You are on page 1of 10

SURFACE AND INTERFACE ANALYSIS

Surf. Interface Anal. 26, 851È860 (1998)

Characterization of Shallow Implants with SIMS


using Electron-beam-assisted Oxygen
Bombardment with Oxygen BackÐll¤

M. Puga-Lambers1,* and P. H. Holloway1,2


1 MICROFABRITECH, University of Florida, Gainesville, FL 32611, USA
2 Department of Materials Science and Engineering, University of Florida, Gainesville, FL 32611, USA

Secondary ion mass spectrometry (SIMS) depth proÐles of shallow implants of boron and arsenic in silicon have
been measured. It was demonstrated that improved depth resolution was achieved at low (1.5 keV) beam energy for
both O ‘ and Cs‘ beams. Oxygen backÐll from the base pressure (10—10 Torr) to 10—6 Torr also improved the
2
depth resolution. Simultaneous electron bombardment during oxygen backÐll further improved the depth resolution,
as measured both by the surface transients in the Si‘ and SiO‘ substrates as well as the B‘ signal. The mechanism
of improvement by electron bombardment during oxygen backÐll was discussed and concluded to be electron-beam-
stimulated oxidation of the Si surface. ( 1998 John Wiley & Sons, Ltd.
KEYWORDS : SIMS ; shallow implants ; oxygen backÐll ; electron beam irradiation ; electron-beam-stimulated oxidation

dopant depth distribution. Cascade mixing e†ects can


INTRODUCTION
be minimized by using heavier primary ions, by lower-
ing the primary ion energy and by increasing the inci-
Characterization of shallow features in integrated cir- dence angle relative to the sample normal up to 60¡.4,5
cuits is currently one of the major challenges to the These conditions signiÐcantly improve depth resolution.
SIMS community. For several decades, SIMS has been In addition to depth resolution limitations, transient
one of the most indispensable techniques for quantiÐca- changes in ion yields and sputter rates occur in the
tion of low-level impurities and dopants in semicon- near-surface region during the initial period of sputter-
ductor materials due to its high sensitivity and excellent ing. This transient condition is associated with pertur-
depth resolution. However, the incessant shrinkage of bation of the sample by the primary ion beam, and with
semiconductor device features to well below 1 lm has matrix e†ects caused by the presence of a native surface
placed new demands on the development of SIMS hard- oxide on the sample. Both changes cause an error in
ware and analytical protocols in order to ensure sub- quantiÐcation of shallow depth proÐles.6 Particularly
nanometer depth resolutions with proÐle accuracy in for shallow implants, the pre-equilibrium region may
the near-surface region.1,2 contain a signiÐcant percentage of the implanted
Shallow doping is formed mainly by low-energy ion dopant species, thus any analysis that excludes this
implantation. Implant energies of \5 keV and, most region will lead to an incorrect dose calibration. Conse-
recently, in the sub-keV range are used for boron. quently, the use of sufficiently low SIMS primary ion
Arsenic is being implanted at energies of O10 keV. This beam energies ( \ 3 keV) is also critical towards estab-
means that most of the implanted dose is located within lishing a rapid equilibrium condition between the
20È30 nm of the sample surface. For example, from primary ion and the samples surface. Recent studies
TRIM calculations, a 11B` implant into Si (100) at 5 have suggested that a primary ion beam energy of less
keV has a projected range of 23.3 nm, and a 10 keV than half the implant energy should be applied for
75As` implant into Si (100) yields a projected range of shallow boron proÐling with a primary oxygen beam.7
12.1 nm. The depth resolution of SIMS shallow This low energy poses difficulties for some instruments
implants may therefore be limited by the penetration in obtaining a well-focused primary ion beam and a
depth of the primary ion beam. For example, the depth current density beam high enough to ensure good detec-
resolution limit has been reported to be D2.5 times the tion sensitivity and detection limits.
projected range of the SIMS primary ions in the Some analytical strategies have been devised to over-
sample.3 Cascade mixing and knock-on e†ects caused come these problems, particularly for boron shallow
by the primary ion impact tend to induce proÐle proÐle measurements in silicon. A combination of low
broadening, which lead to distortion of the measured primary ion energy at oblique impact angles (60¡) and
oxygen backÐll during depth proÐling of boron in
silicon with an oxygen primary beam has been reported
to improve depth resolution and minimize the surface
* Correspondence to : M. Puga-Lambers, MICROFABRITECH,
University of Florida, Gainesville, FL 32611, USA. ion yield transient.2,8 This is due to the continuous for-
¤ Dedicated to Professor Siegfried Hofmann on the occasion of his mation of a thin surface oxide layer during sputtering,
60th birthday. which stabilizes the ionization yields and allows for a
CCC 0142È2421/98/110851È10 $17.50 Received 14 May 1998
( 1998 John Wiley & Sons, Ltd. Revised 24 July 1998
Accepted 27 July 1998
852 M. PUGA-LAMBERS AND P. H. HOLLOWAY

constant matrix to be reached within a few Ó from the 11B` implant into Si. A silicon sample uniformly doped
surface.2,9 It was demonstrated that this technique pro- with 1 ] 1019 cm~3 11B` was used for calibration of
vided uniform sensitivity across the SiO /Si interface the 0.5 keV 11B` implant with an oxygen backÐll to
2
and, consequently, should yield reliable quantiÐcation 10~6 Torr. The arsenic and boron depth proÐles in Si
of the implanted dopant.2 However, the accuracy of the were obtained with a quadrupole Perkin-Elmer 6600
oxygen backÐll technique has been challenged, based on PHI SIMS system. A cesium primary ion beam was
the fact that it does not provide an instantaneous full used for the analysis of arsenic in silicon, whereas
oxidation and therefore is not sufficient to eliminate oxygen bombardment was used for the boron analysis.
transient e†ects altogether.6 Indeed, it has been report- For comparison, boron proÐles were also acquired with
ed that before full oxidation is achieved, a fraction of the Cs` beam. The angle of incidence of either of the
the outermost surface material is sputtered away at dif- primary ion beams with respect to the sample normal
ferent sputter rates due to oxygen adsorption induced was 60¡. Both cesium and oxygen primary ion beam
by exposing the surface to gas.10 The variation in energies were varied from 5 keV down to 1 keV.
sputter rate across this transient region results in signiÐ- Primary ion beam currents were set within the range
cant distortions of the proÐle shape, and subsequently 200È20 nA and the raster size was varied between 350
causes errors in the depth scale calibration.11,12 For and 650 lm, with a 70% gating to reduce crater edge
example, boron proÐles shifted towards the surface by e†ects and to maximize sensitivity. The sputter rates
P4 nm have been observed during oxygen bombard- were determined from crater depth measurements per-
ment with oxygen backÐll, with the amount of proÐle formed with a Tencor Alpha-Step 500 surface proÐler
shift depending not only on the beam conditions but after the SIMS measurements.
also on the oxygen pressure.6,13 An alternative An oxygen backÐll in the main chamber from 10~8
approach towards eliminating transient e†ects is to Torr up to 10~6 Torr was used during analysis of the
encapsulate the surface with a thin layer of amorphous 0.5 keV 11B` implant with the oxygen primary beam. A
silicon.14 In this method, the primary ion beam equi- 60¡ incident electron beam was sometimes applied
librium is established while bombarding the cap layer, simultaneously with the oxygen backÐll during oxygen
but ion yields and sputter rates still vary through the sputtering in order to investigate the e†ects of electron-
underlying native oxide and the substrate. It has been beam-assisted oxidation on the transient region. The
shown that the enhancement e†ects of the native oxide electron beam used for stimulated oxidation was set to
can be reduced substantially by using an oxygen backÐll produce a 2 keV beam of 105 nA. The width of the area
in conjunction with the amorphous cap layer.15 Even so, scanned by the electron beam was 1000 lm.
this method su†ers from some drawbacks, which During cesium bombardment, boron and arsenic
include the presence of extrinsic impurities trapped at were monitored using 39BSi~ and 103AsSi~, respec-
the interface, difficulty in identifying the original surface tively. Mass 11B` was followed for boron analysis
position and extra sample preparation steps.16 during oxygen bombardment. Silicon and the primary
This paper presents SIMS depth proÐle results oxygen beam were followed with 30Si, 58Si , 44SiO and
obtained for shallow arsenic and boron implants 16O. QuantiÐcation of the SIMS proÐles 2was accom-
without using the encapsulation procedure. The plished by processing the raw data into concentration
purpose of this study was twofold : to evaluate the vs. depth using PHI-Matlab data processing software.
e†ects of primary ion mass and beam energy on depth
resolution and peak shape while assessing the lowest
practical energy that we could use with our quadrupole
RESULTS AND DISCUSSION
SIMS system ; and to investigate the possibility of elimi-
nating the surface transient by applying an electron
beam in conjunction with oxygen backÐll during The e†ect of primary Cs` beam energies on As depth
primary oxygen bombardment. A 0.5 keV boron proÐles is shown in Fig. 1. The As depth proÐles were
implant in silicon was used to test this hypothesis. This acquired for Si implanted with an 8 keV As at a dose of
approach is based on the idea that oxidation is strongly 3 ] 1015 cm~2. It can be seen that proÐle broadening
stimulated by electron bombardment.17,18 Several decreased as the primary beam energy was reduced
studies19h22 have been reported on electron-stimulated from 5 to 1.5 keV. Better proÐle depth resolution at
oxidation of Si during oxygen exposure. In one recent lower sputtering energies was expected and observed.4,5
report,22 Auger electron spectroscopy clearly demon- Surprisingly, the As proÐle at 1 keV showed depth
strated the presence of SiÈO bonding at electron-beam- resolution similar to the 1.5 keV proÐle, and at concen-
irradiated Si surfaces. Electron bombardment is known trations below 1 ] 1020 cm~3 it broadened substan-
to dissociate oxygen originally adsorbed in molecular tially. This may be due to either a poor primary ion
form, leading to the formation of SiO .23 An assess- beam shape at 1 keV or sputter-induced roughening.
ment to the e†ectiveness of this method 2 is presented,
Increasing electron gating and raster sizes to exclude
based on the width of the near-surface transient region. crater edge e†ects did not improve depth resolution.
The width of the surface transient region before steady-
state conditions are established (not shown) decreased
EXPERIMENTAL by D8 nm when the Cs` beam energy was reduced
from 5 to 1.5 keV. It is well known that lowering the
primary ion beam energy decreases not only cascade
The samples used in this work were an 8 keV/3 ] 1015 mixing e†ects but also the width of the surface tran-
cm~2 75As` implant into Si, a 5 keV/5 ] 1015 cm~2 sient.24 The depth of the As peak [in Fig. 1(a)] was 10
11B` implant into Si and a 0.5 keV/1 ] 1015 cm~2 nm for all primary beam energies, which was beyond
Surf. Interface Anal. 26, 851È860 (1998) ( 1998 John Wiley & Sons, Ltd.
SIMS OF SHALLOW IMPLANTS WITH OXYGEN BACKFILL 853

Figure 1. The 103AsSiÉ depth profiles from an 8 keV implant in Si for various Cs½ primary SIMS beam energies : (a) complete depth profile ;
(b) peak region.
the transient depth measured under these conditions, as The maximum concentration was observed to be
discussed below. The location of the As peak is in agree- 2.5 ] 1021 cm~3 for a 1.5 keV primary beam. This
ment with the projected range of 10.7 ^ 3.7 nm esti- improvement in depth resolution results from a
mated from TRIM calculations. In Fig. 1(b), the peak in decrease in the surface transition region. Thus, the best
As concentration sharpened and increased gradually in depth resolution was achieved with 1.5 keV Cs`, which
concentration with decreasing energy (except at 5 keV). is well below the implant energy.
( 1998 John Wiley & Sons, Ltd. Surf. Interface Anal. 26, 851È860 (1998)
854 M. PUGA-LAMBERS AND P. H. HOLLOWAY

Figure 2(a) shows the SIMS boron depth proÐles energy conditions (consistent with the projected range
obtained from Si implanted with 5 keV boron to a dose of 23.7 ^ 1.7 nm estimated from TRIM) and showed an
of 5 ] 1015 cm~2 using an O ` primary ion beam. For approximate concentration of 1.5 ] 1021 cm~3. For
2
the purpose of comparison, the proÐles were also both Cs` and O ` bombardment, reducing the primary
acquired with a Cs` primary ion beam [Fig. 2(b)]. The ion beam energy2slightly improved the depth resolution.
boron peak was located at a depth of 22 nm for all This indicates that the beam conditions provided suffi-

Figure 2. Boron depth profiles of a 5 keV implant in Si at various energies using : (a) O ½ to detect 11B½ ; (b) Cs½ to detect 39BSiÉ.
2
Surf. Interface Anal. 26, 851È860 (1998) ( 1998 John Wiley & Sons, Ltd.
SIMS OF SHALLOW IMPLANTS WITH OXYGEN BACKFILL 855

Figure 3. Secondary ion distribution of 30Si½ from the sample in Fig. 2 using O ½ (a) and Cs½ (b) at different energies.
2

cient depth resolution for the proÐles. Boron proÐles deteriorated at concentrations below 1020 cm~3. Using
acquired at 4, 3 and 2 keV show similar depth an O ` primary beam at 1 keV resulted in a poorer
resolution for both Cs` and O `, whereas the 1 keV depth2 resolution at concentrations \7 ] 1018 cm~3.
2
proÐle showed both poorer depth resolution and a This result is probably due to a poor primary ion beam
lower peak concentration than at higher energies. With shape at 1 keV or increased surface-induced roughness
Cs` bombardment, the 1 keV proÐle depth resolution by oxygen sputtering, which degrades depth
( 1998 John Wiley & Sons, Ltd. Surf. Interface Anal. 26, 851È860 (1998)
856 M. PUGA-LAMBERS AND P. H. HOLLOWAY

Figure 4. Comparison of boron depth profiles (a) and 30Si½ surface transient (b) obtained from a 0.5 keV boron implant in Si using O ½
2
and Cs½ at 1.5 and 1 keV.

resolution.25 The proÐles in Fig. 2 also show a change The location of the arsenic and boron maximum
in the slope of the boron distribution at concentrations peaks [Figs 1 and 2] is independent of sputtering
of D1019 cm~3. This is due to channeling e†ects of the energy over all the primary ion beam energies used.
implant into silicon, which increase the penetration This result was observed because the penetration depth
depth of the implanted ions. of Cs` and O ` primary ions was smaller than the
2
Surf. Interface Anal. 26, 851È860 (1998) ( 1998 John Wiley & Sons, Ltd.
SIMS OF SHALLOW IMPLANTS WITH OXYGEN BACKFILL 857

Figure 5. Boron depth profiles of a 0.5 keV boron implant using electron-assisted O ½ bombardment in combination with oxygen backfill.
2

depth of the implant proÐles. The penetration depth of energies. Although using sub-keV primary energies
either of the primary ion beams was estimated from the would have been more appropriate to characterize a 0.5
primary energy and angle of incidence by using an keV boron implant, our choice of conditions was dic-
equation derived in Ref. 24. tated by instrumental constraints and sample avail-
Figure 3 illustrates the surface transient of the silicon ability.
matrix signal 30Si` at various energies obtained with Figure 4(a) shows a comparison of the boron depth
O ` and Cs`, respectively, from the sample implanted proÐles obtained from Si implanted with 0.5 keV boron
2 boron at 5 keV 5 ] 1015 cm~2. The surface
with to a dose of 1 ] 1015 cm~2 using O ` or Cs` primary
enhancement observed with O ` sputtering is larger ion beams at 1.5 and 1 keV and at an2 incidence angle of
2
than with Cs` sputtering, and increases with decreasing 60¡. It is quite clear that reducing the energy from 1.5 to
primary ion energy because the native oxide is mixed to 1.0 keV substantially improved the depth resolution.
a deeper point with increased energy. The width of the Furthermore, O ` proÐling provided a much better
transition region decreased by D15 nm (from 25 nm to depth resolution2than Cs`. This is true because under
10 nm) as the O ` energy was reduced from 5 keV to 1 Cs` bombardment there is a linear dependence between
keV [Fig. 3(a)].2 For the same Cs` energy range, the the decay length and the penetration depth of primary
width of the transition region only decreased by D3 nm ions, which exclusively determines the amount of
(from 6 nm at 5 keV to 3 nm at 1 keV) [Fig. 3(b)]. This mixing.27 Therefore, stronger mixing (i.e. larger decay
agrees with previous Ðndings26 in which a 2 keV Cs` lengths) occurs when the penetration depth of the
beam at 60¡ produced a 2 nm deep transition region primary ion exceeds the depth of the implant proÐle.24
sufficient to resolve a 5 keV boron implant in Si. The In the case of O ` bombardment, the additional
same O ` energy resulted in a much deeper transition 2
incorporation of oxygen into the Si matrix (swelling
2
region width due to the lower mass of oxygen vs. model) causes a change in the matrix density of the
cesium. surface region, leading to smaller decay lengths and,
From the results presented in Figs 1È3, it is feasible to consequently, to improved depth resolution.28
use beam energies down to 1 keV with our quadrupole The width of the transition region before the matrix
SIMS system, although this inherently resulted in a loss signal 30Si` reaches steady state is shown in Fig. 4(b).
in maximum current and problems in focusing the The transition region obtained with 1 and 1.5 keV O `
primary ion beam. Optimizing the beam shape at very 2
beams is D 5 nm deep. The peak of the boron implan-
low beam energies required a substantial increase in tation distribution was measured to be at a depth of 3
raster size in order to minimize crater edge e†ects and nm [Fig. 4(a)], but this value may not be correct for two
to compensate for reduced dynamic range. Obviously, reasons : the calculation of the sputter rate in Figs. 4(a)
this resulted in longer analysis times. In order to investi- and 4(b) did not take into account the increased sputter
gate the possibility of enhanced oxidation by electron rate in the surface oxide, which is about 2.2 times that
bombardment, a 0.5 keV boron implant in Si was in Si ; and, as reported previously, when the penetration
analyzed with O ` and Cs` using 1.5 and 1 keV beam depth of the primary ion exceeds the projected range of
2
( 1998 John Wiley & Sons, Ltd. Surf. Interface Anal. 26, 851È860 (1998)
858 M. PUGA-LAMBERS AND P. H. HOLLOWAY

Figure 6. Comparison of 30Si½ (a) and 44SiO½ (b) surface transients from a 0.5 keV boron implant in Si for only O ½ bombardment or for
simultaneous electron bombardment at different oxygen backfill pressures. 2

the implant, the peak position tends to shift towards the oxygen incorporation and by the native surface oxide.
surface due to continuous oxygen incorporation and, Furthermore, wider transition regions result in larger
consequently, a variation in sputter rate within the o†sets on the depth scale due to sputter rate di†erences
surface oxide region itself.6 Therefore, quantiÐcation of between SiO (with x O 2) and Si. A 1 keV O ` beam
the boron depth proÐles illustrated in Fig. 4(a) is signiÐ- was used to xanalyze this sample under oxygen2 backÐll
cantly a†ected by the ionization yield changes due to conditions (5 ] 10~6 Torr) vs. base pressure
Surf. Interface Anal. 26, 851È860 (1998) ( 1998 John Wiley & Sons, Ltd.
SIMS OF SHALLOW IMPLANTS WITH OXYGEN BACKFILL 859

(5.5 ] 10~10 Torr). The SIMS proÐles [Fig. 5] also electron beam irradiated. A lower sputter rate during
show the e†ect of an electron beam under the same oxygen backÐll results from lower sputter yields for Si,
oxygen backÐll conditions. Oxygen backÐll has been which have been reported to decrease by a factor of two
used extensively to reduce surface transient e†ects, but a to three25,32 as the oxygen pressure was increased. The
simple backÐll is not sufficient to instantaneously form total sputter yields of Si and SiO were reported to level
a stoichiometric oxide.2,6,11,12 By applying electron o† at pressures above 10~6 2Torr, which indicates
irradiation simultaneous with ion sputtering, we stimu- similar compositions for the Si surface region and
lated the dissociation of molecular oxygen and the SiO .32
2
immediate formation of SiO . Figure 6 shows the surface transient signals of 30Si`
2
In Fig. 5, depth resolution improved as the pressure and 44SiO`, respectively, as a function of depth under
was raised from 1.4 ] 10~7 to 5.4 ] 10~6 Torr. It was various oxygen backÐll conditions. The depth values
improved even further by an electron beam at backÐll have not been adjusted for sputter rate changes. At base
pressures of 3.5 ] 10~7 and 2.8 ] 10~6 Torr. The pressure (not included in Fig. 6), the signals from 30Si`
dynamic range of the boron proÐles measured with high and 44SiO` reach their equilibrium levels after a tran-
oxygen pressure was reduced as the oxygen pressure sient depth of 3È4 nm. As the oxygen pressure was
was raised above the base pressure. This was observed increased, the matrix-speciÐc 30Si` transient width was
both with and without an electron beam ; however, the substantially reduced, as expected, reaching D0.3 nm at
electron beam tended to reduce this undesirable e†ect. 10~7 Torr and 0.15 nm at 10~6 Torr. No signiÐcant
Increased raster sizes to exclude crater edge e†ects did reduction was observed in the transient depth for elec-
not improve the dynamic range during oxygen backÐll, tron beam irradiation. It was previously presumed that
in agreement with previous reports.24 In addition, pro- full oxidation was achieved almost instantaneously
Ðles measured shortly after the system recovered back under oxygen backÐll conditions, based upon the fact
to a base pressure of 10~10 Torr still exhibited low that the Si` signal reached equilibrium more rapidly
dynamic ranges, even after 1 day of pumping. This with an oxygen backÐll than at base pressure.32 Under
reduced dynamic range has been attributed to a high oxygen backÐll conditions, the sample was saturated
boron background signal caused by memory with oxygen, thus maintaining a constant ionization
e†ects.24,25 Memory e†ects have been attributed to yield.2 But the 44SiO` transient [Fig. 6(b)] only
increased scattering of primary and secondary ions reached a stable level at much larger depths (2.5 nm at
under oxygen backÐll conditions, a high boron sticking 10~7 Torr and 2.0 nm at 10~6 Torr). This indicates that
coefficient on oxidized surfaces and higher ionization full oxidation was achieved at depths greater than those
yields.24 The boron background level was consistently observed with the 30Si` transient. By expanding the
lower at all pressures when an electron beam was near-surface area, a reduction of D 0.3È0.5 nm in the
applied. Presumably, the electron beam reduces the width of the transition region can be detected for elec-
memory e†ect by stimulating the conversion of boron tron beam irradiation. This supports the fact that simul-
to oxide with a reduced SIMS signal. In addition, the taneous electron beam bombardment accelerates the
improved depth resolution from simultaneous electron oxidation process, in agreement with reports by Kirby
and ion bombardment may be inÑuenced by reduced and Litchman.23
ion-beam-induced roughness. Previous studies29 have
shown that electron beam irradiation simultaneous with
ion beam bombardment suppresses ripple growth in Si.
Wittmaack has reported30 that low-energy oxygen CONCLUSIONS
bombardment (0.5È2 keV) of Si at incidence angles
above 40¡ during vacuum and oxygen backÐll condi-
tions caused rapid surface roughening with ripple for- The SIMS depth proÐles of 8 keV arsenic and 5 or 0.5
mation, which may be suppressed by the electron-beam- keV boron implants into Si (100) wafers were measured.
stimulated oxidation. It was shown that the depth resolution of the implant
Depth resolution is frequently quantiÐed in terms of proÐle improved as the beam energy was reduced from
the decay length.2,7,24,31 Decay lengths were derived 5 keV to 1.5 keV for both O ` and Cs` primary beams.
over an order of magnitude drop along the trailing edge The depth resolution was 2not as good for 1.0 keV
of the boron proÐles in Fig. 5. They were larger than beams, presumably because of trouble in focusing the
those reported previously.2,31 This discrepancy is prob- ion beam at this low energy. Oxygen backÐlling
ably caused by ion-beam-induced roughening and improved the depth resolution but degraded the
memory e†ects as discussed above, and to some extent dynamic range of boron implants. Simultaneous elec-
by ion-beam-induced mixing and sample di†erences. In tron bombardment with depth proÐling improved the
addition, it has been suggested14 that enhanced di†u- depth resolution and maintained the dynamic range of
sion caused by primary ion-beam-induced damage the SIMS analysis. By analysis of both the 30Si` and
during boron analysis can cause the depth resolution to 44SiO` transients, it was demonstrated that the extent
deteriorate. of surface oxidation was accelerated by simultaneous
The sputter rates determined from the proÐles shown electron bombardment.
in Fig. 5 decreased by a factor of three when oxygen
was backÐlled into the chamber up to a pressure of
10~7 Torr. Between 10~7 and 10~6 Torr, the sputter Acknowledgements
rate remained constant, consistent with previous
reports.6 Under identical backÐll conditions, no changes Extensive discussions with Kevin Jones, Erik Kuryliw and Fred Stevie
in sputter rate were detected when the sample was also are gratefully acknowledged.

( 1998 John Wiley & Sons, Ltd. Surf. Interface Anal. 26, 851È860 (1998)
860 M. PUGA-LAMBERS AND P. H. HOLLOWAY

REFERENCES

1. N. S. Smith, M. G. Dowsett, B. McGregor and P. Phillips, in eling of Ultra -Shallow Doping Profiles in Semiconductors ,
SIMS X , edited by A. Benninghoven, B. Hagenhoff and H. W. Research Triangle Park, NC, 6–9 April 1997, Section 10.1.
Werner, p. 363. Wiley, New York (1995). 17. W. Ranke and K. Jacobi, Surf . Sci . 47, 525 (1975).
2. C. W. Magee, J. R. Shallenberger, M. S. Denker, D. F. 18. H. Ibach, K. Horn, R. Dorn and H. Luth, Surf . Sci . 38, 433
Downey, M. Meloni, S. D. Cloherty, S. Felch and B. S. Lee, (1973).
Fourth International Workshop on the Measurement , Charac- 19. B. Carriere, A. Chouiyakh and B. Lang, Surf . Sci . 126, 495
terization and Modeling of Ultra -Shallow Doping Profiles in (1983).
Semiconductors , Research Triangle Park, NC, 6–9 April 1997, 20. R. R. Kunz and T. M. Mayer, J . Vac . Sci . Technol . B5(1), 427
Section 8.1. (1987).
3. P. D. Augustus, G. D. T. Spiller, N. G. Dowsett, P. Knightley, 21. D. C. Sun, Z. Q. Yu, F. M. Li, Y. C. Du and H. Wang, Appl .
G. R. Thomas, R. Webb and E. A. Clark, in SIMS VI , edited by Phys . A 48, 567 (1989).
A. Benninghoven et al ., p. 488. Wiley, New York (1988). 22. M. Matsutani, F. Wakaya, S. Takaoka, K. Murase and K.
4. H. H. Andersen, Appl . Phys . 18, 131 (1979). Gamo, Jpn . J . Appl . Phys . 36, 7782 (1997).
5. K. Wittmaack, Vacuum 34, 119 (1984). 23. R. E. Kirby and D. Litchman, Surf . Sci . 41, 447 (1974).
6. K. Wittmaack and S. F. Corcoran, J . Vac . Sci . Technol . B16, 24. W. Vandervorst and F. R. Shepherd, J . Vac . Sci . Technol . A5,
272 (1998). 313 (1987).
7. M. G. Dowsett, in SIMS XI , edited by R. Lareau, G. Gillen, J. 25. J. W. Erickson and R. Brigham, J . Vac . Sci . Technol . B14, 353
Bennett and F. Stevie, p. 259. Wiley, New York (1998). (1996).
8. S. P. Smith, V. K. F. Chia and M. H. Yang, IEEE Proc . 11th Int . 26. C. W. Magee, ‘Depth profiling of shallow junction depths’,
Conf . on Ion Implantation and Technology , Austin, Texas, Vol. Application Brief . Evans East.
1(1), 1996, p. 512. 27. M. Meuris, W. Vandervorst and J. Jackman, J . Vac . Sci .
9. S. P. Smith, V. K. F. Chia, C. J. Hitzman and G. R. Mount, Technol . A5, 1482 (1991).
Proc . Ion Implant . Technol . 96, 599 (1997). 28. M. Petravic, R. G. Elliman and J. S. Williams, in SIMS VIII ,
10. K. Wittmaack, Surf . Sci . 68, 118 (1977). edited by A. Benninghoven, K. T. F. Janssen, J. Tumpner and
11. K. Wittmaack, Surf . Interface Anal . 24, 389 (1996). H. W. Werner, p. 367. Wiley, New York (1991).
12. M. G. Dowsett, T. J. Ormsby, D. I. Elliner and G. A. Cooke, in 29. M. Hatada, Y. Nakagawa, M. Saeda, A. Karen, K. Okuno, F.
SIMS XI in, edited by R. Lareau, G. Gillen, J. Bennett and F. Soeda and A. Ishitani, in SIMS IX , edited by A. Benningho-
Stevie, p. 371. Wiley, New York (1998). ven, Y. Nihei, R. Shimizu and H. W. Werner, p. 738. Wiley,
13. T. Hoshi, K. Miyoshi and M. Tomita, in SIMS IX , edited by A. New York (1993).
Benninghoven, Y. Nihei, R. Shimizu and H. W. Werner, p. 30. K. Wittmaack, ISSIMS 98 /SIMS II, Beijing, China, 6–10 April
710. Wiley, New York (1993). 1998.
14. J. G. M. van Berkum, E. J. H. Collart, K. Weemers, D. J. 31. P. C. Zalm and C. J. Vriezema, Nucl . Instrum . Methods B64,
Gravesteijn, K. Iltgen, A. Benninghoven and E. Niehuis, J . 626 (1992).
Vac . Sci . Technol . B16, 298 (1998). 32. M. Nakamura, K. Yamada, K. Okuno, F. Soeda and A. Ishitani,
15. J. G. M. van Berkum, in SIMS XI , edited by R. Lareau, G. in SIMS IX , edited by A. Benninghoven, Y. Nihei, R. Shimizu
Gillen, J. Bennett and F. Stevie, p. 187. Wiley, New York and H. W. Werner, p. 207. Wiley, New York (1993).
(1998).
16. K. Iltgen, A. Benninghoven and E. Niehuis, Fourth Internation-
al Workshop on the Measurement , Characterization and Mod-

Surf. Interface Anal. 26, 851È860 (1998) ( 1998 John Wiley & Sons, Ltd.

You might also like