F4.,.

HEWLETT

~E.tI PACKARD

On the Role of Ions in Electron Cyclotron

Resonance Plasma-Enhanced Chemical
Vapor Deposition of Silicon Dioxide
Karen Seaward, Solid State Technology Laboratory
John Turner, K. Nauka, Andre Nel
Integrated Circuit Business Division
HPL-94-94
October, 1994

silicon dioxide, ECR,

PECVD, electron
cyclotron resonance,
plasma

Silicon dioxide films were deposited in an electron cyclotron

resonance plasma onto substrates at temperatures ranging
from 65 to 200C. In order to determine the effect of ions on
the deposition, both ion flux and ion energy were
investigated. The ratio of ion flux.to deposition flux was
found to be a significant parameter and this "flux ratio" was
Mean ion energy was
varied between 8 and 100.
investigated at 10 eV, 50 eV and 75 eV. The silicon dioxide
films were characterized by measurement of wet etch rate,
density, composition, and stress, and by infrared
spectroscopy. It was found that above a flux ratio of about
20, high-quality Si02, was deposited whether or not the
substrate was thermally floating or at 65 to 200C,
indicating that the flux ratio was dominant over the
temperature. Use of rfbias to increase the mean ion energy
to 50 eVor above was effective in producing high-quality
Si02 when the flux ratio was below 20, but not as effective
as using a high flux ratio. Thus high ion flux and low ion
energy were found to be useful in producing Si02 at
temperatures as low as 65C with properties close to that of
thermal silicon dioxide. The role of ions in the deposition
process was found to be densification, removal of -OH (or
hydrogen) and alteration of Si-O bonding.

Internal Accession Date Only

To be Published in the Journalof Vacuum Science and Technology B, February, 1995

Copyright Hewlett-Packard Company 1994

1. INTRODUCTION
Electron cyclotron resonance (ECR) plasmas have been studied and applied to deposition of
silicon dioxide since 1978 1-6. These plasmas are of interest because higher-quality material can
be deposited at lower temperature (-100C) by ECR than by other techniques that are presently
the standard in the semiconductor industry, such as parallel-plate rf plasmas (-250C) or
thermally-activated chemical vapor deposition (-400C). There has been a long-standing interest
in the "mechanism" of the deposition and, recently, in how it differs between high-density (e.g.,
ECR) plasmas and lower-density parallel-plate rf discharges. In particular, there is interest in
the plasma energetics that appear to dominate temperature. Numerous studies have examined
plasma properties and Si02 film properties to elucidate the importance of ions and chemically
reactive neutral species for ECR deposition at low temperature':", Primarily the effect of ion
energy has been studied and it has been found that film quality approaches that of thermal Si02
(900C) with the application of rf bias to the deposition surface. The effect of ion flux may also
be important and the question remains whether or not to choose low ion flux and high ion energy
or high ion flux and low ion energy for the highest quality silicon dioxide deposited at low
temperature.

In comparing ECR and rf parallel-plate plasmas, the most striking difference is the ratio of
excited neutral species to ions. Some parameters used to make estimates are listed in Table I.
For example, in the case of a parallel-plate rf deposition plasma operating at a pressure of 0.5
Torr, we estimate that about lO% of the neutrals are in an excited state. The degree of ionization
is around lO-6 so the ratio of excited neutrals to ions is about lOO,OOO. From a similar estimate
for an ECR plasma operated at 2 mTorr, and assuming that around 50% of the neutrals are
excited and that the degree of ionization is lO,2, the ratio of excited neutrals to ions is about 50.
Since the ion component during deposition in a high density plasma is so much greater than it is
in a parallel-plate rf plasma, in this work the role of ions is investigated. To do this, either ion
energy is changed, by using rf bias, or ion flux is changed, by altering the plasma density near
the deposition surface. Ion flux is measured with a Langmuir probe and silicon dioxide
depositions are characterized by Auger electron spectroscopy, Rutherford backscattering
spectroscopy (RBS), Fourier transform infrared spectroscopy (FTIR) and measurement of stress
and wet etch rate. The best measure for comparing the depositions in which ion flux is varied
turns out to be the ratio of ion flux to deposition flux, which will be called the "flux ratio" and
denoted JlJ ox '
Estimates of J/Jox' which can also be described as the number of ions per molecule of deposited
Si02, can be derived from some recent publications. For example, from the work of Fukuda and
ll 13
coworkers employing an ECR system with a mirror magnetic field , ion flux is derived from
2.
Langmuir probe measurement of the ion current density at 7.5 mAlcm Using a deposition rate
of 1500 Almin, for which high-quality Si02 was obtained without using rf bias, JlJ ox is about 8.
From Andosca and coworkers", also employing a mirror-type ECR source, estimating a lower
1oions/cm3
limit of lxlO
and an upper limit of mo" ions/ern' from measurements in our system,

and using the published deposition rate of 70 Nmin, the flux ratio is 4 to 40. Using a multipolar
ECR source, Buckle and coworkers' deposited Si02 at a rate of 400 Nmin; estimating the ion
density" to be 1-4 x lOll/em', a flux ratio of 2 to 8 is derived. In addition, in their work no
difference was found between depositions at 80C and at 200C. In another style of multipolar
ECR source, a flux ratio of 27 is derived from the work of Plais et al"; this estimate uses their
2
values, for deposition at 800W and 1 microbar, of 1.9 mAlcm for ion current density and 120
Nmin for the silicon dioxide deposition rate. All of these estimates for J/J... are greater than 1
and were derived from data on depositions without substrate bias. The mean ion energy in these
cases is estimated'?" to be tens of eV, for example below 20 eV for pressures above 1 mTorr,
which are typical of ECR deposition.
In our work, a range of flux ratios from 5 to 100 and mean ion energies of 10 eV (the plasma
potential), 50 eV, and 75 eV are investigated. As well, four fixed temperatures are evaluated,
65, 100, 150, and 200C, plus a "floating temperature" that corresponds to plasma heating, in
order to determine whether or not the plasma energetics at the deposition surface override these
temperatures. From this work, as well as published work using similar deposition conditions
wherein similar and higher ion energies were investigated, some guidelines are established for
the flux ratios and ion energies needed to produce high-quality silicon dioxide at low
temperature.

2. EXPERIMENTAL METHODS
Silicon dioxide films were deposited in three different plasma reactors. One is a parallel-plate rf
system shown in Fig. 1. This was used to deposit rf PECVD silicon dioxide for comparison with
ECR deposited Si02. The ECR system was built by Oxford Plasma Technology and is shown in
Fig. 2 with a source that uses a mirror magnetic field and in Fig. 3 with a divergent field source
and a trim magnet downstream from the deposition surface. The mirror source configuration
was used to deposit Si02 with J/Jox varying between 6 and 35 by changing the current in the
lower magnet and by altering the deposition rate by changing the silane flow; rf bias was also
used to change ion energy. In this reactor, the deposition temperature was floating and typically
increased up to 120C. The oxygen to silane ratio was about 3 and depositions were done at 2
mT. The configuration in Fig. 3 was used to produce increased deposition rates and the ion flux
was varied by operating the trim magnet between conditions for magnetic field collimation and
cusp formation; for this configuration, JlJox varied between 8 and 100, the oxygen to silane ratio
was also about 3, and the deposition pressure ranged from 2 to 10 mT. Helium backside cooling
was employed to control the deposition temperature at 65, 100, 150, or 200C; this was
accomplished by first heating the deposition surface using a plasma containing 02 and argon then
relying on a calibrated balance between plasma heating and helium cooling measured with a
Luxtron probe. In both configurations, a Plasma and Materials Technologies fast-injection
Langmuir probe was used to measure ion saturation current near the deposition surface. From
values in mAlcm 2, ion flux was calculated by dividing by the electronic charge. Deposition flux,

Jox' was calculated from the deposition rate and included the density as measured by RBS. Used
this way, Jox is the net deposition flux.
Silicon dioxide films were analyzed by a General lonics RBS system using He" at 2.0 MeV.
Auger analysis using a Perkin-Elmer PHI 660 was used to determine atomic composition; data
were evaluated using sensitivity factors measured on thermally oxidized silicon. The presence
of -OH groups and the nature of bonding in the oxides were evaluated on doubly-polished GaAs
wafers with FTIR on a Nicolet 60SX. Thin film stress was measured on a FleXus 2-30Oi
(Tencor Instruments) and etch rate was determined in 10:1 H20:HF. Additional measurements
of temperature during deposition were done with Omegalabels (Omega Engineering, Inc.) firmly
attached to a silicon wafer; measurements made with these indicators on the back side of wafers
were the same as measurements made on the side facing the plasma and are therefore thought to
be a reasonable estimate of wafer temperature, free of artifacts due to the presence of the plasma.
Most ECR depositions were done with 02 and 10% silane in argon. The rf PECVD depositions
were done with N20 and 2% SiH4 in N2

3. DATA AND INTERPRETATION

In Fig. 4 is shown a comparison of the FfIR spectra of an ECR silicon dioxide deposition and an
rf PECVD deposition. The ECR film was grown using N20 and 10% SiH4 in N2 with the
temperature floating up to 120C during the deposition. The rf PECVD film was grown using
N20 and 2% SiH4 in N2 with the temperature fixed at 200C. The Si-O stretching absorption is
nearly the same for these films, 1071 ern" for the ECR and 1068 em" for the rf sample.
However, the value of the full-width-at-half-maximum (FWHM), which is a qualitative estimate
of the distribution of bonding arrangements", is smaller for the ECR film, 97 ern", than for the rf
film, 127 cm'. In addition, the rf film shows absorption at 3390 em'! corresponding to adsorbed
H20, and absorption at 3620 em')corresponding to Si_OH20 The ECR film also contains Si-OH
but far less than the rf film; when grown with more oxygen and a temperature floating to 200C,
Si-OH does not appear in the FfIR spectra of the ECR oxide. Water is a byproduct of the
reactions that produce silicon dioxide from N20 and SiH4 and the ECR deposition environment
more effectively removes this byproduct from the deposited oxide.
Figures 5-7 show data from silicon dioxide films grown in the mirror ECR configuration
depicted in Figure 2. The ion saturation current at the deposition surface was changed by
changing the current in the second magnet from 0 to 30, 60, 90, and 120 Amps, as shown in Fig.
5. Also indicated in the figure is the flux ratio of ions to deposited Si02 and for these
experiments this ratio varied from 7 to 35. The density of the deposited oxide, as measured by
RBS, changed from 1.8 glcm3 to 2.2 glcm3 as the flux ratio was increased, approaching the value
of 2.27 g/cm' for thermally grown silicon dioxide. Over this range of flux ratios, the
oxygen/silicon ratio remained constant, as indicated in Fig. 6. However, the thin film stress
changed from tensile (at about 40 MPa) to compressive, saturating at about 200 MPa for JlJox

greater than about 25. The corresponding wet etch rates in 10:1 H20:HF went from 250 Alsec at
a flux ratio of 7 down to 8 Alsec at a flux ratio of 35; this latter etch rate is a factor of two higher
than what is measured for thermal silicon dioxide and a factor of 5-10 lower than a diverse
collection of rf PECVD films that were evaluated.
Also shown in Figure 6 are the changes that occur by increasing the mean ion energy. For a flux
ratio of 8, the compressive stress increases from 20 MPa with no rf bias to 150 MPa with 50 eV
ions and 220 MPa with 75 eV ions. At even higher ion energies, the compressive stress
continues to increase. The wet etch rate decreases as the ion energy is increased, also shown in
Figure 6; at a flux ratio of 8, a decrease from 135 to 35 Alsec is measured when the mean ion
energy is increased from 10 to 50 eV and a further decrease to 20 Alsec occurs when the mean
ion energy is increased to 75 eV.
.
In Figure 7 are shown the changes in FfIR spectra as the flux ratio is changed. The Si-O
stretching frequency increases and the full-width-at-half-maximum of this absorption decreases
as the flux ratio increases. At the highest flux ratio of 35, the Si-O stretch frequency is 1074
em" and the FWHM is 91 ern", to be compared to the values measured for thermal silicon
dioxide which are 1077 em" and 83 em". Typical values for rf PECVD oxide are 1060-1070
ern" for the stretching frequency with the FWHM >120 em", Again, the use of higher ion
energy improves the oxide; by increasing the ion energy from 10 to 50 eV at a flux ratio of 8,
the stretching frequency is increased from 1067 em" to 1071 em" and the FWHM is reduced
from 110 em" to 101 em". However, it appears that use of a high flux ratio may produce higherquality material. For example, by using a flux ratio of 28, a stretching frequency of 1072 em"
with a FWHM of 93 em" is measured. Even though the stretching frequency is similar to that of
the deposition at a flux ratio of 8 with rf bias generating 50 eV ions, the FWHM is considerably
smaller for the material grown at the higher flux ratio without rf bias.
The -OH content of the depositions is compared by measuring the relative peak area of the
absorptions occuring between 3350 and 3750 cm'', As expected, the peak area decreases as the
flux ratio increases, as shown in Figure 7. There is a large decrease in -OH content when the
flux ratio is increased from 8 to 16. Increasing the mean ion energy from 10 to 50 eV decreases
the -OH content by 70% when the flux ratio is 8.
In Figure 8, the wet etch rate of silicon dioxide films deposited in both ECR configurations is
shown as a function of the flux ratio. The etch rate values have been normalized to that of
thermal silicon dioxide. For the mirror ECR configuration where the deposition surface was 19
inches from the ECR zone, the sample temperature started at 21C and increased to about 120C;
etch rate data are shown as triangles in Fig. 8. For the high rate configuration depicted in Figure
3 where the deposition surface was 8 inches from the ECR zone, the samples reached about
200C; etch rate data are shown as squares in Fig. 8. The wet etch rate is high at low flux ratio,
and for flux ratios above about 22, the wet etch rate becomes constant at a value of two times
that of thermal silicon dioxide.

More depositions were analyzed at J/Jox = 17 or 31 at a fixed temperature of 65, 100, 150, or
200C. The data for wet etch rate and stress are listed in Table n. As expected, an increase in
temperature results in reduced wet etch rate and increased compressive stress for both flux ratios.
However, the etch rate at J/Jox = 31 and 65C is similar to that of the depostion at J/Jos. = 17 and
150C, indicating that the higher flux ratio is effective at "replacing" temperature for reducing
the etch rate.
In Figure 9, the -OH peak area from FTIR is shown as a function of flux ratio. As for the case
of etch rate, the -OH peak area decreases as the flux ratio increases and it also becomes constant
above a flux ratio of about 20. For flux ratios of 17 and 31, data are shown for depositions done
at 65C and at 200C. The -OH peak areas are the same for these two temperatures, which is
somewhat surprising. This suggests that -OH content depends primarily on flux ratio and not on
temperature, up to 200C.

4. DISCUSSION AND CONCLUSIONS

From the relationships shown in Figures 8 and 9, we conclude that above a flux ratio of about
20, high-quality silicon dioxide is deposited. Above this flux ratio, the wet etch rate is at its
lowest at about twice that of thermal silicon dioxide. The -OH content is also at its lowest value.
By fixing the deposition surface temperature at 65, 100, 150 or 200C, it was found that very
similar silicon dioxide films were produced when the flux ratio was high (J/Jox = 31). This
indicates that the flux ratio dominates the temperature over the range investigated.
With regard to ion energy, only a few experiments were conducted. It was found that the use of
50 eV mean ion energy at a flux ratio of 8 increased the Si-O stretch frequency considerably but
that for a comparable value of stretch frequency, the FWHM was narrower by using a flux ratio
of 28 without rf bias. In addition, as ion potentials rose higher than 50 eV, the oxide films
became more and more compressive, whereas by increasing the flux ratio, the compressive stress
saturated at a reasonable value of 200 MPa.
Taking a look at published work investigating the effect of ion potential on ECR-deposited
silicon dioxide, similar conclusions are drawn about the effect of rf bias. For example, use of 50
eV ion energy in a distributed ECR system reduced the wet etch rate of Si02 films by a factor of
10 over the value for films grown with 10 eV ions; however, etch rate and fixed oxide charge
remained the same for films grown with ions from 50 to 150 eV 12 For deposition conditions
where we have estimated JlJox=2-8 in another distributed ECR system, use of rf bias from -20 to
-100 Vde produced films with the same value of Si-O stretching frequency", In a mirror ECR
deposition system for which we have estimated J/Jox=4-40, the FTIR spectra of silicon dioxide
films deposited with different values of rf bias were examined in detail", An optimum rf-

induced de bias of minus 50 V was found, based on the Si-O stretching absorption and its
minimum FWHM. In another mirror ECR deposition system, the conclusion was drawn that use
of rf-induced dc bias above -50 V produced Si02 films comparable to those deposited by
LPCVD; the comparison was based on wet etch rate, refractive index, and FflR analysis".
Going back to the depositions described earlier, for which lower-quality Si02 was deposited with
a flux ratio of 8 and ion energy of 50 eV and higher-quality Si02 was deposited with a flux ratio
of 28 and ion energy of 10 eV, the suggestion is that raising the ion flux is more beneficial than
raising the ion energy. By looking at these depositions in terms of average energy deposited per
atom," where Ed =E1 x J/JO&, we find that the lower-quality oxide is deposited with Ed = 350 eV
and the higher-quality oxide is deposited with Ed = 280 eV. In other words, there is a difference
depending on the energetic pathway that is chosen for deposition. This conclusion was reached
earlier for a completely different material system, a study of reactive magnetron sputtering of
polycrystalline TiAlN. 22
In conclusion, from our work investigating the importance of the ratio of ion flux to deposition
flux in ECR deposition, a strong trend was found. For values above about 20 for this flux ratio,
high-quality silicon dioxide films were deposited from 65 to 200C. Addition of rf bias was also
studied. From our work and from published literature there is some consensus that low energy
ions, on the order of 50 eV, are sufficient for producing the highest-quality silicon dioxide,
based on thin film characterization by measurement of wet etch rate, stress, and composition,
and FTIR analysis. Returning to the question proposed in the introduction, that of choosing
between a high flux of low energy ions or a low flux of high energy ions, the trend shown in our
work and other literature cited is that a high flux of low energy ions is preferable for producing
high-quality silicon dioxide at low temperature in high density plasmas. As for the implicit
question of the role of ions in ECR deposition, the data indicate that ions densify, change the
bonding of, and drive off -OH (or hydrogen) from depositing silicon dioxide. Lastly, although it
does not add a lot to our understanding of the mechanism of silicon dioxide deposition in a high
density plasma, the flux ratio itself is seen as an effective parameter for process development.

ACKNOWLEDGMENTS
We wish to thank, from Hewlett-Packard, Francoise Mertz for extensive assistance with
experiments, Bill Collins for hardware support, and Paul Soo for Auger Analysis. Also, we are
grateful to Vince Mazzi at Stanford University for setting up the Langmuir probe, and to
Professor J. Leon Shohet at the University of Wisconsin for motivating discussions.

5. REFERENCES
Portions of this work were reported as paper PS-WeA3 at the 40th National Symposium of the
American Vacuum Society on November 17,1993 in Orlando, Florida.
1 Takashi Tsuchimoto, 1. Vac. Sci. Technol.12 1730 (1978)
2 Seitaro Matsuo and Mikiho Kiuchi, Jpn. 1. Appl. Phys. 22 L210 (1983)
3 T. Fukuda, K. Suzuki, S. Takahashi, Y Mochizuki, M. Ohue, N. Momma and T. Sonobe, Jpn.
1. Appl. Phys. 27 L1962 (1988); Takuya Fukuda, Michio Ohue, Naohiro Momma, Kazuo Suzuki
and Tadashi Sonobe, Jp. 1. Appl. Phys. 28 1035 (1989)
4 T.V. Herak, T.T. Chau, D.J. Thomson, S.R. Jejia, D.A Buchanan, and K.C. Kao, 1. Appl.
Phys. 65 2457 (1989)
5 O.A Popov and H. Waldron, 1. Vac. Sci. Technol. A 1914 (1989)
6 Son Van Nguyen and Kevin Albaugh, 1. Electrochem. Soc. 1362835 (1989)
7 M. Kitagawa, T. Hirao, T. Ohmura, T. Izumi, Jpn. 1. Appl. Phys. 28 LI048 (1989)
8 C. Chiang, D.B. Fraser, D.R. Denison, 7th Int. IEEE VLSI Multilevel Interconnection Conf.
(IEEE, New York) 12-13 June, 1990, p. 381
9 K.A Buckle, K. Pastor, C. Constantine, and D. Johnson, 1. Vac. Sci. Technol. B lQ 1133
(1992)
10 R.G. Andosca, W.J. Varhue, and E. Adams, 1. Appl. Phys. 72 1126 (1992)
II T. Fukuda, M. Ohue, T. Kikuchi, K. Suzuki, T. Sonobe, and N. Momma, IEDM 1989 (IEEE),
page 665; T. Fukuda, K. Saito, M. Ohue, K. Shima, and N. Momma, IEDM 1992 (IEEE), page
285
12 O. Joubert, R. Burke, L. Vallier, C. Martinet, and R.AB. Devine, Appl. Phys. Lett. 62228
(1993)
13 S. Kanai, K. Nojiri, M. Nawata, Hitachi Review 40383 (1991)
14 Ion density estimates from data published by Wavemat Inc., Plymouth, Michigan, for MPDR
325i source.
15 F. Plais, B. Agius, F. Abel, 1. Siejka, M. Puech, G. Ravel, P. Alnot, and N. Proust, 1.
Electrochem. Soc. 139 1489 (1992)
16 Morito Matsuoka and Ken'ichi Ono, 1. Vac. Sci. Technol. A Q 25 (1988) and 1. Appl. Phys. 64
5179 (1988)
17 W.M. Holber and J. Forster, J. Vac. Sci. Techno!. A ~ 3720 (1990)
18 Makarem A Hussein, G.A Emmert, N. Hershkowitz, R. Claude Woods, J. Appl. Phys. 72
1720 (1992)
19 F.L. Galeener, Phys. Rev. B 1..2 4292 (1979)
20 AC. Adams in Plasma Deposited Thin Films, ed. 1. Mort and F. Jansen (CRC Press, Boca
Raton, FL, 1968) page 150
21 E.S. Bulat, G. Ditmer, C. Herrick, and S. Hankin, 1. Vac. Sci. Technol. A l.Q 1402 (1992)
22 1. Petrov, F. Adibi, J.E. Greene, L. Hultman, 1.-E. Sundgren, Appl. Phys. Lett. 63 36 (1993)
a)

Table I. Comparison of rf and ECR plasmas.

PARAMETER

rfplasma

ECRplasma

pressure (mTorr)

500

n (ern")

2 x 10

7 X 1013

n* (reactive species)

< lO%n

>50%n

n, (ions)

10-6 n

10-2 n

n*/n

100,000

50

16

Table II. Comparison of

tn; = 17 and J/1ox = 31 at 65, 100, ISO, and 200C.

1.ILr. = 17

~~=31

Temperature

Etch rate

Stress

Etch rate

Stress

65C

45 Nsec

50 MPa tens

17 Nsec

160 MPacomp

100C

34 A/sec

tens/comp

10 A/sec

120 MPacomp

150C

20 Nsec

90 MPacomp

10 Nsec

150 MPacomp

200C

9Nsec

200 MPacomp

6Nsec

210 MPacomp

notes: Tens is tensile; comp is compressive; tens/comp is at the transition and too low to
measure.

N20 and SiH4

upper electrode
with gas distribution
assembly

rf generator
and impedence
matching unit
(13.56 MHz)

lower electrode
assembly with
resistive heating
element

substrate
temperature
control
thermocouple

Figure 1. Schematic of rf PECVD system, from A Kiermasz, AA Chambers, A

McQuarrie and M. Stephens, Semicon East, September 1986, Boston, MA.

10

2.45 GHz,
rectangular waveguide

cylindrical coupler
'-2

...............
.. ,::: .

tn

Q.

en

CD

000
000

ii)o.--..--...-.....-

f\)

0
0

::s
n
CD

0"01

@
19 in

_.

B field (Gauss)

CD

:e~

3_.
n

~~

..

o~

=:01

0= ==

Q)

<

CDf\)

:eo
_.
:::J

- sample

Q.

I
13.56 MHz

:e

Figure 2. Schematic of ECRPECVD system with a mirror source configuration. The

axial magnetic field corresponding to 170 Amps in Ml and varying the current in M2
from 0 to 60 to 120 Amps is included.

11

2.45 GHz,
rectanguJar waveguide

cy Jlndrical coupler

8 field (Gauss)

"-02

.~~:

Langmuir
probe

~
:.:.- 1
.

-.
(f)

m
:::J

0)

c.o

000
000

(')

CD
CTO'I
CD

Bin

0= ==

"

Co

_.

helium .........- - - - wate r in:: ~-=:!!!!::::!!!!!!:&==I

water out

...&.

sample

(')

~...&.

0)0'1

--

10% SiH4/Ar

18Itrim

magn~t

1181

<

CD
~
_.1\)
:::JO

co
~

Figure 3. Schematic of ECR PECVD system with a divergent magnetic field source for
deposition at high rate. The axial magnetic field corresponding to divergence,
collimation and cusp configurations is included.

12

I\)

0
0

a)
b)

3600

2800

2000

1200

400

wavenumber (cml)

Figure 4. Fourier Transform Infrared Spectra of a) an ECR oxide grown with N20 and
10% SiH. in N2 with temperature floating up to 120C and b) an rf PECVD oxide grown
with N20 and 2% SiH. in N2 with temperature fixed at 200C.

13

5-r---------------4

a)

Isat

mA/cm 2

2
1

o
b)
2.2
2.0

g/cm 3

1.8
1.6

30

60

90

120

second magnet current (Amps)

Figure 5. Measurement of a) ion saturation current and b) ECR oxide density as the
current is increased in the second magnet of the ECR configuration shown in Figure 2.
Ion saturation current is measured with a Langmuir probe and oxide density is measured
by RBS. Also shown (in parentheses) is the flux ratio, J/J as a function of the current
in the second magnet.

14

a)

O/Si

.---.----.--.

100

b)

tensile

stress 0
MPa
compressive

-100
-200

c)
etch 200
rate
A/sec
100

20
30
flux/Si02 flux

40

Figure 6. Measurement of a) oxygen-to-silicon ratio, b) oxidestress, and c) wet etch rate

as a function of the flux ratio. Oxygen-to-silicon ratio was determined by both Auger
and RBS. Stress was measured with a FleXus 2-30Oi and wet etch rate corresponds to a
10:1 H20:HF solution. Also shown are the effects of increasing the mean ion energy to
50 and 75 eV by using rf bias during deposition.

15

1075

-r----------------.
a)+50ev

10 1
)

Si-O 1070
stretch
freq.
cm

-1

(116)

1065

(FWHM)

(118)
1060

b)
4
reI.

Si-OH
peak 2
area
1

20
30
flux/Si02 flux

40

Figure 7. Changes. as the flux ratiois increased. in a) the Si-Ostretching frequency and
its full-width-at-half-maximum (shown in parentheses). and b) relative Si-OH peak area.
Included are changes that occur by increasing the mean ion energy to 50 eV by using rf
bias during deposition.

16

1OO~---------------,
Q)
-+-'

80

..c
o

-+-'
Q)

60

-+-'
Q)

40

E
L
0

20

20

40

60

80

100

120

ion flux/Si02 flux

Figure 8. Decrease in wet etch rate in 10:1 H20:HF as the flux ratio is increased. Wet
etch rate values are normalized to that of thermal silicon dioxide. Data for the mirror
ECR configuration are shown as triangles; the temperature floated to 120C for these
depositions. Data for the high rate ECR configuration are shown as squares; the
temperature floated to 200 C for these depositions.

17

4-r--------------------..

e
o
~

~2

:r:

o
I

6S0C
200C

-+-r.,...,..~I'""'I"""Ir-r-r..,...,...........~I'""'I"""Ir_r_r _ _~..__.......r_r_r................._w_~

10

20

30

40

Ion flux/Si02 flux

Figure 9. Decrease in the relative -DB peak area, measured on FTIR spectra, as the flux
ratio is increased. Where temperature is indicated, the temperature was fixed. All other
data are for a floating temperature of 120C.

18