Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
3 CHARACTERIZATION TECHNIQUES RELEVANT TO SUPERHARD MATERIALS Many different types of analytical techniques have been used to characterize the superhard materials. More traditional methods such as optical microscopy, optical absorption, x-ray diffraction (XRD), x-ray fluorescence (XRF), and Auger spectroscopy, have proved very useful in ways that have been discussed at length in other publications (Field, 1979; Sze, 1983; Tsai and Bogy, 1987~. The special characteristics of the superhard materials and their unusual growth methods, however, have led investigators to rely heavily on a number of less commonly used techniques. This chapter provides background regarding the nature of these techniques and the information they provide and points out the need for new characterization methods. The areas covered include techniques for characterizing phases present, surface structure, chemical composition, and properties such as thermal, electrical, and optical. The need for fundamental studies of the diamond-like materials is also discussed. STRUCTURE STUDIES Determination of the different carbon phases incorporated in diamond and diamond-like carbon materials is of importance for controlling the growth or deposition conditions. Several techniques are sensitive to differences between carbon phases, including Auger electron spectroscopy, electron energy loss spectroscopy (EELS), and Raman spectroscopy (Solin and Ramdas, 1970~. Since Auger and EELS are surface analysis techniques, they are sensitive only to the first 3 nm of the surface and hence incapable of characterizing the bulk of the material. Raman spectroscopy appears to be the analytical tool of choice because the diamond Raman neak , , ,# at 1332 cm~' is easily distinguished from peaks related to other forms of carbon' as shown in Figure 3-1. A detailed investigation of the Raman spectra of films prepared under various deposition conditions, correlated with Auger and EELS data, should lead to the development of standard reference materials for calibrating instrumentation in the field (as opposed to the research laboratory). In addition to the Raman signal, a photoluminescence signal may also be present. This fluorescence background! arises generally from defect centers within the films and therefore could provide an analytical tool for their identification. This may be important for films to be used in electronic or optoelectronic applications because free carrier properties such as carrier lifetime anti mobility may be affected by them. The presence of the fluorescence signal does introduce a 61
62 10 8 6 4 2 6 _ ~ in As ,1_ go o 8 6 4 2 o a) DIAMON D J1: X 250 / J A__ Oh o He 8 m cr: )4 . - - a b) GRAPHITE c) MICROCRYSTALLINE GRAPHITE A 0 500 1000 1500 2000 2500 3000 3500 FIGURE 3-1 Raman spectra of (a) diamond, (b) graphite, and (c) microcrystalline graphite. The solid lines in (a) and (b) represent the first- and second-order spectra of crystalline diamond and graphite respectively. The broad line labeled ~ in (a) is the spectrum of a-Si scaled to the diamond frequency to represent the spectrum of amorphous diamond, while the broad line labeled cx in (b) is that due to amorphous graphite (Nemanich et al., 1988~.
63 problem, however, since it represents a background signal that must be subtracted from the Raman signal. If the fluorescence signal is large, experimental uncertainties introduced in the subtraction process could limit the ability to make the Raman data quantitative for this application. The most critical single parameter for understanding the nature of the diamond and DLC materials is the sp3/sp2 ratio. Diamond is characterized by Sp3 bonding between atoms, and graphite is characterized by sp2 bonding. The sp3/sp2 ratio can be determined by several techniques, including nuclear magnetic resonance spectroscopy (Hoch and Reynhardt, 1988), electron energy loss spectroscopy, and optical measurements of the dielectric function. Also of great interest is the distribution of hydrogen over the different types of carbon sites (CH, CH2, ant! CHIC. This can be determined using Fourier transform infrared spectroscopy (FTIR). The diamond-like materials are typically amorphous, and the local environment around each carbon atom may provide information of interest. This can presumably be accomplished by neutron or x-ray diffraction techniques. Extended x-ray absorption fine structure spectroscopy (EXAFS) may also be useful in this context. The concentration of dangling bonds (free radical sites) may be measured by electron spin resonance (ESR). The diamond-like materials often exhibit a medium-range order, which may be small clusters or microcrystallites. High-resolution transmission electron microscopy will undoubtedly be of value in characterizing these features. Surface Ch~i~t~on The surface structure of diamond is extremely stable and is responsible for some of its interesting properties (e.g., low coefficient of friction and negative electron affinity). Characterization techniques useful for examining the structure of diamond surfaces may be conveniently clivicled into three categories, depending on whether they measure geometric structure, electronic structure, or chemical composition. A summary of these techniques follows. Geometric Structure The classical methods for studying the geometric structure of crystalline surfaces are low- energy electron diffraction (LEED) and high-energy electron diffraction (HEED). These methods are well known and require no discussion here. Grazing angle x-ray scattering can also be used to give information about geometric surface structure. To apply this technique to monolayers, one requires a synchrotron x-ray source--e."., the Brookhaven source--and very low scattering angles. In recent years the scanning tunneling microscope (STM) has been developed. It is an extremely powerful method that can resolve individual atoms on surfaces but cannot be applied to insulating materials. The scanning force microscope, on the other hand, does not require a conducting substrate; however, its resolution is significantly less than that of the STM. The expectations are that the scanning force microscope will be able to provide atom resolution within the next few years.
64 Electronic Structure The electronic structure of adsorbed species on surface can be studied by angle resolved photoelectron spectroscopy. Either x-ray or ultraviolet photons can be used. The angular and energy distribution of emitted electrons gives direct information on the electronic structure of the surface, from which geometric structures can also be inferred. Auger spectroscopy may also be useful. Ultrahigh-vacuum ellipsometric spectroscopy can be used to investigate the electronic structure of surfaces. It can provide a measure of the dielectric coefficient tensor and its frequency dependence at optical wavelengths. This provides information about surface electronic states. In principle, the TR region can be done as well. Chemical Composition There are, of course, ~ number of other well-known surface analytical methods that give information on the composition, and sometimes bonding, in the near-surface region. These are secondary ion moss spectroscopy (SIMS), Auger spectroscopy, photoelectron spectroscopy and electron spectroscopy for chemical analysis (ESCA). For highly insulating materials, secondary neutral mass spectroscopy (SNMS) may be useful. Dynamic Processes Inelastic scattering processes give information on the motion of scattering centers. Both inelastic neutron scattering and inelastic laser light scattering can be applied to the study of diamond surfaces. Molecular beam scattering from diamond! surfaces, perhaps using C+ or H+, also might be considered. Scattering intensity as a function of angle and energy will give information about dynamic processes on the surface. Fourier transform infrared spectroscopy is an extremely powerful method for analyzing the vibrational modes of absorbed species. FTIR is now capable of monolayer sensitivity by means of double-reflection infrared anti attenuated total reflection techniques. Raman spectroscopy can also be done in a Fourier transform mode and may prove useful for studying nucleation processes. High-resolution electron energy loss spectroscopy can also be used to obtain information on the vibrational structure of absorbed species. Molecular Modeling Molecular modeling methods are powerful tools for the study of surfaces. Both static structural problems and dynamic processes may be studied. For example, molecular mechanics and more advanced ab initio quantum mechanical techniques can be used to study the state of reconstructed crystalline surfaces. Random, diffusive processes on surfaces can be modeled using stochastic methods directly--e."., through the general Langevin equation for generating trajectories ant! various Monte Carlo algorithms. Also, molecular dynamic techniques can be used to develop trajectories, from which dynamics can be studied. Supercomputers are required for advanced work in either of these areas.
65 ELEMENTAL COMPOSITION It is astonishing that a majority of reports on the DLC and diamond materials still do not include a measurement of the elemental composition of the films. Of particular importance is the amount of hydrogen. Hydrogen can be conveniently determined by isM nuclear reaction analysis (NRA), which provides a depth profile of hydrogen concentration. Hydrogen can also be determined by helium forward scatter spectroscopy, but with a higher detection limit. Analysis of heavy elements is conveniently done by Rutherford backscatter spectroscopy (RBS). Auger spectroscopy is also useful for elemental analysis, and when combined with sputtering can give composition profiles as well. However, both RBS and Auger spectroscopy have detection limits on the order of 10-3 atomic percent. When lower detection limits are required secondary ion mass spectroscopy or neutron activation is useful. Finally, combustion analysis can provide measurements of hydrogen, carbon, oxygen, and nitrogen for samples of milligram size. FUNDAMENTAL MECHANICAL MEASUREMENTS Surprisingly few measurements of the fundamental mechanical properties of CVD diamond and diamond-like materials have been made. Of particular interest would be the measurements of the modulus of elasticity, the thermal expansion coefficient, the thermal conductivity, and the level of internal stress. All of these mechanical properties are of fundamental importance for many proposed applications. The films appear to be very impermeable; however, there are no accurate measurements of the permeabilities of these materials. Applications of diamond and diamond-like materials as diffusion barriers, corrosion protection layers, and vapor barriers will depend on the permeabilities. A simple and very instructive property that should be routinely measured is the mass density. When combined with the knowledge of elemental composition, this permits placing the material of interest in the field of atom number density versus atom fraction hydrogen (Figure 3-2~. This type of plot provides a convenient way for categorizing different classes of hydrocarbons and carbon. Similar compounds, for example the n-alkanes (AL), fall in tight groupings of the diagram. Diamond, DI, is at the upper left-hand corner. The adamantanes, AD, which are 10 and 14 carbon atom molecular diamonds, are at the lower right of the diamond-like hydrocarbon (a-C:H) grouping. The diamond-like hydrocarbons (filled symbols) range from approximately 0.20 to 0.60 atom fraction hydrogen and have number densities greater than 0.19 gram-atom per cubic centimeter. The vertical dashed lines show the theoretical composition range predicted for fully constrained, random hydrocarbon networks. The atom number density of a-C-M increases with hydrogen content, in contrast to the behavior of conventional hydrocarbons. This increase is caused by the increase in average coordination number of carbon from 3 to 4 as the amount of hydrogen is increased. The diamond-like carbons (a-C) appear to be a different class of material, with little or no hydrogen and with atom number densities from approximately 0.19 to 0.28 gram-atom per cubic centimeter. These densities indicate significant Sp3, tetrahedral, bonding. The amorphous carbons (AM) are dominates] by spa, trigonal, bonding and can be formed by laser or electron-beam evaporation. Other symbols are AC, oligomers of acetylene; AD, adamantanes; AR, polynuclear aromatics; and OR, graphite. There are indications from several sources that the diamond-like materials are inhomogerleous. In particular, variations of elemental composition and properties with depth of the film are suspected. This question is rarely addressed but should be considered when assessing film properties. The interface between diamond or diamond-like materials and various substrates
66 ._ ~ 0.20 o - 0.10 0.30 . '/ '. ~ A\4 my\ AC ~ j ~ ~ ~ · 1 1 ~ , 0 0.25 0.50 0.75 Atom fraction hydrogen DI diamond ~ a-C diamond-like carbon O GR graphite AM amorphous carbon ~ AD adamantanes + AL normal alkanes O AC oligomers of acetylene ~ AR polynuclear aromatics *, ., a, *, ., a-ciH diamond-like hydrocarbons FIGURE 3-2 Atom number density versus atom fraction hydrogen (Angus and Hayman, 1988~. is also of great importance. The strength of adhesion and the nature of this interface will often be the dominant factors in determining whether these materials can be used in specific applications. Recent work on the performance of brittle coatings (Evans, et al., 1983) has relevance to diamond. Scratch and Hardness Tests Although the quantitative meaning of a scratch test can be obscure, the qualitative results can be useful as an indicator of relative wear and hardness. It is fairly simple to set up comparative materials for scratch tests: Sapphire (Al2O3), cemented WC, SiC, B4C, cubic EN, and diamond separately mounted metallographically and polished can be used as a base on which the unknown is rubbed (usually a circular motion enhances distinction of new scratches from background). More sophisticated techniques like motion of a diamond point over the sample surface can sometimes be made more quantitative. Indentation techniques on thin films are difficult and should be viewed with skepticism, particularly when the unknown is essentially as hard as diamond. Careful observation of the scratch itself is also sometimes qualitatively helpful in relative hardness evaluation. Woodell abrasion resistance is a somewhat more quantitative measure of hardness based on relative wear of different materials as determined by measurement after simultaneous abrasion under controlled loads. The technique is not as convenient as a simple scratch test (Woodell, 1935).
67 THERMAL CONDUCTIVITY Many of the applications of superhard materials, especially CVD diamond, will rely heavily on their high thermal conductivities (Figure 3-3~. Recent work has shown that the thermal conductivity of CVD diamond depends strongly on the hydrogen/methane ratio used in the microwave plasma CVD process and that the measurements correlate strongly with the DLC 1332 cm-t peak in the Raman spectrum (Ono et al., 1986~. It has also been shown that the thermal conductivity of CVD diamond below room temperature is significantly lower than that of natural diamond crystals (Morelli et al., 1988), possibly because of the small crystallite sizes in CVD diamond. This variability makes it imperative that the thermal properties of CVD diamond be well characterized, not only as a means of establishing design parameters for specific devices but also because it may be possible to use thermal measurements as a means of monitoring the deposition process. 1000 ....... i :-::.~ :-.-: ·:~ :-:-:~:-:-- :-:-:::~:~:~:-:-:-: : :.::.:. ·- :-:-:-:- :- 300 ~ .... . ·.-.-.-.-........ it_ :::: :: HE cu ~ \ Diamond . ~ ~ . as, 100 ~ / \ / S 22"2'"22 1 3 10 30 100 300 1000' TEMPERATURE (I FIGURE 3-3 The thermal conductivity of various materials as a function of temperature. The vertical band represents the temperature region from -25 to l250C.
68 Measurements of thermal properties are important at temperatures below and above room temperature. Diamond has the highest thermal conductivity of any known material at room temperature. The temperature range between 20 and 250 K is important not only because thermal data can provide a measure of the quality of the diamond but also because diamond has considerable potential for applications as a superconductor of heat in this temperature interval. Measurements above room temperature are desirable because electronic devices made of diamond are expected to operate at elevated temperatures. The methods of Ono and coworkers (1986) and Morelli and coworkers (1988) are cumbersome and would probably not be suitable for routine measurement; both employ delaminates! CVD diamond films, which are highly brittle. In the Ono method the film strip is suspended in vacuum between two heated supports. The temperature distribution alone the _ _ = _ ~ ~ , ~ = _ _ _ length of the sample is measured by means of a thermograph, which monitors the specimen through a CaF2 window. From an approximate solution of the steady-state heat diffusion equation, the thermal conductivity of the film can be calculated. The Morelli method is more conventional and allows for measurement of thermal conductivity as a function of temperature down to 20 K. Thermocouples are attached to a strip of delaminated diamond film suspended in a cryostat, and one end of the film is heated to establish a thermal current in the film. From a measurement of the temperatures at two points along the film and other parameters, the thermal conductivity can be calculated. Several methods based on thermal wave generation and detection appear promising for measuring the thermal properties of diamond films. These methods yield the thermal diffusivity, A, which is related to the thermal conductivity, k, by A = k/Cp where C = specific heat and p = density. The first method is called photothermal radiometry (Konstad and Nordal, 1980~. In this technique, a laser beam (CO2 or argon-ion) modulated at a frequency, f, is focused onto the surface of the diamond film, thus heating the surface and causing a thermal wave to propagate into the film. The temperature at the surface is measured with an infrared detector, typically liquid-nitrogen-cooled InSb. The temperature at the surface depends not only on the direct heating from the laser beam but also on thermal energy that returns to the surface because of reflections at the film-substrate interface. (RefIections may also result from defects within the film.) Thermal waves are highly damped, with the penetration depth varying inversely with the square root of f. Thus, by varying f, one can probe to different depths of the film, thereby affecting the amount of thermal energy reflected from the film-substrate interface. By measuring the phase of the surface temperature as a function of f, one can obtain the thermal diffusivity of the film and, hence, the thermal conductivity. The phase of the thermal signal is measured with a vector lock-in amplifier. Measurements can be made over a temperature range of 200C to 900.C by conducting the experiments with the specimen in a furnace (Frederikse and Ying, 1988). With pure diamond films 0.1 to 1 mm thick, frequencies in the range 1000 to 4000 Hz are needed. In the case of thinner films, the frequencies required can be significantly higher, and consequently the measurements can be more difficult to perform. In this case, a variation of the
69 technique could be used. For example, instead of measuring the effect of propagation of heat normal to the film surface, one could measure the effect of heat propagation parallel to the surface. This type of experiment can be done at significantly lower frequencies. Another method! that can be used is the optical beam deflection technique (Boccara et al.. 1980~. Here again, the film surface is heated by a focused, modulated laser beam. In this method, a probe helium-neon laser beam skimming the surface of the specimen is deflected by the thermal gradients in the air, which act as a probe of the specimen temperature. The deflection is measurer! by a quadrant detector. By translating the probe beam position relative to the heated spot on the film surface, one can obtain not only the thermal diffusivity of the film but, in addition, the thermal anisotropy in the film. The thermal properties of diamond are most sensitive to defects near 77 K, the temperature at which the thermal conductivity of pure diamond has a maximum. At this temperature phonon scattering begins to limit the thermal conductivity, and above this temperature only large concentrations of defects would be important. Thus, thermal measurements are expected to yield excellent data regarding defect densities in diamond films. In this case, methods basses on the work of Morelli and coworkers (1988) appear to be most appropriate. Residual stress can be a consequence of the thermal history. ELECTRICAL AND OPTICAL CHARACTERIZATION The performance of semiconductor materials in electronic and optoelectronic devices such as transistors, detectors, light-emitting diodes, lasers, and optical switches is dependent on the extrinsic electronic states in the material. These states may be cleliberately introduced into the material by processes such as doping and ion implantation, or they may be present because of unwanted impurities and defects introduced during processing. Important parameters regarding these states are concentrations and spatial distribution of defect and impurity states; energy levels within the band gap; carrier (electrons ant! holes) capture rates, release rates, and Nobilities; recombination rates and branching ratios for radiative and nonradiative decay; and quantum efficiencies for luminescence and photoconductivity. Some of the techniques that can be used to probe the defect-related electronic states that lie within the forbidden gap of diamond ant, SiC and how these states affect the electronic properties are as follows: · Cathodoluminescence imaging and spectroscopy. These experiments can be carried out in a scanning electron microscope. Cathodoluminescence is optical emission from defect-related or intrinsic electronic states resulting from excitation by an electron beam. By combining scanning-electron imaging with luminescence spectroscopy, information about point defects and impurities, extended defects (e.g., dislocations, stacking faults), and crystal growth habits can be correlated (Collins, 1987~. In addition, cathodoluminescence imaging can be correlated with secondary-electron or other imaging modes of the electron microscope that may provide complementary information about crystal growth and morphology. Cathodoluminescence images and spectra in CVD diamond have been able to identify defect centers that have been observed in bulk diamond (Boccara et al., 1980~. · Electrical conductivity. This measurement provides information about the electrical activity of dopant species. The activation energies of impurities are investigated by measuring the temperature-dependence of the electrical conductivity. As part of this research, various methods of attaching electrical contacts to the specimens will require investigation. . Photoluminescence and photoconductivity excitation spectroscopies. These techniques are used to study defect-related optical absorption. An arc lamp or flash lamp with high UV
70 output, combined with a scanning monochromator, or a tunable excimer-pumped dye laser can provide the means for exciting photocarriers above the band gap. The absorption process is observed either by measuring the total luminescence signal as a function of the incident photon energy or by observing the wavelength-dependence of the photoconductivity. Time-resolved photoluminescence and photoconductivity also provide information about carrier (electron or hole) lifetime and mobilities. A wide variety of defect centers have been identified in natural diamond by means of photoluminescence and cathodoluminescence. Some of these defects such as the neutral vacancy and nitrogen-carbon interstitial complexes have now been observed in CVD diamond. It is important that extensive investigation of these defects and their abundance in CVD diamond be carried out. As is true with other new fields, progress in understanding behavior is impeded by the lack of a "standard" materials of known purity and structure. This is especially true in investigating electronic behavior; the distribution of standardized samples to various laboratories would be a boon. REFERENCES Angus, I. C., and C. C. Hayman. 1988. Low-pressure metastable growth of diamond and diamond-like phases. Science, Vol. 241, p. 913. Boccara, A. C., D. Fournier, and I. Badoz. 1980. Thermo-optical spectroscopy: detection by the "mirage effect. Appl. Phys. Lett., Vol. 36, p. 130. Collins, A. T. 1987. Cathodoluminescence decay time studies of the neutral vacancy in diamond. I. Phys. C: Solic! State Phys., Vol. 20, p. 2027. Evans, A. G., G. Crumiey, and R. Demaray. 1983. On the mechanical behavior of brittle coatings and layers. Oxid. Met. Vol. 20, pp. 193-216. Field, J. E., ed. 1979. The Properties of Diamond. New York: Academic Press. Frederikse, H. P. R., and X. T. Ying. 1988. Heat conductivity of oxide coatings by photothermal radiometry between 293 and 1173.K. Appl. Optics, Vol. 27, no. 22, pp. 4672-4675. Hoch, M. I. R., and E. C. Reynhardt. 1988. Nuclear spin-lattice relaxation of dilute spins in semiconducting diamond. Phys. Rev. B. Vol. 37, p. 9222. Konstad, S. O., and P. E. Nordal. 1980. Photoacoustic and photothermal techniques for powder and surface spectroscopy. Appl. Surf. Sci., Vol. 6, pp. 372-391. Morelli, D. T., C. P. Beetz, and T. A. Perry. 1988. Thermal conductivity of synthetic diamond films. I. Appl. Phys., Vol. 64, pp. 3063-3066. Nemanich, R. I., I. T. Glass, G. Lucovsky, and R. E. Shroder. 1988. Raman scattering of carbon bonding in diamond and diamond-like films. I. Vac. Sci. Tech. A, Vol. 6, p. 1783. Ono, A., T. Baya, H. Fuanmoto, and A. Nishikawa. 1986. Thermal conductivity of diamond films synthesized by microwave plasma CVD. Jap. J. Appl. Phys., Vol. 25, p. LS08.
71 Solin, S. A., and A. K. Ramdas. 1970. Raman spectrum of diamond. Phys. Rev. B. Vol. 1, p. 1687. Sze, S. M., ed. 1983. VESI Technology. New York: McGraw Hill Book Co. Tsai, H. C., and D. B. Bogy. 1987. Critical review, Characterization of diamond-like carbon films and their application as overcoats on thin-film media for magnetic recording. I. Vac. Tech. A, Vol. 5, p. 3287. Tsai, H. C., and D. B. Bogy. 1987. Critical Review: Characterization of diamond like carbon films and their application as overcoats in thin-film application as overcoats in thin-film media for magnetic recording. I. Va. Sci. Technol. A, Vol. 6, pp. 3287-3312. Woodell, C. E. 1935. Method of comparing the hardness of electric furnace products and natural abrasives. Trans. Electrochem. Soc., Vol. 6S, pp. 1 1 1-130.
