Materials science

Nanotubes sorted using DNA

A vast number of DNA sequences are possible, and so finding the few that bind to a particular non-DNA entity is a daunting task. A systematic search algorithm has found sequences that target specific carbon nanotubes.

For nearly two decades, the carbon nanotube has been the poster child of nanotechnology. Researchers have used its exemplary physical and chemical properties in a diverse range of prototype devices, spanning such technologies as alternative energy, biotechnology and computing. Underlying this success is the exquisite sensitivity of the nanotubes' properties to their physical size and atomic structure. However, this sensitivity also creates a fundamental problem: because current syntheses of carbon nanotubes lack atomic-level control, samples produced are mixtures of nanotubes of different sizes and atomic geometries, and thus possess non-uniform properties. This non-uniformity has confounded their use in large-scale commercial applications, which invariably require materials that have consistent, reproducible performance.

Many researchers have therefore devised schemes for sorting carbon nanotubes according to their physical and electronic structures1. Inspiration has often come from bioseparation methods, leading to the use of electrophoresis2, ultracentrifugation3 and chromatography4 techniques. DNA has had a recurring supporting role in these studies because of its ability to disperse carbon nanotubes in biologically compatible aqueous solutions5,6,7. But despite its ability to bind to specific molecules depending on its base sequence, DNA has not been systematically explored as a means of isolating different types of carbon nanotube — until now. On page 250 of this issue, Tu et al.8 describe the heroic efforts that resulted in their identifying more than 20 DNA sequences that each selectively bind a specific carbon-nanotube structure. Their careful study uncovers distinct patterns of DNA sequences that will inform future efforts in nanotube separation, and provides fundamental insight into the chemical interactions between arguably the most important biomolecule and one of the most-studied nanomaterials.

To appreciate the magnitude of the authors' task, consider that custom-made DNA sequences containing 100 nucleotides are readily available commercially. Because there are four DNA bases — adenine (A), thymine (T), guanine (G) and cytosine (C) — this amounts to 4100 (1060) sequences that could be screened for their nanotube-binding properties. This number is almost unfathomably large, and so the authors had to devise a systematic method to focus their search before they could attack this problem experimentally.

Initially, Tu et al. limited their search to DNA molecules containing 28 or 30 bases, thus restricting the number of possibilities to 430 (1018). Although this is a huge improvement over 1060, further refinement was clearly necessary. The authors therefore used a sequence-pattern-expansion scheme to come up with a manageable set of DNA sequences, starting with simple patterns and then adding complexity in a confined, progressive way. The scheme started with molecules that contained only one kind of base, thus yielding four sequences. Complexity was added in the next phase of the scheme — the second-order expansion — when all 16 variants of two-base repeats were added (for example, (AT)15). By following this procedure to a third and fourth order of complexity, Tu et al. constructed a search set containing approximately 350 different DNA sequences.

The authors used each of these sequences to disperse a randomly produced mixture of carbon nanotubes in water. They then used chromatography to separate the resulting 350 solutions into fractions based on the ionic charge of the solutes, and characterized each fraction spectroscopically to see if any of the DNA sequences had formed complexes specifically with a single kind of nanotube. Although most of the sequences had not, a series of DNA molecules that contained alternating patterns of one or more purines (A or G) and pyrimidines (T or C) — such as (GT)15, (TCG)10 and (ATTT)7 — showed a differential affinity for nanotubes as a function of nanotube structure.

Recognizing the successful purine–pyrimidine motifs, Tu et al. performed more experiments in which they varied the length of their DNA sequences, and found that shorter DNA molecules (as short as eight bases) bind to nanotubes with exceptional selectivity. In all, more than 20 distinct DNA sequences selected one kind of carbon nanotube from an as-prepared mixture. The purity of semiconducting nanotubes isolated in this way approached 99%, equalling or exceeding those obtained by all previous carbon-nanotube sorting techniques1.

Although the molecular-recognition mechanism involved in this DNA–nanotube binding8 is not fully understood, highly suggestive trends can be identified from the successful DNA sequences. For example, DNA molecules that contain alternating purine–pyrimidine patterns form stable, well-ordered, two-dimensional sheets through hydrogen bonding (see Fig. 2a on page 252) — structures that resemble the ubiquitous β-sheet motif in proteins. Furthermore, these DNA sheets are expected to form stable cylindrical structures reminiscent of the barrel-shaped structures formed from β-sheets in proteins. Such DNA barrels could thus encapsulate cylindrical carbon nanotubes, presumably with high affinity for nanotubes that have diameters that match the inner diameter of the barrel (see Fig. 2c on page 252). This structural mechanism is different from those of previously described methods for separating carbon nanotubes (which are based on differential chemical binding affinity1) and thus might explain the exceptional purities achieved by Tu and colleagues.

The current study8 marks a considerable advance in the carbon-nanotube field, but major issues remain unresolved. For example, carbon nanotubes are chiral9,10 — each type of nanotube exists as one of two mirror-image forms depending on the direction in which its carbon atoms coil up to form the tube. So far, DNA has not been shown to be able to distinguish between the mirror-image forms of nanotubes, which means that the DNA-separated nanotubes might be sub-optimal for some optical-device applications.

In addition, the DNA sequences identified by Tu et al.8 show higher selectivity for semiconducting carbon nanotubes than for those that have metal-like conductivities, and so further work is required to isolate both types equally using DNA. The authors' approach is also relatively expensive (because of the high cost of DNA), which might limit its use in large-scale applications. The ultimate solution to sorting carbon nanotubes will therefore probably be a hybrid method that combines the best attributes of several different techniques1. In the meantime, Tu and colleagues' approach possesses clear advantages for single-step, low-quantity separations that will be of great interest to research groups around the world.


  1. 1

    Hersam, M. C. Nature Nanotechnol. 3, 387–394 (2008).

    ADS  CAS  Article  Google Scholar 

  2. 2

    Tanaka, T., Jin, H., Miyata, Y. & Kataura, H. Appl. Phys. Express 1, 114001 (2008).

    ADS  Article  Google Scholar 

  3. 3

    Arnold, M. S., Green, A. A., Hulvat, J. F., Stupp, S. I. & Hersam, M. C. Nature Nanotechnol. 1, 60–65 (2006).

    ADS  CAS  Article  Google Scholar 

  4. 4

    Zheng, M. & Semke, E. D. J. Am. Chem. Soc. 129, 6084–6085 (2007).

    CAS  Article  Google Scholar 

  5. 5

    Zheng, M. et al. Nature Mater. 2, 338–342 (2003).

    ADS  CAS  Article  Google Scholar 

  6. 6

    Zheng, M. et al. Science 302, 1545–1548 (2003).

    ADS  CAS  Article  Google Scholar 

  7. 7

    Arnold, M. S., Stupp, S. I. & Hersam, M. C. Nano Lett. 5, 713–718 (2005).

    ADS  CAS  Article  Google Scholar 

  8. 8

    Tu, X., Manohar, S., Jagota, A. & Zheng, M. Nature 460, 250–253 (2009).

    ADS  CAS  Article  Google Scholar 

  9. 9

    Peng, X. et al. Nature Nanotechnol. 2, 361–365 (2007).

    ADS  CAS  Article  Google Scholar 

  10. 10

    Green, A. A., Duch, M. C. & Hersam, M. C. Nano Res. 2, 69–77 (2009).

    CAS  Article  Google Scholar 

Download references

Author information



Rights and permissions

Reprints and Permissions

About this article

Cite this article

Hersam, M. Nanotubes sorted using DNA. Nature 460, 186–187 (2009).

Download citation

Further reading


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.