Nature © Macmillan Publishers Ltd 1998

8

attack Si–Si bonds in a reaction layer close to the interface3–5. Inview of the results of photoemission experiments18,19, the reactionlayer should have a width of 5–10 A. Within the reaction layer,network rearrangement centred on transient threefold-coordinatedoxygen sites leads to a constant cycling of atoms through theinterface. This process disrupts the crystalline Si lattice in thelayer adjacent to the interface. It carries the oxygen atoms whichwere incorporated into Si–Si bonds in the reaction layer into thedisrupted network at the interface, leading to new interface oxide.This mechanism drives some excess Si atoms into the Si layeradjacent to the interface. New Si–Si bonds are also cycled up intothe reaction layer, sustaining the oxidation process as more oxygenis added. This scheme naturally allows for strain relief duringgrowth, and is consistent with medium-energy ion-scatteringresults which show evidence for motion of oxygen atoms afterincorporation into the oxide5,6. The network motion provides anexplanation for Si interstitial (as opposed to vacancy) injectionduring oxidation7. Finally, these atomic-scale motions may be thesource of strain-free growth observed at high temperature17. M

Received 7 May; accepted 17 August 1998.

1. Deal, B. E. & Grove, A. S. General relationship for the thermal oxidation of silicon. J. Appl. Phys. 36,3770–3778 (1965).

2. Rochet, F. et al. The thermal oxidation of silicon: The special case of the growth of very thin films. Adv.Phys. 35, 237–274 (1986).

3. Stoneham, A. M., Grovenor, C. R. M. & Cerezo, A. Oxidation and the structure of the silicon/oxideinterface. Phil. Mag. B 55, 201–210 (1987).

4. Mott, N. F., Rigo, S., Rochet, F. & Stoneham, A. M. Oxidation of silicon. Phil. Mag. B 60, 189–212(1989).

5. Gusev, E. P., Lu, H. C., Gustafsson, T. & Garfunkel, E. Growth mechanism of the thin silicon oxidefilms on Si(100) studied by medium-energy ion scattering. Phys. Rev. B 52, 1759–1775 (1995).

6. Gusev, E. P., Lu, H. C., Gustafsson, T. & Garfunkel, E. in The Physics and Chemistry of SiO2 and the Si-SiO2 Interface—3 Vol. 96-1 (eds Massoud, H. Z., Poindexter, E. H. & Helms, C. R.) 49–58(Electrochemical Soc., Pennington, 1996).

7. Fahey, P., Griffin, P. B. & Plummer, J. D. Point defects and dopant diffusion in silicon. Rev. Mod. Phys.61, 289–384 (1989).

8. Banaszak Holl, M. M. & McFeely, F. R. Si/SiO2 Interface: New structures and well-defined modelsystems. Phys. Rev. Lett. 71, 2441–2444 (1993).

9. Banaszak Holl, M. M., Lee, S. & McFeely, F. R. Core-level photoemission and the structure of the Si/SiO2 interface: A reappraisal. Appl. Phys. Lett. 85, 1097–1099 (1994).

10. Pasquarello, A., Hybertsen, M. S. & Car, R. Si 2p core-level shifts at the Si(001)-SiO2 interface: A first-principles study. Phys. Rev. Lett. 74, 1024–1027 (1995).

11. Pasquarello, A., Hybertsen, M. S. & Car, R. Theory of Si 2p core-level shifts at the Si(001)-SiO2

interface. Phys. Rev. B 53, 10942–10950 (1996).12. Pasquarello, A., Hybertsen, M. S. & Car, R. Structurally relaxed models of the Si(001)-SiO2 interface.

Appl. Phys. Lett. 68, 625–627 (1996).13. Car, R. & Parrinello, M. Unified approach for molecular dynamics and density-functional theory.

Phys. Rev. Lett. 55, 2471–2474 (1985).14. Pasquarello, A., Laasonen, K., Car, R., Lee, C. & Vanderbilt, D. Ab initio molecular dynamics for d-

electron systems: Liquid copper at 1500 K. Phys. Rev. Lett. 69, 1982–1985 (1992).15. Laasonen, K., Pasquarello, A., Car, R., Lee, C. & Vanderbilt, D. Car-Parrinello molecular dynamics

with Vanderbilt ultrasoft pseudopotentials. Phys. Rev. B 47, 10142–10153 (1993).16. Witczak, S. C., Suehle, J. S. & Gaitan, M. An experimental comparison of measurement techniques to

extract Si-SiO2 interface trap density. Solid-State Electron. 35, 345–355 (1992).17. EerNisse, E. P. Stress in thermal SiO2 during growth. Appl. Phys. Lett. 35, 8–10 (1979).18. Himpsel, F. J., McFeely, F. R., Taleb-Ibrahimi, A., Yarmoff, J. A. & Hollinger, G. Microscopic structure

of the SiO2/Si interface. Phys. Rev. B 38, 6084–6096 (1988).19. Lu, Z. H., Graham, M. J., Jiang, D. T. & Tan, K. H. SiO2/Si(100) interface studied by Al Ka x-ray and

synchrotron radiation photoelectron spectroscopy. Appl. Phys. Lett. 63, 2941–2943 (1993).20. Oshishi, K. & Hattori, T. Periodic changes in SiO2/Si(111) interface structures with progress of

thermal oxidation. Jpn J. Appl. Phys. 33, L675–L678 (1994).21. Sarnthein, J., Pasquarello, A. & Car, R. Structural and electronic properties of liquid and amorphous

SiO2: An ab initio molecular dynamics study. Phys. Rev. Lett. 74, 4682–4685 (1995).22. Sarnthein, J., Pasquarello, A. & Car, R. Model of vitreous SiO2 generated by an ab initio molecular

dynamics quench from the melt. Phys. Rev. B 52, 12690–12695 (1995).23. Nose, S. A molecular dynamics method for simulations in the canonical ensemble. Mol. Phys. 52, 255–

268 (1984).24. Hoover, W. G. Canonical dynamics: Equilibrium phase-space distributions. Phys. Rev. A 31, 1695–

1697 (1985).25. Feldman, L. C., Silverman, P. J., Williams, J. S., Jackman, T. E. & Stensgaard, I. Use of thin Si crystals in

backscattering-channeling studies of the Si-SiO2 interface. Phys. Rev. Lett. 41, 1396–1399 (1978).26. Jackman, T. E., MacDonald, J. R., Feldman, L. C., Silverman, P. J. & Stensgaard, I. (100) and (110)Si-

SiO2 interface studies by MeV ion backscattering. Surf. Sci. 100, 35–42 (1980).27. Kosowsky, S. D. et al. Evidence of annealing on a high-density Si/SiO2 interfacial layer. Appl. Phys. Lett.

70, 3119–3121 (1997).28. Filipponi, A., Evangelisti, F., Benfatto, M., Mobilio, S. & Natoli, C. R. Structural investigation of a-Si

and a-Si:H using x-ray absorption spectroscopy at the Si K edge. Phys. Rev. B 40, 9636–9643 (1989).29. Car, R. & Parrinello, M. Structural, dynamical, and electronic properties of amorphous silicon: An ab

initio molecular-dynamics study. Phys. Rev. Lett. 60, 204–207 (1988).

Acknowledgements. We thank A. Mangili for providing visualization support. The calculations wereperformed on the NEC-SX4 of the Swiss Center for Scientific Computing (CSCS).

Correspondence and requests for materials should be addressed to A.P. at IRRMA (e-mail: Alfredo.Pasquarello@epfl.ch).

letters to nature

60 NATURE | VOL 396 | 5 NOVEMBER 1998 | www.nature.com

Molecular-wirebehaviour inp-phenylenevinyleneoligomersWilliam B. Davis*, Walter A. Svec†, Mark A. Ratner* &Michael R. Wasielewski*†

* Department of Chemistry, Northwestern University, Evanston,Illinois 60208-3113, USA† Chemistry Division, Argonne National Laboratory, Argonne,Illinois 60439-4831, USA. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Electron transfer from electron-donor to electron-acceptor mole-cules via a molecular ‘bridge’ is a feature of many biologicaland chemical systems. The electronic structure of the bridgecomponent in donor–bridge–acceptor (DBA) systems is knownto play a critical role in determining the ease of electron transfer1,2.In most DBA systems, the rate at which electron transfer occursscales exponentially with the donor–acceptor distance—effec-tively the length of the bridge molecule. But theory predicts thatregimes exist wherein the distance dependence may be very weak,the bridge molecules essentially acting as incoherent molecularwires3–6. Here we show how these regimes can be accessed bymolecular design. We have synthesized a series of structurallywell-defined DBA molecules that incorporate tetracene as thedonor and pyromellitimide as the acceptor, linked by p-phenylene-vinylene oligomers of various lengths. Photoinduced electrontransfer in this series exhibits very weak distance dependencefor donor–acceptor separations as large as 40 A, with rate con-stants of the order of 1011 s−1. These findings demonstrate theimportance of energy matching between the donor and bridgecomponents for achieving molecular-wire behaviour.

The distance dependence of electron transfer rate constants, kET, isoften described by kET ¼ k0expð 2 bRDAÞ, where k0 is a kinetic pre-factor, RDA is the donor–acceptor centre-to-centre distance, and b isa factor that depends primarily on the nature of the bridge molecule.For instance, b values of 1.0–1.4 A−1 for proteins7,8, <0.2–1.4 A−1

for DNA9,10, 0.8–1.0 A−1 for saturated hydrocarbon bridges11,12,and 0.2–0.6 A−1 for unsaturated phenylene13,14, polyene15–18, andpolyyne19–21 bridges have been reported. Molecules 1–5 (see Fig. 1)are designed to explore the structural and energetic criteria foraccessing the weakly distance-dependent molecular-wire regime.The molecular structures of 1–5 were fully characterized by protonNMR, mass spectral analysis, and ultraviolet-visible spectroscopy.Figure 2a shows the ground-state absorption spectra of 3, 5-(p-tolyl)-TET, and PPV3-PI (here TET indicated tetracene, PI ispyromellitimide, and PPV 1–5 represent p-phenylenevinylene oli-gomers of increasing length). The electronic absorption spectrum ofTET within 1–5 remains the same, indicating that the PPV bridge

N N

O

O

O

O

C8H17

3.

OR

DONOR ACCEPTOR

WIRE

WIRE = 1.

2.

4.

5.

R = 2-ethylhexyl

ORRO

ORRO

RO

Figure 1 Structures of molecules 1–5.

Nature © Macmillan Publishers Ltd 1998

8

letters to nature

NATURE | VOL 396 | 5 NOVEMBER 1998 | www.nature.com 61

molecule is weakly coupled electronically to the donor. The fluor-escence quantum yield of TET itself is 0.16 (ref. 22), yet thefluorescence of TET in 1–5 is strongly quenched, when eachmolecule is dissolved in 2-methyltetrahydrofuran (MTHF). Thisresult indicates that the excited state of TET is being deactivated bya new mechanism in 1–5. Reference molecules that lack the PIacceptor do not quench the lowest excited singlet state of TET.

Electron transfer within these compounds was investigated usingfemtosecond optical pump-probe spectroscopy23. A pump laserpulse (wavelength 485 nm, length 150 fs) was used to selectivelyexcite TET within 1–5 to its lowest excited singlet state. Transientspectra were measured with an overall 180-fs instrument responseusing a white-light continuum probe pulse23. These spectra alldevelop a broad peak at 610 nm and sharper peaks at 720 and870 nm as a function of time (Fig. 2b). The 720-nm peak is assignedto PI− on the basis of previous spectroelectrochemical measure-ments24, while the 610-nm and 870-nm features are characteristic ofthe TETradical cation25. Thus, the fluorescence quenching observedin these electron transfer systems is a consequence of electrontransfer from 1*TET to PI. The measured charge separation (CS)and charge recombination (CR) rate constants for 1–5 are given inTable 1. All of the formation dynamics were fitted to an instrument-limited rise time due to the initial formation of 1*TET, and anexponential rise due to CS. The CR dynamics for 1–4 were fittedusing a single exponential decay, while for 5 a reasonable fit to thedata required a sum of two exponentials (Table 1).

The values of RDA listed in Table 1 were obtained from structuresoptimized using the MM2 force field. In these structures thedihedral angle between TET and the phenyl group connecting itto the bridge molecules is 698, while that between the bridge phenyland PI is 548. The dependence of the CS rate constants on increasingbridge length is given in Fig. 3a, where ln kCS versus RDA is shown for1–5 in MTHF. There appear to be two electron-transfer regimes inthese molecules. For the two shortest bridges, in 1 and 2, there is a

sharp decrease in electron-transfer rate constant with distance,which is consistent with previous experimental results for otherconjugated bridges16,18–21. There is an abrupt change in mechanismstarting with 3, where the CS rate constant for 3 is greater than thatfor 1 even though the donor–acceptor distance of 3 is 13 A longerthan that of 1. A least-squares fit for the three longest bridges yields ab value of only 0.04 A−1. The corresponding distance dependence ofthermal CR is shown in Fig. 3b. In addition to the distancedependence of the electronic coupling between the donor and theacceptor, the free energy of reaction (DG) and the solvent reorgan-ization energy (ls) are also distance dependent primarily as aconsequence of the coulombic interaction between the ions. How-ever, when donor–acceptor distances are significantly greater than10 A, as they are in 2–5, the contribution of the distance dependenceof DG and ls to the overall distance dependence of the electron-transfer rate constant is small26. The CR rate constants for 2–5decrease monotonically with distance, if the slower rate constantfrom the bi-exponential data for 5 is used. However, a break in thistrend is observed if the faster rate constant for CR in 5 is considered.

b

a

500 550 600 650 700 750–0.002

0.000

0.002

0.004

0.006

0.008

TET+ PI –

840 860 880 9000.0000.0010.0020.0030.004

350 400 450 500 550 6000.0

0.2

0.4

0.6

0.8

1.0

A (

a.u.

)∆A

Wavelength (nm)

TET+

Figure 2 Optical absorption spectra. a, Ground-state absorption spectra of 3

(solid line), TET (dashed line), and PPV3-PI (dot-dashed line) inMTHF.b, Transient

absorption spectrum of 3 in MTHF taken 150ps after a 485nm,150 fs, 0.5-mJ pulse.

The sample was contained in a cell of 2mm path length within which the pump

and probe beams overlapped in a 200-mm spot. Inset, near-infrared region of the

spectrum at the same time delay. A, Absorbance of the sample.

10 15 20 25 30 35 40

54

32

1

543

2

1

k CR

(s–1

)k C

S (

s–1)

RDA (Å)

1010

1011

1010

1011

1012a

b

Figure 3 Distance dependence of charge separation and recombination rate

constants. a, Plot of ln kCS versus the donor–acceptor distance, RDA in MTHF

(filled circles). A Levenberg–Marquardt nonlinear least-squares fit to the data for

molecules 3–5 is also shown. b, Plot of ln kCR versus RDA in MTHF (filled circles

and filled triangles). Lines are used to connect the points.

Table 1 Data for charge separation and recombination reactions in 1–5

Molecule RDA

(A−1)−DGCS

(eV)−DGCR

(eV)kCS

(s−1)kCR

(s−1).............................................................................................................................................................................

1 11.1 0.84 1.71 1.31×1011 6.21×1010

.............................................................................................................................................................................

2 17.7 0.77 1.78 2.27×1010 1.39×1010

.............................................................................................................................................................................

3 24.3 0.74 1.81 3.88×1011 1.59×1010

.............................................................................................................................................................................

4 30.9 0.72 1.83 2.63×1011 6.37×109

.............................................................................................................................................................................

5 38.0 0.70 1.85 2.18×1011 1.81×1010 (0.65)5.81×109 (0.35)

.............................................................................................................................................................................The free energies of reaction are obtained from the one-electron redox potentials foroxidation of TET (0.77 vs SCE), reduction of PI (−0.79V vs SCE), RDA, and the lowest excitedsinglet state energy of TET (2.55eV) using procedures given in ref. 23. For 5 the relativeamplitudes of the two components of kCR are given in parentheses.

Nature © Macmillan Publishers Ltd 1998

8

letters to nature

62 NATURE | VOL 396 | 5 NOVEMBER 1998 | www.nature.com

The data suggest that CR in 5 proceeds by two different mechanisms(see below).

The importance to the efficiency of charge injection into the wireof nearly matching the energy of the highest occupied electronicstate of the donor to the lowest unoccupied state of the molecularwire has long been recognized27–29. Yet, experimental demonstra-tions of this phenomenon have been lacking. As the energy gapbetween the highest occupied state of the donor (usually thought ofas the Fermi level of a metallic contact) and an empty eigenstate ofthe bridge decreases, the rate of charge conduction through the wireincreases rapidly. We may apply a similar analysis to the DBAmolecules studied here by examining the changes in the CS rates asthe relevant injection energy gap between the donor and PPV bridgeis modified by the length and electronic substitution pattern alongthe oligomeric PPV backbone. At a simple level of analysis, theelectron transfer event can be envisioned as the movement of anelectron from the lowest unoccupied molecular orbital (LUMO) ofthe photoexcited donor to the LUMO of the acceptor, via the LUMOof the bridge. Figure 4 is a plot of the bridge highest occupiedmolecular orbital (HOMO) and LUMO energies as a function ofbackbone length, as well as those of TET. In 1 and 2 the LUMO–LUMO energy gap of the donor and bridge is substantially largerthan that in the three longer bridges. Lengthening the bridge in 3, 4and 5 causes the LUMO energy of the bridge to approach that ofTET to within 0.1 eV. In addition to the bridge-length dependence,the appended alkoxy groups on the oligomeric backbone alsochange the HOMO and LUMO energies30.

The changes in the donor-to-bridge injection energy are reflectedvery well in the measured CS kinetics. In 1 and 2, the energy gapbetween the donor and bridge LUMOs is large enough that theelectron-transfer event exhibits the normal superexchange mechan-ism of electron transfer; that is, k1 q k2 in equation (1).

1pDBANk4

k2

DþB 2 A

k1

sssd

ttte

k3 ð1Þ

DþBA 2

In the three longest bridges, this energy gap becomes small enoughthat there is very little energy penalty paid for injection of anelectron from the donor into the bridge, and the rates decrease veryslowly with distance. In 5 the energy gap is so small that the twoLUMOs are almost isoenergetic. While this might be expected tolead to very strong electronic mixing between TETand the bridge in5, the 698 dihedral angle between ET and the first phenyl group ofthe bridge strongly diminishes electronic coupling between TETandthe bridges, as is evidenced by the invariance of the lowest excitedsinglet state energy of TET in the series 1–5. The very small energygap suggests that incoherent electron transfer, in which the electrondensity is temporarily dephased on the bridge3–6, may be a con-tributing electron-transfer pathway in 3–5. Thus in equation (1),effectively k2 q k1 for 3–5 leading to formation of D+B−A,which proceeds rapidly to D+BA− with k3. The measured CS rateconstants monitor the slightly endothermic rate-determining step1*DBA → DþB 2 A. There is no evidence from either the transientspectra or the CS kinetics for a significant build-up of D+B−Apopulation during the course of the reaction; that is, k3 q k2.This is reasonable in view of the fact that the reactionDþB2 A → DþBA 2 is strongly exothermic, and therefore isexpected to be very fast. The mono-exponential nature of theobserved CS kinetics indicates also that k2 q k4.

The CR mechanism may contain contributions from both holeand electron transfer. The LUMO–LUMO energy gap between PIand the PPV bridges within 1–5 is >1 eV, so that CR via electrontransfer from PI− to D+ should proceed slowly by the superexchangemechanism. In addition, the HOMO–HOMO energy gap betweenTETand the PPV bridges within 1–4 is >0.2 eV, so that CR via holetransfer should also be dominated by superexchange; that is k5 q k6

in equation (2).

DþBA2Nk8

k6

DBþA 2

k5

sssd

ttte

k7 ð2Þ

DBA

On the other hand, the HOMO–HOMO energy gap between TETand the PPV bridge within 5 is sufficiently small to make holeinjection from TET+ into PPV5 kinetically competitive with super-exchange (k5 < k6) by means of the additional CR mechanismDþBA 2 →k6

DBþA 2 →k7DBA, where k7 q k6. The bi-exponential

CR kinetics observed for 5 suggests that k6 is comparable to k8.These results suggest that the lifetime of the D+BA− state could besignificantly lengthened by increasing the HOMO–HOMO energygap between D and B. This hole-injection CR mechanism isanalogous to the electron-injection mechanism proposed for CSin 3–5, and highlights the importance of the energy-matchingrequirements in promoting the molecular-wire behaviour of thebridge molecule.

We have produced a family of DBA molecules using oligomers ofthe conducting polymer PPV as the bridging motif. Photoinduced

–1.8

–1.6

–1.4

–1.2

–1.0

–0.8

1.8

2.0

2.2

2.4

2.6

2.8

3.0

PPV1

PPV2

TET

TET

PPV3PPV4PPV5

PPV5

PPV4PPV3

PPV2

PPV1

E (e

V)

Figure 4 The LUMO (solid lines) and HOMO (dashed lines) energies of TETand

the five bridge molecules. These energies were derived from the experimental

spectroscopic and electrochemical properties of the individual chromophores.

For the PPV3, PPV4 and PPV5 bridges, all-trans hydrogen-terminated bridge

molecules were synthesized. The oxidation potentials of TET, PPV1 (benzene),

PPV2 (stilbene), PPV3, PPV4 and PPV5 in butyronitrile with 0.1M tetra-n-

butylammonium perchlorate are 0.77, 1.80, 1.32, 1.01, 0.96 and 0.94V versus SCE,

respectively, while their lowest excited singlet state energies are 2.55, 4.82, 3.60,

3.01, 2.90 and 2.75 eV, respectively, as measured by techniques given in ref. 23.

The sum of the oxidation and reduction potentials for each bridge should be

approximately the same as the lowest excited singlet state energy of the bridge.

Using this relationship, we equate the HOMO energy of a bridge molecule to the

negative of its oxidation potential, and calculate its LUMO energy from the

difference between its lowest excited singlet state energy and its oxidation

potential.

Nature © Macmillan Publishers Ltd 1998

8

letters to nature

NATURE | VOL 396 | 5 NOVEMBER 1998 | www.nature.com 63

charge separation in these systems is very weakly distance depen-dent, indicating that the unsaturated bridge acts as an incoherentmolecular wire. If exponential distance dependence is assumed,the b value measured is amongst the lowest found to date. Moreimportantly, we have presented experimental evidence for thecritical role that the energy gap for donor-to-bridge charge injectionplays in promoting molecular-wire behaviour in DBA moleculesthat possess highly conjugated bridges. Given that PPV polymersdisplay unusually high electron and hole mobilities31, the abilityto exercise photochemical control over charge injection into well-defined PPV oligomeric structures may lead to molecular electronicdevices based on these materials. M

Received 22 June; accepted 14 August 1998.

1. Barbara, P. F., Meyer, T. J. & Ratner, M. A. Contemporary issues in electron transfer research. J. Phys.Chem. 100, 13148–13168 (1996).

2. Wasielewski, M. R. Photoinduced electron transfer in supramolecular systems for artificialphotosynthesis. Chem. Rev. 92, 435–461 (1992).

3. Ratner, M. A. & Jortner, J. in Molecular Electronics (eds Jortner, J. & Ratner, M.) 5–72 (Blackwell,Oxford, 1997).

4. Davis, W. B., Wasielewski, M. R., Ratner, M. A., Mujica, V. & Nitzan, A. Electron transfer rates inbridged molecular systems: A phenomenological approach to relaxation. J. Phys. Chem. A 101, 6158–6164 (1997).

5. Okada, A., Chernyak, V. & Mukamel, S. Solvent reorganization in long-range electron transfer: densitymatrix approach. J. Phys. Chem. A 102, 1241–1251 (1998).

6. Pollard, W. T., Felts, A. K. & Friesner, R. A. The Redfield equation in condensed-phase quantumdynamics. Adv. Chem. Phys. 93, 77–134 (1996).

7. Winkler, J. R. & Gray, H. B. Electron transfer in ruthenium-modified proteins. Chem. Rev. 92, 369–379 (1992).

8. Moser, C. C., Keske, J. M., Warncke, K., Farid, R. S. & Dutton, P. L. Nature of biological electrontransfer. Nature 355, 796–802 (1992).

9. Lewis, F. D. et al. Distance-dependent electron transfer in DNA hairpins. Science 277, 673–676 (1997).10. Holmlin, R. E., Dandliker, P. J. & Barton, J. K. Charge transfer through the DNA base stack. Angew.

Chem. Int. Edn. Engl. 36, 2715–2730 (1998).11. Johnson, M. D., Miller, J. R., Green, N. S. & Closs, G. L. Distance dependence of intramolecular hole

and electron transfer in organic radical ions. J. Phys. Chem. 93, 1173–1176 (1989).12. Oevering, H. et al. Long-range photoinduced through-bond electron transfer and radiative recom-

bination via rigid nonconjugated bridges: distance and solvent dependence. J. Am. Chem. Soc. 109,3258–3269 (1987).

13. Helms, A., Heiler, D. & McLendon, G. Electron transfer in bis-porphyrin donor-acceptor compoundswith polyphenylene spacers shows a weak distance dependence. J. Am. Chem. Soc. 114, 6227–6238(1992).

14. Ribou, A.-C. et al. Intervalence electron transfer in pentaammineruthenium complexes of dipyridyl-polyenes, dipyridylthiophene, and dipyridylfuran. Inorg. Chem. 33, 1325–1329 (1994).

15. Arrhenius, T. S., Blanchard-Desce, M., Dvolaitzky, M., Lehn, J. M. & Malthete, J. Molecular devices:caroviologens as an approach to molecular wires—synthesis and incorporation into vesiclemembranes. Proc. Natl Acad. Sci. USA 83, 5355–5359 (1986).

16. Wasielewski, M. R. et al. in Photochemical Energy Conversion (ed. Norris, J. R.) 135–147 (Elsevier, NewYork, 1989).

17. Effenberger, F. & Wolf, H. C. Terminally substituted conjugated polyenes: synthesis and energytransfer properties. New J. Chem. 15, 117–123 (1991).

18. Tolbert, L. M. Solitons in a box: the organic chemistry of electrically conducting polyenes. Acc. Chem.Res. 25, 561–568 (1992).

19. Tour, J. M. Conjugated macromolecules of precise length and constitution. Organic synthesis for theconstruction of nanoarchitectures. Chem. Rev. 96, 537–553 (1996).

20. Grosshenny, V., Harriman, A. & Ziessel, R. Towards the development of molecular wires: electronlocalization, exchange, and transfer in alkyne-bridged multinuclear complexes. Angew. Chem. Int.Edn. Engl. 34, 2705–2708 (1996).

21. Sachs, S. B. et al. Rates of interfacial electron transfer through p-conjugated spacers. J. Am. Chem. Soc.119, 10563–10564 (1997).

22. Burgdorff, C., Ehrhardt, S. & Loehmannsroeben, H. G. Photophysical properties of tetracenederivatives in solution. 2. Halogenated tetracene derivatives. J. Phys. Chem. 95, 4246–4249 (1991).

23. Greenfield, S. R., Svec, W. A., Gosztola, D. & Wasielewski, M. R. Multistep photochemical chargeseparation in rod-like molecules based on aromatic imides and diimides. J. Am. Chem. Soc. 118, 6767–6777 (1996).

24. Viehbeck, A., Goldberg, M. J. & Kovac, C. A. Electrochemical properties of polyimides and relatedimide compounds. J. Electrochem. Soc. 137, 1460–1466 (1990).

25. Perkampus, H.-H. UV-VIS Atlas of Organic Compounds (VCH, Weinheim, 1992).26. Marcus, R. A. On the theory of oxidation-reduction reactions involving electron transfer. I. J. Chem.

Phys. 24, 966–978 (1956).27. Mujica, V., Kemp, M., Roitberg, A. & Ratner, M. Current-voltage characteristics of molecular wires:

Eigenvalue staircase, Coulomb blockade, and rectification. J. Chem. Phys. 104, 7296–7305 (1996).28. Joachim. C. & Vinuesa, J. F. Length dependence of the electronic transparence (conductance) of a

molecular wire. Europhys. Lett. 33, 635–640 (1996).29. Samanta, M. P., Tian, W., Datta, S., Henderson, J. I. & Kubiak, C. P. Electronic conduction through

organic molecules. Phys. Rev. B: Condens. Matter 53, R7626–R7629 (1996).30. Burn, P. L. et al. Chemical tuning of the electronic properties of poly(p-phenylenevinylene)-based

copolymers. J. Am. Chem. Soc. 115, 10117–10124 (1993).31. Hoofman, R. J. O. M., De Haas, M. P., Siebbeles, L. D. A. & Warman, J. M. Highly mobile electrons and

holes on isolated chains of the semiconducting polymer poly(phenylene vinylene). Nature 392, 54–56(1998).

Acknowledgements. This work was supported by the Division of Chemical Sciences, Office of BasicEnergy Sciences, Department of Energy (M.R.W.), the Chemistry Division of the NSF and ONR (M.A.R.),and the Link Energy Foundation (W.B.D.).

Correspondence and requests for materials should be addressed to M.R.W. (e-mail: wasielew@chem.nwu.edu).

A largesourceof atmosphericnitrousoxide fromsubtropicalNorthPacificsurfacewatersJohn E. Dore*†, Brian N. Popp†‡, David M. Karl†& Francis J. Sansone†

* Aquasearch, Inc., 73-4460 Queen Kaahumanu Highway, Suite 110,Kailua-Kona, Hawaii 96740, USA† School of Ocean and Earth Science and Technology, Department ofOceanography, University of Hawaii, Honolulu, Hawaii 96822, USA‡ School of Ocean and Earth Science and Technology, Department of Geology andGeophysics, University of Hawaii, Honolulu, Hawaii 96822, USA. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Nitrous oxide (N2O), a trace gas whose concentration is increasingin the atmosphere, plays an important role in both radiativeforcing and stratospheric ozone depletion1,2. Its biogeochemicalcycle has thus come under intense scrutiny in recent years.Despite these efforts, the global budget of N2O remains unre-solved, and the nature and magnitude of the sources and sinkscontinue to be debated3–5 despite the constraints that can beprovided by characterizations of the gas6,7. We report here theresults of dual-isotope measurements of N2O from the watercolumn of the subtropical North Pacific Ocean. Nitrous oxidewithin the lower-euphotic and upper-aphotic zones is depleted inboth 15N and 18O relative to its tropospheric and deep-oceancomposition. These findings are consistent with a prediction,based on global mass-balance considerations, of a near-surfaceisotopically depleted oceanic N2O source4. Our results indicatethat this source, probably produced by bacterial nitrification,contributes significantly to the ocean–atmosphere flux of N2O inthe oligotrophic subtropical North Pacific Ocean. This source mayact to buffer the isotopic composition of tropospheric N2O, and isquantitatively significant in the global tropospheric N2O budget.Because dissolved gases in near-surface waters are more readilyexchanged with the atmospheric reservoir than those in deepwaters, the existence of a quantitatively significant N2O source at arelatively shallow depth has potentially important implicationsfor the susceptibility of the source, and the ocean–atmosphereflux, to climatic influences.

Field work was carried out at the Hawaii Ocean Time-seriesStation ALOHA8 (228 459 N, 1588 W), a representative oligotrophicdeep-ocean site. We collected seawater samples for isotopic analysesduring the HOT-76 (September–October 1996), HOT-79 (January1997) and HOT-82 (April 1997) cruises. Two archived seawatersamples from the HOT-65 (September 1995) cruise were alsoanalysed to ensure the efficacy of sample storage. We also analysedair samples collected during HOT-76 and from the roof of ourHonolulu laboratory. Dissolved N2O concentrations ranged fromnear air saturation (,6 nM) at the surface to a maximum of nearly50 nM within the deep oxygen minimum at 700–800 m depth(Fig. 1). Concentration profiles were similar for HOT-76 andHOT-79, but showed some increased values below 100 m duringHOT-82. Although the d15N and d18O values of dissolved N2Odisplay some variability between cruises, all three data sets showsimilar trends with depth: values near those of air at the surface,decreasing with depth to a minimum between 100 and 300 m, andthen an increase with depth to 800 m (Fig. 1). The previouslyundescribed minima in the isotopic ratios differ from surfacevalues by as much as 3.3‰ for d15N and 5.7‰ for d18O. Becauseboth isotopic ratios are lighter within this zone than above or below,the N2O in this zone cannot result from a simple mixture ofatmospheric N2O with N2O from deep water. The isotopic-mini-mum layer is located below the local salinity maximum and well