DOI: 10.1021/cg900499c
Growth of Sn-Doped β-Ga2O3 Nanowires and Ga2O3-SnO2 Heterostructures for Gas Sensing Applications
2009, Vol. 9 4471–4479
Lena Mazeina,* Yoosuf N. Picard, Serguei I. Maximenko, F. Keith Perkins, Evan R. Glaser, Mark E. Twigg, Jaime A. Freitas Jr., and Sharka M. Prokes Electronics Science and Technology Division Naval Research Laboratory, Washington, DC 20375 Received May 7, 2009; Revised Manuscript Received August 11, 2009
ABSTRACT: Although β-Ga2O3 thin films and nanowires (NWs) show promise as very stable and reliable active components for high temperature gas sensors, their use at room temperatures is limited due to poor electrical conductivity. To address this problem, we grew Sn-doped β-Ga2O3 nanowires by the vapor-liquid-solid (VLS) approach. Sn-doped β-Ga2O3 NWs with diameters of 100-250 nm retained the monoclinic β-Ga2O3 structure, though photoluminescence (PL) emission was red-shifted by up to 50 nm relative to the deep defect band typically observed for pure β-Ga2O3 NWs. When higher amounts of Sn were introduced, individual Ga2O3-SnO2 heterostructures (HS) self-assembled, to form three distinctive parts: monocrystalline Sn-doped β-Ga2O3, poorly crystalline Sn-doped β-Ga2O3, and polycrystalline Ga-doped SnO2, thus realizing a p-n junction within a single HS. Factors responsible for the self-assembly of Ga2O3-SnO2 HS are the different vapor pressures of Sn and Ga and different growth kinetics of Ga2O3 and SnO2. Inhomogeneity in chemical content and structural composition correlated with distinct optical properties along the length of single HS. When diameters of these HS were less than 100 nm, Sn-doped Ga2O3 sections of the HS exhibited the rarely observed orthorhombic ε-Ga2O3 phase.
Introduction Monoclinic gallium oxide (β-Ga2O3), due to its high thermal resiliency, chemical stability, and low cross-sensitivity to humidity, is suitable for many applications including chemical, environmental, and explosives gas sensors.1 It was observed that Ga2O3 thin films with a slight oxygen deficiency and incomplete crystallinity were preferable for gas sensing.2 Additionally, since interactions with analytes primarily occur at the surface, ideal materials for these applications are nanostructured ones such as nanowires (NWs) which have large and highly reactive surface areas that allow for increased sensitivity per unit area. Since NWs can be readily grown and processed using traditional benchtop equipment the cost of the resultant devices might be significantly reduced as well. Recent studies of β-Ga2O3 NWs showed that they do have higher electrical conductivity as well as enhanced recombination activity with increasing temperature.3,4 However, they demonstrate poor conductivity at room temperature (RT),5 making them impractical for RT device applications relying on conductance response. One can overcome this obstacle through controlled doping and defect engineering. By controlling the conditions (gas, temperature, etc.) during growth and the dopant type, one may be able to enhance the concentration of surface defects (e.g., oxygen vacancies, surface atom coordination, or localized strain regions) that can directly impact their functionality. β-Ga2O3 has an intrinsic n-type conductivity due to oxygen vacancies and/or Ga interstitials.4,6 Tetravalent Sn is a good candidate for doping since it is not only a n-type dopant that will enhance natural n-type conductivity of β-Ga2O3, but also has a similar ionic radius (0.69 A˚) to the octahedrally coordinated Ga (0.62 A˚).7 However, Sn exhibits low solubility in
β-Ga2O3,1,8 and only small amounts of Sn have been shown to be incorporated successfully without segregation effectively improving gas sensitivity of β-Ga2O3 thin films1 and electrical conductivity of single crystals.9-11 For example, adding 0.5 at% Sn to β-Ga2O3 thin films increased1 the electrical conductivity by up to 2 orders of magnitude while also enhancing gas sensitivity,1 allowing not only a decrease in the operating temperature of the device, but also a significant reduction of the sensor gas chip size and associated heating power.1 This work reports Sn-doped β-Ga2O3 NWs, as well as individual Ga2O3-SnO2 heterostructures (HS), which were grown by the vapor-solid liquid (VLS) method. The structural and chemical characterization was performed by scanning and transmission electron microscopies (SEM and TEM, respectively). SEM-based electron dispersive spectroscopy (EDS) and electron back scatter diffraction (EBSD) were coordinated with SEM-based real-color cathodoluminescence imaging (RC-CLi). Additionally, secondary ion mass spectrometry (SIMS), Fourier transform infrared spectroscopy (FTIR), and RT photoluminescence (PL) were also used. Possible mechanisms of the formation of the obtained structures are discussed and their potential applications are highlighted. Experimental Details
*Corresponding author. E-mail:
[email protected]. Tel: 202 404 4520. Fax: 202 764 0546.
Growth. Sn-doped β-Ga2O3 NWs were grown by the VLS method12 at 900 C in a 57 mm inner diameter quartz tube inserted into a horizontal furnace of known temperature gradient. GaSn alloys, 98:2 and 92:8 wt % (Alfa Aesar, 99.999%), were used as sources. Pieces of Si (100) wafers with electron-beam evaporated 20 nm thick Au film were used as substrates and placed in the center of the furnace six inches downstream from the source. The temperature difference between the substrate and the source was ∼80 C. The tube was first evacuated to a pressure of 50 mTorr and then ramped up quickly (20/min) to 900 C. Gases (Ar and O2 with flows ∼1000 and 533 mL/min, respectively) were introduced at about 500 C. The system was held at 900 C under 5-6 Torr of pressure
r 2009 American Chemical Society
Published on Web 09/03/2009
pubs.acs.org/crystal
4472
Crystal Growth & Design, Vol. 9, No. 10, 2009
Mazeina et al.
Figure 1. SEM images of nanowires synthesized using (a) 98:2 wt % and (b) 92:8 wt % GaSn alloy (Sn-doped β-Ga2O3 NWs), and (c) metallic Ga and Sn as sources (Ga2O3-SnO2 HS). (d) High resolution SEM image of a single Ga2O3-SnO2 HS. Scale bar is 2 μm. for about 1 h and then allowed to cool down naturally under a vacuum (∼50 mTorr) without any gas flow. Growth of the Ga2O3-SnO2 HS was also performed as described above but without O2. At similar temperature and pressure, metallic Sn has almost 11 times lower vapor pressure than metallic Ga. To increase the relative pressure of Sn, Sn shots were placed in the hottest temperature zone (at 900 C), whereas Ga was placed three inches upstream in the temperature zone of ∼850 C, thus bringing the difference in the relative pressures down to a factor of 7. Substrates were placed in the center of the furnace next to the Sn shots and 3.5 in. downstream from the Ga. Characterization. The morphologies, crystalline structures, and chemical composition of the NWs and HS were analyzed by SEM and EDS using a LEO SUPRA 55 and by TEM using a JEOL 2200FS also equipped with EDS capabilities. Both mapping and point analysis modes were used to analyze the chemical composition distribution. SEM was performed on as-grown samples without any additional treatment, whereas for TEM the NWs were sonicated off the substrate, suspended in ethanol, and deposited on a Cu TEM grid covered with holey carbon film. The amount of incorporated Sn in β-Ga2O3 NWs was determined by SIMS as carried out by Evans Analytical Group, Inc. The crystalline phases of individual NWs and HS were identified by electron backscatter diffraction (EBSD). Experimental Kikuchi patterns were recorded locally for various individual NWs using a Nordlys II EBSD system (Oxford Instruments) housed in a dualbeam focused ion beam (FEI) SEM system. The EBSD patterns were acquired at an operating voltage of 10 keV and 8.4 nA current. FlamencoTM software (Oxford Instruments) was utilized to record and index the EBSD patterns. Reference patterns of the Si substrate were recorded locally for calibrating the indexing software. The structure of the resultant NWs and HS was also characterized by FTIR. For this purpose, a suspension of NWs or HS was prepared by ultrasonic agitation in ethanol and deposited onto commercial KBr polished crystal preheated at ∼150 C on a hot plate in air for rapid evaporation of ethanol. The spectra were recorded immediately after sample preparation in the 400-1000 cm-1 range with a resolution of 4 cm-1. During the analysis the spectrometer was flushed continuously with nitrogen to minimize contamination by atmospheric water and CO2. A baseline correction was applied before spectrum interpretation. RT PL properties of as-grown samples were determined using the 351 nm line of a Coherent 90-6 Argon ion laser. The PL was analyzed by a 1/4-m double-grating spectrometer with a spot size of ∼1 mm and detected by a UV-enhanced GaAs photomultiplier tube. RC-CLi was acquired using a custom-modified commercial LEO 435 VP SEM equipped with a retractable parabolic mirror. CL was excited using an electron beam at 10 kV accelerating voltage and a current of 1 nA. In order to understand the electrical properties of Sn-doped Ga2O3 NWs compared to undoped ones, we performed conductivity measurements. For this purpose, suspensions of NWs in ethanol were spread on Si substrates with a thermally grown 100 nm SiO2 layer. Subsequent to deposition, metal contacts (5 nm Ti/100 nm Au) to single NWs were made using a standard photolithography lift-off process. An array of such devices was found to incorporate up to five NWs per device. Electrical measurements of
serendipitously obtained single NW devices were performed using a three contact probe station.
Results Sn-Doped β-Ga2O3 NWs. SEM showed that the NWs synthesized using GaSn alloys had morphologies similar to that of pure Ga2O3 NWs (Figure 1a). However, NWs obtained using 92:8 wt % Ga:Sn alloy were somewhat less straight and more kinky (Figure 1b). Both types of NWs had thicknesses of 100-250 nm and lengths up to several micrometers. The maximum amount of Sn that can be incorporated into β-Ga2O3 NWs taking into account the vapor pressure at ∼880-900 C of Sn and a positive deviation from Raoult’s law for SnGa alloys13 is 0.2 and 0.9 wt % for 98:2 and 92:8 wt % GaSn alloys, respectively. For clarity, the dopant concentrations will be further reported using cation fraction x for formulas Ga1-xSnxOz ( δ or cation percent 100•x. The value z is equal either to 1.5 for Ga2O3-based solid solutions or 2 for SnO2-based solid solutions. Thus, possible maximum concentrations of incorporated Sn would be x = 0.002 and 0.007, respectively. EDS analysis of Sn-doped β-Ga2O3 NWs showed that the amount of introduced Sn for samples obtained using both 98:2 and 92:8 wt % GaSn alloys is 0.5-1.0 wt % (x = 0.004-0.008, not shown). However, this amount is right on the edge of the EDS detection limit for Sn at the given conditions. SIMS analysis showed that the Sn concentration did not vary significantly as a function of the proximity from the surface and was the same for both types of grown NWs: 1-1.2 1022 atoms/cm3. Using the density of bulk β-Ga2O3 6.44 g/cm3, the Sn concentration x for both samples was calculated to be between 0.0014 ( 0.0007 and 0.0030 ( 0.0015 for Ga1-xSnxO1.5 ( δ. Structural analysis performed by FTIR spectroscopy showed a spectrum (Figure 2) consistent with the β-Ga2O3 structure.14 The Au peaks at 662 and 450 cm-1 dominate, indicating that the majority of these NWs grow in the [010] direction,15 which exhibits the highest electrical conductivity for β-Ga2O3 NWs.16,17 An additional peak is observed at 586 cm-1 that cannot be identified with any phase and may have appeared due to lattice distortion by the presence of Sn. This peak is the strongest for the sample grown using 92:8 wt % GaSn alloy and appears only as a weak shoulder for the NWs grown using 98:2 wt % GaSn alloy. The RT PL spectrum showed a consistent red shift of the broad defect band relative to pure β-Ga2O3 NWs (Figure 3): 50 nm shift for samples grown using 98:2 wt % GaSn alloy (Figure 3d) and ∼20 nm shift for samples grown using 92:8 wt % GaSn alloy. Since RT PL of the samples grown
Article
Figure 2. FTIR spectrum of pure β-Ga2O3 NWs, Ga2O3-SnO2 HS and Sn-doped β-Ga2O3 NWs grown using 98:2 (1) and 92:8 (2) wt% GaSn alloys.
Figure 3. PL spectra of pure β-Ga2O3 NWs (a) and Sn-doped β-Ga2O3 NWs obtained using 92:8 (b) and 98:2 wt % (c) GaSn alloys.
using 98:2 wt % SnGa showed a larger red shift compared to pure β-Ga2O3, additional structural (EBSD and TEM) and CL studies were performed on these samples. High spatial resolution RT CL measurements of individual Sn-doped β-Ga2O3 NWs grown using 98:2 wt % GaSn alloy also showed a ∼30 nm red shift of the emission band relative to pure β-Ga2O3 NWs (not shown). The spectral distribution of these emission bands is quite uniform across individual Sn-doped β-Ga2O3 NWs, which is good evidence for homogeneous Sn distribution.18 TEM analysis confirmed that Sn-doped NWs are monoclinic β-Ga2O3 NWs (a = 12.23 A˚, b = 3.04, c = 5.80, β = 103.719) and showed that NWs are several micrometers long and 20-150 nm in diameter. Sample preparation of these NWs for TEM can sometimes cause the NWs to conglomerate, as indicated by higher magnification TEM imaging (Figure 4b) where two NWs are shown separated by a small gap. The recorded selected area diffraction (SAD) pattern (Figure 4c) from these NWs is composed of two distinct, overlapping diffraction patterns corresponding to β-Ga2O3 at zone axes (ZA) of [101] and [001]. Simulated diffraction patterns generated for β-Ga2O3 using the JEMS software package are indexed and presented in Figure 4d,e. Fast Fourier-transform (FFT) patterns extracted from two points across the diameter of the larger NW match the simulated SAD patterns (Figure 4d,e). This suggests a tilt boundary
Crystal Growth & Design, Vol. 9, No. 10, 2009
4473
parallel to the longitudinal axis of the larger NW in Figures 4. Also, the experimental SAD pattern indicates the larger NW consists of stacked (010) planes. On the basis of this result and FTIR spectrum, we conclude that the NW grows in the [010] direction, a previously observed growth direction for pure β-Ga2O3 NWs.15 An FFT pattern from the smaller NW shows only two reflections that closely match the position of the 020 reflections (highlighted by circles in all three FFT patterns) of the other NW, suggesting the smaller NW is also elongated parallel to the [010] direction. A much shorter (30 nm diameter) NWs additionally confirmed these Sn-doped NWs are β-Ga2O3 (Figure 6). Ga2O3-SnO2 Heterostructures. VLS growths with metallic Sn and Ga sources exhibited two types of HS, both kinky, with diameters of 20-100 nm and 500-600 nm, and micrometers in lengths (Figure 1c). PL spectra were very broad (not shown) with multiple shoulders indicating inhomogeneity of the materials. CL studies at the SEM resolution showed that each individual HS had highly inhomogeneous CL emission spectra which transitions from blue to green to red (Figure 7a). The red emission is clearly dominant at the top region of the wires, while the regions approaching the Si substrate show increasing blue emission. Spot-CL analysis was performed to verify emission in each individual color emission region (blue, green, red) and will be discussed below. To obtain the chemical composition of the 500-600 nm HS, a mapping mode of EDS was performed (Figure 7). Although this method usually gives semiquantitative results, it did show that the distribution of Sn and Ga is not homogeneous throughout the length of the NW (Figure 7c,d). The part of the NW near the substrate is Ga-rich (Figure 7c), whereas the top part of the NW is Sn-rich (Figure 7d). The middle (transition) part of the NW has nearly equal parts of Sn and Ga. Thus, three distinct zones of the NW can be distinguished in each HS: a Ga-rich zone, a Sn-rich zone, and a transition zone. The presence of the Au droplet on the top of the NW confirms the VLS mechanism (Figure 7e). To determine the influence of varying chemical compositions on the crystallographic structure, spot-EBSD analysis was performed at several positions (labeled in Figure 7b) along the same HS presented in Figure 7. EBSD analysis of the Sn-rich zone (points 1-6, 8) showed a rutile-type SnO2 structure (Figure 8a,b). The transition zone (points 7, 9-13) showed Kikuchi patterns exhibiting weak Kikuchi lines due to poor crystallinity but nevertheless indexable as β-Ga2O3 (Figure 8c,d) while the Ga-rich zone (points 12-21) was strongly crystalline β-Ga2O3 (Figure 8e,f). The SnO2 region exhibited a different orientation for nearly every point identified in Figure 7b and the β-Ga2O3 regions showed variations in orientation too, although with less frequency (every three points). The changes in chemical composition and structure directly correlate with observed changes in the spectral emission distribution along the HS (Figure 7a). The rutile SnO2 region had red emission peaked at ∼600 nm (2.07 eV), which corresponds to the defect emission in SnO2.20,21 The transition zone showed a broad peak with a maximum at 550 nm (2.25 eV). The higher crystalline quality β-Ga2O3 region emitted in the blue region with the peak maximum at 410 nm (3.02 eV), which corresponds to the defect CL
4474
Crystal Growth & Design, Vol. 9, No. 10, 2009
Mazeina et al.
Figure 4. TEM micrographs at (a) low and (b) high magnification of Sn-doped β-Ga2O3 NWs and (c) corresponding SAD pattern exhibiting two overlapping electron diffraction patterns, both simulated for the β-Ga2O3 (d) [101] and (e) [001] zone axis. FFT patterns (f-h) extracted from three points in high-resolution, phase-contrast image exhibit three distinct patterns, two of which match the experimentally observed SAD patterns.
Figure 5. TEM micrographs at (a) low and (b) high magnification of a single, Sn-doped, monoclinic β-Ga2O3 NW. (c) SAD pattern collected from the NW confirmed the β-Ga2O3 phase and matched the FFT (d) extracted from the high-resolution phase contrastimage.
emission of the Sn-doped β-Ga2O3 NWs.18 A more detailed RC-CLi study of these structures will be reported elsewhere.18
The quantitative EDS results are shown in Figure 7g. The concentration of Sn in the Ga-rich “blue” area increases from x = 0.02 to x = 0.06 toward the green “transition” zone, giving a formula of Ga0.94Sn0.06O1.5 ( δ. The incorporation of Sn into Ga2O3 in the “green” transition zone varies from x = 0.10 at the border with the “blue” region to x = 0.33 at the border with the “red” zone. The red-emitting zone of the NW, which is Sn rich, has a Ga concentration changing from 0.23 to 0.11 closer to the tip of the NW, yielding the formula Sn0.89Ga0.11O2 ( δ. TEM analysis was performed to further confirm the structural aspects of the different parts of the Ga2O3-SnO2 HS. Because of the sonication process and fragility of Ga2O3-SnO2 HS, only fragments of different NWs were deposited across the holey carbon film during TEM sample preparation. However, we employed CL analysis of the prepared Cu grid samples prior to TEM analysis in order to identify different (blue and red) luminescent NW fragments as shown in Figures 9-11. The TEM analysis confirmed that blue luminescent fragments of the large (400-600 nm) diameter HS are strongly monoclinic β-Ga2O3 (Figure 9). Higher magnification TEM imaging resolves lattice fringes (Figure 9) with two orientations, as confirmed by FFT (Figure 9c,d). A twin boundary can be observed propagating along the length of the blue luminescent section of the NW. Thus, this NW is a bicrystal similar to the one shown in Figure 4. The two orientations in this bicrystal produce two overlapping diffraction patterns in the
Article
Crystal Growth & Design, Vol. 9, No. 10, 2009
4475
Figure 6. (a) SEM micrograph and (b) Kikuchi pattern recorded by EBSD from a single, Sn-doped β-Ga2O3 NW. The indexed Kikuchi pattern (c) confirms the NW is monoclinic β-Ga2O3.
Figure 7. (a) RC CL and (b) SEM images of the same Ga2O3-SnO2 HS. (c-f) Mapping mode of EDS showing distribution of different elements throughout the length of the HS; numbers in (b) indicate positions where EBSD analysis was made (results are shown in Figure 8). (g) Concentrations of Sn (x) and Ga (1 - x) along the HS taken as marked on (a).
Figure 8. Kikuchi patterns recorded by EBSD at the (a) red luminescent region indexed as (b) rutile SnO2, (c) the weakly green luminescent transition region indexed as (d) monoclinic β-Ga2O3, and (e) the blue luminescent region indexed as (f) β-Ga2O3. Kikuchi patterns recorded in the transition region exhibited more background noise and weaker Kikuchi lines than the blue luminescent regions, possibly due to the presence of substantial amorphous, poorly, or nanocrystalline material.
SAD pattern recorded across the full NW width (Figure 9e). These two overlapping patterns correspond to β-Ga2O3 [150] and [150], simulated and indexed in Figure 9, panels f and g, respectively. Local FFT patterns recorded from the upper (Figure 9d) and lower (Figure 9e) portions of the NW combined with the indexed SAD pattern (Figure 9c) elucidate the crystallographic nature of the twin boundary. Specifically, inversion symmetry is evident, where the upper
portion ([150] ZA) is flipped so that the 514 and 514 reflections (squares in Figure 9c-e) remain unchanged while the 512 and 512 reflections (triangle, Figure 9d,e) exchange positions. Orientation rotation across the width of these NWs was also observed by EBSD. In addition, crystallographic boundaries perpendicular to the NW longitudinal axis were also observed in areas close to the transition zone of the NW (not shown). Finally, TEM results also confirmed
4476
Crystal Growth & Design, Vol. 9, No. 10, 2009
Mazeina et al.
Figure 9. TEM micrographs (a, b) of the blue luminescent section of a Ga2O3-SnO2 NW imaged by CL in the inset of (a). A SAD pattern (e) of the entire NW width consists of two overlying diffraction patterns, both β-Ga2O3, that are [150] and [150]. Lattice fringes (b) indicate a distinct twin boundary, and corresponding FFT patterns from the (c) upper and (d) lower portions of the NW show the inversion symmetry of the observed twin boundary (dotted line in (b). Simulated and indexed diffraction patterns for the (f) [150] and (g) [150] ZA elucidate the inversion symmetry of the twin boundary, where (f) is flipped relative to (g) about the axis between crossing 514 and 514 reflections (denoted by squares) so that the 512 and 512 reflections (denoted by triangles) exchange positions.
The plane spacings for the 010 and 001 reflections (b and c lattice parameters) measured from the [100] ZA SAD patterns reasonably matched literature values determined by X-ray diffraction (see Table 1). Also, both NWs appear to grow along the Æ001æ directions. These ε-Ga2O3 NWs were too thin to yield an indexable Kikuchi pattern by EBSD at 10 kV. Lastly, FTIR of Ga2O3-SnO2 HS yielded spectra with broad bands at 611 and 708 cm-1 that can be attributed to Eu and A2u modes, respectively, of rutile SnO224,25 (Figure 2). The IR mode at 563 cm-1 has been observed for nanoSnO226,27 and corresponds to the surface mode in SnO2.25 Except for the two weak shoulders at ∼450 and 660 cm-1, which could be attributed to β-Ga2O3, no other significant modes can be identified in the spectrum. Discussion
Figure 10. Low (a, b) and high (c) resolution TEM images with SAD pattern recorded near the [113] ZA (d). (a) Inset shows real color CL-image of the NW used for TEM analysis which emits in the red regions of the spectrum.
that the red-luminescent NW fragments were polycrystalline rutile SnO2 (Figure 10). It is interesting to note that TEM analysis of the smaller (20-100 nm) NWs did not produce SAD patterns exhibiting the monoclinic β-Ga2O3 phase but instead were found to match the less abundant polymorph orthorhombic ε-Ga2O3. Two examples of such smaller NWs, 100 nm (Figure 11a-c) and 50 nm (Figure 11d-g) in diameter, show SAD patterns corresponding to the orthorhombic ε-Ga2O3 structure. Sn content was x = 0.01, resulting in the formula Ga0.99Sn0.01O1.48, according to the TEM resolution EDS analysis (Figure 11g).
Our results show that n-type β-Ga2O3 NWs can be easily formed by Sn-doping directly during VLS growth using commercially available GaSn alloys as sources. The defect band in the PL spectra of these NWs is red-shifted by up to 50 nm relative to pure β-Ga2O3 NWs. The predominant native defects in β-Ga2O3 are oxygen and Ga vacancies and interstitials.28,29 Defect blue emission in the luminescence spectrum of β-Ga2O3 is normally associated with the electron-donor recombination at O vacancies, hole-acceptor recombination at Ga vacancies or by carrier recombination at Ga-O vacancy pairs.30,31 Incorporation of Sn4þ happens by the substitution of octahedrally coordinated Ga atoms, creating solid solutions and introducing a positive effective charge in the lattice29 and/ or an additional electron1 according to different models. This leads to various additional defects, such as GaII defect states that can act as donors1 and/or SnIII or SnII will be formed, leading to the formation of additional O vacancies,1 thus creating a potential for electrical conductivity enhancement if no planar defects, surface segregation and/or phase transitions occur. The formation of new defect levels and new
Article
Crystal Growth & Design, Vol. 9, No. 10, 2009
4477
Figure 11. TEM images of a 100 nm diameter Sn-doped Ga2O3 NW at (a) low magnification (with corresponding color CL image as an inset) and (b) higher magnification with corresponding (c) SAD pattern indexed as ε-Ga2O3. Another 50 nm diameter Sn-doped Ga2O3 NW imaged by (d) TEM and analyzed by (e) SAD also shows the NW is ε-Ga2O3. Both NWs appear to grow along the Æ001æ directions. (f) EDS spectra recorded from the 30 nm diameter NW show a Sn content of 0.5 at%. Table 1. Measured b and c Lattice Parameters for ε-Ga2O3 Obtained from Recorded SAD Patterns for Straight Fragments of Ga2O3-SnO2 HS with Sizes 100 and 50 nm (Figure 11, panels b and d, respectively)a a b c a
100 nm wire
50 nm wire
Matsuzaki et al.22
Kroll et al.23
NA 8.50 ( 0.17 9.16 ( 0.18
NA 8.93 ( 0.17 9.18 ( 0.18
5.04 8.73 9.26
5.0566 8.6873 9.3072
Note that a lattice parameter was not obtained in the present study.
transitions between them is the most likely cause of the red shift of Sn-doped β-Ga2O3 NWs relative to the pure ones. We expect that these defects produce shallow levels, since a broad defect PL emission was red-shifted by only ∼0.2-0.4 eV. Electrical measurements of pure and Sn-doped β-Ga2O3 NWs would provide direct experimental evidence of the effect of defects created by Sn incorporation on conductivity. However, the electrical measurements showed different behavior: doped NWs consistently showed a negative differential resistance (see Supporting Information, Figure S1), but results for both types of wires were not reproducible from device to device. Since it has been reported in the literature that NWs grown by VLS are unintentionally doped with the catalyst material,32 it is not excluded that as a result of this doping different parts of the NWs might have different electrical behavior, although this has not been reported in the literature for semiconductor metal oxides. We believe that during the sonication process, the NWs break apart with the resulting shorter NW fragments having different electrical properties based on the assumption above. To describe possible differences in electrical behavior of different parts of the NWs and differences between Sn-doped β-Ga2O3 and pure β-Ga2O3 quantitatively, one would need in situ contact fabrication to a single NW using E-beam nanolithography or focused ionbeam, which is currently underway in our laboratory. When pure Sn and pure Ga are used as sources, higher concentrations of Sn are introduced into the system and Ga2O3-SnO2 HS form as a result. Since the relative pressure of metallic Ga is about 7 times higher at our experimental conditions than that of Sn, Ga dissolved in a greater amounts
in the Au catalyst at the initial stage, forming the Sn-doped β-Ga2O3 NWs (Figure 8e,f). As the Ga concentration decreases in the vapor phase during the formation of the Ga2O3 NWs, the relative Sn concentration in the vapor phase increases. Thus, higher amounts of Sn begin to incorporate into the already formed β-Ga2O3 NWs. Further increase of the Sn incorporation induces strain and distortion, resulting in increasingly poorer crystalline quality as observed by EBSD in the middle (“green” transition) zones HS. Because of relatively high concentrations of Sn in the transition region, we suspect that this region may contain poorly crystalline and/or amorphous SnOx undetectable by EBSD and difficult to identify by TEM. As the Sn continues to incorporate in higher concentrations, a rutile SnO2 phase begins to form. Strain and distortion from the transition zone and high Ga content from the red-emitting zone likely accumulate significant lattice strain and thus the conditions are unfavorable for singlecrystalline SnO2 growth, resulting in polycrystalline SnO2 (Figure 10). Thus, each Ga2O3-SnO2 HS consists of three distinctive parts: monocrystalline Sn-doped β-Ga2O3, poorly crystalline Sn-doped β-Ga2O3, and polycrystalline Ga-doped SnO2. Since Sn is an n-type dopant in Ga2O3 and Ga is a p-type dopants in SnO2, a p-n junction is realized within each individual HS. The maximum concentration of Ga incorporated in the polycrystalline SnO2 parts of the HS was found to be x = 0.11-0.23, which is relatively high when considering the limited solubility of Ga in SnO2 as reported in the literature.8 Only one literature report exists for Ga-doped SnO2 NS33 and for Ga-doped SnO2 thin films34 though without specifying the actual concentration of Ga in SnO2. It is not excluded that some amorphous or fine-grained admixtures of Ga2O3 material are still present within the polycrystalline SnO2. However, EBSD (Figure 8a) and TEM (Figure 10) showed very distinct patterns of SnO2 without any additional phases. The maximum concentration of Sn in single crystalline parts of the HS (“blue” regions) was found to be up to x = 0.06, without any observation of SnO2 segregation or admixtures, which is confirmed by RC-CLi, EBSD, and EDS
4478
Crystal Growth & Design, Vol. 9, No. 10, 2009
analyses. At higher concentrations of Sn (x = 0.11-0.33), the structure becomes distorted and poorly crystalline (“green” regions) but still crystallizes as β-Ga2O3 with possible admixture of amorphous Ga2O3 and/or SnO2. In general, the solubility of Sn in β-Ga2O3 is reported to be very low ; less than 0.5-1 mol % for both bulk powders8 and thin films.1 It was reported that at higher concentrations SnO2 starts segregating at the grain boundaries and SnO2 peaks can be detected by X-ray diffraction.1 However for nanometersized thin films, this solubility is somewhat higher: no surface segregation is observed for 100-nm thin films with 1 mol % of SnO2 incorporated.30 Slightly thicker films did not show any segregation though incorporation of Sn did cause a phase transition from β-Ga2O3 to ε-Ga2O3.22 Similar phase transitions were observed for 100 nm thin films grown on sapphire but with 2.5-4 cation % of incorporated Sn.35 At the same time, similar 100 nm films with the same Sn concentration but grown on silicon substrates showed no β f ε phase transition.35 In our work, ε-Ga2O3 phase was observed only when NW diameters were 20-100 nm with Sn concentration x = 0.01. Taking all these facts into account, we conclude that several factors play a role in β f ε phase transition. First, as the thickness of NWs and thin films decrease and surface areas increase, surface energy starts playing a more significant role in the thermodynamic stability of one phase over another. β-Ga2O3, which is more thermodynamically stable in bulk, becomes metastable relative to ε-Ga2O3 at the nanoscale due to the higher surface energy, as was observed for a variety of other oxide polymorphs.36 Since the surface energy of ε-Ga2O3 is unknown, it is hard to predict at what surface area (i.e., particle size) ε-Ga2O3 becomes more thermodynamically stable. However, a direct confirmation of this interplay was observed during mechanical milling37 where β-Ga2O3 transformed into ε-Ga2O3 when the particle size decreased. However, particle and/or thin film size alone cannot fully explain this transition. Numerous literature reports exist where Ga2O3 NWs with diameters 100 nm and less were obtained at different temperatures and exhibited the monoclinic, β-Ga2O3, structures. Second, β f ε transition is also induced by Sn incorporation22,35,38 which in addition to the particle size and or thin film thickness stabilizes the ε-polymorph over β. It is worth noting, that for thinner films,30 Sn solubility is reported to be higher.1 However, the fact that thin films grown on silicon with up to 4 mol % of SnO2 did not show any ε-phase compared to similar β-Ga2O3 thin films grown on sapphire suggests that β f ε transition is a complicated process induced by certain strains, whether they are size-related, caused by incorporation of Sn or by the interaction with the substrate, or by the combination of all three factors. Potential Applications. One of the main applications considered for Sn-doped β-Ga2O3 NWs and Ga2O3-SnO2 HS grown in this work is gas sensing, where sensing devices with broad sensing capabilities ; such as detecting toxic gases and trace components in gas mixtures ; and high selectivity are needed. Such devices will require gas sensor arrays with multiple materials or with tunable material properties achieved by doping or formation of core-shell and heterostructures. Since interactions with the analyte occur mostly on the surface, surface defects play an important role, especially at nanoscale. For materials with nanometer-dimensions, these surface defects will directly correlate with any structural defects present in NS. Sn-doped β-Ga2O3 NW
Mazeina et al.
might exhibit more of these defects compared to pure β-Ga2O3 NWs, which may enhance the sensitivity of devices based on these materials, as was shown for thin films.1 Similarly, Ga2O3-SnO2 HS show special promise as gas sensing materials not only due to the potential for forming and tailoring surface defects formation, but also because they contain different materials with different sensing capabilities, including doped, poorly, and polycrystalline materials. In fact, Sn-doped β-Ga2O3,1,2,9-11 polycrystalline SnO2,39 and Ga-doped SnO234 showed enhanced gas responses compared to undoped or single crystalline materials. Additionally, a unique mechanism of formation of Ga2O3-SnO2 HS where the Ga2O3-rich part nucleates first has a big advantage for the gas sensing device fabrication as it was recently shown in our laboratory. Since there is significant variability in defect type and chemical composition in single NWs, a more reliable gas sensor would depend on the response from a NW ensemble. In addition, HS are too fragile for contacts fabrication by standard regular photolithography process. Thus, it is may be more beneficial to fabricate capacitance-based sensors40,41 rather than conductance based devices, a strategy currently being explored in our laboratory. Similar sensors have been successfully manufactured based on the capacitance change of carbon nanotubes. Our first preliminary tests have shown that capacitance-based sensors can be also successfully produced using metal-oxide NWs with standard bench techniques that do not require expensive equipment.42 Further studies of these sensors are currently underway. Additionally, some interesting and useful properties might result from the self-assembled p-n junction in Ga2O3-SnO2 HS which can be studied once contacts are fabricated to distinct regions of the HS. Conclusions We have formed Sn-containing Ga2O3 nanostructures using VLS processes and performed extensive structural, chemical, and optical characterization. High quality, straight Sn-doped β-Ga2O3 nanowires (NWs) were grown, containing 0.1-0.3 cation % of Sn. These NWs grew in the [010] direction, which has the highest electrical conductivity, and the defect PL emission was red-shifted by 50 nm relative to pure β-Ga2O3 NWs. We have found that Sn is distributed homogeneously without any surface segregation. We have also formed Ga2O3-SnO2 heterostructures (HS), which self-assemble in Sn-enriched atmosphere and consist of three distinctive parts: monocrystalline Sn-doped β-Ga2O3, poorly crystalline Sn-doped β-Ga2O3, and polycrystalline Gadoped SnO2, thus realizing a p-n junction within a single nanostructure. Factors responsible for the self-assembly of Ga2O3-SnO2 HS are different vapor pressures of Sn and Ga and different growth kinetics of Ga2O3 and SnO2. Changes in chemical and structural composition directly correlate with the observed optical properties. In particular, individual NWs exhibit a transition from a blue luminescent Ga-rich region to a red-luminescent Sn-rich region with a green-luminescent, intermediate region containing both Sn and Ga in substantial concentrations. Incorporation of Sn does not cause any β-Ga2O3 f ε-Ga2O3 phase transition for large-diameter (400-600 nm) Ga2O3-SnO2 HS but does cause significant distortion of the lattice, resulting in stacking faults. However, small diameter (20-100 nm) Ga2O3-SnO2 HS exhibited an orthorhombic
Article
Crystal Growth & Design, Vol. 9, No. 10, 2009
ε-Ga2O3 phase, which is rarely observed and has been previously tied to Sn incorporation into β-Ga2O3. Because of their interesting composition and structural properties, these nanowire systems may provide advantages in gas sensing applications, as well as other device applications requiring in situ formation of p-n junctions. Acknowledgment. This work was supported by the Office of Naval Research. L.M., Y.N.P., and S.I.M. thank the National Research Council (NRC) program for administrative support. Victor Bermudez, Jeremy Robinson, Steve Arnold, and Antti Makinen are thanked for help with instrumental set-ups, analyses, and useful discussions. Supporting Information Available: IV-curve for pure and Sn-doped Ga2O3 nanowires (Figure S1). This material is available free of charge via the Internet at http://pubs.acs.org.
References (1) Frank, J.; Fleischer, M.; Meixner, H.; Feltz, A. Sens. Actuators, B: Chem. 1998, B49, 110–114. (2) Ogita, M.; Saika, N.; Nakanishi, Y.; Hatanaka, Y. Appl. Surf. Sci. 1999, 142, 188–191. (3) Prokes, S. M.; Carlos, W. E.; Glembocki, O. J. Proc. SPIE Intern. Soc. Opt. Eng. 2005, 6008, 60080C/1–60080C/10. (4) Cojocaru, L. N.; Alecu, I. D. Z. Phys. Chem. 1973, 84, 325–331. (5) Li, Y.; Trinchi, A.; Wlodarski, W.; Galatsis, K.; Kalantar-Zadeh, K. Sens. Actuators, B: Chem. 2003, B93, 431–434. (6) Harwig, T.; Wubs, G. J.; Dirksen, G. J. Solid State Commun. 1976, 18, 1223–1225. (7) Shannon, R. D.; Prewitt, C. T. Acta Crystallogr., Sec. B: Struct. Crystallogr. Cryst. Chem. 1969, 25, 925–946. (8) Edwards, D. D.; Mason, T. O. J. Am. Ceram. Soc. 1998, 81, 3285– 3292. (9) Zhang, J.; Xia, C.; Deng, Q.; Xu, W.; Shi, H.; Wu, F.; Xu, J. J. Phys. Chem. Sol. 2006, 67, 1656–1659. (10) Ueda, N.; Hosono, H.; Waseda, R.; Kawazoe, H. Appl. Phys. Lett. 1997, 70, 3561–3563. (11) Suzuki, N.; Ohira, S.; Tanaka, M.; Sugawara, T.; Nakajima, K.; Shishido, T. Phys. Status Solidi C 2007, 4, 2310–2313. (12) Wagner, R. S.; Ellis, W. C. Appl. Phys. Lett. 1964, 4, 89–90. (13) Zivkovic, D.; Manasijevic, D.; Zivkovic, Z. J. Therm. Anal. Calorim. 2003, 74, 85–96. (14) Bermudez, V. M.; Prokes, S. M. Langmuir 2007, 23, 12566–12576. (15) Mazeina, L.; Picard, Y. N.; Prokes, S. M. Cryst. Growth Des. 2009, 9, 1164–1169. (16) Binet, L.; Gourier, D.; Minot, C. J. Solid State Chem 1994, 113, 420–433.
4479
(17) Ueda, N.; Hosono, H.; Waseda, R.; Kawazoe, H. Appl. Phys. Lett. 1997, 71, 933–935. (18) Maximenko, S. I.; Mazeina, L.; Picard, Y. N.; Freitas, J. A., Jr.; Bermudez, V. M.; Prokes, S. M. Nano Lett., accepted. (19) Geller, S. J. Chem. Phys. 1960, 33, 676–684. (20) Calestani, D.; Zha, M.; Zappettini, A.; Lazzarini, L.; Salviati, G.; Zanotti, L.; Sberveglieri, G. Mater. Sci. Eng., C 2005, 25, 625–630. (21) Zhou, X. T.; Heigl, F.; Murphy, M. W.; Sham, T. K.; Regier, T.; Coulthard, I.; Blyth, R. I. R. Appl. Phys. Lett. 2006, 89, 2131091/1– 213109/3. (22) Matsuzaki, K.; Yanagi, H.; Kamiya, T.; Hiramatsu, H.; Nomura, K.; Hirano, M.; Hosono, H. Appl. Phys. Lett. 2006, 88, 092106/1– 092106/3. (23) Kroll, P.; Dronskowski, R.; Martin, M. J. Mater. Chem. 2005, 15, 3296–3302. (24) Katiyar, R. S.; Dawson, P.; Hargreave, M. M.; Wilkinson, G. R. J. Phys. C: Solid State Phys. 1971, 4, 2421–2431. (25) Abello, L.; Bochu, B.; Gaskov, A.; Koudryavtseva, S.; Lucazeau, G.; Roumyantseva, M. J. Sol. St. Chem. 1998, 135, 78–85. (26) Peng, X. S.; Zhang, L. D.; Meng, G. W.; Tian, Y. T.; Lin, Y.; Geng, B. Y.; Sun, S. H. J. Appl. Phys. 2003, 93, 1760–1763. (27) Liu, Y.; Yang, F.; Yang, X. Colloids Surf., A 2008, 312, 219–225. (28) Harwig, T.; Schoonman, J. J. Solid State Chem. 1978, 23, 205–211. (29) Blanco, M. A.; Sahariah, M. B.; Jiang, H.; Costales, A.; Pandey, R. Phys. Rev. B: Condens. Matter Mater. Phys. 2005, 72, 184103/1– 184103/16. (30) Orita, M.; Ohta, H.; Hirano, M.; Hosono, H. Appl. Phys. Lett. 2000, 77, 4166–4168. (31) Harwig, T.; Kellendonk, F. J. Solid State Chem. 1978, 24, 255–263. (32) Allen, J. E.; Hemesath, E. R.; Perea, D. E.; Lensch-Falk, J. L.; Li, Z. Y.; Yin, F.; Gass, M. H.; Wang, P.; Bleloch, A. L.; Palmer, R. E.; Lauhon, L. J. Nat. Nanotechnol. 2008, 3, 168–173. (33) Su, Y.; Xu, L.; Liang, X.-M.; Chen, Y.-Q. Chin. J. Chem. Phys. 2008, 21, 181–186. (34) Tiburcio-Silver, A.; Sanchez-Juarez, A. Mater. Sci. Eng., B 2004, B110, 268–271. (35) Orita, M.; Hiramatsu, H.; Ohta, H.; Hirano, M.; Hosono, H. Thin Solid Films 2002, 411, 134–139. (36) Navrotsky, A. Geochem. Trans. 2003, 34–37. (37) Zhao, M.; Chen, X.-L.; Wang, W.-J; Zhang, Z.-H.; Xu, Y.-P. Chin. Phys. Lett. 2007, 24, 2401–2404. (38) Yoshioka, S.; Hayashi, H.; Kuwabara, A.; Oba, F.; Matsunaga, K.; Tanaka, I. J. Phys.: Condens. Matter 2007, 19, 346211/1– 346211/11. (39) Xu, C.; Tamaki, J.; Miura, N.; Yamazoe, N. Chem. Lett. 1990, 3, 441–444. (40) Snow, E. S.; Perkins, F. K.; Houser, E. J.; Badescu, S. C.; Reinecke, T. L. Science 2005, 307, 1942–1945. (41) Snow, E. S.; Perkins, F. K. Nano Lett. 2005, 5, 2414–2417. (42) Arnold, S. P.; Prokes, S. M.; Perkins, F. K.; Zaghloul, M. E. 2009, Appl. Phys. Lett., accepted.