High-resolution RBS Study of Ultra-low Energy Ion
Implantation for Microelectronic Application
K. Kimura, Y. Oota, K. Nakajima, M. Suzuki
Department of Engineering Physics and Mechanics, Kyoto University Kyoto 606-8501, Japan
Abstract. High-resolution Rutherford backscattering spectroscopy (HRBS) is a powerful technique to measure
elemental depth profiles with depth resolution of atomic level. Application of HRBS to microelectronics, in particular
the issues of ultra-low energy ion implantation, is discussed in this paper. Boron profiles in silicon wafers implanted
with 0.5-keV B+ ions are measured. The observed profile agrees with TRIM simulation very well. Molecular effect in
ion implantation is investigated for 6-keV As2+. It is shown that the radiation damage created by 6-keV As2+ ion
implantation is almost twice larger than that by 3-keV As+ ion implantation. It is also found that the projected range of
6-keV As2+ is several % larger than that of 3-keV As+.
spectrometer and a small accelerator can be used. We
can develop a compact HRBS system including an
accelerator [7]. (3) Scattering cross sections for subMeV He ions are almost one order of magnitude larger
than those for MeV He ions. The large cross section
compensates a relatively small acceptance angle of the
spectrometer.
This allows a reasonably short
measurement time comparable to the conventional
INTRODUCTION
With downscaling of electronic devices, there are
increasing demands for quantitative analysis of thin
films with depth resolution of atomic level. In our
previous work [1], it was demonstrated that the
monolayer resolution can be achieved in Rutherford
backscattering spectroscopy (RBS) using a 90˚ sector
magnetic spectrometer. The present paper shows some
examples of application of the high-resolution RBS
(HRBS) to microelectronics industry, in particular to
the issues of ultra-low energy ion implantation,
including (1) boron depth profiling in ultra-low energy
B+ ion implanted silicon wafers. (2) molecular effects
in ultra-low energy ion implantation.
HRBS SYSTEM
There are several magnetic or electrostatic
spectrometers designed for high-resolution ion beam
analysis [2 - 6]. Although almost all spectrometers
were designed for MeV He ions, we decided to use
sub-MeV He ions as a probe. There are several
advantages for the use of sub-MeV He ions. (1) SubMeV He ions provide better depth resolution than
MeV He ions because the stopping power has a
maximum in the sub-MeV region. (2) A small
FIGURE 1. Schematic drawing of HRBS system.
CP680, Application of Accelerators in Research and Industry: 17th Int'l. Conference, edited by J. L. Duggan and I. L. Morgan
© 2003 American Institute of Physics 0-7354-0149-7/03/$20.00
373
together with a wide energy window (25%) allows a
reasonably short measurement time (~ 10 min). In the
following sections, some examples of HRBS
measurements are shown.
RBS. There are, of course, some disadvantages. In
the sub-MeV region, the cross section deviates from
the simple Rutherford formula due to the screening
effect especially for heavy atoms. However, this is not
a serious problem because the correction of the
screening effect is rather easy [8].
Another
disadvantage is concerning the charge state
distribution. The fraction of He0 in the scattered beam
cannot be negligible in the sub-MeV region (e.g. ~
10 % at 0.4 MeV). For precise measurements, energy
spectrum of all charge states including neutral atoms
must be measured. This cannot be done with a
magnetic or electrostatic spectrometer. Fortunately,
He+ fraction is predominant (~ 65%) and almost
constant over an energy region of 200 – 400 keV [9].
This allows quantitative analysis by measuring only
the He+ spectrum.
BORON DEPTH PROFILING
Ion implantation at ultra-low energies, as low as
0.5 keV, is a key process for the formation of ultra
shallow junctions for device technologies with gate
length below 100 nm. The depth profiles of implanted
B were generally measured by secondary ion mass
spectroscopy (SIMS). The boron profile measured by
SIMS always has a narrow surface concentration peak
[10]. This peak is usually considered as an artifact
because SIMS cannot be accurate in the topmost subnm region. However, whether the peak is an artifact or
not is still in debate. Here, we demonstrate that the
peak is really an artifact by measuring the boron depth
profile with HRBS.
Figure 1 shows a schematic drawing of our HRBS
system. The magnetic spectrometer is basically a 90º
sector type with 26.6º inclined boundaries for double
focusing. The bending radius is 200 mm and the
maximum bending power (ME/q2) is 2.8 MeV. The
exit boundary is slightly modified from a straight line
so that the exit focal plane is perpendicular to the
central ion path. A microchannel-plate positionsensitive-detector (MCP-PSD) of 100 mm length is
placed on the exit focal plane. The energy window of
the spectrometer is 25% of the central energy. The
measured energy resolution is 0.11% including the
energy spread of the incident ion beam, which is good
enough for layer-by-layer analysis. Although the
acceptance angle (~ 0.3 msr) is about one order of
magnitude smaller than that of a typical silicon surface
barrier detector (SSBD), the larger cross sections
Figure 2 shows HRBS spectra for 0.5 keV B+
implanted Si (B+ ion dose Φ = 2 × 1015 cm-2) observed
under random (triangles) and [111] channeling (closed
circles) conditions. A channeling spectrum for the
sample implanted at Φ = 1 × 1013 cm-2 is also shown
(open circles). The arrows show the expected energies
for He ions elastically scattered by Si, O, C and B
surface atoms. The channeling spectrum for Φ = 2 ×
1015 cm-2 has a broad peak around 376 keV, while
there is no structure in the spectrum for Φ = 1 × 1013
cm-2, indicating that the observed broad peak
corresponds to the implanted B atoms. The depth
80000
COUNTS/keV
40000
B
C
CONCENTRATION (at.%)
0.5 keV B implanted Si(001)
0.5-keV B implanted Si(001)
<111> channeling & random
15
-2
Φ = 2× 10 cm
60000
13
-2
Φ = 1× 10 cm
Si
O
×5
20000
0
360
× 10
380
400
420
ENERGY (keV)
440
FIGURE 2 HRBS spectra for 0.5 keV B+ ionimplanted Si. Random and [111] channeling spectra
are shown.
15
cm
-2
100
Si
TRIM
O
B (× 10)
50
0
460
Φ = 2 × 10
C
0
2
4
DEPTH (nm)
6
8
10
FIGURE 3 Depth profiles of compositional elements
in 0.5 keV B+ ion-implanted Si (dose 2 × 1015 cm-2)
derived from the HRBS spectrum.
374
affected by the molecular effect. Because the damage
distribution affects the subsequent diffusion of dopant
atoms, the molecular effect on the damage production
is of particular interest in microelectronics industry.
Here, we measure the damage distribution of Si(001)
samples implanted with 3-keV As+ and 6-keV As2+
ions to see the molecular effect.
profiles of Si, O, C and B in the sample of Φ = 2 ×
1015 cm-2 were derived from the HRBS spectrum and
are shown in Fig. 3. The B-profile has a peak at ~ 3.5
nm and the peak concentration is ~ 7 at.%. There is no
narrow surface concentration peak, indicating that the
narrow surface peak observed in SIMS measurement is
an artifact. It should be noted that using an advanced
technique, a SIMS profile which has no narrow surface
peak was recently measured in agreement with the
present HRBS result [11].
P-type wafers were implanted with 3-keV As+ or 6keV As2+ ions at equivalent arsenic atomic doses of
0.5, 1, 2, or 4 × 1014 cm-2 at room temperature. The
beam current was carefully adjusted so that the flux
(atoms cm-2 s-1) is the same for As+ and As2+. Figure 4
shows HRBS spectra observed under [111] channeling
conditions. Solid curves show the spectra for As+
implanted samples and dashed curves show that for
As2+. The silicon yield increases with increasing As
dose due to the damage production. The silicon yield
for As2+ implantation increases more rapidly than As+,
showing the molecular effect. The spectrum of As+
implanted sample at an atomic dose of 1 × 1014 cm-2
coincides with that of As2+ implanted sample at 0.5
×1014 cm-2. The same relation holds for other doses,
indicating that As2+ creates more damage,
approximately twice larger than As+. This molecular
effect can be ascribed to the cascade overlap during
molecular ion implantation [16]
The B profile calculated by the TRIM code is also
shown by a solid curve for comparison.
The
agreement between the observed and calculated
profiles is reasonably good, showing that the TRIM
simulation is reliable even in this ultra-low energy
regime.
MOLECULAR EFFECT IN ULTRALOW ENERGY ION IMPLANTATION
In ultra low energy ion implantation, the space
charge effect makes the beam handling very difficult.
To avoid this difficulty, molecular ions accelerated to
higher energies but with an equivalent low energy per
atom are used. The behavior of the constituent atomic
ion is, however, different from that of the monoatomic ion of the same velocity due to the existence of
the spatially correlated partner ions. It is well known
that energy loss [12], secondary electron yield [13],
sputtering yield [14] and damage production [15] are
There is a small arsenic peak around 374 keV in
the HRBS spectrum. The depth profile of arsenic
atoms can be derived from the HRBS spectrum.
Figure 5 shows the obtained depth profiles. For
comparison, a depth profile calculated by the TRIM
+
Si(001)-[111]
θ s = 50 °
-3
400keV He
20
20000
+
As CONCENTRATION (× 10 cm )
+
COUNTS/keV
increasing dose
14
-2
0.5, 1, 2, 4 × 10 cm
10000
O
Si
As
0
320
330
340
350
360
ENERGY (keV)
370
10
380
FIGURE 4 HRBS spectra for 3-keV As+ (solid
curves) and 6-keV As2+ (dashed curves) ion-implanted
Si observed under [111] channeling conditions.
8
3 keV As / 6 keV As2
Si(001)
+
As
+
As2
TRIM
increasing dose
14
-2
1, 2, 4 × 10 cm
6
4
2
0
0
5
DEPTH (nm)
10
15
FIGURE 5 As depth profiles for 3-keV As+ (solid
curves) and 6-keV As2+ (dashed curves) ionimplantation. Result of TRIM simulation (dot-dashed
curve) is also shown for comparison.
375
code is also shown for 3-keV As+ implantation at 4 ×
1014 cm-2. The agreement between the observed
profile and the TRIM result is again reasonably good.
REFERENCES
1. K. Kimura, K. Ohshima and M. Mannami, Appl. Phys.
Lett. 64 (1994) 2232.
Looking at the profiles in detail, the profiles for
As2+ implantation seem slightly deeper than those for
As+.
The peak positions of the profiles were
determined by fitting with Gaussian functions. The
obtained peak positions of the profiles for As+
implantation are 5.34 ± 0.06 nm, 5.36 ± 0.03 nm and
5.48 ± 0.02 nm for Φ = 1 × 1014 cm-2, 2 × 1014 cm-2
and 4 × 1014 cm-2, respectively and those for As2+
implantation are 5.78 ± 0.08 nm, 5.77 ± 0.04 nm and
5.77 ± 0.02 nm for Φ = 1 × 1014 cm-2, 2 × 1014 cm-2
and 4 × 1014 cm-2, respectively. The peak position for
As2+ is larger than that of As+ by several %
irrespective of the As dose. A similar effect was found
in molecular dynamics simulation [16].
The
simulation showed that the projected range of 200
eV/atom (Ag)n cluster in Al is larger than that of the
monomer by 11, 67, 115 and 300% for n = 10, 50, 100
and 500, respectively. This effect is called "clear-theway effect". A leading ion clears away target atoms
for trailing ions. This results in larger projected range
of the trailing ions. The present result is the first
experimental evidence of the clear-the-way effect.
2. R. G. Smeenk, R. M. Tromp, H. H. Kersten, A. J. H.
Boerboom, F. W. Saris, Nucl. Instr. And Meth. 195
(1982) 581.
3. T. Enders, M. Rilli and H. D. Carstanjen, Nucl. Instr.
And Meth. B 64 (1992) 817.
4. W. M. Arnoldbik, W. Wolfswinkel, D. K. Inia, V. G. G.
Verleun, S. Lobner, J. A. Reinders, F. Labohm and D. O.
Boerma, Nucl. Instr. And Meth. B 118 (1996) 567.
5. W. A. Lanford, B. Anderberg, H. Enge and B.
Hjorvarsson, Nucl. Instr. And Meth. B 136-138 (1998)
1177.
6. R. Grötzschel, Ch. Klein and O. Kruse, Nucl. Instr. And
Meth. B 183 (2001) 3.
7. K. Kimura, M. Kimura, Y. Mori, M. Maehara and H.
Fukuyama, in Application of Accelerators in Research
and Industry, edited by J. L. Duggan and I. L. Morgan,
AIP Conference Proceedings 475, New York: American
Institute of Physics, 1999, pp. 500-503.
8. M. H. Mendenhall and R. A. Weller, Nucl. Instr. And
Meth. B 58 (1991) 11.
9. J. C. Armstrong, J. V. Mullendore, W. R. Harris and J. B.
Marion, Proc. Phys. Soc. 86 (1965) 1283.
CONCLUSIONS
10. C. W. Magee, D. Jacobson and H.-J. Gossmann, J. Vac.
Sci. Technol. B 18 (2000) 489.
The depth profiles of implanted atoms and damage
distributions in silicon wafers implanted with ultra-low
energy ions (0.5-keV B+, 3-keV As+, 6-keV As2+) were
measured using high-resolution RBS. There is no
narrow surface peak in the boron profile, which is
generally observed in SIMS measurement. It is found
that 6-keV As2+ creates damage twice larger than 3keV As+. It is also found that there is a molecular
effect on the projected range, i.e. the range for 6-keV
As2+ is about 7% larger than that of 3-keV As+
11. T.H Büyüklimanli and C.W. Magee, Appl. Surf. Sci. in
press.
12. W. Brandt, A. Ratkowski and R. H. Ritchie, Phys. Rev.
Lett. 33 (1974) 1325.
13.D. Hasselkamp, in Particle Induced Electron Emission II,
edited by G. Hohler and E. A. Niekisch, Springer Tracts
in Modern Physics Vol. 123 (Springer, Heidelberg, 1991),
p. 1.
14. H.H. Andersen and H.L. Bay, J. Appl. Phys. 45 (1974)
953.
ACKNOWLEDGMENTS
15. J. A. Davies, in Ion Implantation and Beam Processing,
edited by J. S. Williams and J. M. Poate (Academic Press,
Sydney, 1984).
We are grateful to the members of Quantum
Science and Engineering Center of Kyoto University
for use of the 4 MV Van de Graaff accelerator. We
are also grateful to Dr. A. Agarwal for providing the
samples. This work was supported in part by Grant-inAid for Scientific Research from the Ministry of
Education, Culture, Sports, Science and Technology.
16. A. I. Titov, S.O. Kucheyev, Nucl. Instr. and Methods B
149 (1999) 129.
17. Y. Yamamura and T. Muramoto, Phys. Rev. Lett. 31
(1992) 1463.
376