Carbon Nanotube (CNT) Devices

Controlled Growth and Patterning of CNTs

To make useful arrays of carbon nanotube (CNT) devices, it is essential to pattern CNT thin films.  We have developed a number of methods that successfully allow controlled growth, patterning, and placement of CNT thin films for electronic applications.  This can be successfully done using either pre-growth or post-growth patterning techniques.Thelimitations of traditional post-growth patterning methods were circumvented by transfer printing CNTs selectively from the growth substrate to either thermoplastic (PET)) or metallic (Au) stamps, and the resulting CNT thin film patterns are pristine, have high resolution (< 5 μm), and low edge roughness (< 1 μm).Furthermore, the methods we have developed allow placement of patterned CNT films on different plastic substrates.  In the figure below, we show two methods for patterning CVD grown thin films of CNTs.  In (a), we illustrate how a PET stamp is used for subtractive patterning.  A SEM micrograph showing the typical resulting edge resolution is shown in (b).  In

(c), a cold-welding process used for subtractive patterning is also illustrated.  The corresponding edge resolution for this method is shown in SEM micrograph in (d).  The details of these two methods and others can be found in references (1) – (3).

  1. V.K. Sangwan, V.W. Ballarotto, D.R. Hines, M.S. Fuhrer, E.D. Williams, Solid State Electronics 54, 1204 (2010).
  2. V.K. Sangwan, Ph.D. thesis, Ch. 3 (2009).
  3. V.K. Sangwan, A. Benham, V.W. Ballarotto, M.S. Fuhrer, Ant Ural, E.D. Williams, Applied Physics Letters 97, 043111 (2010).

To Top:

Suspended CNTs Devices

These patterning methods have been generalized to make suspended CNT devices to study their intrinsic properties. Suspended CNTs devices show a decrease in substrate-induced effects, demonstrating intrinsic ambipolar behavior with negligible hysteresis in vacuum. We have also found that suspending the nanotubes decreased the spectral noise power by 3 to 10 times, and the

average Hooge’s constant for a suspended CNT-FET was found to be 2.6 x 10-3, over an order of magnitude smaller than identical devices with the CNTs in contact with SiO2. Such pristine suspended CNTs, with decreased influence of the SiO2 surface both in transport characteristics as well as in 1/f noise, will enable further studies of the intrinsic properties of carbon nanotubes.  Complete details can be found in V.K. Sangwan, V.W. Ballarotto, M.S. Fuhrer, E.D. Williams, Applied Physics Letters 93, 113112 (2008)

To Top:

All-CNT Thin-Film Transistor Devices:

Consequently, we are able to develop a viable method to achieve all-carbon electronics.  Solution-processed CNTs as electrodes and pristine CVD-grown CNTs as active layer were incorporated in bottom-gate CNT-based TFTs on large area PET substrate.  A carrier layer-assisted transfer printing approach achieved 100% transfer of CNTs to varieties of substrates, and has a potential of generalization to other nanomaterials.  Resulting CNT-based devices show similar performance to Au-contacted control devices with field-effect mobility in range 1 – 33 cm3/Vs and typical on/off ratio 103.  Ambipolarity of CNT-based devices (compared to p-type control devices) suggests that these CNTs remain relatively undoped even in ambient conditions.

To Top:

Parallel-CNT Field Effect Transistors for High Speed Applications

We have also been developing carbon nanotube (CNT) devices for high speed operation.  To achieve high speed performance, parallel arrays of CNTs are utilized.  The arrays are formed by growing the tubes on mis-cut quartz with standard chemical vapor deposition procedures.  With standard photolithography, the parallel array can be made into the active channel of a high frequency device (see figure, lower right).  Addition of a gate dielectric and third electrode results in a field-effect transistor that can operate in the GHz range.  An optical and SEM micrograph of a completed device is shown below.

The operating speeds of the parallel-CNT FETs have been tested from 10 MHz to 30 GHz.  For example, the devices have been used to successfully mix two RFID-relevant frequencies (10 and 13 MHz).  Below we show the results of mixing a 10 MHz signal with a 20 GHz signal. The upper and lower side bands are readily seen.  Furthermore, the devices have been shown to successfully mix 20 GHz and 29.5 GHz signals.  The up-converted signal at 49.5 GHz is readily observed in the insert of the figure below.  The HF capability of these devices seems to be well beyond 50 GHz.

To Top:


The extremely high mobility of graphene (>20,000 cm2/Vs) suggests that it may be useful for building high speed electronic devices.  Unlike traditional high speed devices like high electron mobility transistors (HEMTs), a two-dimensional electron gas is readily achieved in graphene because of its two-dimensional honeycomb lattice of carbon atoms.  Here we show a graphene flake that has been etched into an acceptable form for a Hall-type transport measurement.  The resulting transfer characteristic is shown on the right.  This particular device shows some hysteretic effects as well as asymmetry in the charge carrier mobility (hole mobility is higher than electron mobility), illustrating some of the problems typically seen in graphene devices.

In addition, the production of electronic grade graphene is limited to time consuming “hunting and gathering” strategies.  There are a number of ideas in the literature for increasing the yield of useful sized graphene (>10 cm2), including epitaxial growth, chemical exfoliation, reduction of graphene oxide to graphene, and chemical vapor deposition methods.
We are working with the UMD (M.S. Fuhrer) to develop practical techniques for producing high quality graphene for electronic applications.  Currently, we are pursuing chemical exfoliation techniques as well as atmospheric-pressure CVD methods for acquiring graphene flakes.

To Top: