Jonathan Tucker from Keithley Instruments looks at the role of pulse mode measurement techniques for testing in the nanodevices sector

Characterising nanodevices like carbon nanotubes (CNTs) and single electron transistors (SETs) presents special challenges in R&D settings. Testing these kinds of devices in a manner that will not introduce measurement errors and characterising their electrical performance without damaging or destroying them can be especially challenging.

Pulse mode measurement techniques offer the best alternative to traditional DC current-voltage (I-V) characterisation because they prevent the joule heating that might otherwise damage or destroy the device under test (DUT) or skew the results, making them meaningless.

At lower levels of joule heating, the temperature change can cause the device to behave uncharacteristically, while at higher levels, it can vaporise experimental, one-of-a-kind devices before they can be measured. In contrast, pulse mode measurement techniques use short duration pulses that minimise the power dissipated within a device, which in turn minimises joule heating.

To characterise the performance of experimental nanodevices, researchers need to be able to make a variety of measurements, such as: current density; differential conductance; conductivity; low and high resistance; and other measures of low power phenomena.

SEM photo of a carbon nanotube attached to a nanomanipulator's probes

To handle this range of requirements, researchers must choose instrumentation with the flexibility to handle a variety of measurement challenges and to adapt readily to evolving test needs.

Over the years, various instruments have been used to produce pulse mode measurements, with varying degrees of success:

• A standard low-level current source used in combination with a nanovoltmeter.
  This approach has the advantage that most labs probably already have these instruments on hand, but standard current sources do not typically have built-in pulse generation functions, so, a separate pulse generator would be required, complicating test set-up and trigger timing, and increasing the system cost and complexity.
• A current source with high speed pulsing capability plus a nanovoltmeter with high precision trigger timing.
  This approach has various advantages. For example, it provides higher sourcing accuracy and lower noise measurements than the standard current source/nanovoltmeter test setup. Pulse timing control is built-in so complete pulse-and-measure cycles as short as 50µs can be performed automatically. Just as important, the timing of sourcing and measurement is co-ordinated seamlessly. The use of a nanovoltmeter provides superior voltage measurement resolution and accuracy, allowing researchers to make low noise measurements at relatively high speeds. This configuration also supports measuring differential conductance at high speeds.
  Unfortunately these types of test set-ups are best suited for testing individual devices. They do not provide device fixturing or the automation capabilities needed to test multiple DUTs quickly and consistently, so users must create their own device handling subsystems, adding to the total system cost and extending the system configuration/implementation period required before testing can begin.
• An integrated characterisation system.
  Fortunately for nano­device designers and manufacturers, low level measurement instrumentation suppliers have recognised the need for integrated characterisation systems with built-in pulse generation and pulse measurement capabilities. Some of these systems are even compatible with commercial fixturing specifically designed for making connections to nanodevices.

The advantages of some of these types of systems include:

• Tightly controlled pulse sourcing and measurement.
  When configured with pulse sourcing and measurement capabilities, a semiconductor characterisation system can be programmed to source pulse widths from 10ns to near DC over two independent channels and measure pulses at up to 1.25Gsample/s per channel.
• High accuracy.
  Modern Source-Measure Units provide 100µV and 10fA (10^-15) measurement accuracy. When equipped with preamplifiers, they allow measuring currents as low as 100aA (10^-18) with confidence.
• A graphical user interface (GUI).
  Even novice users can start acquiring data on experimental devices almost immediately when using a semiconductor characterisation system configured with an intuitive Windows-based GUI.
• Easy connections.
  The newest systems make it simple to connect directly to nanoscale DUTs or through a variety of probe stations, transmission electron microscopes (TEMs), scanning electron microscopes (SEMs), and nanomanipulators/probers.
• The convenience of application-specific tools.
  Add-on nanotechnology toolkits are now available that provide interactive test modules (ITMs) for characterising the most common nanodevice structures: carbon nanotubes, bio-components, carbon nanotube Fets, nanowires, molecular wires, molecular transistors, and multi-pin nanodevices.

No matter which type of instrumentation is used, characterising nanodevices accurately also demands taking other factors into account, including:

• Proper device fixturing.
  Nanomanipulators/probers are positioning and testing tools that accommodate up to four positioners (3D stages) that grasp, move, test, and optimally position micro- and nano-scale samples in SEMs and focused ion beam systems (FIBS). When used with the latest semiconductor characterisation systems, they offer 5nm movement precision with probe tip diameters of less than 20nm and current measuring capability better than 1pA.
• Use of low noise cabling.
  Triboelectric currents are generated in coaxial cables by charges created between a conductor and an insulator due to friction, a result of cables that are flexed, moved or vibrated. Using low noise cable and taking steps to eliminate or minimise cable movement greatly reduces this effect.
• Use of four-wire resistance measurement techniques.
  When the resistance to be measured is relatively low, four-point (or Kelvin) measurements will yield greater accuracy than two-wire measurement techniques, which measure the test lead and contact resistance along with that of the DUT. With Kelvin measurements, a second set of probes is used for sensing. Negligible current flows in these probes, so only the voltage drop across the DUT is measured, making resistance measurements more accurate.

While there are currently no industry standards associated with nanotech device characterisation, industry organisation the IEEE Nanotechnology Council is striving to develop them. When those standards emerge, they are likely to drive the next wave of development of nanotechnology test system design.

Jonathan Tucker is the lead industry consultant for nanotechnology at Keithley Instruments

www.keithley.com