The influence of ammonia on the electrical properties of detonation ...

JOURNAL OF APPLIED PHYSICS 106, 123704 共2009兲

The influence of ammonia on the electrical properties of detonation nanodiamond Mose Bevilacqua, Aysha Chaudhary, and Richard B. Jackmana兲 London Centre for Nanotechnology and Department of Electronic and Electrical Engineering, University College London, 17-19 Gordon Street, London WC1H 0AH, United Kingdom

共Received 7 October 2009; accepted 10 November 2009; published online 28 December 2009兲 Detonation nanodiamonds 共DNDs兲 are an interesting class of materials for sensing applications, but little is currently understood about their electrical properties. Here, aggregated DNDs are explored with impedance spectroscopy and are found to offer near-to-ideal dielectric characteristics, which is intriguing given their nanostructure. When exposed to ammonia, two highly conductive pathways emerge through the material; these appear to be associated with grain boundary and grain interior processes, the latter potentially due to surface transfer doping. This process is reversible given modest temperature increases suggesting DNDs may offer a solid state electrical platform for ammonia sensing applications. © 2009 American Institute of Physics. 关doi:10.1063/1.3272912兴 I. INTRODUCTION

Chemical sensors find widespread use monitoring a large number of processes within safety, pollution, medical engineering, and industrial environments. Sensors need to be fast, specific, reliable, sensitive, inexpensive, and have the possibility for miniaturization.1–3 Ammonia is widely used in industrial processes; sensors for NH3 are particularly important since it is highly toxic, corrosive, and can explode when mixed with air. It is also monitored in a number of medical applications, such as the detection of ammonia in the breath of patients indicating end-stage liver disease.4 Solid-state sensors offer advantages over gas or liquid-based techniques in terms of size and ability to integrate into demanding environments. Diamond, which can be considered as a wide band gap 共5.5 eV兲 semiconductor capable of supporting high carrier mobilities, high electric field breakdown strength, and the highest thermal conductivity of any material,5 is of considerable interest as a platform for chemical sensors. The physical and chemically robust nature of diamond means that it can be deployed within harsh environments. Acoustic wave devices have been produced,6,7 as have sensors based on electrochemical processes.8,9 Diamond surfaces can display strongly differing electrical properties depending on the chemical nature of the surface-terminating groups present. For example, hydrogenated surfaces may be highly conductive, while oxygen termination removes this conductivity.10,11 These observations, when considered alongside the reversible nature of chemical changes to diamond surfaces 共since the diamond lattice itself is so strongly bonded兲 has led to interest in the direct detection of chemical species by monitoring electrical changes to the diamond surface. Some years ago, Gi and co-workers12 showed that polycrystalline diamond films, grown by a chemical vapor deposition 共CVD兲 method, showed decreased surface conductivity in the presence of ammonia, while increasing in the presence of oxidizing gases 共such as NO2兲. More recently, an a兲

Author to whom correspondence should be addressed. Electronic mail: [email protected].

0021-8979/2009/106共12兲/123704/7/$25.00

attempt to increase the sensitivity of this approach to usable levels was reported by Wang et al.13 Polycrystalline CVD diamond films were plasma-etched to produce an array of micron-scale cone structures, which were subsequently hydrogenated; the structures were capable of sensing both NO2 and NH3 gases through surface electrical changes caused by the oxidizing or reducing species. Various forms of nanometer-scale structured diamond materials have recently been synthesized.14 One such material, often referred to as “detonation nanodiamond” 共DND兲, is a powder comprised of 5–10 nm diamond crystals aggregated to the micron scale, formed by detonating an oxygen-deficient trinitrotoluene/ hexogen composition in inert media.15 A model has been proposed16 whereby “core” diamond crystals, created when unburnt carbon experiences the shock wave following detonation, are then surrounded by an aggregate of “sootlike” material after the shock wave passes. The tenacious nature of the aggregates, such as their resistance to acids, was explained on the basis of a diffusion-limited reaction, with little oxidation occurring between the diamond grains. The term “agglutination” was adopted to describe the powder, to avoid the suggestion that the agglomeration was due to van der Waals type forces. The electrical properties of DND are currently poorly understood. In the current study, impedance spectroscopy 共IS兲 has been used to investigate the electronic properties of DND in some detail, and then investigate the changes occurring when DND is exposed to ammonia. The results are discussed both in terms of what changes occur to the DND surfaces when in contact with NH3 and the prospects for DND to be a suitable material for chemical sensing applications. II. EXPERIMENTAL METHODS

DND powder 共SuperSyndia SSX 0–3.5兲 was investigated in the “as-received” condition. Scanning electron microscopy images of a typical particle are shown in Fig. 1. In Fig. 1共a兲, a typical agglomerated “grain” can be seen to have an overall dimension of the order of 3 ␮m, while in Fig. 1共b兲, it is just possible to discern extremely small crystals

106, 123704-1

© 2009 American Institute of Physics

123704-2

J. Appl. Phys. 106, 123704 共2009兲

Bevilacqua, Chaudhary, and Jackman

FIG. 1. Scanning electron microscope images of the DND used revealing 共a兲 typical grain size and 共b兲 the densely agglomerated nature of the DND crystals.

aggregated to grains of 30–100 nm sizes, which in turn make up the 3 ␮m particle. It was impractical to carry out IS on a single particle, hence a multiplicity of grains was placed between two tantalum disks. The total powder thickness between the plates was approximately 0.5 mm. Tantalum was chosen as the metal for the disks due to its refractory properties; a drop of silver paste was applied on top of each disk and then dried in a furnace at 100 ° C for 5 min in order to assure a good contact between electrical probes and disks, and to drive off any water vapor. IS involves the measurement of the real and imaginary components of the films impedance as a function of both frequency and temperature. This enables the different contributions to the overall resistance and capacitance of the films to be isolated.17 The impedance, as a complex number, can be defined as Z共␻兲 = Re共Z兲 + j Im共Z兲,

共1兲

where the real part refers to resistive contributions and the imaginary part to the capacitative. Hence, any contributions to the overall impedance measured that have strongly differing RC components can be identified. In a typical IS analysis, the impedance is measured as a function of frequency, and the real component plotted against the imaginary component as the frequency is changed. These so-called “Cole– Cole” plots may then reveal differing semicircular responses for each RC component of the film; repeating the measurements at differing temperatures enables the activation energy for each component to be determined. In the present case, IS was performed over the frequency range 0.1 Hz–10 MHz 共Solartron 1260 impedance system with Solartron 1296 dielectric interface兲. The voltage applied between the Ta plates was 5 V, giving a field strength of 100 V/cm. All experiments were carried out inside a stainless steel vacuum chamber

FIG. 2. Real impedance plotted against measurement frequency for 共a兲 DND powder measured in air and vacuum at room temperature and 共b兲 DND powder measured at 400 and 550 ° C in air.

共base pressure ⬃10−3 mbar兲, which allowed both electrical screening and control of the ambient gas. Exposure to NH3 involved heating ammonium hydroxide 共50% NH3 by weight兲 and drying the resultant gas through calcium oxide. III. RESULTS A. Untreated DND

Measurements were initially recorded for the DND powder prior to exposure to ammonia. Figure 2共a兲 shows the real impedance plotted against the measurement frequency at room temperature in both air and vacuum; it can be seen that the DND material trapped between the electrodes has a very high dc resistance 共greater than 1011 ⍀兲 and that the ac resistance remains very high beyond a measurement frequency of 1 MHz. The dc resistance value can be seen to be around a factor or two lower in the case of the air measurement compared to vacuum, but this difference is lost once the measurement frequency exceeds 100 Hz. Figure 2共b兲 shows similar data, recorded in air, but at temperatures of 400 and 550 ° C. In each case the measurement was performed several times. Interestingly, the measurements at 400 ° C are both stable when repeated and almost identical to those re-

123704-3

J. Appl. Phys. 106, 123704 共2009兲


semicircle magnitude with increasing temperature can be seen. Measurements continued to be made throughout this temperature range. Figure 3共c兲 shows data recorded at 450– 550 ° C after a total of four measurement cycles. The trend with temperature remains the same, but the overall magnitude of the semicircular response can be seen to have decreased since the first measurements were performed at these temperatures. B. Exposure to ammonia

DND powder was exposed to ammonia vapor 共30 min兲 and the impedance measurements repeated. Interestingly, at room temperature it was now possible to observe two semicircular responses in the Cole–Cole plot, where none were seen prior to ammonia exposure 关Fig. 4共a兲兴. The resistivity range of both semicircles is now much lower than that observed for the DND prior to ammonia exposure, being 1 – 10 M⍀. Heating the ammonia-treated DND powder led to the loss of these semicircles, and the return of characteristics typical for the unexposed DND 关Fig. 4共b兲, 100– 300 ° C; Fig. 4共c兲, 400– 550 ° C兴. These characteristics could be observed after repetitive ammonia exposure-heating cycles. IS was then performed at room temperature after heating an ammonia-exposed sample for one minute at 100, 200, 300, 400, and 550 ° C, as shown in Fig. 5共a兲共100 ° C兲 and Fig. 5共b兲共150– 550 ° C兲. The semicircular response recorded at low frequencies 共right-hand side兲 can be seen to diminish in magnitude following 100 ° C treatment, and to disappear at higher temperatures. The activation energy for both semicircular responses, derived from these data 共Fig. 6兲, gives values that are similar, although they strongly vary with the temperature range 共40–90 meV below 60 ° C, ⬃0.5 eV above兲. IV. DISCUSSION

FIG. 3. Cole–Cole plots for DND powders measured in air at 共a兲 350– 450 ° C, 共b兲 450– 550 ° C and 共c兲 450– 550 ° C after several measurement cycles.

corded at room temperature. However, at 550 ° C the curves can be seen to decrease in resistivity with increasing measurement cycle, with the dc resistance falling by almost two orders of magnitude. Electrical conduction through the films could only be measured at temperatures of 350 ° C and above; Fig. 3共a兲 shows Cole–Cole plots, measured in air, recorded at 350, 400, and 450 ° C. Single semicircular responses can be seen which decrease in overall size with increasing temperature. Similar data are shown in Fig. 3共b兲, but now for measurements at 450– 550 ° C, where the same trend of decreasing

Extrapolation of the line displayed in Fig. 2共a兲 to zero frequency enables the dc resistivity of the film to be determined, revealing a value greater than 1013 ⍀ / sq. This value is typical for the highest quality polycrystalline CVD material18 and remarkably close to values obtained for single crystal diamond. This is an excellent result given that this is nanostructured detonation powder, which has often been considered to comprise a composite of sp3, sp2, and impurity species.15,19 An important characteristic of an insulating film is the dielectric loss tangent 共tan ␦兲, determined as a function of ac frequency. tan ␦ is the imaginary part of the complex dielectric constant as a ratio with its real component; it can be shown that this value is equal to the ratio of the real part of the complex impedance 共Z⬘兲 divided by its imaginary component 共Z⬙兲.20 For the data shown in Fig. 2共a兲, tan ␦ is around 0.3 at 0.1 Hz, falling to 0.02 at 1 MHz for the air measurements, and 0.04 falling to less than 0.01 for the vacuum measurements. It should be remembered that the function tan ␦ is a measure of the ratio of the power dissipated in the dielectric to the power stored, implying low numbers are desirable. Lu et al.21 investigated the dielectric properties of very large grain thick freestanding diamond

123704-4


FIG. 4. Cole–Cole plots for DND powders exposed to ammonia vapor 共30 minutes兲 measured in air at 共a兲 room temperature, 共b兲 100, 200, and 300 ° C, and 共c兲 400 and 550 ° C.

films 共0.3–1.5 mm兲. Over the frequency range we have investigated here, they observed an increasing value of tan ␦ from 0.05 to 0.085, while in our case the trend in the data is the opposite. The decreasing trend is a desirable feature since it shows the dielectric loss is decreasing with increasing frequency. In another report on the use of commercially available thick CVD polycrystalline diamond films, Ibarra and co-workers22 measured a dielectric loss tangent of around 0.01 falling to around 0.001at 10 MHz. In this case the trend is similar to that we observe here, decreasing with increasing frequency, and with similar values, at least for the vacuum measurements performed. The DND samples can be consid-

J. Appl. Phys. 106, 123704 共2009兲

FIG. 5. Cole–Cole plots for DND powders exposed to ammonia vapor measured in air at room temperature after heating for one minute at 共a兲 30– 60 ° C, 共b兲 100– 120 ° C, and 共c兲 150– 550 ° C.

ered as a three-dimensional network of the nanodiamond particles. They are not continuous films; rather they contain many interfaces formed by the surface of the nanodiamond particles and voids where the irregular particles are not perfectly packed. Due to the large specific surface area of the nanodiamond particles, the interfaces should play an important role on the electronic properties of the material. It is thus surprising that the dielectric properties are so similar to single crystal diamond, indicating that this particular DND powder is principally an sp3 material, even at the boundaries between the agglomerated nanocrystals. The most likely rea-

123704-5

J. Appl. Phys. 106, 123704 共2009兲


TABLE I. Resistance and capacitance values determined from fits to the impedance data contained in Fig. 5. Low frequency component

FIG. 6. Arrhenius plots showing activation energy for semicircular responses in Fig. 5 at temperatures between 共a兲 30– 60 ° C and 共b兲 80– 120 ° C.

son for the difference in the values 共both in terms of impedance and tan ␦兲 is the presence of adsorbates on exposed surfaces of the powder when the measurements are carried out in air at room temperature. The fact that little change is evident in the data recorded at 400 ° C 关Fig. 2共b兲兴 indicates good stability of this dielectric material to this temperature. That this is not so at 550 ° C 关Fig. 2共b兲兴 shows that changes to the film are now occurring, as the reduction in impedance values 共and an increase in the value of tan ␦兲 is not reversible. The most obvious explanation would be the onset of sp2 formation in the DND powder at this temperature; this is discussed further below. The Cole–Cole plots in Fig. 3 indicate that electrical conduction through the films can be detected at temperatures greater than 350 ° C. The reduction in the magnitude of the semicircular response with increasing temperature indicates that the powder sample become more conductive as the temperature increases, which is expected behavior. The fact that only a single semicircular response is seen is interesting. In previous studies of diamond using IS, more than one semicircular response has often been observed,21–24 indicating more than one conduction path. In the case of polycrystalline and nanocrystalline diamonds, this observation has been attributed to conduction through both the grains and grain

High frequency component

Temperature

C 共⫻10−11 F兲

R 共⫻107 ⍀兲

C 共⫻10−9 F兲

R 共⫻107 ⍀兲

30 40 50 60 70 80 90 100 110 120

3.15 2.96 2.84 2.79 2.82 2.83 2.85 2.79 2.88 2.90

2.12 2.23 2.28 2.46 2.96 3.85 5.41 9.60 15.60 26.65

2.78 3.35 4.28 4.93 5.40 5.73 6.19 5.64 2.60 5.97

4.53 5.18 5.67 6.41 7.80 10.45 14.94 24.15 45.02 148.52

boundaries. The fact that only one semicircle is seen here indicates a single conduction path exists. Mathematical fitting to the semicircles can lead to the values of R and C associated with it to be determined. For the data contained in Fig. 3 for 350 ° C, the capacitance values all lie in the picofarad range. It has been previously shown that grain interior conduction displays capacitance values of this order, whereas grain boundary conduction more often displays nanofarad values.20 Hence, it is possible to suggest that the surfaces and interfaces between the nanometer-scale diamond crystals within the agglomerated particles are not supporting a second form of electrical conduction when compared to that derived from the crystal interiors at this temperature. However, for the semicircles measured at 550 ° C after a number of measurement cycles display capacitance values in the 1–10 nF range, indicative of grain-boundary-type conduction. It is known that heating diamond in air at temperatures greater than around 400 ° C leads to oxidation of the diamond surface,23 typically increasing the resistivity values determined in any subsequent measurements. This is clearly not the case here. Instead, it can be suggested that the nanometer-scale crystals that are agglomerated to form the micrometer sized particles, are not stable in the sp3 form as they are heated to 550 ° C, and that some degree of sp2 formation, presumably at the boundaries between the agglomerated crystals, is now occurring. An Arrhenius plot derived from the data in Fig. 3 gives an activation energy for this conduction process of 1–1.5 eV, depending on the number of measurement cycles. While it is not possible to assign a physical process to this conductivity from these data alone, it can be noted that activation energies in this range have previously been seen in high temperature impedance studies of polycrystalline diamond.25 The observation of conduction paths, as witnessed by the presence of room temperature semicircular responses in the Cole–Cole plots 共Figs. 5 and 6兲 with relatively low resistances, suggests ammonia has modified the electrical properties of the DNDs. That there are two semicircular responses indicates two differing conduction paths, which are differentiated by their respective capacitance values, while having similar resistivity levels 共Table I兲. It is very interesting to note that at this temperature, picofarad and nanofarad re-

123704-6

J. Appl. Phys. 106, 123704 共2009兲


of a chemisorbed species that can be removed at modest temperatures, which strongly modifies the electronic properties of the DNDs, is encouraging. The most likely origin of this conductivity, some form of surface transfer doping, would suggest that the adsorbate will only be required in monolayer concentrations, suggesting that this process may lead to a highly sensitive solid state chemical detection capability. Moreover, the high surface area offered by the use of compacted DND powders would also enable high sensitivity to be achieved. V. CONCLUDING REMARKS

FIG. 7. FTIR data showing an absorbance peak near 850 cm−1.

sponses suggest a grain-boundary-like conduction process is being accompanied by a grain-interior-like conduction. Heating the ammonia-exposed DNDs above room temperature increases the resistivity associated with both conduction paths 共Fig. 5兲, which are eventually lost completely. The activation energies associated with both appear 共meV兲 at modest temperatures, but when the ammonia-exposed DNDs are subjected to higher temperatures a relatively high activation energy 共0.5 eV兲 appears to be associated with further loss of conductivity within these states. The ability for diamond surfaces to display differing levels of surface conductivity depending on the chemical nature of the surface termination 共typically H or O兲 has been extensively reported,10,11 and that ammonia can affect this property.12 In the latter case hydrogenated diamond surfaces became less conductive when exposed to ammonia, and more conductive when exposed to NO2. Given the observations made here regarding the highly resistive nature of the DNDs prior to ammonia exposure, the presence of any form of hydrogen termination, which would lead to surface conductivity, can be ruled out. Instead it would appear that exposure to ammonia strongly increases the conductivity of this form of diamond. This process is clearly reversible as the conduction is lost upon heating. The presence of the two conduction paths would imply that an ammonia-derived adsorbate may be affecting the subsurface region, causing band bending and a form of surface transfer doping,26 and that this is accompanied by a grain boundary conduction caused by ammonia exposure. The activation energy associated with conduction loss at higher temperatures 共0.5 eV兲 is most likely due to the desorption of surface attached ammonia-derived species; this energy is typical of the bond strength of chemisorbed species.27 The copresence of water or a water-derived species cannot be ruled out, but the exposure to water vapor alone 共i.e., in the absence of NH3兲 did not give rise to similar observations. Indeed, the Fourier transform infrared 共FTIR兲 data 共Fig. 7兲 provides some evidence for the presence of ammonia-derived chemisorption on the DNDs; the peak that appears near to 850 cm−1 having been previously linked with NH2 species.28 The work performed here does not directly relate to the use of DNDs for ammonia sensing, in that controlled flux experiments have not been carried out. However, the observation that ammonia exposure appears to lead to the creation

The commercially supplied form of DND studied here has been found to be highly resistive, and capable of performing as a near-to-ideal dielectric at temperatures below 350 ° C. This is contrary to some previous observations on other DND types, and challenges the idea that the nanometer-sized detonation-formed diamond crystals agglomerate together via a network of non-sp3 carbon that is impervious to annealing and/or acid treatments. Indeed, the measurements performed here suggest that only one form of electrical conduction occurs within these materials, which is grain-interior-like at temperatures below 350 ° C, but may be due to surface/interface conduction at higher temperatures. This implies that these diamond powders may be suitable for several applications where chemically functionalizing nanometer-scale sp3 material is desirable, and for those requiring a dense dielectric powder material under either dc or microwave frequency conditions. At temperatures above 350 ° C, the films resistivity begins to decline irreversibly indicating a limitation to the operational temperature for the powder within any application. This change is most likely to be caused by the onset of sp2 formation on the surfaces of the nanometer-scale diamond crystals. Once exposed to ammonia, ammonia-derived species 共most likely to be NH2兲 appear to chemisorb reversibly to the DNDs with a desorption activation energy of 0.5 eV. The presence of the ammonia-derived species creates two parallel conduction paths, with relatively high conductivity, which appear to relate to grain interior conduction and grain boundary conduction. It seems reasonable to speculate that the former occurs through a form of surface transfer doping. Since the resistivity of the DNDs is changed by around 107 orders upon ammonia exposure, it would appear that there may be a promising future for DND-based solid state ammonia sensors based on this approach. ACKNOWLEDGMENTS

This work has been partially financially supported by the U.K.’s Engineering and Physical Sciences Research Council 共EPSRC兲共Award No. EP/D029651/1兲. 1

B. A. López de Mishima, D. Lescano, T. Molina Holgad, and H. T. Mishima, Electrochim. Acta 43, 395 共1998兲. 2 B. Timmer, W. Olthuis, and A. van den Berg, Sens. Actuators B 107, 666 共2005兲. 3 M. Bendahan, P. Lauque, J.-L. Seguin, K. Aguir, and P. Knauth, Sens. Actuators B 95, 170 共2003兲. 4 S. DuBois, S. Eng, R. Bhattacharya, S. Rulyak, T. Hubbard, D. Putnam, and D. J. Kearney, Dig. Dis. Sci. 50, 1780 共2005兲.

123704-7

Properties and Growth of Diamond, edited by G. Davies 共INSPEC, Institution of Electrical Engineers, London, 1998兲. 6 V. Mortet, O. A. Williams, and K. Haenen, Phys. Status Solidi A 205, 1009 共2008兲. 7 O. A. Williams, V. Mortet, M. Daenen, and K. Haenen, J. Nanosci. Nanotechnol. 9, 3483 共2009兲. 8 X. Ji, C. E. Banks, and R. G. Compton, Electroanalysis 18, 449 共2006兲. 9 J. de Sanoit, E. Vanhove, P. Mailley, and P. Bergonzo, Electrochim. Acta 54, 5688 共2009兲. 10 H. J. Looi, J. S. Foord, and R. B. Jackman, Appl. Phys. Lett. 72, 353 共1998兲. 11 H. J. Looi, L. Y. S. Pang, A. B. Molloy, F. H. Jones, J. S. Foord, and R. B. Jackman, Diamond Relat. Mater. 7, 550 共1998兲. 12 R. S. Gi, T. Ishikawa, S. Tanaka, T. Kimura, Y. Akiba, and M. Iida, Jpn. J. Appl. Phys., Part 1 36, 2057 共1997兲. 13 Q. Wang, S. L. Qu, S. Y. Fu, W. J. Liu, J. J. Li, and C. Z. Gu, J. Appl. Phys. 102, 103714 共2007兲. 14 Synthesis, Properties and Applications of Ultrananocrystalline Diamond, Nato Science Series Vol. 192, edited by D. M. Gruen, O. A. Shenderova, and A. Y. Vul 共Springer, Netherlands, 2005兲. 15 O. A. Shenderova, V. V. Zhirnov, and D. W. Brenner, Crit. Rev. Solid State Mater. Sci. 27, 227 共2002兲. 16 A. Krüger, F. Kataoka, M. Ozawa, T. Fujino, Y. Suzuki, A. E. Aleksenskii, A. Ya. Vul’, and E. Ōsawa, Carbon 43, 1722 共2005兲. 5

J. Appl. Phys. 106, 123704 共2009兲


A. Huanosta and A. R. West, J. Appl. Phys. 61, 5386 共1987兲. J. R. Brandon, S. E. Coe, R. S. Sussmann, K. Sakamoto, R. Sporl, R. Heidinger, and S. Hanks, Fusion Eng. Des. 53, 553 共2001兲. 19 T. Hamilton, E. Z. Kurmaev, S. N. Shamin, P. Y. Detkov, S. I. Chukhaeva, and A. Moewes, Diamond Relat. Mater. 16, 350 共2007兲. 20 L. L. Hench and J. K. West, Principles of Electronic Ceramics 共Wiley, New York, 1989兲, Chap. 5. 21 F. X. Lu, H. D. Zhang, Y. M. Tong, J. X. Yang, C. M. Li, G. C. Chen, and W. Z. Tang, Diamond Relat. Mater. 13, 1714 共2004兲. 22 A. Ibarra, M. Gonzalez, R. Vila, and J. Molla, Diamond Relat. Mater. 6, 856 共1997兲. 23 V. L. Kuznetsov, Yu.V. Butenko, A. L. Chuvilin, A. I. Romanenko, and A. V. Okotrub, Chem. Phys. Lett. 336, 397 共2001兲. 24 H. Ye, O. A. Williams, R. B. Jackman, R. Rudkin, and A. Atkinson, Phys. Status Solidi A 193, 462 共2002兲. 25 C. Feliciangeli, M. C. Rossi, and G. Conte, Diamond Relat. Mater. 16, 930 共2007兲. 26 P. Strobel, M. Riedel, J. Ristein, and L. Ley, Nature 共London兲 430, 439 共2004兲. 27 L. H. Chua, R. B. Jackman, J. S. Foord, P. R. Chalker, C. Johnston, and S. Romani, J. Vac. Sci. Technol. A 12, 3033 共1994兲. 28 B. C. Smith, Infrared Spectral Interpretation: A Systematic Approach 共CRC, Boca Raton, FL, 1999兲, pp. 138–139. 17 18