Text preview for : 2656 Nanoscale2.pdf part of Keithley 2656 Nanoscale2 Keithley Appnotes 2656 Nanoscale2.pdf
Back to : 2656 Nanoscale2.pdf | Home
A G R E AT E R M E A S U R E O F C O N F I D E N C E tally different from the bulk material. This
dictates the use of non-visual measurement
methodologies to uncover unique chemical
and electrical properties. For many of these
properties, the actual electrical quantity be-
ing measured is a low level current or voltage
that is translated to another physical quan-
tity. Direct electrical measurements are pos-
sible on many nanoscopic substances with
probing instruments and nano-manipulators
now available.
Electrical Property Measurements
If a particle becomes small enough, its
physical size may approach the wavelength
Nanoscale Device &
of the material's electrons. Because of quan-
tum mechanical effects, the energy of its
electrons cannot be predicted by the bonding
Material Electrical
normally associated with the bulk material.
For bulk macroscopic materials, electrons
have thermal energies that lie within contin-
Measurements uous energy bands. For nanoscale particles,
the allowable energies within continuous
bands can separate into discrete levels when
the separation between levels approaches
the thermal energy of the electrons. As this
Jonathan tucker happens, the density of states of the material
Keithley instruments, inc. changes. The density of states is a measure
of the number of energy options available
to an electron as it falls into a lower energy
Electricalmeasurementsonnanoscopicma- Nanoparticle characterization level by giving up energy, or as it ascends to
terialsanddevicesareessentialforthede- Methodologies a higher energy level after absorbing energy.
velopmentofpracticalproducts,eventhose As a result of small particle sizes, the at- Since the density of states can be used to ma-
notintendedforelectronicapplications.Us- oms and molecules of nanoscale materials nipulate material properties, its characteriza-
ingtherightinstrumentsandtechniquescan often bond differently than they do in bulk tion is a fundamental research activity.
shortentesttimesandhelpassurecollection substances. While the discovery of bulk Electron energy effects can be deduced
ofusefuldata. properties remains important, measurements from electrical measurements. One example
E
are needed to uncover quantum mechanic is when a nanoscale substance is involved in
lEctrical measurements on na- characteristics that are unique to nanoscale an oxidation-reduction (REDOX) reaction,
noscale materials and devices re- structures. such as the chemical-electrical conversion
veal not only electronic character- Particle size and structure have a major
istics, but also general properties influence on the type of measurement tech-
like a nanoscopic particle's den- nique used to investigate a material. Optical
sity of states. These fundamental properties microscopic techniques have limited value
can be used to predict and manipulate physi- for nanoscale materials. As particles shrink
cal characteristics, such as tensile strength, below micrometer sizes (referred to as me-
color, and thermal conductivity. However, soscopic), visual characterization can be
making meaningful measurements requires done with a scanning electron microscope
highly sensitive instruments and sophisti- (SEM). For nanoscopic materials (particle
cated probing techniques. Instrumentation sizes below100 nanometers), a scanning tun-
designed specifically for nanotechnology neling microscope (STM) can be used. Even
research is increasing, but users must under- smaller particles can be investigated with an Figure 1. Representation of a carbon nano-
stand the types of measurement needed, and Atomic Force Microscope (AFM). tube. These structures exhibit a wide range of
characteristics, giving them unique properties
test system features that facilitate speed and When a particle has nanoscale dimen- that are useful in many types of electronic and
accuracy. sions, its physical behavior is fundamen- physical structures.
Nanoscale Device & Material Electrical Measurements September2005
that takes place in fuel cells or batteries. sorbed (> bandgap) then electrons can jump Differential conductance is simply (di/
Electrical measurements of the number of from the valence band into the conduction dv)/(i/v). The quiescent current vs. voltage
electrons transferred from one species to an- band.) (I-V) characteristics are established through
other determine the reaction rate by tracking the STM's high resistance contact, with a
electrical current and potential with time. Density of States Measurements low level AC modulation on top of the quies-
These measurements can be used to infer Density of states corresponds to the den- cent operating point to measure di/dv. This
particle size, density of states, and other na- sity of a material's energy levels. Highly con- is divided by the quiescent conductance, I/V,
noscopic properties. ductive materials possess a greater density of and plotted against applied voltage.
One of the important properties is the states because of an abundance of free en-
mean free path of an electron (distance trav- ergy levels in the conduction band (i.e., more Other Means of Direct Electrical
eled before it bumps into another atom), allowed energy levels per unit of energy). In- Measurements
which approaches the same order of magni- sulating materials have an electronic struc- For reasons of cost, convenience, and
tude as a nanoscopic particle diameter. This ture with a scarcity of energy levels in the speed, alternatives to the STM are desir-
characteristic affects the material's bandgap conduction band. able for direct electrical measurements. An
and DC resistance. More generally, it de- The three dimensional density of states STM and its high resistance contact can be
termines whether a particle is a conductor as a function of energy can be expressed as: replaced with a nano-manipulator that cre-
(bandgap < thermal energy of the electron), ates a low resistance contact to the nanopar-
an insulator (bandgap > thermal energy of ticle. Nanomanipulators, such as the one
the electron), or a semiconductor. Further- shown in Figure 3a, have as many as four
more, this characteristic can be altered dy- In this equation the quantity r(E) is ex- positioners that grasp, move, and optimally
namically. pressed as the derivative of ns, the density of position a nanoscale sample along four axes.
An example of this is found in carbon states per unit volume with respect to energy, This permits simultaneous manipulation,
nanotubes (CNTs). (See Figure 1.) Typi- E. Thus, r(E) represents the number of elec- imaging and electrical probing of the sample
cally, when CNTs are made, both conduct- tron states per unit volume per unit energy at (Figure 3b).
ing and semi conducting forms occur. When energy E (electron orbital location expressed Because of the complexity involved in
the two forms are separated, the conducting in electron volts). In the expanded equation, connecting individual instruments to nano-
nanotubes can be used, for instance, as field m represents the effective mass of the par- manipulators, it is best to use an integrated
emission display emitters. Semiconducting ticle, and h is the Plank constant. source-measure system with a suitable inter-
nanotubes can be used to make transistor While the result is independent of vol- face and application software. The source-
switches. This is illustrated in Figure 2a, ume (can be applied to any size particle), measure units (SMUs) in these systems have
where a semi-conducting CNT is connected this equation is of limited value if the par- the added advantage of being able to dynam-
between two electrodes that function as a ticle size/structure is unknown. However, ically alter their measurement mode to adapt
drain and source. A third electrode, an in- there are other ways to determine density of to the impedance state of a nanoscale mate-
sulated gate (Figure 2b), is placed directly states experimentally. X-ray spectroscopy rial, which can range from highly conductive
under the entire length of the CNT channel. is frequently used, but a material's density to highly resistive in the case of CNTs.
The introduction of an electric field through of states can also be deduced from electri- There are two possible measurement
the channel (by increasing the voltage on the cal impedance and conductance measure- modes for an SMU: source current/measure
gate) can change the CNT from its semi-con- ments. Prior art has used a scanning tunnel- voltage, or source voltage/measure current.
ducting state to its insulating state. Decreas- ing microscope (STM) to tunnel a current When considering the measurement of low
ing the gate voltage will transition the device through a nanoscopic device. The density of impedance materials and devices (less than
into a conducting state. states is found through differential conduc- 1000 ohms), the source current/measure
(If a suitable amount of energy is ab- tance measurements. voltage technique generally yields the best
results. When measuring high impedance
(greater than 100,000 ohms), the source
voltage/measure current technique is best.
The SMU can switch modes automatically
Nanotube
Source (Au) Drain (Au) as a material's conductive state changes, and
Gate Oxide (SiO2)
measurement resolution can be as good as
femtoamps and nanovolts.
Gate (Si)
Specialized SMU systems are available
with software written specifically for nano-
scale testing. This shortcuts many measure-
ment tasks by providing common routines
(a) (b) for collecting electrical data on a nanotech
Figure 2. A CNT being used to create a new type of transistor switch. (Courtesy of IBM Corporation) device, such as a CNT, bio-device, molecular
September2005 Nanoscale Device & Material Electrical Measurements
ing in increased contact resistance and measurement errors. The best
way to enhance long-term performance of probe tips is to incorpo-
rate periodic cleaning procedures in the test protocol. Some auto-
mated test systems have software that includes probe maintenance
routines.
Probing any nanomaterial or device requires care to avoid non-
ohmic contacts. Non-ohmic contacts create a potential difference
that is not linearly proportional to the current flowing through them.
A typical method for determining ohmic contact on the DUT is to
perform an I-V sweep with the SMU and verifying that it crosses
(a) (b) through zero. If the IV curve does not cross through zero, then ohmic
Figure 3. Nanomanipulator for conducting direct electrical measure-
ments on nanoscale structures. (Courtesy of Zyvex Corporation) contact is highly unlikely. Another method is to change measurement
ranges. Changing ranges, especially when measuring resistance,
electronic component, or a nanowire. Typically, these routines take can change the test currents. Ohmic contact would be indicative of
measurements, plot I-V curves, and have the ability to make differen- the same reading but with higher or lower resolution depending on
tial conductance measurements for determining density of states. whether the range went up or down. Different readings on different
ranges may indicate non-ohmic contact.
connection and accuracy issues Figure 4a illustrates a nanomanipulator making a four-wire con-
A major issue in nanoscale electrical measurements is making nection to a CNT `wire'. Upper and lower probes are used to inject
reliable connections at the right location. At the nanoscopic level, a current through the CNT, while the left and right probes measure
it may be necessary to connect the device under test (DUT) back voltage across a segment of it. Note that the resulting I-V sweep (Fig-
to pads that can be reliably probed. One example of this is particle ure 4b) does cross zero, indicating ohmic contact.
self-assembly from silicon to silicon, where conventional photolitho- Another source of error is self-heating due to excessive electrical
graphic techniques are used to make electrical connection pads for current through the DUT. Such currents may even lead to catastrophic
probing. Particles that are long enough to straddle such pads (for failure of the sample. Therefore, instrumentation must automatically
example, carbon nanowires) can be connected to the pads through limit source current during device testing. Programmable current
externally generated electrostatic fields. and voltage compliance circuits are a standard feature of most SMUs.
In any case, connections to the DUT must not affect measure- In some systems, pulsed current sources are available, which may be
ment accuracy. This is particularly important in low resistance meas- required to avoid self-heating of some low resistance structures.
urements on nanowires and sheet resistivity measurements on films. For high resistance applications, the DUT stimulus typically is a
Typically, low resistance measurements require a four-point probe voltage, and the response current is measured, which can be as low as
(Kelvin) technique to eliminate the effects of lead resistance and en- a few femtoamps. Therefore, instrumentation must provide this level
sure accuracy. The two most commonly used four-point techniques of sensitivity and adequate resolution.
for sheet resistivity are the collinear probe method and the van der Regardless of measurement mode, external sources of error must
Pauw method [1]. SMU-based test systems may include these test be minimized. These errors can arise from stray magnetic fields,
routines and associated calculations in their application software. electrostatic charges, cable connections, thermoelectric EMFs, and
Test signal integrity depends on a high quality probe contact, currents generated by triboelectric and electrochemical effects. To
which is directly related to contact resistance. During the course of protect nanoscale samples from electrostatic charge and magnetic
their use, probe needles wear and may become contaminated, result- fields, as well as maintaining the integrity of the measurement, a
3.00E