Investigation of the transition region between SiO2 layers wet grown at 700 °C and Si

Article abstract of DOI:10.1088/0953-8984/9/14/009

Silicon oxide layers wet grown at 700°C on silicon substrates of integrated circuit quality have been investigated by slow-positron-implantation spectroscopy and Auger electron spectroscopy. The total thickness of pure SiO₂ on top of a SiO₂/Si transition zone is determined by ellipsometric measurements; the positron/Auger results indicate that the transition zone extends over 15-23 nm. Its thickness decreases as the SiO₂ layer is thinned by a wet-chemical process at room temperature. The results are consistent with earlier secondary-neutral mass spectroscopy measurements.

Full text of DOI:10.1088/0953-8984/9/14/009

Home Search Collections Journals About Contact us My IOPscience
Investigation of the transition region between
layers wet grown at
and Si
This content has been downloaded from IOPscience. Please scroll down to see the full text.
1997 J. Phys.: Condens. Matter 9 2947
(http://iopscience.iop.org/0953-8984/9/14/009)
View the table of contents for this issue, or go to the journal homepage for more
Download details:
IP Address: 128.122.253.228
This content was downloaded on 16/04/2015 at 11:48
Please note that terms and conditions apply.

J. Phys.: Condens. Matter 9 (1997) 2947–2954. Printed in the UK
PII: S0953-8984(97)77047-4
Investigation of th◦e transition region between SiO2 layers
wet grown at 700 C and Si
W Anwand†, G Brauer†, P G Coleman‡, A Goodyear‡, H Reuther§ and
K Maserk
† Arbeitsgruppe Positronen-Annihilations-Spektroskopie der Technischen Universita¨t Dresden,
c/o Forschungszentrum Rossendorf, Postfach 510119, D-01314 Dresden, Germany
‡ School of Physics, University of East Anglia, Norwich 4NR 7TJ, UK
§ Forschungszentrum Rossendorf, Institut fu¨r Ionenstrahlphysik und Materialforschung, Postfach
510119, D-01314 Dresden, Germany
k Technische Universita¨t Berlin, Institut fu¨r Mikroelektronik und Festko¨rperelektronik, Jebens-
strasse 1, D-10623 Berlin, Germany
Received 9 August 1996, in final form 10 February 1997
Abstract. Silicon oxide layers wet grown at 700 ◦C on silicon substrates of integrated circuit
quality have been investigated by slow-positron-implantation spectroscopy and Auger electron
spectroscopy. The total thickness of pure SiO2 on top of a SiO2/Si transition zone is determined
by ellipsometric measurements; the positron/Auger results indicate that the transition zone
extends over 15–23 nm. Its thickness decreases as the SiO2 layer is thinned by a wet-chemical
process at room temperature. The results are consistent with earlier secondary-neutral mass
spectroscopy measurements.
1. Introduction
It is desirable that modern semiconductor devices operate under conditions of minimum
power consumption. Of great importance in this context is the quality of interfaces, for
example between SiO2 and a Si substrate. This paper describes new investigations into the
◦
interface region of SiO2/Si after wet oxidation at 700 C.
Even though thermal oxidation of silicon has been one of the basic processes in
electronic device fabrication for about 30 years, there remain great uncertainties about
the exact nature of the growth mechanisms. Of concern is the transport mechanism for
dopants and oxidants as well as their interactions at various interfaces. The thickness
of the transition region has generally been assumed to be several atomic layers; in earlier
positron measurements it has been fixed at 1 nm [1–3]. Investigations using x-ray diffraction
techniques have also pointed to thin transition regions of between 1 and 2 nm [4]. In contrast,
depth profiling of oxygen and silicon (by secondary-neutral mass spectroscopy, SNMS) has
indicated thicker transition regions [5]. For discussion of other techniques used to study
transition regions the reader is referred to review papers [6, 7].
Using infrared spectroscopy (IRS), the vibrational frequency of the TD3 stretching mode
◦
was measured for SiO2 layers thermally grown on Si substrates at 700 C in steam to a
thickness of about 120 nm. Subsequent wet-chemical thinning at room temperature to about
30 nm markedly reduced the vibrational frequency from 1073 to 1065 cm−1. This shift is
due to changes in Si atom spacing and also in the oxidation conditions, which influence the
c
0953-8984/97/142947+08$19.50 ꢀ 1997 IOP Publishing Ltd
2947

2948
W Anwand et al
elasticity of the SiO2 layer and the hydrogen distribution therein. The result is believed to
indicate a densification of the SiO2 layer remaining after the wet-chemical etching process
is complete [8]. For a discussion of IRS studies of density variations of layers produced by
dry oxidation at various temperatures see reference [5].
In general, there is no clear and uniform definition of the SiO2/Si transition zone which
is independent of experimental method. The present paper describes the performance of
positron implantation spectroscopy in conjunction with oxygen and silicon depth profile
determination by Auger electron spectroscopy (AES) in order to estimate thicknesses of
this transition zone for different layers produced by wet oxidation on Si. This is regarded
as a first but essential step towards a more precise determination of the nature of the
transition, which has been discussed in structural, chemical, and electrical terms [5–7].
Figure 1. The layer structure of SiO2/Si.
2. Experimental method
2.1. Sample preparation
A boron-doped Czochralski grown Si(100) wafer of 100 mm diameter and with a specific
electrical resistivity between 15 and 25  cm (integrated circuit (IC) quality) was cleaned
◦
by the standard RCA procedure. Thermal oxidation was performed in steam at 700 C for
16 h. The oxidation process produced an oxide surface layer of about 90 nm thickness.
Subsequently, the oxide layer on the back of the wafer was chemically removed and the
wafer then divided into three parts. Using an aqueous Merck solution containing 6.5%
HF and 34.8% NH4F the oxide layers of two of them were thinned at room temperature
to 63 nm and 29 nm. Finally, the three parts were further divided in order to produce
identical specimens for positron and AES measurements. The layer structure of the samples
is illustrated in figure 1; the total thickness of the pure SiO2 layer plus the SiO2/Si transition
region was measured by ellipsometry.
2.2. Positron implantation measurements
Slow-positron-implantation spectroscopy was performed on the University of East Anglia
computer-controlled positron beam system. Positrons of predetermined energies E are
implanted at depths of up to a few µm in the sample under study [9]. The motion of
positron–electron pairs prior to annihilation causes a Doppler broadening of the photopeak
in the measured energy spectrum of the annihilation photons. This Doppler broadening is
characterized by the lineshape parameter S, which is the ratio of the number of counts in
the central region of the photopeak to the total number of counts in the peak (∼0.5). The

The transition region between SiO2 and Si
2949
(1)
measured value of S at each energy E is thus
S(E) = SsFs(E) + SeFe(E) + SoFo(E) + St Ft (E) + SSiFSi(E)
where Ss, So, St , and SSi are the S-parameters associated with annihilation with electrons
at the surface and in the oxide layer, transition zone, and bulk Si, respectively, and Se is
associated with the annihilation of epithermal positrons. The F(E) are the fractions of the
P
total number of positrons annihilated in each state; thus
Fn = 1 for each E.
Figure 2. The annihilation lineshape parameter S versus incident positron energy E for samples
with oxide layers of thicknesses 28 nm ( ), 68 nm (M), and 89 nm ( ).
◦
•
Typical results for S(E) measured for samples with different oxide layer thicknesses
are shown in figure 2.
The data were analysed using the computer code ‘VEPFIT’ [10]. A reasonable fit could
only be achieved when the transition zone thickness dt was a free parameter—i.e., not fixed,
as in references [2] and [3]. The magnitude of dt is reflected in the value of Ft for each E.
The values of So, St , and SSi were evaluated by fitting a three-layer model to S(E) for
sample 35, and fixed to these results in all further evaluations. The value of Ss was taken
from the measured S-value at E = 100 eV. The positron diffusion lengths in the oxide
layer and transition region were assumed to be 9.5 nm [3] and 0.01 nm (i.e., essentially
zero) [11], respectively. A positron diffusion length L of 100 ± 2 nm for bulk Si was
+
obtained by fitting S(E) for a sample covered with a very thin oxide layer (a few nm only).
Furthermore, values of d(ell) were used to fix the total thickness of oxide+transition layers,
in order to reduce the number of free fitting parameters. As an example, figure 3 shows the
measured S(E) for sample 35 and the fitted curve.
A similar procedure can be used to fit W(E), where W is a second lineshape parameter,
defined as the ratio of the number of counts in the wing regions of the annihilation gamma
photopeak to the total number of counts in the peak. In principle W(E) should yield the
same information as S(E). Although W(E) does not show such sharp structure at low

2950
W Anwand et al
Figure 3. The annihilation shape parameter S versus positron energy E for sample 35. The
solid line represents the fit obtained using VEPFIT. The dotted lines mark the values of E
which correspond to mean positron implantation depths equal to the boundaries of the SiO2/Si
transition region.
energies as does S(E), the percentage change in W as a function of E is about two times
larger than that for S. A typical W-parameter curve is shown in figure 4.
2.3. AES measurements
The AES was carried out using a Fisons field-emission scanning Auger microprobe
MICROLAB 310-F. The primary electrons have an energy of 10 keV and a sample current
of about 10 nA. The Si(LVV), Si(KLL), and O(KLL) Auger peaks were detected in the
direct mode by a hemispherical sector analyser. Sputter depth profiling was performed with
a 3 keV Ar+-ion gun and a current density of about 1 µA mm−2, leading to a sputtering rate
of about 0.33 nm s−1. The sputtering rate itself was estimated by mechanical measurement
of the sputtering crater after the profiling measurement using a DEKTAK profilometer.
During sputtering the sample was rotated to minimize surface roughening and to maximize
the depth resolution [12, 13].
It is well known that both Si Auger peaks are very sensitive to oxide bonding. The
shapes and energies of the peaks are different for atomic and oxidized Si [14, 15], and this
can be used to identify the Si bonding states. For this reason the Si(LVV) Auger peak at
92 eV was detected with an energy resolution of 0.2% while the Si(KLL) Auger peak at
1616.6 eV was measured with an energy resolution of 0.05%. The latter peak moves to
1608 eV while the first one changes its shape completely if the Si is oxidized. For depth
profiling purposes the O(KLL) Auger peak was used and the Si(KLL) Auger peak was
separated into two regions, the first at around 1608 eV to be ascribed to oxidized Si and
the second at around 1616 eV to be ascribed to the atomic Si.
In figure 5 the Si(KLL) Auger peak of sample 24 is displayed for every sputtering step

The transition region between SiO2 and Si
2951
Figure 4. The annihilation lineshape parameter W versus positron energy E for sample 35. The
solid line represents the fit obtained using VEPFIT. The dotted lines mark the boundary of the
SiO2/Si transition region, as in figure 3.
Figure 5. A 3D plot of the Si(KLL) Auger peak measured for every sputtering step for sample
24. The time axis points into the plane of the figure. At the beginning of the measurement the
peak associated with oxidized Si is dominant (Ekin = 1608 eV). At the end of the measurement,
at depths below the transition region, the peak is at 1616 eV, indicating atomic Si.
as a 3D plot, with the time axis pointing into the plane of the figure. The resulting depth
profile is shown in figure 6. In the transition region between SiO2 and Si several steps were
detected. The thickness of the transition range was estimated to be 23 ± 5 nm using the
procedure described in reference [16]; the same value was obtained if the Si(LVV) Auger
peak was used. This value is somewhat larger than that determined by other methods. A
possible explanation is that no deconvolution of the experimental data was performed to
allow for the physical resolution (∼2 nm here) or smearing due to the time duration of

2952
W Anwand et al
Figure 6. AES depth profiles of sample 24.
Si (Siat).
•: oxygen; ꢀ: oxidized Si (Siox); and M: atomic
the sputtering steps (10 s corresponds to 3.3 nm). Deconvolution would lead to profile
sharpening [17].
Applying the same procedure to samples 12, 13, and 35, transition layer thicknesses
of 10, 9, and 21 nm (all ±5 nm) were measured, respectively. Steps were detected in the
depth profiles of samples 25 and 35, and possibly for sample 32, but no steps were seen in
the results for samples 12 and 13, where the layer thicknesses were probably too small for
us to resolve them.
3. Results
The measurement of simple parameters such as S and W yields important information
about open-volume-type defects. The absolute values of the two parameters have little
physical relevance, but it is possible to normalize the parameters in order to allow
comparisons between different samples and different experimental arrangements to be made.
Furthermore, S- and W-values reflect the average momentum of the electrons which the
positrons annihilate. The annihilation peak is narrow (S is high, W is low) if the positrons
annihilate predominantly slow-moving electrons, as in case of valence electrons or electrons
from the bound-positron–electron state (positronium, Ps). The latter has been suggested as
an explanation for the shape of S(E) for SiO2 on Si [3]; the relatively high S-values at
low E are caused by Ps formation in the dioxide layer, and the subsequent decrease by the
inhibition of Ps formation in the SiO2/Si transition region. These effects are reflected in
distinct changes in S(E), in contrast to the relatively flat W(E) (figures 3 and 4). For this
reason the VEPFIT fitting was performed using the S-parameter data only; fits to W(E)
simply confirmed the results obtained using S(E).
It is remarkable that for all of the specimens the mean positron implantation depths
corresponding to the values of E at the local minima in S(E) correspond well to the depths
obtained by summing the thicknesses of the SiO2 layers and the transition regions.

The transition region between SiO2 and Si
2953
Table 1. Row 1: the total thickness d(ell) of the oxide + transition layers on Si, determined by
ellipsometry. Row 2: the SiO2 thickness do(pos) determined by positron spectroscopy. Row
3: the transition region thickness dt (pos) determined as d(ell) − do(pos). Row 4: the transition
zone thickness dt (AES) determined by AES. Uncertainties are shown in parentheses.
Group 1
13
Group 2
23
Group 3
32 35
Sample
12
15
22
24
d(ell) (nm)
26.97
29.70
29.42
63.15
62.86
63.71
88.41
89.06
(0.68) (0.71) (0.67) (0.32) (0.39) (0.16) (0.16) (0.23)
do(pos) (nm)
dt (pos) (nm)
dt (AES) (nm) 10(5)
12(1)
15(2)
—
—
9(5)
12(1)
17(2)
—
47(1)
16(2)
—
46(1)
17(2)
—
—
—
23(5)
68(2)
20(3)
—
66(1)
23(2)
21(5)
The results of the VEPFIT fits and AES measurements of layer thicknesses are shown
in table 1, along with the total oxide layer thicknesses (i.e., SiO2 + transition layer), deter-
mined by ellipsometry at five different positions on each sample. The results obtained
both by positron implantation and by AES indicate the existence of an extensive transition
region between the SiO2 and the bulk Si substrate. It is characterized by a decrease in the
oxygen concentration as a function of depth, so several Si suboxides exist in the region.
The thicknesses of the transition regions obtained by the two methods are much higher than
those reported in earlier papers describing positron studies of similar systems [1–3].
The thickness of the transition region depends on the thickness of the SiO2 layer
remaining after wet-chemical thinning at room temperature. However, to establish a clear
dependence further comprehensive investigations are required.
An unexpected effect was observed by remeasurement of a sample covered with an
88 nm oxide layer. The repeated measurement, performed one year later, showed that
the SiO2/Si transition zone widened from 23 nm to 42 ± 4 nm in the direction towards
the oxide layer. Additional investigations are required in order to understand this ageing
behaviour, which could have important consequences for the physical properties of the
SiO2/Si system.
4. Conclusions and summary
SiO2 layers thermally grown on boron-doped Si substrates of IC quality were investigated
by slow-positron-beam implantation and AES. The results obtained by positron and AES
measurements are in a good agreement. It could be shown that the transition region between
SiO2 and Si extends beyond a few atomic monolayers. The thickness of the zone seems to
increase with increasing SiO2 layer thickness, as already discussed in reference [5]. Further
investigations are necessary to quantify this result.
The difference between the positron diffusion lengths of the different layers reflects
the defect structure. The very short diffusion length in the transition region points to the
presence of a large number of defects which are able to trap positrons, or of an electrical
field. Only measurements done at elevated temperatures could rule out the influence of
built-in electrical fields. Such studies need to be done in the future.
Furthermore, ageing effects on the nature and extent of the transition zone may influence
the physical properties of SiO2-coated Si; these should also be investigated further.
In the first measurements presented here only broad conclusions can be drawn about
wet-grown SiO2 on Si substrates and about the distinct zones within the structure. Further

2954
W Anwand et al
investigations are needed, especially to elucidate the structure of the transition region.
Whereas positron annihilation spectroscopy is sensitive to the defect structure in the material,
AES can be used to detect modifications of the chemical bonding of the atoms as a function
of depth. The results of the positron and AES measurements agree qualitatively with those
from SNMS measurements [5]. Together, the two methods can obtain more information
about the SiO2/Si system and its transition region.
Acknowledgments
The careful sample preparation and the planning and coordination of measurements by
colleagues at the Technology Laboratory of the Institute of Microelectronics and Solid State
Electronics of the Technical University of Berlin are gratefully acknowledged.
References
[1] Katz L E 1988 VLSI Technology ed S M Sze (New York: McGraw-Hill) p 98
[2] Asoka-Kumar P, Lynn K G and Welch D O 1994 J. Appl. Phys. 76 4935
[3] Uedono A, Wei L, Tanigawa S and Ohji Y 1994 Japan. J. Appl. Phys. 33 3330
[4] Dawson J L, Krisch K, Evans-Lutterodt K W, Tang M-T, Manchanda L, Green M L, Brasen D, Higashi G S
and Boone T 1995 J. Appl. Phys. 77 4746
[5] Maser K, Krauser J, Berger H H and Braunig D 1996 Exp. Tech. Phys. 41 101
[6] Sofield C J and Stoneham A M 1995 Semicond. Sci. Technol. 10 215
[7] Hattori T 1995 Crit. Rev. Solid State Mater. Sci. 20 339
[8] Krauser J 1996 Dissertation HMI-B 535, Fachbereich Elektrotechnik der TU Berlin
[9] Anwand W, Kissener H-R and Brauer G 1995 Acta Phys. Pol. A 88 1
[10] van Veen A, Schut H, de Vries J, Hakvoort R A and Ijpma M R 1990 Positron Beams for Solids and Surfaces
ed P J Schultz, G R Massoumi and P J Simpson (New York: AIP) p 171
[11] Au H L, Asoka-Kumar P, Nielsen B and Lynn K G 1993 J. Appl. Phys. 73 2972
[12] Zalar A 1985 Thin Solid Films 124 223
[13] Hofmann S and Stepanova M G 1995 Appl. Surf. Sci. 90 227
[14] Klasson M, Berndtsson A, Hedman J, Nilsson R, Nyholm R and Nordling C 1974 J. Electron Spectrosc.
Relat. Phenom. 3 427
[15] Wagner C D, Passoja D E, Hillery H F, Kinisky T G, Six H A, Jansen W T and Taylor J A 1982 J. Vac. Sci.
Technol. 21 933
[16] 1993 Technical Document GJ/HT-80, FISONS Instruments
[17] Hofmann S and Sanz J M 1984 Springer Topics in Current Physics vol 37 (Berlin: Springer) p 141