Summary - Growth of AlN nanotips by means of

Chapter 4 Growth of AlN nanotips by means of

4.4 Summary

Aluminum nitride nanotips growth on metal (gold, aluminum and platinum) coated or even uncoated silicon substrates via vapor transport and condensation process has been demonstrated. A pure metal or metal silicide phase acts as the nucleation site for the precipitation of crystalline aluminum seed for the aluminum nitride nanotip growth. In case of aluminum nitride nanotip growth on bare silicon, a self catalytic activity of aluminum itself or its silicide phase was visualized. The resultant AlN nanotips exhibit a monodispersed apex angle distribution. Structural properties of the nanotips studied by TEM and XRD suggest that these tips have hexagonal crystal symmetry with a preferred growth direction of (002) along the long axis and a stable (221) plane as the tilted surface.

Chapter 5

Structural Evolution of AlN Nano-structure: Nanotips and Nanorods

In this chapter, Nanostructures produced on the silicon substrates at different temperatures set-points, controlled between 950 and 1200 °C, were studied. An observation in the crystal morphology change from tip-like structure to rod-like crystallites is reported. The surface morphology, structure, and composition properties of the as-deposited samples are characterized in detail. The overall structural and morphological results are discussed in section 5.2 via FESEM, Raman scattering, and XRD measurements, respectively. The platelet growth model, which is governed by the n1/n2 ratio are discussed in section 5.3. Finally, section 5.4 will make a summary according to the obtained results.

5.1. Nanotips and Nanorods

A series of experiments were carried out to study the effect of growth temperature (Td) variation on the shape evolution of the AlN products. Figure 5.1a-d shows SEM images of AlN nano-products grown for 30 min on the Au coated Si substrates at different Td of 950, 1000, 1100, and 1200°C, respectively. At Td = 950°C, the

top of the nanotips, forming a flat section at the top instead of the tip shape. The area of this (001) facet increased with increasing Td as shown in Fig. 5.1b - d. As shown clearly in Fig. 1d, Td = 1200°C produced AlN nanorod (AlNNR) which has a simple six-sided faceted structure. The cross-section of AlNNTs and AlNNRs were both hexagonal.

FIG. 5.1 Typical SEM images of the AlN nanotips on silicon substrates (coated with 15 nm of gold) grown under (a) 950, (b) 1000, (c) 1100, and (d) 1200 °C, respectively.

The cross-sectional SEM images of AlNNTs and AlNNRs shown in Fig. 5.2a and 2b, respectively, indicates their relative dimensions. The inset of Fig. 5.2a and 2b depict the early stages of growth of AlNNTs and AlNNRs, respectively. It indicates that the shapes of the AlN nanostructures were probably decided at the very early stages of the growth. Fig. 5.2c represents the X-ray diffraction analysis of the AlN nanostructures. It shows the strongest reflection of the (002) plane compared to (101) of the h-AlN, indicating a preferential growth perpendicular to the basal plane along [001]. The Raman spectra of the AlNNTs and AlNNRs, as shown in Fig. 5.2d, were collected using Ar⁺ laser as the excitation source. Three clear Raman-active phonon modes can be observed. These are in agreement with those of the h-AlN crystal [126].

FIG. 5.2 Typical cross section SEM image of AlN nanotips grown coated with 15 nm Au coated Si substrate under (a) 950, (b) 1200 °C, respectively. Inset in (a) and (b) show the SEM image of AlN nano-product grown for 25 minutes, respectively. (c) Typical XRD and (d) Raman spectra taken from the AlN nanostructures in Figure 2(a) and 2(b), respectively. The two vertical dashed lines in Figure 2(c) represent bulk AlN positions for (100) and (002) reflections (JCPDS 25-1133).

5.2 Structural revolution of AlN nanostructures

The shape evolution of the AlN nanostructures as a function of Td is interesting as the result (Fig. 5.1) shows. The digestion of the results presented in this paper as well as our earlier work on AlNNTs [166,167] yields certain facts as listed below:

Observation 1: AlN nanostructures have pure Al signatures from the seed crystal

present at its base [166];

Observation 2: The body of the nanostructure is purely hexagonal AlN growing along

[001] and no metallic Al phase is present [166,167];

Observation 3: A high aspect ratio (>10) nanotip structure, having apex angles of

~14°, produced at Td = 950 °C modifies into a low aspect ratio (~5) flat top ((001) facet) AlNNRs at Td = 1200 °C. Any proposed growth model should explain these listed facts of which the last one is of most importance.

We begin with a discussion on a diffusion mediated growth mechanism which is the most obvious as we are dealing with a temperature activated process. The growth of AlN nanostructures proceeds by the transport of Al vapors to the growth region as the temperature inside the quartz tube is ramped up. Note, although a thin Au layer, intended to be a catalyst, was used on the Si substrate, it was not detected at the tip of the nanostructures or in its body. Again the AlNNTs could be obtained even without

pursue the vapor-solid (VS) route instead. We introduce a diffusion mediated growth mechanism here. Al vapors landing on the growth surface will migrate with a large diffusion length, LAl, and can be adsorbed and deposited on the substrate as Al itself when Td is low (but above 660°C, the melting point of Al powders). This results in the basal Al islands since Td is not sufficient to dissociate the nitriding gas NH3. This explains observation 1. The Al vapors may then encounter and react with NH3 or its dissociation products when Td is higher (dissociation temperature for NH3 is) producing AlN following eqn. 1:

3 2 3

2Al+ NH = AlN+ H (1) ∆G = - 9.7 kJ/mol at 950 °C ∆G = -14.6 kJ/mol at 1200 °C

Then AlN will migrate in molecular form with a diffusion length, LAlN. Consequently, an effective diffusion length Leff is defined as

Leff = LAl + LAlN (2)

where LAl and LAlN are temperature activated according to the following [168]

LAl ~ _⎟⎟

t and E are the diffusion times and activation energies for surface diffusion, respectively, and kB is the Boltzmann constant. The AlN molecule can be deposited at

a low energy site and contribute to the growth or can be desorbed also. This happens

in a condition close to tAl ~ tAlN, and LAl ~ LAlN. Thus, to permit the growth of AlN

from a predominantly metallic Al seed, LAlN ≥ LAl. Above the dissociation temperature

of NH3 the thermodynamics of the system (negative Gibb’s free energy for eqn. 1)

promotes the formation of AlN. This explains observation 2.

To address observation 3 we discuss the anisotropy as well as strains that might be

present during the growth. We are now studying the growth of AlN on Al seed crystal.

The initial islands may have edges beveled at an angle θ, with respect to the substrate,

determined by the surface energetics [169]. Let us discuss first the diameters of the

AlN nanostructures grown at different Td. Figure 5.3 shows a schematic drawing of

the cross-section and top view of the hexagonal close packed (hcp) wurtzite structure

of AlN crystal with the Al- terminated (001) face and the N- terminated (001) face.

From our TEM and XRD results we have already established that we have a [001]

growth direction for all kinds of nanostructures grown at different Td. Number of

dangling bonds available per oncoming growth precursors, say Al, on the Al (001)

and N (001) face are 3 and 1, respectively (Fig. 5.3a). The lateral facets have two kinds of surfaces, namely A and B, which have 2 and 1 dangling bonds, respectively,

per oncoming Al atom. Hence on the average each Al atom has 1.5 dangling bonds

availability of dangling bonds, diffusion lengths on the lateral faces of the hexagonal

islands will be less than that on the N-terminated

( )

⁰⁰¹ face [170]. This will lead to a anisotropy in the growth rate resulting in large platelets/islands of AlN following eqn.

(1). Since the diffusion length is temperature activated, a larger Td will produce larger

but thin platelets. This exactly explains why the AlNNTs produced at 950 °C (Fig.

5.1a) had smaller base diameters than the AlNNRs (Fig. 5.1d) prepared at 1200 °C.

However, this also means that the AlNNRs are closer to equilibrium than the AlNNTs

which is understandable because of a larger diffusion length of the AlN precursors at

higher Td. Following the model proposed by Tersoff and Tromp [169], the island size

(a0) is related to the surface energy (Γ) and a stress factor (c) by the following:

where h is the height of the initial island and ξ = . This model predicts that

if ch>>Γ , then the island sizes a

0 will be smaller. So the AlNNTs having a smaller base diameter are the stressed ones. However the model presumed a slant edge (θ ≠

90°) island and cannot directly predict a perfect (θ = 90°) wide base rod-like crystal, since then ξ = 0 and a0→0. There are indications that such perfect (θ = 90°) islands

[171]can be the seeds for nanorod growth via the VS mechanism.

FIG. 5.3 (a) The cross-sectional view of hexagonally close packed wurtzite crystal

structure of AlN. (b) The plan-view of a bilayer of hcp AlN. There are two types of

lateral growth directions marked as A and B.

Now consider the vertical growth of the AlN nanostructures along the [001]

direction which is an Al-terminated face (Fig. 5.3a). For Td ≥ 950 °C, the dissociation

rate of NH3 was very fast and the Al vapor pressure was high, suggesting tAl → 0 and

0 ← LAl << LAlN; so Leff = LAlN. When Al atoms arrive at the growth face, every Al

atom has three bonds connecting with three N atoms. That means, EAlN (001) >> EAlN

( )

⁰⁰¹ and also EAlN (001) > EAlN(010), resulting in growth rates R > R >

[ . A higher growth rate along the Al-terminated [001] direction is observed than that along the normal to the lateral faces. When this condition is satisfied, we expect

an aspect ratio >1.

The beauty of crystal growth is depicted in Figure 5.4. The AlNNT produced at 950

°C (Fig. 5.4a) clearly shows a stacking of different sized platelets along the growth

direction generating the tip shape. The platelets/islands behaves similar to a building

block and a repetitive stacking of these, layer by layer, gives rise to the nanostructures.

A similar stacking of double bilayer steps has been observed along the sidewalls of

InN pyramids also [171]. However the existence of these distinct platelets started to

disappear from Td = 1000 °C (Fig. 5.4b) signifying deposition along the lateral facets

of the platelets. A significant portion of this deposition may be due to adatom hopping

down a step. Fig. 5.4a – d demonstrates the growth along the radial direction of the

nanostructures due to adatom hopping as a result of a reducing Ehrlich- Schwoebel

barrier, and increased diffusion lengths as Td is increased. When Td = 1100 °C, the top

(001) facet clearly appears and the lateral face growth becomes more pronounced

giving it a smoother appearance than that observed during 1000 °C. However the

conical nature of the structure is still maintained. As Td creeps up to 1200 °C, the

diffusion length was sufficient to push the system towards equilibrium resulting in

larger hexagonal platelets and growth along the lateral facets generating the pillar like

structure of the AlNNRs with a near perfect uniformity in diameter. Note that the

existence of the Ehrlich- Schwoebel barrier at low Td will also give rise to a large

driving force along the [001] growth direction [172,173]. In addition, the stress along

the basal plane of the platelets, as indicated from the smaller base diameters of the

AlNNTs, is also known to drive the growth along a direction perpendicular to it. This

explains why the AlNNTs have a higher aspect ratio than the AlNNRs.

FIG. 5.4 (a-d) High resolution SEM images of the AlN nanostructures produced at

different growth temperatures; (e) HRTEM image of the edge of AlNNTs showing the

step edges, including the step height (h) and step spacing (λ); a schematic aiding in the estimation of the semi apex angle (φ/2) is also shown; (f) HRTEM image of the

edge of AlNNRs showing uniform diameter without tapering via step edges.

Lastly, the origin of the tip shape has to be addressed for the growth at 950 °C.

Since the growth units were clearly identified as islands/platelets stacked on each

other, it is only obvious that a certain arrangement of these produced the tip shape.

Careful TEM observations proved what we conjectured in our earlier work [166].

This is a pyramidal growth of 3D islands growing on top of each other (Fig. 5.4e). A

rapid 2D nucleation, which may be due to precursor supersaturation, can lead to these

3D islands/platelets. The tilted surface of the AlNNTs consist of a train of parallel

steps of polyatomic step heights (h), with a small step spacing (λ) (Figure 5.4e) and a high step density (p = h / λ ) [10]. Clearly, h=n₁×d₀₀₁, and λ=n₂×d₁₁₀, where

d001 and d₁₁₀ are the lattice spacings along and perpendicular to the growth direction, respectively, as determined from TEM studies [174] and n1, n2 are integer. The apex

angle (φ) of the AlNNTs, as determined microscopically, is dictated by the large h/λ

ratio and hence small magnitudes of the ratio, though present, are neglected.

Calculating h/λ from the TEM micrograph shown in Fig. 5.4e we can roughly

estimate a set of φ values for our AlNNTs which falls within 8-22° in agreement with

that measured from the SEM images. Note that the apex angles mentioned here are

2D projections of the solid angle at the apex. For a particular material system the

lattice spacings are constant and hence h/λ is a function of n1/n2. Hence following this

example, GaN, InN, AlN) can be predicted based only on the n1/n2 ratio. The AlNNRs

does not exhibit these step structures (Fig. 5.4f). The growth model proposed in ref. 8

is inadequate in a sense that reducing precursor concentration is a general reaction

feature and if the model is true should necessarily produce nanotips, which is not the

case. However their observation is very similar to us since they can generate the

nanotips and nanorods as a function of growth temperature alone and that has been

attributed to diffusion mediated process. Our results indicate rather a stress related

phenomenon for the step spacing (λ) and a high Ehrlich-Schwoebel barrier to preserve

the steps that build up the AlNNTs. The positions for the (002) and (100) reflections

from the nanostructures are shifted towards lower and higher 2θ values, respectively,

from their bulk counterparts as marked in Fig. 5.2c. Note that the [110] and [100]

have identical 2θ positions in the XRD and that [110] is the radial direction of the nanostructures shown in the TEM images in Fig. 5.4e and f. The tensile and

compressive stress along the axial and radial direction, respectively, of the AlNNTs

and AlNNRs is clearly visible from the XRD spectra (Fig. 5.2c). The AlNNRs

produced at 1200 °C manages to counter the stress deformation while AlNNTs

produced at a lower temperature yields to it by getting tapered as growth progresses.

5.3 Magic ratio

Following the report in chapter 5.2, we deliver a model of platelet growth of AlN

nanostructures which will lead to a train of parallel steps of polyatomic step heights

(h), with a small step spacing (λ). It is obvious that the step height and spacing should be the integer multiple of lattice spacing of d₀₀₁and d₁₁₀, respectively, are known if

the growth direction was fixed at [001]. The angle (θc) between the existing facet and

basal plane can be compute by the following expression.

the cleaved facet and the basal plane in the crystallography point of view, CaRIne v

3.1), existing facet (adapted from CaRIne v 3.1), apex angle and reported literatures.

The digestion of the results presented in table 5.1 as well as listed references on

AlN provide certain information as lined up below: First, the apex angle (φ) will be

determined by the a simple n1/n2 ratio (or h/λ); second, θc are highly match with θr,

especially for the n1/n2 ratio equal to 0.5 [164], 1 [175], and 3.5 [166], respectively

which were well documented; third, the apex angle will not be a any number, it is

seems to be predictable following the table.

Surprisingly, this model can applied for group-III nitride nanostructures with long

axis at [001] direction as shown in table 5.2, 5.3, and 5.4. The identical calculation

was made for the GaN, InN and BN, respectively. Here we propose a regulation

which can be applied to predict and describe the tip formation and apex angle

variation of group-III nitride nanostructures. However, the reason for the apex angle

variation is not clear for the moment.

Table 5.1 n1/n2 ratio of AlN

Ref.: Pearson’s Handbook of Crystallographic Data, p213.

Table 5.2 n1/n2 ratio of GaN

Ref.: Pearson’s Handbook of Crystallographic Data, p3492.

Table 5.3 n1/n2 ratio of InN

Ref.: Pearson’s Handbook of Crystallographic Data, p4017.

Table 5.4 n1/n2 ratio of BN

Ref.: Pearson’s Handbook of Crystallographic Data, p1572.

5.4 Summary

To summarize, the growth of high purity and dense hexagonal aluminum nitride

nanotips and nanorods on gold coated silicon substrates via a thermal chemical vapor

deposition system has been demonstrated. A single growth parameter, namely the

growth temperature, is able to dictate the formation of nanotips or nanorods in this

vapor transport and condensation process. Optimal growth temperature for these

nanotips and nanorods were 950°C and 1200°C, respectively. The AlN nanotips and

nanorods grow by vapor solid mechanism and their preferred growth direction (long

axis) is [001]. Composition and structures of the nanostructures have been confirmed

by TEM, XRD, and Raman spectroscopy. A diffusion mediated growth model

incorporating an Ehrlich-Schwoebel barrier and compressive stress at lower growth

temperatures has been proposed to explain the formation of high aspect ratio nanotips

with substantial microscopic and structural evidence.

Chapter 6

Luminescence of AlN nanotips

In recent times, wide band-gap III-nitrides have attracted considerable interest due to their outstanding thermal and chemical stability, leading to major advances for the devices which are active in the short-wavelength and capable of operation at high temperatures and in hostile environments [9]. Recently, the synthesis of one-dimensional nanostructured AlN opened potential applications in electronic/optic devices [177,178]. Therefore, understanding of electronic/ optic properties of AlN is of a great interest. Previous studies on the optical properties of AlN were well-documented. The near band-edge transition was investigated by using cathodoluminescence (CL) [179-181], photoluminescence (PL) [182,183], thermoluminescence (TL) [184,185], and optical reflectance [186] measurements on bulk as well as epi-AlN samples. Defect-related transitions were characterized in these systems as well [184,185,187-190]. Optically simulated luminescence measurements using visible or infra-red light and TL studies assisted with excitation spectral analysis showed that defect related emission peaks are due to radiative recombination process involved in complex formed due to oxygen related impurity (ON) and aluminum vacancies (VAl) [184,185].

Single crystalline, free-standing, nearly strain free and high quality nanostructures are preferred over thin film or bulk materials, since they serve as ideal systems for measurements of the fundamental properties of AlN [191]. However, there are only a few reports on the optical properties of nano-size AlN [192]. Therefore, it is useful to study and understand the luminescence features of AlN nano-structure. In this chapter, the optical properties nanostructures were studied by various optical techniques.

Temperature dependent cathodoluminescence results are discussed in section 6.1. The photoluminescence and photoluminescence excitation properties were reported in section 6.2. Section 6.3 displays thermoluminescence and thermoluminescence excitation spectra. Absorption phenomenon was exhibited is the section 6.4. A luminescence model of AlN was proposed in section 6.5. Finally, section 6.6 will make a summary according to the obtained results.

6.1 Temperature dependent cathodoluminescence (CL)

Temperature-dependent CL studies (Fig. 6.1) showed two broad emission bands at 3.4 and 2.1 eV. The peak at 3.4 eV dominates the spectral range from 1.8 to 6.2 eV.

This peak appears at the same wavelength as the well-known peak due to oxygen-related defects. According to Youngman et al., [187,193] the dominant line

AlN lattice. Interestingly, the position of the oxygen related defects luminescence in AlN can vary from 3.3 to 4.0 eV as the concentration of oxygen content are changed.

Based on M. Strassburg et al. [190], it was presumed that nitrogen vacancies and Al interstitial point defects are responsible for the 2.1 eV emission band. The CL spectra obtained from R. A. Youngman [193], where two peaks at 3.4 eV and 2.1 eV were evident suggesting that it is likely that our materials had the same level of oxygen content dissolved in the AlN lattice. The origin of these two peaks will be discussed in more detail subsequently, with the analysis of the midgap luminescence of AlNNTs.

The inset in Fig. 6.1 shows a low temperature CL spectrum of AlNNTs measured at high resolution in the near-band-edge energy range. This short wavelength emission close to ~6 eV is related to the transitions at the fundamental absorption edge. This emission can be further deconvoluted into two peaks (inset, Fig. 6.1), centered at 6.07

在文檔中一維奈米氮化鋁的製備與光電特性分析 (頁 76-0)