STEM-Cathodoluminescence of SnO2 nanowires and powders

https://doi.org/10.1016/j.snb.2016.08.145Get rights and content

Highlights

  • STEM-Cathodoluminescence is applied to SnO2 for the first time.

  • Spectrum imaging revealed strong spatial variation of emission spectra.

  • 1.7–2.0 eV emission is most intense at crystal edges and assigned to surface defects.

  • 2.4–2.7 eV emission is most intense at crystal centers and assigned to bulk defects.

Abstract

For the first time, ultra-high spatial resolution STEM-cathodoluminescence has been applied to SnO2 nanowires and nanoparticles in order to study the spatial distribution of point defects that form deep levels in the band gap and often dictate their gas-sensor performance. SnO2 nanowires consistently had emission signals centered at 1.76, 1.99, and 2.45 eV. Emission at 1.99 eV was found to be more intense with the STEM probe at the edges of nanowires and in higher surface-area nanoparticles compared to emission most intense at 2.45 eV with the probe in the center of a nanowire or particle. This result is contrary to recent studies proposing that both emission energies are associated with two different surface oxygen vacancy types. The ability to spatially map the luminescence of a single nanostructure also facilitated much less-ambiguous spectral deconvolution consistent across many particles which gives confidence in properly assigning the energies of each defect state. Individually-probed SnO2 nanowires were found to have much more consistent and reproducible defect states compared to commercial nanoparticles which has implications in stability as gas sensors and efforts to model interactions between analyte gases and the oxide surface.

Introduction

Cathodoluminescence (CL) is a powerful tool for analyzing the emissive defect states in semiconducting or insulating materials. An electron beam is used to excite electrons in the sample material to a higher energy state, usually the conduction band, and then the electrons recombine with holes and emit photons. These photons are collected and compiled into a spectrum where each peak represents a radiative recombination where the peak center energy is the difference between the initial and final energy levels, typically from the conduction band minimum (CBM) to a mid-gap defect state or the valence band maximum (VBM). SnO2 nanomaterials have been used extensively in applications such as gas sensing and photocatalysis, where knowledge of the defect states and electronic properties is important for predicting and improving performance [1], [2], [3]. Korotcenkov et al. [4] used a scanning electron microscope (SEM) equipped with a CL detector to show that the CL spectra of bulk-doped SnO2 powders were directly related to their sensor properties. They propose that CL properties can be used to characterize new materials for gas sensing applications, and further as a simple quality control measure during sensor manufacturing. They suggest that the CL spectrum from a given sample can be used as an indicator of crystallinity, dopant incorporation, and defect concentrations when compared to well-characterized reference samples. Although SnO2 nanostructures have been characterized using CL detectors in the SEM [4], [5], [6], high spatial resolution CL in a scanning transmission electron microscope (STEM) has not yet been reported. In a STEM, the electron beam can be focused to sub-nanometer probe sizes, enabling CL emission of a nanostructure to be mapped with pixels as small as the atoms in the material. However, it is the interaction volume and the carrier diffusion length that limit the effective spatial resolution of the technique depending on the material properties and operating conditions [7]. The interaction volume using an 80 keV electron beam in a ∼100 nm thin specimen is typically much smaller than in a thick SEM sample because the incident electrons only lose a small fraction of their energy before being transmitted through the lower surface. This study will show that sharp spectral changes can be observed across distances as small as 10 nm.

To the authors’ knowledge CL of single SnO2 nanowires has only once been reported [8], using an SEM-CL system with relatively large (300 nm diameter) nanobelts, though spatially-resolved spectra were not taken within any single nanowire and their focus was mostly on temperature and size-dependent behavior.

SnO2 is naturally oxygen-deficient due to low defect formation enthalpies, which contributes to its n-type behavior. These oxygen vacancies create defect energy levels that lie within the band gap region. In SnO2, bulk oxygen vacancies can be singly-ionized, doubly-ionized, or neutral based on the number of electrons the vacancy has lost to the conduction band [9]. It is generally accepted that the singly- and doubly-ionized bulk oxygen vacancies form many shallow donor levels between 0.03 and 0.30 eV below the conduction band minimum, with slight variations between several studies [5], [10], [11]. Recently, Trani et al. [12] used density functional theory (DFT) to predict that bulk oxygen vacancies should form a flat band approximately 1 eV from the valence band maximum (VBM), much deeper than the previous experimental studies have found. Prades et al. [5] used DFT to study surface oxygen vacancies of the most common rutile SnO2 faceting surfaces: (110), (100), (101), and (001). They reported only two different oxygen vacancy configurations available on these surfaces: either 100° coordinated or 130° coordinated oxygen atoms, corresponding to bridging oxygen and in-plane oxygen vacancies, respectively. The energies of these vacancies were found to differ only slightly between different common facets, averaging to 1.4 eV (bridging) and 0.9 eV (in-plane) above the VBM. Taking into account that recombination can occur from the CBM or the bulk shallow levels to the surface vacancies, this gives four average energies to be expected: 2.70, 2.48, 2.20, and 1.98 eV, assuming a band gap of 3.6 eV, as shown in Fig. 1 (left).

Maestre et al. [10] among others, reported that the types of surface vacancies present depends on the final synthesis or annealing temperature, among other possible variables. They also found that in sintered polycrystalline SnO2 the 1.94 eV peak was associated with oxygen vacancies because it was quenched following a 1500 °C anneal, while this same anneal enhanced the green emission near 2.25 eV, suggesting it is associated with the formation of well-defined crystal facets. Zhou et al. [13] used time-resolved x-ray excited optical luminescence (XEOL) on SnO2 nanoribbons to show that high-energy states around 2.7 eV had much faster emission decay than two or more lower-energy peaks around 2.45 eV and 2.0 eV (shown in Fig. 1, center). This suggests that the higher-energy, faster-decaying states are related to recombination from the band edge to a surface state (0.9 eV from the VBM), while the slower-decaying states originate from the shallow trap levels because the excited electrons must first populate the normally unoccupied shallow donor levels, and then recombine into the surface state band. This somewhat contradicts the results of Prades et al. [5] that would predict two fast-decaying states and two slow-decaying states, but Zhou does not report a 2.2 eV fast-decay peak alongside the 2.7 eV peak. Furthermore, if all of the lower-energy slow-decaying states are from varying shallow donor levels to the 2.7 eV band, then this would predict bulk shallow levels as much as 1 eV below the CBM. Kar et al. [9] used ultrafast pump-probe spectroscopy (UPPS) to show that the excitation beam quickly saturates singly-ionized bridging vacancies 1.3–1.6 eV below the CBM due to fast recombination (∼150 fs) from the lowest conduction band states, which then decay into lower singly-ionized states that lie about 1.6 eV below the CBM. They then propose that these trapped electrons more slowly decay into the valence band causing an emission of about 2.1 eV or trapping an additional electron to become a neutral vacancy, as shown in Fig. 1 (right). This partially agrees with Zhou in that both showed emission near 2-2.1 eV to be one of the slowest processes, though they disagree as to the initial and final states causing this emission. Lettieri et al. [14] point toward a direct role of surface bridging oxygen vacancies as a source of recombination centers in PL measurements by using environmentally-controlled PL in conjunction with first-principles DFT calculations. The DFT analysis also shows the density of states (DOS) and band structure for stoichiometric and vacancy-populated (101) SnO2 surfaces, which predict the formation of mid-gap bands approximately 2.1–2.7 eV below the CBM with the introduction of surface bridging oxygen vacancies. It is important to remember that while simulations work with idealized versions of the bulk and surface states, the experimental results reported above and in this work are subject to much variation based on differences in processing methods and measurement techniques. Some studies that have different interpretations may indeed be working with materials that have defects with different concentrations or energy levels.

These defect-related emission peaks are typically all that is observed in SnO2 using luminescence techniques because rutile SnO2 has a dipole-forbidden direct-gap at 3.6 eV under normal conditions [1]. However, some near-band edge emission (∼3.4 eV) has been rarely reported [15], [16] in SnO2 nanowires. Chen et al. [17] even observed sharp fine-structure between 3.30 and 3.37 eV at 10 K, including three orders of phonon replicas. It is thought that the nanowire size and morphology alter the electron orbitals in a way that creates some allowed near-gap transitions. These recent discoveries have grown interest in SnO2 for its high-energy field-emission properties [6].

It is important to note that the radiative transitions observed with luminescence techniques do not directly give information that predicts the electronic behavior of the materials [18]. The conductivity of SnO2 is primarily related to the donor bands from bulk and surface oxygen vacancies. The spatial distribution, concentration and configuration of these vacancies will affect the observed luminescence. However, electrical measurements may be needed to understand the correlations between luminescence and electronic behavior. The emissive behavior can be a good direct indicator of the expected reactivity of the surface with analyte gases and will be useful in modeling these surface-gas reactions in gas sensing materials. Even more directly, the dependence of emissive defects on the gas atmosphere is useful in the newer field of contactless or optical gas sensors [14]. Collectively, these works show that the over-simplified representation of these materials with sharp crystalline interfaces and perfect defect-free band edges and alignment, which is so often used for predicting charge carrier movement, is not suitable and in the future should include defect states at the surface and in the bulk [4].

Section snippets

Experimental methods

SnO2 nanowires were grown by a high-temperature gold-catalyzed vapor-liquid-solid (VLS) mechanism. (100) p-type Si wafers with small oxidation layer were diced into 1 cm2 squares and sputter-coated with either 1.4 nm or 10 nm of gold using a Leica ACE600. The substrate was loaded onto a flat alumina plate into a quartz tube with 1′′ inner diameter and 0.02 grams of SnO powder (99.9% Sigma Aldrich) were sprinkled into a half cylinder (open-top) of alumina loaded 4 cm upstream from the substrate. The

STEM imaging and x-ray diffraction analysis

XRD was performed on SnO2 nanowires still attached to the Si growth substrate using grazing incident diffraction to maximize the nanowire film signal while minimizing the substrate signal. The powders were mounted in glass low-background holders. All three samples showed nearly identical XRD patterns (Fig. 4a) corresponding to SnO2 cassiterite (P42-mnm). The nanowire sample had small peaks at 44.6 and 56 ° which were attributed to the (200) reflection of the Au nanoparticle tips and the (311)

Conclusions

STEM-cathodoluminescence allows spectral changes to be spatially resolved on a scale of less than 10 nm. This technique has shown great promise in this study and should be applied to other semiconducting oxide nanostructures to develop knowledge of defect states for gas sensing, but could also aid other fields using similar structures such as catalysis, photocatalysis, and nanomaterial-based photovoltaics [22], [23], [24]. In the present work, SnO2 nanowires showed two dominant peaks and a third

Funding

This work was funded by a NASA Space Technology Research Fellowship and a Facilities Grant from the Institute for Materials Research (IMR) at The Ohio State University.

Acknowledgements

The authors would first like to acknowledge Dr. David Stowe and Gatan UK for the provision of the Vulcan system to CEMAS under and evaluation agreement to make these measurements possible. The authors also would like to thank Prof. Roberto Myers for helpful discussions and John Jamison for assisting in PL experiments.

Derek R. Miller is a Ph.D. candidate at The Ohio State University. His research focuses on growing and synthesizing new oxide nanostructures for applications in gas sensing, photovoltaics, and biocompatible surfaces. His recent focus is on fabrication and characterization of oxide nano-heterostructures to enhance performance in these applications.

References (24)

  • G. Salviati et al.

    Cathodoluminescence spectroscopy of single SnO2 nanowires and nanobelts

    Phys. Status Solidi

    (2005)
  • A. Kar et al.

    Tailoring the surface properties and carrier dynamics in SnO2 nanowires

    Nanotechnology

    (2011)
  • Cited by (21)

    • Role of Oxygen Vacancies in Nanostructured Metal-Oxide Gas Sensors: A Review

      2019, Sensors and Actuators, B: Chemical
      Citation Excerpt :

      Although nanomaterials are now widely used in the community, bulk properties are still commonly assumed when it comes to oxygen vacancies. Recent advancements in characterization techniques have revealed the complex nature of defect states in metal oxide nanomaterials compared to their bulk counterparts and how the synthesis method can also introduce variability [12]. Given the central role that oxygen vacancies have in gas sensing with metal oxide nanostructures, it is important to bring together the relevant ideas and break down the important considerations taking into account the details that some may leave out of their studies to better inform future research.

    • Synthesis of ZnO-decorated SnO<inf>2</inf> nanopowder with enhanced photocatalytic performance

      2019, Optik
      Citation Excerpt :

      Manjula et al. [11] reported that their synthesized SnO2 nanopowders exhibited blue and green emissions with wavelengths centered at 459 nm, 485 nm, 530 nm and 545 nm, respectively. Miller et al. [26] prepared SnO2 nanowires using gas-liquid-solid technique; they found that the nanowire had a strong red emission with a wavelength of about 620 nm (1.99 eV), and the wavelengths of emissions were different for different regions on the nanowire. Shewale et al. [27] found that the F-doped SnO2 films prepared by spray pyrolysis showed strong violet emissions with wavelengths of 396 and 404 nm, respectively, accompanied by a red emission with a wavelength centered at 700 nm.

    • Synthesis of coryphantha elephantidens-like SnO<inf>2</inf> nanospheres and their gas sensing properties

      2019, Current Applied Physics
      Citation Excerpt :

      Although SnO2 has a wide band and is easy to obtain appropriate electrical characteristics, the performance in practice is also strongly dependent on its size, morphology and microstructure. Over the past few years, various SnO2 micro-nanostructures have been synthesized, for example, nanoflowers [16–18], nanotubes [19], nanoleafs [20], nanowires [21–24], hollow spheres [25,26], and octahedrons [27]. Furthermore, SnO2 semiconductor is reported to have promising performance in detection of n-butanol [28], acetone [29], carbon monoxide [30], methanol [31], dimethyl disulfide [32] and hydrogen [33].

    View all citing articles on Scopus

    Derek R. Miller is a Ph.D. candidate at The Ohio State University. His research focuses on growing and synthesizing new oxide nanostructures for applications in gas sensing, photovoltaics, and biocompatible surfaces. His recent focus is on fabrication and characterization of oxide nano-heterostructures to enhance performance in these applications.

    Robert E.A. Williams is a Senior Research Associate Engineer at the Center for Electron Microscopy and Analysis (CEMAS) in the Department of Materials Science and Engineering at The Ohio State University. He received his Ph.D. from The Ohio State University in 2010.

    Sheikh A. Akbar is a Professor in Materials Science and Engineering at The Ohio State University. He received his Ph.D. in 1985 from Purdue University. He teaches courses on Sensor Materials, Electronic Ceramics, and Thermodynamics. He is the founding director of the NSF Center for Industrial Sensors and Measurements (CISM) and his main research interests include thin film sensors, metal oxide sensors, and nano-textured biomedical coatings.

    Pat A. Morris is an Associate Professor in Materials Science and Engineering at The Ohio State University. She received her B.S. from The Ohio State University in 1980 and Ph.D. in 1986 from MIT. Dr. Morris was a Technical Leader working on chemical sensors at Dupont from 1999 to 2005. Her current research interests focus on oxide nanoparticles deposited on miniaturized sensors using ink-jet printing.

    David W. McComb is the Director of the Center for Electron Microscopy and Analysis (CEMAS), an Ohio Research Scholar and Professor of Materials Science and Engineering at The Ohio State University. A chemistry graduate from the University of Glasgow, David completed his PhD in Physics at the University of Cambridge.

    View full text