5.4 Growth and phase determination of SiC-SiO2 nanocones Figure 5.1 High Resolution SEM image of dense, uniform SiC nanocones At the end of the deposition, a deep blue film could be seen
Trang 1Chapter 5 Growth of silicon carbide nanocones
In this chapter, we continue to use the VLS method to grow silicon carbide nanocones
5.1 Introduction
SiC is an important group IV semiconductor material with a bandgap of 2.3
eV It has received much attention for its applications in high frequency, high power and high temperature devices due to its high breakdown electric field, electron mobility and good thermal conductivity [1-5] Theoretical calculations and experimental results show that the electricity and strength of SiC nanorods are better than those of large whiskers and bulk crystalline SiC, [6-7] the elastic constant can reach the theoretical value of 600 Gpa for (111) oriented SiC [6]
5.2 Motivation
Controlling nanostructure formation is one of the key steps in fabricating nanomaterials and has been the focus of intensive research efforts For example, creating architectured assembly of dimensional nanostructures is essential for device integration in field emission arrays, sensor arrays, multi-tip arrays for dip-pen lithography, photonic waveguides etc [8-10] There have been extensive research efforts on the synthesis and assembly of nanorods and nanotubes for such
Trang 2mechanical or electronic applications However there have been relatively few systematic studies on ways to precisely engineer the shape and form of the components for functional nanoscale mechanical devices For example, a conical structure offers substantially higher mechanical and thermal stability than a narrow cylinder Structurally, nanocones can have nanometer-sized tips and micrometer sized bases, rendering their manipulation easier than nanotubes
Nanocones are also mechanically stiffer and less prone to bending and thermal shock, making them ideal candidates for scanning probe tips, or as nano-syringes for quantum dot injection into biological cells Zhang [11] reported the chemical vapor deposition (CVD) synthesis of tubular graphite cones consisting
of annular rings of graphene planes concentric with a hollow interior The conical structure is due to the progressively shorter terrace edges going from the inside to outside, arising from the sequential growth of shorter secondary layers on the inner layers as the central tube grows upward Khrisnan [12] used arc discharge methods
to synthesize geometrically precise, hollow nanocones that consisted of folded conical graphene planes Vladmir [13] balanced the growth and etching effects of the acetylene-ammonia mixtures to synthesize carbon nanocones made of cylindrical carbon nanofibres surrounded by carbon precipitates on the outside All the nanocone structures reported thus far are straight-growing, crystalline cones made of carbon
Trang 3Many researchers have been trying various methods to synthesize one-dimensional SiC rods The detailed methods are listed in table 5.1
Table 5.1 Methods for synthesizing SiC nanorods and nanowires
Dai et al SiC nanorods nanotube template and SiO or Si+I2 1995[14] Meng et al SiC nanorod within
Lai et al SiC nanorods hot filament CVD on Si substrate using
mixture of Si, SiO2 and carbon above
SiC rod as the anode to arc discharge 2001[19]
Gao et al Needle-Like SiC
Trang 4temperatures of more than 1000 ˚C were needed for these reactions
In this study, we report for the first time the synthesis of silicon oxide-ensheathed silicon carbide nancones using tetramethylsilane as the single source precursor at temperatures under 900 ˚C It is discovered that the nanocones show a propensity to undergo bending at various angles and we will discuss the growth mechanism
5.3 Growth conditions for SiC nanocones
Silicon wafers coated with 20 nm of nickel film were introduced into the MW-PECVD system The nickel-coated Si(100) substrate was pretreated at 800 ˚C using a pure hydrogen plasma in order to remove the impurities and oxide film from the substrate prior to growth A single source precursor, tetramethylsilane, was used for the deposition of the SiC nanocones The details of sample pretreatment and the growth conditions used in the MP-PECVD system are listed as follows:
H2 pre-treatment Growth condition
Trang 55.4 Growth and phase determination of SiC-SiO2 nanocones
Figure 5.1 High Resolution SEM image of dense, uniform SiC nanocones
At the end of the deposition, a deep blue film could be seen on the Ni -coated
Si substrate Visualization of the surface deposits by SEM reveals the growth of both straight and bent conical fibres with an overall length of several micrometers, as shown in figure 5.1 Beneath the nanocones, the substrate is covered by ball-like deposits about 0.5 µm in diameter The base of the cone is micrometer-sized and is anchored to the micron-sized ball whilst the tip is nanometer-sized
Trang 6500 1000 1500 2000 2500 20
1200 cm−1 respectively [23,24] The second band is originally IR-inactive but can be
Trang 7activated by disorder-induced mode coupling The intensity of SiO2 peaks after deposition becomes stronger compared to the peaks due to native oxide before deposition, which indicates the film consists of large quantity of amorphous SiO2 The additional peak at 820 cm-1 was observed after deposition which is the characteristic stretching vibration of crystalline SiC [25]
Figure 5.3 (a) TEM image of the tip of the cone The size of the tip is around 20 nm, one straight rod with diameter 10 nm is at the center of the cone; (b) High resolution TEM image which shows that the inner nanorod is crystalline cubic SiC growing along the <111> direction, the resolved {111} lattice planes are separated by 0.25 nm
The internal microstructure of the nanocone was studied using TEM Figure 5.3(a) shows a low magnification TEM image of a nanocone The inset shows the whole morphology of this nanocone The peripheral wall of the cone is smooth and a nanorod with diameter of 10 nm is concentric to the cone, tipped by a catalyst particle A high magnification view reveals that the center of the cone has a coaxial crystalline rod of about 10 nm diameter The HRTEM image of this coaxial rod in
(a)
(b)
Trang 8figure 5.3(b) shows lattice fringe separations of 0.25 nm consistent with the cubic β-SiC {111} interplanar separation, whilst the outer coat is amorphous TEM observation shows that the inner SiC nanorod grows preferentially along the <111> direction
Figure 5.4 Energy filtered maps of silicon carbide/silicon dioxide nanocones The maps show that the center rod is crystalline silicon carbide and the outer layer is silicon dioxide
The elemental distribution was verified by EELS mapping in figure 5.4 The elemental maps were obtained from the L23 edge of Si, and the K edge of C and O Silicon is found in both the body and the tip of the cone although the Si signal at the tip arises from the overlap of the Ni M and Si L edge Carbon existing at the center
Trang 9conical coat is silicon oxide Since the atomic concentration of silicon in SiO2 is lower than that in SiC, the brightness indicates that the density of silicon atoms in the outer sheath is lower than that of the center The metal particle at the tip is nickel Elemental mapping clearly proves that center core of the nanocone is cubic SiC, while the amorphous outer sheath layer is SiO2
Figure 5.5 Electron energy loss spectra (a) collected at region ‘A’, focusing on the rod; (b) collected at region ‘B’, focusing on the side of a cone
More information on the SiC-SiO2 nanocones is obtained by interpreting the energy loss near edge structure (ELNES) of the Si, C and O edges The near edge energy loss structure arises from the excitation from core shell electrons to the vacant levels above the Fermi level The fine structure and position of the Si-L2,3 edge depends on the ionicity of the bond with the ligand atoms We can identify and differentiate the SiX (X=O, C, P, N) compound using ELNES as a fingerprint
Trang 10method [26] Background substracted EEL spectra taken from side and tip of the cone are shown in figure 5.5 Focusing the electron beam on the narrower region ‘A’
of the cone where the rod is more prominent produces a spectrum with C and Si edges characteristic of SiC Focusing on the thicker region ‘B’ on the side produces
an EELS spectrum with Si and O edges, which are characteristic of the SiO4tetrahedral cluster The Si L-edge consists of two sharp peaks at 107.3 and 113.9 eV, and a third broad peak at 129.4 eV which is separated by about 22 eV from the first peak Such ELNES structures of SiC and SiO2 had also been previously recorded by L.A Garvie [27] The small C-K edge signal may come from the supporting holey carbon film
The composition of the SiC-SiO2 cone was determined by the quantification method which is described in chapter 3 The atomic ratio of the inner SiC rod and side SiO2 wall of the cone is Si/C=1.0:(0.9±0.12) and Si/O=1.0:(2.12±0.25) respectively All these results indicate the SiC-SiO2 nanocone consists of the nearly stoichiometric SiC nanorods covered by an amorphous SiO2 layer
The presence of nickel catalyst at the tip of the SiC rod suggests tip-catalyzed growth following the classic vapor-liquid-solid mechanism [28] The single source precursor tetramethylsilane decomposed into SiC vapor and diffused into the Ni catalyst particles Since the {111} plane of β-SiC has the lowest surface free energy, estimated to be about 2830 erg/cm2, which is much lower than those of other planes
Trang 11such as the {110} (3450 erg/cm2), {211} (3990 erg/cm2) and {100} (4890 erg/cm2), most of the SiC nanorods grow along the preferred <111> direction [29]
The presence of an outer amorphous SiOx coat indicates that lateral growth
of amorphous SiOx occurred on the SiC rod template simultaneously There was no intentional introduction of oxygen in the growth environment Some leakage of the oxygen into the chamber may have promoted the oxidation of the single organic source Unlike the hexagonal phase carbon nanotube, the stacking of cubic crystal planes to form a rod in the case of β-SiC creates a high density of reactive edge sites
on the surface, these are also inevitably oxidized during the continuous adsorption of reactive SiO2 ions in the plasma environment [15, 16, 20, 22] Depending on the rate
of lateral growth of SiOx relative to the vertical growth of the SiC rod, a rod-in-cone structure results The schematic growth process of SiC-SiO2 nanocones is shown in figure 5.6
Figure 5.6 Schematic showing the growth process of SiC nanocones
Trang 12Since the lateral growth of the SiO2 results from the rapid precipitation of gaseous SiOx species in the plasma ball on the SiC rod, the position of the substrate
in the chamber and applied bias play important roles in the growth process If we place the substrate away from the center of the plasma ball, coaxially structured SiC-SiO2 nanorods instead of nanocones are obtained These coaxial nanorods also could be found when the bias was not directly loaded on the sample The balance between plasma etching and precipitation speed of SiO2 accounts for the morphology of the products
5.5 Structural properties of bent SiC-SiO2 nanocones
As mentioned, some bent nanocones were observed in the SEM images Figure 5.7(a) and (b) depict a nanocone with smooth curvature bent at 80˚ Since the SiC wire acts as a template for the lateral growth of SiOx, the turning SiC wire causes a similar turn in the conical SiOx deposit around it, resulting in the growth of
a bent nanocone The amorphous nature of the SiO2 deposit allows the smooth merging of interfaces between the segments of the bent cones that would otherwise
be difficult for crystalline systems The growth of the bent nanocone is a phenomenon restricted to the unique material combination of a cubic phase wire and
an amorphous oxide sheath Thus far, there have been no reports of bent nanocones for the graphitic system because the growth mechanism precludes the sharp angular
Trang 13bending of crystalline graphene sheets in the c-axis The amorphous phase has no such restrictions and remarkable branching and merging between interfaces of silicon oxide nanowires to form branching networks have been recently reported
Figure 5.7 TEM images showing (a) low mag image of bent SiC-SiO2 nanocones; (b) magnified image of bent area; (c) a SiC nanowire bent at 70˚; (d) a SiC nanowire bent at 110˚
Trang 14Figure 5.8 HRTEM and dark field image of bent SiC nanorod junction area
A
B
C
Trang 15However, the cumulated amorphous coating at the bent area is too thick to permit enough electrons to pass though, preventing high quality electron imaging of this elbow area In order to understand the intrinsic properties of the bent cones, bent SiC nanorods with a thinner SiO2 sheath was prepared Placing the sample away from the plasma ball reduces the rate of SiOx precipitation, and nanofibres instead
of nanocones are obtained We observe that the β-SiC nanorods has an intrinsic propensity to undergo changes with certain angles, such as 70º and 110º Two typical examples are shown in figure 5.7(c) and (d) Previous studies of SiO2-ensheathed SiC nanorods revealed that the growth axis changed frequently between the {x11} family planes (x = 1, 2 or 3) in the course of growth to minimize the surface energy, resulting in a zigzag coursing along the rod [30] If the density of twins is dominant, the nanorod grows along [111] The growth direction changes to [211] if a defect free block is grown, and to [311] if there is a constant translation between adjacent growing blocks The bending angles match the angles between {111} planes very well The corresponding plane angles of {111} and other common faces are listed in table 2 [29] Considering that the SiC growth follows the VLS mechanism, SiC nanorods may grow along the <111> direction normal to the lowest surface energy {111} family of planes The {111} family of planes consist of
four planes: (111), ( 111 ), (111) and (111) and their opposites The switch of directions between these planes results in the bending of the nanorods
Trang 16Table 5.2 Corresponding plane angles in β-SiC crystal
Trang 17To achieve an understanding of how deformation stress and twinning defects propagate to influence the bending of the cone, a detailed HRTEM image of the bent nanorod junction is shown in figure 5.8 A schematic drawing of the SiC crystal structure is shown in figure 5.9 The schematic shows the three layer stacking sequence of ASiAcBSiBCCSiCC with alternating Si and C layers Since the stacking sequence along the [111] direction produces large quantities of dangling bonds on the side of the crystal, the sides of the SiC nanorod are passivated by SiO2 in the plasma ball Hence, 1-2 nm of SiO2 layer is inevitably observed on the side of the SiC nanorod in the HRTEM image
A selected area diffraction pattern of the nanorod demonstrates that the rod is
a single crystal containing twins, and the two sections of the bending nanowires grow along the [111] and [11 ] directions A dark field image shows that both 1nanowires have stacking faults, including the parts before bending These stacking faults and amorphous segments shown in HRTEM image are likely to change the growth direction from [111] to [11 ], the defect free region 1
The SiC wire-in-silicon oxide nanocones may offer applications for structural applications due to their mechanical strength For example the catalyst at the SiC nanowire tip can be removed, exposing a SiC nanowire which has its stem encapsulated by a mechanically stiff SiO2 coat for support Engineering straight SiC wire-in-silicon oxide nanocones however is difficult since it is not known whether
Trang 18the initial impetus for a change in growth direction of the SiC nanowire is purely random, i.e plasma process fluctuations, or entropy-driven etc It is possible that defects in stacking were influenced by oxygen ions in the plasma, because a much higher density of bent nanowires was observed in this work compared to previous studies One possibility to direct the growth of the nanocone may be using templates with nano-sized channels to externally constrict the bending angles
5.6 Conclusion
In summary, we have synthesized a rod-in-cone structure using CVD Our results suggest that the SiC nanorod is inevitably ensheathed by an amorphous silicon oxide coat to passivate the reactive edges In a plasma chemical vapor deposition system, lateral growth of the amorphous oxide deposits can result in a conical sheath around the coaxially aligned cubic phase rod
A change in the growth direction of this rod, due to stacking faults and twinning defects intrinsic to the cubic phase, results in the growth of a bent nanocone The switch of directions between {111} planes results in the bent nanocones Such bent nanocones may offer nanomechanical applications as nanocantilevers
Trang 19[7] Macmillan, N.H J Mater Sci 1992, 7, 239
[8] Piner R.D.; Zhu, J.; Xu, F.; Hong, S.H.; Mirkin, C.A Science 1999, 283, 661 [9] Teo, K.B.K.; Chhowalla, M.; Amaratunga, G.A.J.; Milne, W.I.; Pirio G, Legagneux, P.; Wyczisk, F.; Pribat, D.; Hasko, D.G Appl Phys Lett 2002, 80,
2011
[10] Greene, L.E.; Law, M.; Goldberger, J.; Kim, F.; Johnson, J.C.; Zhang, Y.F.; Saykelly, R.J.; Yang, P.D Angew Chem Int Ed 2003, 42, 3031
[11] Zhang, G.Y.; Jiang, X.; Wang, E.G Science 2003, 300, 472
[12] Krishnan, A.; Dujardin, E.; Treacy, M.M.J.; Hugdahl, J.; Lynum, S.; Ebbesen, T.W Nature 1997, 388, 451
Trang 20[13] Merkulov, V.I.; Guillorn, M.A.; Lowndes, D.H.; Simpson, M.L.; Voelkl, E Appl Phys Lett 2001, 79, 1178
[14] Dai, H.; Wong, E.W.; Lu, Y.Z.; Fan, S.S.; Lieber, C.M Nature 1995, 375, 769 [15] Meng, G.W.; Zhang, L.D.; Mo, C.M.; Zhang, S.Y.; Qin, Y.; Feng, S.P.; Li, H.J
Trang 21[25] Ehara, T.; Notake, K.; Handa, K Diamond Relat Mater 2001, 10, 1287 [26] Auchterlonie, G.L.; Mckenzie, D.R.; Cockayne, D.J.H Ultramicroscopy 1989,
31, 217
[27] Garvie, L.A.; Craven, A.J.; Brydson, R Am Minera 1984, 79, 411
[28] Wagner, R S.; Ellis, W C Appl Phys Lett 1964, 4, 39
[29] Bootsma, G.A.; Knippenberg, W.F.; Verspui, G J Cryst Growth 1971, 11,
297
[30] Wang, Z.L.; Dai, Z.R.; Gao, R.P.; Bai, Z.G.; Gole, J.L Appl Phys Lett 2000,
77, 3349
Trang 22Chapter 6 Novel heterogeneous reaction route to Cu
chalcopyrite thin films The last three chapters focus on the synthesis of inorganic nanostructural materials through VLS mechanism In this chapter, a novel heterogeneous synthesis route to copper chalcopyrite thin films by vapor-liquid method will be developed and discussed
6.1 Introduction
Thin solar cell technology with Cu multinary chalcopyrite absorbers has received considerable interest over the past 20 years due to the potential of these semiconductor materials to act as the absorber layer in solar cells CuInSe2 has been the most intensely studied as it has achieved, on a laboratory scale, conversion efficiency close to 19% from a recent development of CuInSe2 and related compounds [1] CuInSe2 thin film has a direct band gap of about 1.0 eV at room temperature and possesses a high absorption coefficient of 104-105 cm-1 [2-4] Having this large absorption coefficient, even a 1 µm thick CuInSe2 film would be able to absorb nearly 93% of radiation falling upon it [5] It can be prepared as either
a n or p type semiconductor by adjusting its stoichiometry It has good electrical and thermal stability [6] CuInS2 has similar physical, chemical and electrical properties
as CuInSe2 but it is studied to a lesser extent Its direct band gap is about 1.5 eV A calculation by Meese et al [7] concluded that solar energy efficiencies between 27 and 32% are theoretically achievable in CuInS2 p/n homojunctions Based on this theoretical efficiency, the band gap of 1.5 eV is nearly the optimum for efficient
Trang 23cm [9] and the solar energy conversion efficiency of 10% make CuInS2 a suitable choice in photovoltaic applications
For other chalcopyrite materials, InS (2.44 eV) and In2S3 (2.07 eV) are medium band gap semiconductors GaS and Ga2S3 (3.4 eV) have potential applications as a passivating coating on GaAs [10, 11] CuGaS2 (2.49 eV) may be a promising material for production of light-emitting devices in the green light region [12]
6.2 Motivation
Since thin films of Cu ternary chalcopyrite have been regarded as a potential candidate for photovoltaic applications [13, 14], various routes to prepare these sulfide materials are well documented, such as solid state reaction synthesis [15], the homogenous precipitation method [16], microwave irradiation of stoichiometric amounts of the corresponding elements [17],annealing of coevaporated In-Cu films
in H2S or H2Se atmosphere [18], solvothermal routes[19, 20] and spray CVD from a single source precursor [21]
Among these synthesis routes, the single source precursor process has received much interest due to its potential approach to the deposition of thin films in MOCVD systems The single source approach presents a cleaner and simpler technique Compared to the conventional MOCVD precursors, these novel single source compounds exhibit the following advantages [22]:
Air and moisture stability
Low toxicity
Pre-reaction is limited, only single source precursor in the supply stream
Trang 24Ideal volatility for MOMBE
Impurity incorporation into films may be controlled by ligand design Ligand and metal are closely associated
Low temperature growth is possible
Therefore, in last two decades, chemists have been interested in synthesizing stable and efficient molecular precursors for various metal chalcopyrite materials [13, 14] Unlike their III / V analogues, studies on single source precursors for group III chalogenides are limited
[In(SCONEt2)3] that has been synthesized and characterized by O’Brien and coworkers [23] has a low boiling point of 98.5 ˚C, which makes it volatile and easily evaporated This precursor has been used for LPCVD and In2S3 films which were successfully deposited
Thiolate, thiocarbamate and thiocarboxylate ligands have been used to extensively synthesize single source precursors for various metal sulfide materials [16-19] Hampden-Smith and his coworkers have reported that thin films of β-In2S3could be obtained through aerosol assisted chemical vapor deposition from solutions
of the compound [HL][In(SC{O}Me)4] (L = 3,5 dimethyl pyridine) over the silicon substrate [30]
In recent years, more ternary single source precursors (type I-III-VI2) have been synthesized and characterized [(Ph3P)2Cu(µ-ER)2M(ER)2] (M = Ga and In; E
= S, Se; R = Et and Bui) and [Bu2In(SPr)Cu(S2CNPri2)] have been used to prepare thin films of CuInS2, CuInSe2 and CuGaS2 [21, 24-27] MOCVD from [Bu2In(SPr)Cu(S2CNPri2)] leads to the formation of CuInS2, CuIn5S8, mixtures of CuInS2 and CuIn5S8 or mixtures of In6S7 and CuInS2 depending on the carrier flow
Trang 25rate, base pressure and temperature employed [23,28] Hepp and co-workers [29] has even synthesized and characterized liquid single source precursors for the ternary semiconductor CuInS2 in order to improve the volatility of the precursor for CVD However, a possible disadvantage of such precursors is that their volatility is rather low and an assisted delivery method such as aerosol assisted CVD or flash evaporation has to be used Moreover, despite the potential advantage of these ternary compounds, the control of stoichiometry in complex compounds is difficult which usually results in a low-quality film This is attributed to some ternary species (Metal-Metal-S-ligand) having lower surface diffusion rates than the separate groups and species [22]
In general, the volatility of a binary precursor would be higher than a ternary single source precursor In this chapter, we present a novel and hitherto unknown heterogeneous reaction route to high quality ternary CuInS2, CuInSe2 and CuGaS2thin films on Cu coated Si substrate Binary neutral In(EPh)3 (E= S, Se) precursors and trialkylammonium salts of group 13 metals (Ga, In) thiocarboxylate compounds have been employed for deposition of the corresponding ternary metal sulfides (selenides) on Cu substrates through the heterogeneous reaction route
This technique eliminates the use of toxic and hazardous gases like H2Se and
H2S and presents some advantages over the conventional preparative methods There
is no need for precise control of flow rates as in multiple source MOCVD or spray CVD and also the stoichiometry of the final product can be controlled and maintained during the deposition
The objective of this project is to deposit high quality ternary thin films with the good crystallite, high purity and uniform grain size that are essential for the film
Trang 26device applications
6.3 Experiment
The synthesis process for the four precursors and the CVD experiment are demonstrated in chapter 2.2.2.1 Thin films of CuIn(Ga)E2 were grown by thermal
[Et3NH][Ga(SC{O}Ph)4]·H2O (2), In(SPh)3 (3) to In(SePh)3 (4) at 80, 100, 220 and
200 ˚C respectively onto Cu or Ni coated Si substrates inside a CVD chamber The appropriate evaporation temperatures were determined from the melting point and decomposition profile of the compounds in TGA studies
6.4 Growth of CuMeS2 (Me=In, Ga) from metal thiocarboxylate compounds 6.4.1 Description of precursor
The precursor compounds [Et3NH][In(SC{O}Ph)4]·H2O (1) and [Et3NH][Ga(SC{O}Ph)4]·H2O (2) are fairly stable Despite this, they were stored under nitrogen at 5 ˚C to avoid any decomposition
The representative structures of these two compounds are shown in figure 6.1 and figure 6.2 respectively It is clearly seen that each metal atom is connected to four thiocarboxylate ligands to form a tetrahedral metal center The detail crystallographic data are listed in Table 6.1