See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/224454371

Reaction analysis of initial oxidation of silicon by UV-light-excited ozone and

the application to rapid and uniform SiO2 film growth

Article  in  Journal of Applied Physics · March 2007

DOI: 10.1063/1.2433750 · Source: IEEE Xplore

CITATIONS READS

6 158

4 authors, including:

Tetsuya Nishiguchi
Meidensha
33 PUBLICATIONS   220 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Development of a new market of High purity Ozone gas View project

All content following this page was uploaded by Tetsuya Nishiguchi on 09 August 2016.

The user has requested enhancement of the downloaded file.

 JOURNAL OF APPLIED PHYSICS 101, 034909 共2007兲

Reaction analysis of initial oxidation of silicon by UV-light-excited ozone

and the application to rapid and uniform SiO2 film growth
Aki Tosaka,a兲 Hidehiko Nonaka, and Shingo Ichimura
National Institute of Advanced Industrial Science and Technology (AIST),
Research Institute of Instrumentation Frontier, Tsukuba Central-2, 1-1-1 Umezono, Tsukuba,
Ibaraki 305-8568, Japan
Tetsuya Nishiguchi
Meidensha Corporation, 515 Higashimakado Kaminakamizo, Numazu, Shizuoka 410-8588, Japan
共Received 2 July 2006; accepted 30 November 2006; published online 9 February 2007兲
UV-light-excited O3 prepared by irradiation of nearly 100% pure O3 with a KrF excimer laser 共␭
= 248 nm, irradiated area= 30⫻ 10 mm2兲 was utilized for low-temperature Si oxidation. The initial
oxidation rate was determined, and the activation energy was shown to be almost zero 共0.049 eV兲.
To clarify the optimum oxidation conditions, the dependence of the SiO2 film growth rate on the
total photon number and the photon density was investigated. The evolution of O3 density after
UV-light irradiation was experimentally measured, and the O共 1D兲 density change is discussed.
O共 1D兲 density changes are successfully explained by using a second-order reaction model,
indicating that a pulse supply of oxygen atoms is essential in the initial oxidation process. The
uniform oxidation of 8 in. Si wafer has been carried out using a wafer-transfer type chamber by
irradiating the wafer with KrF excimer laser light expanded linearly to the wafer width by a concave
lens. © 2007 American Institute of Physics. 关DOI: 10.1063/1.2433750兴

I. INTRODUCTION O2 + h␯共␭ = 172 nm兲 → O共 3 P兲 + O共 1D兲,

The low-temperature Si-oxidation method has been ex-

tensively studied for use in the fabrication of poly-Si thin-
film transistors formed on glass or plastic substrates. The
plasma enhanced chemical vapor deposition 共PECVD兲
method is the most common way to stack oxide films; how-
ever, the quality factor of films prepared by this method is
not as good as with the thermal oxide method. The use of
high annealing temperatures to improve chemical vapor
deposition 共CVD兲 film and interface quality is limited by the
deformation temperature of the substrates 共e.g., T ⬍ 200 ° C
and ⬍400 ° C for plastic and glass substrates, respectively兲.
As one of the solutions to utilize CVD film, it has been
reported that high-quality SiO2 thin-film fabrication before
stacking the CVD film improves film quality.1,2 To fabricate
a high-quality “interface film,” a several nanometer film to
control the interface, it is necessary to develop a low-
temperature Si-oxidation method.
In high-temperature 共艌800 ° C兲 oxidation of Si in an O2
atmosphere, molecular oxygen diffuses through the oxide
film and reacts with silicon at the SiO2 / Si interface via dis-
sociative adsorption to silicon at the interface.3,4 On the other
hand, in low-temperature oxidation, some other method for
the generation of atomic oxygen is necessary because O2
does not dissociate.
UV radiation has been used to dissociate O2 molecules at
a lower temperature.5–9 The following are some of the disso-
ciation processes:
a兲
Present address: Institute of Multidisciplinary Research for Advanced Ma-
terials, Tohoku University, 2-1-1 Katahira, Aoba-ku, Sendai, Miyagi 980-
8577, Japan; electronic mail: akiគtosaka@mac.com
O2 + h␯共␭ = 126 nm兲 → O共 3 P兲 + O共 1S兲.
O共 1D兲 and O共 1S兲 are excited states of atomic oxygen, which
are expected to strongly oxidize silicon. In fact, Fang et al.8
showed that the growth rate of SiO2 was very high
共8 nm/ min兲 in an O2 / Ar atmosphere irradiated with light
共␭ = 126 nm兲. However, quality factors of such films, such as
limited leakage current density through the oxide film, are
not as high as those of films prepared by the thermal oxide
method. The dissociation of O2 by 172 or 126 nm light is
less effective because 共1兲 the photodissociation rate is low,
共2兲 light at these wavelengths cannot penetrate through air,
and 共3兲 irradiation at these wavelengths induces defects in
SiO2 film.
O3 oxidation is also expected to be a potentially useful
process for silicon oxidation at a low temperature because O3
is easily decomposed into atomic oxygen at the silicon wafer
at temperatures as low as 400 ° C. Nishiguchi et al.10 re-
ported that the SiO2 films fabricated using nearly pure O3.
The electrical properties such as the interface trap density
and the leakage current are 5 ⫻ 1010 cm−2 / eV and
10−9 A / cm2 at 5 MV/ cm, respectively, which shows that O3
oxidation method is acceptable for use in the most recent
gate oxides. However, growing SiO2 films thicker than 2 nm
below 400 ° C is difficult because of the low oxygen-atom
density due to the low decomposition rate of O3 on low-
temperature SiO2 surfaces.10
The Si oxidation method using UV-light-excited O3 that
we previously proposed satisfies the need for both high
growth rate and good film quality.11 The experimental results
for 12⫻ 12 mm2 silicon tips show that 共1兲 the growth rate of

0021-8979/2007/101共3兲/034909/7/$23.00 101, 034909-1 © 2007 American Institute of Physics

Downloaded 13 Feb 2007 to 150.29.219.145. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp
034909-2 Tosaka et al.
FIG. 1. Schematic view of a wafer transfer type chamber.
SiO2 film is 5.2 nm/ 10 min at 300 ° C and 3.6 nm/ 10 min at
70 ° C, and that 共2兲 the film quality is as good as that of the
thermal oxide using O2. From an industrial standpoint, how-
ever, the method must be applicable for the oxidation of
large wafers.
In this paper, we report the initial oxidation of Si using
UV-light-excited O3 and the application to larger Si wafer
oxidation. First, we describe the method used to prepare
nearly 100% O3 and UV-light-excited O3. Next, we refer the
initial oxidation using UV-light-excited O3. To analyze the
time-resolved distribution of atomic oxygen, we made time-
resolved O3 density measurements, and the supplied O共 1D兲
density will be discussed. Finally, we describe a wafer-
transfer-type oxidation chamber and report on the results ob-
tained with the chamber.
II. PRODUCTION OF UV-LIGHT-EXCITED OZONE
UV-light-excited O3 was prepared from almost 100%
pure O3 irradiated with light from a KrF excimer laser 共␭
= 248 nm兲. KrF excimer laser light was used because 共1兲
irradiation by 248 nm light is expected to generate excited
atomic oxygen, O共 1D兲,9 which should be a powerful oxidiz-
ing agent; 共2兲 a high density of O共 1D兲 will be formed be-
cause of the high photon density of KrF excimer laser light
共in our case, ⬃1016 cm−3兲; and 共3兲 248 nm light will not
induce defects in SiO2 film, because the absorption coeffi-
cient of ideal SiO2 for 248 nm light is practically zero.12
The nearly pure, highly concentrated 共HC兲 O3 was sup-
plied from a pure-ozone generator 共Meidensha Corp.,
MPOG-SM1C1兲.13 HC O3 gas was produced by the follow-
ing procedure: O3 gas of 5%–7% purity prepared with a con-
ventional ozonizer was introduced into an ozone vessel
cooled to 90 K. Since O3 exists as a liquid and O2 as a gas at
90 K, pure liquid O3 was obtained by pumping out the O2
gas. Then, nearly pure O3 gas was produced in the chamber
by increasing the vessel temperature 共typically to
110– 125 K兲.
Silicon was oxidized by UV-light-excited O3 in a mecha-
nochemically aluminum chamber as shown schematically in
Fig. 1. The base pressure of the chamber was 10−3 Pa or
lower, and the O3 pressure was kept between 200 and 800 Pa
during the oxidation. The flow rate of the O3 gas was main-
tained between 100 and 170 SCCM 共SCCM denotes cubic
centimeter per minute at STP兲. The KrF excimer laser
共Lambda Physik, Compex 110, irradiation area 10
⫻ 30 mm2, 180– 260 mJ/pulse, 1 – 100 Hz兲 was positioned to
Downloaded 13 Feb 2007 to 150.29.219.145. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp
J. Appl. Phys. 101, 034909 共2007兲
FIG. 2. The growth rate of initial oxidation of SiO2 film. The oxidation
temperature is RT 共filled circles兲, 200 ° C 共open squares兲, and 260 ° C 共filled
triangles兲.
irradiate the Si wafer from above. The laser light was a pulse
of approximately 10 ns duration. Note that the net irradiation
time was not equal to the oxidation time. The irradiation area
could be expanded by using a cylindrical concave lens 共f =
−45.2 mm兲. The HF terminated silicon wafer was heated
with a halogen lamp through an opaque glass substrate
holder from the bottom side of the wafer and the typical
temperatures are in the range of room temperature to
260 ° C. The linear motion system used to transfer the Si
wafer is described in Sec. VI.
The oxidation temperature was measured with a K-type
thermocouple spot welded to the surface of the Si wafer. The
irradiation by the KrF excimer laser heated the outermost
surface of the Si wafer. On average, the temperature increase
caused by the laser light irradiation, as measured by thermal
tape attached to the wafer, was less than 50 ° C. In this paper,
therefore, the temperature rise of the Si surface due to laser
irradiation is not fully taken into account.
III. INITIAL OXIDATION OF SILICON BY UV-LIGHT-
EXCITED O3
To determine whether the UV-light-excited O3 process
was effective enough for the formation of the “interface
film” mentioned in the introduction on a large-size Si wafer,
the initial growth rate of the process was investigated. The
growth rate is shown in Fig. 2. The oxidation was carried out
in a temperature range between room temperature and
300 ° C.
The growth rate can be represented by a linear kinetics
equation, d = d0 + At, where d is the SiO2 thickness, A the
linear rate constant, t the oxidation time, and d0 the thickness
that has grown barrierlessly in the initial period of the UV-
ozone oxidation. The calculated value of d using A
= 0.076 nm/ min 共at RT兲, 0.136 nm/ min 共at 200 ° C兲,
0.159 nm/ min 共at 260 ° C兲, and d0 = 1.75 nm are shown also
in Fig. 2 and fit the experimental results well. The value of
d0 = 1.75 nm is thicker than the nascent oxide film, which is
typically 1 nm for O3. The thick initial oxide film means that
the oxidation occurs within an unusually short period of
time. The experiment showed that a 1.7± 0.1-nm-thick SiO2
film grew in only 20 s. From these results, we estimate that a
1.7-nm-thick SiO2 film can be fabricated within 6.7 min by
oxidation of an 8 in. wafer using the oxidation system de-
scribed in Sec. VI.
034909-3 Tosaka et al.
FIG. 3. Arrhenius plots of linear rate constants A.
An activation energy 共Ea兲 of 0.049 eV was obtained
from the Arrhenius plot shown in Fig. 3. Though only a few
reports on initial oxidation of Si are available, this Ea is
much lower than that determined for other methods, such as
the Kr and O2 mixture plasma process 共0.14 eV兲14 or the HC
O3 process 共0.32 eV兲.10 The low Ea indicates that UV-light-
excited O3 is applicable to low-temperature oxidation.
IV. DEPENDENCE OF SIO2 FILM GROWTH ON
PHOTON DENSITY AND TOTAL PHOTON NUMBER
The area irradiated by the KrF excimer laser light 共10
⫻ 30 mm2兲 must be expanded with a lens to oxidize a large
Si wafer. Therefore, clarifying the dependence of growth
thickness of SiO2 film on the number of irradiated photons or
the photon density is important. When the photon density is
constant the film thickness apparently depends on the total
photon number of irradiated photon, but as O3 is excited
impulsively by the laser light, a nonlinear effect might occur
when the photon density is changed. The relation between
SiO2 film thickness and the total photon number is shown in
Fig. 4. The number of total photons is the product of the
photon density and the irradiation time and the repetition
rate. As seen in the figure, the thicknesses of SiO2 films
grown at various photon densities show a scattering behav-
ior, and no particular correlation is apparent.
The dependence of SiO2 film thickness on photon den-
sity is shown in Fig. 5. The repetition rate was varied so as to
keep the total photon number constant for different photon
densities. The photon density was adjusted with the lens
It was shown that there are two oxidation growth modes
in the case of Si oxidation using UV-light-excited O3 and the
Downloaded 13 Feb 2007 to 150.29.219.145. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp
J. Appl. Phys. 101, 034909 共2007兲
FIG. 4. The relation between SiO2 film thickness and number of irradiated
photons.
the transition time is 5 min.11 Therefore, the relation between
the SiO2 film thickness and photon density was made in the
two oxidation modes 共⬍5 min and ⬎5 min兲. For oxidation
times longer than 5 min 关Fig. 5共b兲兴, there seems to be a
following relation between the SiO2 thickness and photon
density. When the constant total photon numbers are 1.3
⫻ 1021 and 8 ⫻ 1020 cm−2, the lower photon density results in
thicker SiO2 films. When the constant total photon number is
4.5⫻ 1020 cm−2, the SiO2 film thickness seems to have a
maximum value around photon density= 5 ⫻ 1015 cm−2. On
the other hand, for oxidation times shorter than 5 min 关Fig.
5共a兲兴, there is no relationship between SiO2 film thickness
and photon density.
The thickness of grown SiO2 film is mainly determined
by the supply and diffusion of atomic oxygen. For oxidation
times above 15 min, SiO2 film growth must be limited by
oxygen diffusion. The rate constant of the reaction of O共 1D兲
with O2 is high 共2.4⫻ 10−10 cm3兲,9 and the lifetime of O共 1D兲
will be very short 共e.g., the lifetime can be calculated to be
86 ns at 200 Pa by the reaction rate constant兲. Therefore, the
short lifetime of O共 1D兲 may cause a pulsed supply of O共 1D兲
to the Si wafer. The fact that higher photon density 共with low
repetition rate兲 results in thinner SiO2 films may indicate that
there is an excess of oxygen atoms that cannot be used for
oxidation within the lifetime of O共 1D兲 because of a limita-
tion of diffusion. To discuss the effect of the supply of O共 1D兲
on SiO2 film growth, understanding the behavior of O共 1D兲 is
necessary. In the next section, the time-resolved distribution
of the O共 1D兲 density will be discussed.
FIG. 5. The relation between SiO2
film thickness and photon density. The
total numbers of irradiated photon are
共a兲 4.8⫻ 1020 cm−2 共solid squares兲,
6.7⫻ 1018 cm−2 共open triangles兲, 共b兲
1.3⫻ 1021 cm−2 共open circles兲, 4.5
⫻ 1020 cm−2 共solid triangles兲, and 8
⫻ 1020 cm−2 共open squares兲. The oxi-
dation times are 共a兲 5 min 共solid
squares兲, 3 min 共open triangles兲, 共b兲
17.2 min 共open circles兲, 15 min 共solid
triangles兲, and 15 min 共open squares兲.
034909-4 Tosaka et al.
FIG. 6. The experimental set up for the time-resolved distribution measure-
ment of O3 density.
V. ESTIMATION OF THE CHANGE OF O„ 1D… DENSITY
BY EXPERIMENTAL MEASUREMENT OF THE
TIME-RESOLVED DISTRIBUTION OF THE O3 DENSITY
For application to rapid and large-size wafer oxidation,
determining the time-resolved and spatially resolved distri-
bution of the O共 1D兲 density is helpful. However, because the
deexcitation occurs nonradiationally, the direct detection of
O共 1D兲 is not easy. Therefore, we instead measured the time-
resolved distribution of the O3 density and estimated the
time-resolved distribution of the O共 1D兲 density.
A. Experimental setup
The time-resolved distribution of the O3 density was
measured by using two KrF excimer lasers. One served as an
“excitation light,” and the other as a “probe light.” The
power densities of the KrF laser were 70 and 2 mJ/ cm2, and
the irradiation areas were 10⫻ 30 and 10⫻ 10 mm2, respec-
tively. The pulse duration of both laser lights was about
10 ns.
The experimental setup is shown schematically in Fig. 6.
The features of the chamber and the preparation method of
UV-light-excited O3 are described in Sec. II. A synthetic
quartz window that was made for the O3 density measure-
ment was placed on the top of the chamber. Boxes made of
synthetic quartz were welded to the window to enclose the
process region. The dielectric multilayer mirrors that reflect
the probe light were set so as to pass through the process
region. The probe light passed through the O3 atmosphere for
a length of 43 mm. Excitation light was irradiated into this
region vertically from the top, as in the case of silicon oxi-
dation.
The intensity of the probe light was measured using a
photodiode 共Hamamatsu Photonics S1226-18BU, rise time
0.15 ␮s兲. The intensity of the probe light through a vacuum
is defined as I0. The probe light intensity traveling through
O3 atmosphere is weakened to I1 because O3 molecules ab-
sorb light. After the excitation-light irradiation, the intensity
of the probe light I2 becomes greater than I1 because the
excitation light, whose photon density is much greater than
that of the probe light, decomposes a large portion of the O3.
A time-resolved measurement of O3 density after excitation-
light irradiation was made by detecting the intensity of the
probe light while varying the delay time of the probe light
with respect to the excitation light.
The ratio of I2 / I0 is converted to O3 density N by
Downloaded 13 Feb 2007 to 150.29.219.145. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp
J. Appl. Phys. 101, 034909 共2007兲
FIG. 7. 共a兲 The experimental results of O3 density at 300 Pa 共filled circles兲
and 200 Pa 共filled triangles兲 with simple analysis 共shown in Sec. V B兲 and at
300 Pa 共open circles兲 and 200 Pa 共open triangles兲 with detailed analysis
共shown in Sec. V D兲. The solid curves shows the calculated O3 density at
300 and 200 Pa obtained by the simple calculation described in Sec. V E 1.
The dotted and broken curves show the calculated results by considering
more detailed reaction process. 共b兲 The experimental results and calculated
results at the first stage.
I2 = I0 exp共− N␴0x兲, 共1兲
where x is the path length and ␴0 is the absorption cross
section 共1 ⫻ 10−17 cm2 for ␭ = 248 nm兲. In the case of laser
light irradiation which has a high photon density, ␴0 must be
corrected by taking into account the saturation absorption
effect. The corrected ␴ is expressed as
␴0
␴= , 共2兲
I
1+
Isat
where Isat is the correction term given by15,16
Isat = h␯Nl. 共3兲
The corrected ␴ is 1%–4% smaller than ␴0 in a pressure
range between 150 and 350 Pa. We confirmed that the calcu-
lated values of I1 / I0 using the corrected ␴ when the power
density of the probe light was 2 mJ were in fair agreement
with the experimentally measured I1 / I0 共without the excita-
tion light兲 at pressures lower than 330 Pa.
B. Decomposition of O3 molecules after UV-light
irradiation
The experimental results of the time-resolved distribu-
tion measurements of O3 density are shown in Fig. 7. The O3
pressure before the irradiation of the excitation light was
either 200 Pa 共filled circles兲 or 300 Pa 共filled triangles兲. The
O3 density was calculated from Eqs. 共1兲–共3兲 with a path
length of 43 mm. The results show that there were three time
034909-5 Tosaka et al. J. Appl. Phys. 101, 034909 共2007兲

regions of O3 density change 关Fig. 7共a兲兴: 共A兲 The density

started to decrease immediately after the excitation-light ir-
radiation to about 10−5 s, 共B兲 O3 density greatly decreased
between approximately 10−5 and 10−4 s, and 共C兲 O3 density
started to increase after approximately 10−4 s. The third re-
gion 共C兲 was the recovery process due to the supply of fresh
O3 gas into the process region.

C. Decomposition of O3 molecules after UV-light

irradiation
In this subsection, the decomposition of O3 and the O3
density calculation method are presented. The decomposition
of O3 by irradiation of the excitation light is expressed as
follows:
O3 + h␯共␭ = 248 nm兲 → O共 1D兲 + O2共 1⌺+g 兲. 共4兲
FIG. 8. The schematic view of the probe light and the excitation light
The number of absorbed photon density can be calculated by
configuration.
using Eqs. 共1兲–共3兲 and the excitation KrF laser power
共60 mJ/ cm2 at a single shot兲. The estimated value for the
absorbed photon density was 1 ⫻ 1016 cm−3 at 200 Pa and 6.5 mm for xab, 30 mm for xbc, and 6.5 mm for xcd共=xab兲.
2 ⫻ 1016 cm−3 at 300 Pa. The intensity of the probe light at d共Id兲 is expressed as fol-
O共 1D兲 and O2共 1⌺+g 兲 are further consumed by reacting lows:
with residual O3 molecules as follows:9
Id = Ia exp共− 2N␴6.5xab兲exp共− N⬘␴30xbc兲. 共13兲
O共 1D兲 + O3 → O2 + 2O共 2 P兲, 共5兲

→O共 3 P兲 + 2O3 , 共6兲 The O3 density at excitation light area 共N⬘兲 is given by

→2O2共 1⌬g兲, 共7兲 N⬘ = −

1
␴bcxbc
ln冋 Id
Ia exp共− 2N␴abxab兲
. 册 共14兲

→O2共 1⌺+g 兲 + O2共 3⌺−g 兲, 共8兲

The results for O3 density calculated using Eq. 共14兲 are
→2O2共 3⌺−g 兲, 共9兲 shown in Fig. 7 as open circles 共200 Pa兲 and open triangles
共300 Pa兲. Using this corrected density 共Id兲 which satisfies the
O2共 1⌺+g 兲 + O3 → O + 2O2 , 共10兲 condition, Id ⬍ Ia exp共−2N␴abxab兲 can be analyzed.

→2O2共 1⌬g兲 + O3 , 共11兲

→2O2共 3⌺−g 兲 + O3 . 共12兲
The total rate constant of the reactions of O共 1D兲 with O3 共kO兲
and the total rate constant of the reactions of O2共 1⌺+g 兲 with
O3 共kO2兲 are 2.4⫻ 10−10 and 2.5⫻ 10−11 cm3 / molecule s,
respectively.17,18 Because definite branching ratios of the re-
actions are not available, we calculated the O3 density using
the total rate constant.
D. Precise analysis of the O3 density
The O3 density shown in Sec. V B was obtained from
the ratio of the probe light 共I2 / I0兲, which was experimentally
measured, and a calculation using Eqs. 共1兲–共3兲 with a path
length of 43 mm. Although the actual path length of the
probe light was 43 mm, the area of the excitation light was
30 mm, as shown schematically in Fig. 8. In this subsection,
we introduce a correction to the O3 density in order to ana-
lyze the reaction involving O共 1D兲 more accurately. The
probe light passed through a “no excitation light area” before
and after the “excitation light area.” The path lengths were
E. The O3 density calculation
1. O„ 1D… and O2„ 1⌺+g… reactions with O3
The time-resolved distributions of O共 1D兲 and O2共 1⌺+g 兲
densities can be calculated by means of the second-order
reaction model using total rate constants, as described in Sec.
V C. The reaction rate of O共 1D兲 with O3 was much larger
than that of O2共 1⌺+g 兲 with O3. Therefore, for the first time
region 关as in Sec. V B 共A兲兴, we must calculate only the de-
crease in O3 density due to reaction with O共 1D兲. The change
of O共 1D兲 density 关O共 1D兲兴 can be expressed as
d
− 关O共 1D兲兴 = kO关O共 1D兲兴关O3兴, 共15兲
dt
=kO共mO − xO兲共mO3 − xO兲, 共16兲
where mO is the initial density 共t = 0兲 of O共 1D兲, mO3 is that of
O3, and xO is the decrease of O共 1D兲. Since xO = 0 at t = 0, kO
can be calculated by

Downloaded 13 Feb 2007 to 150.29.219.145. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp
034909-6 Tosaka et al. J. Appl. Phys. 101, 034909 共2007兲

2. Correction by O2 molecules generated by the

reaction between O„ 1D… and O3
The simple calculation results described above did not fit
the experimental results well. The calculation of O3 density
should involve more-complex reactions. In this subsection,
we take into account O2 molecules generated by the reaction
of O共 1D兲 with O3. Since the rate constant of these generated
O2 molecules with O3 is not available, we assumed that it
was equal to kO2. At time t, the generated O2 density 关which
is not primary generated O2共 1⌺+g 兲兴 is 2xO. Therefore the
FIG. 9. The calculated results of O共 1D兲 density in the cases of O3 pressure
change of O2 density is calculated by
are 300 and 200 Pa. d
− 关O2兴 = kO2关O2兴关O3兴 共26兲
dt

k Ot = 冕0
xO
1
共mO − xO兲共mO3 − xO兲
dxO , 共17兲 =kO2共mO2 − xO2 + 2xO兲共mO3 − xO − xO2兲. 共27兲

The decrease of O2 density is calculated by

1 mO共mO3 − xO兲 共mO2 + 2xO兲共mO3 − xO兲兵1 − exp关− 共mO2 − mO3 + 3xO兲kO2t兴其
=− ln , 共18兲 x O2 = . 共28兲
mO − mO3 mO3共mO − xO兲 共mO2 + 2xO兲 − 共mO3 − xO兲exp关− 共mO2 − mO3 + 3xO兲kO2t兴

and the decrease of O共 1D兲 共xO兲 is expressed as The results for O3 density calculated by Eqs. 共19兲 and 共28兲
are shown in Fig. 7 as dotted curves. The results fit the cor-
mOmO3 − mOmO3 exp关− 共mO − mO3兲kOt兴 rected experimental results well 共open symbols兲.
xO = . 共19兲
mO − mO3 exp关− 共mO − mO3兲kOt兴
F. The time distribution of O„ 1D… density
Next, for the second time region 关as in Sec. V B 共B兲兴 the
reaction between O2共 1⌺+g 兲 and O3 is calculated. The change The agreement between the calculated values 共dotted
of density of O2 is also expressed as curves兲 and the experimental results 共open symbols兲 shown
in Fig. 7 indicates that our calculation, which involved the
d reactions of O共 1D兲, reflected well the actual O共 1D兲 density.
− 关O2兴 = kO2关O2兴关O3兴. 共20兲
dt The calculated O共 1D兲 density is shown in Fig. 9. The density
of O共 1D兲 decreased to 1% of the initial value at t = 5
In this region, O3 density should be calculated by taking into ⫻ 10−7 s, indicating that the supply of O共 1D兲 to the silicon
account xO; wafer surface was intermittent with a short duration of ap-
关O3兴 = mO3 − xO − xO2 , 共21兲 proximately 10−6 s.
As shown in the previous section, SiO2 thickness did not
where xO2 is the decrease of O2共 1⌺+g 兲. The change of 关O2兴 is depend on the photon density at a constant total photon num-
ber in the initial oxidation region 共⬍5 min兲, whereas the
d thickness did depend on the photon density after the initial
− 关O2兴 = kO2共mO2 − xO2兲共mO3 − xO − xO2兲. 共22兲
dt oxidation region 共⬎5 min兲. In the initial oxidation region 共
⬍5 min兲, the SiO2 film thickness was small enough for the
In this case, kO2t can also be written as
oxygen atoms to diffuse to the interface without a consider-

k O2t = 冕 0
xO
2 1
共mO2 − xO2兲共mO3 − xO − xO2兲
dxO2 . 共23兲
able barrier. Therefore, the SiO2 thickness did not depend on
the photon density at a constant photon number. On the other
hand, after the initial oxidation region 共⬎5 min兲, the oxida-
In this reaction region, we assumed that xO was not depen- tion rate was limited by the diffusion of atomic oxygen
dent on xO2. Therefore, xO2 can be calculated by through the thicker SiO2 film. Although a detailed discussion
of oxygen diffusion is difficult, we can state that the mecha-
mO2共mO3 − xO兲兵1 − exp关− 共mO2 − mO3 + xO兲kO2t兴其 nism of oxygen diffusion for the case of an oxidation method
x O2t = . using a pulsed supply of atomic oxygen 共O3 and pulsed KrF
mO2 − 共mO3 − xO兲exp关− 共mO2 − mO3 + xO兲kO2t兴
excimer laser light兲 may be different from that using a con-
共24兲 tinuous supply of atomic oxygen 共e.g., O2 and Ar lamp8兲.
The O3 density at t is calculated by
VI. OXIDATION OF AN 8 IN. WAFER
关O3兴 = mO3 − xO − xO2 . 共25兲
A wafer-transfer-type oxidation chamber 共Fig. 1兲 was de-
The calculated time-distribution of O3 density is shown in veloped for large-wafer oxidation. Our chamber includes the
Fig. 7 as a solid curve. The calculated results did not fit the following features: 共1兲 the irradiation area of KrF laser light
corrected experimental results 共open triangles and circles兲. is expanded in one direction by a cylindrical concave lens

Downloaded 13 Feb 2007 to 150.29.219.145. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp
034909-7 Tosaka et al.
FIG. 10. SiO2 film thickness profile. Oxidation temperature is 200 ° C and
the transfer speed is 9.5 mm/ min. The outline of an 8 in. wafer was also
drawn.
共f = −45.2 mm兲, and 共2兲 the silicon wafer is transferred in one
direction. The transfer speed can be varied between 3 and
30 mm/ min. Si wafers up to 8 in. can be oxidized in this
chamber. The O3 supply system and other features are de-
scribed in Sec. II.
The SiO2 thickness distribution on an 8 in. Si wafer us-
ing this system is shown in Fig. 10. The oxidation tempera-
ture was 200 ° C, and the scan speed was 9.5 mm/ min. The
average thickness of the SiO2 film was 2.1 nm, and the de-
viation of the SiO2 film thickness was 0.1 nm, which indi-
cates that the fabrication of a uniform SiO2 film is possible
with this system.
The relation between the film thickness and the wafer
transfer speed is shown in Fig. 11. Since the initial oxidation
rate by the UV-light-excited O3 was high, as mentioned in
Sec. III, the thickness of SiO2 films grown at a transfer speed
of 12 mm/ min was 1.75 nm. We have not yet measured the
electrical characteristics for a whole wafer sample since our
measurement system is only capable for small chip samples
and the film thickness is anyhow too small for the measure-
ment using the simple metal-insulator-semiconductor struc-
ture 共MIS兲 structure. However, since the quality factors of
the film fabricated by UV-excited O3 were as good as those
of a film fabricated by a thermal oxide process,11 the
1.7-nm-thick SiO2 film is expected to be an “interface film”
that improves the interface character between a CVD stacked
film and poly-Si. The application to poly-Si oxidation by
FIG. 11. The relation between SiO2 film thickness and transfer speed.
Downloaded 13 Feb 2007 to 150.29.219.145. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp
View publication stats
J. Appl. Phys. 101, 034909 共2007兲
UV-light-excited O3 is a future issue. In order to apply the
method to processes in most advanced industries such as the
gate oxide formation, further optimization of the process
conditions such as the control of the laser beam profile and
the ozone flow rate must be done.
VII. SUMMARY
A low-temperature oxidation method using UV-light-
excited O3 was developed. A 1.7-nm-thick SiO2 film could
be grown in only 20 s at room temperature. The activation
energy of the oxidation using the method was almost zero
共0.049 eV兲. The oxidation rate depended on the photon den-
sity when the oxidation time was longer than 5 min.
To estimate the evolution of O共 1D兲 density after an ex-
citation laser pulse, the time-resolved distribution of O3 den-
sity was experimentally observed. The O3 density change
was calculated using a second-order reaction model, and the
calculated results reproduced the experimental results well.
We concluded that O共 1D兲 was supplied to the Si wafer sur-
face as a pulse. We developed a wafer-transfer-type oxidation
chamber, and by using the chamber, we successfully grew
SiO2 films on 8 in. Si wafers by the UV-light-excited O3
oxidation method.
ACKNOWLEDGMENT
This work was supported by the New Energy and Indus-
trial Technology Development Organization 共NEDO兲.
1
Y. Nakata, T. Okamoto, T. Hamada, T. Itoga, and Y. Ishii, Extended Ab-
stracts of IWGI2001, November 1–2 共International Workshop on Gate
Insulator, Tokyo, 2001兲, p. 120.
2
Y. Ebiko and Y. Mashima, Extended Abstracts of IDW02, December 4–6
共The Ninth International Display Workshop, Hiroshima, 2002兲, p. 271.
3
U. Hofer, P. Morgen, W. Wurth, and E. Umbach, Phys. Rev. Lett. 55, 2979
共1985兲.
4
P. Gupta, C. H. Mak, P. A. Coon, and S. M. George, Phys. Rev. B 40, 7739
共1989兲.
5
I. W. Boyd, V. Craciun, and A. Kazor, Jpn. J. Appl. Phys., Part 1 32, 6141
共1993兲.
6
J. Y. Zhang and I. W. Boyd, Appl. Surf. Sci. 186, 64 共2002兲.
7
A. Fukano and H. Oyanagi, J. Appl. Phys. 94, 3345 共2003兲.
8
Q. Fang, J. Y. Zhang, and I. W. Boyd, Appl. Surf. Sci. 208–209, 369
共2003兲.
9
D. L. Baulch, R. A. Cox, P. J. Crutzen, J. R. F. Hampson, J. A. Kerr, J.
Troe, and R. T. Watson, J. Phys. Chem. Ref. Data 11, 327 共1982兲.
10
T. Nishiguchi, H. Nonaka, S. Ichimura, Y. Morikawa, M. Kekura, and M.
Miyamoto, Appl. Phys. Lett. 81, 2190 共2002兲.
11
A. Tosaka, T. Nishiguchi, H. Nonaka, and S. Ichimura, Jpn. J. Appl. Phys.,
Part 2 44, L1144 共2005兲.
12
E. D. Palik, Handbook of Optical Constants of Solids 共Academic, New
York, 1985兲.
13
S. Ichimura, H. Nonaka, Y. Morikawa, T. Noyori, T. Nishiguchi, and M.
Kekura, J. Vac. Sci. Technol. A 22, 1410 共2004兲.
14
T. Ueno, A. Morioka, S. Chikamura, and Y. Iwasaki, Jpn. J. Appl. Phys.,
Part 2 39, L327 共2000兲.
15
R. W. Keyes, IBM J. Res. Dev. 7, 334 共1963兲.
16
R. J. Harrach, T. D. M. Vicar, G. I. Kachen, and L. L. Steinmetz, Opt.
Commun. 5, 175 共1972兲.
17
S. T. Amimoto, A. P. Force, and J. R. Wisemfeld, Chem. Phys. Lett. 60,
327 共1978兲.
18
R. Gilpin, H. I. Schiff, and K. H. Welge, J. Chem. Phys. 55, 1087 共1971兲.