In-Situtemobservation of a Microcrucible Mechanism of Nanowire Growth

In-situTEMObservation of a Microcrucible Mechanism of Nanowire Growth

Rebecca Boston1,2, Zoe Schnepp3, Yoshihiro Nemoto4, Yoshio Sakka4, Simon R. Hall1,*

1Complex Functional Materials Group, School of Chemistry, University of Bristol, Bristol BS8 1TS, UK;2Bristol Centre for Functional Nanomaterials, Centre for Nanoscience and Quantum Information, Tyndall Avenue, Bristol, UK;3School of Chemistry, University of Birmingham, Birmingham, B15 2TT, UK; 4National Institute for Materials Science, 1-2-1 Sengen, Tsukuba, Ibaraki, 305-0047, Japan. *Correspondence to:

Abstract

The growth of metal oxide nanowires canproceed via a number of mechanisms such as screw-dislocation, vapor-liquid-solid orseeded growth. Transmission electron microscopy (TEM) can resolve nanowires, but invariably lacks the facility for direct observation of how nanowiresform. We used a TEM equipped with an in-situ heating stage to follow the growth of quaternary metal oxide nanowires. Real-time video imaging revealedbarium carbonate nanoparticles diffusing through a porous matrix containing copper and yttrium oxides to subsequently act as catalytic sites for the outgrowth of metal oxide nanowires on reaching the surface. The results suggest that sites on the rough surface of the porous matrix act as microcrucibles and thus provide insights into the mechanisms that drive metal oxide nanowire growth at high temperatures.

One sentence summary

A long-postulated, but never seen crystal growth mechanism has been observed using an electron microscope equipped with a furnace.

The fabrication of nanowires of functional materials has become an important goal for their application in miniaturized circuits as diodes and transistors, coaxial gates and sensors (1). Fabrication methods such as soft lithography have yielded ternary or more complex metal oxide nanowires(2), however these types of techniques rely on an underlying substrate which may have a detrimental effect on the properties of the material. Simple metal oxide nanowires such as ZnO (3) or TiO2 (4) have been grown using a variety of methods, although the unsupported crystallization of more complex quaternary and quinternary oxide functional materials such as, for example, magnetoresistive La0.67Ca0.33MnO3 (LCMO), piezoelectric PbZr0.52Ti0.48O3 (PZT) and the superconductors YBa2Cu3O7-δ (Y123) and Bi2Sr2Can-1CunO2n+4+x (BSCCO), is difficult at the nanoscale, owing to the formation of stable impurity phases and/or the lack of a suitable nanoparticulate catalytic seed (1). Wires and whiskers of BSCCO have been grown previously at the macro-scale (5,6) using a rapidly cooled glassy Bi-rich BSCCO precursor seeded with Al2O3 powder in order to provide sites of nucleation and outgrowth. The mechanism for this growth, however, has yet to be fully characterized, and while several processes have been proposed,the microcrucible mechanism, is generally the favoredone (7).

The microcrucible mechanism is a two-phase growth process which relies on the presence of a liquid and a solid phase. The liquid phase dissolves ions out of a solid matrix, and once supersaturated, growth occurs at the exposed surface of the droplet with the nanowirebeing extruded from the bulk (5). The microcrucible itself is comprised of the solid underlying substrate, which forms a crucible-like structure to contain and concentrate the molten precursor, and act as a conduit for the provision of more precursor materials as the nanowire forms. Crystals produced by this mechanism would therefore have well-defined faceted leading edges, rather than the commonly observed tapering tip of crystals grown via the VLS mechanism, which typically have a catalytic nanoparticle at the tip that is consumed during the course of nanowire growth (7).

Here we present direct observation of the microcrucible growth mechanism in quaternary metal oxide Y2BaCuO5 (Y211) nanowires, using a TEM fitted with an in-situ heating stage. Full details on the materials and methods are available as supplementary materials on Science Online (8), but briefly the samples used were prepared from stoichiometric mixtures of yttrium, barium and copper nitrate with sodium alginate in de-ionized water. The resulting composite was dried for 12 hours at 60 °C then calcined in a chamber furnace for 6 hours at 500 °C, before being heated a second time to 800 °C for two hours at the same ramp rate. Once cool, these samples were then dispersed onto TEM grids and introduced into either a JEM-2100F TEM or a JEOL 200-ARMF TEM each fitted with an AHA-21 Aduro Protochips heating stage.

In order to assess which growth mechanism occurred, we first determined the crystallochemical nature of our system at the stage prior to the in-situ TEM heating. Figure S1 shows powder XRD confirming the complex mix of phases present immediately prior to nanowire growth (8). BaCO3 is a major crystalline phase at this point, along with variouscrystalline copper and yttrium phases. The XRD background is flat, particularly between 20-30 degrees 2θ, thereby suggesting only a minor amorphous contribution to the matrix.The crystalline matrix is porous (Figure S2) and the chemical composition of the system was confirmed using energy dispersive x-ray analysis (EDXA), which shows the presence of only yttrium, barium, copper and oxygen (Figure S3) (8).

The results of the in-situ TEM heating experiments are shown in Figure 1. Nanowires were observed to grow with sharply faceted ends and a uniform cross-section, indicative of a microcrucible mechanism (7). For nanowire growth to proceed via this mechanism,however,there needs to becontinuous crystallization of the molten phase occurring with support from an underlying matrix. In this work we have identified that this molten phase comes from BaCO3 nanoparticles, which melt and diffuse to the surface of the precursor matrix (Figure S5 and Movie S1 (8)). On the heating stage of the TEM in vacuo, BaCO3 nanoparticles became molten at an applied temperature of 450 °C and so the ramp rate was slowed from 20 °C min-1 to 4 °C min-1in order to ensure time for adequate heat transfer from the grid to the sample. By diffusing through regions containing the other two precursor ions, the BaCO3nanoparticle is additionally able to solubilize the Y and Cu cations required for Y211 formation,thereby increasing ionic concentration at the eventual outgrowth site at the surface of the matrix. There is precedent for this as it has been shown that using the high affinity of carbon for certain metals (9), molten metal nanoparticles are able to act as catalytic sites for the uptake of carbon with subsequent carbon nanotube outgrowth (10).On reaching the surface of the porous matrix, the molten BaCO3nanoparticle initiates nanowire outgrowth and no molten drop is observed at the growing tip of the nanowire. Under continuing heating, nanowires were observed to grow over the course of 10 minutes (Figure 2). Movie S2, taken at a higher resolution, better shows the initial stages of growth at high temperature (8). The faceted end of the emerging wireindicates that the free end of the wire is solid during the growth process. The uniformity of the cross-section of the wire throughout the initial stages of the growth indicates that crystallization is occurring within a region of fixed diameter corresponding to the diameter of the microcrucible (Figure 1B). Figure S6Ashows a partially grown nanowire which was rapidly cooledjust after the start of outgrowth, when the nanoparticle at the base of the nanowire was still molten (8). Figure S6Bis a high-resolution image of the base of this nanowire and illustrates the chemical composition of the component parts of the microcrucible, which wereidentified crystallographically from the lattice fringes (8). The nanowire itself was determined to be Y211, with regions which indexed to BaCO3 and CuO being found in the underlying material. Crucially, the majority of the material separating the crystalline porous matrix from the nanowire was found to be amorphous. Thisconfirms that at the onset of rapid cooling this was a liquid, thereby providing further support for the microcrucible mechanism. EDXA analysis of this amorphous region showed that it was composed of Y, Ba and Cu.

Since the walls of the microcrucible contain the ions that are consumed in the formation of the nanowire, as the nanowire grows it would be expected that the dimensions of the microcrucible itself would change, leading to alteration of the nanowire morphology.The continual evolution of the liquid-solid interface of the microcrucible is a dynamic process which leads to creep of the interface and the concomitant morphogenesis of non-classical crystal structures.Changes in the structure of the microcrucible have been observed previously in the growth of BSCCO and Y123 whiskers at the micro-scale, with the resultant creation of single crystal morphologies as diverse as bows and rings (11,12). In this work, microcrucible creep resulted in two distinct nanowire morphologies;those which underwent growth in both length and width, and those where two nearby microcruciblesjoined together to form wires with “stepped” ends.Figure 3shows nanowires with stepped ends as a result of two microcrucibles joining together (Figure 3A and 3B), and the progression of a growing wire over the course of five minutes, which shows the walls of the microcrucible breaking down (Figure 3C and D). Since the nanowire in Figure 3C is short, and at high temperature, as the microcrucible supporting it breaks down (allowing for a droplet of greater diameter) the molten material is able to wet the side edge of the nanowire, causing a rapid and uniform increase in width. A degree of control of the liquid-solid interface is exhibited here, as the BaCO3nanoparticles are of low size polydispersity and therefore limit the type and amount of creep in the system and consequently the diversity of structural features in the nanowires.

Crystallographic analyses of the nanowires were conducted in order to determine the direction of growth, and to ascertain whether they areindeed single crystals of Y211. Figure 4A shows a typical wire that has a single crystalline region along the length, with the zone axis determined as the [25] direction and thedirection of growth perpendicular to the(102) crystal planes of Y211 (Figure 4B). Lattice fringes due to the (102) crystal planes can be observed at the termination of a typical Y211 nanowire (Figure 4C).

In previous studies on the synthesis of quaternary oxide nanowires (13,14),carbon-rich moieties such as citrate and acetate were used in the syntheses, which tend to persist for longer under calcination (15,16) than do the nitrates used here. This in turn leads to a lower density matrix, enabling nanoparticles to leave the surface and act as catalytic sites on the leading edges of the outgrowing nanowires. There then follows tapered growth as the nanoparticle is consumed, producing morphologies more reminiscent of a VLS process. The loss of nitrates at a lower calcination temperature means that in this work, the matrix is denser at the point at which nanowire growth begins. We deduce therefore that here, the matrix isstill porous and reticulated, but denser and will therefore tend to entrap Ba-rich nanoparticles at the surface and lead to the microcrucible growth observed. It is likely that previous reports where nitrate precursors were used to form thecomplex oxides of La3Ga5SiO14 (17) and La0.67Sr0.33MnO3(18), via a porous matrix, the faceted-ended nanowires produced were alsothe result of a microcrucible mechanism.

Through judicious design of the synthetic protocol, we have demonstrated the direct observations of a microcrucible growth mechanism and confirm that it is a viable method for the growth of complex oxide nanowires. The successful formation of nanowires is predicated on the presence of a catalytic nanoparticle and a porous matrix that enables migration of the former through the latter, leading to nanowire outgrowth at the surface. The uniform cross-section which arises from the microcrucible mechanism means that the nanowires produced in this waywill have the same physical properties along their entire length, leading to more uniform current-carrying ability, ferroicbehavior and tensile strengthforthe future use of complex functional oxide nanowires in applications.

References and Notes

1. G. A. Ozin, Nanochemistry—a chemical approach to nanomaterials. Cambridge, UK: Royal Society of Chemistry (2005).

2. K. Xu, J. R. Heath, Nano Lett. 8, 3845 (2008).

3. Z. L. Wang, J. Phys. Cond. Mater. 16, R829 (2004).

4. J. D. Berrigan, T-S. Kang, Y. Cai, J. R. Deneault, M. F. Durstock, K. H. Sandhage, Adv. Funct. Mater21, 1693 (2011).

5. I. Matsubara, R. Funahashi, T. Ogura, H. Yamashita, K. Tsuru, T. Kawai, J. Cryst. Growth141, 131 (1994).

6. P. Badica, K. Togano, J. Mater. Res.20, 3358 (2005).

7. P. Badica, K. Togano, S. Awaji, K. Watanabe, H. Kumakura,Supercond. Sci. Tech. 19, R81 (2006).

8. See supplementary materials on Science Online.

9. Y. Cao, L. Xiao, M. L. Sushko, W. Wang, B. Schwenzer, J. Xiao, Z. Nie, L. V. Saraf, Z. Yang, J. Liu, Nano Letters 12, 3783 (2012).

10. A. Gorbunov, O. Jost, W. Pompe, A. Graff, Carbon,40, 113 (2002).

11. P. Badica et al.Supercond. Sci & Tech. 25, 105003 (2012).

12. S. Cagliero et al.Supercond. Sci & Tech. 25, 125002 (2012).

13. S. R. Hall, Adv. Mater. 18, 487 (2006).

14. Z. A. Schnepp, S. C. Wimbush, S. Mann, S. R. Hall, Adv. Mater. 20, 1782 (2008).

15. L. A. Chick, L. R. Pederson, G. D. Maupin, J. L. Bates, L. E. Thomas, G. Exarhos,Materials Letters1-2, 6 (1990).

16. H. C. Zeng, S. K. Tung, Chem. Mater.8, 2667 (1996).

17. Z. Schnepp, J. Mitchells, S. Mann, S. R. Hall, Chem. Comm.46, 4887 (2010).

18. Z. Schnepp, S. C. Wimbush, S. Mann, S. R. Hall, CrystEngComm.12, 1410 (2010).

Acknowledgements

SRH and RB acknowledge the EPSRC, UK (Grant EP/G036780/1) and the Bristol Centre for Functional Nanomaterials for project funding. SRH and RB also acknowledge the support of the EPSRC under their ‘Building Global Engagements in Research' scheme and the Electron and Scanning Probe Microscopy Facility and Electron Microscopy Group in the Schools of Chemistry and Physics, University of Bristol respectively. ZS would like to thank the National Institute for Materials Science in Japan for the award of an ICYS Postdoctoral Fellowship. All authors would like to thank Masaki Takeguchi for his help with all aspects of TEM operation at NIMS.A part of this work was supported by NIMS microstructural characterization platform as a program of "Nanotechnology Platform" of the Ministry of Education, Culture, Sports, Science and Technology(MEXT), Japan.

Supplementary Materials

Materials and Methods

Figures S1-S6

Movie S1

Movie S2

Figure Legends

Fig. 1.TEM images of Y211 nanowires grown via a microcrucible mechanism.(A) Low-resolution TEM image showing multiple nanowires with faceted edges, (B) High-resolution image of one nanowire, clearly showing a uniform rectilinear cross-section. Scale bar in (A) is 500 nm, in (B) 50 nm.

Fig. 2. TEM images taken from Movie S1 showing nanowire growth.Each frame represents 1 minute intervals showing the initial stages of growth of a YBCO nanowire at 500 °C. Scale bar is 100 nm.

Fig. 3. Images showing mechanisms of nanowire expansion due to creep in the microcrucible system.(A) TEM images with (B) schematic showing how microcrucible intergrowth to produce wider nanowires and (C) TEM images with (D) schematic, of a nanowire spontaneously broadening as a result of microcrucible expansion on continuing heating at 500 °C. Scale bars in (A) are 100 nm, and in (C) 200 nm.

Fig. 4. TEM images of fully-formed nanowires grown from the porous matrix.(A), (C) TEM images of as-grown YBCO nanowires with (B) electron diffraction pattern from nanowire in (A). Diffraction and lattice fringes are indexed using JCPDS card No. 01-079-0653. Scale bar in (A) is 20 nm, in (C) 5 nm.

1