Evaluation of FNO and F3NO as substitute gases for semiconductor CVD chamber cleaning

Article abstract of DOI:10.1149/1.1616000

Two types of FNO compounds (FNO and F₃NO) were evaluated as candidates for new chemical vapor deposition (CVD) chamber cleaning gases. NF₃ and C₂F₆ were measured as the reference. Like NF₃, as these gases have no carbon in their molecules, no perfluoro carbon (PFC) is thought to be emitted. FNO is a compound highly susceptible to hydrolysis. F₃NO is expected to decompose more easily than NF₃ in the atmosphere because its N-F bond has been weakened by introducing an N=O bond into the molecule. Hence, the contribution to global warming of these compounds is expected to be small. Performance of these gases was evaluated by measuring their etch rates and their exhaust gases. The results showed that the etch rate of F₃NO is virtually the same as that of NF₃, whereas the etch rate of FNO is about 1/2 that of NF₃. However, from the results of exhaust gas analysis, it was found that an unexpected side reaction had occurred in the chamber, and therefore, it was confirmed that it is important to take this property into account in designing applications.

Full text of DOI:10.1149/1.1616000

Journal of The Electrochemical Society, 150 ͑11͒ G707-G710 ͑2003͒
G707
0013-4651/2003/150͑11͒/G707/4/$7.00 © The Electrochemical Society, Inc.
Evaluation of FNO and F3NO as Substitute Gases
for Semiconductor CVD Chamber Cleaning
T. Yonemura,a,z K. Fukae,a,c Y. Ohira,a Y. Mitsui,a
T. Takaichi,a A. Sekiya,b and T. Beppua
aSemiconductor CVD Chamber Cleaning Project, Research of Innovation Technology for the Earth, Tsukuba,
Ibaraki 305-8565, Japan
bNational Institute of Advanced Industrial Science and Technology, Research Center for Developing
Fluorinated Greenhouse Gas Alternatives, Tsukuba, Ibaraki 305-8565, Japan
cKanto Denka Kogyo Company, Limited, Shibukawa, Gunma 377-8513, Japan
Two types of FNO compounds ͑FNO and F3NO) were evaluated as candidates for new chemical vapor deposition ͑CVD͒ chamber
cleaning gases. NF3 and C2F6 were measured as the reference. Like NF3 , as these gases have no carbon in their molecules, no
perfluoro carbon ͑PFC͒ is thought to be emitted. FNO is a compound highly susceptible to hydrolysis. F3NO is expected to
decompose more easily than NF3 in the atmosphere because its N-F bond has been weakened by introducing an NϭO bond into
the molecule. Hence, the contribution to global warming of these compounds is expected to be small. Performance of these gases
was evaluated by measuring their etch rates and their exhaust gases. The results showed that the etch rate of F3NO is virtually the
same as that of NF3 , whereas the etch rate of FNO is about 1/2 that of NF3 . However, from the results of exhaust gas analysis,
it was found that an unexpected side reaction had occurred in the chamber, and therefore, it was confirmed that it is important to
take this property into account in designing applications.
© 2003 The Electrochemical Society. ͓DOI: 10.1149/1.1616000͔ All rights reserved.
Manuscript received October 28, 2002. Available electronically September 22, 2003.
rates of gases. The measurement of the etch rates ͑Å/min͒ was car-
ried out by etching approx. 10000 Å of SiO2 film deposited in ad-
vance on a 6 in. sample wafer and calculating the difference in film
thickness measured before and after the plasma discharge. It is
thought that the higher the etch rate of the compound is, the higher
its reactivity with SiO2 is and the more advantageous the compound
is for plasma CVD cleaning. For film thickness measurements, a
spectroscopic reflectometric film thickness measuring instrument
͑model 3000 manufactured by Nanometrics Japan͒ was used. Film
thickness was measured at nine points on the wafer and the etch rate
was calculated from the average of these values. As well as the etch
rate, within-wafer uniformity ͑WINWU͒ was obtained by the fol-
lowing equation. Poor uniformity means that plasma is not stably
generated, and it is not desirable as the CVD chamber cleaning
process
Perfluoro carbons ͑PFCs͒ such as C2F6 or CF4 have been com-
monly used for chamber cleaning of plasma-enhanced chemical va-
por deposition ͑PECVD͒ systems. However, these gases have a long
atmospheric lifetime and a large global warming potential ͑GWP͒.
Thus, they are designated as greenhouse gases which should be
reduced according to the Kyoto Protocol.1 NF3 has also been used as
a part of the CVD cleaning gases. Although it emits no PFC, as it
has no carbon in the molecule, its GWP is also large. Moreover,
because of its chemical stability, a considerable amount of energy is
required for its abatement. The objective of our project2 which
started in October 1998 is to reduce emission of greenhouse sub-
stances by selecting optimum alternatives.
In the present paper, evaluation of FNO compounds ͑FNO and
F3NO) is discussed as a part of the new alternative gas development
program. Like NF3 , FNO and F3NO do not emit PFC, either. Fur-
thermore, by introducing the NϭO bond into its molecule, it is
anticipated that the atmospheric lifetime of these gases will become
shorter and its contribution to global warming will be less than that
of NF3 .
In order to conduct the evaluation of cleaning gases perfectly, it
may have to be carried out in an actual PECVD system. Ideally, the
cleaning should be carried out after the actual deposition and the
cleaning performance should be evaluated. In the present study, we
evaluated the cleaning performance with an experimental plasma
tool. The cleaning performance was evaluated by measuring the etch
rate of SiO2 film which was deposited in advance on a sample wafer
placed on the electrode. Although the magnitude of the etch rate
may not have completely corresponded to the length of the cleaning
time, it was considered to be sufficient for simple evaluation of the
performance of cleaning gases. To evaluate the impact of these gases
on the environment, components of exhaust gas were measured by
Fourier transform infrared ͑FTIR͒ spectroscopy.
WINWU U Ϯ%͒ ϭ
Emax Ϫ Emin͒/ Emax ϩ Emin͒ 100
͓͑ ͑ ͔
͑
Emax maximum etching quantity ͑Å)
Emin minimum etching quantity ͑Å)
͓1͔
The exhaust gas was analyzed with an FTIR spectroscope ͑Mat-
toson Infinity Gold͒. A 6 in. quartz plate was placed on the electrode
instead of a sample wafer and plasma was discharged. Dilute N2 was
run in the dry pump at about 15.5 L/min, and sampling was done
from the latter of the pump at a rate of 7 L/min. After running it
through the FTIR cell (BaF2 , cell length of 2 cm͒, it was returned to
the exhaust line.
In the present study, the conditions of the plasma discharge were
determined based on our previous experiment using C2F6 .3 In the
present experiment, the effect of the gas concentration was studied
under the fixed conditions of chamber pressure of 250 Pa, total gas
flow rate of 300 standard cubic centimeters per minute ͑sccm͒, elec-
trode temperature of 300°C, radio frequency ͑rf͒ power of 750 W,
and distance between electrodes of 50 mm. NF3 and C2F6 were
measured as the reference gases. F3NO, FNO, and NF3 were diluted
by Ar. In the case of C2F6 , it was diluted by O2 .
Experimental
FNO and F3NO used for the evaluation were purchased from
Advance Research Chemicals, Inc., and their purity was 99%. All
experiments were conducted by using the capacitive coupled plasma
͑CCP͒ system ͑ANELVA͒. This system is outlined in Fig. 1. The
performance of cleaning gases was evaluated by comparing etch
Results and Discussion
In Table I chemical properties of FNO and F3NO are shown.
FNO and F3NO are gases at room temperature. Figure 2 presents the
etch rates as a function of gas concentration.
z E-mail: yonemura@cvd-rite.gr.jp

G708
Journal of The Electrochemical Society, 150 ͑11͒ G707-G710 ͑2003͒
Figure 1. Schematic drawing of the experimental apparatus.
Figure 2. Etch rates as a function of gas concentration. Chamber pressure,
250 Pa; Total gas flow rate, 300 sccm; Electrode temperature, 300°C; rf
power, 750 W; distance between electrodes, 50 mm; Etching time, 30 s.
F3NO showed virtually the same degree of etch rate as NF3
which is considered to be an excellent cleaning gas used at present.
However, plasma became more unstable and the etch uniformity
decreased when the gas concentration was over 40%. This phenom-
enon was similar to that of NF3 . In the case of FNO, although the
number of F in the molecule is 1, its maximum etch rate was found
to be about half that of NF3 . Furthermore, plasma discharge re-
mained stable up to the concentration of 80%, and when the etch
rates were compared in a range of uniformity within 10%, perfor-
mance of FNO was about 80% of that of NF3 (NF3 , 30%: 12509
Å/min, FNO, 80%: 10008 Å/min͒.
Analysis of exhaust gas was conducted with an FTIR in order to
examine the impact of these gases on the environment. The compo-
nents of exhaust gases were analyzed under the conditions in which
the etch rates were virtually the same at about 12000 Å/min
(NF3 /Ar: 30%, 12509 Å/min, F3NO/Ar: 30%, 12164 Å/min,
C2F6 /O2 : 30%, 12509 Å/min͒. Regarding FNO, as the etch rate did
not reach 12000 Å/min, the exhaust gas was measured at 40% ͑etch-
ing rate; 6369 Å/min͒ for the sake of convenience. The results are
shown in Fig. 3.
In the case that C2F6 is used, large quantities of CF4 and C2F6
whose GWP values are large are emitted. In the case that FNO and
F3NO are used, components of the exhaust gases are only nonde-
composed raw gases, SiF4 and HF ͑similar to the situation when
NF3 is used͒. Also, in the case that NF3 is used, the GWP value of
the nondecomposed gas, NF3 , is large and NF3 is chemically stable,
and therefore considerable energy is required to eliminate it, thus
posing a problem. FNO emitted will be abated easily with an alkali
scrubber as it is hydrolyzed even by moisture in the atmosphere.
Although the abatement of F3NO with a scrubber may be difficult, it
is thought that the atmospheric lifetime of F3NO is shorter than that
of NF3 and less energy will be required to abate F3NO than NF3
when the structure is considered.
͑30%͒ and C2F6 /O2 ͑50%͒ on the electrode were virtually equal
͑approx. 12000 Å/min͒, whereas F3NO/Ar ͑30%͒ showed a much
lower concentration of SiF4 in the exhaust. In the case of FNO, also,
the quantity of SiF4 corresponding to the etch rate ͑6369 Å/min͒ was
not detected. To investigate this phenomenon in more detail, the
trend of SiF4 concentration in exhaust gas as a function of the gas
concentration was examined. The results are shown in Fig. 5a, b,
and c.
In the etching of SiO2 by fluorine compounds, as SiO2 to be
etched will be emitted as SiF4 to outside the system, the higher the
etch rate of SiO2 is, the higher the SiF4 concentration in the exhaust
gas becomes. As clear from Fig. 5a, etch rate and SiF4 concentration
are proportional in the cases of NF3 and C2F6 .
On the other hand, in the cases of F3NO and FNO, a phenom-
enon was different from the cases of NF3 and C2F6 . Although the
etch rate increased with the gas concentration, the SiF4 concentra-
tion in the exhaust gradually decreased ͑Fig. 5b and c͒ by contraries.
Also, note that during the FTIR analysis, even after the gas supply
was stopped, the weak absorption of SiF4 was detected for some
period of time.
It is known that FNO reacts with SiF4 forming (NO)2SiF6 .7
Thus, it is assumed that the following reaction might have occurred
in the chamber
2FNO ϩ SiF → NO͒ SiF
͓2͔
͑
4
2
6
Figure 4 shows the relation between etch rate and SiF4 concen-
tration in exhaust gases. The etch rates of F3NO/Ar ͑30%͒, NF3 /Ar
Table I. Chemical properties of FNO4 and F3NO.5,6
F3NϭO
͑Nitrogen
F-NϭO
trifluoride
͑Nitrosyl fluoride͒
oxide͒
Boiling point
Melting point
Density
Ϫ59.9°C
Ϫ132.5°C
1.326 ͑liquid at bp͒,
1.719 ͑solid͒
229 K
Ϫ85.0°C
Ϫ160.0°C
¯
Critical temperature
Critical pressure
Critical density
Critical volume
302.6 K
64 bar
75 bar
490 kg/m3
593 kg/m3
146.7 cm3/mol
82.3 cm3/mol
Figure 3. Concentration of exhaust gases.

Journal of The Electrochemical Society, 150 ͑11͒ G707-G710 ͑2003͒
G709
Figure 4. Relation between etch rate and SiF4 concentration in exhaust
gases.
To verify this hypothesis, FNO and SiF4 were mixed at the ratio
of 2:1 in a cylinder, and it was found that the reaction proceeded
almost quantitatively even at room temperature. Consequently, in
the etching of SiO2 by using FNO, this reaction was considered to
occur between SiF4 generated and FNO. This reaction will be en-
couraged as the concentration of FNO became higher because the
decomposition rate of FNO decreased with increasing concentration.
(NO)2SiF6 is solid at room temperature. However, even if this sub-
stance is generated in the chamber, it seems to unlikely to be depos-
ited on the hot electrode because it sublimates at 353 K.8 Yet, it is
likely that deposition occurred at the latter line such as piping where
the temperature was much lower; this may be responsible for the
unexpected behavior of SiF4 in the present study.
On the other hand, when F3NO and SiF4 were mixed at the ratio
of 2:1, they did not react directly. However, the relation between
etch rate and SiF4 concentration was similar in both cases of FNO
and F3NO. Therefore, it is considered that FNO is generated in the
decomposition process of F3NO (F3NO → 2F ϩ FNO) and that
the same reaction occurs as the case of FNO.
Moreover, this (NO)2SiF6 is known to be a material highly sus-
ceptible to hydrolysis9 following the nest reaction
NO͒ SiF ϩ H O → SiF ϩ NO ϩ NO ϩ 2HF
͓3͔
͑
2
6
2
4
2
Hence, observation of the absorption of SiF4 lasting for some
period of time by the FTIR even after the gas supply was stopped
can be attributed to gradual hydrolysis of (NO)2SiF6 deposited on
the piping and other parts.
As the foregoing discussions show, when using FNO and F3NO
as cleaning gases, it is critical to construct applications by taking
into account the generation of (NO)2SiF6 . Conversely, the property
of generating a solid having a propensity for sublimation may be
applicable for anisotropic etching.
Figure 5. ͑a͒ The trend of SiF4 concentration in exhaust gas as a function of
the gas concentration (C2F6 and NF3). ͑b͒ The trend of SiF4 concentration in
exhaust gas as a function of the gas concentration (F3NO). ͑c͒ The trend of
SiF4 concentration in exhaust gas as a function of the gas concentration
͑FNO͒.
Conclusion
Two types of FNO compounds ͑FNO and F3NO) that have little
impact on global warming were evaluated as candidates for the new
CVD chamber cleaning gases by measuring the etch rates and ana-
lyzing the components of the exhaust gases. The etch rate of F3NO
proved was equal to that of NF3 . Its exhaust contained no substance
contributing to global warming, except for a small amount of F3NO
whose GWP was expected to be smaller than that of NF3 from the
consideration of the molecular structure. Because FNO has just one
F in its molecule, its etch rate was about half that of NF3 . As FNO
is easily hydrolyzed even by moisture in the atmosphere, FNO has
no impact on global warming. Therefore, from a standpoint of the
global warming properties, those FNO compounds are advanta-
geous.

G710
Journal of The Electrochemical Society, 150 ͑11͒ G707-G710 ͑2003͒
On the other hand, as a result of detailed analysis of the exhaust
gases of FNO and F3NO, the behavior of SiF4 in the exhaust gas led
us to think that there is a high possibility that a solid (NO) SiF
References
1. Framework Convention on Climate Change, Conference of the Parties, Third Ses-
sion, Agenda Item Five, Kyoto, Japan ͑1997͒.
͓
͔
6
2
with its propensity for sublimation ͑353 K͒ might have been depos-
ited on low temperature parts such as the piping in the latter stage of
the chamber. Therefore, when FNO or F3NO is used for chamber
cleaning, it is important to take this characteristic into consideration
in designing applications.
2. T. Beppu, Paper presented at the Seventh International Semiconductor Environ-
mental Safety and Health Conference, Dresden, Germany June 24-29, 2000.
3. S. Suzuki, E. Wani, T. Sunada, K. Shibata, T. Morimura, Y. Mitsui, Y. Kosano, T.
Takaichi, T. Beppu, and A. Sekiya, in The Global Climate Change, PV 2000-20, p.
231, The Electrochemical Society Proceedings Series, Pennington, NJ ͑2000͒.
4. O. Ruff, W. Menzel, and W. Neumann, Z. Anorg. Chem., 208, 293 ͑1932͒.
5. W. B. Fox, J. S. MacKenzie, E. R. Mccarthy, J. R. Holmes, R. F. Stahl, and R.
Juurik, Inorg. Chem., 7, 2064 ͑1968͒.
Acknowledgment
This study was conducted as part of the research projects com-
missioned by the New Energy and industrial technology Develop-
ment Organization ͑NEDO͒.
6. W. B. Fox, J. S. MacKenzie, N. Vanderkooi, B. Sukornick, C. A. Wamser, J. R.
Holmes, R. E. Eibeck, and B. B. Stewart, J. Am. Chem. Soc., 88, 2604 ͑1966͒.
7. O. Glemser and U. Biermann, Chem. Ber., 100, 1184 ͑1967͒.
8. A. Kigoshi, Thermochim. Acta, 12, 179 ͑1975͒.
The Semiconductor CVD Chamber Cleaning Project assisted in meeting
the publication costs of this article.
9. Handbook of Inorganic and Organometallic Chemistry Pt. B9: Si Silicon Supple-
ment, p. 146, Springer-Verlag, New York ͑1997͒.