atX= 7 5 km andX= 90 km 9. The heterogene~ty w t h n the Franciscanaccretonary terranes [J A. Goff and A. R. Levander, J Geophys Res. 101,8489 (1 996)] is not resolved. Upper crustal velocities change (by 0 1 to 0.2 km/s) across the SAF, representng the difference between an old ac- cretionary wedge and terranes involved n subduc- tion untl 2 to 3 Ma. 10. Apparent lower-crustal offsets n reflect~ondata could be caused by basns near the fault zones or by changes in crustal velocit~es across the faults. How- ever, to explain travel-time anomalies n w~de-angle reflections and refractions by shallow structure would requre basins offset from the faults, because these arrivals travel laterally through the crust. The s~mplest explanation for all the observations is that the anomalies arse near the base of the crust. 11. The kinked velocty model has a Moho slope of 10" west of the surface trace of the SAF and a Moho slope of 5" east of the SAF surface trace, so that the knk descrbes the pont where the slope changes The depths of the Moho at the endpoints of the slopes are: 12 km at X = -77 km, 21 km at X = -20 km, and 29 km at X = 80 km. 12. The rms travel-tme misf~t for 3100 lower crustal and upper mantle pcks is 164 ms for the stepped model, with areduced x20f 1.4,compared w~th 186 msand 1.9 for the kinked model. Most of the addtonal error arises In the onshore-offshore P , phase. 13 Depth mgraton focuses reflected energy at the re- flector locat~on [W. A. Schne~der, Geophysics 43, 49, (1978)l. It 1s routinely appl~ed to near-vert~cal data, our appl~cation to onshore-offshore data links marine reflecton data and a single-fold land secton. The d~fferent data types cause changes n appear- ance of the section at X = -30 km and X = 1 km on the profle. 14. K. P. Furlong, W D. Hugo, G. Zandt, J. Geophys Res. 94, 3100 (1 989) 15. T. M. Brocher et a/. , Science 265, 1436 (1 994). 16. W S. Hobrook, T. M Brocher, U. S, ten Brink, J. A. Hole, J. Geophys. Res. 101, 2231 1 (1 996). 17. A. S. Meltzer and A R. Levander, ibid 96, 6475 (1 991); J. M Howe, K. C. Mller, W. U. Savage, ibid. 98, 8173 (1993); K. C Miller, J M Howie, S. D. Ruppert, /bid. 97, 19961 (1 992); C. F. Lafond and A. Levander, ibid. 100, 22231 (1 995). 18. H. M. Kesey and G A Carver,ibid. 93,4797 (I 988); D. A. Castillo and W L. Ellsworth, /bid 98, 6543 (1 993). 19. B. M. Page and T. M. Brocher, Geoiogy 21, 635 (1 993). 20. R. G. Bohannon and T. Parsons, Geol. Soc. Am. Buil. 107, 937 (1 995) 21. R. C. Jachens, A. Griscom, C. W. Roberts, J. Geo- phys. Res. 100, 12769 (1 995). 22. Th~s is withn the limits of DeMets et ai. (7) but IS hgher than the 95% mits of 1 to +5 mm/year for present shortening of Argus and Gordon (1). 23. A J. Cavert, Can. Jour. Earth Sci. 33, 1294 (1 996). 24. S W Smth, J. S Knapp, R. C McPherson, J. Geo- phys. Res. 98, 8153 (1 993). 25. Assumptons that ad~abatic upwelling of mantle n t o the slab w~ndow extends to the base of the accre- tionary rocks and f~llsthe space vacated by the Gordaslab allow us to use published calculat~ons [D. McKenze and M J. Bcke, J. Petrol. 29, 625 ( I 988)] to estimate the melt generated. For 1300°C mantle, this sabout 2 km, and to explan the thickness of the LCL would require temperatures 100" to 150°C h~gher Petrolog~c consequences of h~gher mantle temperatures are discussed by McKenzie and Bickle. 26 To matchthe wide-angle velocity structure and near- verica reflectiv~ty requires thin bodes with too great a matera contrast for a sold-solid contact so that fluids must be present. Our nterpretation that these represent melt ntrusions is based on the tectonic , setting [A. Levander et a / , in preparation]. 27. We are grateful to the many people who assisted n the 1993 and 1994 f~eld programs, the IRIS- PASSCAL program, and Lamont-Doherty Earth Observatory. Earthquake locations were obtained from the northern California earthquake network operated by the U.S Geological Survey and the Seismographic Staton of U.C. Berkeley. The seis- was funded by Natona Science Foundation Con- mic data are avalable through the IRIS Data Manage- tinental Dynamics program grants EAR-9218968 ment Center We thank other members of the MTJ and EAR-952693. sesmic experiment workng group and R. G. Gor- don for comments on this manuscript This study 16 July 1997, accepted 3 October 1997 Synthesis of Nanoparticles and Nanotubes with Well-Separated Layers of Boron Nitride and Carbon K. Suenaga, C. Colliex," N. Demoncy, A. Loiseau, H. Pascard, F. Willaime Polyhedral and tubular graphitic nanoparticles made of carbon layers and boron nitride (BN) layers have been synthesized. These particles were observed in the soot collected on the anode deposit formed by arcing a hafnium diboride rod with graphite in a nitrogen atmosphere. Elemental profiles with subnanometer-scale resolution revealed a strong phase separation between BN layers and carbon layers along the radial direction. Most of these tubes have a sandwich structure with carbon layers both in the center and at the periphery, separated by a few BN layers. This structure provides insight into the atomistic mechanism of nanotube growth in the boron-carbon-nitrogen ternary system and may lead to the creation of nanostructured electronic devices relying on the con- trolled production of heteroatomic nanotubes. T h e family of graphitic nanoparticles with tubular or spherical shape has expanded rapidly slnce the discovery of carbon nano- tubes 11) and carbon onions 12)- which . , , ,, consist of a few concentric cylindrical or s~herical carbon lavers. Their vure BN an- alogs have now been successiully synthe- sized 13-5). The electronic orooerties of , , . . these nanoparticles open up new possibili- ties for makine nanoscale electronic devic- - es, in particular from the tubular form (6). On the basis of theoretical vredictions sue- - gesting that the electronic properties of car- bon nanotubes wlll range from metallic to semiconducting with a small gap, depend- ing on the tube diameter and chirality (7), the idea of making electronic switches by connecting Dure carbon nanotubes was flrst - proposed (8). This concept was recently generalized to heterojunctions between K. Suenaga, Laboratoire de Physique des Solides, URA 002, Un~vers~te de Paris-Sud, Bgt. 510, 91405 Orsay, France. C. Coll~ex, Laboratore de Physique des Sol~des, URA 002, Un~vers~te de Paris-Sud, B2.t. 510, 91405 Orsay, France, and Laboratore Aime Cotton, UPR 3321, Cam- pus d'orsay, B8t. 505, 91405 Orsay, France. N. Demoncy, Laboratoire des Solides Irradies, CEA- CNRS, Ecole Polytechnque, 91 128 Paaiseau Cedex, France. and Laborato~re de Phvs~aue des Solides. Offce B,C,N, nanotubes with different chemical compositions (9). The advantage of such nanotubes is that their electronic properties are primarily determined by composition and are thus relatively easy to control. For example, BN nanotubes are predicted to be semiconducting, with a w~de gap close to the 5.8-eV gap of bulk hexagonal BN (10). In this context, uniformly doped carbon nanotubes, as well as nanotubes with other chemical compos~tions such as BC,N or BC, (1 1), would be interesting for their electronic properties. However, their syn- thesis has not yet been achieved in a con- trollable fashion, although portions of such tubes ainong a majority of pure carbon tubes have been reported (1 2-1 4 ) . Here, we report the synthesis of a soot containing polyhedral and tubular nanopar- ticles that consist of well-separated BN lay- ers and carbon layers. In previous arc-dis- charge syntheses of B-C-N tubes, it was shown that when the anode contains car- bon, the graphitic products contain mostly carbon (12-14). Either a low doping by B and N (less than 2%) or very low concen- trations of B-rich or BN-rich nanotubes have been reported. In our study, this draw- back was overcome bv mod~fvine the oripi- , " atonal d3Etudes et de ~echerches ABrospatiales, BP rial geometry used to ~rodllce pure BN 72, 92322 Ch8tilon, France. A. Loiseau, Laboratoire de Physque des Sodes, Offce (5). A graphite cathode-used instead of National d3Etudes et de Recherches ABros~atales, BP the original HfB, rod-is arced wlth an 72, 92322 Ch8tlon, France. HfB, anode in a-nitrogen atmosphere. In H. Pascard, Laborato~redes Sodes Irradies, CEA- CNRS, Ecole Polytechnque, 91 128 Palaseau, France. the present configurationl the three F Wlaime, Section de Recherches de Metallurge ~ h y - uents have different sources: the anode for slque, CEA/Saclay, 91 191 Gif-sur-Yvette, France. boron, the cathode (which slightly vaporiz- *To whom correspondence should be addressed. es in the present case) for carbon, and the ncemag.org SCIENCE VOL 278 24 OCTOBER 1997 653