Interaction between electronic structure and strain in Bi nanolines on Si(001)

Abstract

Heteroepitaxial strain can be a controlling factor in the lateral dimensions of 1-D nanostructures. Bi nanolines on Si(001) have an atomic structure which involves a large sub-surface reconstruction, resulting in a strong elastic coupling to the surrounding silicon. We present variable-bias STM and first principles electronic structure calculations of the Bi nanolines, which investigates this interaction. We show that the strain associated with the nanolines affects the atomic and electronic structure of at least two neighbouring Si dimers, and identify the mechanism behind this. We also present partial charge densities (projected by energy) for the nanoline with clean and hydrogenated surroundings and contrast it to the clean Si(001) surface. Comment: 10 pages, 3 figures, submitted to Surface Science

Full text

Page 1

arXiv:cond-mat/0212514v1 [cond-mat.mtrl-sci] 20 Dec 2002
Interaction between electronic structure and
strain in Bi nanolines on Si(001)
J.H.G. Owena,b,∗,1, K.Mikia,band D.R. Bowlerc,2
aNanotechnology Research Institute (NRI), National Institute of Advanced
Industrial Science and Technology (AIST), 1-1-1 Higashi, Tsukuba. Ibaraki
305-8562, Japan
bNanomaterials Laboratory (NML), National Research Institute for Materials
Science (NIMS) Sengen 1-2-1, Tsukuba. Ibaraki 305-0047, Japan,
cDepartment of Physics and Astronomy, University College London, Gower Street,
London WC1E 6BT, UK
Abstract
Heteroepitaxial strain can be a controlling factor in the lateral dimensions of 1-D
nanostructures. Bi nanolines on Si(001) have an atomic structure which involves a
large sub-surface reconstruction, resulting in a strong elastic coupling to the sur-
rounding silicon. We present variable-bias STM and first principles electronic struc-
ture calculations of the Bi nanolines, which investigates this interaction. We show
that the strain associated with the nanolines affects the atomic and electronic struc-
ture of at least two neighbouring Si dimers, and identify the mechanism behind this.
We also present partial charge densities (projected by energy) for the nanoline with
clean and hydrogenated surroundings and contrast it to the clean Si(001) surface.
Key words:
∗Corresponding author
Email addresses: james.owen@materials.ox.ac.uk (J.H.G. Owen),
miki.kazushi@aist.go.jp (K.Miki), david.bowler@ucl.ac.uk (D.R. Bowler).
URL: http://www.cmmp.ucl.ac.uk/∼drb/research.html (D.R. Bowler).
1Present address : Dept. of Materials, Oxford University, Parks Rd,Oxford, OX1
3PH, UK
2Also at: London Centre for Nanotechnology, Department of Physics and Astron-
omy, Gower Street, London WC1E 6BT, UK
Preprint submitted to Surface Science1 February 2008

Page 2

1 Introduction
Nanometer-scale electronic technologies require not only the formation of
nanoscale devices, but also nanoscale interconnections between the devices.
At present, nanoscale structures may be made via at least three different ap-
proaches, all of which are likely to used in any scheme: top-down lithographic
methods, such as e-beam lithography and AFM lithography; bottom-up meth-
ods, which involve positioning of ex situ “prefabricated” structures such as
conducting molecules, semiconductor nanowires[1] and carbon nanotubes[2]
using, e.g., scanning probes or micro-fluidics; or by in situ self-assembly, as
in semiconductor quantum dots, or silicide nanowires[3]. We are pursuing a
hybrid self-assembly-based fabrication route, whereby a Bi nanoline[4,5] is
used as a nanowire or as a template for nanowires of other metals. These
nanolines are very long—over 400 nm in some cases—and straight: a kink in
a nanoline has never been seen. Unlike silicide nanowires[3], their width is
constant, occupying the space of 4 dimers (1.5 nm) in the Si(001) surface.
However, variable-bias STM of the nanolines indicates that they have a band
gap which is larger than the surrounding surface, and suggests that they are
not conductive[4,5].
Recently, we used a combination of atomistic structure calculations and ex-
periments to identify the structure of the nanoline[6,7] (which we have called
the Haiku structure). An example of the Bi nanoline on a clean Si(001) surface
is shown in Fig. 1, with the Haiku structure shown in the inset. It has a com-
plex reconstruction in the silicon substrate reaching down five layers below
the surface, though the surrounding surface is not reconstructed. The depth
of the reconstruction suggests that the nanoline may be coupled strongly to
its surroundings. Knowledge of the nanoline structure allows us for the first
time to make detailed calculations of the electronic structure of the nanoline,
which bear out the observations made in STM. The calculations also provide
an explanation for the additional features that we have observed in recent
atomic-resolution variable-bias STM of the Bi nanoline on the clean Si(001)
surface at room temperature.
In this letter, we present high resolution, variable bias images of a Bi nano-
line, and DFT calculations which explore the electronic structure of the Bi
nanoline. In the next section, we discuss details of the experimental and the-
oretical techniques used. We then present the atomic-resolution images of the
Bi-exposed Si(001) surface and interpret them in light of the various calcula-
tions.
2

Page 3

Fig. 1. A 50 nm × 50 nm STM image of a small section of the Si(001) surface
containing a short nanoline, and numerous small double-dot features (A). Note
that some of the double-dot features have formed small clusters in a diagonal or
c(4 × 4) arrangement (B). The dotted box marks the area shown in Fig. 2. Inset:
The Haiku structure for the Bi nanoline[6]. A pair of subsurface 7-membered rings
of Si form the basis of this unusual reconstruction.
2 Experimental and Theoretical Methods
The Si(001) substrate was cleaned using a standard process[8] before being
transferred into vacuum. The Si surface was prepared by flashing repeatedly to
1100◦C for a few seconds, until there was only a small pressure rise. The clean
surface was checked with STM before Bi deposition began. Bi was evaporated
from an effusion cell, a typical dose being Bi at 470◦C for 10 mins. STM
images were taken at the deposition temperature between 570-600◦C, and at
room temperature, using a JEOL 4500 XT UHV STM. The standard recipe
for the formation of the Bi nanolines is to anneal the Si(001) surface close
3

Page 4

to 600◦C, under a Bi flux of around 1 ML/min.[4]. In this case, the surface
was annealed at a slightly lower temperature, around 570◦C, and quenched to
300◦C as soon as the presence of nanolines had been confirmed. In this way,
we hoped to capture Bi nanolines while still growing, and hence make some
observations which would lead to a likely nucleation mechanism. To minimise
surface contamination, the sample was held at 300◦C while the Bi cell was
cooled and the chamber pressure returned to its base level, before the sample
was cooled to room temperature.
The density functional theory (DFT) calculations were performed in the Gen-
eralised Gradient Approximation (GGA)[9,10] using the VASP code[11], with
ultrasoft pseudopotentials, a plane wave cutoff of 150eV (sufficient for en-
ergy difference convergence) and a Monkhurst-Pack k-point mesh with 4×4×1
points. The unit cell used contained a single Bi nanoline, and had ten layers
of Si, with twenty atoms in each layer (forming a single dimer row ten dimers
long with the p(2×2) reconstruction) with the bottom two layers constrained
to remain fixed and dangling bonds terminated in hydrogen. This unit cell was
of sufficient size to for convergence of energy with cell depth, and long enough
that the end Si dimers were representative of the clean, undistorted surface.
3Electronic structure of Bi nanoline
Figure 1 shows a typical image from our STM experiments. Note that the
Si(001) surface is clean, and that the image was taken at room temperature
(in contrast to all previous experiments, where either the substrate was hy-
drogenated or the temperature was ∼ 570◦C). There are two key features
to note: first, the Bi nanoline, which is the double line extending diagonally
across the image from top left to bottom right; second, the features (marked
‘A’ in Fig.1) which appear as a double dot, and which are scattered across
the surface; note that these features tend to align diagonally with each other,
even forming a local c(4×4) pattern in places (marked ‘B’). In this paper, we
will concentrate on the electronic structure of the nanoline. We are actively
investigating a possible structure for the double-dot features, which may be a
precursor to the nanoline, and we shall present our results in future work.
In order to probe more closely the physical and electronic structure of the
nanoline, a series of atomic-resolution variable bias STM images (in the area
shown by the dotted box in Fig. 1) were taken and are shown in Fig. 2.
The STM bias voltages are given in the figure caption. Filled-states images
range from -2.0V down to -0.3V. An empty-states image at a bias voltage of
+1.5V is also presented. As has been previously reported[4,5], the Bi nanoline
appears bright at large bias voltages of either sign, but as the bias voltage is
reduced, the contrast of the nanoline reduces so that at around -0.8V, the
4

Page 5

Fig. 2. 10nm × 10nm insets from the area shown in the dotted black box in Fig. 1.
The sample bias voltages used are -2.0V, -1.2V, -0.6V, -0.3V, +1.5V and -0.8V, in
(a-f) respectively. As the sample bias is reduced, between (a) and (d), the nanoline
changes contrast from light to dark relative to the surrounding Si(001). Over this
range, some of the dimers in the nanoline (marked A in (c) and (e)) exhibit a
different voltage contrast. At very low biases, around -0.3V, an enhancement of the
dimers around the nanoline, similar to that seen around a missing dimer defect in
clean Si(001)[13], is seen. This is visible (and marked schematically as B) in (d). In
(f), the resolution is sufficient to see that the corrugation of the Si dimers closest to
the nanoline is increased, suggesting a greater separation, and hence tensile strain.
relative contrast of the nanoline and the Si substrate is the same, and at lower
bias voltages, the nanoline becomes dark relative to the clean Si surface. By
contrast, images of the Bi nanoline on the H:Si(001) surface[12] always show
the nanoline as a bright feature.
At very low biases, as in Fig. 2(d), substrate dimers either side of the nanoline
become enhanced, i.e. appear to be brighter than the rest of the Si dimers,
out to a distance of at least two dimers. (This is shown schematically in (d),
by the lighter and darker rectangles.) This phenomenon of enhancement of
neighbouring dimers at low bias voltages has been seen previously for single
missing dimer defects (1DV) on Si(001)[13]. In that case, the enhancement
was explained by the distortion of these dimers away from their normal struc-
ture by the strain field of the 1DV, resulting in a local change of the electronic
structure, and the top-most occupied states being moved higher in energy. The
tensile strain around the Bi nanoline is expected to be the cause of the en-
hancement seen here; this enhancement is modelled below. In Fig. 2(f) (which
5

Page 6

shows extremely high resolution), tensile strain near the Bi nanoline is seen
directly: the darkening between the first and second Si dimers is more pro-
nounced than between the second and third Si dimers, indicating an increase
in separation between these dimers. In our DFT modelling, the dimer-dimer
spacing close to the nanoline is increased by 3%, a significant strain.
Another feature visible in the images is a change of contrast between Bi dimers
in the nanoline. While at elevated temperatures the Bi nanoline always appears
uniform, at room temperature, higher resolutions may be achieved, and some
subtle details resolved. Some dimers in the Bi nanoline in Fig. 2(a) appear
different to the rest of the line (they seem to have a dark boundary around
them); these separated dimers correspond to brighter dimers on the otherwise
dark nanoline in the lower bias images (especially Fig 2(c)) and darker patches
(marked as ‘A’) in the empty states image shown in Fig. 2(e). The reason
for these features have not been determined. However, they are not seen in
elevated-temperature images of the clean Bi nanoline, even at the same bias
voltages, while similar dark patches were observed during experiments with
adsorbed hydrogen and oxygen[12]. It is therefore quite possible that they are
the result of contamination by ambient water or hydrogen while cooling down
to room temperature.
We have calculated the electronic structure of the Haiku model for the Bi
nanoline using DFT, as described above. In order to understand the voltage
contrast seen in STM, we have projected out the charge density associated
with various states within certain energies of the Fermi level. We show the
partial charge densities associated with all states within 0.2eV, 0.6eV and
1.0eV of the Fermi energy in Figure 3 along with the complete charge density,
for three systems: (a) the clean Si(001) surface; (b) the Haiku model of the
Bi nanoline; and (c) the Haiku model with a hydrogenated Si(001) surface.
For the last, the first image is missing as the first states occur 0.55eV below
the Fermi level (a side-on view of the Haiku structure is shown in its place).
The partial densities are shown in a plane passing through a Si dimer atom.
As the results are for a statically buckled Si(001) surface, alternate dimers are
“up” and “down” (except on the hydrogenated surface, where the buckling
does not occur). Only the up atoms appear bright, resulting in an appearance
of alternate missing dimers; the difference is most clearly seen by comparison
of the electron charge densities with the structural model of the Haiku shown
in the top right-hand corner of Fig. 3.
The partial densities are important for the comparison with experiment, since
the states close to the Fermi level will contribute strongly to the STM images
at low biases. Localisation of a state near the Fermi level in the area around
particular atoms or bonds (in comparison to bulk or clean Si) is indicative of a
local strain, because strain forces structures away from their equilibrium state,
and hence minimum energy configuration. We first identified this connection
6

Page 7

Fig. 3. Contour plots of charge density for the states (i) 0.2eV, (ii) 0.6eV and (iii)
1.0eV below EF, and (iv) the total charge density for: (a) the clean Si(001) surface
(first column); (b) the Haiku structure on the clean surface (second column), and (c)
the hydrogenated surface (third column). The 0.2eV image for the hydrogenated
surface has been replaced with a ball-and-stick model of the Haiku structure, as
there are no states within 0.2eV of EF. See text for a full discussion.
between strain and STM contrast (or states moved higher in energy) in the
case of the single missing dimer defect on Si(001)[13], and we will use it here
to elucidate the substrate distortions around the Haiku structure. In the case
of the clean Si(001) surface (first column of Fig. 3), the states near the Fermi
level (a,i) are mainly localised on the surface dimers, and the density fades
away beneath the surface layer. This is as expected, since the dimerisation of
the Si(001) surface is associated both with dangling bonds (at the surface)
and with strain (sub-surface). By contrast, the states near the Fermi level
for the Bi nanoline on clean Si(001) (b,i) are reduced in density below the Si
dimers, and are entirely excluded from the Haiku structure and substructure.
Instead, the two Si dimers immediately to the side of the Haiku show a greatly
enhanced charge density in the states within 0.2eV of the Fermi level, both
by comparison with their neighbours (the left-most dimers in (b,i)) and with
the clean Si(001) surface (a,i), which is seen in STM as the brightening of the
dimers either side of the Haiku at very low biases (Fig. 2(d) in particular).
The localisation of these low-energy states indicates that the substructure of
the Haiku, far from being a strained structure is more relaxed than the layers
7

Page 8

beneath a clean Si(001) surface. Furthermore, it demonstrates that the strain
induced by the Bi nanoline is concentrated on the Si dimers either side of the
Haiku structure, and on the second-to-third layer bonds beneath the dimers.
The electronic structure of the Bi nanoline with a H-terminated Si(001) surface
is shown in Fig. 3(c). The major effect is the removal of the surface states
associated with the Si dimer π-bonds, and the removal of the buckling on
the surface. The Si atoms immediately beneath the Bi nanoline now show
an increased charge density relative to the neighbouring Si atoms, which is
clearly visible in the 0.6eV and 1.0eV images (c,ii & iii), indicating that the
region below the Bi nanoline is more strained than bulk Si. Furthermore,
enhancement can be seen in layers as deep as the sixth layer, indicating the
deep strain caused by the presence of the Bi nanoline. This difference in the
electronic states suggests that the contrast seen with the clean surface is due
to the interaction of the buckled dimers and the nanoline (and in particular
the “up” atom of a buckled dimer) which is removed with a hydrogenated
surface.
The lack of any states on the Bi near the Fermi level is clearly the cause of
the darkening of the nanoline relative to the Si(001) surface seen in STM at
low voltages. Indeed, the first states seen on the nanoline lie over 0.5eV away
from the Fermi level (which agrees qualitatively with the bias voltage of 0.8V
where parity of appearance between the nanoline and the substrate occurs). In
the full charge densities(Fig. 3(iv)), the density of states associated with the
Bi dimers, combined with their higher physical height, explains their relative
brightness in STM at higher bias voltages. In the case of the hydrogenated
surface, the Bi dimers have a similar charge density as in the clean surface,
but in this case, the Si π-bonds have been eliminated, so the Bi dimer remains
bright in STM at all biases[12]. The large density associated with the nanoline
in the 0.6eV(c,ii) and 1.0eV(c,iii) images underline the mechanism behind the
brightening of the Bi nanoline.
4 Conclusions
We have presented high resolution room temperature STM images of Bi nano-
lines on a clean Si(001) surface for the first time. These images allowed us to
probe the electronic structure of the nanoline and the surrounding Si(001)
surface, and, by extrapolation, the localised strain associated with the Haiku.
We found that the two Si dimers neighbouring the nanoline show enhancement
at low bias voltages, and we confirmed that the nanoline is darker than the
surrounding surface for biases of less than 0.8V. We have also presented first
principles electronic structure calculations of the Haiku structure with both a
clean and a hydrogenated Si(001) surface, and we have explained the voltage
8

Page 9

contrast of the nanolines using these calculations.
The localisation of the states near the Fermi level seen in the projected charge
densities suggests that the Si atoms immediately surrounding the Haiku struc-
ture are strained (in part at least due to the interaction with the buckled Si
dimers), while the Si substructure immediately beneath the Haiku itself is
somewhat relaxed (though the hydrogenated surface plots suggest that it is
slightly strained relative to perfect bulk Si). This goes some way to explaining
the stability of the subsurface 5-7-5 ring structures, which serve as a highly
effective relief mechanism for the epitaxial stress exerted by adsorbed Bi. The
absence of states close to the Fermi level suggests that the nanolines are likely
to block surface conduction perpendicular to the nanoline, and are not likely
to act as a nanowire.
The Bi nanolines are clearly not suitable for conduction on their own. However,
as we have shown before[12], hydrogen will adsorb preferentially on the Si(001)
surface and not on the Bi, giving a natural, automatic masking technique. This
should allow us to adsorb metals on the nanolines, creating exceptionally high
quality nanowires. This research is under way, and will be presented in future
work.
Acknowledgements
DRB thanks the Royal Society for funding through a University Research Fel-
lowship. Calculations were performed at the HiPerSPACE Centre at UCL
(JREI grant JR98UCGI). JHGO was supported by the Japanese Science
and Technology Agency (JST) as an STA Fellow. This study was performed
through Special Coordination Funds of the Ministry of Education, Culture,
Sports, Science and Technology of the Japanese Government (Research Project
on active atom-wire interconnects). We would like to thank Bill McMahon for
bringing to our attention many of the examples of 5-7-5 structures, and for
sharing unpublished STM data.
References
[1] Y.Huang, X.Duan, Y.Cui, L.J.Lauhon, K.-H.Kim, C.M.Lieber, Science 294
(2001) 1313.
[2] A.Bachtold, P.Hadley, T.Nakanishi, C.Dekker, Science 294 (2001) 1317.
[3] Y. Chen, D.A.A.Ohlberg, G. Medeiros-Ribeiro, Y.A.Chang, R.S.Williams,
Appl. Phys. Lett. 76 (2000) 4004.
9

Page 10

[4] K.Miki, J.H.G.Owen, D.R.Bowler, G.A.D.Briggs, K.Sakamoto, Surf. Sci. 421
(1999) 397.
[5] K.Miki, D.R.Bowler, J.H.G.Owen, G.A.D.Briggs, K.Sakamoto, Phys. Rev. B 59
(1999) 14868.
[6] J.H.G.Owen, K.Miki, H.Koh, H.W.Yeom, D.R.Bowler, Phys. Rev. Lett. 88
(2002) 226104.
[7] D.R.Bowler, J.H.G.Owen, J. Phys.:Condens. Matter 14 (2002) 6761.
[8] K.Miki, K.Sakamoto, T. Sakamoto, Surf. Sci. 406 (1998) 312.
[9] Y.Wang, J.P.Perdew, Phys. Rev. B 44 (1991) 13298.
[10] J.P.Perdew, J.A.Chevary, S.H.Vosko, K.A.Jackson, M.R.Pederson, D.J.Singh,
C.Fiolhais, Phys. Rev. B 46 (1992) 6671.
[11] G.Kresse, J.Furthm¨ uller, Comp. Mat. Sci. 6 (1996) 15.
[12] J.H.G.Owen, D.R.Bowler, K.Miki, Surf. Sci. Lett. 499 (2002) L124.
[13] J.H.G.Owen, D.R.Bowler, C.M.Goringe, K.Miki, G.A.D.Briggs, Surf. Sci. Lett.
341 (1995) L1042.
10