Nuclear Instruments and Methods in Physics Research B 164±165 (2000) 979±985
www.elsevier.nl/locate/nimb
2D Monte Carlo simulation of proton implantation
of superconducting YBa2 Cu3 O7ÿd thin ®lms
through high aspect ratio Nb masks
Nianhua Peng
a
a,*
,
Ivan Chakarov b, Chris Jeynes a, Roger Webb a, Wilfred Booij c,
Mark Blamire c, Michael Kelly a
Surrey Centre for Research in Ion Beam Applications, School of Electronic Engineering, Information Technology and Mathematics,
University of Surrey, Guildford GU2 5XH, UK
b
SILVACO International, Santa Clara, CA 95054, USA
c
Department of Materials Science and Metallurgy, University of Cambridge, Pembroke Street, Cambridge CB2 3QZ, UK
Abstract
Perturbation of proton beam damage pro®le due to sidewall interactions in very high aspect ratio implant masks has
been studied using Monte Carlo simulations. The model structure is composed of amorphous Nb metal mask, crystalline high temperature superconducting YBa2 Cu3 O7ÿd (YBCO) thin ®lm, and amorphous LaAlO3 substrate. The
simulation results reveal the existence of enhanced proton beam penetration in target materials due to sidewall interactions. Ó 2000 Elsevier Science B.V. All rights reserved.
1. Introduction
The discovery of cuprate superconductors in
the late 1980s [1,2] stimulated worldwide studies
on high temperature superconductivity. With the
superconducting transition temperature Tc well
above liquid nitrogen temperature in many new
superconductors, it is possible that applications of
superconductivity based on high temperature superconductors can work at a more feasible temperature and cost for wider applications.
Fabrication of superconducting/normal/supercon-
*
Corresponding author.
E-mail address: ees1np@ee.surrey.ac.uk (N. Peng).
ducting (SNS) sandwiched Josephson junction
devices with high Tc superconductor thin ®lms, as
the core of any superconducting electronics applications, has been demonstrated successfully in
many laboratories [3±6], though the manufacturing yield has poor throughput. The diculty in
making high quality high Tc Josephson junction
devices is closely related to the intrinsic short coherence length, at about nanometer scale for
YBa2 Cu3 O7ÿd (YBCO), and large penetration
depth of all high Tc superconductors. Consequently, workable SNS Josephson junctions have
to be made with a thin normal state region on a
nanometer scale. This is in sharp contrast to the
size of conventional Nb based Josephson junctions
[7].
0168-583X/00/$ - see front matter Ó 2000 Elsevier Science B.V. All rights reserved.
PII: S 0 1 6 8 - 5 8 3 X ( 9 9 ) 0 1 1 9 4 - 5
980
N. Peng et al. / Nucl. Instr. and Meth. in Phys. Res. B 164±165 (2000) 979±985
It is well known that high temperature superconductivity in cuprates is sensitive to crystal defects and chemical non-stoichiometries. For
example, creation of oxygen vacancies in YBCO
can lead to superconducting±non-superconducting±insulating phase transitions. Elemental substitutions, such as 3d transition metal elements for
Cu, depress the superconducting phase transition
temperature Tc in general, though the solubility of
such elemental substitutions is limited [8]. This
opens a new approach for the fabrication of superconducting devices, such as Josephson junctions, if the suppression of superconductivity can
be controlled up to a scale comparable with the
dimension of the coherence length. Appropriate
implantation of superconductors causes suppression of the superconducting transition temperature
Tc through radiation damage e€ects, so a SNS
junction can be made by implantation using a
suitable mask structure or using focused electron
and ion beams. Further investigations are underway in many laboratories worldwide using masked
ion implantation damage processes, allowing
faster manufacturing of device batches, due to the
recent developments of lithography technology for
the fabrication of implantation masks in the nanometer range [9±16].
It is very clear that the radiation damage distribution of high temperature superconductors,
such as YBCO, is important for the performance
of devices fabricated under energetic ion beam
bombardment. Considering the small physical size
of the mask structure (down to nanometer range)
involved here, the damage behavior is very interesting but not fully understood. To con®ne the
lateral distribution of the damaged area within a
small area, and with high homogeneity, a high
energy proton beam, and thin (50 nm) superconducting YBCO ®lms are proposed for the
Josephson junction fabrication to avoid the lateral spreading of ion beam scattering and to
maintain high spatial de®nition of the damaged
region. The purpose of this paper is to report on
the development of a simulation program to investigate the e€ects of implantation on Nb
masked YBCO thin ®lms, and to report on a
problem found by the simulator of a mechanism
for ion to penetrate the mask, and consequently
the need to make the mask much thicker than
expected.
2. The Monte Carlo simulator: CRYSTAL
The Monte Carlo simulation program CRYSTAL was written to meet the requirements for
modelling of damage accumulation and channelling implantation in crystal silicon. The program is
highly parallelized and vectorized, which results a
signi®cant reduction in computation times. It has
been adapted in order to take account of the
complex crystal structure of YBCO. The detailed
description of CRYSTAL has been reported elsewhere [17]. The schematic structure model is illustrated in Fig. 1. To simplify the calculation, the
LaAlO3 substrate is approximated by an amorphous structure in®nitely thick. The assumption of
an amorphous substrate is arbitrary, but its e€ect
on the damage distribution in YBCO thin ®lm is
expected to be small. On top of the LaAlO3 substrate is 50 nm YBCO with orthorhombic crystal
structure. The ion beam can be tilted by an angle h
relative to the sample normal. Then the normal
and the beam direction lie in the ``implantation
plane''. The ac plane of the YBCO is also normal
to the surface, and the angle between the implantation plane and projection plane is /. The ac
Fig. 1. A model structure for Monte Carlo simulation. The
layered structure is composed of 400 nm Nb mask with 50 nm
opening slot in the middle, 50 nm YBCO thin ®lm and LaAlO3
substrate.
N. Peng et al. / Nucl. Instr. and Meth. in Phys. Res. B 164±165 (2000) 979±985
plane is chosen in this work as the projection
plane, with h ˆ 15° and / ˆ 90°. No divergence of
the incident ion beam is assumed.
All simulations are carried out at room temperature with 105 trajectories recorded for each
simulation normalized to 1016 ions/cm2 implant
dose, except those speci®ed otherwise. Self-annealing e€ect of damage created by irradiation is
excluded.
3. Simulation results
It is well established that the defect concentration created by a radiation damage process is
proportional to its energy deposition, particularly
in the low dose range [18]. For superconducting
phase transition temperature Tc , to a good approximation, it is this damage level that a€ects Tc
rather than the speci®c ion beam energy or implant
dose. A series of YBCO thin ®lms have been
subjected to 50 keV proton irradiation with doses
from 1015 up to 4 1016 ions/cm2 , and the superconducting phase transition temperature Tc has
been determined by standard 4 terminal AC electrical resistance measurements. Energy depositions
have been derived from Monte Carlo simulations.
A plot of energy deposition against Tc is shown in
Fig. 2. It is clear that for an energy deposition less
Fig. 2. Correlation between superconducting phase transition
temperature Tc and simulated energy deposition by 50 keV
proton beam irradiation with various implant doses, for a 50 nm
YBCO thin ®lm on LaAlO3 substrate without Nb mask.
981
than 1021 eV/cm3 the superconducting transition
temperature Tc will be greater than 80 K, but Tc is
sharply suppressed for energy depositions greater
than 1021 eV/cm3 . An excellent linear correlation
of implant dose and Tc suggests that the variation
of self-annealing e€ect upon the implant dose can
be ignored for the doses concerned here.
Range calculations using SUSPRE 1 reveal that
about 350 nm thick Nb ®lm should be enough to
stop a 50 keV proton beam completely. Simulation
results from CRYSTAL are consistent with this
conclusion. For a 400 nm Nb mask on a 50 nm
YBCO layer, all the proton beam has been stopped
by Nb mask, Fig. 3(a), and no implant damage can
be observed inside the YBCO layer, Fig. 4(a). This
has also been con®rmed experimentally.
However, gradual opening of a central slot on
the Nb mask leads to a signi®cant ``leakage'' of
incident beam into the YBCO thin ®lm and
LaAlO3 substrate. A striking general feature is
that a 400 nm Nb mask seems unable to stop the
beam completely, with an extended damaged
pro®le underneath the Nb mask being developed,
as shown in Figs. 3 and 4. For a central opening
width of 5 nm, Figs. 3(b) and 4(b), the lateral
distribution of energy deposition greater than
1019 eV/cm3 extends up to 250 nm, the variation in
energy deposition along the forward direction of
ion beam is also obvious, though the badly damaged central area with signi®cant Tc suppression,
that is, with energy deposition greater than
1021 eV/cm3 , as shown in Fig. 4(b), is still con®ned
to a small area. Similar features have also been
obtained for a model with a 2 nm slot width. A
reasonably well de®ned central damaged region
can be observed with a central opening width
wider than 20 nm, as shown in Fig. 4(c), the dimension of the central region with energy deposition greater than 1021 eV/cm3 is con®ned to a 200
nm region. In contrast to the extended damage
underneath the Nb masks in Fig. 4(b) and (c),
the extension of damage range into the area
1
SUSPRE is a program based on energy deposition in the
surface and is available from R.P. Webb, SCRIBA, SEEITM,
University of Surrey, UK.
982
N. Peng et al. / Nucl. Instr. and Meth. in Phys. Res. B 164±165 (2000) 979±985
Fig. 3. Simulated net doping distributions (in log scale) in YBCO thin ®lms with various 400 nm Nb mask structures: (a) without
central opening; (b) 5 nm central opening; (c) 20 nm central opening; (d) semi-in®nite Nb mask.
Fig. 4. Simulated implant damage distributions (in log scale) with various mask structures: (a) without central opening; (b) 5 nm central
opening; (c) 20 nm central opening; (d) semi-in®nite Nb mask. Only energy depositions which are greater than 1021 eV/cm3 are
presented here.
N. Peng et al. / Nucl. Instr. and Meth. in Phys. Res. B 164±165 (2000) 979±985
983
Fig. 5. Variations of implant damage distribution upon mask thickness and incident beam energy with 50 nm central slot: (a) 400 nm
Nb mask, 50 keV proton beam, and 1016 ions/cm2 proton implant; (b) 1200 nm Nb mask, 50 keV proton beam, and 7:5 1016 ions/cm2
proton implant; (c) 400 nm Nb mask, 30 keV proton beam, and 1:75 1016 ions/cm2 proton implant.
underneath a semi-in®nite Nb mask is still there,
but is much weaker (<1021 eV/cm3 ) compared to
other structures with a central opening slot, indicating an enhanced penetration of proton beam
due to the sidewall interactions.
In fact, a slot width of 5 nm or even 2 nm is
dicult to make in practice with a 400 nm thick
Nb mask. The simulation results described above
also show that such a narrow slot has no advantage in achieving a well de®ned narrow and
homogeneous damaged region. We now consider
at a simple but di€erent approach to reduce the
extended damage by increasing the Nb mask
thickness with a standard 50 nm central opening slot. The simulation results are shown in
Fig. 5.
As expected, the lateral beam damaged area
decreases as the Nb mask thickness increases. As
far as the proton beam implant e€ect is concerned,
a minimum 1200 nm Nb mask layer seems necessary to stop nearly all the proton implant inside
the Nb mask layers, Fig. 5(b). It is interesting to
point out here that roughening of the Nb mask
slot sides helps to stop the implant ions a little bit
more e€ectively, obviously due to the e€ective
damping of sidewall interaction cascade. So the
simulation with smooth surfaces gives an upper
limit to the implant damage spreading process. It
is also easy to understand that the damage level is
lower in the YBCO target with a thick Nb mask,
as more proton ions have been stopped by the
mask. Ignoring the self-annealing process, the reduction in damage level can be easily compensated
by increasing the implant dose. In fact the nominal
dose used for Fig. 5(b) is 7.5 times higher than that
for the structure with only 400 nm Nb mask,
Fig. 5(a), in order to give a comparable damage
level in the central area con®ned by the mask
opening width.
Reduction in beam energy can also help to stop
the beam penetration, and also to reduce the range
of damage spreading. The results for a standard
400 nm Nb mask with 50 nm opening are shown
in Fig. 5 with incident beam energies of 30 keV
(Fig. 5(c)) and 50 keV (Fig. 5(a)). To compensate
the small reduction in damage level just inside the
central slot de®ned area with decrease in incident
beam energy, the normalized implant dose has
been increased by a factor of 1.75 for the 30 keV
case compared to 50 keV implant. It is clear that
984
N. Peng et al. / Nucl. Instr. and Meth. in Phys. Res. B 164±165 (2000) 979±985
the dimension of the central region with energy
deposition higher than 1021 eV/cm3 has been
decreased signi®cantly from about 250 nm for 50
keV down to about 200 nm for 30 keV beam
irradiation.
4. Discussion and conclusions
The damage pro®les of energetic ion irradiation
under varied mask structures have been studied
before both experimentally and theoretically for
semiconducting materials [19±21]. It is obvious
that the contribution of mask structure is getting
more important as the size of mask structure goes
down to the nanometre scale. We carried out
Monte Carlo simulation study on the sidewall interaction process, using superconducting YBCO
thin ®lm covered with varied Nb masks as an example, but the physics clearly applies to other
material systems.
Simulation results described above establish
clearly that sidewall interaction for high aspect
ratio mask structure is signi®cant. This interaction
is intrinsic for the structure model discussed here,
and its contribution towards the damage and dopant distributions can be modi®ed with changing
initial incident beam energy, aspect ratio and surface ®nishing of the mask structure. Previous
studies on the e€ect of mask edges is more or less
similar to the semi-in®nite mask structure described above. It is natural to anticipate that such
a e€ect will be less important as the width of
opening goes up, i.e., the aspect ratio goes down.
Similar interaction should be expected for other
systems, as the e€ect is basically dominated by the
geometrical factor of the mask structure.
The main consequence of this sidewall e€ect is
that the lateral damage range in the target material
underneath the mask structure is far larger than
the masked area. Provided the beam energy is high
and target ®lm thickness is not very thick, the most
prominent e€ect of sidewall interaction will be the
inhomogeneous distribution of both damage level
and net doping concentration along the lateral
direction. This may be a serious problem for nanometer scale electronics device fabrication where
homogeneous distribution of doping is important.
However, for YBCO sample considered here, the
entire damaged YBCO track can be better described as a series of stripes with varied superconducting transition temperatures; it is still likely
to form a well de®ned central damaged area with
suppressed superconducting transition temperatures or non-superconducting at all, and thus leave
a narrow working temperature and normal state
resistance window for the formation of Josephson
weak links. This has been con®rmed by our in-situ
implant experiments, which with the detailed description of Josephson junction fabrication and
characterization, have been reported elsewhere
[22]. Simulations of the optimal combinations of
beam energy and mask structures are still underway, as well as the incorporation of self-annealing
processes at high implant dose ranges.
Acknowledgements
This work is fully supported by EPSRC research grants, GR/L66199, GR/L65963, GR/
M85210, GR/M84756, and GR/L78512.
References
[1] J.G. Benorz, K.A. Muller, Z. Phys. B 64 (1986) 189.
[2] M.K. Wu, J.R. Ashburn, C.J. Torng, P.H. Hor, R.L.
Meng, L. Gao, Z.J. Huang, Y.Q. Wang, C.W. Chu, Phys.
Rev. Lett. 58 (1987) 908.
[3] D. Dimos, P. Chaudhari, J. Mannhart, F.K. LeGoues,
Phys. Rev. B 41 (1990) 4038.
[4] J. Gao, Y. Boguslavskij, B.B.G. Klopman, G.J. Gerritsma,
H. Rogalla, Appl. Phys. Lett. 59 (1991) 2754.
[5] G. Koren, et al., Appl. Phys. Lett. 64 (1994) 247.
[6] T. Satoh, J. Fujita, T. Yoshitake, H. Tsuge, Appl. Phys.
Lett. 62 (1993) 1685.
[7] T.P. Orlando, K.A. Delin, Foundations of Applied
Superconductivity, Addison-Wesley, Reading, MA, 1991.
[8] N.H. Peng, Ph.D. thesis, University of Cambridge, UK,
1992, and references therein.
[9] G.J. Clark, F.K. LeGouse, A.D. Marwick, R.V. Laibowitz, R. Koch, Appl. Phys. Lett. 51 (1987) 1462.
[10] R. Barth, A.H. Hamidi, B. Hadam, J. Hollkott, D.
Dunkmann, J. Auge, H. Kurz, Microelectron. Eng. 30
(1996) 407.
[11] C.H. Chen, I. Jin, S.P. Pai, Z.W. Dong, R.P. Sharma, C.J.
Lobb, T. Venkatesan, K. Edinger, J. Orto€, J. Melngailis,
Z. Zhang, W.K. Chu, J. Vac. Sci. Technol. B 16 (1998)
2898.
N. Peng et al. / Nucl. Instr. and Meth. in Phys. Res. B 164±165 (2000) 979±985
[12] W.Y. Liang, J. Yuan, Y. Yan, A.J. Pauza, Mater. Sci. Eng.
B 41 (1996) 5.
[13] B.A. Davidson, B. Hinaus, M. Rzchowski, K. Siangchaew,
M. Libera, J.E. Nordman, IEEE Trans. Appl. Superconductivity 27 (1997) 2518.
[14] W.E. Booij, A.J. Pauza, E.J. Tarte, D.F. Moore,
M.G. Blamire, Phys. Rev. B 55 (1997) 14600.
[15] S. Morohashi, T. Utagawa, Y. Enomoto, IEEE Trans.
Appl. Superconductivity 27 (1997) 2522.
[16] G.B. Assayag, J. Gierack, S. Flament, C. Dolabdjian,
F. Gire, E. Lesquey, C. Gunther, J.F. Hamet, C. Prouteau,
D. Robbes, Rev. Sci. Instr. 67 (1996) 446.
985
[17] I. Chakarov, R. Webb, Radiat. E€. Def. Solids 33 (1994)
447.
[18] B.D. Weaver, E.M. Jackson, G.P. Summers, E.A. Burke,
Phys. Rev. B 46 (1992) 1134.
[19] M.M. Faye, J. Beauvillain, Ph. Salles, L. Laana, A. Yahia
Messaoud, A. Martines, J. Appl. Phys. 72 (1992) 5117.
[20] G. Hobler, Nucl. Instr. and Meth. B 96 (1995) 155.
[21] V. Privitera, V. Raineri, M. Saggio, F. Priolo, E. Rimini,
Nucl. Instr. and Meth. B 96 (1995) 144.
[22] W.E. Booji, N.H. Peng, F. Kahlmann, R. Webb,
E.J. Tarte, D.F. Moore, C. Jeynes, M.G. Blamire,
Proceedings of EUCAS, 1999, Spain.