- Email: [email protected]
carbon nanowall composite films
Diamond & Related Materials 27-28 (2012) 40–44
Contents lists available at SciVerse ScienceDirect
Diamond & Related Materials journal homepage: www.elsevier.com/locate/diamond
Low threshold field emission from nanocrystalline diamond/carbon nanowall composite films☆ C.Y. Cheng, M. Nakashima, K. Teii ⁎ Department of Applied Science for Electronics and Materials, Interdisciplinary Graduate School of Engineering Sciences, Kyushu University, Kasuga, Fukuoka 816-8580, Japan
a r t i c l e
i n f o
Available online 25 May 2012 Keywords: Carbon nanowalls Nanocrystalline diamond Graphene Plasma CVD Undulation Field emission Raman spectroscopy
a b s t r a c t Two methods of substrate scratching pretreatment using diamond powder are employed to control the wall spacing in carbon nanowalls (CNWs). The surface after scraping for undulation has continuous undulant scratches with a number of residual diamond grains exclusively along the scratches, while that after scratching with ultrasonic vibration it shows irregular distributions of residual diamond grains and scratches, depending upon the size of the diamond powder. Nanocrystalline diamond film/CNW composites are obtained with either pretreatment method by microwave plasma-enhanced chemical vapor deposition. With increase of the duration of scratching, the morphology of the deposits changes from CNWs to a film/CNW composite and lastly to CNWs on a film, accompanied by an overall increase in wall spacing. The turn-on field for field emission decreases from 2.1 V/μm without scratching down to 1.2 V/μm with scratching due to suppression of electric field screening between the walls as evidenced by the larger field enhancement factor up to ~ 2700. © 2012 Elsevier B.V. All rights reserved.
1. Introduction An increasing interest in vacuum microelectronic devices such as point electron sources, vacuum diodes, and ultrathin flat panel displays has stimulated research on field emission cold cathodes with the low turn-on field, high current density, and excellent durability and stability over the past few decades. Carbon nanotubes have been identified as particularly suitable for field emitters due to their intrinsic high aspect ratios, compared to the traditional emitters like metal and Si tips designed for the Spindt-type emitters. However, carbon nanotubes often show emission current deterioration due to structural deformation at the nanotube tips, mainly caused by Joule heating under excessively imposed emission current [1–4]. The deformation from thin crystalline nanotube bundles to thick amorphoustype fibers is accountable for decreasing the electrical conductivity and the local field enhancement [3]. Carbon nanowalls (CNWs) consist of two-dimensional networks of almost vertically aligned graphitic walls [5]. The walls consist of stacks of nanographite domains, interconnected with amorphous carbon [6]. Due to the high aspect ratio, high surface area, and high in-plane continuity of the unique wall structures, CNWs are expected as the potential candidate for large-area field emitters free from Joule heating, templates for fabricating other nanomaterials, and electrodes for fuel cell and lithium-ion battery [7–13]. High n-type conductivity ☆ Presented at the Diamond 2011, 22st European Conference on Diamond, DiamondLike Materials, Carbon Nanotubes, and Nitrides, Garmisch-Partenkirchen. ⁎ Corresponding author. Tel./fax: + 81 92 5837097. E-mail address: [email protected] (K. Teii). 0925-9635/$ – see front matter © 2012 Elsevier B.V. All rights reserved. doi:10.1016/j.diamond.2012.05.009
in nitrogen-incorporated n-type CNWs provides wide variations in electronic use [14,15]. Quantum tunneling at a surface is normally facilitated by increasing a local electric field Elocal, which is related to the macroscopic field Emacro applied between the sample (cathode) and the anode by Elocal = βEmacro, where β is the field enhancement factor. The emission turn-on field of CNWs at room temperature has been shown to be below ~ 3 V/μm [7–10], which is as low as that of carbon nanotubes. However, the local field onto sharp edges of CNWs is not well enhanced due to electric field screening between the densely grown walls [9]. In fact, the measured β value was found to be smaller than the aspect ratio of the walls [10]. In the previous work, we demonstrated the formation of nanocrystalline diamond film/CNW composites with a substrate scratching pretreatment using diamond powder by microwave (MW) plasmaenhanced chemical vapor deposition (CVD) [16]. Methane was highly diluted with Ar to increase the flux of C2 radicals, enabling simultaneous growth of the nanocrystalline diamond film and CNWs. The wall spacing was increased due to interception of the wall continuity by the island film growth, and then the field screening was suppressed [16]. However, the wall spacing could not be controlled due to random or irregular distribution of residual scratches and diamond grains introduced by ultrasonic vibration. In particular, we believe that residual diamond grains offer subsequent growth sites for diamond. It is desirable to develop a more controllable method to align residual diamond grains in a certain direction. In this study, we investigate the effect of two different scratching pretreatments on the resulting microstructure of CNW films. The electron emission properties of the CNW films are then studied. Two methods of scratching pretreatment, that is, 1-dimensional undulation
C.Y. Cheng et al. / Diamond & Related Materials 27-28 (2012) 40–44
and 2-dimensional ultrasonic vibration, are compared. The former has originally been developed for heteroepitaxial growth of 3C-SiC on Si substrates [17], while the latter has widely been used for diamond deposition on heterogeneous substrates.
41
cathode consists of the deposits to be tested and the anode consists of an indium–tin-oxide coated glass plate, with a Teflon interelectrode spacer 100 μm thick. The deposits needed no pretreatments to trigger the emission but a higher turn-on field in the first measurement due probably to removal of adsorbates or electronic changes.
2. Experiment 3. Results and discussion The deposition experiments were carried out on 11 × 11 mm2 n-type Si(001) wafer substrates without catalysts in a modified AsTeX MW plasma CVD apparatus with a quartz bell-jar reactor. Two methods of scratching pretreatment were used as follows; (1) for undulation, the substrates were dipped in a hydrofluoric acid solution to remove the native oxide layer, scraped with diamond powder of 0.25 μm in average grain size (Tomei Diamond Co.) diluted with ethanol on an emery paper toward the 110 direction under a loading of approximately 0.1 kg/cm 2, and then rinsed in ethanol and deionized water to remove excess residue, (2) for ultrasonic vibration, the substrates were cleaned in ethanol in an ultrasonic generator, dipped in a hydrofluoric acid solution, scratched with two types of diamond powder of 0.25 μm or 25 μm in average grain size diluted with ethanol in an ultrasonic generator, and rinsed with deionized water for removal of excess residue. The number of scrapings for undulation and the ultrasonic vibration period were varied in the range of 0 to 200 times and 0 to 20 min, respectively. The substrates were then mounted on a Mo holder equipped with a Ta wire heater and a water-cooling system. During deposition, the MW power, total pressure, and total gas flow rate were kept at 800 W, 13.3 kPa, and 200 sccm. A mixture of 86%Ar–10%N2–4%CH4 was used as the source gas. The deposition temperature at the substrate surface was measured with an infrared pyrometer and kept at 1150 ± 20 °C. At temperatures less than 1100 °C, CNWs were hardly deposited. Atomic force microscopy (AFM), scanning electron microscopy (SEM), and Raman spectroscopy with 532 nm of visible laser excitation were used for structural characterization of the deposits. AFM images were captured in tapping mode at room temperature in atmosphere by using Si tip probes with the radius of curvature of 10 nm. Field emission characteristics were measured at room temperature in high vacuum (10 − 4 Pa) for a parallel-plate configuration, where the
AFM images of the Si substrate surfaces after scratching pretreatment (before deposition) are shown in Fig. 1(a)–(d). The pristine Si surface was very smooth with the root-mean square (rms) roughness of about 0.1 nm. The surface after 100 times of scraping shows continuous undulant scratches toward the 110 direction [Fig. 1(a)]. The ridge to ridge width and the ridge to valley height of the scratches are in the range of tens to hundreds of nanometers and several tens of nanometers, respectively. Residual diamond grains of ~ 0.25 μm are mostly aggregated into larger grains of several hundreds of nanometers. A large portion of diamond grains are left along the scratches, while the rest of them are still dispersed irregularly over the surface. The surface after 200 times of scraping shows a larger ridge to valley height of the scratches as shown by the stronger color contrast, with aggregated grains exclusively along the scratches as well [Fig. 1(b)]. The rms roughness of about 41 nm for Fig. 1(b) is indeed larger than about 22 nm for Fig. 1(a). The surface after 10 min of vibration with small diamond grains (~0.25 μm) shows many aggregated grains of several hundreds of nanometers [Fig. 1(c)]. The surface after 10 min of vibration with large diamond grains (~25 μm) shows only a small number of residual grains of sub-microns to a few microns, while the number of submicron-sized scratches corresponding to the dark area increases [Fig. 1(d)]. Top-view SEM images of the deposits obtained with and without scratching pretreatment are shown in Fig. 2(a)–(f). Without scratching pretreatment, the deposit consists of simple CNWs [Fig. 2(a)]. The walls are wavy and exhibit a high in-plane continuity and a multi-tier indepth structure. The thickness of the walls is as small as 10 nm, while the spacing between the walls ranges from tens to a few hundreds of nanometers. The walls are almost vertically aligned and the top surface
(a)
(b)
(c)
(d)
Fig. 1. AFM images of the Si substrate surfaces after scratching pretreatment; undulation with (a) 100 times and (b) 200 times of scraping, ultrasonic vibration with (c) small and (d) large diamond grains for 10 min.
42
C.Y. Cheng et al. / Diamond & Related Materials 27-28 (2012) 40–44
of the walls occupies almost the same altitude with a deviation of ±0.5 μm as shown in the cross sectional image [Fig. 2(b)]. The height of the walls reaches ~20 μm for 20 min of deposition. The dense wall growth would force the walls to align vertically, similar to the socalled “crowding effects” in vertical growth of carbon nanotubes [18]. With scratching pretreatment, the deposit becomes a composite of CNWs and an underlying fine-grained film. The underlying film consists of graphitic amorphous carbon and diamond nanocrystals, as shown previously by transmission electron microscopy [16], and contains small segments of walls depending on the scratching conditions. The thickness of the walls tends to increase up to several tens of nanometers. For the deposits with 100 times of scraping [Fig. 2(c)] or 10 min of vibration and small diamond grains [Fig. 2(e)], the underlying film regions cover the space between the walls and, hence, the wall spacing is favorably widened to several hundreds of nanometers by interception of the lateral wall continuity. The underlying film limits the wall connection and seemingly reinforces the walls to align vertically by the crowding effect [18]. On the other hand, for the deposits with 200 times of scraping [Fig. 2(d)] or with 10 min of vibration and large diamond grains [Fig. 2(f)], the underlying film is not well visible because the walls are grown over the film. A large portion of the walls are inclined as they are no longer reinforced by the underlying film, thus only a small portion of the walls are aligned vertically. Originally, we expected the undulation method to grow the walls in the direction of aligned scratches as diamond would be grown from residual diamond grains along the scratches. However, as shown in the above results, no alignment of the walls was observed. The diamond-containing films
were grown uniformly over the surface, not selectively along the scratches. This could be due to the presence of residual diamond grains dispersed irregularly. Raman spectra of the deposits obtained with and without scratching pretreatment are shown in Fig. 3. The spectra consist mainly of two peaks: the G band peak at around 1580 cm− 1 from the bond stretching of all pairs of sp2 atoms in both rings and chains, and the D band peak at around 1350 cm− 1 from the breathing modes of sp2 atoms in rings and it is disorder-activated [19]. The G band is accompanied by a shoulder peak at around 1620 cm− 1, which has been attributed to the D′ band of some disordered graphitic carbons [20]. The D′ band arises from the highest frequency feature in the density of states, which is forbidden under defect-free conditions [21]. The diamond peak normally at around 1332 cm− 1 is not observed due to the small amount and size of diamond. The intensity ratio of the D band to the G band (ID/IG) measures inplane correlation length of the sp2 cluster (La): the ID/IG ratio increases moving from graphite to nanocrystalline graphite, ID/IG∝1/La [22], while it decreases moving from amorphous carbon to tetrahedralamorphous carbon, ID/IG∝La [19]. Its dependence upon La shows a transition at La of approximately 20 Å [19]. The ID/IG ratio for CNWs is considered to follow the former relation ID/IG∝1/La [22], because of their highly graphitic character. The full width at half maximum of the G peak [FWHM(G)] is a measure of structural disorder of the sp2 phase: the larger FWHM(G) indicates the higher disorder [19]. A list of the ID/IG ratio and FWHM (G) is given as the inset in Fig. 3. Basically, the ID/IG ratio and FWHM(G) are smaller than those of simple CNWs
(a)
(d)
(b)
(e)
(c)
(f)
Fig. 2. SEM images of the deposits with and without scratching pretreatment; [(a) and (b)] without scratching, undulation with (c) 100 times and (d) 200 times of scraping, ultrasonic vibration with (e) small and (f) large diamond grains for 10 min.
Intensity (arb. units)
C.Y. Cheng et al. / Diamond & Related Materials 27-28 (2012) 40–44
ID /IG
FWHM(G) (cm-1)
(a)
0.32
29
(b)
0.46
38
(c)
0.30
25
(d)
0.48
33
(e)
0.35
32
43
(a)
G
D (a) (b) (c)
(d)
(b)
(e) 1100
1300
1500
1700
Raman shift (cm-1) Fig. 3. Raman spectra of the deposits with and without scratching pretreatment; (a) without scratching, undulation with (b) 100 times and (c) 200 times of scraping, ultrasonic vibration with (d) small and (e) large diamond grains for 10 min. The inset shows the ID/IG ratio and FWHM (G).
reported by other groups [8,20]. This suggests larger sp 2 cluster sizes and higher sp 2 order of our deposits. Both the ID/IG ratio and FWHM(G) for the spectra (b) and (d) are similar and the largest among all, indicating that these two deposits contain more disordered sp 2 phase. This is consistent with the SEM images [Fig. 2(c) and (e)], where the two deposits have similar morphologies with a larger coverage of the underlying film. Field emission characteristics of the deposits obtained with and without scratching pretreatment are shown in Fig. 4(a). The turn-on field for the onset of emission is measured at a current density of 10 nA/cm 2 with a reproducibility of ±0.4 V/μm. Without scratching pretreatment, the turn-on field is 2.1 V/μm and the current density reaches 1 × 10 − 5 A/cm 2 at 3.0 V/μm. With scratching pretreatment, the turn-on field decreases down to 1.2 V/μm, and the current density increases to more than 1 × 10 − 4 A/cm 2 at 3.0 V/μm, depending a little on the method of scratching. Thus the emission performance of the composite films is higher than that of simple CNWs. The Fowler– Nordheim (FN) plots [23] corresponding to Fig. 4(a) are shown in Fig. 4(b). They are composed of two straight lines, indicating that the emission is apparently governed by a metallic-type FN tunneling [23]. The relationship among the slope of the FN plot (Δ), the effective potential barrier height (Φ), and β is expressed as Δ = −6840Φ 3/2/β, where Δ and Φ are in V/μm and eV, respectively. The local field is enhanced by β compared to the macroscopic applied field. In Fig. 4(b), Δ in the low field region is steeper than that in the high field region. A slope decrease in the high field region may be referred to as depletion of available trapped electrons to be emitted [24–26]. A list of the turn-on field, and Δ and β in the low field region derived from Fig. 4(a) and (b) is given in Table 1. It is believed that field emission occurs at localized sites rather than uniformly over the surface. The field lines focus exclusively onto the sharp wall edges since the wall edges occupy the higher location than the underlying film. Here we used Φ = 5.0 eV for graphite to calculate β on the assumption that electrons are emitted from defect states at the Fermi level of the walls, such as dangling bond, π bonding, and π* antibonding states incorporated with nitrogen [10,15]. If the emission occurs above the Fermi level, similar to nanocrystalline diamond
Fig. 4. (a) Field emission characteristics of the deposits with and without scratching pretreatment. (b) The Fowler–Nordheim (FN) plots corresponding to (a). I and E are the emission current and the applied field, respectively.
film [27], the value of Φ should be lower. In Table 1, the higher the value of β becomes, the lower the turn-on field is. This means that the emission is governed by field enhancement. The value of β should ideally correspond to the aspect ratio of the walls. The measured β = 1250 for simple CNWs is smaller than the aspect ratio of ~ 2000 measured from the SEM images [Fig. 2(a) and (b)] due probably to the field screening between the walls. In contrast, the measured β = 1700–2700 for the composite films is larger than that for simple CNWs and several times larger than the aspect ratios of several hundreds to around 1000 depending upon the growth condition. Thus the field screening is suppressed for any of the composite films. The turn-on field is plotted as a function of number of scrapings for undulation or ultrasonic vibration period in Fig. 5. For undulation, the
Table 1 Turn-on field, and Δ and β in the low field region. β was obtained by assuming Φ = 5.0 eV for graphite. Pretreatment
Turn-on field (V/μm)
Δ (V/μm)
β
No scratching Undulation (100 times) Undulation (200 times) Vibration (small grains, 10 min) Vibration (large grains, 10 min)
2.1 1.4 1.4 1.6 1.2
− 61 − 45 − 44 − 41 − 28
1250 1710 1720 1860 2700
44
C.Y. Cheng et al. / Diamond & Related Materials 27-28 (2012) 40–44
Fig. 5. Turn-on field as a function of number of scrapings for undulation [(i), top axis] or ultrasonic vibration period [(ii) small and (iii) large diamond grains, bottom axis].
continuous undulant scratches with a number of residual diamond grains exclusively along the scratches, accompanied by a large increase in surface roughness. The ultrasonic vibration treatment with small diamond grains left a number of diamond grains, while that with large diamond grains left only a small number of diamond grains. The morphology of the deposits was found to depend more on the duration of the scratching method and less on the scratching method itself. With increase of the duration of scratching, the deposits exhibited a morphological change from CNWs to a film/CNW composite and finally to CNWs on a film, accompanied by an increase in wall spacing. The emission turn-on field decreased down to 1.2–1.4 V/μm and the current density increased above 1 × 10− 4 A/cm 2 with either pretreatment method. The emission was enhanced due to suppression of the field screening as shown by a large increase in the field enhancement factor β. The results confirm that the emission performance is governed by the field enhancement and can be controlled by the wall spacing. Scratching treatments do not directly affect field emission. Rather, they change the structure and orientation of the resulting deposits, which consequently affect field emission. References
turn-on field shows a plain decrease from 2.1 V/μm without scratching to 1.6 V/μm with 10 times of scraping and is saturated at 1.4 V/μm for a further increase in the number of scrapings. For vibration, the turn-on field decreases monotonically from 2.1 V/μm without scratching to 1.2 V/μm for a further increase in vibration period. The slope of the plot with large diamond grains is apparently steeper than that with small diamond grains. As shown in the SEM images in Fig. 2, the wall spacing tends to increase with the increase of the duration of scratching, while the morphology changes from CNWs to a film/CNW composite and lastly to CNWs on a film. The possible mechanism is considered as follows. Without scratching, simple CNWs are formed on bare substrates. The wall spacing cannot be controlled as the dense wall growth would force the walls to align vertically. With scratching, the increasing duration of scratching leaves a larger number of diamond grains on the substrate, which in turn causes a higher nucleation density and a faster coalescence of the island films. For a short duration of scratching, the walls are generated from the boundary regions of the islands as the nucleation density is not high. In this case, the wall spacing is increased as the lateral wall connection is intercepted by the islands. For a long duration of scratching, the walls are grown on the underlying film as the nucleation density is too high to allow the walls to grow at the boundaries of the islands. In this case, the effective wall spacing for field enhancement is increased because only a smaller portion of the walls are aligned vertically. Scratching treatments do not directly affect field emission. Rather, they change the structure and orientation of the resulting deposits, thus affecting field emission. 4. Conclusions The morphology and field emission properties of nanocrystalline diamond/CNW composite films were examined with two methods of scratching pretreatment. The undulation treatment produced
[1] S.T. Purcell, P. Vincent, C. Journet, Vu Thien Binh, Phys. Rev. Lett. 88 (2002) 105502. [2] Z.L. Wang, R.P. Gao, W.A. De Heer, P. Poncharal, Appl. Phys. Lett. 80 (2002) 856. [3] J.H. Lee, S.H. Lee, W.S. Kim, H.J. Lee, J.N. Heo, T.W. Jeong, C.W. Baik, S.H. Park, SeGi Yu, J.B. Park, Y.W. Jin, J.M. Kim, H.J. Lee, J.W. Moon, M.A. Yoo, J.W. Nam, S.H. Cho, J.S. Ha, T.I. Yoon, J.H. Park, D.H. Choe, Appl. Phys. Lett. 89 (2006) 253115. [4] C.-W. Baik, J. Lee, J.H. Choi, I. Jung, H.R. Choi, Y.W. Jin, J.M. Kim, Appl. Phys. Lett. 96 (2010) 023105. [5] Y.H. Wu, P.W. Qiao, T.C. Chong, Z.X. Shen, Adv. Mater. (Weinheim, Ger.) 14 (2002) 64. [6] K. Kobayashi, M. Tanimura, H. Nakai, A. Yoshimura, H. Yoshimura, K. Kojima, M. Tachibana, J. Appl. Phys. 101 (2007) 094306. [7] Y. Wu, B. Yang, B. Zong, H. Sun, Z. Shen, Y. Feng, J. Mater. Chem. 14 (2004) 469. [8] A.T.H. Chuang, J. Robertson, B.O. Boskovic, K.K.K. Kozio, Appl. Phys. Lett. 90 (2007) 123107. [9] A. Malesevic, R. Kemps, A. Vanhulsel, M.P. Chowdhury, A. Volodin, C.V. Haesendonck, J. Appl. Phys. 104 (2008) 084301. [10] S. Shimada, K. Teii, M. Nakashima, Diamond Relat. Mater. 19 (2010) 956. [11] L. Giorgi, Th. Dikonimos Makris, R. Giorgi, N. Lisi, E. Salernitano, Sens. Actuators, B 126 (2007) 144. [12] E. Luais, M. Boujtita, A. Gohier, A. Tailleur, S. Casimirius, M.A. Djouadi, A. Granier, P.Y. Tessier, Appl. Phys. Lett. 95 (2009) 014104. [13] K. Mase, H. Kondo, S. Kondo, M. Hori, M. Hiramatsu, H. Kano, Appl. Phys. Lett. 98 (2011) 193108. [14] W. Takeuchi, M. Ura, M. Hiramatsu, Y. Tokuda, H. Kano, M. Hori, Appl. Phys. Lett. 92 (2008) 213103. [15] K. Teii, S. Shimada, M. Nakashima, A.T.H. Chuang, J. Appl. Phys. 106 (2009) 084303. [16] K. Teii, M. Nakashima, Appl. Phys. Lett. 96 (2010) 023112. [17] H. Nagasawa, K. Yagi, T. Kawahara, J. Cryst. Growth 237 (2002) 1244. [18] C.J. Lee, D.W. Kim, T.J. Lee, Y.C. Choi, Y.S. Park, Y.H. Lee, W.B. Choi, N.S. Lee, G.S. Park, J.M. Kim, Chem. Phys. Lett. 312 (1999) 461. [19] A.C. Ferrari, J. Robertson, Phys. Rev. B 61 (2000) 14095. [20] Z.H. Ni, H.M. Fan, Y.P. Feng, Z.X. Shen, B.J. Yang, Y.H. Wu, J. Chem. Phys. 124 (2006) 204703. [21] R.J. Nemanich, S.A. Solin, Phys. Rev. B 20 (1979) 392. [22] F. Tuinstra, J.L. Koenig, J. Chem. Phys. 53 (1970) 1126. [23] R.H. Fowler, L. Nordheim, Proc. R. Soc. Lond. A 119 (1928) 173. [24] G.A.J. Amaratunga, S.R.P. Silva, Appl. Phys. Lett. 68 (1996) 2529. [25] K. Teii, S. Matsumoto, J. Robertson, Appl. Phys. Lett. 92 (2008) 013115. [26] K. Teii, R. Yamao, S. Matsumoto, J. Appl. Phys. 106 (2009) 113706. [27] T. Ikeda, K. Teii, Appl. Phys. Lett. 94 (2009) 143102.
Contents lists available at SciVerse ScienceDirect
Diamond & Related Materials journal homepage: www.elsevier.com/locate/diamond
Low threshold field emission from nanocrystalline diamond/carbon nanowall composite films☆ C.Y. Cheng, M. Nakashima, K. Teii ⁎ Department of Applied Science for Electronics and Materials, Interdisciplinary Graduate School of Engineering Sciences, Kyushu University, Kasuga, Fukuoka 816-8580, Japan
a r t i c l e
i n f o
Available online 25 May 2012 Keywords: Carbon nanowalls Nanocrystalline diamond Graphene Plasma CVD Undulation Field emission Raman spectroscopy
a b s t r a c t Two methods of substrate scratching pretreatment using diamond powder are employed to control the wall spacing in carbon nanowalls (CNWs). The surface after scraping for undulation has continuous undulant scratches with a number of residual diamond grains exclusively along the scratches, while that after scratching with ultrasonic vibration it shows irregular distributions of residual diamond grains and scratches, depending upon the size of the diamond powder. Nanocrystalline diamond film/CNW composites are obtained with either pretreatment method by microwave plasma-enhanced chemical vapor deposition. With increase of the duration of scratching, the morphology of the deposits changes from CNWs to a film/CNW composite and lastly to CNWs on a film, accompanied by an overall increase in wall spacing. The turn-on field for field emission decreases from 2.1 V/μm without scratching down to 1.2 V/μm with scratching due to suppression of electric field screening between the walls as evidenced by the larger field enhancement factor up to ~ 2700. © 2012 Elsevier B.V. All rights reserved.
1. Introduction An increasing interest in vacuum microelectronic devices such as point electron sources, vacuum diodes, and ultrathin flat panel displays has stimulated research on field emission cold cathodes with the low turn-on field, high current density, and excellent durability and stability over the past few decades. Carbon nanotubes have been identified as particularly suitable for field emitters due to their intrinsic high aspect ratios, compared to the traditional emitters like metal and Si tips designed for the Spindt-type emitters. However, carbon nanotubes often show emission current deterioration due to structural deformation at the nanotube tips, mainly caused by Joule heating under excessively imposed emission current [1–4]. The deformation from thin crystalline nanotube bundles to thick amorphoustype fibers is accountable for decreasing the electrical conductivity and the local field enhancement [3]. Carbon nanowalls (CNWs) consist of two-dimensional networks of almost vertically aligned graphitic walls [5]. The walls consist of stacks of nanographite domains, interconnected with amorphous carbon [6]. Due to the high aspect ratio, high surface area, and high in-plane continuity of the unique wall structures, CNWs are expected as the potential candidate for large-area field emitters free from Joule heating, templates for fabricating other nanomaterials, and electrodes for fuel cell and lithium-ion battery [7–13]. High n-type conductivity ☆ Presented at the Diamond 2011, 22st European Conference on Diamond, DiamondLike Materials, Carbon Nanotubes, and Nitrides, Garmisch-Partenkirchen. ⁎ Corresponding author. Tel./fax: + 81 92 5837097. E-mail address: [email protected] (K. Teii). 0925-9635/$ – see front matter © 2012 Elsevier B.V. All rights reserved. doi:10.1016/j.diamond.2012.05.009
in nitrogen-incorporated n-type CNWs provides wide variations in electronic use [14,15]. Quantum tunneling at a surface is normally facilitated by increasing a local electric field Elocal, which is related to the macroscopic field Emacro applied between the sample (cathode) and the anode by Elocal = βEmacro, where β is the field enhancement factor. The emission turn-on field of CNWs at room temperature has been shown to be below ~ 3 V/μm [7–10], which is as low as that of carbon nanotubes. However, the local field onto sharp edges of CNWs is not well enhanced due to electric field screening between the densely grown walls [9]. In fact, the measured β value was found to be smaller than the aspect ratio of the walls [10]. In the previous work, we demonstrated the formation of nanocrystalline diamond film/CNW composites with a substrate scratching pretreatment using diamond powder by microwave (MW) plasmaenhanced chemical vapor deposition (CVD) [16]. Methane was highly diluted with Ar to increase the flux of C2 radicals, enabling simultaneous growth of the nanocrystalline diamond film and CNWs. The wall spacing was increased due to interception of the wall continuity by the island film growth, and then the field screening was suppressed [16]. However, the wall spacing could not be controlled due to random or irregular distribution of residual scratches and diamond grains introduced by ultrasonic vibration. In particular, we believe that residual diamond grains offer subsequent growth sites for diamond. It is desirable to develop a more controllable method to align residual diamond grains in a certain direction. In this study, we investigate the effect of two different scratching pretreatments on the resulting microstructure of CNW films. The electron emission properties of the CNW films are then studied. Two methods of scratching pretreatment, that is, 1-dimensional undulation
C.Y. Cheng et al. / Diamond & Related Materials 27-28 (2012) 40–44
and 2-dimensional ultrasonic vibration, are compared. The former has originally been developed for heteroepitaxial growth of 3C-SiC on Si substrates [17], while the latter has widely been used for diamond deposition on heterogeneous substrates.
41
cathode consists of the deposits to be tested and the anode consists of an indium–tin-oxide coated glass plate, with a Teflon interelectrode spacer 100 μm thick. The deposits needed no pretreatments to trigger the emission but a higher turn-on field in the first measurement due probably to removal of adsorbates or electronic changes.
2. Experiment 3. Results and discussion The deposition experiments were carried out on 11 × 11 mm2 n-type Si(001) wafer substrates without catalysts in a modified AsTeX MW plasma CVD apparatus with a quartz bell-jar reactor. Two methods of scratching pretreatment were used as follows; (1) for undulation, the substrates were dipped in a hydrofluoric acid solution to remove the native oxide layer, scraped with diamond powder of 0.25 μm in average grain size (Tomei Diamond Co.) diluted with ethanol on an emery paper toward the 110 direction under a loading of approximately 0.1 kg/cm 2, and then rinsed in ethanol and deionized water to remove excess residue, (2) for ultrasonic vibration, the substrates were cleaned in ethanol in an ultrasonic generator, dipped in a hydrofluoric acid solution, scratched with two types of diamond powder of 0.25 μm or 25 μm in average grain size diluted with ethanol in an ultrasonic generator, and rinsed with deionized water for removal of excess residue. The number of scrapings for undulation and the ultrasonic vibration period were varied in the range of 0 to 200 times and 0 to 20 min, respectively. The substrates were then mounted on a Mo holder equipped with a Ta wire heater and a water-cooling system. During deposition, the MW power, total pressure, and total gas flow rate were kept at 800 W, 13.3 kPa, and 200 sccm. A mixture of 86%Ar–10%N2–4%CH4 was used as the source gas. The deposition temperature at the substrate surface was measured with an infrared pyrometer and kept at 1150 ± 20 °C. At temperatures less than 1100 °C, CNWs were hardly deposited. Atomic force microscopy (AFM), scanning electron microscopy (SEM), and Raman spectroscopy with 532 nm of visible laser excitation were used for structural characterization of the deposits. AFM images were captured in tapping mode at room temperature in atmosphere by using Si tip probes with the radius of curvature of 10 nm. Field emission characteristics were measured at room temperature in high vacuum (10 − 4 Pa) for a parallel-plate configuration, where the
AFM images of the Si substrate surfaces after scratching pretreatment (before deposition) are shown in Fig. 1(a)–(d). The pristine Si surface was very smooth with the root-mean square (rms) roughness of about 0.1 nm. The surface after 100 times of scraping shows continuous undulant scratches toward the 110 direction [Fig. 1(a)]. The ridge to ridge width and the ridge to valley height of the scratches are in the range of tens to hundreds of nanometers and several tens of nanometers, respectively. Residual diamond grains of ~ 0.25 μm are mostly aggregated into larger grains of several hundreds of nanometers. A large portion of diamond grains are left along the scratches, while the rest of them are still dispersed irregularly over the surface. The surface after 200 times of scraping shows a larger ridge to valley height of the scratches as shown by the stronger color contrast, with aggregated grains exclusively along the scratches as well [Fig. 1(b)]. The rms roughness of about 41 nm for Fig. 1(b) is indeed larger than about 22 nm for Fig. 1(a). The surface after 10 min of vibration with small diamond grains (~0.25 μm) shows many aggregated grains of several hundreds of nanometers [Fig. 1(c)]. The surface after 10 min of vibration with large diamond grains (~25 μm) shows only a small number of residual grains of sub-microns to a few microns, while the number of submicron-sized scratches corresponding to the dark area increases [Fig. 1(d)]. Top-view SEM images of the deposits obtained with and without scratching pretreatment are shown in Fig. 2(a)–(f). Without scratching pretreatment, the deposit consists of simple CNWs [Fig. 2(a)]. The walls are wavy and exhibit a high in-plane continuity and a multi-tier indepth structure. The thickness of the walls is as small as 10 nm, while the spacing between the walls ranges from tens to a few hundreds of nanometers. The walls are almost vertically aligned and the top surface
(a)
(b)
(c)
(d)
Fig. 1. AFM images of the Si substrate surfaces after scratching pretreatment; undulation with (a) 100 times and (b) 200 times of scraping, ultrasonic vibration with (c) small and (d) large diamond grains for 10 min.
42
C.Y. Cheng et al. / Diamond & Related Materials 27-28 (2012) 40–44
of the walls occupies almost the same altitude with a deviation of ±0.5 μm as shown in the cross sectional image [Fig. 2(b)]. The height of the walls reaches ~20 μm for 20 min of deposition. The dense wall growth would force the walls to align vertically, similar to the socalled “crowding effects” in vertical growth of carbon nanotubes [18]. With scratching pretreatment, the deposit becomes a composite of CNWs and an underlying fine-grained film. The underlying film consists of graphitic amorphous carbon and diamond nanocrystals, as shown previously by transmission electron microscopy [16], and contains small segments of walls depending on the scratching conditions. The thickness of the walls tends to increase up to several tens of nanometers. For the deposits with 100 times of scraping [Fig. 2(c)] or 10 min of vibration and small diamond grains [Fig. 2(e)], the underlying film regions cover the space between the walls and, hence, the wall spacing is favorably widened to several hundreds of nanometers by interception of the lateral wall continuity. The underlying film limits the wall connection and seemingly reinforces the walls to align vertically by the crowding effect [18]. On the other hand, for the deposits with 200 times of scraping [Fig. 2(d)] or with 10 min of vibration and large diamond grains [Fig. 2(f)], the underlying film is not well visible because the walls are grown over the film. A large portion of the walls are inclined as they are no longer reinforced by the underlying film, thus only a small portion of the walls are aligned vertically. Originally, we expected the undulation method to grow the walls in the direction of aligned scratches as diamond would be grown from residual diamond grains along the scratches. However, as shown in the above results, no alignment of the walls was observed. The diamond-containing films
were grown uniformly over the surface, not selectively along the scratches. This could be due to the presence of residual diamond grains dispersed irregularly. Raman spectra of the deposits obtained with and without scratching pretreatment are shown in Fig. 3. The spectra consist mainly of two peaks: the G band peak at around 1580 cm− 1 from the bond stretching of all pairs of sp2 atoms in both rings and chains, and the D band peak at around 1350 cm− 1 from the breathing modes of sp2 atoms in rings and it is disorder-activated [19]. The G band is accompanied by a shoulder peak at around 1620 cm− 1, which has been attributed to the D′ band of some disordered graphitic carbons [20]. The D′ band arises from the highest frequency feature in the density of states, which is forbidden under defect-free conditions [21]. The diamond peak normally at around 1332 cm− 1 is not observed due to the small amount and size of diamond. The intensity ratio of the D band to the G band (ID/IG) measures inplane correlation length of the sp2 cluster (La): the ID/IG ratio increases moving from graphite to nanocrystalline graphite, ID/IG∝1/La [22], while it decreases moving from amorphous carbon to tetrahedralamorphous carbon, ID/IG∝La [19]. Its dependence upon La shows a transition at La of approximately 20 Å [19]. The ID/IG ratio for CNWs is considered to follow the former relation ID/IG∝1/La [22], because of their highly graphitic character. The full width at half maximum of the G peak [FWHM(G)] is a measure of structural disorder of the sp2 phase: the larger FWHM(G) indicates the higher disorder [19]. A list of the ID/IG ratio and FWHM (G) is given as the inset in Fig. 3. Basically, the ID/IG ratio and FWHM(G) are smaller than those of simple CNWs
(a)
(d)
(b)
(e)
(c)
(f)
Fig. 2. SEM images of the deposits with and without scratching pretreatment; [(a) and (b)] without scratching, undulation with (c) 100 times and (d) 200 times of scraping, ultrasonic vibration with (e) small and (f) large diamond grains for 10 min.
Intensity (arb. units)
C.Y. Cheng et al. / Diamond & Related Materials 27-28 (2012) 40–44
ID /IG
FWHM(G) (cm-1)
(a)
0.32
29
(b)
0.46
38
(c)
0.30
25
(d)
0.48
33
(e)
0.35
32
43
(a)
G
D (a) (b) (c)
(d)
(b)
(e) 1100
1300
1500
1700
Raman shift (cm-1) Fig. 3. Raman spectra of the deposits with and without scratching pretreatment; (a) without scratching, undulation with (b) 100 times and (c) 200 times of scraping, ultrasonic vibration with (d) small and (e) large diamond grains for 10 min. The inset shows the ID/IG ratio and FWHM (G).
reported by other groups [8,20]. This suggests larger sp 2 cluster sizes and higher sp 2 order of our deposits. Both the ID/IG ratio and FWHM(G) for the spectra (b) and (d) are similar and the largest among all, indicating that these two deposits contain more disordered sp 2 phase. This is consistent with the SEM images [Fig. 2(c) and (e)], where the two deposits have similar morphologies with a larger coverage of the underlying film. Field emission characteristics of the deposits obtained with and without scratching pretreatment are shown in Fig. 4(a). The turn-on field for the onset of emission is measured at a current density of 10 nA/cm 2 with a reproducibility of ±0.4 V/μm. Without scratching pretreatment, the turn-on field is 2.1 V/μm and the current density reaches 1 × 10 − 5 A/cm 2 at 3.0 V/μm. With scratching pretreatment, the turn-on field decreases down to 1.2 V/μm, and the current density increases to more than 1 × 10 − 4 A/cm 2 at 3.0 V/μm, depending a little on the method of scratching. Thus the emission performance of the composite films is higher than that of simple CNWs. The Fowler– Nordheim (FN) plots [23] corresponding to Fig. 4(a) are shown in Fig. 4(b). They are composed of two straight lines, indicating that the emission is apparently governed by a metallic-type FN tunneling [23]. The relationship among the slope of the FN plot (Δ), the effective potential barrier height (Φ), and β is expressed as Δ = −6840Φ 3/2/β, where Δ and Φ are in V/μm and eV, respectively. The local field is enhanced by β compared to the macroscopic applied field. In Fig. 4(b), Δ in the low field region is steeper than that in the high field region. A slope decrease in the high field region may be referred to as depletion of available trapped electrons to be emitted [24–26]. A list of the turn-on field, and Δ and β in the low field region derived from Fig. 4(a) and (b) is given in Table 1. It is believed that field emission occurs at localized sites rather than uniformly over the surface. The field lines focus exclusively onto the sharp wall edges since the wall edges occupy the higher location than the underlying film. Here we used Φ = 5.0 eV for graphite to calculate β on the assumption that electrons are emitted from defect states at the Fermi level of the walls, such as dangling bond, π bonding, and π* antibonding states incorporated with nitrogen [10,15]. If the emission occurs above the Fermi level, similar to nanocrystalline diamond
Fig. 4. (a) Field emission characteristics of the deposits with and without scratching pretreatment. (b) The Fowler–Nordheim (FN) plots corresponding to (a). I and E are the emission current and the applied field, respectively.
film [27], the value of Φ should be lower. In Table 1, the higher the value of β becomes, the lower the turn-on field is. This means that the emission is governed by field enhancement. The value of β should ideally correspond to the aspect ratio of the walls. The measured β = 1250 for simple CNWs is smaller than the aspect ratio of ~ 2000 measured from the SEM images [Fig. 2(a) and (b)] due probably to the field screening between the walls. In contrast, the measured β = 1700–2700 for the composite films is larger than that for simple CNWs and several times larger than the aspect ratios of several hundreds to around 1000 depending upon the growth condition. Thus the field screening is suppressed for any of the composite films. The turn-on field is plotted as a function of number of scrapings for undulation or ultrasonic vibration period in Fig. 5. For undulation, the
Table 1 Turn-on field, and Δ and β in the low field region. β was obtained by assuming Φ = 5.0 eV for graphite. Pretreatment
Turn-on field (V/μm)
Δ (V/μm)
β
No scratching Undulation (100 times) Undulation (200 times) Vibration (small grains, 10 min) Vibration (large grains, 10 min)
2.1 1.4 1.4 1.6 1.2
− 61 − 45 − 44 − 41 − 28
1250 1710 1720 1860 2700
44
C.Y. Cheng et al. / Diamond & Related Materials 27-28 (2012) 40–44
Fig. 5. Turn-on field as a function of number of scrapings for undulation [(i), top axis] or ultrasonic vibration period [(ii) small and (iii) large diamond grains, bottom axis].
continuous undulant scratches with a number of residual diamond grains exclusively along the scratches, accompanied by a large increase in surface roughness. The ultrasonic vibration treatment with small diamond grains left a number of diamond grains, while that with large diamond grains left only a small number of diamond grains. The morphology of the deposits was found to depend more on the duration of the scratching method and less on the scratching method itself. With increase of the duration of scratching, the deposits exhibited a morphological change from CNWs to a film/CNW composite and finally to CNWs on a film, accompanied by an increase in wall spacing. The emission turn-on field decreased down to 1.2–1.4 V/μm and the current density increased above 1 × 10− 4 A/cm 2 with either pretreatment method. The emission was enhanced due to suppression of the field screening as shown by a large increase in the field enhancement factor β. The results confirm that the emission performance is governed by the field enhancement and can be controlled by the wall spacing. Scratching treatments do not directly affect field emission. Rather, they change the structure and orientation of the resulting deposits, which consequently affect field emission. References
turn-on field shows a plain decrease from 2.1 V/μm without scratching to 1.6 V/μm with 10 times of scraping and is saturated at 1.4 V/μm for a further increase in the number of scrapings. For vibration, the turn-on field decreases monotonically from 2.1 V/μm without scratching to 1.2 V/μm for a further increase in vibration period. The slope of the plot with large diamond grains is apparently steeper than that with small diamond grains. As shown in the SEM images in Fig. 2, the wall spacing tends to increase with the increase of the duration of scratching, while the morphology changes from CNWs to a film/CNW composite and lastly to CNWs on a film. The possible mechanism is considered as follows. Without scratching, simple CNWs are formed on bare substrates. The wall spacing cannot be controlled as the dense wall growth would force the walls to align vertically. With scratching, the increasing duration of scratching leaves a larger number of diamond grains on the substrate, which in turn causes a higher nucleation density and a faster coalescence of the island films. For a short duration of scratching, the walls are generated from the boundary regions of the islands as the nucleation density is not high. In this case, the wall spacing is increased as the lateral wall connection is intercepted by the islands. For a long duration of scratching, the walls are grown on the underlying film as the nucleation density is too high to allow the walls to grow at the boundaries of the islands. In this case, the effective wall spacing for field enhancement is increased because only a smaller portion of the walls are aligned vertically. Scratching treatments do not directly affect field emission. Rather, they change the structure and orientation of the resulting deposits, thus affecting field emission. 4. Conclusions The morphology and field emission properties of nanocrystalline diamond/CNW composite films were examined with two methods of scratching pretreatment. The undulation treatment produced
[1] S.T. Purcell, P. Vincent, C. Journet, Vu Thien Binh, Phys. Rev. Lett. 88 (2002) 105502. [2] Z.L. Wang, R.P. Gao, W.A. De Heer, P. Poncharal, Appl. Phys. Lett. 80 (2002) 856. [3] J.H. Lee, S.H. Lee, W.S. Kim, H.J. Lee, J.N. Heo, T.W. Jeong, C.W. Baik, S.H. Park, SeGi Yu, J.B. Park, Y.W. Jin, J.M. Kim, H.J. Lee, J.W. Moon, M.A. Yoo, J.W. Nam, S.H. Cho, J.S. Ha, T.I. Yoon, J.H. Park, D.H. Choe, Appl. Phys. Lett. 89 (2006) 253115. [4] C.-W. Baik, J. Lee, J.H. Choi, I. Jung, H.R. Choi, Y.W. Jin, J.M. Kim, Appl. Phys. Lett. 96 (2010) 023105. [5] Y.H. Wu, P.W. Qiao, T.C. Chong, Z.X. Shen, Adv. Mater. (Weinheim, Ger.) 14 (2002) 64. [6] K. Kobayashi, M. Tanimura, H. Nakai, A. Yoshimura, H. Yoshimura, K. Kojima, M. Tachibana, J. Appl. Phys. 101 (2007) 094306. [7] Y. Wu, B. Yang, B. Zong, H. Sun, Z. Shen, Y. Feng, J. Mater. Chem. 14 (2004) 469. [8] A.T.H. Chuang, J. Robertson, B.O. Boskovic, K.K.K. Kozio, Appl. Phys. Lett. 90 (2007) 123107. [9] A. Malesevic, R. Kemps, A. Vanhulsel, M.P. Chowdhury, A. Volodin, C.V. Haesendonck, J. Appl. Phys. 104 (2008) 084301. [10] S. Shimada, K. Teii, M. Nakashima, Diamond Relat. Mater. 19 (2010) 956. [11] L. Giorgi, Th. Dikonimos Makris, R. Giorgi, N. Lisi, E. Salernitano, Sens. Actuators, B 126 (2007) 144. [12] E. Luais, M. Boujtita, A. Gohier, A. Tailleur, S. Casimirius, M.A. Djouadi, A. Granier, P.Y. Tessier, Appl. Phys. Lett. 95 (2009) 014104. [13] K. Mase, H. Kondo, S. Kondo, M. Hori, M. Hiramatsu, H. Kano, Appl. Phys. Lett. 98 (2011) 193108. [14] W. Takeuchi, M. Ura, M. Hiramatsu, Y. Tokuda, H. Kano, M. Hori, Appl. Phys. Lett. 92 (2008) 213103. [15] K. Teii, S. Shimada, M. Nakashima, A.T.H. Chuang, J. Appl. Phys. 106 (2009) 084303. [16] K. Teii, M. Nakashima, Appl. Phys. Lett. 96 (2010) 023112. [17] H. Nagasawa, K. Yagi, T. Kawahara, J. Cryst. Growth 237 (2002) 1244. [18] C.J. Lee, D.W. Kim, T.J. Lee, Y.C. Choi, Y.S. Park, Y.H. Lee, W.B. Choi, N.S. Lee, G.S. Park, J.M. Kim, Chem. Phys. Lett. 312 (1999) 461. [19] A.C. Ferrari, J. Robertson, Phys. Rev. B 61 (2000) 14095. [20] Z.H. Ni, H.M. Fan, Y.P. Feng, Z.X. Shen, B.J. Yang, Y.H. Wu, J. Chem. Phys. 124 (2006) 204703. [21] R.J. Nemanich, S.A. Solin, Phys. Rev. B 20 (1979) 392. [22] F. Tuinstra, J.L. Koenig, J. Chem. Phys. 53 (1970) 1126. [23] R.H. Fowler, L. Nordheim, Proc. R. Soc. Lond. A 119 (1928) 173. [24] G.A.J. Amaratunga, S.R.P. Silva, Appl. Phys. Lett. 68 (1996) 2529. [25] K. Teii, S. Matsumoto, J. Robertson, Appl. Phys. Lett. 92 (2008) 013115. [26] K. Teii, R. Yamao, S. Matsumoto, J. Appl. Phys. 106 (2009) 113706. [27] T. Ikeda, K. Teii, Appl. Phys. Lett. 94 (2009) 143102.