NANO LETTERS
Catalytic and Catalyst-free Synthesis of CdSe Nanostructures with Single-Source Molecular Precursor and Related Device Application
2009 Vol. 9, No. 1 437-441
Youxiang Zhang, Yun Tang, Kwan Lee, and Min Ouyang* Center for Nanophysics and AdVanced Materials and Department of Physics, UniVersity of Maryland, College Park, Maryland 20742 Received November 6, 2008
ABSTRACT Air-stable single-source molecular precursor is applied for controlled size and morphology synthesis of one-dimensional and quasi-onedimensional CdSe nanostructures. Two different growth approaches are compared to control the growth of nanostructures. When combined with well-defined Au colloidal catalysts, the use of single-source molecular precursor allows diameter control synthesis of monodispersed CdSe nanowires from 10-30 nm via a vapor-liquid-solid mechanism. In addition, a variety of CdSe nanostructures with different morphologies can be achieved and tuned without assistance of metallic catalysts by carefully manipulating dynamic thermal decomposition process of single-source molecular precursor. The new level of synthetic control afforded by our present work opens up new opportunities for using as-synthesized CdSe nanostructures as model systems for fundamental studies as well as building blocks for larger scale functional device assembly. Importantly, we demonstrate that a single CdSe tripod can be natively configured as a nanoscale phototransistor in which photocurrent created between two tripod arms can be efficiently modulated by applying a gate voltage through the third arm.
Semiconductor nanostructures, including quantum dots, nanowires, nanobelts and nanotubes, have attracted significant attention over the past decade due to their unique mechanical, electronic, optical, and magnetic properties and represent a broad class of building blocks for functional device architectures.1-7 Critical to their fundamental properties and device functionality are the precise morphology and dimensionality control of semiconductor nanostructures, which has been a forefront research area and will lead to big impact on progress of nanoscience and nanotechnology. One of the examples is provided by the extensive studies of cadmium selenide (CdSe). CdSe is an important group II-VI direct band gap semiconductor material with attractive optoelectronic properties.8,9 Different solution- and gas-phase synthetic methods have been applied to control the shape and the size of zero-dimensional (0D), one-dimensional (1D), and quasi-1D CdSe nanostructures,10-17 whose physical properties can be engineered by the introduction of quantum confinement leading to a wide range of applications in electronics,18 biosensing/bioimaging,19 spintronics,20 and so forth. Herein we report for the first time the controlled synthesis of a variety of 1D and quasi-1D CdSe nanostructures by using single-source molecular precursor based on * To whom correspondence should be addressed. E-mail: mouyang@ umd.edu. 10.1021/nl803352p CCC: $40.75 Published on Web 12/04/2008
2009 American Chemical Society
different growth mechanisms. In particular, we are able to control diameters of CdSe nanowires with near monodispersed distribution through a catalyst assisted vaporliquid-solid (VLS) mechanism, while CdSe nanostructures with rich morphology can be obtained by tuning synthetic parameters in the catalyst-free chemical vapor transport and condensation (CVTC) process. Existence of various CdSe nanostructures offers opportunity to explore device applications with integrated functionality. As an example, we show nanoscale phototransistor effect configured from a simple CdSe tripod nanostructure. Single-source molecular precursor has been applied as organometallic sources for preparing different semiconductor thin films21 and quantum dots in solution.22 More recently, it has been further demonstrated that they can be used for the growth of undoped and manganese-doped CdS and ZnS nanowires in high yield with high quality optical properties.23,24 Compared with other methods such as laser ablation and high temperature thermal evaporation, advantages of single-source molecular precursor for synthesizing nanostructures include well-defined stoichiometry, existed metalchalcogenide bond in precursor offering a convenient reactive intermediate for gas-phase synthesis, and low sublimation temperatures providing high dynamic control over synthetic conditions. For the case of CdSe in this study, air-stable solid
Figure 1. Schematics of two synthetic approaches of CdSe nanostructures via single-source molecular precursor. (a) VLS based process; (b) CVTC based process.
molecular precursor Cd[(SePiPr2)N]2 is synthesized in our group via modified process,21 serving as precisely defined source for Cd and Se reactants for synthesizing different structures. Two different synthetic approaches (Figure 1) are taken to prepare a variety of CdSe nanostructures by controlling thermal decomposition of as-synthesized CdSe molecular precursor with and without the presence of metallic catalysts. In both approaches molecular precursor powder is placed upstream of the substrate at the entrance to a horizontal tube furnace where the lower temperature produces gas phase reactants without thermal decomposition. SiO2/Si substrate is placed downstream of the precursor to collect assynthesized CdSe nanostructures. The whole system is degassed for about two hours before the furnace is elevated to 950-1050 °C (measured at the center of furnace) at a programmable rate of ∼30 °C/min. Argon gas is sent through the system at a rate of 20-50 sccm to act as a carrier gas to transport sublimated reactant vapor to the region of substrate within the tube furnace where the decomposition, nucleation, and growth occur. Positions of molecular precursor and substrate are intentionally controlled in order to achieve desired sublimation and growth temperatures in a specific experiment. In the first synthetic approach, the SiO2/Si substrate is first functionalized in 0.1% w/v aqueous polyL-lysine solution (Ted Pella, product no. 18026), and then commercial available monodispersed Au colloids with different sizes (Ted Pella, unconjugated colloids) are diluted and dispersed onto the substrate and served as metallic catalysts for nanowire growth. Compared with experimental conditions applied in the first synthetic approach, the growth apparatus remains the same in the second approach except that position of SiO2/Si substrate is typically located at different temperature/reactant vapor pressure zones in the furnace without presence of any metallic catalyst.25 Figure 2a shows typical scanning electron microscopy (SEM) image of CdSe nanowires prepared in high yield by the first synthetic approach. All the nanowires are found to be sheathed with an amorphous layer of 3-5nm in thickness. High resolution transmission electron microscopy (HRTEM) image clearly shows that most of straight nanowires possess single crystalline (0001) lattice planes perpendicular to the nanowire axis (Figure 2b).26 On the basis of the VLS mechanism,2 the molecular precursor undergoes thermal 438
Figure 2. Controlled diameter synthesis of CdSe nanowires from VLS mechanism with single-source molecular precursor. (a) typical SEM image of CdSe nanowires grown from 30 nm Au colloids. Scale bar: 2 µm. Inset, TEM micrograph of a single nanowire showing a Au catalyst at the end. Scale bar: 10 nm. (b) HRTEM image of VLS-grown nanowire showing single crystalline lattice. Scale bar: 1nm. (c-f) diameter histogram of CdSe nanowires grown from 10, 15, 20, and 30 nm Au colloids. The solid lines show a Gaussian fit of the wire diameter distribution.
decomposition in the substrate regions (613-652 °C) and forms liquid alloys with Au colloid catalysts when supersaturated CdSe nanowires would precipitate out at the end of Au catalysts. This VLS mechanism is further confirmed in this synthetic approach based on two experimental observations. One is the presence of Au nanoclusters at the end of as-synthesized nanowires as shown in the inset of Figure 2a; the other is the control experiments carried out Nano Lett., Vol. 9, No. 1, 2009
under the same growth condition but without Au colloids, showing no growth of CdSe nanowires. More importantly, the function of Au catalyst is not only to mediate the nanowire growth but also to localize the reactants and determine the diameter of nanowires, implying that nanowires with a narrow size distribution could be obtained by exploiting well defined catalysts.27-29 To address the issue of size selectivity in the VLS process with singlesource molecular precursor, we demonstrate here the controlled diameters of CdSe nanowires by using different sized Au colloids as catalysts. In order to quantify diameter distribution of CdSe nanowires histogram analysis of nanowire diameters with different sized Au colloids is shown in Figure 2c-f, which demonstrates narrow distribution of nanowires as well as obvious correlation with the colloid catalyst sizes. For nanowires grown from 10, 15, 20, and 30 nm Au colloids, the average diameter and size distribution are 11.2 ( 1.9, 17.1 ( 2.7, 21.5 ( 2.9, 31.4 ( 3.2 nm, respectively. Obviously, nanowire diameter mirrors the size of Au colloid catalyst. This result shows that near monodispersed CdSe nanowires can be well controlled through the catalysts assisted VLS process by using the single-source molecular precursor. By manipulating synthetic conditions we observe that CdSe nanowires can also be obtained without the presence of Au colloid catalysts based on different growth mechanism (Figure 1b). Figure 3a shows typical SEM images of nanowires synthesized at 695 °C with sublimation temperature of precursor at 200 °C. Further detailed HRTEM analysis reveals that most of nanowires synthesized in this approach are grown along 〈0001〉 (Figure 3b) which is similar to that in the VLS grown nanowires, but small amount of nanowires along 〈112j0〉 orientation can also be obtained (Figure 3c). While the dominant morphology of CdSe nanostructures synthesized by the first approach is the VLS grown nanowires, this second CVTC approach offers not only non-VLS CdSe nanowires but also nanostructures in the form of nanocomb (590-620 °C), nanobelts (730-865 °C), multibranched structures (680-750 °C), nanotubes (815-865 °C), nanoflowers (770-865 °C), and nanodiskettets (815-865 °C) without the assistance of metallic catalysts, which are characterized by SEM and TEM as shown in Figure 3d-i. It is worthy noting that even though CdSe nanobelts, some multibranched structures, nanotubes, and nanoflowers were reportedrecentlyfollowingadifferentVLSgrowth,16,17CdSenanocomb as well as nanodiskette structures are the first time observed. While the exact mechanism of this catalyst-free synthetic process is not clear we believe that CdSe reactant vapor sublimed from molecular precursors at lower temperature moves with the carrier gas, directly deposits on a substrate at a higher temperature region, nucleates, and grows into different nanostructures. More detailed structural analysis of these unique catalyst-free CdSe nanostructures are currently under investigation, which might provide more insight into the growth mechanism of different morphology. However, existence of different nanostructures in the CVTC process as well as their dependence on temperature, nonNano Lett., Vol. 9, No. 1, 2009
Figure 3. Controlled morphology synthesis of CdSe nanostructures from CVTC process with single-source molecular precursor. (a) SEM image of CdSe nanowires synthesized without the presence of Au colloids. Scale bar: 500 nm. (b) HRTEM image of CVTC grown nanowire showing single crystalline lattice plane. Arrow highlights nanowire growth direction along 〈0001〉. Scale bar: 3 nm. (c) HRTEM image of CVTC grown nanowire with different oriented single crystalline lattice plane. Arrow highlights nanowire growth direction along 〈112j0〉. Scale bar: 2 nm. (d) SEM image of nanocomb structure. Scale bar: 1 µm. (e) SEM image of nanobelt structure. Scale bar: 500 nm. (f) SEM image of tripod structure. Scale bar: 200 nm. (g) TEM image of nanotube structure. Scale bar: 100 nm. (h) SEM image of nanoflower structure. Scale bar: 25 µm. (i) SEM image of nanodiskette structure. Scale bar: 300 nm.
equilibrium vapor pressure and growth time further suggests that their structures might be determined and controlled by the growth kinetic. Particularly, one advantage of using single-source molecular precursor is its low sublimation temperature (∼200 °C), which make it possible to intentionally create extremely high local nonequilibrium reactant vapor pressure near substrate and further leads to the increase of chemical potential (∆µ) of the nucleation process of nanostructures. This provides a unique opportunity to tune the critical nuclei size by simply manipulating nonequilibrium vapor pressure in a CVTC process. We take CdSe tripod (Figure 3f) as an example to demonstrate such kinetic control. Experimentally we achieve vapor pressure control by keeping the position of substrate unchanged while altering the spacing between molecular precursor powder and substrate (the smaller the spacing, the higher the nonequilibrium pressure 439
Figure 4. Dynamic control of arm diameter of CdSe tripod in CVTC process by manipulating precursor-substrate spacing (x). (Inset) Schematic experimental configuration.
above the substrate).30 Our results show clearly dependence of arm diameter of tripod nanostructures on the spacing (x) between molecular precursor source and the substrate (Figure 4). This result is in qualitative agreement with classic nucleation theory that the increase of local vapor pressure relative to the equilibrium vapor pressure of the bulk deposit could induce a larger positive ∆µ and further leads to the smaller critical nucleus size.31 While diameter of CdSe tripod arms can be continuously controlled from nanometer- to micrometer-scale by manipulation of nonequilibrium reactant vapor pressure, length of arms can be tuned independently up to submillimeter by controlling reaction time. More detailed structural characterization of as-synthesized CdSe tripod is shown in the Supporting Information (Figure S1). Controlled synthesis of various CdSe nanostructures opens up opportunities for fundamental studies as well as device applications. For example, CdSe tripod shown in the Figure 3f provides a native entity for a nanoscale transistor. We carry out photoconductivity measurement at low temperature of CdSe tripod structures. Single electron transport measurement has been carried out on CdTe tetrapods and reveals existence of both hopping and delocalization transport regimes.32 More intriguingly the CdTe tetrapod is found to behave as a new single electron transistor by treating three arms of tetrapods as source, drain and gate electrodes, respectively, while leaving the fourth arm dangling. In this regard, our tripod represents a better candidate as integrated nanoscale transistor because it has only three arms therefore potential issues such as electrostatic charging existed in tetrapod structures can be avoided. CdSe tripods shown in the Figure 3f are sonificated, dispersed in ethanol, and deposited onto an oxidized p-Si wafer (Nova Electronic Materials, Ltd.). Occasionally tripods with three arms projecting downward the substrate can be found (while the branching point pointing upward) and can be located by the alignment markers predeposited on the substrate. E-beam lithography is then carried out, followed by e-beam (thermal) deposition to fabricate 300 nm Ti/100 nm Au electrodes contacting onto three individual arm respectively (no thermal annealing process is applied in all devices reported here to avoid diffusion of electrode materials as well as charge 440
Figure 5. Nanoscale phototransistor prepared from a CdSe tripod nanostructure. (a) Photocurrent measured between two arms of a CdSe tripod under visible light illumination with different intensity; (b) modulation of photocurrent between source and drain by applied gate voltage from the third arm. (Inset) SEM image of a typical nanoscale phototransistor configured from a CdSe tripod.
injection into the tripods), which can be electrically defined as source, drain, and gate terminals, respectively. The device is loaded into an opto-cryostat for low temperature (5 K) photoconductivity measurements in low vacuum. In a typical photoconductivity measurement a homemade femtosecond laser with wavelength at 510 nm is applied. A typical SEM image of a CdSe tripod contacted with three Au electrodes (labeled as 1(S), 2(D), and 3(G)) is shown in the inert of Figure 5. Figure 5a shows selected I-V curves through source 1(S)drain 2(D) arm pair under different light intensity illumination while keeping the third arm floating. It is worthy noting that measurement of pair 2-3 and 1-3 shows no obvious difference, and the data are reproducible with time. In the dark condition, current is in the range of pA at small VDS. However, under the illumination of visible light (510 nm), significant increase of current is typically observed in both positive and negative VDS. Increase of photocurrent is proportional to the light intensity applied. More importantly, we notice that photocurrent gain can be controlled by applying voltage to the third arm, which behaves as a gate electrode in field effect transistor configuration. In order to avoid potential charge interference between gate electrode and source and drain electrodes, we selectively illuminate only source and drain arms including central branch point (highlighted by green area in the SEM image, inset of Figure 5b); therefore in our measurement only current between source and drain is enhanced due to illumination of green light. Figure 5b shows photocurrent as a function of arm gate voltage under light intensity of 3.3 mW/cm2. Dependence of IDS on the VG is clearly observed. At negative gate voltage, the photoconductivity is significantly enhanced, while positive gate voltage reduces the Nano Lett., Vol. 9, No. 1, 2009
photoconductivity. Because the gate branch is not illuminated by visible light during the measurement we believe that modulation of photoconductivity of source-drain channel is due to gate voltage effect instead of current coupling between the gate electrode and source and drain electrodes. Furthermore, while gating mechanism requires more systematic studies including such as temperature dependence, we believe that the gating effect is due to the gate branch instead of electrostatic interaction of the Au electrodes and the branching point, especially considering that size of tripod is much bigger than that of tetrapod devices reported before that could lead to large separation between branching point and Au electrodes. In summary, we report here that by using single-source molecular precursor size of nanowires can be controlled in a precise way based on VLS mechanism when the well defined Au catalyst is applied, while more rich 1D and quasi1D nanostructures (with different morphology and dimensions) can be independently obtained and controlled without catalysts by maneuvering kinetic synthetic condition such as temperature and pressure. Our methods offer new level of control over the synthesis of inorganic nanostructures in gas phase and can be implemented to other material systems with potential for much greater structural and functional complexity. In particular, we demonstrate photoconductivity from as-synthesized CdSe tripod can be controlled by using one arm as a sensitive gate. Considering that CdSe is a direct bandgap semiconductor and is expected to show strong photoconductivity at visible light, our result might enable a novel nanoscale field effect phototransistor operation scheme enabled by CdSe tripod nanostructures. Acknowledgment. Supported by NSF CAREER Grant (DMR-0547194), NSF MRSEC seed Grant (DMR 0520471), Sloan Fellowship, Powe Junior Faculty Enhancement award, ONR YIP award (N000140710787), Beckman YIP Grant (0609259093), and startup fund from the University of Maryland (College Park). We also thank Chenggang Tao for chemical drawing, and facility supports from Center for Nanophysics and Advanced Materials (CNAM), Maryland Nanocenter and its NISP laboratory (the NISP laboratory is supported in part by the NSF as a MRSEC shared Experimental Facility). Supporting Information Available: Supporting Figure S1. This material is available free of charge via the Internet at http://pubs.acs.org. References (1) Cui, Y.; Duan, X.; Huang, Y.; Lieber, C. M. In Nanowires and Nanobelts: Materials, Properties and DeVices; Wang, Z. L.; Kluwer Academic/Plenum Publishers: New York, 2003; p 3. (2) Hu, J.; Odom, T. W.; Lieber, C. M. Acc. Chem. Res. 1999, 32, 435. (3) Lieber, C. M. Scientific American, September, 2001, p 58. (4) Xia, Y.; Yang, P.; Sun, Y.; Wu, Y.; Mayers, B.; Gates, B.; Yin, Y.; Kim, F.; Yan, H. AdV. Mater. 2003, 15, 353. (5) Wang, Z. L. Annu. ReV. Phys. Chem. 2004, 55, 159. (6) Alivisatos, A. P. Pure. Appl. Chem. 2000, 72, 3. (7) Alivisatos, A. P. Science 1996, 271, 933. (8) Hodes, G.; Albu-Yaron, A.; Decker, F.; Motisuke, P. Phys. ReV. B 1987, 36, 4215.
Nano Lett., Vol. 9, No. 1, 2009
(9) Ichimura, M.; Sato, N.; Nakamura, A.; Takeuchi, K.; Arai, E. Phys. Status Solidi 2002, 193, 132. (10) Murray, C. B.; Norris, D. J.; Bawendi, M. G. J. Am. Chem. Soc. 1993, 115, 8706. (11) Peng, X.; Wickham, J.; Alivisatos, A. P. J. Am. Chem. Soc. 1998, 120, 5343. (12) Peng, X.; Manna, L.; Yang, W.; Wickham, J.; Scher, E.; Alivisatos, A. P. Nature 2000, 404, 59. (13) Milliron, D. J.; Hughes, S. M.; Cui, Y.; Manna, L.; Li, J.; Wang, L. W.; Alivisatos, A. P. Nature 2004, 430, 190. (14) Duan, X. F.; Lieber, C. M. AdV. Mater. 2000, 12, 298. (15) Ma, C.; Wang, Z. L. AdV. Mater. 2005, 17, 1. (16) Ma, C.; Ding, Y.; Moore, D.; Wang, X.; Wang, Z. L. J. Am. Chem. Soc. 2004, 126, 708. (17) Wang, G. X.; Park, M. S.; Liu, H. K.; Wexler, D.; Chen, J. Appl. Phys. Lett. 2006, 88, 193115–1. (18) Ridley, B. A.; Nivi, B.; Jacobson, J. M. Science 1999, 286, 746. (19) Alivisatos, P. Nat. Biotechnol. 2004, 22, 47. (20) Ouyang, M.; Awschalom, D. D. Science 2003, 301, 1074. (21) Afzaal, M.; Crouch, D.; Malik, M. A.; Motevalli, M.; O’Brien, P.; Park, J.; Woollins, J. D. Eur. J. Inorg. Chem. 2004, 2004, 171. (22) Cumberland, S. L.; Hanif, K. M.; Javier, A.; Khitrov, G. A.; Strouse, G. F.; Woessner, S. M.; Yun, C. S. Chem. Mater. 2002, 14, 1576. (23) Barrelet, C. J.; Wu, Y.; Bell, D. C.; Lieber, C. M. J. Am. Chem. Soc. 2003, 125, 11498. (24) Radovanovic, P. V.; Barrelet, C. J.; Gradecak, S.; Qian, F.; Lieber, C. M. Nano Lett. 2005, 5, 1407. (25) We notice that both substrate temperature and reactant vapor pressure are important for controlling CdSe nanostructures. Manipulation of vapor pressure is achieved by controlling the sublimation temperature of precursor powder. Occasionally a ceramic boat is also covered on top of both substrate and precursor powders in order to fine tune nonequilibrium reactant vapor pressure during the synthesis of CdSe nanostructures with different morphologies. (26) Nanowires with different orientations (such as ) are occasionally observed but these nanowires are typically curly with a lot of defects. (27) Gudiksen, M. S.; Lieber, C. M. J. Am. Chem. Soc. 2000, 122, 8801. (28) Gudiksen, M. S.; Wang, J.; Lieber, C. M. J. Phys. Chem. 2001, 105, 4062. (29) Cui, Y.; Lauhon, L. J.; Gudiksen, M. S.; Wang, J. F.; Lieber, C. M. Appl. Phys. Lett. 2001, 78, 2214. (30) In this control experiment, position of substrate remains unchanged in order to maintain constant nucleation and growth temperature, while location of molecular precursor powder is altered to achieve local reactant pressure control in the substrate region. When precursor powder is moved closer to the substrate, there are two effects leading to higher nonequilibrium reactant vapor pressure above the substrate. One is the increase of sublimation temperature. We carefully calibrate temperatures at different locations within the furnace. Since we use single zone furnace, temperature at the position far away from the furnace center shows sensitive dependent on the distance to the center. The closer to the furnace center, the higher the local temperature can be achieved. We take advantage of this temperature distribution. Because the substrate is placed downstream of the precursor source, when the precursor powder is moved closer to the substrate it becomes also closer related to the furnace center; therefore sublimation temperature of precursor is increased, which contributes to higher local nonequilibrium reactant vapor pressure; The other effect is related to the distance itself between source and substrate. Nonequilibrium reactant vapor pressure is maximum right above the precursor source, then decays away from the source location. When the spacing between source and substrate becomes smaller, vapor pressure above the substrate is increased. Therefore by carefully controlling the spacing between precursor source and substrate, different nonequilibrium local reactant vapor pressure can be achieved above the substrate: the smaller the spacing between substrate and precursor source, the higher reactant vapor pressure above the substrate. (31) Campbell, W. B. In Whisker Technology; Levitt, A. P., Ed.; WileyInterscience: New York, 1970; Vol. 1, pp 5-47. (32) Cui, Y.; Banin, U.; Bjork, M.; Alivisatos, A. P. Nano Lett. 2005, 5, 1519.
NL803352P
441
