Morphology Dependence of Raman Properties of Carbon Nanotube

Jan 11, 2011 - Investigation of phenol electrooxidation in aprotic non-aqueous solvents by using cyclic and normal pulse voltammetry. László Kiss , ...
0 downloads 0 Views 3MB Size
ARTICLE pubs.acs.org/JPCC

Morphology Dependence of Raman Properties of Carbon Nanotube Layers Formed on Nanostructured CeO2 Films Heng Li,†,‡ Andrea Petz,^ Hui Yan,† Jia Cai Nie,†,§ and Sandor Kunsagi-Mate*,‡ †

Department of Physics, Beijing Normal University, 100875, Beijing, People’s Republic of China Department of General and Physical Chemistry and ^Department of Inorganic Chemistry, University of Pecs, H-7624 Pecs, Ifjusag 6, Hungary ‡

ABSTRACT: Multiwalled carbon nanotubes (MWCNTs) were deposited from solution phase onto the nanostructured CeO2 films grown on a sapphire substrate. High-resolution images by scanning electron microscopy showed that MWCNTs laid on the top of CeO2 islands. Raman shifts of these combinatorial materials showed considerable dependence on the morphology of the underlying CeO2 nanostructures. Quantum chemical calculations validated the resonant nanotube vibrations, especially for the stationary longitudinal wave at typical surface morphology, as the origin of this unexpected property of the Raman signal. Considering that the CeO2 surface morphology is fully controlled by the experimental setup of growth, properties of the CNT layers are tunable toward the requirements of practical applications.

1. INTRODUCTION Intrinsic small dimensions and remarkable electronic properties make nanotubes (NTs) promising building blocks for molecular or nanoscale devices, which may overcome fundamental physical and economic limitations of conventional Sibased very large scale integration (VLSI) fabrication techniques. For instance, carbon nanotubes (CNTs) are a sort of fascinating nanomaterials which have attracted wide interest in many areas of science and technology since 1991.1,2 A recent trend in research shows a growing interest in vibration and wave propagation in CNTs due to their various applications in nanoscale devices.3-5 NTs need to be assembled into hierarchical arrays over large-scale areas. As active components (such as transistors or sensing elements), they often need to be interfaced with other device components (such as electrodes). However, at the nanoscale, how to place the NTs at desired locations with targeted shapes, directions, and densities has been one of the longstanding unsolved problems of NT-based molecular devices.6 Considering that NT is a typical one-dimensional (1D) system of which the radius is nm-scaled while the length usually can reach several micrometers, a nanostructured surface can be a fruitful substrate for the fabrication of selective patterns of the nanoscaled NTs.7 To experimentally investigate the issue about the fabrication of selective patterns of the nanoscaled NTs, we fabricated a layer of CNTs on a nanostructured CeO2 surface by liquid-phase epitaxy (LPE) and examined how the interaction between the CNTs and the different CeO2 surfaces affects the vibrational properties of CNTs by micro-Raman spectra and compared the results with the theoretical ones. r 2011 American Chemical Society

2. EXPERIMENTAL METHODS All of the CeO2 films were grown by pulsed laser deposition technique and annealed at high temperature (1050 C) to form self-assembled nanostructures, as described earlier.8,9 X-ray diffraction (XRD) and high-resolution scanning electron microscopy (SEM) were used to examine the crystalline properties and surface morphology, respectively. Hydroxylated multiwalled CNTs (MWCNT-OH, purity > 95%) with diameters of 8-10 nm and lengths over 5 μm were purchased from Guangzhou Heji Trade Co. (China) and were used for sample preparation. The powder was dissolved in MeOH, and LPE was then applied for deposition of CNTs onto the CeO2 surface. Accordingly, the CNT-MeOH saturated solution was dropped onto the surface of the nanostructured CeO2 films, and during evaporation of MeOH, the CNTs adsorbed onto the CeO2 surface. The amount of the CNTs deposited was controlled by the different number of drops. We used 20 drops for every sample, each drop 10 μL, and the interval between two drops was 15 min. All of the process was carried out at room temperature. Micro-Raman investigations were performed by a LabRam 600 spectrometer (Jobin-Yvon). The optimal confocal spatial and depth discrimination was down to 1 μm. The sample surface was illuminated with 632.8 nm monochromatic light with less than 10 mW/cm2 power density. Received: August 24, 2010 Revised: October 17, 2010 Published: January 11, 2011 1480

dx.doi.org/10.1021/jp108023f | J. Phys. Chem. C 2011, 115, 1480–1483

The Journal of Physical Chemistry C

ARTICLE

Table 1. Nominal Thickness (t), Mean Thickness (h), Mean Distance (d), and Corresponding Raman Shift (G Peak)a sample name

a

nominal thickness t (nm)

mean island height h (nm)

mean island distance d (nm)

Raman shift (cm-1)

sapphire

0

0

¥ (no film)

CeO2-1

6.79

8.82

45.43 ( 12.29(27.2%)

1586.8

CeO2-2

7.18

9.33

36.82 ( 5.95(16.1%)

1586.3

CeO2-3

7.49

9.81

49.40 ( 7.74 (15.7%)

1581.7

CeO2-4

7.65

11.44

53.79 ( 8.71(16.2%)

1580.1

CeO2-5

8.22

12.14

48.47 ( 7.01(14.5%)

1582.4

CeO2-6

8.80

16.62

77.36 ( 32.31(41.8%)

1585.3

CeO2-7 CeO2-8

9.39 13.48

16.82 24.07

72.65 ( 16.62(22.9%) 44.79 ( 13.65(30.5%)

1586.1 1586.5

CeO2-9

38.95

37.69

0 (connected)

1586.0

1586.2

The mean distance of the CeO2 islands was obtained by a statistical analysis of over 100 islands of the SEM images. The standard deviation is given in brackets. The nominal thickness of the as-grown CeO2 films, t, was determined by the number of laser shots and the calibrated growth rate. The mean island heights (nanonet thickness) of the CeO2 films were calculated from XRD results.

3. MODELING METHOD For simulation of the long-wave vibration of MWCNTs, we applied our setup previously tested for studying the vibration and adsorption behaviors of single-walled carbon nanotubes (SWCNTs).2,10 To determine the appropriate multiwalled structure, the following properties of MWCNTs were considered: the diameter of the SWCNT was about 0.75-3 nm;1 it can be considered as the core of the MWCNT. The number of MWCNT’s layers usually ranged from 2 to 50.1 The distance between two neighboring layers was 0.34 nm, which is equal to the graphite interlayer distance.11,12 The diameter and length could range from 2 to 100 nm and 0.1 to 50 μm, respectively.13,14 Due to the similarity in the stationary long-wave vibration of armchair and zigzag nanotubes, in this work, only the armchair multiwalled structures were considered; for the simulation, fragments of the (8,0) nanotube with 155 nm length (∼12800 atoms) were considered first according to our previous simulations.8 Starting from this SWCNT, calculations were performed for up to six layers. The number of the layers was limited by the computational effort. Then, as the second group of calculations, the second layer of the structure calculated in the first step was used as the core tube with a diameter of 0.968 nm. As a result, for example, the outer diameter of the SWCNT in the first series varied between 0.628 and 2.328 nm, while the number of layers varied from 1 to 6. This procedure was repeated seven times. The vibrational dynamics of the nanotube were investigated according to our previous work.10 Molecular dynamics (MD) simulations were performed by the AMBER force field. To find an appropriate initial condition for MD, the “heating” algorithm implemented in HyperChem code15 was used. This procedure smoothly heats up the molecular system from 0 K to the temperature T at which MD simulations are desired to perform. The starting geometry for this heating phase is a static initial structure. The temperature step and the time step in the heating phase were set to 2 K and 0.1 fs, respectively. After an additional 100 fs equilibration period at the given temperature, the MD simulations were started; 5000 data points during a 5 ps time interval and with 0.1 fs resolution were collected and used for further evaluation to calculate the characteristic data of vibrational dynamics. Simulations were performed at 293.16 K. 4. RESULTS Using pulsed laser deposition (PLD) and ex situ high-temperature O2 annealing,8 a self-assembly process induced a surface

reorganization in CeO2 deposited on R-cut sapphire substrates. When the as-grown CeO2 film was thin, a highly ordered phase was formed as quasi-2D islands of CeO2 with most of the sapphire surface exposed. With an increase of the CeO2 film thickness, the isolated CeO2 islands started to coalesce into large rectangular islands and connected with each other, forming an ordered quasi-2D nanonet with almost uniform nanoscale thickness. With further increase of the as-grown CeO2 film thickness, the connected nanonet ripened into a nanohole pattern. When the CeO2 film thickness exceeded a critical value, an atomically flat surface was formed by the reorganization. The nominal thickness of the as-grown CeO2 films, t, was determined by the number of laser shots and the calibrated growth rate. A group of CeO2 samples with different nominal thicknesses were chosen for further investigations (see Table 1). XRD patterns indicate that both the as-grown and the annealed CeO2 films were epitaxial with single-phase character and (00l) orientation (data not shown). After annealing, the full width at half-maximum (βc) of the most intense peak CeO2(002) in the XRD pattern was routinely used to determine the mean height h (out-of-plane grain size) of CeO2 islands using the Scherrer formula 0:9λ h ¼ cos θ ð1Þ βc According to the above description, we prepared several samples where the CNTs condensed on the CeO2 layers, which however possessed different morphologies (see, e.g., Figure 1). It can be clearly seen from Figure 1 that CNTs are condensed on the top of CeO2 islands (oriented parallel to the surface of the substrate) in both cases of the CeO2 nanonet (Figure 1a) and nanohole (Figure 1b). The mean value of the island distances, namely, the distance between the edges of two adjacent islands (see Figure 1), was defined as d. The mean distance between nanoislands was obtained by a statistical analysis of over 100 spots of the SEM image for each sample. The measured distances showed Gauss distribution half-widths that were usually smaller than 16% and larger than 30% for nanonet and nanohole patterns, respectively. The standard deviation is given in brackets (see Table 1) and also shown by the error bars in Figure 2. It is obviously necessary to examine how the morphology of the CeO2 layer affects the electronic structure of the formed combinatorial material. Raman studies were then applied for these investigations. Using the micro-Raman technique (Jobin-Yvon/LabRam600), all of the CeO2 samples showed detailed Raman spectra. The 1481

dx.doi.org/10.1021/jp108023f |J. Phys. Chem. C 2011, 115, 1480–1483

The Journal of Physical Chemistry C

ARTICLE

Figure 3. Sketch map of longitudinal long-wave vibration of a nanotube.

Figure 4. Wavelength of standing waves obtained for SWCNTs possessing different numbers of nanotube layers. The variation ranges of the outer diameters of CNTs are indicated.

Figure 1. FE-SEM images of CNTs deposited onto the nanostructured CeO2 surface for samples of two typical kinds of morphology, nanonet (a) and nanohole (b) patterns, respectively. The red arrows indicate the CNTs condensed on top of the CeO2 islands. The mean island distance d was defined as shown in both (a) and (b).

Figure 2. The position of the G peak of CNTs in micro-Raman spectra versus the mean distance between CeO2 nanoislands (standard deviation shown by the error bars). Inset: Original micro-Raman data of the CNTs’ G peak (indicated by arrows) obtained from five typical CeO2 samples.

difference between these samples is the different mean thickness (i.e., mean island height) of the CeO2 nanostructures (see Table 1). The micro-Raman spectra of the MWCNTs showed several characteristic peaks located at around 1330 (D peak), 1580 (G peak), 1370, and 1400 cm-1 (substrate peaks). The G and D

peaks are attributed to the C-C sp2 interaction (E2g) and the structural disorder in graphite material, respectively.16,17 Evaluating the Raman spectra, significant difference between the Raman shift of the G peaks associated with the different samples was observed (Figure 2 and Table 1). Generally, when on a bare sapphire substrate (namely, there is not any CeO2 film) or when the nominal thickness of the as-grown CeO2 is very large (e.g., 38.95 nm, CeO2-9) (there is not any CeO2 nanostructure), the G peak of the CNTs is nearly the same. However, when the mean thickness of CeO2 reaches a specific value (∼7.65 nm, CeO2-4), the Raman shift will show a sharp decrease (see also Figure 2 inset). Although this decrease is rather small, about 7 cm-1, it is crucial if we consider that the experimental error is within 0.4 cm-1. To ensure that this sharp down shift of the G peak is true, the measuring process was repeated as follows: for each sample, at least three regions were measured; especially for four samples CeO2-3,4,5,6, where the down shift of the G peak happened, at least five regions were examined. In different regions on each sample, the amount of CNTs that were examined is not always the same, resulting in different G peak intensity. However, for the same sample, the difference in the Raman shift between different locations is within 0.4 cm-1. Considering the experimental error is about 0.4 cm-1, the scattering of the data points (the variation of the G peak shift) in Figure 2 is within the experimental resolution, that is, the G peak position is nearly the same for the same sample. Molecular dynamics simulations highlighted the following internal moving of the nanotube: the cross section of the nanotube can be represented always by such an ellipsoid of which the main axes show periodic changes with time.10 The shape of the cross section of the nanotube is changed not only in the time domain but also along the long axis of the nanotube. The distance between the nearest same cross sections, L, was recorded for all 1482

dx.doi.org/10.1021/jp108023f |J. Phys. Chem. C 2011, 115, 1480–1483

The Journal of Physical Chemistry C calculated models (Figure 3). This distance is equal to the wavelength of standing waves when constraint forces fix the nanotubes at the same distance. Figure 4 summarizes these results.

5. DISCUSSION To explain this down shift of the G peak, we first consider the strain-induced effect. The residual strain, which is induced by the nanotube bending, is usually considered to affect the band structure and hence the optical properties too. Recently, Ren’s results18 showed a ring-diameter-dependent Raman G band splitting phenomenon in carbon nanotube ring structures. However, obvious splitting or shifting will happen only when the diameter of the CNT ring structures is smaller than 200 nm. In our case, as we can see, for example, from Figure 1b, such a ring diameter is usually over 600 nm; thus, the bending-induced G peak shift can be neglected. In previous theoretical work,19 vibrational wavepackets for the radial breathing mode (RBM) and the G mode were clearly observed, in particular, their anharmonic coupling, resulting in a frequency modulation of the G mode by the RBM. Quantum chemical modeling20 shows that this effect is due to a corrugation of the CNT surface, leading to a coupling between longitudinal and radial vibrations. Using this idea, the significant downshift of the G peak in our measurements probably can be explained as follows. By gradually changing the thickness of the CeO2 film, the mean distance (d) between CeO2 nanoislands was also changed. However, it is not always true that this distance will decrease when increasing the nominal thickness of the CeO2 film. This is because with further increase in t, the CeO2 nanonet grows up, and some nanoholes may form, of which the size is sometimes even bigger than the gap between disconnected CeO2 islands. Sometimes, this distance would reach some specific values that are close to the period of the corrugation of the CNT surface; therefore, the G peak might show a resonance, for example, a downshift. Actually, the adsorption of the CNT onto any surface disturbs the long-wave vibration and increases the coupling between longitudinal and radial vibrations. Therefore, the shift of the G peak reduces only when stationary waves are formed (Figure 2). The theoretical results (Figure 4) validate this explanation. In detail, the wavelength of the standing waves increases with an increase of the number of the nanotube layers and tends to converge to a well-determined value with further increase of the nanotube diameters. Namely, this increase can be negligible above a diameter of 10 nm. These calculations highlight the presence of characteristic long-wave vibration (standing wave) of MWCNTs with about a 50-70 nm wavelength. This standing wave remained undisturbed in the nanotube when it was fixed onto the CeO2 surfaces by the islands, of which the distance was equal to the wavelength of the standing wave. If the distance between the islands differs from the above values, the long-wave vibration couples with other vibrational modes of the nanotubes. In other words, the site of the G peak in the Raman spectra can be modified when this distance reaches a specific value (e.g., 55 nm; see also Figures 1 and 4). This finding suggests that the interaction between CNTs and the nanostructured CeO2 film is significantly affected by the surface morphology of the CeO2 substrate. It has a very important practical consequence; the vibrational property of the NTs can be tuned by modifying the surface morphology of the substrate; in detail, the coupling between longitudinal and radial vibrations can be highly affected by changing the mean distance between the nanostructures on the surface of the substrate, resulting in a significant decrease of, for example, the G peak of

ARTICLE

the CNT, which would be crucial when fabricating NT layerbased nanodevices.

6. CONCLUSION New MWCNTs-CeO2-sapphire combinatorial materials were prepared via LPE. SEM images validated that the adsorbed CNTs were parallel to the substrate surface and laying on the CeO2 nanostructures. Analysis on the Raman properties highlighted that the interaction between CNTs and CeO2 nanostructures significantly influenced the C-C stretch of the adsorbed CNT layer. The specific Raman shift can be controlled by the surface morphology of the underlying CeO2. This finding offers potential application of these layers in wide industrial scale. ’ AUTHOR INFORMATION Corresponding Author

*E-mail: [email protected]. Notes §

E-mail: [email protected].

’ ACKNOWLEDGMENT The authors acknowledge Dr. Gerhard Frank (Institute for Microcharacterization, University of Erlangen, Germany) for the micro-Raman measurements. This work was supported by the Chinese-Hungarian Intergovernmental S&T Cooperation Programme (Project No. CH-4-32/2008: CN-54/2007). This work was also supported partly by the National Natural Science Foundation of China (Grant Nos. 50772015 and 10974019) and the Special Program for the Ph.D. Subjects in the University of the Ministry of Education of China (Grant No. 200800270004). ’ REFERENCES (1) Iijima, S. Nature 1991, 354, 56.  cs, P.; Kollar, L.; Kunsagi-Mate, S. Fullerenes, (2) Peles-Lemli, B.; A Nanotubes, Carbon Nanostruct. 2008, 16 (4), 247. (3) Peles-Lemli, B.; Matisz, G.; Kelterer, A.-M.; Fabian, W. M. F.; Kunsagi-Mate, S. J. Phys. Chem. C 2010, 114, 5898–5905. (4) Li, C.; Chou, T. W. Phys. Rev. B 2006, 73, 245407.  lvarez, R.; Cabrillo, C. Phys. Rev. B 2006, 73, 075425. (5) Chico, L.; Perez-A (6) Dai, L. M.; Patil, A.; Gong, X. Y.; Guo, Z. X.; Liu, L. Q.; Liu, Y.; Zhu, D. B. ChemPhysChem 2003, 4, 1150. (7) Yan, Y.; Chan-Park, M. B.; Zhang, Q. Small 2007, 3, 24. (8) Nie, J. C.; Yamasaki, H.; Mawatari, Y. Phys. Rev. B 2004, 70, 195421. (9) Nie, J. C.; Hua, Z. Y.; Tu, Q. Y.; Yamasaki, H. Physica C 2007, 1353, 460. (10) Kunsagi-Mate, S.; Nie, J. C. Surf. Sci. 2010, 604, 654. (11) Ebbesen, T. W.; Aiayan, P. M. Nature 1992, 358, 220. (12) Carroll, D. L.; Redlich, P.; Ajayan, P. M. Phys. Rev. Lett. 1997, 78 (14), 2811. (13) Hernadi, K.; Fonseca, A.; Nagy, J. B.; et al. Synth. Met. 1996, 77, 31. (14) Peigney, A.; Laurent, C.; Dobigeon, F.; et al. J. Mater. Res. 1997, 12 (3), 613. (15) HyperChem Professional 7, HyperCube: Gainesville, FL, 2002. (16) Kasuya, A.; Sasaki, Y.; Saito, Y.; Tohji, K.; Nishina, Y. Phys. Rev. Lett. 1997, 78, 4434. (17) Thomsen, C.; Reich, S. Phys. Rev. Lett. 2000, 85, 5214. (18) Ren, Y.; Song, L.; Ma, W. J.; Zhao, Y. C.; Sun, L. F.; Gu, C. Z.; Zhou, W. Y.; Xie, S. S. Phys. Rev. B 2009, 80, 113412. (19) Gambetta, A.; Manzoni, C; Menna, E.; et al. Nat. Phys. 2006, 2, 515. (20) Tretiak, S.; Saxena, A.; Martin, R. L.; Bishop, A. R. Phys. Rev. Lett. 2002, 89, 097402. 1483

dx.doi.org/10.1021/jp108023f |J. Phys. Chem. C 2011, 115, 1480–1483