See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/260077300

Ultrahigh vacuum scanning tunneling microscope manipulation of single gold

nanoislands on MoS2 for constructing planar nanointerconnects

Article in Journal of vacuum science & technology B · September 2007

DOI: 10.1116/1.2781573

CITATIONS READS

19 134

4 authors, including:

JianShu Yang
Paul Drude Institute for Solid State Electronics
37 PUBLICATIONS 593 CITATIONS

SEE PROFILE

All content following this page was uploaded by JianShu Yang on 07 February 2014.

The user has requested enhancement of the downloaded file.

 Ultrahigh vacuum scanning tunneling microscope manipulation of single gold
nanoislands on Mo S 2 for constructing planar nanointerconnects
Jianshu Yang, Jie Deng, N. Chandrasekhar, and C. Joachim

Citation: Journal of Vacuum Science & Technology B 25, 1694 (2007); doi: 10.1116/1.2781573
View online: http://dx.doi.org/10.1116/1.2781573
View Table of Contents: http://scitation.aip.org/content/avs/journal/jvstb/25/5?ver=pdfcov
Published by the AVS: Science & Technology of Materials, Interfaces, and Processing

Redistribution subject to AVS license or copyright; see http://scitation.aip.org/termsconditions. Download to IP: 193.49.32.253 On: Fri, 07 Feb 2014 13:53:51
 Ultrahigh vacuum scanning tunneling microscope manipulation of single
gold nanoislands on MoS2 for constructing planar nanointerconnects
Jianshu Yang,a兲 Jie Deng, and N. Chandrasekhar
IMRE, A*STAR, 3 Research Link, Singapore 117602, Singapore
C. Joachim
IMRE, A*STAR, 3 Research Link, Singapore 117602, Singapore
and Nanoscience Group, CEMES-CNRS, 29 Rue J. Marving, BP 94347, 31055 Toulouse Cedex, France
共Received 18 April 2007; accepted 13 August 2007; published 18 September 2007兲
The manipulation of single metallic nanoislands with a precision better than 0.5 nm on a MoS2
surface is demonstrated. Optimizing the metal growth conditions yields triangular-shaped
nanoislands of 30 nm in lateral size and 12 nm in height on the MoS2 surface. The manipulation of
a single nanoisland was performed using the tip apex of a scanning tunneling microscope. The
feedback loop conditions to achieve this manipulation are discussed. Fully planar four-pad
nanostructures were constructed, and the apex of each triangular nanoisland of the nanostructure is
pointing toward a central 10⫻ 20 nm2 MoS2 working area where the surface atomic cleanliness is
preserved. © 2007 American Vacuum Society. 关DOI: 10.1116/1.2781573兴

I. INTRODUCTION preparing identical metal nanoislands and have selected for

The measurement of current-voltage characteristics on a
single atomic wire1 or a single molecule2 demands an atomic
precision of the interconnection between the atomic wire 共or
molecule兲 ends and the surface atoms of the contacting
pads.3,4 This is not achievable with modern e-beam
nanolithography5 or with the nanostencil technique.6 Both
are limited to a lateral precision of about a nanometer and do
not preserve the atomic cleanness of the surface supporting
the atomic wire or molecule. In a planar configuration, the
precise positioning of ultraflat metallic nanoislands absorbed
at the surface of a semiconductor opens the way to construct
artificial nanostructures for atomic-scale interconnects, fabri-
cating the surface atomic wire by vertical scanning tunneling
microscope atom manipulation.7,8 Without damage, and due
to the soft tip approach that the scanning tunneling micro-
scope 共STM兲 feedback loop system is permitting, each
nanoisland of the nanostructure can be electrically contacted
from the top using an ultrasharp tip positioned under a scan-
ning electron microscope 共SEM兲 for precise navigation of
each tip on the surface.9
To date single adsorbate manipulation with the tip apex of
a STM has been performed with atomic-scale precision on
single atoms10 and single molecules.11 Manipulation of larger
objects containing up to a few 106 atoms such as metallic
nanoislands can be performed, but not with such lateral
precision.12–17 In many cases,13–17 it is difficult to access the
full range of STM parameters governing the manipulation of
such large objects because they are not identical on the sur-
face as compared to a given atom or molecule. This intro-
duces dispersion in the manipulation conditions that pre-
cludes any systematic studies of the adsorbed manipulation
conditions. In the work below, we have taken extreme care in
a兲
Electronic mail: js-yang@imre.a-star.edu.sg
STM manipulations only those of identical lateral size,
shape, and height.
To construct a planar multipad metallic nanostructure, we
start from the self-assembly of metallic gold nanoislands on
a MoS2 surface.12 After careful control of the growth condi-
tions to obtain flat triangular Au nanoislands, we demonstrate
the manipulation of single nanoislands on the MoS2 surface
by the STM tip with a precision better than 0.5 nm. The
triangle shape of each nanoisland is important to control for
a very small lateral extension of the surface in the center of
the constructed nanostructures delimited by the tip of each
nanoisland. The experimental tunneling current and feedback
loop conditions for this manipulation are detailed. As an ap-
plication, the construction of a four-metallic-pad nanostruc-
ture is presented. The structure consists of four triangular
nanoislands pointing toward a 10⫻ 20 nm2 atomically clean
and flat MoS2 working area.
The STM images and the nanoisland manipulation were
carried out at room temperature using an Omicron ultrahigh
vacuum 共UHV兲-STM with a base pressure of 3 ⫻ 10−8 Pa.
The STM tip was made by chemically etching a
0.25-mm-diameter tungsten wire in 2M NaOH. Freshly
etched tips were transferred into the vacuum chamber and
cleaned by electron bombardment. The typical radius of cur-
vature of the tip apex is 10 nm, as measured using a trans-
mission electron microscope.
II. SAMPLE PREPARATION
The 1 cm2 MoS2 wafer was fabricated from bulk molyb-
denite extracted in Australia. The MoS2 substrate was freshly
cleaved from this wafer. Prior to the nanoisland growth, the
surface was degassed at 400 ° C for 3 h to remove surface
impurities. The gold nanoislands were prepared by thermal
evaporation at 400 ° C with a deposition rate of 0.02 nm/ s.
The total thickness of deposited gold was 1 nm, as measured

1694 J. Vac. Sci. Technol. B 25„5…, Sep/Oct 2007 1071-1023/2007/25„5…/1694/6/$23.00 ©2007 American Vacuum Society 1694

Redistribution subject to AVS license or copyright; see http://scitation.aip.org/termsconditions. Download to IP: 193.49.32.253 On: Fri, 07 Feb 2014 13:53:51
 1695 Yang et al.: Ultrahigh vacuum scanning tunneling microscope manipulation
FIG. 1. 共a兲 Constant-current STM image of the MoS2 surface showing a
majority of 30 nm lateral-size Au nanoislands recorded at I = 20 pA and V
= 1.0 V for G = 24%. To avoid the multiple-tip effect, the tip was cleaned by
pulsing the bias voltage up to 3 V before imaging. 共b兲 Constant-current
image on a 22 nm lateral-size nanoisland and 共c兲 on a 30 nm nanoisland
with the same scanning conditions as in 共a兲.
by a quartz microbalance. After the gold deposition, the
sample was maintained at 400 ° C for another 1 h to facili-
tate the self-assembly of crystalline Au nanoislands on the
MoS2 surface. The nanoisland growth conditions were opti-
mized to achieve a majority of almost equilateral triangular
nanoislands that measure 30 nm on a side. A small number
of 22 nm triangles 共⬍3 % 兲 were still found on the MoS2
surface by STM. The full process was optimized by imaging
each batch using a 10 keV SEM equipped with a field-
emission gun SEM. At optimized conditions, each of the
triangular nanoislands has a sharp apex with radii of curva-
ture measuring 4.5 and 3.3 nm for the 30 and 22 nm lateral-
size nanoislands, respectively 共see Fig. 1兲.
On the MoS2 surface, the self-assembled Au nanoislands
are STM imaged at a very low tunneling current intensity I,
typically lower than 20 pA, with a bias voltage V of 1 V, as
presented in Fig. 1共a兲. When the tip apex is clean enough to
preserve the atomic resolution of the MoS2 surface and when
it does not pick up any of the residual small nanoislands
共lateral size typically ⬃5 nm兲 while scanning, the STM im-
ages are stable over night and are identical to the SEM im-
ages of the same surface as presented in Fig. 6. On the MoS2
surface, the atomic resolution is preserved around the nanois-
JVST B - Microelectronics and Nanometer Structures
Redistribution subject to AVS license or copyright; see http://scitation.aip.org/termsconditions. Download to IP: 193.49.32.253 On: Fri, 07 Feb 2014 13:53:51
1695
land, showing the absence of a wetting layer around those
nanoislands. Figures 1共b兲 and 1共c兲 present the STM images
of two nanoislands with lateral dimensions of 30 and 22 nm,
with an apparent heights of 12 and 9 nm, respectively.
III. STM MANIPULATION OF A SINGLE AU
NANOISLAND
As compared to the imaging conditions given above, in-
creasing the tunneling current intensity I by two orders of
magnitude caused the brushing and subsequent removal of
the Au nanoislands from the MoS2 surface in the area
scanned by the STM tip. This phenomenon has been reported
by several groups for different nanoclusters and surface
materials.12–17 It corresponds to the fact that by increasing I,
the tip apex end atom is very close to the surface/island in a
regime where the van der Waals interaction dominates. In
this case, other lateral atomic-scale protuberances of the tip
apex can also interact with the adsorbed nanoislands, and
there is no control over the direction of the interactions be-
tween the tip apex and nanoislands. Therefore, upon scan-
ning, nanoislands are manipulated at many locations at the
same time. Between this brushing mode and the stable-image
condition, there is a large range of possible STM feedback
loop conditions to fine tune the magnitude of the force acting
on a single nanoisland. To our knowledge, this fine tuning
has not been explored for nanoclusters on semiconductor sur-
faces with the prospect of manipulating only one cluster at a
time with a positioning precision better than 0.5 nm.
To determine the threshold current and feedback loop gain
共G兲 to manipulate a single nanoisland on MoS2, the follow-
ing experimental sequence was followed. First, well-isolated
triangular nanoislands, 30 nm in lateral size, were identified
and imaged at a very low I with large STM feedback loop
gain G. All manipulations were performed on 30 nm nanois-
lands having the same lateral size and apparent height. Sec-
ond, the STM was switched from the imaging mode to a
single scan mode with a scan always generated at the same
predetermined location and unidirectional. During a single
scan, the STM tip passes over the nanoisland at a point in the
middle of a nanoisland facet, as presented in Fig. 2共a兲. We
call this the “point of localized interaction between the tip
and the island.” Third, up to a maximum of N = 700 identical
scan lines were performed exactly at this “local interaction
point,” while carefully noting any resultant displacement of
the targeted nanoisland. To minimize possible deformation of
the nanoisland facet at the impact location, all the scans were
recorded in a constant-current mode. For V = 1 V and below
I = 20 pA, as many as 700 scans performed during a time
interval of 180 s do not affect the position of the nanoisland,
even at low G.
To observe the first occurrence of nanoisland manipula-
tion by the STM tip, I and G were systematically varied for
V = 1 V starting from I = 20 pA and a large G. The experi-
mental procedure is illustrated in Fig. 2共b兲 for I = 80 pA and
G = 50%. Here, the nanoisland starts to move after N = 38
scans. After N = 256 scans, the nanoisland had been moved
2.5 nm laterally, having been displaced from its initial posi-
 1696 Yang et al.: Ultrahigh vacuum scanning tunneling microscope manipulation
FIG. 2. Representation of the time succession of the N = 256 identical,
20 nm-long scan lines recorded from A to B in 180 s on a Fig. 1共c兲-like
nanoisland for 共a兲 I = 20 pA, V = 1.0 V, G = 24%, 共b兲 I = 80 pA, V = 1.0 V,
G = 50%, and 共c兲 I = 1 nA, V = 1.0 V, G = 50%. In 共b兲, the nanoisland starts to
move after N = 38 line scans. The N = 256 identical STM lines scans pre-
sented in 共a兲–共c兲 were all recorded at the same impact location on a Fig. 1共c兲
nanoisland.
tion by the STM tip. It should be noted that this observation
is not to be confused with the lateral drift of the scan due to
some instabilities of our STM. Its lateral drift was measured
to be one order of magnitude lower, that is, 2.7 nm/ h. The
256 scans were recorded in 180 s. During this time period
involving manipulation of this particular nanoisland, the ap-
parent drift was only 0.12 nm, recorded at I = 20 pA and V
J. Vac. Sci. Technol. B, Vol. 25, No. 5, Sep/Oct 2007
Redistribution subject to AVS license or copyright; see http://scitation.aip.org/termsconditions. Download to IP: 193.49.32.253 On: Fri, 07 Feb 2014 13:53:51
1696
FIG. 3. Statistics of the nanoisland manipulation on MoS2. 共a兲 Average num-
ber 具N典 of identical scan lines recorded at the same position and required to
manipulate one nanoisland as a function of the tunneling current I, for two
values of the feedback loop gain 共circles: G = 50%; triangle: G = 24%兲. 共b兲
Probability of manipulating a nanoisland after N identical scan lines as a
function of the tunnel current I for G = 50% 共square: N ⬍ 10; triangle: N
⬍ 100; lozenge: N ⬍ 200兲. No manipulation was observed below I = 20 pA,
V = 1.0 V in this constant-current mode.
= 1 V. In contrast to the smooth lateral shift seen in Fig. 2共b兲,
we find at I = 1 nA and V = 1 V that the nanoisland jumps by
steps of 2 to 5 nm in a stick-and-slip-like motion, as pre-
sented in Fig. 2共c兲.
After this initial determination of the threshold current
and feedback to move a nanoisland, we explored the full
range of feedback parameter variations with the objective of
speeding up the manipulation process and increasing the dis-
tance a nanoisland can be STM manipulated, as demon-
strated in Fig. 2共c兲, while retaining the manipulation preci-
sion, as demonstrated in Fig. 2共b兲. On the same impact point
and for the same tip apex, Fig. 3共a兲 indicates the average
number 具N典 of identical STM scan lines required to move a
nanoisland as a function of I for a fixed G. For a 100%
chance of success during a single STM line scan, the thresh-
old current must be larger than 0.1 nA for V = 1 V. At large
gain and low current, the tip apex trajectory is avoiding the
nanoisland edge and the chance to manipulate it is very
small. Closing the loop results in a larger interaction between
 1697 Yang et al.: Ultrahigh vacuum scanning tunneling microscope manipulation 1697

the tip apex and the nanoisland edge. It should be mentioned

here that this threshold varies slightly from one tip to an-
other. A better statistical analysis of the manipulation phe-
nomenon is presented in Fig. 3共b兲, where the probability of
manipulating a nanoisland after a certain number of scans
repeated on the same impact point is given for as a function
of I. For I ⬎ 200 pA, a nanoisland can be manipulated with
100% success, irrespective of the feedback loop gain G. We
noticed that the magnitude and sign of the bias voltage had
no direct influence on the determined threshold for manipu-
lation of a nanoisland for V lower than 1 V. We therefore
decided to keep this low tip-surface bias voltage.
Simple investigations were performed such as bringing
two nanoislands close together to attempt coalescence, rotat-
ing a nanoisland by positioning the tip on an apex of the
triangular nanoisland, searching for MoS2 surface defects at
the initial position of a nanoisland, and manipulating a few
of the 22 nm lateral-size nanoislands. Our conclusion is that
a nanoisland is a solid object that is very stable because no
coalescence was observed. The 22 nm nanoisland can also be
manipulated, but at lower current. Although there are surface
atomic defects on natural MoS2 samples, they are not corre-
lated to the position of the nanoislands. Finally, at larger
current and identical to STM single-molecule
manipulation,11 Fig. 4 presents the STM image recorded dur-
ing manipulation of a single nanoisland in a pushing mode,
where the manipulation trace between the two stable posi-
tions of a single nanoisland has been observed.

IV. CONSTRUCTION OF A SINGLE

NANOSTRUCTURE
In Fig. 5, a sequence of images is presented demonstrat-
ing how four nanoislands can be manipulated to construct
fully planar metallic four-nanopad structures. The end apex
of each pad is pointing toward a central MoS2 surface area
where atomic corrugation is recovered. First, we identified
four nanoislands, A, B, C, and D, having the same apparent
lateral size and height and a suitable orientation to form the
four pads, as described below. Then, we identified the 15
nanoislands that need to be cleared out one by one from the
target area. Notice that there are only three possible direc-
tions for manipulating a nanoisland on the MoS2 surface due
to their triangular shape because the tip apex must scan to- FIG. 4. STM images recorded before, during, and after the pushing of a
ward a facet of the nanoisland. Therefore, the manipulation 30 nm nanoisland showing the clear trace of the nanoisland trapped under
the tip apex and scan along the surface during each scan in 共b兲. Scale bar:
direction is chosen by electronically rotating the scan direc- 30 nm.
tion so that it is perpendicular to the selected facet. The
consequence is that not all the nanoislands can be selected as
candidates for manipulation. But there are enough nanois-
lands on the surface that we have always found a way to nanoislands with distance less than 2 nm and a four-pointed
select the A, B, C, and D nanoislands. star using 30 nm nanoislands whose apices point outward.
Around the four pads, the cleaning creates a free area of After construction, such a nanostructure is stable over weeks
150⫻ 150 nm2, easily recognized by SEM 关see Fig. 6共b兲兴. and can be located on the MoS2 surface by SEM observa-
The closest distances between neighboring nanoislands in the tions after an appropriate marking using STM brushing-
four-nanopad structure are 3, 6, and 10 nm. The diagonals of mode manipulation to generate a simple bar code fabricated
the four-nanopad structure are 10 and 20 nm. Several other by sweeping out the nanoislands from large and square areas,
nanostructures can be constructed along similar lines. For large enough to be imaged easily by a SEM. Comparison
instance, we created a junction between the apex of two between STM and SEM images of another four-pad con-

JVST B - Microelectronics and Nanometer Structures

Redistribution subject to AVS license or copyright; see http://scitation.aip.org/termsconditions. Download to IP: 193.49.32.253 On: Fri, 07 Feb 2014 13:53:51
 1698 Yang et al.: Ultrahigh vacuum scanning tunneling microscope manipulation 1698

FIG. 5. Sequence of STM images taken during the construction of a four-

pads nanostructure on a MoS2 surface. The nanoislands A, B, C, and D are
first identified to be the part of a future four-pad nanostructure and all the
other 15 nanoislands to be pushed away from this nanostructure area. 共a兲
The surface after the Au nanoisland self-assembly process and identification.
共b兲–共f兲 are various stages of the construction where the 15 other nanoislands
have also been manipulated one by one, but out of the central area. At the
final stage, one apex of each triangle is pointing toward the center of the
nanostructure, where an atomically clean 12⫻ 24- nm2 MoS2 surface can be
recovered. Imaging conditions are as described in Fig. 1, and manipulation
conditions are as described in Fig. 2. Scale bar: 60 nm.
FIG. 6. Comparison between the UHV-STM and SEM images of a STM-
constructed four-pad nanostructure different from the one constructed in Fig.
5. On the MoS2 wafer, the navigation under the SEM to reach the exact
struction is presented in Fig. 6. Apart from the STM tip location of this four-pad structure results from marking the MoS2 surface
effect, which renders a nanoisland as having a larger appar- with the STM after completing the four-pad construction.
ent width in the STM images, all the nanoislands are found
at the same place after the transfer from the UHV-STM to the
SEM. the van der Waals interactions. In the first regime, the nanois-
land progressively charges up to a point where the electro-
V. CONCLUSION static repulsion is large enough to overcome the nanoisland
We demonstrated how single gold nanoislands of 30 nm surface adhesion. In the second regime, by increasing I and
in size and 12 nm in height can be manipulated at room lowering the gain, the repulsive interaction is not progres-
temperature with an UHV-STM, on a semiconductor MoS2, sive. When it exceeds the adhesion of the nanoisland, a stick-
and with an appropriate choice of STM feedback parameters and-slip motion is observed.18,19
reaching a precision better than 0.5 nm. We foresee that the precise manipulation of single ultraflat
A preliminary interpretation of the manipulation statistics metallic nanoislands on a semiconductor surface is a new
indicates two distinct interaction regimes. At low tunneling way of fabricating planar metallic contact pads to intercon-
current, the tip apex end atom Coulomb interactions with the nect an atomic wire or a molecule to macroscopic probes,
nanoisland dominates, and at large tunneling currents, it is preserving the atomic cleanliness of the surface. Commer-

J. Vac. Sci. Technol. B, Vol. 25, No. 5, Sep/Oct 2007

Redistribution subject to AVS license or copyright; see http://scitation.aip.org/termsconditions. Download to IP: 193.49.32.253 On: Fri, 07 Feb 2014 13:53:51
 1699 Yang et al.: Ultrahigh vacuum scanning tunneling microscope manipulation 1699

7
cially available UHV compatible multi-STM systems, where S. Hosoki, S. Hosaka, and T. Hasegawa, Appl. Surf. Sci. 60/61, 643
共1992兲.
each tip apex is positioned independently on a surface using 8
A. J. Mayne, D. Riedel, G. Comtet, and G. Dujardin, Prog. Surf. Sci. 81,
top SEM imaging,9 provide direct electrical access to the 1 共2006兲.
constructed four-pad nanostructures without a nanolithogra- 9
See, for example, the UHV-4 Probes system from Omicron
phy step. 共www.Omicron.de兲.
10
D. Eigler and E. Schweizer, Nature 共London兲 344, 524 共1990兲.
11
T. A. Jung, R. R. Schlittler, J. K. Gimzewski, H. Tang, and C. Joachim,
ACKNOWLEDGMENT Science 271, 181 共1996兲.
12
The authors would like to thank the A*STAR VIP pro- T. Ichinokawa, T. Ichinose, M. Tohyama, and H. Itoh, J. Vac. Sci. Tech-
nol. A 8, 500 共1990兲.
gram, “Atom Technology for Bottom-up Molecular Electron- 13
P. E. Sheehan and C. M. Lieber, Science 272, 1158 共1996兲.
ics Devices,” for financial support during this work. 14
C. Ritter, M. Heyde, B. Stegemann, K. Rademann, and U. D. Schwarz,
Phys. Rev. B 71, 085405 共2005兲.
1 15
A. Yazdani, D. M. Eigler, and N. D. Lang, Science 272, 1921 共1996兲. S. D. Sartale, K. L. Lin, C. I. Chiang, M. F. Luo, and C. C. Kuo, Appl.
2
C. Joachim, J. K. Gimzewski, R. R. Schlittler, and C. Chavy, Phys. Rev. Phys. Lett. 89, 063118 共2006兲.
16
Lett. 74, 2102 共1995兲. P. J. Durston, R. E. Palmer, and J. P. Wilcoxon, Appl. Phys. Lett. 72, 176
3
F. Moresco, L. Gross, M. Alemani, K.-H. Rieder, H. Tang, A. Gourdon, 共1998兲.
17
and C. Joachim, Phys. Rev. Lett. 91, 036601 共2001兲. K. H. Park, J. S. Ha, W. S. Yun, and E. H. Lee, J. Vac. Sci. Technol. A 17,
4
G. V. Nazin, X. H. Qiu, and W. Ho, Science 302, 77 共2003兲. 1441 共1999兲.
5 18
M. S. M. Saifullah, T. Ondarcuhu, D. F. Koltsov, C. Joachim, and M. Scanning Probe Microscopy and Spectroscopy: Theory, Techniques, and
Welland, Nanotechnology 13, 659 共2002兲. Applications, edited by D. Bonnell 共Wiley-VCH, New York, 2000兲.
6 19
M. M. Deshmukh, D. C. Ralph, M. Thomas, and J. Silcox, Appl. Phys. C. J. Chen, Introduction to Scanning Tunneling Microscopy, Oxford Se-
Lett. 75, 1631 共1999兲. ries in Optical and Imaging Sciences 4 共New York, 1993兲.

JVST B - Microelectronics and Nanometer Structures

Redistribution subject to AVS license or copyright; see http://scitation.aip.org/termsconditions. Download to IP: 193.49.32.253 On: Fri, 07 Feb 2014 13:53:51
View publication stats