Using JPF1 format

JOURNAL OF APPLIED PHYSICS

VOLUME 89, NUMBER 5

1 MARCH 2001

Electron field emission for ultrananocrystalline diamond films A. R. Krauss Materials Science and Chemistry Divisions, Argonne National Laboratory, Argonne, Illinois 60439

O. Aucielloa) Materials Science Division, Argonne National Laboratory, Argonne, Illinois 60439

M. Q. Ding Beijing Vacuum Electronics Research Institute, P.O. Box 749, Beijing 100016, People’s Republic of China

D. M. Gruen and Y. Huang Materials Science and Chemistry Divisions, Argonne National Laboratory, Argonne, Illinois 60439

V. V. Zhirnov Semiconductor Research Corporation, Research Triangle Park, North Carolina 27709

E. I. Givargizov Institute of Crystallography, 11733, Moscow, Russia

A. Breskin, R. Chechen, and E. Shefer Weizmann Institute, Rehovoth, Israel

V. Konov, S. Pimenov, and A. Karabutov General Physics Institute, Russian Academy of Science, Moscow, Russia

A. Rakhimov and N. Suetin Moscow State University, Nuclear Engineering Department, Moscow, Russia

共Received 2 June 2000; accepted for publication 30 August 2000兲 Ultrananocrystalline diamond 共UNCD兲 films 0.1–2.4 ␮m thick were conformally deposited on sharp single Si microtip emitters, using microwave CH4 –Ar plasma-enhanced chemical vapor deposition in combination with a dielectrophoretic seeding process. Field-emission studies exhibited stable, extremely high 共60–100 ␮A/tip兲 emission current, with little variation in threshold fields as a function of film thickness or Si tip radius. The electron emission properties of high aspect ratio Si microtips, coated with diamond using the hot filament chemical vapor deposition 共HFCVD兲 process were found to be very different from those of the UNCD-coated tips. For the HFCVD process, there is a strong dependence of the emission threshold on both the diamond coating thickness and Si tip radius. Quantum photoyield measurements of the UNCD films revealed that these films have an enhanced density of states within the bulk diamond band gap that is correlated with a reduction in the threshold field for electron emission. In addition, scanning tunneling microscopy studies indicate that the emission sites from UNCD films are related to minima or inflection points in the surface topography, and not to surface asperities. These data, in conjunction with tight binding pseudopotential calculations, indicate that grain boundaries play a critical role in the electron emission properties of UNCD films, such that these boundaries: 共a兲 provide a conducting path from the substrate to the diamond–vacuum interface, 共b兲 produce a geometric enhancement in the local electric field via internal structures, rather than surface topography, and 共c兲 produce an enhancement in the local density of states within the bulk diamond band gap. © 2001 American Institute of Physics. 关DOI: 10.1063/1.1320009兴

I. INTRODUCTION

from the diamond surface is not well understood. In particular, the following questions have not been definitively answered: 共1兲 Is the tunneling barrier located at the substrate– diamond or the diamond–vacuum interface? 共2兲 Is there a local enhancement of the electric field at the emission sites, and if so, is this enhancement related to the presence of surface asperities? 共3兲 Is the tunneling current associated with emission from valence band, conduction band, or interband states? One of the principal experimental difficulties in answering these questions pertains to the fact that the Fowler–Nordheim16 expression for electron current density

Diamond films are being extensively investigated for application to field emission sources1–9 critical to the operation of a variety of vacuum microelectronic devices and high current, high frequency tubes. Diamond-coated Si microtip emitters appear to be able to sustain higher emission currents 共presumably because of the high thermal conductivity of diamond兲, and exhibit lower threshold fields10–15 than uncoated Si or Mo tip arrays, although the electron emission process a兲

Electronic mail: [email protected]

0021-8979/2001/89(5)/2958/10/$18.00

2958

© 2001 American Institute of Physics

Downloaded 19 Jun 2001 to 132.76.33.15. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/japo/japcr.jsp

Krauss et al.

J. Appl. Phys., Vol. 89, No. 5, 1 March 2001

depends on both the effective work function and the local electric field at the emission site in a manner which does not permit independent determination of these two quantities. This local field may differ from that of the macroscopic field. By simultaneously measuring the emission current density and the kinetic energy distribution of the electrons, it is possible to independently determine the work function and the local electric field at the emission site.17 Recent measurements using this method revealed relatively high work functions 共5.0–5.3 eV兲 and high field enhancement factors for both diamond like-carbon and nanocrystalline chemical vapor deposition 共CVD兲 diamond films. The cause of the field enhancement on nominally smooth, flat films was attributed to asperities resulting from arcing during the initial application of high voltage. In the present work, we have used CVD methods to synthesize diamond films on high aspect ratio sharp Si tips, which provide an intrinsically high field enhancement associated with the very small radius of curvature of the tip. The emission from these coated tips turns on smoothly without arcing, which results in no enhancement in the surface roughness as compared with the as-deposited state, and as confirmed by scanning electron microscopy 共SEM兲 observation after electron emission measurements. Conventional CVD deposition methods such as hot filament CVD 共HFCVD兲6,9,14 and microwave plasma-enhanced CVD 共MPECVD兲 with CH4 –H2 gas mixtures7,8 typically produce micron-sized diamond grains either at the ends of the tip or on the shank14 as shown in Fig. 1. Such nonconformal diamond coatings result in a degree of stabilization of the emission current and reduction of the threshold field, but are still subject to thermal runaway due to the poor thermal conductivity of Si, and high aspect ratio of the tips. For an array of microtips, many of the tips will have no diamond coating at all, while the size and location of the diamond ‘‘blob’’ varies from tip to tip. It is desirable from both an experimental and a device point of view to produce conformal, smooth diamond coatings with uniform, controllable thickness, capable of yielding low threshold fields, and high electron emission currents with good reproducibility, lateral uniformity, and temporal stability. A solution to the problem of producing conformal, smooth diamond films has been provided by our group via the development of a microwave plasma CVD technique capable of yielding crystalline diamond thin films with 2–5 nm average grain size that are totally devoid of intergranular amorphous or graphitic phases.18 In order to distinguish this material from the material commonly described in the literature as ‘‘nanocrystalline diamond’’ with 50–100 nm grain size and significant intergranular non-diamond phases,19 we have adopted the term ‘‘ultrananocrystalline diamond’’ 共UNCD兲. The UNCD deposition method involves the use of microwave plasmas with CH4 /Ar or C60 /Ar gas mixtures, typically with no more than 1%–2% H2 added to assure plasma stability in the ASTeX PDS-17 reactor.20,21 Optical emission spectroscopy analysis of the CH4 /Ar and C60 /Ar plasmas revealed that carbon dimers are the dominant species in the plasma,22 and are considered to be the diamond growth species, and that atomic hydrogen abstraction pro-

2959

cesses are not involved in the UNCD growth mechanism.23 However, up to 20% H2 can be added to the plasma before the dominant growth process changes from the C2 insertion mechanism22 to the conventional CH3* –H0 process.24 The average grain size increases up to approximately 100 nm, but the grain boundary morphology22 appears to be unchanged throughout the 0%–20% H2 range.25 Our UNCD dimerinsertion growth process is characterized by little regasification of nascent diamond grains by atomic hydrogen and very high nucleation densities and renucleation rates. Consequently, UNCD films exhibit little if any grain coarsening as the film thickness increases and are ideal candidates for producing conformal, smooth coatings on Si microtips. One of the issues in relation to diamond films for field emission sources is the effect of film thickness on the field emission properties. If the diamond film is too thin, it may not be possible to obtain a continuous coating, whereas an extremely thick film may result in poor emission properties due to a resistive drop within the film, resulting in a reduction in the field at the diamond–vacuum interface at moderately large current densities. This would result in deviation from the Fowler–Nordheim current versus voltage relation and an apparent space charge saturation effect.11 Therefore, it is important to determine the film thickness that yields the best field emission performance. Additionally, studies of the thickness effect on field emission may provide information about the mechanism of field emission from diamond films. Unfortunately, there is very little information in the literature regarding the effect of the diamond film thickness on the field emission characteristics of Si-coated tips. Zhirnov et al. studied the effect of coating thickness on field emission using a compacted diamond powder deposited by dielectrophoresis, and reported a space charge effect at a film thickness above 1 ␮m.11 There appear to be no previous studies of the dependence of the field emission from tips coated with dense, conformal crystalline diamond coatings. One of the main reasons for this lack of information is the difficulty in coating tips with a conformal thin diamond film using conventional diamond deposition methods. Using the deposition process developed in our laboratory,12,13,20 we were able to coat high aspect ratio single Si tip emitters with smooth and conformal UNCD thin films with 2–5 nm grain size and controlled film thickness. The diamond films deposited on the Si microtips were characterized using scanning electron microscopy 共SEM兲, transmission electron microscopy 共TEM兲, electron energy loss spectroscopy 共EELS兲, and Raman spectroscopy. The main goal of the work discussed in this article is to study the effect of film thickness and Si microtip radius on the field emission for two very different forms of diamond coatings. This comparison results in the identification of a possible electron emission mechanism for UNCD films. II. EXPERIMENTAL PROCEDURES

Single-crystal silicon tip emitters were grown on 1⫻1 mm2 silicon posts.26 First, Si whiskers were grown by a Si–Au eutectic vapor–liquid–solid technique. Then, the whiskers were sharpened by oxidation and subsequent


2960


Krauss et al.

FIG. 2. Series of SEM images of a 25 nm diameter Si tip emitter coated with a UNCD film with incremental coating thickness. The tip on the left is uncoated and successive images represent coatings ranging from 100 nm to 2.4 ␮m in thickness.

FIG. 1. Nonconformal diamond film growth using HFCVD on Si microwhiskers.

Hartree–Fock 共HF兲 etching to form high aspect ratio sharp tips with a radius of curvature less than 25 nm. The resultant tips were 50–200 ␮m in height. Single tips were prepared for this study by growing tip arrays, and then breaking off all but one of the tips while viewing in a SEM. The Si tip emitters were cleaned in 20% HF for 30 s to remove the native oxide and other surface contaminants. Then, the tip emitters were seeded for 3 min in a suspension of 5 nm diamond powder in ethanol using a dielectrophoresis method.10,11 Following the seeding procedure, the tip emitters were immediately loaded into the MPECVD reactor 共ASTeX-PDS-17兲 for diamond deposition. A mixture of CH4 and Ar gases was admitted to the deposition chamber through mass flow controllers. The total pressure in the chamber was kept constant at 100 Torr, while the flow rates were 1 and 99 sccm for CH4 and Ar, respectively. The microwave-input power was 800 W, and the substrate temperature was 800 °C. It has been previously demonstrated.12,13,20,21 that these conditions lead to the growth of phase-pure diamond films with equiaxed 2–5 nm grains.

The deposition was carried out for 20 min to attain a diamond film thickness of about 0.1 ␮m for the first deposition step. The study of the film thickness effect on field emission was done by sequential growth of diamond layers on the same tip emitter. The sequence of experiments involved the following repetitive steps: 共1兲 deposition of a 100 nm thick diamond layer, 共2兲 I–V measurements, 共3兲 SEM imaging, and 共4兲 further deposition of a 100 nm layer. The process was repeated iteratively until the maximum film thickness was obtained. Tips of various radii and length were studied. As shown in Fig. 1, these tips were uniformly coated with the UNCD film and had hemispherical UNCD caps with radii varying from 0.1 to 2.4 ␮m. A second set of four samples was prepared by coating an array of microtips using HFCVD. As shown in Fig. 2, the diamond coating grew as discrete grains on either the shank, or on the very end of the microtips. A single microtip with a symmetrically positioned spherical grain was selected from each array, and all the remaining microtips were broken off. Samples were prepared with diamond grains of various radii, and closely matched diamond grain sizes were selected on microtips of various radii. The field emission characteristics of the single microtip emitters were investigated using a computer-controlled current–voltage testing system.12,13 The emission current was measured using an anode 1.89 mm in diameter with rounded edges to suppress arcing at sharp edges. The anode– cathode 共tip emitter兲 spacing was determined by using a high magnification charge coupled device 共CCD兲 camera. The anode voltage was increased under computer control until the emission current reached a preset value 共usually 10⫺5 A兲 and then decreased to zero. The anode–cathode spacing 共h兲 and the tip emitter geometry are illustrated in Fig. 3. The radius of the Si microtip is designated ‘‘r’’ in Fig. 3, and the radius of the hemispherical diamond cap is designated ‘‘R.’’


Krauss et al.


FIG. 3. Schematic showing the geometry of the coated tip emitter and the position of the anode.

The tip-anode spacing was set at a value equal to the length 共h兲 of the microtip 共typically 50 to 200 ␮m兲. Under these conditions, for an applied voltage V, the electric field at the emitting surface is given by F⫽ 共 V/2h 兲 * 共 h/r 兲 ⫽V/2r

共1a兲

F⫽ 共 V/2h 兲 * 共 h/R 兲 ⫽V/2R,

共1b兲

or

depending on whether the tunneling barrier is located at the Si–diamond interface or the diamond-vacuum interface. Due to the sharpness of the microtips, they make excellent TEM specimens. As-deposited diamond films 共about 0.1 ␮m thick兲 on sharp Si tips can be directly used for TEM studies without conventional TEM sample preparation. The TEM image 共Fig. 4兲 shows that the Si tip is centered within

FIG. 4. TEM image of a Si tip coated with a UNCD film about 0.1 ␮m in thickness. Note the smoothness and conformality of the UNCD film. The average roughness is estimated to be in the range of 20–40 nm.

2961

the UNCD coating and that the value of R is therefore equal to the diamond coating thickness. A Phillips CM-30 TEM operated at 300 keV was used to study the microstructure of the UNCD films. The nature of the carbon bonding in the diamond film was determined using a Gatan EELS. A Renishaw Raman microscope with a wavelength of 632 nm was also employed to supplement the EELS study. Measurements of the density of interband states were carried out using a calibrated electron collection system coupled to a grating monochromator light source. The photoelectron current was measured by an electrometer as a function of the wavelength of light incident on the diamond sample. Scanning tunneling microscopy 共STM兲 measurements were carried out on UNCD films on flat Si substrates using the STM in various modes of operation. In the normal mode, the tip is placed in close proximity to the surface, such that it is within the surface tunneling barrier and produces a topographic image. In a second mode of operation, the sample-tip distance is increased so that electron emission completely penetrates the surface potential barrier and a mapping of the electron emission intensity is obtained. Details of the emission mapping procedure are reported elsewhere.27,28 In a second STM experiment,29 topographic mapping is used to determine high and low points in the topography. One high point and one low point are selected, and the tip is placed over each of these regions while the bias voltage is scanned and the emission current is measured.

III. RESULTS AND DISCUSSION A. Characterization of a UNCD-coated Si tip

The SEM images of a typical single tip emitter before and after successive coatings with UNCD diamond films are shown in Fig. 1. Fig. 1共a兲 shows a bare 150 ␮m long microtip with a radius of curvature of 25 nm. The seeded tips 共without UNCD coating兲 were visually indistinguishable from the uncoated tips. The microtip was coated with sequential layers of UNCD until a thickness of 2.4 ␮m was reached. Some of the SEM images are shown in Figs. 1共b兲– 2共e兲. The thickness of the successive UNCD diamond layers was determined by measuring the radii of the coatings from the SEM images, assuming the tip apex was positioned at the center of the diamond film hemisphere as shown in the TEM image of Fig. 4. A unique feature distinguishing all UNCD diamond layers is that they were conformal and smooth in the 0.1–2.4 ␮m range investigated here, as opposed to the case of Si tips coated by conventional CH4 /H2 diamond growth methods. We attribute this remarkable result to: 共1兲 the dielectrophoretic seeding using the nanosized diamond powder suspension, and 共2兲 the UNCD diamond growth process.20 The average roughness of the film was estimated to be in the same 20–40 nm range as that of the UNCD films grown on flat, smooth Si wafers.22 The selected area electron diffraction 共SAED兲 pattern shown in Fig. 5 reveals the crystalline character of the UNCD film. The three rings present in Fig. 5 correspond to the 共111兲, 共220兲, and 共311兲 orientations, respectively. The spot patterns are from the single crystal Si


2962

Krauss et al.


FIG. 5. SAED pattern displays three rings, identified as diamond 共111兲, 共220兲, and 共311兲 orientations, showing a polycrystalline diamond nature.

microtip substrate. The UNCD films deposited on the needleshaped Si tips have microstructure similar to those grown on flat Si substrates.12,13,20–22 EELS analysis was performed by focusing the electron beam onto the areas around the tip apex. Figure 6 shows a typical EELS spectrum, in which only the ␴* peak at about 303 eV due to s p 3 bonded carbon 共diamond兲 is present. The absence of the ␲* peak around 286 eV resulting from sp 2 bonded carbon indicates that the film is a highly s p 3 bonded diamond film. To supplement the EELS studies, Raman analysis was conducted using a focused laser beam with a beam size 2 ␮m in diameter. Figure 7 shows a Raman spectrum taken from the tip apex region. The spectrum is similar to that of previous UNCD films on flat substrates.22 The sharp peak at 1332 cm⫺1 overlapping a broad fluorescence peak is characteristic of UNCD diamond, with broadening of the peak related to the nanocrystalline grain size.20 The less intense peak at 1580 cm⫺1 is associated with s p 2 bonded

FIG. 6. EELS spectrum showing only the ␴* feature at 303 eV due to sp 3 bonded carbon with no presence of the ␲* peak at 286 eV resulting from sp 2 bonded carbon.

FIG. 7. A Raman spectrum taken from an UNCD-coated Si surface reveals a sharp intense feature at 1332 cm⫺1 overlapped with a broad peak associated with the UNCD. The feature at 1580 cm⫺1 corresponds to sp 2 bonded carbon.

carbon, but because of the 50:1 sp 2 /sp 3 sensitivity ratio20–22 of Raman spectroscopy, corresponds to only a few percent of the total electronic bonding. A recent tight-binding pseudopotential calculation30,31 has predicted that UNCD films have grain boundaries that are ⬃3.56 Å in width, in accord with high resolution TEM data.21 This calculation also predicts that the grain boundaries are ⬃40%–80%. locally sp 2 bonded, and that there is a continuum of states at the grain boundary within the 5.5 eV diamond band gap. Because of the small grain size and consequent high ratio of grain boundary-to-bulk atoms, the grain boundary carbon atoms are sufficient to completely account for the experimentally observed sp 2 character. B. Field emission characterization

In a previous article,32 we have shown that Si tip emitters coated with 0.1 ␮m thick UNCD films exhibit a substantial reduction in the threshold field and an increase in the maximum emission current compared with uncoated Si microtips. In this work, we have expanded the previous work by systematically studying the effect of film thickness on field emission properties of UNCD-coated Si tip emitters. Figure 8 shows a set of I–V curves for a Si microtip emitter coated with sequential UNCD layers with a total thickness from 0.1 to 2.4 ␮m. The data indicate that for all UNCD coating thicknesses in this range, the threshold field is much smaller and the emission current much higher than the corresponding values for the uncoated Si microtip. The emission characteristics of the seeded tip 共dielectrophoresis only, without UNCD coating兲 were indistinguishable form those of the uncoated tips. Since we are trying to determine whether tunneling occurs at the Si–diamond or the diamond–vacuum interface, it is not clear whether the local field enhancement factor is determined by the Si tip radius 共r兲 or the diamond coating radius R. The experimental results are therefore presented in terms of the voltage at which an emission current of 1 ⫻10⫺7 A is obtained as a function of film thickness produced on different tips. By measuring the threshold voltage



FIG. 8. Typical I–V curves taken from a Si tip emitter coated with various UNCD film thicknesses: 共a兲 0 ␮m, 共b兲 0.1 ␮m, 共c兲 0.5 ␮m, and 共d兲 2.4 ␮m. Each set of data contains four curves 共two up-and-down scans兲. Note that all the I–V curves corresponding to UNCD-coated Si tips are shifted to low turn-on voltages.

rather than emission at a higher current, voltage drops within the resistive film and space charge effects are minimized. Results for the HFCVD-coated microtips are presented in Fig. 9 as a function of radius of the diamond ball for several microtips with the same 25 nm radius. The threshold voltage increases monotonically with diamond film thickness, suggesting that emission is controlled by the field at the diamond–vacuum interface. However, Fig. 10 shows the threshold voltage for Si microtips of various radii coated with diamond films of the same 共0.5 ␮m thickness. The threshold voltage now varies monotonically with the Si tip radius, suggesting that there is also a tunneling barrier at the Si–diamond interface. As shown in Fig. 11, the tips conformally coated with UNCD display a completely different behavior. This figure contains sets of data taken from three Si microtips with the following geometry: 共a兲 r⫽25 nm, h⫽150 ␮ m, 共b兲 r ⫽25 nm, h⫽60 ␮ m, and 共c兲 r⫽300 nm, h⫽60 ␮ m. The threshold voltages for the uncoated tips depend strongly on the tip radius, but for even the thinnest UNCD coatings 共100 nm兲, the threshold voltage becomes nearly independent of

FIG. 9. Threshold field for HFCVD diamond coated 25 nm diameter Si microtips as a function of the radius of the diamond film ball at the end of the tip.

Krauss et al.

2963

FIG. 10. Threshold field for HFCVD diamond coated Si microtips of varying radii with an approximately 0.5 ␮m radius diamond film ball at the end of the tip.

both the Si tip radius and the UNCD coating thickness. For all three microtips, there is a broad minimum in the threshold voltage for films with thickness in the 0.2–0.5 ␮m range, followed by a slight increase for films thicker than 0.5 ␮m. Consequently, it appears that the local field enhancement is an intrinsic property of the UNCD film, rather than the result of topographic field enhancement associated with either the Si tip sharpness or the radius of curvature of the diamond film coating. A comparison of the I–V characteristics of the UNCDcoated tip emitters 共Figs. 8 and 11兲 with previous results11 on tips coated by dielectrophretic deposition at room temperature11 show that: 共1兲 All the UNCD-coated emitters with film thickness from 0.1 to 2.4 m exhibited much better emission than that of the bare Si tip emitter. 共2兲 The UNCDcoated tips exhibit a minimal drop in emission current with coating thickness ⬎1 ␮m compared with films produced entirely by dielectrophoresis.11 This implies that, as suggested by the tight-binding calculations,30,31 electron transport through UNCD may be quite different from transport

FIG. 11. Turn-on voltage vs film thickness for three Si tip emitters coated with UNCD films. Note the large drop in the turn-on voltage for UNCD coatings from 0 to 0.1 ␮m thick, followed by a broad minimum centered at about 0.5 ␮m and a subsequent slow rise for film thickness in excess of 0.5 ␮m.


2964

Krauss et al.


FIG. 12. I–V curves from UNCD-coated Si tip emitters taken with an emission current of up to 100 ␮A for film thicknesses: 共a兲 0.5 ␮m, 共b兲 1.0 ␮m, and 共c兲 2.4 ␮m. Note that for a UNCD film thickness up to 2.4 ␮m do not exhibit the ‘‘N’’ shape characteristic of poor emission.

through the consolidated nanocrystalline powder, where space charge effect occurs when the film is thicker than 1 ␮m. Figure 12 shows the high current I–V curves for three Si tip emitters coated with UNCD films of three thickness: 共a兲 0.5 ␮m, 共b兲 1.0 ␮m, and 2.4 ␮m, respectively. The 100 ␮A maximum measured value is three orders of magnitude higher than the maximum current yielded by an uncoated Si tip emitter, which generally burns out at an emission current of Ⰶ1 ␮A, and does not represent the maximum current that can be sustained by the UNCD-coated tips. Clearly, the thicker the film, the higher the voltage required to reach a given current. This variation may be attributable to the resistive drop within the UNCD films. By taking slices of Fig. 12 at various current values, it is possible to solve self-consistently for the voltage drop within the film using the familiar relation V s ⫽V t ⫹IR⫽V t ⫹Ir␳ /A,

共2兲

where V s is the potential at the diamond–vacuum interface, V t is the potential at the diamond–Si interface 共assumed equal to the externally applied potential兲, and R is the resistance across the film of thickness r. ␳ is an effective resistivity which may in principle vary with both the film thickness and emission current density, and I is the average current through the film terminating in an effective emitting area A. We treat the quantity ␳ /A as a single unknown variable in the following calculation, and it is adjusted to selfconsistently yield a common value for V s at a given emission current for all three samples. These values are self-consistent to within ⬃7%. At specific current values, the potential drop within the film is found to be linear with film thickness, although the variation with current density exhibits closer to a square root relation as show in Fig. 13. This latter variation could be the result of either decreasing resistivity or increasing emission area as the emission current increases. We have conducted a long-term stability test for a single Si tip coated with a 2.4 ␮m thick UNCD film. As shown in Fig. 14, the test was run for more than 400 h. The initial current was 100 ␮A for a fixed voltage of 1200 V with a 60

FIG. 13. Values calculated from Eq. 共1兲 for the IR drop within the UNCD film as a function of film thickness.

␮m gap. The current decreased slowly to 57–60 ␮A over a period of 48 h and then remained nearly constant at that value for the next 367 h. Further stability tests are planned to investigate possible surface conditioning effects that may be responsible for the initial decrease of the emission current and to further probe the emission stability properties of the UNCD coatings. The power density corresponding to the observed emission current is several Mw/cm2 if it is assumed that all of the measured current originates from the hemispherical cap of the coated microtip. This is a physically unrealistic value, and strongly suggests that the emission current in fact originates from the entire UNCD-coated surface of the Si microwhisker. Considering that the Si whisker is 60 ␮m in length, emission from the entire coating would reduce the requisite power density by 2–3 orders of magnitude. The assumption that the entire coating is emitting is consistent with previously reported findings12 that smooth UNCD coatings on flat Si substrates exhibit emission thresholds as low as 2 V/␮m. It was further calculated from the Fowler–

FIG. 14. Time dependence of the emission current from a single UNCDcoated Si microtip with a UNCD coating 1.0 ␮m thick.



Krauss et al.

2965

FIG. 17. Topographic and electron emission line scans for the cross section shown in Figs. 16共a兲 and 16共b兲. FIG. 15. Quantitative photoyield measurements at various locations on the surface as a function of photon energy for an UNCD film grown using an Ar–C60/20% H2 plasma at 800 °C.

Nordheim analysis of the smooth UNCD films that either the assumption of a physically unreasonable value for the work function 共⬃20 meV兲 or an extremely high local field enhancement factor 共⬎5,000兲 is required to explain the smooth film.12 This observation strongly suggests that the local electric field enhancement reflects an intrinsic property of the material, 共rather than surface topography兲 of the smooth films, in accord with the present result that neither the radius of the Si microtip nor the UNCD hemispherical cap has significant effect on the threshold field. In order to explain the observed electron-emitting properties of UNCD, a viable model must meet several key requirements: Even though the TEM data show that UNCD contains no graphitic or amorphous carbon secondary phases, there must be a conducting path from the substrate to the surface. In addition, the previous smooth film Fowler– Nordheim data12 strongly suggest that there is a very large enhancement in the local electric field at the emission sites and/or an effective work function that is significantly less than the diamond band gap. The UNCD films are observed to be electrically conducting33 and the tight binding calculation30,31 identifies the grain boundaries as the probable conduction path. The tight binding calculation also shows that at the grain boundary, there is a continuous density of states throughout the

FIG. 16. STM topographic 共a兲 and electron emission 共b兲 maps for a common area of an Ar–C60/20% H2 UNCD film on a flat Si substrate. Scanning area is 6⫻6 ␮m2.

bulk diamond band gap, resulting in a significant lowering of the potential barrier that must be traversed by field-emitted electrons. Finally, the abrupt character and extreme narrowness of the grain boundaries results in a conducting path with an extremely high aspect ratio and consequently a strong enhancement in the local electric field at the grain boundary/ diamond/vacuum interface as proposed by ‘‘triple point’’ models.34 In the case of the UNCD films,12 this electric field enhancement is associated with the internal structure of the UNCD film, and is not related to topographic asperities. The morphology of the UNCD films with conducting sp 2 grain boundary regions ⬍0.4 nm thick in an insulating sp 3 matrix is very reminiscent of recently studied amorphous carbon films35 in which it was found that field electron emission originated from small sp 2 regions in an sp 3 matrix, and the ease of electron emission depended on the size of the sp 2 regions, rather than the sp 2 /sp 3 ratio. The optimum size for the sp 2 region was determined to be ⭐1 nm, comparable with the observed grain boundary width in UNCD. The optimimum size was ascribed to a trade off between the need for a small structure to provide maximum field enhancement and the need for carrying the emission current.35 Similar dimensional correlations were ascribed to the optimized electron emission in microcrystalline diamond and carbon nanotubes. In a related experiment using STM measurements,27,28 the electron emission from diamond films on smooth Si substrates was found to originate from grain boundaries/ diamond–graphite interface regions. Photoelectron quantum yield measurements have shown that UNCD films grown in an Ar–C60 plasma with up to 20% added H2 exhibit an enhancement in the intergap density of states that is correlated with the observed threshold field. It should further be noted that this enhancement in densities of state is observed only for photon energies less than the 5.5 eV band gap as shown in Fig. 15. STM images of the surface topography and field emission distribution maps for the same area of UNCD films grown in an Ar–C60 plasma with 20% added H2 film are shown in Fig. 16, and matching line scans for these areas are shown in Fig. 17. It is obvious that the regions of most intense electron emission do not correspond to topographic


2966


Krauss et al.

FIG. 18. 共a兲 Topographic STM image of an Ar–C60/20% H2 UNCD thin film with adjacent regions of topographic maxima and minima indicated. 共b兲 I vs V curves for the two regions marked in 共a兲.

asperities, and that most of the asperities correspond to minima in the emission current. Maxima in the emission current appear to be correlated with either inflection points or minima in the local topography. Although it is not possible to unambiguously identify the regions of maximum current with the grain boundaries observed by TEM, the result is consistent with emission from sites in which either there is a locally reduced tunneling barrier of an internal, morphologically related enhancement of the local electric field, rather than a field enhancement related to surface topography. Additional STM data is shown in Fig. 18共a兲 in which two regions corresponding to adjacent high and low points in the topographic image are indicated. Figure 18共b兲 shows the I–V plots corresponding to the two sites. Again, the electron emission originates at the minimum in the local surface topography.

Although there are some ambiguities in the interpretation of STM data in this manner, the overall picture obtained from both theoretical and a variety of experimental sources is consistent with a grain boundary emission model in UNCD. It must be emphasized however, that the morphology of UNCD films is both quite different from other forms of diamond film, and in many respects much easier to characterize than that of so-called nanocrystalline diamond, diamond-like carbon and ‘‘amorphic diamond’’ materials. It remains to be seen whether the model postulated for the UNCD films can be extended to these other electron-emitting forms of carbon. However, recent data on electron emission as a function of the morphologies of the sp 2 and sp 3 regions35 suggest that the morphological properties of UNCD that give rise to electron emission in UNCD may be similar to those features that give rise to optimized electron emission in a number of dif-


Krauss et al.


ferent forms of amorphous carbon films, as well as other carbon-based materials including microcrystalline diamond and carbon nanotubes. IV. CONCLUSIONS

We deposited UNCD films on high aspect ratio Si tip emitters using MPECVD in combination with a dielectrophoretic seeding process. SEM images showed that the UNCD films are extremely smooth and conformal to the sharp tip emitter. Field emission studies showed a substantial reduction in emission turn-on voltage for the diamond-coated tip emitters and a large enhancement in emission current for films with thickness ranging from 0.1 to 2.4 ␮m, as compared to the values for uncoated Si tip emitters. The threshold fields are found to be independent of film thickness or surface topographic feature size, and are ascribed to the intrinsic morphology of the UNCD film. A model is presented in which emission occurs at the grain boundary–vacuum interface or from a region very close to the grain boundary, and experimental photoyield and STM data support this correlation. The conclusions drawn with regard to the role of grain morphology in electron emission from UNCD are in accord with recent findings on the optimization of electron emission in amorphous carbon films. ACKNOWLEDGMENTS

The work was funded by the DOE Office of Basic Energy Sciences under Contract No. W-31-109-ENG-38, and DARPA/ONR under Contact No. N00014-96-C0283. 1

C. Wang, A. Garcia, D. C. Ingram, and M. E. Kordesh, Electron. Lett. 27, 1459 共1991兲. 2 M. W. Geis, N. N. Efremov, J. D. Woodhouse, M. D. McAleese, M. Marchywka, D. G. Socker, and J. F. Houshedez, IEEE Electron Device Lett. 12, 456 共1991兲. 3 J. E. Jaskie, MRS Bull. 21, 59 共1996兲. 4 M. W. Geis, J. C. Twichell, and T. M. Lyzcasz, J. Vac. Sci. Technol. B 14, 2060 共1996兲. 5 D. Zhou, A. R. Krauss, L. C. Qin, T. G. McCauley, D. M. Gruen, T. D. Corrigan, R. P. H. Chang, and H. Gnaser, J. Appl. Phys. 82, 4546 共1997兲. 6 W. Zhu, G. P. Kochanski, S. Jin, and L. Seibles, J. Appl. Phys. 78, 2707 共1995兲. 7 M. W. Geis, N. N. Efremow, K. E. Krohn, J. C. Twichell, T. M. Lyszczarz, R. Kalish, J. A. Greer, and M. D. Tabat, Nature 共London兲 393, 431 共1998兲. 8 V. V. Zhirnov and J. J. Hren, MRS Bull. 23, 43 共1998兲. 9 J. Liu et al., J. Vac. Sci. Technol. B 13, 422 共1995兲.

2967

10

W. B. Choi, J. J. Cuomo, V. V. Zhirnov, A. F. Myers, and J. J. Hren, Appl. Phys. Lett. 68, 1 共1996兲. 11 V. V. Zhirnov, G. J. Wojak, W. B. Choi, J. J. Cuomo, and J. J. Hren, J. Vac. Sci. Technol. A 15, 1733 共1997兲. 12 A. R. Krauss, D. M. Gruen, D. Zhou, T. G. McCauley, L. C. Qin, T. D. Corrigan, O. Auciello, and R. P. H. Chang, Mater. Res. Soc. Symp. Proc. 495, 299 共1998兲. 13 T. G. McCauley et al., Mater. Res. Soc. Symp. Proc. 498, 227 共1997兲. 14 E. I. Givargizov, V. V. Zhirnov, A. V. Kuznetsov, and P. S. Plekhanov Mater. Lett. 18, 61 共1993兲; E. I. Givargizov, V. V. Zhirnov, N. N. Chubun, and A. N. Stepanova, J. Vac. Sci. Technol. B 15, 450 共1997兲. 15 A. F. Myers, S. M. Camphausen, J. J. Cuomo, J. J. Hren, J. Liu, and J. Bruley, J. Vac. Sci. Technol. B 14, 2024 共1996兲. 16 R. H. Fowler and L. W. Nordheim, Proc. R. Soc. London, Ser. A 119, 173 共1928兲. 17 O. Gro¨ning, O. M. Ku¨ttel, P. Gro¨ning, and L. Schlapbach, J. Vac. Sci. Technol. B 17, 1970 共1999兲. 18 L. C. Qin, D. Zhou, A. R. Krauss, and D. M. Gruen, Nanostruct. Mater. 10, 649 共1998兲. 19 B. V. Spitsyn, L. L. Bouilov, and B. V. Derjaguin, J. Cryst. Growth 52, 219 共1981兲. 20 D. M. Gruen, S. Liu, A. R. Krauss, and X. Pan, J. Appl. Phys. 75, 1758 共1994兲. 21 R. Csencsits, C. D. Zuiker, D. M. Gruen, and A. R. Krauss, Solid State Phenom. 51–52, 261 共1996兲. 22 D. Zhou, D. M. Gruen, L. C. Qin, T. G. McCauley, and A. R. Krauss, J. Appl. Phys. 84, 1981 共1998兲. 23 P. C. Redfern, D. A. Horner, L. A. Curtiss, and D. M. Gruen, J. Phys. Chem. 100, 11654 共1996兲. 24 S. J. Harris and D. G. Goodwin, J. Phys. Chem. 97, 23 共1993兲. 25 L. C. Qin 共private communication兲. 26 E. I. Givargizov, J. Vac. Sci. Technol. B 11, 449 共1993兲. 27 A. V. Karabutov, V. D. Frolov, S. M. Pimenov, V. I. Konov, Diamond Relat. Mater. 8, 763 共1999兲. 28 V. D. Frolov, A. V. Karabutov, V. I. Konov, S. M. Pimenov, A. M. Prokhorov, J. Phys. D 32, 815 共1999兲. 29 A. T. Rakhimov, N. V. Suetin, Z. S. Soldatov, M. A. Timofeyev, A. S. Trifonov, V. Khanin, and A. Silzars, J. Vac. Sci. Technol. B 18, 76 共2000兲. 30 P. Keblinski, D. Wolf, F. Cleri, S. R. Phillpot, and H. Gleiter, MRS Bull. 23, 36 共1998兲; P. Keblinski, D. Wolf, S. R. Phillpot, and H. Gleiter, J. Mater. Res. 13, 2077 共1998兲; F. Cleri, P. Keblinski, L. Colombo, D. Wolf, and S. R. Phillpot, Europhys. Lett. 46, 671 共1999兲. 31 P. Zapol, M. Sternberg, T. Frauenheim, D. M. Gruen, and L. A. Curtiss, Proceedings of the Sixth International Symposium on Diamond Materials, 196th Meeting of the Electrochemical Society, 17–22 October 1999. 32 M. Q. Ding, A. R. Krauss, O. Auciello, D. M. Gruen, Y. Huang, V. V. Zhirnov, and E. I. Givargizov, Proceedings of IVMC 99, July 1999, Darmstadt, Germany, p. 282. 33 S. Bhattacharya et al. 共unpublished兲. 34 M. W. Geis, J. C. Twichell, and T. M. Lyszczarz, J. Vac. Sci. Technol. B 14, 2060 共1996兲. 35 A. Lie, A. C. Ferrari, T. Yagi, and J. Robertson, Appl. Phys. Lett. 共in press兲.


Using JPF1 format - Weizmann Institute of Science

Using JPF1 format - Weizmann Institute of Science

Suggest Documents

Using APL format - Weizmann Institute of Science

Using JCP format - Weizmann Institute of Science