You are on page 1of 6


Synthesis and Characterization of Carbon NanotubeNanocrystal Heterostructures

Sarbajit Banerjee and Stanislaus S. Wong*,,
Department of Chemistry, SUNY at Stony Brook, Stony Brook, New York 11794, and Materials and Chemical Sciences Department, BrookhaVen National Laboratory, Building 480, Upton, NY 11973
Received October 25, 2001; Revised Manuscript Received December 17, 2001

2002 Vol. 2, No. 3 195-200

Oxidized single-walled carbon nanotubes (SWNTs) have been reacted with cadmium selenide (CdSe) nanocrystals, capped with mercaptothiol derivatives, as well as with titanium dioxide (TiO2) nanocrystals, functionalized with 11-aminoundecanoic acid to form nanoscale heterostructures, characterized by transmission electron microscopy and infrared spectroscopy. The reaction with acid-terminated CdSe nanocrystals and acid-terminated tubes was facilitated with the aid of intermediary linking agents, such as ethylenediamine and semicarbazide, in an amideforming reaction in the presence of 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide hydrochloride, EDC. Based on electronic absorption spectroscopy, charge transfer is thought to proceed from the nanocrystal to the nanotube in the CdSenanotube system, whereas in the TiO2nanotube system, charge transfer is expected to occur from the nanotube to the nanocrystal.

Single-walled carbon nanotubes (SWNTs)1 have been the focus of intensive study due to their unique structuredependent electronic and mechanical properties.2,3 They are thought to have potential applications as catalyst supports in heterogeneous catalysis,4 as high strength engineering fibers,5 and as molecular wires for the next generation of electronics devices.6 SWNTs have been used as molecular wires, due to their long electron mean free paths and their ballistic transport properties.7 Depending on their helicity and diameter, carbon nanotubes can be either metallic or semiconducting. It is also predicted that the band gap of semiconducting nanotubes decreases with increasing diameter.8 Moreover, a single defect can change the structure of a tube from a metallic to a semiconducting variety. A novel strategy of altering the electronic properties of nanotubes is to chemically functionalize9 them with a moiety or structure whose intrinsic properties are electronically configurable. One such structure is the family of semiconductor nanocrystals,10 such as CdS and CdSe, alternatively known as quantum dots, which exhibit strongly sizedependent optical and electrical properties. The high luminescence yield of these materials11 as well as the potential of adjusting emission and absorption wavelengths by selecting for nanocrystal size make quantum dots attractive for constructing optoelectronic devices, for instance, with tailored
* Corresponding author. Phone: 631-632-1703; 631-344-3178. E-mail:; SUNY at Stony Brook. Brookhaven National Laboratory. 10.1021/nl015651n CCC: $22.00 Published on Web 01/12/2002 2002 American Chemical Society

properties. In fact, the electronic structure of semiconductor nanocrystallites exhibits distinctive quantum effects;12 the band gap of these materials increases with decreasing particle size. In this letter, we report the synthesis and characterization of SWNTs covalently joined to CdSe and TiO2 semiconductor nanocrystals by short chain organic molecule linkers. In each case, purified, oxidized nanotubes were reacted with nanocrystals, derivatized with either amine or acid terminal groups (Figures 1 and 2). By judiciously varying the nature of the organic capping groups on the nanocrystal surface and the organic bifunctional linkers, it should be possible to modulate interactions between the nanotubes and the nanocrystals, with implications for self-assembly. Such composites are expected to be useful for applications as diverse as molecular electronics, photocatalysis, solar energy conversion, and as probes for scanning force microscopy. The optical and physical properties of these nanocomposites are currently being actively investigated and will be subsequently reported. Starting materials and reagents were characterized by UV-visible spectroscopy (Beckmann Coulter DU600 and ThermoSpectronics UV1), transmission electron microscopy (TEM, Philips CM12), scanning electron microscopy (SEM, Leo 1550 with a field emission column), and energy-dispersive X-ray spectroscopy (EDS). For microscopy analyses, a portion of the solid was dispersed in ethanol by mild sonication and evaporated over a 300 mesh Cu grid, coated with a lacey carbon film. The nanotube-

Figure 1. Schematic of the addition of titanium dioxide nanocrystals to oxidized single-walled carbon nanotubes (OCNTs). (a) Raw SWNTs (I) were treated in a permanganate-sulfuric acid mixture to create opened, oxidized carbon nanotubes [SWNT-COOH] (II). Amineterminated TiO2 nanoparticles (III) were prepared by the forced hydrolysis of titanium(IV) isopropoxide in the presence of 11-aminoundecanoic acid. (b) (II) and (III) were linked in the presence of EDC to form the adduct (IV). The bonds at the interfaces are not drawn to scale.

nanocrystal adduct was analyzed by TEM, FT-IR spectroscopy (Mattson Galaxy FT-IR 3000), and UV-visible spectroscopy. Raw SWNT bundles (Carbolex; length range of 0.8-1.2 m and widely varying width; SEM images in Figure S1 (Supporting Information)) were thoroughly oxidized according to existing procedures involving acidic potassium permanganate solution and hydrochloric acid.13 This treatment14 has been shown to coat the nanotube sidewalls with keto, carboxylic, aldehyde, and alcoholic groups. TEM analyses indicated that the oxidation process not only removed amorphous carbon but also the majority of metal particles (Figure 3a) as well. The purified tubes were then dried in a vacuum oven at 180 C. Carboxylic acid groups are expected to be the predominant group along the sidewalls.14 This was followed by acid etching in 2.6 M HNO3 to further shorten the tubes. TEM analysis showed that oxidation and etching removed most of the amorphous carbon and metal catalyst particles. The remaining tubes obtained, for the most part, were relatively free of impurities, such as particulates, and either one or both of the tube caps were opened. The tubes were also substantially shortened after acid etching, ranging in length from 100 to 200 nm. Wurtzite CdSe nanocrystals capped with trioctylphosphine oxide (TOPO) were synthesized by existing procedures, using purified dimethylcadmium as a precursor.15 After dissolution in hexane and ethanol, a sharp low energy exciton was centered at 550 nm in the absorption spectrum (Figure 4b) of the orange-red sample. The sample prepared was relatively monodisperse with an average particle size of 32 ( 3 , a figure derived from comparison of our data with experimen196

Figure 2. Schematic of the addition of CdSe nanocrystals to OCNTs. TOPO capped nanocrystals (V) were prepared by established methods using organometallic precursors. (a) TOPO capping was substituted by a thiol ligand to form an acid-terminated CdSe nanocrystal (VI). Substituted thiocarboxylic acids used included p-mercaptobenzoic acid, thioglycolic acid, and 3-mercaptopropionic acid, as represented by different R groups shown. (b) (VI) was linked to OCNTs (II) by an ethylenediamine linker (VII) in the presence of EDC to form the adduct (VIII). The bonds at the interfaces are not drawn to scale.

tal, room-temperature optical absorption spectra from previous reports15,16 and confirmed by TEM data (Figure S2,
Nano Lett., Vol. 2, No. 3, 2002

Figure 3. Transmission electron micrographs taken at 120 kV on a 300 mesh Cu grid with a lacey carbon film. (a) A purified singlewalled carbon nanotube bundle. Scale bar denotes 30 nm. (b) Oxidized carbon nanotube bundle circumscribed by capped TiO2 particles. Scale bar represents 45 nm. (c) Oxidized tubes linked to CdSe nanocrystals capped with 4-mercaptobenzoic acid. The crystallites are concentrated at the functionalized open ends, with some coverage on the sidewalls. Scale bar is 150 nm. (d) Oxidized nanotube linked to CdSe nanoparticles capped with 3-mercaptopropionic acid. Image at high magnification shows prolate nanocrystals scattered along the length of the tube. Scale bar signifies 20 nm.

Supporting Information). The presence of TOPO capping was demonstrated by a strong phosphorus signal in the energy dispersive spectrum. However, TOPO molecules are not strongly bound to the nanocrystals themselves, as they can be washed away by polar solvents such as methanol.15 To create a relatively robust ligand shell, CdSe quantum dots were further derivatized with several other moieties, prepared through a ligand exchange reaction. Thioalcohols and thio acids have been shown to be effective size-regulating and stabilizing agents for cadmium chalcogenide nanoparticles.17,18 Moreover, they strongly attach to the CdSe surface, yielding a uniform surface coverage of passivating caps; indeed, we have taken advantage of the richness of thiol chemistry to vary terminal functional groups at the nanocrystal surface, which is crucial for chemical manipulation of these nanostructures.19-21 To demonstrate the generalizability of our derivatization methodology, we chose a range of different mercaptocarboxylic ligands, of different hydrocarbon chain lengths and composition, to ensure a certain level of solubility in DMF, the reaction medium. These included22 p-mercaptobenzoic acid, thioglycolic acid, and 3-mercaptopropionic acid (AlNano Lett., Vol. 2, No. 3, 2002

drich). The UV-visible spectrum in methanol showed a small blue shift upon capping with thiols in the ligand exchange reaction; the peaks of the optical features were also somewhat broadened, especially with samples capped with the aliphatic mercaptoacid derivatives (Figure 4b). While the former situation indicated a slight decrease in the average size of the crystallites (5-10%) presumably due to the smaller size of these thiols with respect to the starting TOPO molecules, the latter observation indicates an increased broadness in size distribution (standard deviation in size from 8% to 15%), which has been attributed to a loss of some surface Cd and Se species.15 The success of the capping reactions was further confirmed by the elemental signature of sulfur in the EDS spectrum (Figure S3, Supporting Information); the size distribution of the particles, deduced by optical means, was further confirmed by TEM. We attempted capping of CdSe with longer chain mercaptothiols as well, most notably with 11-mercapto1-undecanol and 16-mercaptohexadecanoic acid. These, however, were found not to link particularly well to nanotubes, potentially because of steric and configurational considerations involving the lower numbers of as well as

Figure 4. UV-visible absorption spectra of nanotube-nanocrystal composites. (a) TiO2 nanoparticle covalently joined to a carbon nanotube. The spectrum corresponding to the carbon nanotube (black line) is relatively featureless; that of the TiO2 nanocrystal (blue line) has a peak at 259 nm, whereas the nanotube-nanocrystal adduct product (green line) has peaks at 227, 239, 249, and 288 nm. (b) CdSe nanocrystal covalently linked to a carbon nanotube. The spectrum corresponding to the carbon nanotube (black line) is featureless; that of the CdSe nanocrystal (purple line) has a peak at 550 nm, while the nanotube-nanocrystal adduct product (red line) has peaks at 256 and 274 nm.

the geometric packing of the long alkyl chains, which most likely prevented easy access of a substantial number of reactive surface ligands on the nanocrystals to the tubes themselves. Spherical functionalized titania nanoparticles were prepared by an adaptation of a previously existing procedure.23,24 To summarize, these particles were synthesized by the slow hydrolysis of titanium tetraisopropoxide (Aldrich) in the presence of an alcoholic solution of 11-aminoundecanoic acid (Aldrich). The UV-visible spectra of transparent TiO2 solutions (Figure 4a) in ethanol showed the onset of absorption below 400 nm25,26 and an absorption maximum at 259 nm, which agrees closely with previous reports. Transmission electron and scanning electron microscopy showed that the size of the nanoparticles was 35 ( 5 , consistent with optical data as well. EDS spectroscopy (Figure S3, Supporting Information) confirmed the expected elemental signatures in the capped anastase TiO2 particles. When forming quantum dot-nanotube structures, intermediary linking agents, ethylenediamine (en) and semicarbazide (sc), both of which are diamines, were used to attach the capped CdSe nanoparticles to the nanotubes. In a typical synthesis, the oxidized nanotubes (OCNTs) were dispersed in DMF by sonication for 5 min, after which an aliquot of 0.1 M (2-N-morpholino) ethanesulfonic acid (MES) (Aldrich) in distilled water buffer was added, followed by a few ml of neat en or 0.1 M solution of sc in DMF and 5 mL of 50 mM 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide hydrochloride (EDC) (Aldrich) in DMF. EDC promotes the formation of an amide bond between carboxylic groups on the OCNT

and either en or sc. The reaction mixture was sonicated briefly to maintain a dispersion of nanotubes, stirred vigorously for 48 h under nitrogen, and then filtered over a 0.2 m nylon 6,6 membrane. The resulting solution was washed repeatedly first with ethanol and then water, prior to drying under vacuum for 12 h. The functionalized amine-terminated OCNTs were then soaked and sonicated in DMF with subsequent addition of 50 mM of EDC in DMF to which 0.1 M MES in distilled H2O had been added. The main linking step thereafter to synthesize quantum dot-nanotube heterostructures involved reacting the previously prepared acid-capped CdSe nanocrystal solids with these modified tubes, followed by stirring for 3 days at room temperature. The EDC effectively attached the amine groups, emanating from the functionalized OCNTs, to the carboxylic acid groups, terminating from the surface capping ligands on the CdSe nanocrystal. After filtration over a 0.2 m nylon 6,6 membrane, the solid precipitate was washed repeatedly with water and ethanol; the filtrate was clear and did not show any CdSe absorption peaks, suggesting that the reaction proceeded to completion in the solid phase. The resulting solid was then dried under vacuum for 4 h. An alternative method of creating the nanotube-nanocrystal composite has been to react the mercaptocarboxylic acid capped nanocrystal with ammonium hydroxide at 0 C in a process27 to form an amide terminated system (without the need for intermediary linkers) that can then be directly linked to the OCNTs. A third, equally reliable strategy we used involved initially attaching the acid-terminated nanocrystal to the linker molecule (either en or sc) prior to reaction with the acidterminated OCNTs. In a typical synthesis to create a composite containing carbon nanotubes and titanium oxide nanocrystals using analogous reactions described above, solutions of capped, amine-terminated nanoparticles and oxidized, acid-terminated nanotubes were directly mixed and sonicated in DMF in the presence of 0.1 M MES in H2O and 50 mM EDC in DMF. The reaction mixture was stirred vigorously for 24 h, allowed to settle for an extra 24 h, and ultimately filtered over a 0.2 m nylon 6,6 membrane. The nanocomposite solid was then washed with water and EtOH repeatedly and dried under vacuum. Figures 1 and 2 show stylized schematics of the reactions involving both types of nanocrystals with carbon nanotubes. TEM examination at an accelerating voltage of 120 kV showed nanocrystals linked to the OCNTs, forming discrete nanocomposites. The nanocrystals tend to be concentrated at the open caps and the ends, where there is the largest concentration of carboxylic groups28,29 and thus, the highest probability of amide bond formation.30 A commonly observed structure was a cluster of nanoparticles, especially in the case of CdSe, localized at the outer edges of the tubes (Figure 3c). In other cases, especially with longer and thinner tubes, numbers of nanocrystals tended to disperse along the sidewalls, presumably attaching to defect sites on the oxidized nanotube surface (Figure 3d). The titanium dioxide system showed substantially greater coverage of the tubes with the nanoparticles (Figure 3b). It is known that carboxyl
Nano Lett., Vol. 2, No. 3, 2002

groups adsorb strongly onto bare titanium dioxide surfaces.31 Hence, larger numbers of these particles can interact with nanotubes and in effect, amide bond formation may not be required for an underivatized, uncapped titanium oxideOCNT adduct to be directly produced. Indeed, we postulate that this is the reason for enhanced coverage with TiO2 particles. EDS analysis (Figure S3, Supporting Information) on both nanocomposite systems was consistent with the presence of appropriately functionalized nanocrystals attached to nanotubes as proposed. Estimating the precise stoichiometry of the reactions, though, is difficult due to a lack of knowledge about the exact spatial location of the carboxylic groups on the nanotubes as well as variations in the numbers and types of defect sites, specific to each individual tube. All of these factors do have an influence on the position and configuration of nanocrystals in the resulting nanocomposites. In light of the above, we have at times overestimated the quantity of nanocrystallites required to react with the nanotubes, and the samples show some amount of dispersed unreacted nanocrystals. These patterns of nanocrystal clustering on the nanotubes were not observed in control experiments, without the use of EDC, where the tubes themselves were found to remain totally separate from the nanocrystals. FT-IR spectra provide further evidence for bonding (Figure S4, Supporting Information). The amide bond region shows several sharp and distinctive peaks. The peaks are numerous but sharp and unbroadened and likely correspond to distinctive modes of bonding between the OCNTs and the nanocrystals. All the spectra (KBr pressed pellet) show a peak at 1581 cm-1, corresponding to the IR active phonon mode of nanotubes that has been previously reported.32 In addition, the sharpest peak in all the spectra occurs at 1638 cm-1, located in the amide bond region. We postulate that this does in fact correspond to an amide bond, potentially correlating with unreacted amine groups attached to the modified nanotube ends as well as with amide bonds of the linker groups associated with the capping ligands. Previous observations of amide bonds in carbon nanotubes linked to amines have been noted at 1663 and 1642 cm-1.33 The region of the mid-IR around 1615-1580 cm-1 also has a number of distinctive peaks, which leads us to believe that at least some of the bonding may occur through an additional ionic, noncovalent COO-NH3+ interaction, since previous work has attributed CO in this region to carboxylate anion stretch modes in such interactions.34 Peaks observed in the 3000-2850 cm-1 range are likely a result of C-H stretches within the intermediary linkers used as well as within the aliphatic chains, when aliphatic thiocarboxylic acids were used as capping groups. This assignment is confirmed by the disappearance of these peaks in the specific example of mercaptobenzoic acid, associated with the SWNT-CONH-C6H4S-CdSe adduct, where neither en nor sc was used and the CdSe nanocrystal in that case was effectively capped by an aromatic thioamide. Aliphatic C-H stretches are shifted 10-15 cm-1 from C-H stretches in neat en or sc. Similarly, the C-H bends, noted in en at 1459 cm-1 and in sc at 1438 and 1484 cm-1, are in
Nano Lett., Vol. 2, No. 3, 2002

a range from 1422 to 1469 cm-1 in the adduct. A shift upon product formation is also seen for C-C aromatic stretches in the mercaptobenzoic acid ligand from an initial value of 1554 cm-1 to a range of 1543 to 1578 cm-1 depending on the mode of linkage to the nanotube. These results are consistent with a similar type of interpretation of IR spectra for derivatized SWNTs, which has been previously reported.35 We expect that OCNTs may be able to act as hole scavengers for titanium oxide nanocrystals in an analogous way as other derivatized carboxylic acids have been shown to do (Figure 4a).36 This suggests that photoactive modulation of the band gap on the nanotube should be possible, which could then facilitate reductive processes on the titania surfaces. The absorption spectrum for OCNT is featureless in the 200-800 nm region, while TiO2 nanocrystals show absorption below 400 nm with a peak at 259 nm. The adduct product shows peaks at 227, 239, 249, and 288 nm. While the sharp peak, located at 249 nm, might correspond to unreacted TiO2, we postulate that the broad band centered about 288 nm, coupled with depletion of the band at 259 nm, corresponds to a charge-transfer transition involving electron transfer to the titanium dioxide surface and lattice, thereby suggesting that charge separation has been achieved. For our system, in contrast to titanium dioxide, which has a large band gap and a low lying conduction band, CdSe has a high lying conduction band and a low band gap; hence, it may be expected that charge injection occurs from the nanocrystal to the nanotube in that case.25,31 Evidence for this hypothesis (Figure 4b) comes from the disappearance of the lowest energy exciton in the adduct at around 550 nm, corresponding to the band gap excitation in the initial nanocrystal reagent solution. Indeed, for the adduct, distinctive, intense, and sharp charge transfer bands appear at lower wavelengths (<300 nm), specifically at 256 and 279 nm, and the absorbance spectrum, though featureless, at higher wavelengths is increased from that of individual nanotubes alone. In a few cases, as with the mercaptopropionic acid ligand, the absorption spectra show a small shoulder around 515 nm, characteristic of CdSe itself. As there is still some variability in nanocomposite synthesis, specifically in controlling the precise number of crystallites binding to each nanotube, it prevents a more quantitative description of the depletion of the lower energy states in the semiconductor nanocrystals. Further studies of charge transfer in these adducts will be reported in future work. In this letter, we have reported the synthesis of a number of novel nanostructure composites with the potential for tailoring the electronic properties of SWNTs. The synthesis is robust and we have demonstrated its applicability to a wide variety of organic linkers and capping ligands. Moreover, this synthetic strategy is readily extendable to other chalcogenide as well as other types of nanocrystals such as those derived from metal oxides. Generating a heterojunction incorporating nanotubes and nanocrystals is significant, not only because no such device has been reported previously but also because interesting charge-transfer behavior has been

demonstrated from nanocrystal to nanotube (in the case of CdSe) or from nanotube to nanocrystal (with TiO2). Future work will take advantage of the tunability, in terms of size, shape, and chemistry, of nanotubes and nanocrystals, to create a sharp junction interface, whose properties are manipulable and hence, predictable. For instance, singleelectron charging effects and energy-level or conductance quantization can be controlled to varying degrees at room temperature, and the results can be used to increase understanding of factors affecting carrier mobility, such as electronic structure, carrier trapping, and delocalization. Acknowledgment. We acknowledge support of this work through startup funds provided by the State University of New York at Stony Brook as well as Brookhaven National Laboratory. Acknowledgment is also made to the donors of the Petroleum Research Fund, administered by the American Chemical Society, for support of this research. We also thank Dr. James Quinn for his guidance and help with the SEM and TEM work. Supporting Information Available: (a) SEM images of raw nanotubes, (b) TEM images of unreacted CdSe nanocrystals, (c) EDX data correlating with TEM images (Figure 3) of nanocrystal-nanotube composites observed and with unreacted nanocrystals, and (d) representative FT-IR spectra of these composites are presented. This material is available free of charge via the Internet at References
(1) Iijima, S.; Ichihashi, T. Nature 1993, 363, 603. (2) Dresselhaus, M. S.; Dresselhaus, G.; Eklund, P. C. Science of Fullerenes and Carbon Nanotubes; Academic Press: New York, 1996. (3) Odom, T. W.; Huang, J.-L.; Kim, P.; Lieber, C. M. Nature 1998, 391, 62. (4) Planeix, J. M.; Coustel, N.; Coq, B.; Ajayan, P. M. J. Am. Chem. Soc. 1994, 116, 7935. (5) Calvert, P. Nature 1992, 357, 365. (6) Saito, S. Science 1997, 278, 77. (7) White, C. T.; Todorov, T. N. Nature 1998, 393, 240. (8) Martel, R.; Schmidt, T.; Shea, H. R.; Hertel, T.; Avouris, P. Appl. Phys. Lett. 1998, 73, 2447. (9) Banerjee, S.; Wong, S. S. Nano Lett. 2002, 2, 49. (10) Brus, L. E. Appl. Phys. A 1991, 53, 463. (11) Dabbousi, B. O.; Rodriguez-Viejo, J.; Mikulec, F. V.; Heine, J. R.; Mattoussi, H.; Ober, R.; Jensen, K. F.; Bawendi, M. G. J. Phys. Chem. B 1997, 101, 9463. (12) Alivisatos, A. P. J. Phys. Chem. 1996, 100, 13226. (13) Hiura, H.; Ebbesen, T. W.; Tanigaki, K. AdV. Mater. 1995, 7, 275. (14) Bond, A. M.; Miao, W.; Raston, C. Langmuir 2000, 16, 6004. (15) Murray, C. B.; Norris, D. J.; Bawendi, M. B. J. Am. Chem. Soc. 1993, 115, 8706. (16) Peng, X.; Wickham, J.; Alivisatos, A. P. J. Am. Chem. Soc. 1998, 120, 5343. (17) Vossmeyer, T.; Katsikas, L.; Giersig, M.; Popovic, I. G.; Diesner, K.; Chemseddine, A.; Eychmuller, A.; Weller, H. J. Phys. Chem. 1994, 98, 7665.

(18) Rogach, A. L.; Kornowski, A.; Gao, M.; Eychmuller, A.; Weller, H. J. Phys. Chem. B 1999, 103, 3065. (19) Swayambunathan, V.; Hayes, D.; Schmidt, K. H.; Liao, Y. X.; Meisel, D. J. Am. Chem. Soc. 1990, 112, 3831. (20) Peng, X.; Wilson, T. E.; Alivisatos, A. P.; Schultz, P. G. Angew. Chem., Int. Ed. Engl. 1997, 36, 145. (21) Eychmuller, A.; Rogach, A. L. Pure Appl. Chem. 2000, 72, 179. (22) Chen, C.-C.; Yet, C.-P.; Wang, H.-N.; Chao, C.-Y. Langmuir 1999, 15, 6845. In a typical reaction, carried out on a standard Schlenk apparatus under an inert atmosphere, a few mg of TOPO capped nanocrystallites and 20 mg of the chosen mercaptocarboxylic acid were added together in methanol with stirring, after which an aliquot of 0.1 N tetramethylammonium hydroxide in propanol was dripped in carefully until the pH of the resulting solution was close to 10. Thereafter, the solution was heated at 57 C for 6-12 h until it became optically clear. The nanocrystals were precipitated by gradual dropwise addition of THF, purified by repeated dissolution in methanol, and then, reprecipitated in THF. The resulting solid was obtained by filtering over a 0.2 m nylon 6,6 membrane. (23) Cassagneau, T.; Fendler, J. H.; Mallouk, T. E. Langmuir 2000, 16, 241. Specifically, a millimolar ethylene glycol solution of 11aminoundecanoic acid was heated to 70 C for a few minutes, after which a millimolar solution of 2-propanol containing titanium(IV) isopropoxide was added under vigorous stirring. Distilled water (0.5 mL) was added slowly to complete the reaction; stirring was continued at room temperature for a further hour. The carboxylic acid moiety of 11-aminoundecanoic acid tends to strongly interact with the positively charged surface of TiO2 during the hydrolysis and condensation reaction, yielding amine-terminated titania particles. The colloidal suspension was filtered over a 0.2 m nylon 6,6 membrane; the yield for the reaction was initially low, as the source of titanium was not converted into well-dispersed TiO2 nanoparticles. However, allowing the supernatant to settle for a week followed by centrifugation at 9000 rpm for 40 min permitted the isolation of a very stable colloidal dispersion of titanium dioxide. (24) Kotov, N. A.; Dekany, I.; Fendler, J. H. J. Phys. Chem. 1995, 99, 13065. (25) Hao, E.; Yang, B.; Zhang, J.; Zhang, X.; Sun, J.; Shen, J. J. Mater. Chem. 1998, 8, 1327. (26) Colvin, V. L.; Goldstein, A. N.; Alivisatos, A. P. J. Am. Chem. Soc. 1992, 114, 5221. (27) March, J. AdVanced Organic Chemistry; John Wiley and Sons: New York, 1992. (28) Ajayan, P. M.; Ebbesen, T. W.; Ichihashi, T.; Iijima, S.; Tanigaki, K.; Hiura, H. Nature 1993, 361, 333. (29) Liu, J.; Rinzler, A. G.; Dai, H.; Hafner, J. H.; Bradley, R. K.; Boul, P. J.; Lu, A.; Iverson, T.; Shelimov, K.; Huffman, C. B.; RodriguezMacias, F.; Shon, Y. S.; Lee, T. R.; Colbert, D. T.; Smalley, R. E. Science 1998, 280, 1253. (30) Wong, S. S.; Joselevich, E.; Woolley, A. T.; Cheung, C. L.; Lieber, C. M. Nature 1998, 394, 52. (31) Lawless, D.; Kapoor, S.; Meisel, D. J. Phys. Chem. 1995, 99, 10329. (32) Mawhinney, D. B.; Naumenko, V.; Kuznetsova, A.; Yates, J. T., Jr.; Liu, J.; Smalley, R. E. J. Am. Chem. Soc. 2000, 122, 2383. (33) Chen, J.; Hamon, M. A.; Hu, H.; Chen, Y.; Rao, A. M.; Eklund, P. C.; Haddon, R. C. Science 1998, 282, 95. (34) Chen, J.; Rao, A. M.; Lyuksyutov, S.; Itkis, M. E.; Hamon, M. A.; Hu, H.; Cohn, R. W.; Eklund, P. C.; Colbert, D. T.; Smalley, R. E.; Haddon, R. C. J. Phys. Chem. B. 2001, 105, 2525. (35) Hamon, M. A.; Chen, J.; Hu, H.; Chen, Y.; Itkis, M. E.; Rao, A. M.; Eklund, P. C.; Haddon, R. C. AdV. Mater. 1999, 11, 834. (36) Thurnauer, M. C.; Rajh, T.; Tiede, D. M. Acta Chem. Scand. 1997, 51, 610.



Nano Lett., Vol. 2, No. 3, 2002