Novel Pressure-Induced Structural Transformations of Inorganic Nanowires

1. Introduction

Nanostructured materials in different morphologies such as dots, wires and belts are of fundamental importance because of their wide range of tunable electrical, optical and mechanical properties that the bulk materials do not possess. These properties that critically depend on the nano-structures predominantly determined by the synthetic approaches, however, can be substantially modified by compression (Chen et al., 2002; Chen & Herhold, 1997; Guo et al., 2008; He et al., 2005; Jacobs et al., 2001; Jiang & Gerward, 2000; Jiang et al., 1999; Jiang et al., 2001; Jiang et al., 1998; Park et al., 2008; Swamy et al., 2006; Tolbert & Alivisatos, 1994; Tolbert & Alivisatos, 1995; Wang et al., 2005; Wang et al., 2001). Investigations of the structures and phase transformations of nanomaterials under high pressures have received increasing attention simply because high pressure has proven to be a powerful driving force to produce new structures and, therefore, new nanomaterial properties (San-Miguel, 2006). The most interesting aspect of high-pressure studies on nanomaterials is the observation that compressed nanomaterials exhibit significantly different behaviours than their corresponding bulk counterparts, such as the size-dependent phase transformations observed for nano-scale CdSe (Tolbert & Alivisatos, 1994; Tolbert & Alivisatos, 1995), SnO₂ (He et al., 2005; Jiang et al., 2001) and TiO₂ (Swamy et al., 2006). In addition, morphology can play an important tuning role in the pressure-induced transformations of nanostructured materials. For instance, ZnS nanobelts have been found to exhibit a much wider stability region up to 6.8 GPa for the wurtzite phase, in strong contrast to bulk ZnS, which is much more stable in the sphalerite phase (Wang et al., 2005).

In this chapter, we focus our interests on one-dimensional inorganic nanostructured materials, such as nanowires, due to their wide range of applications. Table 1summarizes all high-pressure studies on one dimensional inorganic nanomaterialsso far. In particular, we will discuss the novel and unusual pressure behaviors oftwo inorganic nanowires with great technological importance: tin dioxide (SnO₂)and gallium nitride (GaN). Our findings in these studies demonstrate the potentialof using the combination of pressure and morphology as a powerful approach to tunethe structures and properties of one-dimensional nanomaterials required for specific applications.

Material	Morphology	Dimension *	Pressure (GPa)	Characterization Method	Reference
ZnS	Nanobelt	t: ~10 nm; w: ~1 μm; l: 100 μm	0-11.4	X-ray diffraction	(Wang et al., 2005)
Nanorod	w: ~10 nm	0-19.4	Raman, Photoluminescence	(Li et al., 2007)
w: ~10 nm	0-37.2	X-ray diffraction	(Li et al., 2011)
BN	Nanotube	d: 20-50 nm	0-16	Raman	(Saha et al., 2007; Saha et al., 2006)
d: ~ 50 nm	0-19.1	X-ray diffraction	(Muthu et al., 2008)
d: ~ 100 nm	0-34.6	FTIR	(Dong & Song, 2010)
ZnO	Nanowire	w: 60-100 nm; l: tens of μm	0-21.5	Raman,; X-ray diffraction	(Yan et al., 2009)
Nanotube	d:10-70 nm	0-21.5	X-ray diffraction	(Hou et al., 2009)
SnO2	Nanowire	w: 50-60 nm; l: several μm	0-37.9	Raman,; X-ray diffraction	(Dong & Song, 2009)
Nanobelt	t: tens of nm; w: ~1μm; l: several μm	0-36.2	Raman,; X-ray diffraction	(Dong & Song, 2009)
TiO2	Nanoribbon	t: ~20 nm; w: 50-200 nm; l: tens of μm	0-30.9	Raman,; X-ray diffraction	(Li et al., 2010)
GaN	Nanowire	w: tens of μm; l: < 100 nm	0-65	X-ray diffraction	(Dong & Song, 2010)

Table 1.

Summary of high-pressure studies on one-dimensional inorganic nanomaterials.

* d: diameter; w: width; l: length; t: thickness

2. Experimental details

2.1. Materials

SnO₂ nanobelts and nanowires were synthesized using chemical vapor deposition on silicon substrate, starting with SnO powders (99%, Alfa Aesar). The experimental details have been described elsewhere (Wang et al., 2007; Zhou et al., 2006). The morphologies and chemical composition of SnO₂ nanobelts and nanowires were examined by scanning electron microscopy (SEM) (Instrument model: Leo/Zesis 1540XB FIB/SEM) and energy dispersive X-ray spectroscopy. Fig. 1 a and 1cshow the dimensions and morphologies of the as-made SnO₂ nanobelts and nanowires. The nanobelts are several tens of nanometers thick, several micrometers long and a few hundred nanometers up to 1 μm wide. The nanowires were 50-60 nm in diameter and several microns long.

The GaN nanowires were synthesized by passing ammonia through a mixture of Ga and Ga₂O₃ at high temperature in the presence of Au nanoparticles as the catalysts. The experimental details were reported previously (Zhou et al., 2005). The nanowires were thoroughly characterized by SEM. The SEM images (Fig. 2) reveal that the GaN nanowires have wire-like structure with zigzag periodic units. Most of the wires have a length of tens to hundreds of microns and their thickness is several tens of nanometers.

Figure 1.

SEM images of SnO₂ nanobelts before compression (a) and after decompression (b) as well as SnO₂ nanowires before compression (c) and after decompression (d) with scales shown in each panel. (From Dong & Song, 2009)

Figure 2.

SEM images of GaN nanowires before compression (a) and after decompression (b) with scales labeled in each panel. (From Dong & Song, 2010).

2.2. High pressure techniques

Diamond anvil cell (DAC) is a fundamental apparatus to achieve static high pressures. Recent rapid advances in the DAC technology have allowed the generation of extreme conditions in a broad P-T range with great controllability and accuracy (Hemley & Mao, 2002). The transparency of diamonds over a wide wavelength range allows the use of various optical techniques to examine the high-pressure behaviors of materials in situ. Over the past a few decades, in particular, new emerging analytical probes including optical spectroscopy, synchrotron and neutron sources have enabled structural characterization of materials with unprecedented spatial, temporal and spectral resolutions (Hemley & Mao, 2002).

Fig. 3 shows a typical DAC apparatus where two brilliant cut diamonds are used as anvils to exert static pressure up to several million atmospheres (or several hundred GPa) with only moderate force. Such extreme pressures can be accurately determined by monitoring ruby fluorescence lines using the following relationship (Mao et al., 1978):

$$P=\frac{1904}{B}\left[{\left(1+\frac{\Delta \lambda }{694.24}\right)}^B-1\right]$$ E1

where P is the pressure in GPa, Δλ is the ruby R₁ line shift in nm, and parameter B is 7.665 for quasi-hydrostatic conditions and is 5 for non-hydrostatic conditions.

Figure 3.

Photo and schematics of a symmetric diamond anvil cell.

In the experiment for SnO₂ nanomaterials, a symmetric DAC with a pair of type I diamonds and a 400-micron culet was used. A hole with a diameter of 150 microns was drilled on a stainless steel gasket and used as the sample chamber. A few ruby chips were loaded with the sample as the pressure calibrant. The samples were loaded without pressure transmitting medium (PTM) for Raman measurements, whereas silicon oil was used as the PTM for X-ray diffraction measurements. GaN nanowires were compressed up to 65 GPa using a symmetric diamond anvil cell (DAC) which consists of a pair of type I diamonds with 200 μm culets and cubic boron nitride seats. The sample and a small ruby ball were enclosed in an 80-μm-diameter hole in a tungsten gasket. Silicone oil was used as the PTM.

2.3. In situ Raman spectroscopy

Vibrational spectroscopies compatible with DAC apparatus, including Raman and Fourier Transform Infrared (FTIR) microspectroscopy have been widely used to characterize high-pressure structures of materials in situ. In particular, Raman spectroscopy is a sensitive structural characterization probe providing rich information about molecular geometries, bonding properties, phase identities and transformations, lattice dynamics, and crystallinity of materials, etc. Raman measurements were performed on SnO₂ nanobelts using a customized Raman micro-spectroscopy system. A 488 nm line from an Innova Ar⁺ laser (Coherent Inc.) was used as the excitation source and was focused to less than 5 μm on the sample by an Olympus microscope. The Rayleigh line was removed using a pair of notch filters. The scattered light was dispersed using an imaging spectrograph equipped with a 1800 lines/mm grating, achieving a resolution of 0.1 cm^-1. The scattered light was then recorded using an ultrasensitive liquid-nitrogen-cooled, back-illuminated charge-coupled device (CCD) detector from Acton. The system was calibrated using neon lines with an uncertainty of ± 1 cm^-1.

2.4. Synchrotron X-ray micro-diffraction

Synchrotron light source in the high-energy hard X-ray region is an indispensable probe for DAC based materials characterizations. Such hard X-ray available in almost all synchrotron facilities worldwide enables in situ diffraction measurements on micron-sized materials under extreme P-T conditions with unparalleled accuracy for the elucidation of crystalline structures. For SnO₂ nanomaterials, the angle dispersive X-ray diffraction measurements were carried out at the X17C beamline at the National Synchrotron Light Source (NSLS) at Brookhaven National Laboratory (BNL). A high-energy, fixed-exit monochromator with Sagittelly-bent double Si crystal Laue mode was used to optimize the high-energy synchrotron X-ray from 20 keV to 40 keV, with an incident X-ray wavelength of 0.4066 Å. A pair of Kirkpatrick-Baez (KB) mirrors consisting of Si crystals coated with Pt and a focal length of 100 mm was used to focus the white X-ray beam at a glancing angle of approximately 1 mrad. This focused a 180 μm×180 μm incident beam to a 25 μm (horizontal) × 25 μm (vertical) beam on the sample. A MAR CCD X-ray detector was used to collect the 2D Debye-Scherrer patterns. The gonioemeter geometry and other diffraction parameters were calibrated using CeO₂ standard diffraction. Each diffraction pattern was obtained during an average exposure time of 5 to 10 minutes. The two-dimensional Debye-Scherrer patterns were converted to one-dimensional diffraction patterns using Fit2D software and Rietveld refinements were performed using GSAS package.

The angle-dispersive X-ray diffraction measurements on GaN nanowires were performed at the undulator sector 16-ID-B, High-Pressure Collaborative Access Team (HPCAT), Advanced Photon Source (APS) of Argonne National Laboratory (ANL) using an incident wavelength of 0.3680 Å and beam size of 15 μm×10 μm. The procedures of data analysis for GaN nanowires are similar to those for SnO₂ nanomaterials.

3. SnO₂ nanobelts and nanowires under high pressures

3.1. Properties, structures, and applications of SnO₂

As an n-type semiconductor with a large band gap (E_g = 3.6 eV at 300 K), SnO₂ has been studied extensively for applications in transparent conducting electrodes, lithium ion batteries and gas sensors (Watson et al., 1993). Specifically, because SnO₂ exhibits high sensitivity for detecting CO and NO_x gases, vigorous studies on SnO₂-based gas sensors have been undertaken (Watson et al., 1993). In these studies, SnO₂ in orthorhombic structures was found to exhibit better sensitivity for specific gases than the tetragonal rutile-type structure (Arbiol et al., 2008; Sangaletti et al., 1997). However, natural SnO₂, known as cassiterite, always exists as the rutile-type structure and it is generally difficult to obtain the orthorhombic phase directly from minerals (Chen et al., 2006). Therefore, finding new methods for producing SnO₂ with orthorhombic structures, such as by pressure tuning (Haines & Leger, 1997), is of particular interest in the sensor industry. In addition, studies suggest that nanostructured SnO₂ might exhibit significantly enhanced performance for certain gas-sensing applications than thick films or bulk materials (Wang et al., 2008). Therefore, many nanostructured SnO₂ of different morphologies, including nanoparticles, nanowires, nanorods and nanobelts, have been synthesized (Calestani et al., 2005; Calestani et al., 2005; Zhou et al., 2006) and their optical and electrochemical properties evaluated. Spectroscopic studies such as Raman measurements (Sun et al., 2003) and photoluminescence (Zhou et al., 2006) show that SnO₂ nanobelts exhibit unique optical properties that are different than those from bulk materials. These recent studies motivated us to undertake the high-pressure investigations on the behaviors of one-dimensional nanostructured SnO₂, resulting in the observation of novel, and unexpected nano-effects.

3.2. Raman spectrum of SnO₂ nanobelts

Raman measurements were performed on SnO₂ nanobelts with selected spectra depicted in Fig. 4. SnO₂ nanobelts at ambient pressure had a regular rutile-type structure (space group P4₂/mnm or $$D_{4h}^{14}$$ , Z=2) and, therefore, the irreducible representation predicted the Raman active modes to be A_1g, B_1g, B_2g and E_g, with three bands observed at 473 cm^-1 (E_g), 631 cm^-1 (A_1g) and 773 cm^-1 (B_2g), consistent with previous Raman measurements on nanostructured SnO₂ (Sun et al., 2003). Additional weak Raman bands at 500 cm^-1 and 692 cm^-1 can be assigned as A_2u (TO) and A_2u (LO) modes, both of which are IR active, whereas the band at 544 cm^-1 was a Raman forbidden B_1u mode. These abnormal Raman bands are characteristic of SnO₂ nanobelts and are not observed in the Raman spectrum of bulk SnO₂ (Sun et al., 2003). Four additional Raman active modes were observed below 300 cm^-1 (not shown here), which can be attributed to the impurities of the substoichiometric Sn₂O₃/Sn₃O₄ phases in the synthetic process (Wang et al., 2007). Interestingly, no strong Raman active modes were observed for the SnO₂ nanowires, which is in contrast to a Raman study by Zhou et al. (Zhou et al., 2006) on single crystalline SnO₂ wires. It is well known that the optical properties of nanomaterials, especially their Raman features, are very sensitive to a number of factors including size, morphology, synthetic route, purity, as well as short-range structures and environment (Gouadec & Colomban, 2007). As a result, different Raman measurements on the same materials (e.g., various metal oxides) but with different nano-parameters have been reported and analyzed extensively (Arora et al., 2007; Gouadec & Colomban, 2007). Upon compression, all SnO₂ nanobelts Raman modes exhibited a blue shift with decreasing intensities along with profile broadening. At the highest pressure, only the A_1g mode was observed to have a significantly broadened profile (middle spectrum of Fig. 4). Upon decompression, the Raman profiles changed very gradually as the pressure decreased. As the SnO₂ nanobelts returned to ambient pressure (lower spectrum of Fig. 4), the E_g and B_2g modes were recovered but the bands remained broadened. These observations indicate the partial reversibility of the SnO₂ nanobelts optical responses to compression and decompression, which is coincident with the pressure-induced morphology modifications. The SEM images obtained before compression (Fig. 1a), and after decompression (Fig. 1b), suggest that the belts are crushed to shorter sections that are thicker than the original belts before compression. In contrast, the SnO₂ nanowires exhibited more dramatic changes in morphology as a result of compression, i.e., the wire shapes were no longer recognizable (Fig. 1d).

Figure 4.

Raman spectra of SnO₂ nanobelts in the spectral region of 300 to 1000 cm^-1 collected at ambient pressure (top), at 39.5 GPa (middle) and upon decompression (bottom). The assignments of the observed Raman modes are labeled above the ambient-pressure spectrum. (From Dong and Song, 2009)

3.3. X-ray diffraction of SnO₂ nanobelts

In situ high-pressure angle dispersive X-ray diffraction measurements were performed on SnO₂ nanobelts on compression up to 38 GPa followed by decompression. Representative diffraction patterns are depicted in Fig. 5. Fig. 6 shows the quantitative analysis of the nanobelts diffraction patterns at 14.8, 19.2 and 31.8 GPa and upon complete decompression (i.e., recovered). Starting at ambient pressure, the diffraction pattern of SnO₂ nanobelts indicates an excellent crystalline phase that adopts a rutile structure (P4₂/mnm) that is the same as that for the bulk material (Haines & Leger, 1997), with cell parameters of a=4.7218Å and c=3.1802 Å. All the diffraction patterns can be indexed with a single rutile-type phase until compressed to near 15 GPa (Fig. 6a). The broadening of the (101), (200) and (211) reflections of the rutile phase suggests that an orthorhombic phase has formed (CaCl₂-type structure with space group Pnnm). Rietveld refinement suggests that there was only a slightly modified cell parameter from the rutile structure: a=4.6543 Å, b=4.5744 Å and c=3.1483 Å. This pressure-induced phase transformation was similar to that observed with bulk SnO₂ (Haines & Leger, 1997), but the transformation pressure was higher. SnO₂ nanobelts existed in this single phase up to 19.2 GPa beyond which new phase transformations were observed (Fig. 6b). The new reflections at 2θ of 6.606º and 8.129º were characteristic of a new orthorhombic α-PbO₂ phase (Pbcn) at (110) and a cubic fluorite phase (Pa $$\overline{3}$$ ) at (111) directions, respectively. From 19.2 GPa to the highest pressure in the present study, SnO₂ existed as a mixture of these three phases (Fig. 6c). This observation is in strong contrast to the phases observed for bulk SnO₂ materials, in which both the CaCl₂-type and α-PbO₂ phases are reported to exist between 12 and 21 GPa (Haines & Leger, 1997). Upon decompression, the three-phase mixture was found to persist down to 7.5 GPa, when the cubic fluorite phase disappeared, with a subsequent back transformation to the rutile structure. Upon complete decompression, the SnO₂ nanobelts were composed of mixtures of rutile and α-PbO₂ phases (Fig. 6d).

Figure 5.

X-ray diffraction patterns of SnO₂ nanobelts at selected pressures on compression (a) and decompression (b). Pressures in GPa are labeled along each pattern. The Miller indices are shown for the tetragonal structure at 1.9 GPa, the orthorhombic structure at 14.8 GPa and the cubic structure at 36.2 GPa on compression, and at 36.2 GPa and 0.2 GPa on decompression, respectively.

Figure 6.

Rietveld refinement of X-ray diffraction patterns of SnO₂ nanobelts at 14.8 GPa (a), 19.2 GPa (b), 31.8 GPa (c) and from recovered sample (d). The inset in (c) shows the original 2D Debye-Scherrer patterns with one quadrant. The red cross is experimental X-ray intensity whereas the green solid line is the calculated diffraction pattern based on refinement with the black curve at the bottom showing the difference between the calculated and observed intensities. The vertical bars with different colors indicate the characteristic reflections of different phases labeled in the front.

3.4. X-ray diffraction of SnO₂ nanowires

The X-ray diffraction patterns SnO₂ nanowires at selected pressures upon both compression and decompression are depicted in Fig. 7. The Rietveld refinement analysis for patterns at pressures of 14.8, 20.0 and 34.4 GPa and upon complete decompression is shown in Fig. 8. Upon compression and decompression, as can be seen, SnO₂ nanowires exhibit unexpected pressure responses that are different than those for nanobelts or bulk materials. Starting with the same rutile structure (e.g., Fig. 8a), transformation to a CaCl₂-type orthorhombic structure (Fig. 8b) was observed only when it was compressed to 17 GPa, which is a higher transition pressure than that for the nanobelts and much higher (ΔP> 5GPa) than that for the bulk materials. When further compressed to 25 GPa, a fluorite-type phase was found to contribute to the overall diffraction pattern and coexisted with the CaCl₂ type phase all the way to the highest pressure. A striking observation that the α-PbO₂ phase was missing in the entire compression region was noted as shown in the Rietveld refinement analysis of the diffraction pattern, which unambiguously suggests that the SnO₂ nanowires are composed of only CaCl₂-type and fluorite-type phases at 34.4 GPa (Fig. 8c). Upon decompression, the CaCl₂-type phase transformed back to the rutile-type phase at 16 GPa, which is much earlier than that observed for either nanobelts or bulk materials. Further decompression results in more surprising transformations - the fluorite-type phase persists all the way to near-ambient pressure, whereas the α-PbO₂ phase that was missing during compression was observed when it was decompressed to 10.6 GPa and was recovered at ambient pressure (Fig. 8d).

Figure 7.

X-ray diffraction patterns of SnO₂ nanowires at selected pressures on compression (a) and decompression (b). The legends follow Fig. 5.

Figure 8.

Rietveld refinement of X-ray diffraction patterns of SnO₂ nanowires at 14.8 GPa (a), 20.0 GPa (b), 34.4 GPa (c) and from recovered sample (d). The legends follow Fig. 6.

3.5. Equation of states

Rietveld refinement performed on all diffraction patterns indicated that the rutile and CaCl₂-type phases were the dominant phases both for nanobelts and nanowires, whereas the α-PbO₂ and flurorite phases contribute to the mixed phases only to a certain extent. Therefore, we fit the third-order Birch equation of state (EOS) based on only the dominant phases of nanobelts and nanowires (Fig. 9) to estimate the compressibility using (Birch, 1978):

$$P=\frac{3}{2}{B}_0\left[{\left(\raisebox{1ex}{$V$}\!\left/ \!\raisebox{-1ex}{$V_0$}\right.\right)}^{-7/3}-{\left(\raisebox{1ex}{$V$}\!\left/ \!\raisebox{-1ex}{$V_0$}\right.\right)}^{-5/3}\right]\times \left\{1+\frac{3}{4}\left(B_0^{\text{'}}-4\right)\left[{\left(\raisebox{1ex}{$V$}\!\left/ \!\raisebox{-1ex}{$V_0$}\right.\right)}^{-2/3}-1\right]\right\}$$ E2

where V₀ is the original cell volume at ambient pressure, and V is the cell volume at pressure P. $$B_0$$ and $$B_0^{\text{'}}$$ are the bulk modulus and its first pressure derivative, respectively, at ambient pressure.

The bulk modulus and its first derivative were found to be B₀=169.3 GPa and B′=8.4 for the nanobelts whereas those for the nanowires were B₀=225.3 GPa and B′=8.1, respectively. We note that the compressibility of the nanobelts was significantly higher than that for the bulk material (B₀= 204 GPa, B′= 8.0) (Haines & Leger, 1997), whereas the nanowires were less compressible.

Figure 9.

Pressure-volume relations for SnO₂ nanowires (open circles) and nanobelts (solid circles). The solid lines are fittings using 3^rd order Birch equation of state (see text).

The size- and morphology-induced alteration of SnO₂ compressibility characterized by bulk moduli can be understood in parallel with other nanomaterials. CeO₂ nanoparticles exhibit a prominent enhancement of the bulk modulus compared with that for bulk materials (Wang et al., 2001; Wang et al., 2004), whereas no obvious difference in compressibility was observed for ZnS nanocrystals (Jiang et al., 1999). In contrast, the compressibility of PbS and γ-Al₂O₃ (Chen et al., 2002) was found to increase with decreasing nanoparticle size. Furthermore, strongly contrasting compressibility was observed for TiO₂ nanoparticles, i.e, the bulk modulus of the rice-shaped particles was reduced whereas that of the rod-shaped particles was enhanced by more than 50% relative to that of the bulk materials (Park et al., 2008). Therefore, multiple factors determine the mechanical properties of nanomaterials. In this case, by carefully examining SnO₂ nanobelts at ambient pressure using SEM and Raman imaging (Wang et al., 2007), other tin oxides (SnO_x) attached to the nanobelt surface could contribute to the defect in the SnO₂ crystal lattice and may therefore decrease the stiffness. Compared with nanobelts, SnO₂ nanowires carry much fewer or no defects and are more strictly one-dimensional in morphology, which may correlate with their general size-dependent compressibility (Wang et al., 2001). These arguments are corroborated by a previous comparative study of nanobelts and nanowires (Calestani et al., 2005).

3.6. Discussion

Fig. 10 summarizes the strongly contrasting pressure-induced phase transformations of SnO₂ nanobelts and nanowires compared with bulk materials. The differences in the pressure-induced phase transitions between nanostructured and bulk SnO₂ materials have been observed for other morphologies, primarily nanoparticles. However, contrasting results were reported by different groups - He et al. found that transition pressure increases with decreasing SnO₂ nanocrystal size (He et al., 2005), whereas Jiang et al. observed no obvious size-dependent transition pressure differences (Jiang et al., 2001). For other nanostructured materials such as oxides, sulfides or elements, transition pressure shifts have been found to go in both directions. The majority of these nanomaterials (e.g., CdSe (Tolbert & Alivisatos, 1994; Tolbert & Alivisatos, 1995), ZnS (Jiang et al., 1999) and PbS (Jiang & Gerward, 2000)) exhibit higher transition pressures than bulk materials do and their transition pressures also increase with decreasing nanocrystal size (Tolbert & Alivisatos, 1995). However, other nanocrystal oxides, such as γ-Fe₂O₃ (Jiang et al., 1998) and CeO₂ (Rekhi et al., 2001), have reduced transition pressures. The general understanding of the 'size' effect in pressure-induced transformations is used to examine a thermodynamic function, i.e., the Gibbs free energy change (ΔG) with the major contributing factors: the ratio of the volume collapse (PΔV), the surface energy differences (AΔγ) and the internal energy differences (ΔU) (Jiang, 2004; Wang et al., 2005). It is believed that enhanced transition pressures in nanomaterials indicate that surface energy differences are playing a dominant role, whereas reduced transition pressures might be associated with a compression process overwhelmed by volume collapse. The 'size' effect also seems to extend to one-dimensional nanomaterials such as ZnS nanobelts (Wang et al., 2005). Indeed, it was found that the reduced ZnS nanobelt thickness resulted in a higher transition pressure (Wang et al., 2005). Therefore, these principles can be adopted to explain some of the pressure-induced transformations observed in this study. The higher transition pressures observed both in nanobelts and nanowires indicate that there is a prominent surface energy effect on nanostructured SnO₂. In particular, the onset pressures involving the rutile-to-CaCl₂ type transition are approximately 11.8, 15.0 and 17.0 GPa for bulk materials, nanobelts and nanowires, respectively. Because there is no significant change in the unit cell volume in this transition, the contribution of the first factor, i.e., the ratio of the volume collapse to the overall ΔG, is negligible. Considering that the internal energy differences (ΔU) are typically small or negative (Jiang, 2004), the surface energy differences mainly are therefore believed to be responsible for the enhanced transition pressures for nanostructured SnO₂. Gauging from the observed transition pressures, the surface energy differences in the nanowires were estimated to be 62.5% higher than in the nanobelts. Qualitatively, the significantly enhanced nanowire transition pressures can be interpreted by further 'reduced size' from nanobelts with reduced width and thickness. Indeed, the band widths of the nanowire reflections are slightly higher in the nanobelts in general, consistent with the size-induced broadening observed for other nanomaterials (Wang et al., 2004).

In addition to size effects, morphology has also proven to be an important factor for regulating nanomaterial structure and stability, either by early or delayed phase transitions (Wang et al., 2005). However, the drastically contrasting phase stability regions observed for different morphologies of SnO₂, especially for the completely missing α-PbO₂ phase, are unprecedented.

Figure 10.

Summary of pressure-induced phase transformations for SnO₂ nanobelts and nanowires upon compression (plain vertical bars) and decompression (hatched vertical bars) compared with those for bulk SnO₂ material. The different colors label different phases: rutile (red), CaCl₂ type (green), flurorite type (yellow) and α-PbO₂ type (blue).

On the basis of the above principles, one may speculate that the α-PbO₂ transition pressure might be significantly elevated (i.e., >> 38 GPa), which requires further experimental and theoretical justification. However, such huge pressure increases may well induce other new SnO₂ phases to form (Shieh et al., 2006). Therefore, the α-PbO₂ phase is likely a metastable phase that cannot simply be interpreted by thermodynamic principles alone. Furthermore, the observed prominent hysteresis, which was characterized by significantly different forward and backward transition pressures, is likely a consequence of different transformation barriers. While the stabilities of difference phases are only determined by thermodynamic functions, the actual transformation pressure may be predominantly governed by kinetics, which scales with the width of the hysteresis. It would therefore be interesting to investigate hysteresis and kinetics as a function of temperature. Indeed, the hysteresis for the nanocrystal CdSe transitions was found to narrow as temperature increased (Chen & Herhold, 1997). The combination of pressure, size, morphology, thermodynamics and kinetics has led to the formation of a multi-dimensional structure-property domain with extremely broad tunabilities. Our findings indicate that certain structures and/or phases can be switched 'on' or 'off' at selected pressure regions with selected morphologies via selected paths. Applying pressure to nanomaterials with different morphologies, therefore, has profound implications for producing controlled structures with desirable properties, such as those for gas sensors whose sensitivity has a preferential correlation to the orthorhombic α-PbO₂-type structure of SnO₂ (Arbiol et al., 2008). However, detailed transformation mechanisms, especially the origins of the surprising reversibility and metastability require further theoretical investigation.

4. GaN nanowires under high pressures

4.1. Properties, structures, and applications of GaN

Gallium nitride (GaN) is a wide band gap semiconductor (3.5 eV) of great technological importance due to its potential applications in high brightness blue/green light emitting diodes (LEDs) and room temperature blue laser (Nakamura et al., 1997). These applications usually require GaN in the wurtzite and zinc blende phases depending on the synthetic techniques (Lei et al., 1992). In addition, GaN is also characterized to have high hardness, low compressibility, high ionicity and high thermal conductivity. These properties make GaN a promising candidate for optoelectronic devices operating under extreme conditions such as at high pressure (Liu et al., 1999). At ambient condition, GaN crystallizes in a wurtzite (B4) structure with a space group of P6₃mc (Schulz & Thiemann, 1977). It is well established that GaN transforms to the rocksalt (B1) structure (space group Fmm) at high pressure as suggested by recent theoretical and experimental reports ( Cui et al., 2002 ; Jorgensen et al., 2003; Munoz & Kunc, 1991; Pandey et al., 1994; Perlin et al., 1992; Saib & Bouarissa, 2007; Ueno et al., 1994; Xia et al., 1993; Xiao et al., 2008). For instance, Perlin et al. identified such phase transformation at 47 GPa using Raman scattering and X-ray absorption spectroscopy (Perlin et al., 1992). By synchrotron energy-dispersive X-ray diffraction, Xia et al. reported the first observation of rocksalt phase at an onset pressure of 37 GPa (Xia et al., 1993). Additional later studies confirmed that GaN with different forms all underwent the wurtzite to rocksalt transformation on compression, however, with large discrepancies in transition pressures (e.g., 42-54 GPa for experimental results (Halsall et al., 2004; Perlin et al., 1992; Ueno et al., 1994) and 37-55 GPa for some theoretical studies (Abu-Jafar et al., 2000; Cai & Chen, 2007; Mujica et al., 2003; Munoz & Kunc, 1991; Pandey et al., 1996; Serrano et al., 2000)). GaN can also be synthesized in the nanocrystalline form, either as nanowires or as quantum dots (Jorgensen et al., 2003; Liu et al., 2003; Zhou et al., 2005). Compared to bulk GaN, nanostructured GaN exhibits attractive properties and thus enhanced performance for applications such as optoelectronic devices because of its excellent tunability as a direct band gap semiconductor (Qian et al., 2005). Therefore, studies on structural tuning of nano-GaN such as by the application of pressure in comparison with the bulk materials are of fundamental interests. To date, very few experiments were performed on nanostructured GaN at high pressure. Only Jorgensen et al. has investigated the phase transformations and compressibility of GaN nanocrystals in comparison with the bulk GaN using X-ray diffraction (Jorgensen et al., 2003). Here we discuss the in situ high-pressure X-ray diffraction measurements on one-dimensional nanostructured GaN in the form of zigzagged nanowires where interesting, abnormal pressure behaviors were observed.

4.2. GaN nanowires upon compression

Selected X-ray diffraction patterns of GaN nanowires collected upon compression to 65 GPa are shown in Fig. 11a. The X-ray diffraction pattern collected at near ambient pressure (i.e., 0.5 GPa) can be indexed with a hexagonal wurtzite structure (P6₃mc) with cell parameters a=b=3.1916 Å and c=5.1733 Å, consistent with the previous X-ray diffraction measurement on GaN nanowires (Zhou et al., 2005). In contrast to the diffraction pattern of nanocrystalline GaN which is characterized with significantly broadened reflection profiles (Jorgensen et al., 2003), however, the narrow and sharp reflections for GaN nanowires here suggest an excellent crystalline phase that resembles bulk GaN. Upon compression, the wurtzite phase was found to persist to 65 GPa indicated by the consistent indexing of the first six reflections associated with this phase. At 55 GPa, a new reflection appeared at 10.4519º, which can be indexed as (2 0 0) for the rocksalt phase, suggesting the onset of phase transformation. The phase transformation can be further evidenced by the depletion of the strong (1 0 0) reflection at the lowest 2θ angle and the dominant reflection (1 0 1) of wurtzite phase which coincides with reflection (1 1 1) of the rocksalt phase. We note that the phase transition pressure of 55 GPa is higher than that in most previous studies for bulk GaN (Ueno et al., 1994; Xia et al., 1993) but lower compared to that for nanocrystalline GaN which was found to be around 60 GPa (Jorgensen et al., 2003). Size and morphology dependent enhancement of transition pressures have been observed in other nanostructured materials (Dong & Song, 2009). Now it is the general understanding that the surface energy that plays an important role in nanostructures contributes to the enhanced transition pressures. The GaN nanowires in the current study had an initial wire-like morphology and converted to smaller nanoparticles on compression with an average size of 50-200 nm as shown in Fig. 2b, resulting an increase of surface area. Comparing with the particle size of nanocrystalline GaN (i.e., 2-8 nm ) studied previously (Jorgensen et al., 2003), it can be inferred that surface energy of GaN increased from bulk to nanowires and to nanoparticles with decreasing sizes, giving the corresponding different transition pressures.

Figure 11.

Angle-dispersive X-ray diffraction patterns (λ=0.3680 Å ) of GaN nanowires at selected pressures upon compression (a) and decompression (b). The pressures in GPa are labelled along each pattern.

Another significant difference between this study and previous studies on bulk GaN is that the wurtzite-to-rocksalt transformation for GaN nanowires is far from complete even at 65 GPa. Quantitative Rietveld analysis of GaN nanowires diffraction pattern at 65 GPa shown in Fig. 12a indicates that wurtzite and rocksalt phases co-exists at this pressure with respective abundance of 88% and 12%. In contrast, an abundance of near 100% for the rocksalt phase was reported for bulk GaN at a pressure less than 60 GPa in most of the previous studies. For example, Perlin et al. found that the wurtzite phase disappears at 54 GPa (Perlin et al., 1992), while Halsall et al. reported such a pressure to be 49 GPa (Halsall et al., 2004). The highest pressure for complete transformation to rocksalt phase for bulk GaN was reported to be 58.8 GPa (Cui et al., 2002) and 60.6 GPa (Ueno et al., 1994), respectively. Our observation is consistent with that for nanocrystalline GaN where the wurtzite to rocksalt phase transition was found incomplete even at 63.5 GPa (Jorgensen et al., 2003). Similar situations were also observed in other one-dimensional nanomaterials, such as in multi-wall boron nitride nanotubes (MW-BNNT) (Dong & Song, 2010). The required pressures for the hexagonal-to-wurtzite structural transformation to be completed are 23 GPa for bulk BN and > 35 GPa for MW-BNNTs, respectively, which can be understood from the topological and mechanical aspects of BNNT nanostructures (Dong & Song, 2010). In addition to the intrinsic properties of nanostructures, non-hydrostaticity may also contribute to the “sluggish” phase transformations. We realize that silicone oil used as the pressure transmitting media in this study, which forms a glass phase even at very low pressures (Ragan et al., 1996), is far from ideal. The non-hydrostaticity in manifested by the significant broadening of all reflections especially at high pressures as shown in the Debye-Scherrer 2D

Figure 12.

Rietveld refinement of x-ray diffraction pattern at 65 GPa (a) and 20.9 GPa (b). The redcross is experimental X-ray intensity and the green line is the calculated diffraction patterns based on refinement with the blue curve at the bottom showing the difference between the calculated and observed intensities. The vertical bars denote the indexed reflections for each phase with the space groups labeled beside. The characteristic reflection for the cubic rocksalt phase is labeled as c-(2 0 0). (From Dong and Song, 2010)

diffraction patterns in Fig 13. For instance, the full width at half maximum (FWHM) of reflection (1 0 1) increased dramatically from 0.09º at 0.5 GPa to 0.25º upon compression 65 GPa. Furthermore, severe distortions in the Debye-Scherrer 2D diffraction patterns for reflections (1 0 0), (0 0 2) and (1 0 1) were observed in the high-pressure region (Fig. 13b), indicating pressure-induced enhancement of lattice strain. All these factors may interactively affect the phase stabilities and thus the transformation pressures.

Figure 13.

Debye-Scherrer 2D X-ray diffraction patterns of GaN nanowires at ambient pressure (a), 65 GPa (b), 20.9 GPa upon decompression (c) and ambient pressure upon complete decompression (d).

4.3. GaN nanowires upon decompression

X-ray diffraction patterns of GaN nanowires at selected pressures upon decompression from 65 GPa to the ambient pressure are shown in Fig. 11b. An interesting yet unusual phase transformation was found during decompression. Upon releasing pressure from 65 GPa, the highest pressure achieved in the current study, the abundance of rocksalt phase was found to increase significantly. For instance, Rietveld refinement (Fig. 12b) suggests that the fraction of rocksalt phase has increased upon releasing the pressure from 12 % at 65.0 GPa to 29% even at 20.9 GPa. Bellow 20.9 GPa, the rocksalt phase gradually transforms back to the wurtzite phase, but was retained even at very low pressures (e.g., <16.4 GPa) as indicated by the noticeable intensity of the (2 0 0) reflection as the dominant reflection of the rocksalt phase. The recovered phase at ambient pressure can be identified as a single wurtzite phase, but the significantly broadened diffraction pattern (Fig. 13d) suggests pressure-induced reduction of the grain size, consistent with the SEM image of the recovered GaN nanomaterial (Fig. 2b). All the previous studies found that B4-B1 phase transformation is reversible for both bulk GaN and nanocrystalline GaN (Cui et al., 2002; Halsall et al., 2004; Perlin et al., 1992) indicating that the wurtzite phase is thermodynamically stable in the low pressure region. However, the strongly contrasting, unprecedented decompression behavior of GaN nanowires exhibiting a large hysteresis suggests that the rocksalt phase is a metastable phase. Prominent hysteresis, which was characterized by significantly different forward and backward transition pressures, is likely due to different kinetic barriers that impede the sharp transitions often involving a metastable phase. Here, we note that this phase of GaN behaves similarly as one of the high-pressure phases (i.e., α-PbO₂ phase) for SnO₂ nanowires (Dong & Song, 2009).

4.4. Equation of state

The unit cell parameters of GaN nanowires at different pressures are plotted in Fig. 14a with the unit cell volume against the pressure for the wurtzite phase plotted in Fig. 14b in comparison with previous studies. As can be seen, the unit cell parameters, the a/c ratio, and the corresponding P-V curve of GaN nanowires exhibit a noticeable discontinuity in the pressure region of 11-20 GPa. In the low pressure region (i.e., < 11 GPa), the equation of state (EOS) for GaN nanowires is almost identical for that of bulk GaN material. Above 11 GPa, obvious change in the P-V relation of GaN nanowires suggest a smaller compressibility or larger bulk modulus. Comparing with previously established EOS for bulk GaN (B₀=187-237 GPa) (Jorgensen et al., 2003; Ueno et al., 1994), and for nanocrystalline GaN (B₀=319 GPa)(Jorgensen et al., 2003), the bulk modulus for GaN nanowires in the pressure region above 20 GPa is estimated to be in between, consistent with that an increase of the surface area may enhance the bulk modulus. This observation can be further understood from the fact the GaN nanowires break into nanoparticles when the external stress surpasses the yield strength at around 20 GPa, resulting in the size-induced enhancement of the bulk modulus. The correlation between the reduction of the particle size and the increasing bulk modulus has been found in other nanomaterials, such as AlN nanocrystals (Wang et al., 2004). A similar discontinuity was also found in the P-V curve for nanocrystalline CeO₂.(Wang et al., 2004) However, a decrease in the bulk modulus of nanocystalline CeO₂ was found above ~20 GPa, a pressure that was believed to signify the onset of a size-induced weakening of the elastic stiffness of nanocrystalline CeO₂. Therefore, the discontinuity in the EOS for both nanocystalline CeO₂ and GaN nanowires can be understood from the nano-size effect while the different shapes of the EOS curves may be associated with opposite pressure-induced changes in particle size. It is probable, too, that the P-V behavior observed in the region of 11-20 GPa is associated with the combined effect of the original nanowires and the yield-generated nanoparticles. The high yield strength observed in the GaN nanowires thus furthers the understanding of the morphology-tuned improvement in yield strength and hardness in hard materials applications. Moreover, the choice of silicone oil as the pressure-transmitting medium may also contribute to the discontinuity observed in the P-V curve. An anomalous behavior inherent in silicone oil at ~12 GPa, which is believed to be the result of phase transition (Chervin et al., 1995; Klotz et al., 2009), seems to have further complicated the abnormal behavior of GaN nanowires in the pressure region of 11-20 GPa. It would therefore be interesting to use better pressure transmitting media, such as helium, to study compressibility of GaN nanowires in a broader hydrostatic pressure region.

Figure 14.

Unit cell parameters of GaN nanowires as a function of pressure (a) and unit cell volume as a function of pressure for GaN nanowires (open squares) in comparison with that for bulk and nanocrystal GaN (b). Solid lines are fitted EOS curves for bulk GaN using bulk moduli of 187 GPa (bulk₁) and 237 GPa (bulk₂) from (Jorgensen et al., 2003) and (Ueno et al, 1994) respectively. The dashed line is fitted EOS curve for nanocrystalline GaN using bulk modulus of 319 GPa from (Jorgensen et al., 2003).

5. Summary

In summary, two one-dimensional inorganic materials with various morphologies and sizes were investigated under high pressures. In particular, pressure-induced structural evolutions in nanostructured SnO₂ in the form of nanobelts and nanowires were studied in diamond anvil cells using Raman spectroscopy, angle dispersive X-ray diffraction and SEM. We found that nanostructured SnO₂ exhibits drastically disproportionate high-pressure behaviors compared with bulk materials, which suggests that nanostructured SnO₂ has significantly different optical, chemical and mechanical properties. These morphology-induced differences for some of the phase transformations can be explained by surface energy differences as the dominant thermodynamics factor, while other phases are primarily mediated by kinetics. These principles may serve as a general guideline for producing novel functional materials with desired stability and/or metastability that may yield promising industrial applications, particularly for semiconductor and chemical sensor uses.

For GaN nanowires, we have investigated high-pressure structures and properties of the first one-dimensional nanostructured GaN using angle-dispersive X-ray diffraction in situ. A wurtzite to rocksalt phase transformation is observed to start at 55 GPa and but is far from complete even at pressures up to 65 GPa. Upon decompression, the abundance of the rocksalt phase was found to increase and then decrease until ambient pressure, at which only the wurtzite phase was recovered, indicating a reversible transformation but with large hysteresis. A discontinuity in the EOS for GaN nanowires is found in the pressure region of 11-20 GPa. These abnormal pressure behaviors of GaN nanowires were compared with previous high pressure studies on bulk and nanocrytalline GaN and can be understood from the nano-size and morphology dependent thermodynamic and kinetic properties of GaN nanowires with the complication of the nonhydrostatic conditions.

6. Future perspectives

Despite that interesting high-pressure behaviors of the two inorganic nanowires were observed and analyzed, more in-depth understanding of these unusual behaviors at quantitative level is still needed. In strong contrast to the extensive high-pressure studies on zero-dimensional nanoparticles, only very few one-dimensional inorganic nanomaterials were investigated at high pressures up-to-date as outlined in Table 1. Therefore, further systematic studies on different inorganic 1D nanostructures with different dimensions under different compression conditions are needed. Additional experimental endeavor with more structural characterization approaches aided with theoretical modeling will help to understand and predict the structures, properties, and transformation mechanisms of these nanomaterials and to shed light on their new promising applications.

Acknowledgments

This work is supported by a Discovery Grant, a Research Tools and Instruments Grant from Natural Science and Engineering Research Council of Canada, a Leaders Opportunity Fund from Canadian Foundation for Innovation and an Early Researcher Award from Ontario Ministry of Research and Innovation. The authors acknowledge Prof. T. K. Sham for supplying the material, Drs. D. Shakhvorostov and C. Murli for experimental assistance, Drs. J. Hu and Y. Meng for technical assistance, and Drs. H. Liu and K. K. Zhuravlev for helpful discussions. The original experiments were performed at X17C beamline at the National Synchrotron Light Source at Brookhaven National Laboratory and at HPCAT (Sector 16), Advanced Photon Source, Argonne National Laboratory.