DOI: 10.1126/science.1087691, 1519 (2003);301 Science

et al.Michael S. StranoFunctionalizationElectronic Structure Control of Single-Walled Carbon Nanotube

This copy is for your personal, non-commercial use only.

clicking here.colleagues, clients, or customers by , you can order high-quality copies for yourIf you wish to distribute this article to others

  here.following the guidelines

can be obtained byPermission to republish or repurpose articles or portions of articles

  ): April 1, 2014 www.sciencemag.org (this information is current as of

The following resources related to this article are available online at

http://www.sciencemag.org/content/301/5639/1519.full.htmlversion of this article at:

including high-resolution figures, can be found in the onlineUpdated information and services,

http://www.sciencemag.org/content/suppl/2003/09/11/301.5639.1519.DC1.html can be found at: Supporting Online Material

http://www.sciencemag.org/content/301/5639/1519.full.html#relatedfound at:

can berelated to this article A list of selected additional articles on the Science Web sites

http://www.sciencemag.org/content/301/5639/1519.full.html#ref-list-1, 2 of which can be accessed free:cites 17 articlesThis article

502 article(s) on the ISI Web of Sciencecited by This article has been

http://www.sciencemag.org/content/301/5639/1519.full.html#related-urls13 articles hosted by HighWire Press; see:cited by This article has been

http://www.sciencemag.org/cgi/collection/chemistryChemistry

subject collections:This article appears in the following

registered trademark of AAAS. is aScience2003 by the American Association for the Advancement of Science; all rights reserved. The title

CopyrightAmerican Association for the Advancement of Science, 1200 New York Avenue NW, Washington, DC 20005. (print ISSN 0036-8075; online ISSN 1095-9203) is published weekly, except the last week in December, by theScience

on

Apr

il 1,

201

4w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from

o

n A

pril

1, 2

014

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fr

om

on

Apr

il 1,

201

4w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from

o

n A

pril

1, 2

014

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fr

om

on

Apr

il 1,

201

4w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from

tions should be an important feature ofpolymer theories attempting to modelstresses in extensional flows. We anticipatethat the effects of hydrodynamic interac-tions will be crucial for description of moreflexible synthetic polymers such as poly-styrene, with a smaller ratio of persistencelength to hydrodynamic radius, and hencelarger extensibility (L/Rg) ratios.

Finally, conformational hysteresis mayplay a role in turbulent-drag reduction, aneffect discovered by B. A. Toms a half cen-tury ago (32): high molecular weight poly-mers mixed with fluids at an extremely dilutelevel (�1 part per million by weight) canreduce the drag resistance in turbulent flowby as much as 80% (33). Two types of ex-planations of this effect have been proposed.The first conjecture, originally proposed byLumley (34), argues that the drag reductionoccurs at the boundary between the turbulent-core region and the laminar zone near thepipe surface. Polymers that have been ex-tended by (transient) elongational flows canenter the boundary layer and reduce the mo-mentum transfer between the rapidly movingfluid and the laminar layer. Polymers wouldremain extended for longer periods of time intheir stretched state because of conformation-al-dependent drag. Hysteresis would furthermagnify this effect. For synthetic polymersthat exhibit a large amount of turbulent-dragreduction, �stretch/�coil is estimated to be �18.In contrast, Tabor and de Gennes have argued(35) that polymers in turbulent flows experi-ence rapidly varying extensional flows so thatthe coil-stretch transition disappears entirely.Instead, they propose that energy is trans-ferred in turbulent flows though a cascade ofeddies to smaller size scales where it is final-ly dissipated. Long polymers interrupt thiscascade by storing some of this energy in theform of an elastic modulus that is then deliv-ered back to the moving fluid.

References and Notes1. R. G. Larson, The Structure and Rheology of Complex

Fluids (Oxford Univ. Press, New York, 1999).2. H. P. Babcock, R. E. Teixeira, J. S. Hur, E. S. G. Shaqfeh,

S. Chu, Macromolecules 36, 4544 (2003).3. R. B. Bird, C. F. Curtiss, R. C. Armstrong, O. Hassager,

Dynamics of Polymeric Liquids (Wiley, New York, ed.2, 1987), vol. 2.

4. P. G. de Gennes, J. Chem. Phys. 60, 5030 (1974).5. G. G. Fuller, L. G. Leal, Rheol. Acta 19, 580 (1980).6. D. Hunkeler, T. Q. Nguyen, H. H. Kausch, Polymer 37,

4257 (1996).7. M. J. Menasveta, D. A. Hoagland, Macromolecules 24,

3427 (1991).8. E. C. Lee, S. J. Muller, Macromolecules 32, 3295

(1999).9. T. T. Perkins, D. E. Smith, S. Chu, Science 276, 2016

(1997).10. D. E. Smith, S. Chu, Science 281, 1335 (1998).11. E. J. Hinch, in Proc. Symp. Polym. Lubrification 233,

241 (1974).12. R. I. Tanner, Trans. Soc. Rheol. 19, 557 (1975).13. E. J. Hinch, Phys. Fluids 20, s22 (1977).14. Y. V. Brestkin, Acta Polymerica 38, 470 (1987).15. J. J. Magda, R. G. Larson, M. E. Mackay, J. Chem. Phys.

89, 2504 (1988).

16. J. M. Rallison, E. J. Hinch, J. Non-Newtonian FluidMech. 29, 37 (1988).

17. X. J. Fan, R. B. Bird, M. Renardy, J. Non-NewtonianFluid Mech. 18, 255 (1985).

18. J. M. Wiest, L. E. Wedgewood, R. B. Bird, J. Chem.Phys. 90, 587 (1988).

19. P. G. de Gennes, Scaling Concepts in Polymer Physics(Cornell Univ. Press, Ithaca, NY, 1979).

20. F. Reif, Fundamentals of Statistical and Thermal Phys-ics (McGraw-Hill, New York, 1965).

21. G. K. Batchelor, J. Fluid Mech. 44, 419 (1970).22. We have omitted a numerical constant on the order

of unity in this expression.23. D. E. Smith, T. T. Perkins, S. Chu, Macromolecules 29,

1372 (1996).24. T. T. Perkins, D. E. Smith, R. G. Larson, S. Chu, Science

268, 83 (1995).25. DNA molecules were imaged with a Micromax

512BFT camera from Roper Scientific, using a ZeissAxioplan microscope equipped for epifluorescencewith a 40� 1.0 numerical aperture objective oil-immersion lens. We used a 0.31� demagnifying lensto provide a field of view of �480 �m. For polymerextensions greater than our field of view, we trans-lated our microscope stage in the direction of mo-lecular stretch to discern total extended lengths. Thetime scale for translation was on the order of sec-onds, which was much faster than the time scale oftransient molecule dynamics for the range of εprobed.

26. Materials and methods are available as supportingmaterial on Science Online.

27. B. J. Bentley, L. G. Leal, J. Fluid Mech. 167, 219(1986).

28. Polymer relaxation times were measured by firststretching the polymer molecules at high De andthen stopping the flow. The extent of the visualimage of the molecule is tracked as a function oftime, and the final 30% of the relaxation is fit to adecaying exponential �x � x� � Aexp(t / r) � B,where x is dimensional polymer extension, r is thelongest polymer relaxation time, and A and B are

fitting constants. Observation times t obs � ε/ε onthe order of several hours were required for fluidstrains of about 10 to 15 units. Therefore, we addeda small concentration of Sytox dye (MolecularProbes) to our inlet buffer solutions. Dye moleculesbound to DNA exchange with free dye in solution sothat fresh dye molecules replenish older, photo-bleached dye molecules. We also used a mechanicalshutter to minimize light exposure from a mercurylamp illuminator and an oxygen-scavenging glucoseoxidase–catalase enzyme system to minimize photo-bleaching. Combining these techniques, we achievedstable polymer relaxation times for at least 7 hours ofobservation time of a single DNA molecule.

29. R. M. Jendrejack, J. J. de Pablo, M. D. Graham, J. Chem.Phys. 116, 7752 (2002).

30. C. C. Hsieh, L. Li, R. G. Larson, J. Rheol., in press.31. J. Rotne, S. Prager, J. Chem. Phys. 50, 4831 (1969).32. B. A. Toms, in Proc. Int. Congr. Rheol. 2, 135 (1949).33. P. S. Virk, Am. Inst. Chem. Eng. J. 21, 625 (1975).34. J. Lumley, J. Polym. Sci. Macromol. Rev. 7, 263

(1973).35. P. G. de Gennes, Introduction to Polymer Dynamics

(Cambridge Univ. Press, Cambridge, 1990).36. We thank M. Gallo and E. Chan at U.S. Genomics for

generosity in genomic-length E. coli DNA samplepreparation and G. Fuller and R. Larson for usefuldiscussions. Supported in part by the Materials Re-search Science and Engineering Center Program ofthe NSF (DMR-0213618 and DMR-9808677), the AirForce Office of Scientific Research, and theNSF. C.M.S. was supported in part by an NSF gradu-ate fellowship.

Supporting Online Materialwww.sciencemag.org/cgi/content/full/301/5639/1515/DC1Materials and MethodsSOM TextFigs. S1 to S3

24 April 2003; accepted 8 August 2003

Electronic Structure Control ofSingle-Walled Carbon Nanotube

FunctionalizationMichael S. Strano,1*† Christopher A. Dyke,2* Monica L. Usrey,1

Paul W. Barone,1 Mathew J. Allen,2 Hongwei Shan,2

Carter Kittrell,2 Robert H. Hauge,2 James M. Tour,2,3†Richard E. Smalley2,4†

Diazonium reagents functionalize single-walled carbon nanotubes suspended inaqueous solution with high selectivity and enable manipulation according toelectronic structure. For example, metallic species are shown to react to thenear exclusion of semiconducting nanotubes under controlled conditions. Se-lectivity is dictated by the availability of electrons near the Fermi level tostabilize a charge-transfer transition state preceding bond formation. Thechemistry can be reversed by using a thermal treatment that restores thepristine electronic structure of the nanotube.

The main hurdle to the widespread application ofsingle-walled carbon nanotubes is their manipu-lation according to electronic structure (1). Allknown preparative methods (2–4) lead to poly-disperse materials of semiconducting, semime-tallic and metallic electronic types. Recent ad-vances in the solution-phase dispersion (5, 6),along with spectroscopic identification usingband-gap fluorescence (7) and Raman spectros-

copy (8), have greatly improved the ability tomonitor electrically distinct nanotubes as sus-pended mixtures and have led to definitive as-signments of the optical features of semiconduct-ing (7), as well as metallic and semimetallic,species (8).

We now report selective reaction pathwaysof carbon nanotubes in which covalent chemi-cal functionalization (9) is controlled by differ-

R E P O R T S

www.sciencemag.org SCIENCE VOL 301 12 SEPTEMBER 2003 1519

ences in the nanotube electronic structure. Wedemonstrate the utility of these chemical path-ways for manipulation of nanotubes of distinctelectronic types by selective functionalizationof metallic nanotubes. This chemistry is amarked departure from previously developedmechanisms for nanotube selectivity that arederived from fullerene chemistry, namely, thosebased on the carbon pyramidalization angle (10,11). Controlling nanotube reaction pathways inthis way should allow for the separation ofsemiconducting from metallic and semimetallicnanotubes with high selectivity and scalability,as well as the direct fabrication of devices of aparticular electronic type. In contrast to recentwork reporting enrichment of metals over semi-conductors (12, 13), we show a nearly completeselectivity for particularly metallic species.

The diversity in electronic structure ofcarbon nanotubes arises from the quantiniza-tion of the electronic wave vector of theone-dimensional (1D) system through theconceptual rolling of a graphene plane into acylinder forming the nanotube (2, 4). Thevector in units of hexagonal elements con-necting two points on this plane defines thenanotube chirality in terms of two integers: nand m. When �n m� � 3q, where q is aninteger, the nanotube is metallic or semime-tallic, and the remaining species are semicon-ducting with a geometry-dependent bandgap(14). Although largely unrealized in previousstudies, subtle differences in geometric struc-ture of carbon nanotubes lead to markedchanges in the rates of solution-phase reac-tivity of these species. We find that water-soluble diazonium salts (15), which havebeen shown to react with carbon nanotubes(9, 16, 17), can extract electrons from nano-tubes in the formation of a covalent aryl bond(Fig. 1A) and thereby demonstrate highlychemoselective reactions with metallic versusthe semiconducting tubes.

This bond forms with extremely high affin-ity for electrons with energies, Er, near theFermi level, Ef of the nanotube (Fig. 1B). Thereactant forms a charge-transfer complex at thenanotube surface, where electron donation fromthe latter stabilizes the transition state and ac-celerates the forward rate. Once the bondsymmetry of the nanotube is disrupted bythe formation of this defect, adjacent car-bons increase in reactivity (Fig. 1C), and

the initial selectivity is amplified as theentire nanotube is functionalized.

Under carefully controlled conditions (18),this behavior can be exploited to obtain highlyselective functionalization of metallic and semi-metallic nanotubes to the exclusion of the semi-conductors. Figure 2 shows the ultraviolet–visible–near-infrared (UV-vis-nIR) absorptionspectra of aqueous suspended nanotubes aftersuccessive additions of 4-chlorobenzenediazo-nium tetrafluoroborate after steady state. Thespectrum monitors the valence (v) to conduction

(c) electronic transitions denoted (vn3cn)where n is the band index. Figure 2 indicates thev13c1 transitions of the metallic and semime-tallic nanotubes from roughly 440 to 645 nm, aswell as the v13c1 and v23c2 transitions of thesemiconducting nanotubes in the ranges from830 to 1600 nm and 600 to 800 nm, respectively.These separated absorption features allow for themonitoring of valence electrons in each distinctnanotube; as the species reacts to form covalentlinkages, electrons are localized and these max-ima decay. Under controlled additions, only

1Department of Chemical and Biomolecular Engineer-ing, University of Illinois–Urbana/Champaign, Urbana,IL 61801, USA. 2Department of Chemistry, Center forNanoscale Science and Technology, and Center forBiological and Environmental Nanotechnology, 3De-partment of Mechanical Engineering and MaterialsScience, 4Department of Physics, Rice University,6100 Main Street, Houston, TX 77005, USA.

*These authors contributed equally to this work.†To whom correspondence should be addressed. E-mail: strano@uiuc.edu (m.s.s.); res@rice.edu (R.E.S.);tour@rice.edu (J.M.T.)

e

DOS

Cl

ClN+NF4B–

EF–∆ET

EF

EF–∆ET

EF

ClCl

rgy

En

erg

y

NN+NNFF4BB–

Metallic (|n-m| = 3q)M

-conducting (|n-m|Semi ≠ 3q)

A B

C

Fig. 1. (A) Diazoniumreagents extract elec-trons, thereby evolv-ing N2 gas and leavinga stable C™C covalentbond with the nano-tube surface. (B) Theextent of electrontransfer is dependenton the density ofstates in that electrondensity near EF leadsto higher initial activ-ity for metallic andsemimetallic nano-tubes. (C) The arene-functionalized nano-tube may now exist asthe delocalized radicalcation, which couldfurther receive elec-trons from neighbor-ing nanotubes or reactwith fluoride or diazo-nium salts.

Fig. 2. (A) UV-vis-nIR spectrum of sodium dodecyl sulfate–suspended carbon nanotubes after theaddition of various amounts of 4-chlorobenzenediazonium tetrafluoroborate (in mol/1000 mol carbon).Under controlled conditions, semiconductor transitions are unaffected, whereas metals react with highselectivity. (B) Expanded view of the metallic region. The tetrafluoroborate salt causes some batho-chromic shifting of the longer wavelength features because of changes in the surfactant adsorbed phase.

R E P O R T S

12 SEPTEMBER 2003 VOL 301 SCIENCE www.sciencemag.org1520

metallic transitions initially decay (Fig. 2), indi-cating highly preferential functionalization ofmetallic nanotubes. This selectivity is remark-able given that these transitions arise from elec-trons that are much lower in energy comparedwith the v13c1 and v23c2 transitions of thesemiconductors. The selective decay of thesemetallic transitions is distinct from reversibleelectronic withdraw (19) or generic “doping”processes (20), as previously reported. Selectiv-ity is also confirmed by the preservation of band-gap fluorescence of the semiconducting nano-tubes, which is observed to be highly sensitive tochemical defects. In Fig. 2A, we attribute thechange in the feature near 1350 nm to corre-spond to a nonspecific solvatochromic shift re-sulting from addition of the BF4

counterion thathas been observed previously (21).

The functionalization increases the intensityof a phonon mode at 1330 cm1 in the Ramanspectrum, as shown in Fig. 3A at 532 nm exci-tation. Its prominence corresponds with the con-version of an sp2 C to an sp3 C on the nanotubeduring the formation of an sp3 C™sp2 Cnanotube™aryl bond. The so-called D-band in-volves the resonantly enhanced scattering of anelectron via phonon emission by a defect that

breaks the basic symmetry of the graphene plane(2, 4). This peak is not observed to increase asthe result of adsorption of hydronium ions (19)or surfactants (5) on the nanotube sidewall. Weobserve that the height of this peak increasessharply with increasing functionalization, thendecreases along with the C™C tangential mode asthe system loses electronic resonance (Fig. 3B).These results allow for spectroscopic correlationof the number of sidewall functionalizationevents to this phonon intensity at low conversion.

The addition of the moiety to the sidewall ofthe nanotube disrupts the oscillator strength thatgives rise to resonantly enhanced, low-frequencyRaman lines that are distinct for species of aparticular diameter. This causes the mode todecay accordingly as the particular (n,m) nano-tube reacts. Figure 4 analogously shows the so-lution-phase Raman spectra of the mixture at532 nm with each reactant addition after steadystate. The relative rates of the decays of thesefeatures reveal unprecedented reactivity differ-ences between chiral semimetallic species. Here,Raman spectroscopy probes nanotubes withnearly identical transition energies, and thesedifferences reveal a curvature-dependent stabili-zation of the charge-transfer complex that mayultimately be exploited to separate semimetallicand metallic species. When all v13c1 transi-tions of semimetallic and metallic species havedecayed (Fig. 2), only one low-frequency Ramanmode that we have previously assigned to the(9,2) semiconductor (19) remains unaffected.These results also independently confirm therecent spectroscopic assignment of thesefeatures (7, 8).

Carbon nanotube chemistry has been correct-ly described with a pyramidization angle formal-ism (10). Here, chemical reactivity and kineticselectivity are related to the extent of s characterdue to the curvature-induced strain of the sp2-hybridized graphene sheet. Because strain ener-gy per carbon is inversely related to nanotubediameter, this model predicts that smaller diam-eter nanotubes will be the most reactive, with theenthalpy of reaction decreasing as the curvaturebecomes infinite. Although such behavior ismost commonly the case, our findings under-score the role of electronic structure in determin-ing the reactivity of the nanotube. Because sucha structure is highly sensitive to chiral wrapping,chemical doping, and charged adsorbates, aswell as to nanotube diameter, there is a consid-erable diversity of these various pathways, inaddition to a simple diameter dependence.

Thermal pyrolysis of the reacted material at300°C in an atmosphere of inert gas cleaves (22)the aryl moieties from the sidewall and restoresthe spectroscopic signatures of the aromatic,pristine nanotubes (9). Figure 5 compares theRaman spectra before and after recovery andthermal pyrolysis at 633 nm (Fig. 5). This wave-length was used because it probes a mixture ofmetals and semiconductors for samples preparedby CO disproportionation (8). The radial phonon

1200

0 5 10 15 20 25

1400 1600

ii

iii

iiii

iii

TM peak

D peak

Inte

nsi

ty

Rel

ativ

e in

ten

sity

Raman shift (cm–1)

Functional groups/carbons x 103

A Bii

iii

iiii

2000

1500

1000

500

0

300

200

100

0

1200 1300 1400

Fig. 3. (A) Raman spectrum at 532-nm excitation, showing the growth of the “disorder” mode withincreasing functionalization from 0 (i) to 5.6 (ii) to 22.4 (iii) groups attached per 1000 carbonatoms. (B) The intensity of the tangential mode (TM) � 0.1 decreases as resonance enhancementof the scattering event is lost with increasing reaction. The disorder mode (D) increases sharplythen decays because of the same loss of enhancement.

(9,2)

(9,3)

(13,1)

175 225 275 325

(9,6)

(10,4)

(9,2)

(9,3)

(9,2)

(9,6)(10,4)

100

0

200

100

0

300

200

100

0

Rel

ativ

e in

ten

sity

Raman shift (cm–1)

A

B

C

Fig. 4. (A) Low–wave number Raman spectra at532-nm excitation of the starting solution. Fourmetallic nanotubes (black) are probed at thiswavelength and one semiconductor (red) via aradial mode sensitive to nanotube diameter. (B)After 5.6 groups attached per 1000 carbons,functionalization disrupts this mode, as seen bythe decay particularly of the small-diametermetals. This initial evidence of selective reac-tivity among metals provides a handle for sep-aration of these species. (C) After a ratio of22.4, all metallic modes have decayed, leavingonly the single semiconductor, in agreementwith Fig. 2B.

(13,4)(9,9)

150 200 250 300

(12,3)(10,3)

(7,6)

(7,5)

2700

1800

900

0

Inte

nsi

ty

Raman shift (cm–1)

Fig. 5. Raman spectra at 633 nm probing bothmetals and semiconducting nanotubes beforereaction (solid line) and after recovery andthermal pyrolysis (dotted line). The reversibilityof the chemistry implies that intrinsic electron-ic and optical properties of the pristine nano-tubes can be recovered.

R E P O R T S

www.sciencemag.org SCIENCE VOL 301 12 SEPTEMBER 2003 1521

modes are nearly completely restored after ther-mal treatment. Similarly, electronic transitions inthe absorption spectrum are restored, indicatingthe loss of the side group and a restoration of theoriginal electronic structure of the nanotube (22).Hence, this selective chemistry can be used as areversible route to separate, deposit, or chemi-cally link nanotubes of a particular electronicstructure, and the original optical and electroniccharacteristics can then be recovered.

References and Notes1. P. Avouris, Acc. Chem. Res. 35, 1026 (2002).2. M. S. Dresselhaus, G. Dresselhaus, P. C. Eklund, Sci-

ence of Fullerenes and Carbon Nanotubes (AcademicPress, San Diego, CA, 1996).

3. M. J. Bronikowski, P. A. Willis, D. T. Colbert, K. A.Smith, R. Smalley, J. Vac. Sci. Technol. 19, 1800(2001).

4. R. Saito, G. Dresselhaus, M. S. Dresselhaus, PhysicalProperties of Carbon Nanotubes (Imperial CollegePress, London, 1998).

5. M. S. Strano et al., J. Nanosci. Nanotechnol. 3, 81 (2003).6. M. J. O’Connell et al., Science 297, 593 (2002).7. S. M. Bachilo et al., Science 298, 2361 (2002).8. M. S. Strano, Nanoletters 3, 1091 (2003).9. J. L. Bahr, J. M. Tour, J. Mater. Chem. 12, 1952 (2002).

10. S. Niyogi et al., Acc. Chem. Res. 35, 1105 (2002).11. Z. F. Chen,W. Thiel, A. Hirsch,Chemphyschem4, 93 (2003).12. M. Zheng et al., Nature Mater. 2, 338 (2003).13. D. Chattopadhyay, I. Galeska, F. Papadimitrakopoulos,

J. Am. Chem. Soc. 125, 3370 (2003).14. S. Reich, C. Thomsen, Phys. Rev. B 62, 4273 (2000).15. C. Bravo-Diaz, M. Soengas-Fernandez, M. Rodriguez-Sara-

bia, E. Gonzalez-Romero, Langmuir 14, 5098 (1998).16. C. A. Dyke, J. M. Tour, J. Am. Chem. Soc. 125, 1156 (2003).17. J. L. Bahr et al., J. Am. Chem. Soc. 123, 6536 (2001).18. A recirculating flow reactor was used to transfer

sodium dodecyl sulfate–suspended carbon nanotubesat pH � 10 at a flow rate of 150 ml/min through acuvette with inlet and outlet ports. Continuous UV-vis-nIR spectra were generated after the addition of ametered amount of diazonium aryl chloride tetraflu-oroborate. Addtions were made in 0.05 mM incre-ments after the system had reached steady state.

19. M. S. Strano et al., J. Phys. Chem. B 107, 6979 (2002).20. M. E. Itkis et al., Nanoletters 2, 155 (2002).21. Increasing the ionic strength of the surfactant-sus-

pended nanotubes screens the charged repulsion ofthe sulfate head groups and causes the adsorbedlayer to adopt a configuration that allows greateraccess of water to the surface. This causes a charac-teristic red shift of the absorption transitions that ismost prominent for the v13c1 transitions of thesemiconductors. Because of the convolution of thetransitions in the spectrum, this shifting is seen as achange in intensity of one region in particular. Theresults can be qualitatively duplicated by the additionof 100 mM NaCl to the solution.

22. Thermogravimetric analysis, as well as absorptionand Raman spectra of the reacted and thermallyrestored material, are available as supporting mate-rial on Science Online.

23. We thank J. White for assistance with Raman spec-troscopy M.S.S. acknowledges the financial support ofthe University of Illinois, School of Chemical Scienc-es. Support at Rice University was provided by theNSF Focused Research Group on Fullerene NanotubeChemistry (DMR-0073046), the NSF Center for Bio-logical and Environmental Nanotechnology (EEC-0118007), NASA URETI NCC-01-0203, Air Force Of-fice of Scientific Research (F49620-01-1-0364), andthe Office of Naval Research Polymer Division. Sup-port from NASA (NCC9-77) for development of theHiPco method is also gratefully acknowledged.

Supporting Online Materialwww.sciencemag.org/cgi/content/full/301/5639/1519/DC1Materials and MethodsFigs. S1 to S3

5 June 2003; accepted 5 August 2003

Anomalous Nitrogen IsotopeRatio in Comets

Claude Arpigny,1* Emmanuel Jehin,2 Jean Manfroid,1

Damien Hutsemekers,1 Rita Schulz,3 J. A. Stuwe,4

Jean-Marc Zucconi,5 Ilya Ilyin6

High-resolution spectra of the CN B2 ��X2 �� (0,0) band at 390 nanometersyield isotopic ratios for comets C/1995 O1 (Hale-Bopp) and C/2000 WM1(LINEAR) as follows: 165 � 40 and 115 � 20 for 12C/13C, 140 � 35 and 140 �30 for 14N/15N. Our N isotopic measurements are lower than the terrestrial14N/15N � 272 and the ratio for Hale-Bopp from measurements of HCN, thepresumed parent species of CN. This isotopic anomaly suggests the existenceof other parent(s) of CN, with an even lower N isotopic ratio. Organic com-pounds like those found in interplanetary dust particles are good candidates.

Determination of the abundance ratios of thestable isotopes of the light elements in differentobjects of the solar system (SS) provides im-portant clues regarding origin of the SS and itsearly history. Comets are among the best-pre-served specimens of the primitive solar nebulaand, as such, they can play an outstanding role.The ground-based determination of their C andN isotope ratios is based on the comparison ofthe intensities of spectral features of the variousisotopic species. The observed molecules areCN and C2 in the optical domain and HCN inthe submillimeter range. These measurementsare difficult in both domains due mainly to theweakness of the emissions of the low-abun-dance species (1).

We observed comet C/2000 WM1 (here-after designated as “WM1”) in March 2002with the Ultraviolet-Visual Echelle Spectro-

graph (UVES) (2) mounted on the 8.2-m UT2(Kueyen) telescope of the European SouthernObservatory Very Large Telescope (ESOVLT) array at Cerro Paranal, Chile, in orderto measure the C and N isotopic ratios fromthe CN Violet (0,0) band (Fig. 1).

Cometary emissions are produced by absorp-tion of the solar light followed by re-emission oflines of different frequencies, a process calledresonance-fluorescence. The synthetic fluores-cence spectra of the various isotopomers 12C14N,13C14N, and 12C15N (13C15N is negligible) werecomputed for each observing circumstance (Ta-ble 1) (1) according to the scheme described byZucconi and Festou (3). The isotope ratios areestimated by fitting the average observed CNspectrum with a linear combination of the syn-thetic spectra of the three species (1). The finalvalues are 12C/13C � 115 � 20 and 14N/15N �140 � 30 (Fig. 2). The “errors” cited give thedeviations of the values for which acceptable fitsare obtained using various procedures and vari-ous sets of lines (1). These ratios are consistentwith estimates of Hale-Bopp in which we iden-tified the presence of 12C15N (4) and for whichwe have now derived 12C/13C � 165 � 40 and14N/15N � 140 � 35.

The C ratio was measured in situ by theVEGA and GIOTTO spacecraft in P/Halley(5) and from ground-based, submillimeter,

1Institut d’Astrophysique et de Geophysique, Sart-Tilman, Batiment B5c, B-4000 Liege, Belgium. 2Euro-pean Southern Observatory, Casilla 19001, Santiago,Chile. 3ESA/RSSD, ESTEC, Post Office Box 299, NL-2200 AG Noordwijk, Netherlands. 4Leiden Observa-tory, NL-2300 RA Leiden, Netherlands. 5Observatoirede Besancon, F25010 Besancon Cedex, France. 6As-tronomy Division, Post Office Box 3000, FIN 90014University of Oulu, Finland.*To whom correspondence should be addressed. E-mail: Claude.Arpigny@ulg.ac.be

Fig. 1. The UVES averageobserved spectrum (1) ofthe R branch of the CN(0,0) band in cometC/2000 WM1. Some ofthe rotational lines are la-beled with the quantumnumber N of the lowerlevel of the correspondingtransition. The small fea-tures between the R linesare essentially P lines ofthe (1,1) band. “Hidden”among these lines, at astill smaller level, are theR lines of 12C15N and13C14N.

R E P O R T S

12 SEPTEMBER 2003 VOL 301 SCIENCE www.sciencemag.org1522