating temperature is
increased to 700oC and sharply increases to 5.0 as the temperature approaches 800oC. On the
other hand, the dielectric constant of low-k films deposited using DEMS/O2 does not degrade
until the annealing temperature is increased to 800oC. This indicates that DEMS-based low-k
films with O2 as an oxidant gas have a superior thermal resistance to that deposited only
using DEMS precursor. This result seems to imply that low-k films deposited using DEMS
and DEMS/O2 have different bonding structures, which exhibits a different thermal resistance.
To further investigate the difference in the bonding structure for these low-k films, FTIR and
X-ray Photoemission spectroscopy (XPS) analyses were conducted depicted in pervious
paragraph. Low-k film deposited using only DEMS gas has more –CH3 terminal bonds and
more C-Si4-XHx (x<2) bonds to form a cross-linking structure. In contrast, the deposited low-k
films contain less C-Si4-XHx (x<2) bonds and mono-mthylsilane group to form a micro-void
structure as O2 gas was added to the reaction. The speculated schematics of the low-k film
chemical bonding structures deposited using (I) DEMS, (II) DEMS/O2 are shown in Figure
5-5-7. Furthermore, it is worth to note that the dielectric constant of DEMS-based low-k films
deposited using DEMS/O2 further decrease after annealing with temperature below 600oC.
187
The decreasing dielectric constant could be attributed to the desorption of the small amount
water in the film and rearrangement of the amorphous structure during annealing. It was
observed from FTIR spectra that no significant change in the concentration of C-H bonds and
Si-CH3 bond is thermally stable up to 600oC. We believe that the decreasing dielectric
constant of low-k films deposited by DEMS/O2 was a result of the formation of open ring
structures in the films caused by the loss of water and CHx organic materials during annealing
at 400-600oC. Since the thermal resistance of Si-CHx-Si bonds is lower than Si-CH3, the
Si-CHx-Si bonds were converted to CHx organic materials and desorped during 400-600oC
annealing. This is positive to reduction the dielectric constant, but is negative to the thickness
stability. As the annealing temperature is increased to 700oC, the Si–CH3 absorbance peak
dramatically decreases and there is an increase in the oxide character of the film, and a loss of
methyl groups after the annealing process as shown in Figure 5-5-8. Interestingly, the
enhancement in Si-O characteristics is less for the low-k films deposited using DEMS/O2 film,
which also showed much smaller increases in the dielectric constant. This suggests that the
Si-CH3 bonding did not decompose until the thermal temperature above 700oC. In addition,
low-k film deposited only using DEMS, where Si is bonded with multi-methyl group (-CH3)
bonding would degrade easily during the higher temperature annealing.
The change in the dielectric constant of DEMS-based low-k films under the moisture
stress test in terms of O2/DEMS ratios is shown in Figure 5-5-9. It indicates a slightly
188
increase in the refractive index and the dielectric constant after a 168 h PCT for two different
deposition conditions. Compared with low-k films deposited using 3MS as precursor[73],
low-k films deposited using DEMS as precursor shows a better moisture resistance, implying
that this film has a surface hydrophobic property as a result of more –CH3 terminal bonds.
Additionally, the dielectric constant of low-k film deposited only using DEMS gas increases
to 2.93 from 2.86, which is slightly lower than that prepared by DEMS/O2 with a dielectric
constant of 2.96. More surface hydrophobic -CH3 bonds in low-k films prepared using DEMS
gas, hindering the moisture penetrate into the film, causes this divergence. More interesting to
note, the stress of low-k film tends to decline to neutral value after the moisture test.
Additionally, Thermal Desorption Spectrum (TDS) analysis indicated that the H2O peak was
observed at about 250oC and the desorption amount of H2O was greater for the low-k films
prepared using DEMS/O2 reaction. These imply that the deposited low-k film still contains
moisture in terms of physical absorption and low-k film produced only using DEMS gas have
better moisture resistance, consistence with the previous observation.
In the interconnect integration fabrication, the ILD layer etching and photo-resist
stripping processes are indispensable steps. Figure 5-5-10 shows the etching rate of
DEMS-based low-k film as a function of O2/DEMS ratios, performed using Ar/C5F8/N2 gas.
The etching rate of DEMS-based low-k films slightly increases with an increasing O2/DEMS
ratio. The higher etching rate for DEMS-based low-k films with a higher O2 flow rate is
189
suspected to porosity structure in the low-k film; that is, the film is less dense (refractive
index is lower for DEMS-based low-k films with a higher O2 flow rate). On the other hand,
the dielectric constant of post-etching DEMS-based low-k films almost remains the stable
value as the as-deposited films independent of the O2 flow rate. This indicates that low-k film
deposited using DEMS and O2 gas can efficiently withstand the treatment of the etching
chemical gas.
In conventional photo-resist stripping step process, O2 plasma ashing is commonly
implemented because of its better efficiency in removing the polymer. However, all porous
low-k films would suffer a degradation of the dielectric constant after exposing O2 plasma
ashing process17. Therefore, the effect of O2 plasma ashing on low-k films prepared by
DEMS or DEMS/O2 was investigated in this study. Figure 5-5-11 shows the change of
thickness and the dielectric constant of DEMS-based low-k film as a function of O2/DEMs
ratios. The thickness reduction can be negligible since the maximum thickness reduction is
about 2.5% for films deposited only using DEMS reactant. Additionally, the thickness
reduction is less 1% for low-k films deposited using DEMS/O2 reactant. A plausible
explanation is that terminated methyl groups in DEMS-based low-k films are oxidized in the
O2 plasma condition, in line with the following Eq. (5-5-1)~(5-5-3):
-Si-CH3 + H3C-Si- + O Æ -Si-OH + OH-Si- +CO +H2O (5-5-1)
-Si-OH + OH-Si- Æ -Si-O-Si- +H2O (5-5-2)
190
-Si-CHn-C-Si- + O Æ -Si-O-Si- +CO +H2O (5-5-3)
The forming -Si-O-Si- bonds are almost as dense as -Si-C-Si- as the bond length of Si-O
is similar to that of Si-C bonds. As a result, the thickness change underO2 oxidation can be
negligible, which might account for the lower reduction in the DEMS/O2 reaction because the
low-k films deposited using DEMS/O2 contains more Si-C-Si bonds.
Exception of the above oxidation, O2 plasma treatment has also been found to resulting
in dangling bonds arising from the enhanced breaking of Si-H bonds by oxygen radicals, and
damage through the following reaction, resulting in increased dielectric constant.
-Si-H + O Æ -Si-OH (5-5-4)
In contrast to the thickness reduction, the degradation in the dielectric constant of low-k
films deposited using DEMS/O2 gas become serious as O2 flow rate is increased. On the other
hand, low- k films deposited using DEMS gas show a lower change in the dielectric constant.
To solve the dielectric constant degradation issue for the deposited low-k films, the
dielectric constant of DEMS-based low-k films with post O2 plasma treatment was measured
after dry etch method, as shown in Figure 5-5-12. The purpose of dry etch is to remove the
dense layer, which is oxidized on the top of low-k film during the O2 plasma ashing process.
As can be seen, after the dry etch, the dielectric constant of low-k film deposited using DEMS
or DEMS/O2 reduced to about 3.3, which is close to the original as-deposition value. It
clearly shows that the low-k film inside is not degraded during the O2 ashing process. SEM
191
also displays the distinct two-layers inside the low-k film after the O2 plasma ashing. The
depth of the oxidation film increases with increasing O2 exposure time. This dense surface
can be removed during sputter etching prior to the metal deposition in the practical
fabrication process. As a result, control of the O2 plasma time and pre-sputter etching is a
feasible method for recovering the dielectric constant. Another approach to alleviate dielectric
constant degradation is using NH3 or N2 plasma treatment on the as-deposited low-k films.
This N2 plasma treatment would induce the N atom doping in low-k films and replace the
Si-H bonds to form a thin Si-N layer on the film surface. The dielectric constant of low-k
films prepared by DEMS/O2 is slightly increased from 2.78 to 2.86 after N2 plasma treatment.
The new forming layer can effectively impede the attack on the dielectric constant of low-k
film during the O2 plasma ashing process. Figure 5-5-13 shows the leakage current density at
2 MV/cm and dielectric constant of the N2 plasma treated low-k films before and after being
exposed to the O2 plasma treatment. The leakage current density remains at a stable value of
about 1.40E-8 A/cm2 at 2 MV/cm, which is slightly lower than the as-deposited low-k film
(1.65E-8 A/cm2 at 2 MV/cm) dut to the formation of Si-N bonds. In addition, the dielectric
constant of N2 plasma-treated low-k films deposited with 0.05 ratios of DEMS/O2 maintains
the stable value when compared to film without N2 plasma treatment. The results indicate this
functional group produced by N2 plasma treatment ensures that the low-k films have a
superior electrical performance to against O2 plasma ashing damage.
192
5-5-4 Summary
The resistance to heat, moisture stress and chemical test for organo-silicate glass (OSG) low-k film deposited using DEMS and various O2 flow was investigated. Low dielectric constant organo-silica-glass (OSG) film deposited using DEMS and O2 is shown to be the most reliable. The dielectric constants are stable even after a heating test at 700oC and a pressure cooler test for 168 h, and are superior to other PE-CVD low-k films deposited by other precursors. This excellent stability ensures the low-k film deposited using DEMS is suitable for application as multilevel interconnects, showing long-term reliability after fabrication. However, the O2 plasma ashing process leads to a dielectric degradation in deposited low-k film during photoresist removal processing. A N2 plasma treatment is proposed as a method of preventing the damage from an O2 plasma attack on the low-k film deposited using DEMS/O2 gas
193
Table 5-1-1. Physical properties data for He and Ar Carrier gas.
* Ionization energy denotes the Ionization energy of the nth electron.
Carrier gas First Second Third
He 24.59 54.42 4 2.551 13.43 4.68 x 10-4
Ar 15.76 27.63 40.74 40 3.542 3.15 9.03 x 10-4
Thermal conductivity (cal/cm-s-K) Ionization Energy (eV)
Molecular Weight (a.m.u)
Lennard-Jones Collision diameter(Å)
Mass diffusivity (cm2/s)
194
Table 5-2-2. Percentage of element from XPS Quantification (%).
O C Si O C Si
He 34.7 30.9 34.4 34.3 33.1 32.7
Ar 33.1 31.8 35.1 33.4 34 32.6
STD 33.6 31.6 34.8 33.1 33.2 33.8
Take-off 0o Take-off 60o
Carrier gas
195
Table 5-3-3. Related Binding energy and Percentage for Si2p peak.
Si(2p) Si4+-O4 Si4+-O3 Si4+-O2 Si4+-O1 Si4+-O0
Binding Energy(eV) 103.4 102.2 101.1 100.2 99.1
He 2% 23% 49% 21% 5%
Ar 2% 24% 47% 21% 6%
STD 2% 24% 48% 21% 6%
196
Figure 5-1-1. A series of reaction process steps of the low-k film formation using 3MS and O2.
Gas Phase Reaction
:
(CH3)3SiH + O2
Reaction gas
:
Intermediates:
(CH3)3Si+, CH3Si3+ ,O *………...
Gas Phase Diffusion
:
O2 , Carrier gas (Ar, He) Migration
Adsorption Reaction
Diffusion of by-product
197
Figure 5-1-2. The effects of the deposition temperature on the deposition rate of the low-k film processes.
0 200 400 600 800 1000 1200
200 250 300 350 400 450