Transport measurements of single wall carbon nanotube multiterminal devices with normal and ferromagnetic contacts

Spin based electronics or spintronics is a field having the electron's spin
degree of freedom as a subject. It is about how to write, transfer and read
information using the electron spin. The birth of spintronics is considered
to be the discovery of the giant magnetoresistance (GMR) in 1988 [1] and
since then a major progress has been achieved in the field [2, 3]. The best
example of this progress is the development of so called spin-valves. Modern
day spin-valves are based on the GMR and they are used for measuring small
magnetic fields. Their most common application is as sensors in hard disk
reading heads.
Spintronics can conceptually be divided in two parts. The first one is
about generating and detecting spin polarized electrons, which is normally
done using ferromagnetic materials, but can also be done using optical methods
[3]. The latter part is about coherent transfer of spin information. It
is of fundamental importance to understand how spin infomation can be
transfered coherently over larger distances.
In recent years new nanoscale allotropes of carbon have been discovered.
In 1985 the first fullerene, the buckyball was discovered [4] and 1991 carbon
nanotubes (CNT) were discovered by Sumio Iijima [5]. CNTs behave as onedimensional
conductors and the coherence length of the electron in them is
very long, especially in individual SWCNT, where the electrons have been
found to be coherent over the distance of 3 �m [6]. Moreover, carbon is believed
to have long spin coherence length, due to low spin orbit coupling and
no nuclear spin of its main isotope 12C . This all makes CNTs an interesting
platform for spin transport studies.
The first work on CNT spin-valve devices was done on multiwall carbon
nanotubes (MWCNTs) contacted by Co electrodes [7]. By applying magnetic
field to the device the magnetization of the Co electrodes can be changed between
parallel and antiparallel mutual orientation. The resistance for parallel
and antiparallel mutual orientation, RP and RA respectively, are measured
and the TMR, which is defined as follows
TMR = (RA - RP)/
is calculated. The TMR of this first CNT spin-valve was 9% at maximum
and it was positive (i.e RA > RP ) [7, 8].
Negative TMR signal was later measured in similar devices, i.e. MWCNTs
contacted with Co electrodes. The maximal size of the TMR signal in these
devices was 36% for a low current bias, but higher current bias resulted in
lower TMR signals [9]. The origin of the di�erent sign of the TMR was not
clear by then.
The first CNT spin devices fabricated in our lab wereMWCNTs contacted
by Pd1-xNix (x ~ 0:7) 1. These ferromagnetic contacts were transparent,
having room temperature resistance of 5:6 k[omega]. What was new about these
devices was that they were equipped with a back gate and could be tuned
between di�erent transport regimes [10]. More importantly it was shown that
TMR was dependent on the back gate voltage [11]. Further studies revealed
that the TMR signal was either negative or positive dependent on applied
gate voltage, but the origin of this behavior was not well understood [12].
When the signal changes in TMR were studied single wall carbon nanotubes
(SWCNT) grown in-house by chemical vapor deposition (CVD) using
methane as a carbon source became available. The CVD growing process
had been optimize to produce individual SWCNT [13]. Individual CVD
grown SWCNTs were connected with PdNi contacts. In such device it was
shown that the TMR signal was correlated with the coulomb oscillations of
the quantum dot which is formed in the SWCNT between the contacts. In
SWCNT the quantum dot behavior is much simpler than in MWCNT and
the TMR could be tuned smoothly from positive to negative values by the
gate voltage [12, 14]. This work demonstrated for the first time the control
of spin transport in a three terminal device.
There are still many open questions concerning SWCNT spin devices.
There are mainly two issues that one should be concerned about when constructing
a SWCNT spin valve device. The first one is the switching characteristics
of the electrodes. The switching in the devices contacted with PdNi
contacts is not always clear indicating that the electrode consists of many
magnetic domains.
The latter one is due to spurious effects in the SWCNT spin-valves. Such
effects could be magneto-coulomb effect [15] or tunnelling anisotropic magnetoresistance
(TAMR). Spurious effects could cause a "false TMR signal",
i.e. a switching behavior in the signal as a function of applied field that that
does not originate from transport of spin.
The focus of the this work was mainly to address these issues but some
work was also done on how to process of individual SWCNT devices. PdNi
electrodes were studied in order to understand their switching behavior better.
We worked to optimize the switching characteristics of the spin-valve
devices, by trying other contact materials on the SWCNTs.
One way of avoiding spurious e�ects is to make multi-terminal devices.
It has been shown in metallic nanostructures that by measuring non-local
spin signals, artefacts can be avoided. Non-local spin transport measurements
have been done on SWCNT contacted by four Co contacts [16]. The
multiterminal devices made in this work have two normal contacts and two
ferromagnetic contacts. They are gateable with a back-gate enabling it to
study the behavior of the three quantum dots that are formed in each segment
of the tube between the contacts.
Outline of this thesis
- Chapter 2 is on the basics of spintronics. It includes a short description
on ferromagnetism and on anisotropic magnetoresistance (AMR)
and for historical resons giant magnetoresistance (GMR) is briefly discussed.
The tunnelling magnetoresistance is explained and Julliére's
model.
- Chapter 3 is on carbon nanotubes. It is focused on single wall carbon
nanotubes (SWCNT), their structure and their electronic properties.
- Chapter 4 is on processing of SWCNT devices. The first part of the
chapter is on SWCNT production and characterization of the SWCNT
material. A lot of time was invested in the lab in finding the best way
to obtain individual SWCNT for our nanotube project. Both main
approaches tested, i.e spreading tubes from suspension solution and
CVD growth are described. In the latter part it is generally described
how to make SWCNT devices.
- Chapter 5 is on SWCNT based spin valves. The idea behind the
SWCNT is discussed (the statement of the problem) and then measurements
using different ferromagnetic contact materials are discussed.
Temperature dependence on TMR in SWCNT is discussed in the last
section of the chapter.
- Chapter 6 is on measurements on multiterminal devices. Non-local
and semi-nonlocal measurements are shown and discussed.
- Chapter 7 is a summary of the thesis.
Details on experimental setups and recipes can be found in appendices.